GLOBAL DYNAMICS OF SOCIAL POLICY

## **Causal Mechanisms in the Global Development of Social Policies**

Edited by **Johanna Kuhlmann Frank Nullmeier**

**Global Dynamics of Social Policy** CRC 1342

## Global Dynamics of Social Policy

#### **Series Editors**

Lorraine Frisina Doetter University of Bremen Bremen, Germany

Delia González de Reufels University of Bremen Bremen, Germany

Kerstin Martens University of Bremen Bremen, Germany

Marianne Sandvad Ulriksen University of Southern Denmark Odense, Denmark

Tis open access series welcomes studies on the waves, ruptures and transformative periods of welfare state expansion and retrenchment globally, that is, across nation states and the world as well as across history since the inception of the modern Western welfare state in the nineteenth century. It takes a comprehensive and globalized perspective on social policy, and the approach will help to locate and explain episodes of retrenchment, austerity, and tendencies toward de-welfarization in particular countries, policy areas and/or social risk-groups by reference to prior, simultaneous or anticipated episodes of expansion or contraction in other countries, areas, and risks. One of the aims of this series is to address the diferent constellations that emerge between political and economic actors including international and intergovernmental organizations, political actors and bodies, and business enterprises. A better understanding of these dynamics improves the reader's grasp of social policy making, social policy outputs, and ultimately the outcomes of social policy.

More information about this series at https://link.springer.com/bookseries/16294 Johanna Kuhlmann • Frank Nullmeier Editors Causal Mechanisms

in the Global Development of Social Policies

*Editors* Johanna Kuhlmann University of Bremen Bremen, Germany

Frank Nullmeier University of Bremen Bremen, Germany

ISSN 2661-8672 ISSN 2661-8680 (electronic) Global Dynamics of Social Policy ISBN 978-3-030-91087-7 ISBN 978-3-030-91088-4 (eBook) https://doi.org/10.1007/978-3-030-91088-4

© Te Editor(s) (if applicable) and Te Author(s) 2022. Tis book is an open access publication. **Open Access** Tis book is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this book are included in the book's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the book's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

Te use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specifc statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use.

Te publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. Te publisher remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

Tis Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG. Te registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

## **On this Series**

Te intervention of states in felds such as health, social security and work dates back to the nineteenth century, and became more dynamic over time. Imperial Prussia, a social policy pioneer, frst showcased its progress at the Paris World Fair in 1900: the Prussian exhibit drew large crowds eager to fnd out more about state pensions. Clearly, social policy had become a matter of great interest to states and citizens alike.

Other nations soon embarked on implementing discrete social policies, thus turning the twentieth century into a time of remarkable welfare state expansion. Te end of World War II marked a new departure, as an increasing number of countries outside the Western hemisphere began to introduce social policy measures. States not only copied established forms of welfare, but often developed measures *sui-generis* to meet their specifc needs. While episodes of policy retrenchment and ruptures can be observed over time, recent developments point to an expansion of social policies in low-to-upper-middle-income countries of the Global South. Social policy has thus become a global phenomenon.

It is generally accepted that the state is responsible for welfare and that domestic politics and ideas have been a primary driver of its expansion. However, in an increasingly interconnected world, social policy is implemented at the national level but infuenced by international developments and relations. It is shaped by trade, migration, war and colonialism. Just as people travel, policy ideas follow. Tese factors merit scholarly attention and demand interdisciplinary collaboration to generate new insights into the global dimension of social policy.

Tis is what the Global Dynamics of Social Policy book series sets out to accomplish. In doing so, it also contributes to the mission of the Collaborative Research Center 1342 (CRC) "Global Dynamics of Social Policy" at the University of Bremen, Germany. Funded by the German Research Foundation, the CRC leaves behind the traditionally OECDfocused analysis of social policy to stress the transnational interconnectedness of developments.

Te book series showcases scholarship by colleagues worldwide who are interested in the global dynamics of social policy. Studies can range from in-depth case studies, comparative work and large quantitative research. Moreover, the promotion of scholarship by young researchers is of great importance to the series.

Te series is published in memory of Stephan Leibfried to whom our research on state and social policy at the CRC is indebted in countless ways.

Bremen, Germany Lorraine Frisina Doetter Odense, Denmark Delia González de Reufels Kerstin Martens Marianne Sandvad Ulriksen

## Praise for *Causal Mechanisms in the Global Development of Social Policies*

"Causal mechanisms meet social policy research in this masterful volume that ofers clear theoretical and methodological guidance, while providing concrete examples from across the world. Kuhlmann, Nullmeier, and the impressive group of scholars they have gathered ofer an innovative mechanism-based approach to study social policy processes that elegantly fuses the actions of policy actors with complex theory-centered causal sequences. Tis book's theoretical and methodological insights make it a must-read for social policy scholars across disciplines."

—Sara Niedzwiecki, *University of California, Santa Cruz, USA*

## **Contents**




## **Notes on Contributors**

**Armando Barrientos** is Professor Emeritus of Poverty and Social Justice at the Global Development Institute at the University of Manchester and Mercator Fellow at the CRC 1342, University of Bremen.

**Stephen Devereux** is a professorial fellow at the UK Institute of Development Studies, co-director of the Centre for Social Protection, and Mercator Fellow at the CRC 1342, University of Bremen.

**Ertila Druga** is a doctoral researcher within the CRC 1342 project B08 "Transformation of Health Care Systems in Central and Eastern Europe", University of Bremen.

**Delia González de Reufels** is Professor of Latin American History at the Institute of History and principal investigator of the CRC 1342 projects B01 "Mechanisms of Social Policy Dynamics" and B02 "Emergence, Expansion, and Transformation of the Welfare State in the Cono Sur in Exchange with (Southern) Europe (1850–1990)", University of Bremen.

**Karin Gottschall** is Professor of Sociology at SOCIUM Research Center on Inequality and Social Policy and principal investigator of the CRC 1342 project B07 "Transnational Service Provision in Long-Term Care between Eastern and Western Europe", University of Bremen.

**Andreas Heinrich** is a postdoctoral researcher within the CRC 1342 project B06 "External Reform Models and Internal Debates on the New Conceptualisation of Social Policy in the Post-Soviet Region" at the Research Centre for East European Studies, University of Bremen.

**Teresa Huhle** is a former postdoctoral researcher within the CRC 1342 project B02 "Emergence, Expansion, and Transformation of the Welfare State in the Cono Sur in Exchange with (Southern) Europe (1850–1990)", University of Bremen, and currently a postdoctoral researcher at the Department of Iberian and Latin American History, University of Cologne.

**Gulnaz Isabekova** is a researcher and doctoral student within the CRC 1342 project B06 "External Reform Models and Internal Debates on the New Conceptualisation of Social Policy in the Post-Soviet Region" at the Research Centre for East European Studies, University of Bremen.

**Johanna Kuhlmann** is a postdoctoral researcher at SOCIUM Research Center on Inequality and Social Policy and within the CRC 1342 project B01 "Mechanisms of Social Policy Dynamics", University of Bremen.

**Tao Liu** is Professor of Social Policy at the School of Public Afairs, Zhejiang University, China, and principal investigator of the CRC 1342 project B05 "Dynamics of Chinese Social Policy. Interplay of National and International Infuences", University of Bremen.

**Ante Malinar** is a doctoral researcher within the CRC 1342 project B08 "Transformation of Health Care Systems in Central and Eastern Europe", University of Bremen.

**Armin Müller** is a postdoctoral researcher within the CRC 1342 project B05 "Dynamics of Chinese Social Policy. Interplay of National and International Infuences", University of Bremen and Jacobs University Bremen.

**Kristin Noack** is a PhD researcher within the CRC 1342 project B07 "Transnational Service Provision in Long-Term Care between Eastern and Western Europe", University of Bremen.

**Frank Nullmeier** is Professor of Political Science at SOCIUM Research Center on Inequality and Social Policy and principal investigator of the CRC 1342 project B01 "Mechanisms of Social Policy Dynamics", University of Bremen.

**Kerem Gabriel Öktem** is a postdoctoral researcher within the CRC 1342 project B01 "Mechanisms of Social Policy Dynamics", University of Bremen.

**Heiko Pleines** is Professor of Comparative Politics and Head of the Department of Politics & Economics of the Research Centre for East European Studies, and principal investigator of the CRC 1342 project B06 "External Reform Models and Internal Debates on the New Conceptualisation of Social Policy in the Post-Soviet Region", University of Bremen.

**Heinz Rothgang** is Professor of Health Economics and Health Policy at SOCIUM Research Center on Inequality and Social Policy and principal investigator of the CRC 1342 projects A04 "Global Developments in Health Care Systems and Long-Term Care as a New Social Risk" and B07 "Transnational Service Provision in Long-Term Care between Eastern and Western Europe", University of Bremen.

**Anna Safuta** is a postdoctoral researcher within the CRC 1342 project B07 "Transnational Service Provision in Long-Term Care between Eastern and Western Europe", University of Bremen.

**Klaus Schlichte** is Professor for International Relations and World Society at the Institute for Intercultural and International Studies (InIIS) and principle investigator of the CRC 1342 projects B01 "Mechanisms of Social Policy Dynamics" and B09 "Te Rise, Decay and Renaissance of Social Policy in Africa", University of Bremen.

**Sebastian Sirén** is a PhD candidate in Sociology at the Swedish Institute for Social Research, Stockholm University.

**Tobias ten Brink** is Professor of Chinese Economy and Society at Jacobs University Bremen and principal investigator of the CRC 1342 project B05 "Dynamics of Chinese Social Policy. Interplay of National and International Infuences", University of Bremen.

**Kressen Tyen** is a postdoctoral researcher within the CRC 1342 project B09 "Te Rise, Decay and Renaissance of Social Policy in Africa" at the Institute for Intercultural and International Studies (InIIS), University of Bremen.

## **List of Figures**


#### **xviii List of Figures**


## **List of Tables**


## **Part I**

**Introductory Section**

## **1**

## **Introduction: A Mechanism-Based Approach to Social Policy Research**

**Frank Nullmeier and Johanna Kuhlmann**

## **1 Introduction**

How to explain developments in social policy is a matter of longstanding debate. As Baldwin noted in 1990 (36), "[s]cores of theories compete to explain why it [the welfare state, FN/JK] exists at all, dozens of comparative analyses account for its variations, legions of narratives detail how individual examples contradict or confrm general hypotheses". Not surprisingly, more than 30 years later social policy research continues to be distinguished by its plurality of approaches. Many of them have been developed to capture developments in OECD welfare states. Yet, since social policy scholars are now turning more to researching the *global* dynamics of social policy, the question of to what extent these existing

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Projektnummer 374666841—SFB 1342.

F. Nullmeier • J. Kuhlmann (\*)

University of Bremen, Bremen, Germany

e-mail: frank.nullmeier@uni-bremen.de; johanna.kuhlmann@uni-bremen.de

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_1

social policy approaches can be applied to countries and regions beyond the OECD world comes to the fore (Kpessa and Béland 2013; Lavers and Hickey 2016; Veit et al. 2017; Schmitt et al. 2020). At the same time, aspects that have long been less evident in social policy research become more visible when studying the global dynamics of social policy, for example the role of transnational policymaking (Schmitt 2020; Leisering 2019) or autocracies (Mares and Carnes 2009; Knutsen and Rasmussen 2018; Eibl 2020). Tese approaches can, in turn, also inform analyses of countries that are known to be the "usual suspects" in social policy research. As a result, the list of approaches for social policy research continues to become longer.

When studying the global dynamics of social policy through qualitative case studies, this plurality can come as a challenge: How can we—in a meaningful way—compare social policy developments in very diferent countries and regions, relating fndings from such arguably diferent countries as, for example, Uganda, Turkey, and Malaysia, or Bolivia and South Korea? How can we compare developments within old-age provision to those in employment policy? And is it possible to liken developments from the nineteenth century or colonial times to social policies in the twenty-frst century? Which analytical level is suitable for such endeavours? Given this background, this edited volume seeks to fnd new ways of explaining social policy. It introduces causal mechanisms as the key concept of such a new explanatory approach. Te key argument that we present throughout this book is that causal mechanisms can generate explanations that can complement, expand, deepen, and, in some cases, even correct analyses that rely on established approaches to social policy.

Causal mechanisms facilitate the identifcation of causal chains and patterns, but not at the level of political systems, regimes, structures, policy felds, or long-term processes. Rather, causal mechanisms trace smaller regularities at the level of sequences, which can be observed in political processes at diferent times, in diferent places, and on diferent topics. Importantly, these kinds of regularities enable a more than descriptive access to the development of social policy. To put it in the words of Charles Tilly: "great social regularities do not occur at the level of whole structures, full sequences, or total processes but in the detailed social mechanisms that generate structures, sequences, and processes" (Tilly 2008, 123). Tus, the conceptual advantage of causal mechanisms is that they provide the basis for a decomposed form of comparison at sub-levels of political processes, since processes can be broken down into smaller units and sequences can be analysed in a far more diferentiated manner. What this mechanism-based approach does not imply, importantly, is that the mechanisms identifed need to be integrated afterwards into one particular approach to social policy.

A focus on causal mechanisms can thus especially (but not only) promote qualitative approaches in social policy research. A quantitative research strategy has been found to be attractive for researchers because of its use of extremely sparse ontological assumptions about the social and political world. Whatever factor might play a role in explaining social policies is turned into a variable that can be measured at diferent scales. Tis contributes to the extreme variability and adaptability of quantitative research, but interpreting its results coherently becomes difcult. For qualitative research, which draws its momentum from a critique of the quantitative, probabilistic, and correlation-based approach (George and Bennett 2005; Goertz and Mahoney 2012), imitating such a theoretical and ontological approach can become a serious obstacle to further developing social policy research.

Until now, research on social policy has only rarely been combined with an approach that *explicitly* uses causal mechanisms as a theoretical or methodological concept. Difusion research has relied on a well-known set of mechanisms to analyse social policy developments (Obinger et al. 2013), but more often than not these mechanisms are not clear when it comes to the key requirement of mechanism-based research, namely to detail the basic elements of a causal process and how they are linked to produce an outcome (Hedström and Ylikoski 2010). Similar points of critique could be raised for mechanisms of gradual institutional change (Streeck and Telen 2005). In recent years, however, mechanism-based research has gained traction. Obinger, Petersen, and Starke (2018) focus on a set of mechanisms that explain social policy expansion in the context of war, thereby distinguishing the phases of war preparation, mobilisation, and the post-war period. Leisering (2019) introduces the mechanisms of "cultural linkages", "theorisation", and "quantifcation" as three mechanisms of global social policy difusion. And a recent issue of *Social* 

#### **6 F. Nullmeier and J. Kuhlmann**

*Policy & Administration* applies a mechanism-based approach to studying transnational social policy dynamics in the Global South (for an overview see Kuhlmann and ten Brink 2021). We argue that there is great potential for social policy research in continuing such eforts, especially for comparisons between case studies of historically widely divergent situations, diferent felds of social policy, and very diferent countries.

In this introductory chapter, we will present an approach to causal mechanisms that is modular, process-oriented, and actor-centred. It guides the analysis in this book's chapters. We proceed as follows: First, we root our approach to mechanism-based research more broadly in the history of causal mechanisms as a social sciences concept by distinguishing four central strands of research. Subsequently, we present our approach to causal mechanisms which includes the idea of modularisation as well as a distinction between elementary and complex causal mechanisms. Tereafter we discuss the question of what social policy research can gain from a mechanism-based approach, further detailing our key argument that causal mechanisms can improve social policy research in a number of ways. Finally, we give a brief overview of the chapters in this volume.

### **2 Causal Mechanisms as a Concept in the Social Sciences**

Contributions to the literature on causal mechanisms usually include a defnition of a causal mechanism, and the growing literature on the topic has resulted in many—sometimes complementary, sometimes competing—attempts to defne the term. In fact, it is striking that the existing literature on causal mechanisms by no means ofers one single defnition of the term. Already in 2001, Mahoney listed 24 diferent defnitions of causal mechanisms (Mahoney 2001, 579), and extracted 3 diferent understandings of the term (Mahoney 2001, 578–82). In their review article on causal mechanisms from 2010, Hedström and Ylikoski (2010) listed nine "alternative defnitions". While diverse defnitions of single concepts can be considered a scientifc problem, it is important to note that a standardised defnition of causal mechanisms would not solve all of the (admittedly many) challenges that conceptual work on causal mechanisms is struggling with. In a similar vein, Hedström and Swedberg have argued that "it is not so much the defnition per se that is important, as the type and style of theorizing it encourages" (1996, 299). In the following, we will outline the history of the concept of causal mechanisms in the social sciences (for an even broader history, see Glennan and Illari 2018). Tis will help us to elaborate on the main theoretical options that we have when conducting mechanism-based research, and to lay the foundation for further advancing the conceptual debate. Looking at the multitude of refections on social and causal mechanisms, we will diferentiate between four strands of the discussion since the 1970s, before situating ourselves within this debate.

#### **2.1 First Strand: A Methodology of Qualitative Research**

Te concepts of "causal mechanism" and "process tracing" delineate more recent debates in political science on a specifc methodology of qualitative research (in particular: Beach and Pedersen 2019; Bennett and Checkel 2015; Goertz 2017; Goertz and Mahoney 2012). Te main rationale for creating a new qualitative research methodology was the attempt to refute an original, case study-oriented approach to research, which was prominently articulated in *Designing Social Inquiry* (King et al. 1994). Here, Gary King, Robert O. Keohane, and Sidney Verba postulated *one* causal inference logic for both quantitative and qualitative research. Studies that do not follow this universal logic would not be—so the not only implicit reasoning—valuable in scientifc terms. Te "imperialism" of such an approach triggered, with a certain time lag, an intense counter-movement (Brady and Collier 2004; George and Bennett 2005; Gerring 2006; Ragin 2000), which fnally resulted in the book *A Tale of Two Cultures. Qualitative and Quantitative Research in the Social Sciences* (Goertz and Mahoney 2012) as a manifesto for a separate methodology for qualitative research (see also Goertz 2017). By moving towards an independent methodology for qualitative research, the concept of causal mechanisms became central. Tis led researchers to search for an alternative understanding of causality and analyses that are not probabilistic. In addition, authors working more in a historical tradition used the term "process tracing" for political science and thus approached the concept of mechanisms from yet another angle (Beach and Pedersen 2019; Starke 2015; Trampusch and Palier 2016)—although the understandings of process tracing as a method are as various as the defnitions of the mechanisms that it aims to trace (see the compilation of defnitions in Trampusch and Palier 2016).

In their joint work, Goertz and Mahoney (2012) defne causal mechanisms in purely methodological terms, without referring to the social theory literature in greater detail:

Instead, we can understand causal mechanisms to mean the intervening processes through which causes exert their efects. We propose that any relatively well-developed theory will provide a discussion of causal mechanisms. Tis is equally true for theories tested in the quantitative and qualitative research traditions: they propose ideas about the causal mechanisms that link independent variables to dependent variables. (Goertz and Mahoney 2012, 100)

Notably, the basic elements of research are still variables:

For the purposes of illustrating process tracing here, I use the term mechanism to refer to a factor that intervenes between a cause and outcome. I treat mechanisms in the same way as causes and outcomes; they are particular events or specifc values on variables. Mechanisms are diferent from causes and outcomes because of their temporal position: they stand between a cause and outcome in time. Tus, in the expression *X* ⇢ *M* ⇢ *Y*, the letters refer to events or specifc values on variables, with X being treated as the cause, M as the mechanism, and Y as the outcome. (Mahoney 2015, 206)

Mechanisms are thus variables that can be arranged chronologically between cause and outcome, that is a certain subgroup of intervening variables, whereby in the graph X ⇢ M ⇢ Y, the arrows are the interesting elements which have not yet been sufciently clarifed (Goertz 2017, 32). Summing up, in this understanding variables are explained in terms of their temporal position in a causal chain, which gives the term mechanism a purely analytical status.

#### **2.2 Second Strand: Generative Mechanisms in Critical Realism**

Te lesser known yet oldest strand of a mechanism concept in the social sciences can be found in critical realism, which continues to exert a strong infuence in Anglo-Saxon sociology. Roy Bhaskar is generally regarded as the founder of this approach. In his early work (1975), he used the term "generative mechanisms" to indicate what science should explore. Critical realism emerged as a theory of science that opposed the search for general laws that had previously been common in the social sciences (the socalled Hempel–Oppenheimer model). However, the roots of this approach go back to debates on the philosophy of science in the feld of linguistics in the 1960s, which were then dominated by Noam Chomsky's approach to a "transformational grammar" (Chomsky 1965). Mechanism is a term that Chomsky uses frequently in his work. One of Chomsky's central considerations, however, was the critique of the behaviouralist scheme of stimulus and response. According to his argumentation, reaction mechanisms are not triggered by external stimuli. Rather, there are generative capacities in individuals which ensure that they can generate grammatically correct sentences. Te corresponding mechanisms are therefore not reaction mechanisms to environmental stimuli, but *generative* mechanisms. Te refections from linguistics became a research paradigm in the social sciences.

I have argued that the causal structures and generative mechanisms of nature must exist and act independently of the conditions that allow men access to them, so that they must be assumed to be structured and intransitive, i.e. relatively independent of the patterns of events of men alike. (Bhashar 1975, 56)

For critical realism, causal structures or generative mechanisms are units of a not directly perceivable world that generates the events that happen and that people can experience. Tis is their causal power. Today, Margaret Archer (1995, 2015), in the explicit tradition of Roy Bhaskar, includes the concept of "generative mechanisms" and "causal power" as an integral part of her discussion of the sociological micro-macro or structure-agency problem and the justifcation of a "morphogenetic sociology".

In these recent contributions, the structuralist roots of the concept of generative mechanisms become efective. It is not individuals who play the decisive role here, but structures and macro-phenomena that have their own causal power. To date, such discussions have not played a major role in the political science literature. Our conception of mechanismbased explanation picks up the notion that action is not triggered by external stimuli, but by individual capacities. However, we strictly reject the structuralist orientation of this strand of theory building.

#### **2.3 Third Strand: Analytical Sociology**

Te third strand, dating back to the late 1970s, is the strand of "Analytical Marxism" and "Analytical Sociology". Initially it also dealt with methodological questions but was diametrically opposed to the structuralist conceptions of critical realism. Te initial search for clarity and precision—therefore the term "analytical" (Roemer 1986)—turned into a primarily methodological debate with game theory and the rational choice models as a more suitable basis for the social sciences, which also had the advantage of being able to provide a microfoundation (van Parijs 1993, 70–85) for Marxist economic theory. In this dispute between functionalism, structuralism, and rational choice explanation (a variable-based methodology did not play a major role at that time), the concept of causal mechanisms evolved (Elster 1983, 1986, 1989). In *Nuts and Bolts for the Social Sciences* (Elster 1989), the concept of mechanisms was a kind of compromise in view of the immanent difculties of Elster's attempt to think rational choice theory, its anomalies, and limits within a unifed framework. Elster also transcended the mere methodological discussion by turning to single elements of rational choice theories and social psychological theories as well as classical sociological concepts such as social norms. Tis contributed to attempts to develop a toolbox of causal mechanisms (Elster 1989, 1999, 2015, 2017). Elster thus paved the way for a more social theoretical development of the concept of causal mechanisms, which eventually led to analytical sociology. In an early text, Elster describes the role of mechanisms in social science explanations:

To explain is to provide a mechanism, to open up the black box and show the nuts and bolts, the cogs and wheels of the internal machinery. […] Te role of mechanisms is twofold. First, they enable us to go from the larger to the smaller: from molecules to atoms, from societies to individuals. Secondly, and more fundamentally, they reduce the time lag between explanans and explanandum. A mechanism provides a continuous and contiguous chain of causal or intentional links; a black box is a gap in the chain. (Elster 1983, 24)

Central to Elster are, frst, the microfoundation, and second, closing the temporal gap between cause and efect. Later, Elster moved away from this defnition of the role of causal mechanisms in favour of an interpretation that is more strongly based on regularities (Elster 1999, 2015). Our version of a mechanism-based approach presented here, however, more closely follows Elster's early understanding of mechanisms.

Beginning with Hedström and Swedberg's anthology (1998) and Hedström's *Dissecting the Social* (2005), a broader stream of research developed that claimed the name "Analytical Sociology". Apart from Jon Elster, it built on the work of Raymond Boudon, James Coleman and Tomas Schelling and their contributions to the micro-macro issue, while maintaining an action-centred theoretical foundation for social science (Hedström 2005, 6). Additionally, scholars within the analytical sociology tradition argued that referring to a rationalist theory of action provided a far more viable justifcation for middle-range theories as advocated by Merton: "Building upon the foundations laid by them, an analytical middle-range approach to sociological theory can be developed that avoids the somewhat empiricist and eclectic tendencies of Merton's original middle-range approach" (Hedström 2005, 8–9).

#### **2.4 Fourth Strand: Historical Sociology and Historical Institutionalism**

James Mahoney is a leading researcher in the frst strand of literature on the methodology of qualitative research. However, his methodological work is anchored in the tradition of historical institutionalism, which brings together infuences from Marxism and historical sociology. In the historical institutionalist work by Teda Skocpol, Dietrich Rueschemeyer, John Stephens, Evelyne Huber, Wolfgang Streeck, Peter Hall, and Kathleen Telen—to name prominent researchers within this research strand—the value of a mechanism-based explanation is tested on major macrosociological issues. Charles Tilly occupies a special position in this line of research, because his methodological considerations evolved during the refection on his own wide-ranging historical research on revolutions, state development, and protest events. Here, research practice informs and shapes the methodology. Tis priority of empirical research also leads to lists of relevant causal mechanisms (such as environmental mechanisms, cognitive mechanisms, and relational mechanisms), especially in Tilly's collaborative work with Doug McAdam and Sidney Tarrow in *Dynamics of Contention* (McAdam et al. 2001). Tis connection between political and historical sociology comes perhaps closest to a concept of mechanism that could tackle central questions of political science.

Te work of Renate Mayntz and Fritz W. Scharpf on "Actor-centred Institutionalism" can also be considered to belong to this institutionalist strand, as their approach is also concerned with middle-range theories and explanations that consider collective actors. In his book *Games Real Actors Play: Actor-centered Institutionalism in Policy Research* (1997), Scharpf strongly relies on game theory, but he combines it with considerations from governance theory to look for meso-level mechanisms as elements for *modular explanation*. Tus, explanations are based on combining diferent causal mechanisms as modules. It follows from this idea of modularisation that basic mechanisms can be combined into more complex causal mechanisms (see below). For Renate Mayntz (Mayntz 2004, 2017, 2019, 2020), methodological questions play a greater role, as do the sociological debates on the macro-micro problem, whereby the mesolevel of organisations and collective actors is of great importance to her. Mayntz raises doubts about the ability to solve all relevant questions based on methodological individualism, as well as scepticism towards system theoretical approaches (Mayntz 2017). She has long been interested in individual events, especially those that seem surprising when measured against everyday expectations. Importantly, Mayntz (2019) demonstrates that social science cannot be preoccupied with producing general theories. In fact, there are good reasons to study how certain events unfolded in a particular single case. Ideographic, descriptive investigations which go into detail are therefore legitimate, and studying a single case can make perfect sense. Te path to modular explanations and acknowledging that also single political events, when remarkable, should be studied and explained in single cases studies, are methodological allusions that go beyond references to the rational choice paradigm.

### **3 A Modular and Actor-Centred Conception of Causal Mechanisms**

Te development of a process-oriented, actor-centred concept and a modular approach to causal mechanisms takes up ideas from three out of these four strands of research on causal mechanisms. Only the structuralist tradition of generative mechanisms is not compatible with our more actor-centred understanding of causal mechanism, as well as our process orientation.

First, the *process orientation* is elaborated in the methodological considerations in qualitative research on mechanisms and process tracing. However, with a focus on new testing procedures to check the validity of results, this strand risks turning into a discussion among methodology specialists with only limited efects on the discipline as a whole. Terefore, it is important that researchers develop a clear idea of which mechanisms could be identifed, and what signifcance these mechanisms can have for research in their feld. Second, the *focus on actors* and enriching the understanding of causal mechanisms with social theory follows analytical sociology and actor-centred institutionalism. And fnally, the idea of *modular explanation*, which turns out to be extremely relevant for research practice, is based on Elster's and Scharpf's considerations. Closely related to this is the idea that social science research must also be able to explain individual cases without withdrawing from comparative research. Here, our approach follows the course of historical institutionalism and the methodological considerations by Mayntz. It is modularisation that makes the explanation of individual cases fruitful for comparative purposes, at the level of particular mechanisms.

*Process Orientation*: We follow the strand of qualitative research in political science with regard to its strong process orientation, as embodied in particular in the literature on process tracing. Te basic elements of political processes are *events*. Events are spatiotemporally determined phenomena that can be distinguished from states as more permanent properties of entities. In contrast to objects and their states, events have something that is momentary: "Events prefer to pass away. On the other hand, every event brings about a total change in past, present, and future—simply because it gives up the quality of being present to the next event and becomes a past for it (i.e., for its future)" (Luhmann 1995, 287). Te term *causal chain* refers to the chronological sequence of causally relevant events that occur between an initial state and the outcome, for example a political decision of a public authority. Acknowledging causal relevance is the frst step in the analysis of a process. If researchers can demonstrate from the data that the individual and collective actors who were involved in a process perceive certain events as the reason for their own activities, they can assume causal relevance. By collecting additional data, researchers must show in the course of the further research process whether the presumed causal relevance can be confrmed or not. Te starting point of an analysis, the *initial state*, is determined by the research interest, the result of previous scientifc analyses, or particular expectations about causal relationships. Te end point of a causal chain is the *outcome*. Like the initial state, it is determined by the researcher's interest and previous knowledge about the respective case. A single element in the causal chain is called a *causal link*.

*Focus on Actors*: In mechanism-based research, a highly debated question is how mechanisms unfold and what their key "entities" and "activities" are (Machamer et al. 2000). Our approach to the entities of mechanisms is actor-centred. Rather than socio-economic conditions, institutionalised rules or other supposed determinants of policy developments, we consider policy actors, their actions, and interaction as the key drivers of social policy. Terefore, our conception of causal mechanisms focuses on the activities of actors and the causal relationships between these activities. Identifying events that represent activities is central to this theoretically guided analysis of mechanisms. An *activity* is an event that can be attributed to a specifc individual or collective actor. In organisations and other forms of collective actors, actions are not those of an organisation as a collective entity or those of all its members. Rather, the interpretations, preferences, and goals of action stated by representatives or elites of these collective actors are recognised as those of their organisation. For a detailed understanding of inner-organisational developments, it is therefore necessary to refer to the individual level. For analysing political decision-making processes, it is of great importance whether actions are attributed to a collective actor or if they express the intention of an individual member of an organisation. Activities can be explained by elementary causal mechanisms (see below).

Of course, events that cannot be attributed to individual or collective actors may also be highly relevant. Tese events are usually the result of many people's activities, but they cannot be attributed to specifc collective or individual actors. Examples of such events include election results, price relations, and income distributions. Tey describe the results of the interplay of activities of an (un)determinably large number of people who do not represent a collective actor. To explain such a subsequent event, for example an election result, identical mechanisms must be assumed for a large number of persons, or diferent mechanisms for different groups of persons.

Another question is whether there might be mechanisms that are determined by social situations, such as a crisis, an economic downturn, a long unresolved and contentious issue, or the pressure of certain problems. Te strongest argument here might be that situations themselves generate adjustments. Tey "force" actors to act in a certain way and impose a certain, sometimes very limited scope of action on them. However, an actor-centred understanding would focus on how actors interpret the situation, and not on the situation itself. Although the situation exists independent of the actors' interpretation, it can only engender further activities via actors' perceptions and understandings of the situation. Each reaction to a situation is based on the perception, interpretation, and action orientation of those actors who fnd themselves in the situation and have to deal with it.

For applying an actor-centred approach to causal mechanisms in comparative analyses, one way forward can be to distinguish *types of actors*, for example heads of government, trade unions, medical associations, or conservative parties. Building such a typology of actors can be challenging, especially when dealing with very diferent periods of time, regions, and policy felds, as in the present volume. Here, the research purpose and the scope of comparison determine the typology's degree of abstraction. Te clustering of causal mechanisms according to diferent types of actors is one aim of the concluding chapter (Chap. 14).

*Modularisation*: A modularised explanation focuses on the idea that a single causal link can be explained by elementary causal mechanisms, that sequences of such causal links can be explained by complex causal mechanisms, and that the entire causal chain between the initial state and the outcome can be explained by a combination of several complex causal mechanisms. *Elementary causal mechanisms* start at the level of individual and collective actors and comprise only one causal link, namely the production of activities (demands, programmes, protests, decisions, etc.). *Complex causal mechanisms* comprise several successive steps and are composed of a sequence of activities, which in turn can be explained by elementary causal mechanisms. In the following, we start by presenting modularisation at the level of elementary causal mechanisms.

*Elementary Causal Mechanisms*: An elementary causal mechanism designates a specifc form of perception, interpretation, and action orientation. Tese are understood in a very broad sense. Perception and interpretation encompass all forms of attention that are directed towards something, and the processing of what is perceived. Here, cognitive aspects with descriptive, explanatory, and prognostic elements have to be considered, as well as evaluative and normative elements. Taken together, perception, interpretation, and action orientation can explain an activity undertaken by an individual or collective actor. Te following typology of elementary causal mechanisms starts from the existing literature in sociological theory. Similarities to Max Weber's four-fold division of action types are present (instrumentally rational, value-rational, afective, and traditional) as well as references to the models of Homo Oeconomicus, Homo Sociologicus, the emotional man, and the (Gofman's) identity claimant (Schimank 2016; Little 2016). Te following list of *six elementary causal mechanisms* gives a frst impression of how a toolbox of elementary causal mechanisms could be developed. Tey are characterised by a coupling of (well-known) action models with a perceptioninterpretation component:


age, being the best within a group, being diferent), which can be combined with more cognitive or more emotional forms of perception and interpretation.

Tese six elementary causal mechanisms might serve as basic building blocks for explaining political processes on the micro-level.

*Complex causal mechanisms*: Elementary causal mechanisms explain causal links between two activities. However, a political decision-making process usually comprises a multitude of causal links. Reconstructing all causal links between an initial state and an outcome might, to a certain extent, be feasible in individual case studies. However, it is very likely that the number of causal links is so high that comparisons with other cases become difcult. Instead of comparing decision-making processes on an elementary level, research can therefore also focus on identifying sequences in a political process. Such sequences can be explained by complex causal mechanisms. Complex causal mechanisms comprise several successive steps and are composed of a sequence of activities, which in turn can be traced back to elementary causal mechanisms.

Te outcompeting mechanism as a complex causal mechanism, which is further detailed and empirically traced in Chaps. 3 and 4, can serve here to illustrate our theoretical considerations. It explains how competition between political parties in a democratic setting leads to social policy expansion. Depending on the analytical level of the analysis, several elementary causal mechanisms can be identifed within this complex causal mechanism, such as a calculatory orientation of party elites to push for social policy expansion in the frst place, or a norm orientation of party elites who consider social policy expansion appropriate due to fairness considerations. Likewise, with regard to voters who vote for the party that is suggesting social policy expansion, elementary causal mechanisms might include a calculatory orientation as well, with voters thinking that they would beneft in material terms. Or they might—like the party elites—hold a norm orientation, a comparative orientation, assessing social policy proposals with regard to other groups or policies, or possibly even an emotional orientation, linking social policy expansion to feelings of pride or the feeling of being acknowledged in the policy process. Te elementary causal mechanisms that are at play within a complex causal mechanism need to be identifed in empirical research. Tis can often mean a time-consuming research process. Terefore, analyses at the level of complex mechanisms should be a priority for social policy research. We argue that the number of complex causal mechanisms can, in principle, be infnite. A frst attempt at bringing together complex causal mechanisms relevant to social policy research will be made in Chap. 14.

Summing up, by distinguishing between elementary and complex causal mechanisms, policy processes can be disentangled into individual steps and sequences that lead to a certain efect, which can thus be analysed in more detail. While elementary causal mechanisms are mostly in the realm of social theory, complex causal mechanisms are highly relevant for social policy research, which is why they are also the primary focus of the case studies in this book. Understood as distinct process sequences, they are located at an abstraction level that can be identifed in diferent countries, at diferent points in time, and in diferent policy areas, allowing for a modular explanation.

We illustrate our modularised, process-oriented, and actor-centred approach of mechanism-based explanation in Fig. 1.1.

**Fig. 1.1** Causal mechanisms and causal chain. (Source: Own presentation)

## **4 What Can We Expect from a Mechanism-Based Approach to Social Policy Research?**

By combining a process orientation, a focus on actors, and modularisation, the concept of mechanisms ofers new perspectives for the entire feld of social policy research. Working with causal mechanisms can be an asset in macro-quantitative comparative social policy research as well as in case study-centred work on individual countries or social policy programmes. It can function as a *complement*, an *expansion*, it can *add depth* or even be a *corrective* to existing research approaches.

Te results generated in quantitative studies by using pooled time series analysis, logistic regression or other methods are often not easy to interpret because there is a time gap between measuring the independent variables and the dependent ones (Schmitt 2019, 356). Te signifcance of a causal relationship can only be assessed if several intermediate steps are assumed, for which quantitative studies, however, do not provide any evidence. Macro-quantitative social policy research can thus be *complemented* by mechanism-based studies examining these intermediate steps, which are assumed in the temporal gap between causes and efects but treated as a black box in quantitative research. Even if not all cases included in the quantitative studies will (or should!) be examined in this way, mechanism-based studies indicate which intermediate steps or which causal chains one should be able to expect between independent and dependent variables. It would thus be possible to improve the interpretation of causal efects through more precise knowledge of causal chains and mechanisms in individual cases.

Macro-quantitative studies rely on data on the relevant variables, which are often available only for a rather limited number of countries, and for a limited period of time. When aiming to analyse the historical dimension of social policy, or to include further countries of the Global South, this line of research often experiences limits of data availability. By adopting an approach that is more strongly oriented towards individual social policy programmes and decisions, however, it is possible to *expand* the scope of social policy research, as there might be historical accounts or reports on these decision-making processes and the actors that were involved, and in some cases also fles. Tese non-quantifed data can then be used. To ensure that this does not lead to a transition to merely descriptive-narrative research, a mechanism-based approach that ensures comparability between individual studies is central. In this way, it becomes possible to identify whether causal chains exist that can also be observed in countries or during periods for which sufcient quantifed data are available.

Te actor-centredness and modularisation of a mechanism-based approach also facilitates closer scrutiny of the micro-level of social policy processes and so can provide a microfoundation for previous theories. Tis enables researchers to *add depth* to the results of social policy research. In particular, this is important for qualitative social policy research, particularly for case study design. Case studies can directly focus on identifying causal mechanisms, which requires that process tracing is applied systematically. Together with a cross-case classifcation of the relevant actors in social policy, comparisons between case studies can be facilitated. If certain types of actors are referenced, and the same procedure for identifying causal mechanisms is applied across studies, comparisons across very diferent cases are possible. However, these comparisons are not about generalisation. Te question is not about regularities between variables, but about specifc sequences in causal chains that can be explained through the same complex causal mechanisms. It is not cases that are being compared, but (occurrences of) complex causal mechanisms. Tus, beyond a logic of regularities, references can be discovered between social policy developments in very diferent felds, countries, and historical periods.

A fnal beneft of a mechanism-based approach is to *correct* existing approaches and theories. Te extant repertoire of theoretical approaches often leads to a tendency to use case studies to "confrm" or "reject" respective approaches. Scholars either select certain concepts from existing approaches because they consider them to have great explanatory power, or they even take the assumptions of certain approaches for granted prior to empirical investigation. Te downside of this is that a comprehensive comparative perspective on the level of case studies is made rather difcult. To advance social policy research that is able to cover diferent regions, sectors, and time periods, causal mechanisms can also be used as a new explanatory approach that is able to overcome some of the weaknesses of case-centred research on social policy. Many of the existing approaches focus on specifc actors and/or variables. Yet, empirical results from social policy research certainly show one thing: Tere is not one valid explanation for (almost) all social policy programmes and countries that is only based on the factors highlighted in a single approach. What is more, no approach is superior to others. Rather, what social policy research shows is that approaches are being confrmed in some cases, while being disconfrmed in others. Against this background, the mechanism-based approach and modularisation establish a kind of "meta-level" that allows the combination of existing approaches. Moreover, by introducing mechanisms that were not included or focused on in existing theoretical approaches, it also contributes to correcting some of their "blind spots".

### **5 Overview on the Chapters of This Book**

Te individual chapters of this edited volume analyse social policies from very diferent countries around the globe in both single and comparative case studies. Te country chapters are structured into four parts, dealing with social policies in Asian countries (Part II), African countries (Part III), European countries (Part IV), and Latin American countries (Part V) (following the distinction of geographic regions from the United Nations Statistics Division). What is more, the chapters not only cover diferent countries, but also diferent felds of social policy, such as oldage provision, health, unemployment, work injury, long-term care, and social assistance. Many chapters focus on the development of social insurance institutions, yet the volume also includes chapters on non-contributory forms of social policy, such as social assistance or public health. In sum, the chapters thus demonstrate the great explanatory power of the mechanism-based approach that we have outlined in the preceding sections for diferent felds and institutional arrangements within the realm of social policy. In most analyses, complex causal mechanisms are clearly in the foreground. Yet, the level of detail with which the mechanisms are spelled out depends, among others, on the time span covered by the empirical analysis, the number of countries, as well as the felds and programmes that were analysed.

Tis section gives a short introduction to the diferent chapters, especially with regard to the countries and social policies that they focus on. Te mechanisms that the authors identify in the diferent chapters are presented and discussed in the concluding chapter of this volume (Chap. 14).

Following this introduction, the second part of the book deals with the development of social policies in Asian countries. *Ten Brink*, *Müller*, and *Liu* analyse the development of social protection schemes for urban workers in China, which were adopted as part of longer-term reform policies. More specifcally, they trace the complex causal mechanisms that can explain the introduction of the Urban Employees' Basic Pension Insurance in 1997, the Urban Employees' Basic Medical Insurance in 1998, and the Work-Related Injury Insurance in 2004 (Chap. 2). *Kuhlmann* and *Nullmeier* analyse the development of two types of contribution-based pension systems, focusing on the cases of South Korea and Vietnam (who have a social insurance scheme) and Sri Lanka and Malaysia (who rely on national provident funds). Despite well-known problems in all four countries—most importantly with regard to efective coverage—the systems have been maintained or even expanded since their introduction, which the authors explain by several complex causal mechanisms (Chap. 3). *Öktem* zooms in on the development of the unemployment insurance programme in Turkey, which was established in 1999. He focuses on the complex causal mechanisms that have transformed the programme's initial focus from unemployment protection to active labour market policies (Chap. 4). Finally, *Heinrich, Isabekova*, and *Pleines* apply the mechanism-based approach to countries in the post-Soviet region—thus covering both Asian and European countries—and trace the complex causal mechanisms that contributed to the introduction of mandatory health insurance in some countries within that region after the end of the Soviet Union (Chap. 5).

In the third part of the book, we turn our attention to social policies in African countries. Focusing on the highly fragmented and exclusionary social insurance schemes in Tunisia and Uganda, *Tyen* and *Schlichte*

show that their roots can be traced back to decolonialisation, and analyse the complex causal mechanisms that were at play throughout this process (Chap. 6). In contrast, *Devereux*'s analysis does not focus on single African countries, but illuminates the role of international development agencies in promoting social protection policies, and the strategies these actors use to convince African policymakers of their ideas, focusing on one complex causal mechanism (Chap. 7).

Te fourth part of the book focuses on social policies in European countries. Analysing the reform process in the healthcare system in Croatia in the early 1990s, *Malinar* identifes the complex causal mechanisms that contributed to the development of a hybrid healthcare system that can also be understood as a counter-reaction to the previous communist system. Notably, Malinar's analysis also zooms in on some elementary causal mechanisms (Chap. 8). Like Malinar, *Druga* focuses on healthcare reform processes in the early 1990s, analysing the Albanian case. In her analysis, she focuses especially on one complex causal mechanism that characterises the interaction between Albanian policymakers and the World Bank as a transnational actor, while also zooming in on some elementary causal mechanisms (Chap. 9). *Safuta*, *Noack*, *Gottschall*, and *Rothgang* analyse the processes that followed the introduction of long-term care insurance in Germany in 1995–1996. Focusing on the crucial role that migrant workers play here, the authors trace the complex causal mechanisms that contributed to a specifc form of migrantisation within the feld (Chap. 10).

In the ffth part, social policies in Latin American countries are in the foreground. In their historical analysis, *González de Reufels* and *Huhle* show that the four Latin American Medical Congresses (1901–1909), as transnational events for the medical profession, played an important role for establishing public health policies in Chile and Uruguay, and highlight the complex causal mechanisms that were at play in this process (Chap. 11). Health is also the focus of *Sirén*'s analysis. His chapter traces the complex causal mechanisms within the political process on universal health coverage in Bolivia, which unfolded against the background of a highly fragmented and exclusionary healthcare system (Chap. 12). In the fnal chapter of this part, *Barrientos* adopts a mechanism-based approach to study whether conditional income transfers in Latin American countries lead to political responses by the recipients of these transfers and, subsequently, better political inclusion (Chap. 13).

In the concluding chapter (Chap. 14), *Kuhlmann* and *Nullmeier* draw together the mechanisms that were traced in the individual chapters, and present a structured compilation of the complex causal mechanisms that have been identifed throughout the volume. Moreover, they outline future research avenues for mechanism-based approaches in social policy research.

#### **References**


———. 1989. *Nuts and Bolts for the Social Sciences*. Cambridge, New York: Cambridge University Press.


Transnational Ideas and Domestic Politics. *International Journal of Social Welfare* 25 (4): 388–398.


———. 2017. Handeln und Struktur, Akteur und System: Die kausale Rekonstruktion von sozialen Makrophänomenen am Beispiel der Finanzkrise [Action and Structure, Actor and System: Te Causal Reconstruction of Social Macro-Phenomena Using the Example of the Financial Crisis]. *Zeitschrift für theoretische Soziologie* 6 (1): 5–26.

———. 2019. Überraschende historische Ereignisse als theoretisches Problem [Surprising Historical Events as Teoretical Problem]. In *Ordnung und Fragilität des Sozialen. Renate Mayntz im Gespräch*, ed. Ariane Leendertz and Uwe Schimank, 297–315. Frankfurt am Main: Campus.

———. 2020. Causal Mechanism and Explanation in Social Science. *MPIfG Discussion Paper* 20/7. Cologne: Max Planck Institute for the Study of Societies.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **Part II**

**Causal Mechanisms and Social Policies in Asian Countries**

## **2**

## **The Introduction of Pension, Accident, and Health Insurance in Urban China**

**Tobias ten Brink, Armin Müller, and Tao Liu**

## **1 Introduction1**

Tis chapter aims to explain the introduction of three types of contribution-based social insurance in the People's Republic of China (PRC). We focus on the explanation of three signifcant policy events, in particular, the introduction of the 1997 Urban Employees' Basic Pension

T. ten Brink (\*) • A. Müller

Jacobs University Bremen, Bremen, Germany e-mail: t.tenbrink@jacobs-university.de; arm.mueller@jacobs-university.de

T. Liu Zhejiang University, Hangzhou, China e-mail: taoliu1975@zju.edu.cn

<sup>1</sup>A previous version of this chapter was published as a working paper by the SOCIUM Research Centre on Inequality and Social Policy and the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", University of Bremen (ten Brink et al. 2020).

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

Insurance, the 1998 Urban Employees' Basic Medical Insurance, and the 2004 Work-Related Injury Insurance. Tese three events each represented an important milestone in a longer reform process, resulting in a redesign of the respective insurance models. Furthermore, as Fig. 2.1 illustrates, the three events were all followed by substantial increases in coverage.

To explain the three policy events, we build on recent advances in mechanism-based explanations in political science and sociology, detailed in the Introduction (Chap. 1) to this edited volume (also see Goertz and Mahoney 2012; Hedström and Ylikoski 2010; Kuhlmann et al. 2020). To unpack the causal mechanisms that facilitated the introduction of the three social insurance schemes under scrutiny, we reconstruct the reform dynamics in each national policy subsystem. We start with the *initial conditions*, that is those structural trends in the economy or the fscal system, reform pressures arising from demographic change, and other threat perceptions that triggered the reform process in the frst place. We then trace relevant actor patterns, denoted in this chapter as *activities*; key sector-specifc domestic *events* such as compromises between competing

**Fig. 2.1** Participation in urban employees' insurance schemes 1989–2008. (Sources: MoLSS 2001, MoHRSS 2018; CHFP 2018; MoH 2005)

actor groups or camps; and *events of interdependence*, indicating signifcant international infuences on the domestic policy process. Particularly the latter entail the existence (or not) of complementarities between national and international factors, such as established international insurance models serving as domestic frames of reference. Te causal mechanisms we analyse mostly operate at the meso-level of organisational actors. We especially focus on ministries, the State Council, as well as the key political leaders within these organisations, and on nationalinternational interaction rather than local issues.2

In a nutshell, our policy analysis fnds *three key mechanisms* which can plausibly help to explain the three respective policy events and their commonalities and diferences. First, a *policy experimentation mechanism* fostered learning from domestic and international experiences, by testing diferent variants of insurance schemes at the local level. Tere are two forms of this mechanism in our case: On the one hand, in a context of political confict, the policy experimentation mechanism is more *strategic*, used to either promote diferent policy options, as in the example of pension insurance, or to delay policy implementation, as in the example of health insurance (see Heilmann 2008). In the absence of confict, on the other hand, it is a more *neutral* means of gaining information and learning (work accident insurance).

Second, we fnd an *elite cooperation mechanism*. Again, we fnd two versions: One that originated in a long-term and broad ideational *consensus* among the governing party-state elite (work accidents) and another that was based on an enforced *compromise* between competing groups within the party-state (pensions).

Te third mechanism we fnd comes into play only in the health feld: Te *top-leader intervention mechanism*. Tis mechanism involves the attempt by an individual actor within the administration to promote a new policy direction. Moreover, we observe that the presence or absence

<sup>2</sup>Data was collected as part of our ongoing research, in which we enquire into how internal and external infuences shaped Chinese social policy in the reform era. To examine key mechanisms that have afected the introduction of urban social insurance systems, in particular the evolution of ideas and the choice of models, we conducted a comprehensive review of policy documents and other primary and secondary literature. Insights from the authors' previous research were also utilised. ten Brink was responsible for the pension case, Liu for the work accident case, and Müller for the health insurance case.

of complementarity between the international environment and the domestic actor constellation had a decisive efect on how the policy experimentation mechanism and the elite cooperation mechanism in particular played out.

In the next section, we frst provide a short historical background and inform about the existing academic literature that our analysis builds on. In the main part, we then reconstruct the key causal mechanisms in each of the cases. In the conclusion, we summarise the fndings, present our contribution to the literature, and identify the limitations of the study.

### **2 Historical Background and Research on China's Social Security System**

Te People's Republic of China (PRC) fundamentally restructured its social security system during the transformation from a command economy into a state-permeated type of capitalism (ten Brink 2019). Before the 1980s, state-owned enterprises (SOEs) were responsible for protecting workers from social risks such as old age, accidents, and illness in urban China (danwei system; see Dillon 2015). But although reform pressure was already high in the 1980s, and many reform proposals were made in the early phases of this transformation, it was only in the 1990s that serious steps were undertaken to adopt a comprehensive social insurance system in urban China.

In the 1990s, in the midst of what China scholars call the second phase of reform (see Naughton 2007), the objective of the party-state leadership was to transform the remnants of the old eroding social security system (especially the *danwei* system in the cities) into a contributionbased social insurance system partially based on Western and East Asian models. From the 2000s onwards, the aim increasingly became the implementation of universal insurance schemes, and new policies were adopted to cover the rural population and informal workers (see Ngok and Chan 2016; Zhou 2017; Zheng 2008).

Tus, our three case studies in fact only cover a small, yet important segment of a much larger, and much more complex, process—in which social insurance institutions have been established as parts of the broader structure of a new welfare state. Tis process of building a welfare state in China is ongoing, albeit at diferent speeds and with large variation when it comes to system design, coverage, and generosity, for instance. Indeed, in this chapter, we merely examine social insurance institutions for *urban* Chinese workers with a formal employment contract. Te majority of the population in the 1990s, which either lived in the countryside or as migrant workers in the cities, had only very limited state-run social protection. Social security reform for the rural population and informal workers was initially organised into diferent systems, and mostly came at a later point. It is important to note that the urban-rural divide so typical for reform China was already established in the 1960s with a strict system of household registration (*hukou*) that ascribed people either a rural or urban residency status. Tis restrictive household registration system perpetuated the disparities between urban and rural incomes, standards of living, and social safety nets.

With the establishment of a contribution-based social insurance system with Chinese characteristics, an academic literature emerged which has placed a spotlight on many of its key features and development trajectories. Tis holds for socioeconomic studies, which stress the efects of market-oriented reforms and related waves of redundancies, and for a social security vacuum which made it necessary to introduce employerindependent contributory pension and health insurance (Liu 2002; West 1999; Whiteford 2003). Additionally, rapid demographic change and a declining signifcance of the family facilitated the expansion of social policy (Fong 2004; Ngok and Chan 2016). Besides these arguments, political-institutional approaches also focus on the capacity for action among party-state actors, on interest groups, and on other political factors (Duckett 2001; Frazier 2010; Müller 2016). Te success or failure of experimentation under hierarchy, that is of local pilot projects, also infuenced the reform processes, and previous research has already highlighted the neutral and strategic facets of policy experimentation (Heilmann 2008; Shi 2012). Further studies show a causal relation between social protests and defcits in the legitimacy of the Chinese authorities, and social policy reform (Solinger 1999).

Notably, the existing literature on social insurance reform in China largely refers to national infuences as explanatory factors. To capture both national and transnational infuences, we tie in with an emerging literature that investigates international infuences on the formation of social insurance schemes in China (see Hu 2012; Dillon 2015). For instance, Duckett (2019) reveals the role played by international events and organisations on China's New Rural Cooperative Medical Schemes, and Leisering et al. (2017) demonstrate how Chinese actors have synthesised disparate ideas from the United States and Western Europe with Chinese traditions to create a distinctive model of social assistance (*dibao*), thereby complementing the literature on policy difusion. Moreover, we also build on mechanism-based explanations in the research on Chinese social policy, which has been gaining ground recently (see Müller 2016).

## **3 Causal Mechanisms in the Introduction of Urban Basic Pension Insurance**

In 1997, the Chinese government introduced a new urban employee basic pension insurance framework under State Council Document No. 26, with the goal of adopting a unifed basic pension system for all urban employees. Te 1997 decision united a system that had previously been highly fragmented. While a basic model of mixing social pooling and individual accounts was already decided by the State Council in 1991, resulting in some coverage increase, it took six years to overcome a heavily localised system with numerous mixes of social pooling and individual accounts. Te 1997 unifcation also triggered a relatively successful rise in coverage.

Te causal chain leading to the adoption of a mixed model of social pooling and individual accounts was dominated by two mechanisms. First, since the early 1990s, strategic policy experimentation to promote diferent model mixes stimulated learning. However, persistent confict between the two prevailing camps delayed the process of national policy unifcation. Second, strong complementarities between the domestic pressure to reform SOEs, a centrepiece of economic reform from the mid-1990s onwards, and international infuences especially by the then dominant World Bank pension reform model caused a political compromise on basic pensions. An elite cooperation mechanism emerged that fostered policy unifcation. Moreover, strong national-international complementarities fuelled further reform eforts thereafter (see Fig. 2.2).3

**Fig. 2.2** Causal chain pension insurance. (Source: Own presentation)

<sup>3</sup>Note that despite the 1997 basic pension unifcation success, important dysfunctionalities such as limited portability and compliance, pension debt, empty individual accounts, or corruptive practices of local ofcials were not eliminated (Frazier 2010; Li 2014). Te 1997 policy event also failed to stipulate penalties for non-compliance, which was linked to limited administrative capacity (Béland and Yu 2004). While pension policy was unifed, pension administration remained fragmented.

#### **3.1 Initial Conditions and Threat Perception**

Chinese economic reform in the 1980s put the necessity to change pension policy onto the agenda. China had established its urban pension scheme in 1951 as part of its labour insurance system, following the Soviet model with insurance fnanced solely by enterprises, and therefore largely for SOE employees. Obviously, the old system was no longer functional. In the 1980s, a large proportion of the urban population was exposed to the risks of having no old-age coverage at all, especially in the emerging non-state enterprise sector. "In addition, enterprises supporting a large number of retirees were disadvantaged in market competition, with some fnding themselves on the verge of bankruptcy" (Li 2014, 283). Te old enterprise-based provision of pensions served as an obstacle to both SOE competitiveness and labour mobility (West 1999).

Te feld of pensions constituted a problem for social stability from early on in the reform period. Around 1980, there were already two million Chinese workers who had reached pension age but were not allowed to retire by their employers (Chen and Wang 2010). Te containment of rising pension costs mostly for SOEs also was high on the agenda at an early stage. Additional reform pressure had arisen from rapid demographic change (West 1999; Whiteford 2003).

In the 1980s, however, diferent and competing agendas in the responsible government authorities made serious reform difcult. Tere was an agreement on the underlying rationale to maintain social order and political stability and, as a developing country, to keep the beneft level as low as possible to avoid fnancial burdens on the party-state. Regarding the design of new schemes, however, there was substantial disagreement. Consequently, policymakers and policy intellectuals learned from competing international models that coexisted at the time, with no one having a clearly dominant position. On the one hand, the Ministry of Labour (MoL) was eager to learn from Western (German) models of social insurance, and more particularly from a pay-as-you-go system, in which the current cohort of contributors pays for pensioners through contributing to a pooled social insurance fund. Te MoL was closely connected to the International Labour Organization (ILO) since the 1980s, and an ILO ofce in Beijing was opened in 1985.

On the other hand, the National Economic System Reform Commission (ESRC, a ministry-level authority, and precursor of the National Development and Reform Commission) was attracted to the global epistemic community of (neo)liberal economists and eager to learn from the World Bank. Te World Bank was promoting an individual accounts-based pension system as a means to facilitate SOE reform. Intellectual and technical support was provided, for instance the formulation of comprehensive actuarial projections. Tis promotion of an alternative to a mostly state-run social insurance pension was complemented by the attraction of the Singapore model, in essence compulsory savings accounts for the purpose of a pension. Resembling developments in health (see Sect. 5), the inclusion of the Singapore model on the policy agenda also made it possible to interpret this as a creative mixing of Western and non-Western models and to thereby appease domestic critics of Western models.

From the mid-1980s onwards, the MoL promoted social pooling of pension funds to replace the work unit as the focal point of the social basic pension pillar and to even out dissimilar burdens of enterprises. Experimentation with local social pooling began in a few selected provinces. Te ESRC supported experiments as early as July 1989, in Fujian province, which "set up individual accounts on top of social insurance for workers" in non-SOEs (Hu 2012, 620). Interestingly, already in 1991, a basic model of mixing social pooling and individual accounts was promoted by the State Council (Guofa 1991, 33). From then on, the battle was about the "right mix" between social pooling and individual accounts. In the following, we identify two causal mechanisms that have signifcantly impacted the process.

#### **3.2 Mechanism 1: Strategic Policy Experimentation**

Since a number of important players were involved in the competition of ideas and infuence in the pension feld, and functional authority was overlapping, the debate over suitable reform measures was intense, especially between the MoL and the ESRC. From 1989 onwards, each institution supported numerous local experiments. Policy experimentation here had the following aspects: Te local pilots typically were useful for gaining information ("neutral" experimentation) and learning, for example, about diferent combinations of social pooling and individual accounts. However, most of those experiments also had a strong strategic aspect. Tey were a means for promoting diferent policy options in a situation in which both the MoL and the ESRC lacked the political support to produce a collective, binding decision at the national level.

In 1993, the New Socialist Market Economy initiative confrmed the mix of pooling and individual accounts the State Council had already promoted in 1991 (Zhonggong Zhongyang 1993). All local experiments now had to be variants of this model. Te MoL's "pro-pooling" experiments, based on ideas from the ILO, favoured a small individual account (equal to 3–5% of payroll) on top of a larger social pooling component, a model that was later coined the "Guangdong" or "Hainan model". Conversely, the ESRC's "pro-individual account" approach led to strategic experimentation with larger individual account components (equal to 16–17% of payroll), later coined the "Shanghai model". Te ESRC branded this approach as a middle way between planning and free market, and also stressed Southeast Asian infuences, by emphasising and transferring the Singaporean experience (Gong 2003; Hu 2012). It also had the support of national leader Zhu Rongji, whose promotional eforts however were resisted by the MoL, which was "probably supported by Vice-Premier Zhou Jiahua" (Gong 2003, 247), and thus had a diferent efect than in the health system where top-leader intervention strongly impacted on the policy process (see Sect. 5).

In 1995, the insights from strategic experimentation were incorporated into State Council Circular No. 6, which required provinces to select one of the two models mentioned above to establish local basic pension schemes. Te result was system fragmentation. Both models involved individual accounts and social pooling; however, their proportion was diferent (larger vs. smaller individual accounts) and system components could be combined and implemented in diverse ways. While in almost all provinces individual accounts now were adopted, it led to the creation of hundreds of schemes all over China, signifcantly handicapping labour mobility, for example, from SOEs to the non-state sector (Béland and Yu 2004). Te Circular also failed to achieve an agreement about the transfer of employer contributions and "delegated local authorities to decide the detailed transfer plan" (Gong 2003, 87). In practice, both beneft and contribution rates varied signifcantly.

In summary, strategic experimentation led to policy learning and important changes in the urban pension system. However, it also fostered fragmentation due to unresolved conficts regarding the right pension mix, resulting in coverage stagnation. Another causal mechanism, the elite cooperation mechanism based on a productive complementarity between domestic and international factors, however, fostered another change of direction shortly afterwards.

#### **3.3 Mechanism 2: Elite Cooperation Through Enforced Compromise**

If not to the same degree as in health (with implementation delayed by strategic experimentation and substantial implementation gaps at the local level), experimentation with pension schemes still contained enough confict and opposition between key stakeholders to hinder a reform breakthrough. Te second mechanism we identify here however caused a real step forward. Under pressure to reform the SOEs and given a dominant international model for pension reform with enough leeway for compromise upon its transferral to China, elite cooperation and consensus fnding developed.

With economic reform accelerating again after 1992, the reform of SOEs soon became a key objective. Increased competition, not only created by the rise of the non-state sector but more and more also by international investors, resulted in a drop in proft margins, particularly for local SOEs, and rising debt levels (Naughton 2007, 304–05). At this point, the government's economic policy strategists developed a plan to address the situation that went down in the annals of history as "grasping the large, releasing the small" (*zhuada fangxiao*). From 1995 to 2000, three-quarters of all local SOEs had been subject to restructuring measures ranging from rationalisation to formal privatisation. A dramatic reduction of state-sector employment (as well as in urban collectives) resulted, which led to a serious "confict over how to distribute the social costs of SOE reforms" (Frazier 2010, 19).

SOE reforms thus heavily afected, and indeed accelerated, urban pension reform. According to leading policymakers, the unifcation of pensions across enterprises of various ownership forms and the active participation of the non-state sector with a younger demographic structure could reduce the overall pension costs for SOEs (Gong 2003). Additionally, benefts for pensioners would stimulate domestic consumption more than that of dismissed older workers with lower propensities to consume. Most importantly, unifed pension schemes for diferent enterprise forms, industries, and localities promised higher labour mobility, "particularly from the state sector with privileges to the non-state sector" (Gong 2003, 202).

Tis strong pressure to move ahead with pension reform was complemented by a change at the international level. Here, the former coexistence of competing pension models was replaced by the advent of one dominant model, propagated inter alia by the World Bank. Its "threepillar" model now served as the dominant international reference frame. In particular, in addition to the 1994 publication of the book *Averting the Old Age Crisis* (World Bank 1994), a comprehensive report on *China's Pension System Reform* (World Bank 1996a) was published after a mission to China. By criticising the fragmentation of the 1995 system, it became a key reference thereafter and was widely read at the highest levels of the party-state, including by President Jiang Zemin himself (Frazier 2010, 60).

Tis "event of interdependence" facilitated domestic compromise between the promoters of social pooling and the promoters of individual accounts (see Fig. 2.2).

Upon the World Bank's criticism on the fragmented pension system, … the State Council required the ESRC and the Labor Ministry to investigate provincial pension systems to fnd a way to unify them. Te two ministries organized a research group … According to the research group, most of the localities agreed to unify the pension system. (Hu 2012, 630)

#### **2 The Introduction of Pension, Accident, and Health Insurance…**

A new unifed model was eventually agreed upon. Te State Council Document No. 26 of 1997 on the *Establishment of a Unifed State Pension Programme for Enterprise Employees* required all enterprises to accept the same contribution and entitlement rules. As a compromise, a mediumsized individual account was established.4 Individual accounts, equal to 11% of payroll, would be funded by both enterprises (3%) and employees (8%). Social pooling was funded by enterprises, which could not contribute more than 20% of the total wage as pension funding. It was also promised that the fnancial burden of pension commitments would be more evenly spread across diferent enterprise forms (Béland and Yu 2004; Li 2014). While the compromise entailed a strong role for individual accounts, all sides could eventually live with it (on the interpretative leeway in World Bank proposals that left enough room for compromise between the opposing domestic camps, see: World Bank 1996a, 19–31).

Accordingly, the party-state elite could be unifed. Rather than the previous inter-ministry confict, elite cooperation at the national level and between central and local authorities now became efective. Notably, this constellation was prolonged. Te unifed pension policy paved the way for geographic unifcation, and until 2000, "state enterprises, collective enterprises, foreign-invested enterprises and private enterprises were all part of the system, which was further fne-tuned later" (Li 2014, 286). In principle, every urban employee with stable, contract-based employment was now covered by some sort of pension scheme (the self-employed were only able to contribute after 2003, on a voluntary basis). And indeed, it became easier for many workers to move between localities, as pension contributions were now managed at the provincial level.

What made for this prolongation and deepening of reform? In a nutshell, the elite cooperation mechanism was reinforced through another *event of interdependence*, namely the 1997/1998 Asian Financial Crisis (which caused a decrease in demand), and an upsurge in pension protests. Te protests were based on the still poor implementation of pension policy leading to delayed or reduced pensions for instance, and one

<sup>4</sup> Interestingly, the compromise on a medium-sized individual account matched with the practice of some provincial governments that had already introduced medium-sized individual accounts of 10–12% after 1995 (Gao 2006). Tis also eased consensus fnding at the national level. We thank Tian Tong for sharing this detail from her PhD research.

incident was particularly relevant as it shocked the highest echelons of the party-state: On a tour to North East China early in 1998 to visit ongoing SOE reform eforts, a special train carrying then Premier Zhu Rongji and several ministers "was blocked by local pensioners. Some of them lay down on the rail track to plead for a regular payment of pensions. It was a real shock" (Gong 2003, 193).

In conclusion, after the mid-1990s, a "golden opportunity" (Gong 2003, 190) for a basic pension reform compromise emerged. A favourable domestic constellation matched with an international policy environment that provided an elastic reference frame for erstwhile conficting camps.

## **4 Causal Mechanisms in the Introduction of Work Accident Insurance**

Tis section investigates the trajectory of work accident insurance in China and the introduction of the *Regulations on Work Accident Insurance* by the State Council in 2004. While some provisional regulations in this feld had been in place since 1996 such as *Trial Measures of Work Accident Insurance for Employees*, and coverage increased slowly from 1993 onwards, the 2004 Regulation represented a milestone with substantial impact: Trough its inclusion of non-SOE employees, and migrant workers (at least theoretically), coverage in this scheme rose from 45 million in 2003 and 85 million in 2005 to 239 million workers in 2018 (MoHRSS 2019).5

Te causal chain leading to the adoption of the regulations on work accident insurance in 2004 was dominated by two mechanisms: First, beginning in the late 1980s, the neutral policy experimentation mechanism stimulated learning on introducing insurance schemes based on

<sup>5</sup>Costs were borne in full by enterprises, varying between 0.5 and 2% of total wages. Again, the 2004 policy event should not lead one to underestimate remaining problems in the following years such as the huge number of workers not covered under ofcial registered employment contracts (and thus, a mass of workers not having a de facto chance to be insured), employers that did not transfer money into insurance funds, and other local implementation gaps (Cheng and Darimont 2005).

social pooling. Second, the elite cooperation mechanism brought about by a broad consensus among scholars and ofcials on the design of a work accident insurance scheme facilitated reform since the late 1990s. Strong complementarities between Chinese and international epistemic communities helped to further reinforce elite cooperation and consensus building. A dominant international model of public social insurance as opposed to commercial insurance schemes was especially relevant in this feld.

Compared to health and pensions, however, the introduction of work accident insurance came several years later. Why? We argue that reform eforts were halted in the mid-1990s because reform dynamics in this feld were not perceived as crucial to social stability. Since the end of the 1990s, fnally, the re-employment of the two mentioned mechanisms mobilised enough drive to nationally adopt the insurance scheme—elite cooperation was further encouraged by wider public debates on work injuries, and new experimentation especially fostered learning of technical details, with Germany's system as a reference frame and the capability to locally adapt these (see Fig. 2.3).

#### **4.1 Initial Conditions and Threat Perception**

Since the adoption of the economic reform, labour insurance provided by SOEs and collective enterprises had eroded in China (Hu 2009, 322–77). Protective measures against work-related injuries, accidents, and occupational diseases were fundamentally weakened as well. Although in the 1980s some SOEs still provided some compensation once work accidents had occurred, however, the beneft level for compensation was equivalent to the 1950s' level and could not cover the real needs of households with family members sufering from accidents. Non-state enterprises regularly ofered no protection at all. If a severe work injury occurred, many enterprises were unable to bear the huge costs for compensating injured or dead workers. Some employers even fed with their money and escaped their obligations to compensate or had to declare bankruptcy after the occurrence of deadly work accidents (Hu 2009).

#### **2 The Introduction of Pension, Accident, and Health Insurance…**

Lack of protection frequently resulted in serious disputes throughout the 1980s: "Disturbance related to work accidents" (*nao gongshang*) was a euphemism for widows and relatives publicly displaying the corpses of dead workers and blocking companies' production through verbal quarrels and conficts with employers, aiming at receiving higher compensation. Enterprises in some high-risk sectors, such as mining, were especially at risk of falling into crisis when deadly accidents occurred (Hu 2009). Taken together, these developments led to a new demand for social protection against work accidents.

#### **4.2 Mechanism 1: Neutral Policy Experimentation**

Te search for a work accident insurance programme began in the late 1980s, at around the same time as Chinese ofcials and experts started to explore a health insurance and a pension insurance system. Te MoL began to research the matter in 1988, and frst experiments were conducted. Te process was complemented by regulations on occupational diseases by the Ministry of Health (MoH), as well as disease-specifc prevention programmes and sector-specifc safety regulations. In 1990, the decision to create an insurance system was integrated into the eighth fveyear plan. Moreover, government ofcials from the MoL searched for a social protection system beyond individual enterprises and preferred a social pooling scheme that was able to collect and pool resources to compensate for injured workers. Notably, from the beginning, transnational interdependence infuenced the process. Western social insurance models that pool work-related risks of individual enterprises were especially used as a reference model for establishing a work accident insurance in China (Zheng 2008, 269).

At the end of the 1980s and into the 1990s, various counties and cities in the provinces of Liaoning, Fujian, Hainan, and Guangdong piloted work accident insurance schemes based on social pooling. One of the most successful and thus famous pilot projects at that time was located in Tieling (Liaoning), which already achieved a social pooling scheme at city level in 1991 (Hu 2009, 334). Since the 1990s, similar pilot projects have been further promoted nationwide, mainly designed by the MoL, its local counterparts, and local governments.

In comparison, particularly to piloting in health insurance, experimentation here was mainly to gain information about the operational experiences of a pooling-based accident insurance model. Due to the low confict intensity between ministerial actors in this policy subsystem, local pilots could be mostly utilised to collect practical implementation experience. Notably, strategic experimentation with substantially diferent models like in pension or health insurance was the exception to the rule. Only a few actors promoted commercial insurance. When in the early 1990s some ofcials suggested that the management of work accident insurance could be transferred to commercial insurance companies and some insurance companies provided pilot programmes for employers' liability insurance, the MoL stepped in and corrected this approach with a frm attitude. Backed by strong support among experts, a consensus emerged that work accident protection should be based on social pooling, not commercial frms (Liu and Leisering 2017, 118). Troughout the 1990s, the debate on a private work accident insurance gradually became weaker in mainstream discourse.

#### **4.3 Mechanism 2: Elite Cooperation Through Consensus**

A second aspect that contributed to the policy event of 2004 was elite cooperation, based on a long-term and broad policy consensus. As mentioned above, the initial consensus in the early 1990s set the reform direction towards a pooling-based insurance. Another complementary consensus emerged at the turn of the millennium. A group of key fgures from the new Ministry of Labour and Social Security (MoLSS, founded in 1998), the think tanks of the MoLSS, and some elite universities like Renmin University constituted major "pushing hands" of this consensus building. Tis group co-constituted a "modern" and "rational" work accident insurance system in the early 2000s, which was to reshape the confguration of traditional work accident insurance and move the emphasis of this system from merely monetary compensation to prevention measures and comprehensive recovery programmes after the occurrence of work accidents. Chinese elites began to orient themselves strongly towards the German model of statutory work accident insurance with its emphasis on prevention, compensation, and rehabilitation (Liu and Leisering 2017).

A dominant international model for the feasibility of a public social insurance scheme served as a key reference frame for long-term consensus. Te ILO with its partner organisation, the International Social Security Association (ISSA), played a key role here. Both organisations were also strongly infuenced by the German model of work accident insurance and both have promoted its worldwide difusion. Scholars and experts from European social insurance states like Germany, France, and Switzerland constituted the strongest "expert community" in these organisations (Liu 2018). Te World Bank did not promote an alternative model in this feld.

#### **4.4 Reform Delay: From the Late 1990s to the 2004 Regulations**

Although the area of work accident insurance is an arena with low intensity of contestation in comparison to health for instance (see Sect. 5), it was adopted comparatively late. In a nutshell, reform eforts were halted in the mid-1990s because reform dynamics in this feld were not perceived as crucial to social stability. In contrast to general social risks like old age and illness, the risks of work accidents are rather contingent and specifc, that is, occupation related. Especially workers in high-risk sectors such as mining are exposed to work accidents. Indeed, in the mid-1990s, work accidents were not perceived as an acute threat for society as a whole or as a challenge to social and political stability. For these reasons, it was not put at the top of policymakers' agenda by the mid-1990s, when health and pension reform absorbed many political capacities. Te elite cooperation mechanism thus lost momentum for some time.

However, a threat perception was back on the agenda in 2002/2003, this time with enough pressure for the elite cooperation mechanism to play a role. An increase in work accidents, especially in mining, attracted a great deal of attention—with increased coverage by both national and international media on the sufering of miners. A "naming and shaming" efect, with increased communication among Chinese ofcials and scholars on these issues, fnally accelerated the adoption of the regulations on work accidents insurance in 2004.

## **5 Causal Mechanisms in the Introduction of Urban Basic Health Insurance**

In 1998, the State Council issued a decision on the Urban Employees' Basic Medical Insurance. It resembled the decision on pension insurance in 1997. Most notably, both decisions made a mixed model of social pooling and individual savings accounts the national norm. But overall, the coverage of health insurance remained low far into the 2000s for two reasons: First, health insurance appeared on the agenda later than pension insurance and rose in priority only in the context of crisis episodes. Second, due to bureaucratic confict, the expansion proceeded overall more slowly in the 1990s.

Te causal chain leading to the adoption of a mixed model of social pooling and Singaporean-style Medical Savings Accounts (MSAs) was dominated by two mechanisms. First, the introduction of MSAs came onto the policy agenda through a mechanism of top-leader intervention. Second, various ministries6 attempted to get it of the agenda again by delaying the process through strategic experimentation with alternative models. Moreover, there was an absence of complementarity with the international environment in healthcare due to the lack of a neat, simple, and practical model induced by the complexity of the policy subsystem (see also: Weyland 2007) and the minimalist universalism the World Bank promoted at the time, which contributed to the overall lower coverage of health insurance.

<sup>6</sup>Most notably, the MoH reportedly primarily opposed reforms curtailing the revenues of public hospitals, and the Ministry of Labour held an ambiguous stance on experimenting with various models (Duckett 2011).

#### **5.1 Initial Conditions and Threat Perception**

Health reforms for cost control appeared on the policy agenda very early in the reform process. Te introduction of user fees actually dates back to the Cultural Revolution period, when the central government ordered several rounds of price decreases in healthcare to make health services more afordable. During this period, hospitals were allowed to run their own pharmacies and charge user fees for the drugs sold, part of which they could keep as a proft to fund their operations. Starting in the late 1970s, the Ministry of Health moved towards allowing individual practitioners in basic medical services (Dong et al. 1999; Li 2008).

In risk protection, the Ministry of Health (MoH) oversaw a gradual and stepwise reform approach in the two main protection systems for urban residents: Labour Insurance for workers (under the jurisdiction of the MoL), and Government Employee Insurance for civil servants and employees in government and administrative organs (under its own jurisdiction). Te latter provided very generous benefts funded from government budgets, and the former provided less generous benefts for workers in SOEs and some collective enterprises. For the most part, these initiatives were dominated by cost control measures and the introduction of co-payments. In 1988, the State Council asked a group of eight ministries led by the MoH to experiment with reforms of the health insurance systems. Two lines developed under this initiative: First, special health insurance systems for pensioners; and second, pooled insurance for serious illnesses of employees, which was tested in four pilot localities in 1988.7 Te efectiveness of the latter line of reform on cost containment was disputed (Duckett 2001, 2011; Zheng 2008).

Te Tian'anmen protest and its suppression in 1989 became a gamechanger for health reforms. Large protests in urban China had substantial national and international repercussions. Within the PRC, the leadership of the Communist Party was suddenly sensitised for the destabilising potential of urban unrest and became more careful with economic and labour market reforms. Research on workers' opinions in the aftermath

<sup>7</sup>Among the predecessors of these initiatives were local insurance pools organised by companies outside the state sector in Beijing (Aitchison 1997, 139–42).

of 1989 revealed that healthcare was a major concern for them. As Duckett (2011, 137–38) reports, this substantially raised central government's threat perception with regard to the feld of healthcare. Under these circumstances, the previous line of health reforms under the MoH became less acceptable. In particular, the MoH's reforms were seen as inefective in containing escalating health costs and too protective of hospitals and the medical profession in profting from rising costs, at the expense of workers' social security. Furthermore, it was clear that if SOE reforms were to be broadly implemented, the protection of workers' health risks would need to be taken care of.

Te year 1989 also saw a signifcant event of interdependence between national and international factors, which changed the course of development. Te international repercussions of the Tian'anmen protest were far-reaching, including a widespread diplomatic isolation and a temporary stop of ongoing development projects. Beijing began to intensify its relations with developing countries as a response, and it began emphasising non-Western models of development. Singapore was one of the clearly identifable success cases of development in Asia that had preserved an authoritarian political system. Its welfare practices were dominated by the provident fund model of mandatory individual savings accounts. Te case of healthcare here is particularly insightful, because Singapore's MSAs were a singularity, whereas capital-based schemes were more common in a global context.8 So after 1989, Party leaders' threat perception in healthcare rose, while at the same time the Singapore model became a more attractive and politically acceptable policy option (Hu 2012).

#### **5.2 Mechanism 1: Top-Leader Intervention**

Fundamental policy change was soon underway. In 1991, liberal politician and previous mayor and party-secretary of Shanghai Zhu Rongji was appointed Vice Premier of the State Council. Te year before, he had gone to Singapore to investigate the provident fund9 and became its

<sup>8</sup>For the infuence of the 1989 events on the process of pension insurance reform, see: Hu (2012).

<sup>9</sup>We thank Yuxin Li for sharing this detail from her PhD research.

leading advocate thereafter. Te State Council decided that pension insurance would henceforth combine a social pooling component with individual savings accounts. In 1992, Deng Xiaoping gave a strong signal for returning to the course of economic reforms after the short interregnum that followed 1989, which meant that SOE reforms were back on the agenda. In the same year, the authority over health reforms at the State Council was reshufed: Te Ministry of Health lost its leadership role and was replaced by the ESRC for a brief period before the MoL took the main initiative (Zheng 2002, 135).

Zhu Rongji directly intervened in the health reforms, pushing the Singapore model. Te ESRC was a forceful institution in favour of the Singapore model in pension reforms, but in healthcare, Aitchison (1997) reports that all ministries in the leading small group on health reforms disliked Zhu's reform plan and subsequently negotiated him down to prolonged policy experimentation. At the same time, as we will see below, the inter-ministerial competition between the MoH and the MoL arguably weakened their bargaining position. Pilot projects joining MSAs and social pooling subsequently developed under the jurisdiction of both ministries, while at the same time, the implementation of MSAs in other insurance programmes and localities was often neglected.10

#### **5.3 Mechanism 2: Strategic Experimentation to Delay**

Tere was a signifcant diference between the pension and health policy felds: In pensions, there were already a substantial number of countries applying capital-based models of old-age protection, and the World Bank also began to promote a mix of methods in the 1990s. In healthcare, however, capital-based social protection was an exception, and the international policy subsystem was chaotic in the sense that there was no consensus about an optimal approach to protection (Howlett and Ramesh 1998). While there was a general normative consensus on universal access and universal coverage, there are multiple ways to realise this with fundamentally diferent distributive consequences. Furthermore, the systems

<sup>10</sup>Overall, more research is needed to shed light on the details of this policy process.

of risk protection and service provision in healthcare are closely interdependent, which makes for a much larger number of possible models than in pensions, and there is no universally agreed-upon system of classifcation. Some scholars even generally doubt the utility of "models" in healthcare due to the high complexity of health systems (Bali and Ramesh 2017; see also: Weyland 2007; for an overview of existing classifcations and their ft for the Global South, see: de Carvalho et al. 2021). Te World Bank promoted a universalist, minimalist approach of "essential health services" funded through taxation and focusing on benefts for the poor (Kaasch 2015). But this option was fundamentally incompatible with the political aim of pacifying the privileged SOE workers precisely because of its redistributive, minimalist universalism. SOE workers would have perceived it as social decline to lose their privileged status in social protection after decades of preferential treatment. Furthermore, tax funding was a problematic option at the time of deepening fscal crisis. Te absence of a dominant international model in healthcare that could be reconciled with the divide-and-rule approach of the central government prohibited a productive complementarity between domestic bureaucratic confict and international ideas, as occurred in pension insurance.

As noted above, at ministerial level, the MoH and the MoL both disliked Zhu Rongji's proposal of MSAs. However, both were also competing with one another for jurisdiction over health insurance. Te MoH was in charge of health protection for civil servants as well as health insurance in rural areas, whereas the MoL was in charge of labour health insurance. Shenzhen was the frst city to test an insurance scheme joining social pooling and MSAs run by the local Bureau of Health in 1992. Te city had received the mandate to reform social insurance from the State Council in 1989. Preparations had already begun in 1991, and the administrative organ for health insurance had already been set up at the Bureau of Health. Te insurance combined social pooling and MSAs, as in the MoL's proposal, but the local management was by the health administration (World Bank 1996b, 60–61; Zhou 2017, 137–38). Also in 1992, the MoL issued two reform proposals, one about health insurance for urban employees (*zhigong yiliao baoxian*) including MSAs, and the other under the name of the former MoH initiative: Catastrophic

#### **2 The Introduction of Pension, Accident, and Health Insurance…**

disease pooling (*dabing tongchou*) (See also: Zheng 2002).11 A number of pilot cities were subsequently chosen for further experiments: Te MoL selected Pingdingshan in Henan Province and Chongqing, the MoH selected the city of Jiujiang in Jiangxi,12 and the city of Zhenjiang in Jiangsu Province volunteered as a pilot (Aitchison 1997, 89–91). Te experiments in Jiujiang and Zhenjiang gained a high profle, and the bureaucratic actors at national level reached a compromise on the mixed model of pooling and MSAs in the wake of their evaluation in 1996, ignoring explicit criticism by the World Bank. Tis led to the extension of pooling to 57 more cities, and the collective decision followed in 1998.

Te process of experimentation delayed a broader implementation of MSAs. Troughout the 1990s, various Chinese cities operated social pooling funds, largely without the implementation of MSAs. As Aitchison notes about the catastrophic health pooling scheme in Xicheng District in Beijing around 1995:

Te central government stipulated that any health care reform plan should involve individual accounts (…) Neither Xicheng district Bureau of Labor, nor Beijing city Bureau of Labor, seemed enthusiastic in implementing the individual account. In Xicheng district, the Bureau of Labor recommended that individual accounts should be set up by each enterprise separately, but did little to ensure that this was done. At the city level, the city Bureau of Labor planned to implement the individual account "in about two years" [from 1994]. (Aitchison 1997, 148–49)

Reports on similar schemes in other cities often make no mention of MSAs,13 and the classic Labour Insurance schemes continued operating without MSAs as well. Catastrophic health pooling remained listed as a separate insurance programme in ofcial statistics until 2000 (MoLSS 2001), indicating that the implementation of a unifed model proceeded slowly even after the 1998 decision. More generally speaking, the

<sup>11</sup>Te original documents could not be procured.

<sup>12</sup> Jiujiang was also the site of a World Bank health project, so the MoH had good connections in the city.

<sup>13</sup>Zheng points to substantial diferences in these pooling programmes and indicates that some also introduced MSAs in the 1990s (Zheng 2002).

**Causal process leading to mixed model in health insurance**

**Fig. 2.4** Causal chain health insurance. (Source: Own presentation)

implementation was difcult "because social insurance departments are unable to overcome vested interests and ensure the participation of enterprises and ofcials, audit enterprises, regulate health service providers and prevent fraud" (Duckett 2001, 302). As Fig. 2.1 illustrates, coverage picked up somewhat in 2000, and only increased more substantially since 2001—when the priority of the health system rose in general, and preparations for extending health insurance to rural China were initiated (Müller 2016) (Fig. 2.4).

## **6 Conclusion**

Tis chapter has identifed the following causal mechanisms to explain the introduction of three contribution-based social insurances in China: Te policy experimentation mechanism, the top-leader intervention mechanism, and the elite cooperation mechanism. Moreover, our fndings demonstrate that the presence or absence of complementarity between the international environment and the domestic actor constellation had a decisive efect on how the policy experimentation mechanism and the elite cooperation mechanism, in particular, played out.

In the urban pension insurance scheme, we observe a mechanism of strategic policy experimentation, resulting from experimental competition between "pro-pooling" and "pro-individual account" groups, which fostered learning in the 1990s. Furthermore, an elite cooperation mechanism, triggered by a forced compromise between the two competing groups who anticipated future conficts in this feld, facilitated the functional unifcation of a heavily fragmented system as well as coverage increase. A strong complementarity between the domestic pressure to reform SOEs and international infuences, especially from the then dominant World Bank pension model which had enough leeway to facilitate compromise on its adaptive transferral to China, helped to enforce cooperation.

In the work accident insurance scheme, policy experimentation was neutral, and not strategic. Rather than experimenting with competing models or contrasting confgurations of distinct models, pilot projects fostered the learning of technical details and the capability to adapt these to local circumstances. Additionally, elite cooperation through a broad consensus resulted in the comprehensive adoption of a functional insurance model and in coverage increase. A dominant international model on the feasibility of public social insurance served as a key frame of reference for long-term consensus in this policy subsystem.

In the urban health insurance scheme, top-leader intervention by Vice Premier Zhu Rongji aimed to achieve policy reform by using a hierarchical approach. Tis intervention, however, was opposed by the ministries involved in health reform. Implementation was delayed by a mechanism of strategic policy experimentation, and the ultimate compromise of mixing social pooling and MSAs may have succeeded only because Zhu exploited the rivalry of the MoL and the MoH regarding jurisdiction over health insurance. More research is needed to illuminate the details of this process. Troughout the 1990s, the classic labour insurance and various other urban health insurance funds continued to operate without MSAs, and coverage of the Urban Employees' Basic Medical Insurance only grew slowly even after the collective decision of 1998. All this was facilitated by a lack of national-international complementarity due to the absence of an international consensus on specifc health protection models.

Regarding the role of complementarity between the international environment and the domestic actor constellation, we fnd that the presence of a dominant international model facilitated domestic elite cooperation in pensions, whereas the absence of such a dominant model resulted in prolonged confict in healthcare. Similarly, due to the existence of an established international model promoted by the ILO, policy experimentation was more neutral in accident insurance and geared towards the generation of knowledge. Political and academic elites reached a consensus on a public social insurance scheme, which facilitated cooperation. In pension insurance, domestic promoters of social pooling and advocates for individual accounts both found points of reference in the international environment. In health insurance, the situation was similar. However, pensions and healthcare difered with regard to international complementarities: In the former, a dominant international model created productive complementarity which mediated bureaucratic confict. In this context, with strategic policy experimentation, it was possible to establish a dominant model which combined elements of the competing camps, leading to enforced cooperation between elites. In health insurance, conversely, top-leader intervention to establish a similar mixed model failed to produce enforced cooperation due to lack of complementarity with the international environment, where there was no dominant model that matched local demand. Implementation was delayed by a prolonged phase of experimentation, and coverage only grew slowly even after the collective decision.

Our chapter thereby contributes to the literature on China's social security reform, and to an emerging literature on mechanism-based explanations of transnational social policy dynamics (Leisering 2019; Obinger et al. 2018; Orenstein 2008; but also see Benz and Dose 2010, who distinguish between diferent governance mechanisms that resemble those described in this chapter). Notably, our fndings complement insights into mechanisms which explain the expansion of social protection programmes, such as the *crisis management by going further mechanism* (Chap. 3). In fact, the transition from the company protection model of Maoism to the social insurance system of the twenty-frst century can be interpreted along these lines. Te pressure for expansion here was strongest in pensions, as a growing number of citizens reached pension age and needed social support. As the average age of workers in SOEs difered depending on when they were founded, the introduction of pooling across public and, later, private enterprises helped consolidate social protection in old age. A similar argument can be applied to accident insurance, with the diference that the distribution of risk across companies depended more on the type of occupation (e.g. high-risk occupations such as mining). In healthcare, in contrast, risk was distributed in a comparatively random way, and only a small share of any company's workers were afected by serious illnesses each year. In pension and accident insurance, comparatively homogeneous groups such as elderly employees or miners had some potential for collective action. In healthcare, on the other hand, the individuals afected tended to be more isolated and less concentrated in social groups. Tis situation reduced the potential for immediate collective action.14

Regarding the limitations of this study, for reasons of accessibility and space, we were unable to reconstruct the causal chains in the three case studies with all their case-specifc aspects and idiosyncrasies. More research is needed, for example, on the role of the insurance industry. In future research, we would also be keen to include unemployment and maternity insurance, which constitute the two other areas of insurance that have infuenced social policy arrangements in China. Moreover, the diferent social nature of the three forms of risk we analyse, and its implications, should be more thoroughly addressed. Last but not least, it should be investigated whether the mechanisms identifed here can also be identifed in other countries, especially in emerging economies with non-liberal political systems or in defective democracies. Te same applies to in-depth comparative research to explore variation in the way nationalinternational complementarities impact on social policy dynamics.

<sup>14</sup>Lack of health protection has, however, been reported to fuel participation in underground churches practising divine healing (Oblau 2011) and health-oriented Qi-Gong cults (Kupfer 2008). Both are perceived as destabilising by the leadership of the party-state, but their connection to social protection is less direct and immediate than in the other felds. Tis may help explain why it took longer for healthcare to become more of a priority.

## **References**


———. 2019. Quanguo gongshang baoxian canbao renshu da 2.39 yi ren [Te Number of Participants on Accident Insurance Reaches 239 Million People]. http://www.gov.cn/zhengce/2019-05/13/content\_5391241.htm. Accessed January 15, 2021.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **3**

## **Causal Mechanisms in the Development of Contribution-Based Pension Systems in South Korea, Vietnam, Sri Lanka, and Malaysia**

**Johanna Kuhlmann and Frank Nullmeier**

## **1 Introduction**

Since the turn of the century, social pension programmes have evolved as a new instrument for old-age provision in many Eastern, South-Eastern, and Southern Asian countries (Asher and Bali 2015; Barrientos 2015). Tis is remarkable given that some of these countries introduced social insurance systems only in the 1990s, partly following the International Labour Organization's (ILO) considerable eforts to establish socialinsurance-based pension schemes around the globe (Deacon 2015). In

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

J. Kuhlmann (\*) • F. Nullmeier

University of Bremen, Bremen, Germany

e-mail: johanna.kuhlmann@uni-bremen.de; frank.nullmeier@uni-bremen.de

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_3

recent years, many Eastern, South-Eastern, and Southern Asian countries have seen encompassing welfare state reforms (Hujo 2014; Haggard and Kaufman 2008; Asher and Kimura 2015). While much research has focused on the development of social pensions in these countries (Gliszczynski 2015; Leisering 2019; Barrientos 2015; Böger and Leisering 2020), it is important to keep in mind that social pensions have not replaced the older and often still much more comprehensive contributionbased systems. Against this backdrop, this chapter focuses on these contribution-based pension systems. Te trajectories of the contributionbased pension systems in many countries in these regions follow a similar pattern. Although they are facing considerable problems in expanding efective coverage and providing a decent pension after retirement, these countries have undertaken considerable political eforts to maintain or even expand those systems. We aim to explain these dynamics through a modular and mechanism-based approach and thereby specifcally seek to demonstrate the potential of mechanism-based analysis in comparative research. To this end, we compare two diferent types of contributionbased pension systems: pension systems that largely follow a Bismarckian logic and pension systems that largely rely on national provident funds (the latter have only received scant attention in the comparative social policy literature so far; but see Dixon 1989; Lindeman 2002; Kaseke et al. 2011). Generally, provident funds can be understood as a social policy legacy from British colonial rule. Most share the following characteristics: (1) Tey are compulsory for those in formal employment with larger employers. (2) Tey operate as defned-contribution schemes managed as part of the public sector. (3) Contributions are paid by both employees and employers. (4) By design, they accumulate interest based on defned rates. (5) Instead of regular payments, most provident funds rely on lump-sum payments for employees that enter retirement age (Kaseke et al. 2011). Tus, in contrast to Bismarckian pension schemes, they generally involve no redistributive elements and are not a pay-as-you-go system but only individual accounts. Moreover, lump-sum benefts have been criticised for not guaranteeing a regular income for the whole duration of retirement.

We have chosen the pension systems of South Korea, Vietnam, Sri Lanka, and Malaysia for our analysis, thus following a most diferent systems design for our comparison of contribution-based pension systems. Two of the countries have a Bismarckian social insurance scheme (South Korea, Vietnam), one with a rather high (South Korea) and one with a rather low level of coverage (Vietnam). Te two other countries have a national provident fund (Malaysia, Sri Lanka), again one with a rather high (Malaysia) and one with a rather low level of coverage (Sri Lanka). Te aim of this chapter is to show that similar mechanisms and outcomes can be found in these diferent contribution-based pension systems and thus highlight the set of key complex causal mechanisms that are at work across the diferent variants of contribution-based pension systems.

Te four countries difer with regard to a number of additional factors. Tese include, frst, their income level. Second, the countries also difer with regard to their types of political system and frequency of government changes. While South Korea and Sri Lanka are characterised by party competition, the Socialist Republic of Vietnam has always been governed by the Communist Party, and Malaysia experienced its frst government change only in 2018. Te national time frames of the study range from the initial establishment of a pension system for workers outside of the state sector up until today. Although this implies a relatively long time frame for countries with a national provident fund (which was established in Malaysia in 1951 and in Sri Lanka in 1958), Bismarckian pension systems covering workers in the private sector were introduced more recently in South Korea (1986/1988) and Vietnam (1995). Yet, irrespective of these diferent time periods, none of the selected countries have been able to achieve universal coverage, and there are signifcant differences between legal and efective coverage in each country. Data from the ILO illustrates this point: According to the most recent World Social Protection Report 2017–2019 (ILO 2017), legal coverage of mandatory contributory systems for old-age provision as a percentage of the workingage population is 70.9% in South Korea, 48.6% in Malaysia, 32.9% in Sri Lanka, and 33.1% in Vietnam. In contrast, efective coverage (indicating active contributors to a pension scheme in the working-age population between 15 and 64 years of age) was 53.7% in South Korea, 28.1% in Malaysia, 18.9% in Sri Lanka, and 20.6% in Vietnam.1 Although this data is not fully comparable, it nevertheless shows that important coverage gaps remain.

Te puzzle for researchers is to explain a permanently low efective coverage rate in spite of ongoing political eforts to maintain or even expand these contribution-based systems in all countries. Our analysis of the pension systems in the four countries reveals six complex causal mechanisms that explain important dynamics that are at play. In a nutshell, (1) the *outcompeting mechanism* explains the expansion of social policy as a result of party competition in democratic settings. Conversely, (2) the *gaining acceptance spiral mechanism* reveals how autocratic regimes introduce or expand social policies in order to increase their legitimacy. (3) Te *evasion mechanism* explains the interaction between national policymakers and international organisations and how national policy-makers announce reforms in line with recommendations from international organisations but do not implement them. As these political mechanisms are not limited to the realm of contribution-based pension systems—and, in the case of the evasion mechanism, not even to the realm of social policy—we defne these as general political mechanisms. In addition to these three mechanisms, we identify three more that are characteristic of contribution-based pension systems: (4) Te *double beneft mechanism* explains how money accumulated in a pension fund not only serves oldage provision but also helps to fnance other government purposes, thus leading to an expansion of the pension system. (5) Te *crisis management by going further mechanism* explains how a strategy of expanding social policies to include new groups can help to maintain the system's fnancial sustainability in times of crisis. (6) Te *alarmed middle classes mechanism* explains how fears among privileged groups of higher costs or lower benefts lead to an only moderate expansion of the pension system.

Te chapter is structured as follows. We start by presenting the causal efects and key causal mechanisms that we briefy identifed above. Te third section gives a historical overview of the pension systems of South Korea, Vietnam, Sri Lanka, and Malaysia, and traces the causal mechanisms that

<sup>1</sup>Note that data from national sources and for other time spans may suggest slightly diferent numbers.

are at play. Te fourth section provides our conclusion. Our analysis relies on a broad assessment of the secondary literature as well as on national documents and policy reports from international organisations.

### **2 The Development of Pension Systems: Causal Effects and Mechanisms**

Te trajectories of Eastern, South-Eastern, and Southern Asian pension systems indicate that contribution-based pension schemes are being maintained irrespective of the problems of using these as a basis to build a coherent pension system that covers most individuals. Te persistence of these systems we defne as the *continuation efect*. To specify this broader efect, we distinguish more specifc causal efects that characterise the development of social policy in the respective countries: Te *ratchet efect* indicates that contribution-based schemes once introduced are not politically abolished, transferred to other schemes, or fundamentally retrenched. Te *expansion efect* describes the extension of the pension system to additional groups; the *moderate expansion efect* is a variant of this that refers to an expansion that is limited by political dynamics. Finally, we identify an *embellishment efect*, a special form of interaction between national governments and international organisations, by which national governments signal agreement to avoid confict but do not efectively implement the measures in question. All efects result from diferent complex causal mechanisms that can be combined to explain the development of pension policy in the four countries. We thus rely on the set of mechanisms outlined above to show how a modular explanation can be achieved (Chap. 1). We frst focus on three more general political mechanisms that we identifed throughout the study of the four country cases before introducing three mechanisms that are more specifc to contribution-based pension systems. Note that this does not imply that all mechanisms play a role in all four cases. Rather, the mechanisms can be combined in diferent ways to explain the trajectories of the diferent pension systems in the countries in question.

#### **2.1 Outcompeting Mechanism**

Te outcompeting mechanism explains the expansion of the pension system. It can occur wherever there is competition in elections, especially party competition. Te relationship between party competition and social policy is a long-standing topic in the comparative social policy literature, which in this respect often focuses on diferences between parties to explain social policy developments (e.g. Schmitt and Zohlnhöfer 2019). Yet, party diferences might also blur if several parties aim to expand social policies. If several parties in a country have the prospect of winning the elections, the elites of these parties choose strategies that intensify political competition. Tere might be diferent reasons why the elites of competing parties decide to introduce or expand a social insurance scheme, among them economic growth, increased tax revenues, or hopes of a future increase in state revenues, not to mention broader political or societal changes that alter existing cleavage structures (Wong 2004). In any case, these party elites will only propose introducing or expanding a social insurance system if they also think that there will be a large number of potential voters who will welcome this policy, whether for personal economic reasons, as an expression of their country's advanced development, or out of enthusiasm for a normatively convincing political programme.

One variant of the outcompeting mechanism involves two or more strategically interacting parties. Since Party A may know that it makes sense for Party B to promise the introduction or expansion of social insurance institutions, and Party B may know that this also applies to Party A, both parties will outbid each other in their promises to extend social benefts (Derthick 1979). However, this is only successful if the parties can hope that voters will bear the costs of the expansion strategy later on or that rising costs will not be considered as mistakes or failures for which a certain party will be held accountable. Tis condition may be neglected, however, when party elites can aford to shorten the time horizon of their strategic considerations to the time of the election campaign and do not feel bound by norms of good governance.

#### **2.2 Gaining Acceptance Spiral Mechanism**

Tis mechanism can play a role in non-democratic countries and lead to the same expansion efect as in the case of the outcompeting mechanism in democratic ones. In the literature, it is widely accepted that not only democratic governments adopt or expand social policy measures but that autocratic governments also have reasons to do so. In particular, it has been highlighted that autocratic governments might adopt social policy measures to gain the support of particular groups as "a special form of co-optation" (Knutsen and Rasmussen 2018; see also Mares and Carnes 2009). Autocratic governments do not rely on repression and co-optation alone for survival. Tey also maintain stability by cultivating belief in their legitimacy (Gerschewski 2013). Autocratic regimes might opt for introducing or expanding social benefts if they see scope for additional public expenditure and view social policy as being a more efective means of lending themselves legitimacy than the pursuit of additional economic growth. Once established, these social policies can set in motion an acceptance spiral that draws on further social policy expansion. Tis acceptance spiral might become even more important if improving the living conditions of the population is a key ideological pillar of that autocratic regime, as in many socialist regimes (Wurster 2019). Moreover, even if social policy reforms do not signifcantly increase acceptance among the population, governments may still have reason to further expand social policies. Lack of acceptance may lead party elites to outbid each other in their reform eforts to bolster their support among the population.

#### **2.3 Evasion Mechanism**

Te evasion mechanism (Kuhlmann and Nullmeier 2021) sheds light on the role of international organisations in national reform processes. Tis aspect has been increasingly researched in comparative social policy in recent years (Orenstein 2008; Deacon 2015; Leisering 2019), with mixed fndings on the overall role that international organisations actually play in national social policy reform (see also Chap. 8, Chap. 9). In fact, while international organisations might be deeply involved in social policy reform, it is not unusual for the adopted policies to primarily refect national considerations. And even in this case, both national and transnational actors may well assess the cooperation as having been highly successful. Tis we defne as the embellishment efect. If, for instance, an international organisation urges a country to restructure its social security systems and to introduce or expand a social insurance scheme, national actors might employ strategies of circumventing recommendations by international organisations. Tey typically do not do so openly because of international pressure for reform. What national actors do instead is either stress that their planned policies are in line with international recommendations on social insurance schemes or develop policy compromises that make vague references to the recommended social insurance schemes. In a mixture of rational calculation and the normative belief that social policy legislation is a matter of national sovereignty, political elites resist international organisations' attempts to shape the direction of reforms while avoiding open dissent so as not to risk ofending these international organisations. To accomplish this, national policy actors obscure the fact that they are not (fully) following the respective recommendations and are not adopting the social insurance scheme in national policy accordingly so as not to endanger important contacts with the international organisations.

#### **2.4 Double Beneft Mechanism**

Pension schemes are often characterised by a double beneft mechanism. Te double beneft of pension funds lies in disbursing old-age provision to the population on the one hand and accumulating a considerable amount of money for the state on the other. Tis leads to the said ratchet efect of the pension system. Governments ensure the public fnancing of the pension fund because they want to use the money to fnance the public debt or promote the country's economic development (Holliday 2000; Koreh 2017; see also Kuhlmann and Nullmeier 2021, 2022).2

<sup>2</sup> In Kuhlmann and Nullmeier (2021), we focused on the *policy-areas interdependence mechanism*. It is strongly linked to the double beneft mechanism. However, whereas the term "policy-areas inter-

#### **3 Causal Mechanisms in the Development of Contribution-Based…**

Te size of the fund whets rational actors' appetite for accessing it for other purposes. Te prospect of sizeable amounts of discretionary funds also provides a strong incentive for the government to push for expanding the system if this entails additional contribution payments and thus increases the fund. Capital assets increase in particular if contributions have already been paid but no payments have yet been made due to a statutory minimum savings period or waiting period. Under these conditions, the government as well as opposition parties will fnd any expansion of coverage highly reasonable, as expanding the insured population becomes a viable policy option because of the attractiveness of further increasing the possibilities of state fnancing. Te impetus for enlargement will even be stronger if the government has to win elections. In this case, the promise of new or more encompassing social benefts may correspond with securing the fnancing of public debt or enhancing the national capital stock. Te other side of the coin is that abolishing contribution-based schemes becomes highly unattractive for a government that has once enjoyed the opportunity to tap such a reservoir to fnance its public debt or other government expenditures.

Te money from the pension fund also allows it to serve other purposes. For example, policy-makers can choose to use money from the pension fund to relieve pressure on a tight labour market by facilitating early retirement (Trampusch 2005). Tis constitutes a special variant of the double beneft mechanism.

#### **2.5 Crisis Management by Going Further Mechanism**

Tis mechanism can become relevant when a government is faced with a situation in which a social security system covering only a limited number of persons runs into massive fnancial trouble either because the number of insured persons is falling, or the contributions of insured persons are too low in relation to the promised benefts, or high infation rates are diminishing the fund's real value, or the funds are simply too small to

dependence" focuses on the overlapping of diferent policy felds, the double beneft mechanism stresses the material advantages that money from the pension system can have for several policy felds.

generate good investment rates in the fnancial market. In these cases, policy-makers can weigh several policy options, such as dismantling or transforming the system, cutting benefts, raising contribution rates, or providing massive state subsidies, for instance, in the form of tax breaks. If the government considers expanding the circle of insured persons to be another viable political option, this will become the preferred choice to emerge from rational calculation because it involves low burdens for both the group of already insured persons and the state budget. In this situation, rational government actors will increase the degree of inclusion to ensure the stability of the system, thereby leading to an expansion efect. Tus, the crisis of the old system with low coverage rates becomes the driving force for including other segments of the population.

#### **2.6 Alarmed Middle Classes Mechanism**

In countries with a large group of informal workers, formal workers can regularly be considered as members of the middle classes. As a privileged part of the population, these formal workers are among the frst groups to be included in a contribution-based pension scheme (e.g. Huber and Stephens 2012), and they do not want to lose these privileges. If a contributory insurance system has been introduced in a country for one group, extending this system of social protection to other parts of the population (such as farmers, fshermen, the self-employed, or informal workers) may appear to be a legitimate concern in the eyes of the groups already included in the system. Te middle classes will typically consider these demands argumentatively convincing and worthy of recognition on normative grounds, as equal rights for all combined with some form of performance justice, a characteristic especially of social insurance institutions, are widely recognised.

Tese middle classes therefore initially approve government plans to expand social protection schemes. However, when the conditions of expanded coverage begin to take shape in detail, it often becomes clear that this expansion will also have material consequences for those already insured. Especially if the new groups are considered to be fnancially weaker, this might stoke fears that expanded coverage could entail higher contribution rates or reduced benefts for all in the long term. Moreover, if the government grants subsidies to include new, fnancially less powerful groups into the system, the middle classes might perceive this as unfair. Such a discrepancy between initial normative approval of an expansion and the fears and concerns related to the enactment of the actual reform can incite anxiety among the middle classes. Tis constellation will allow for a rather moderate expansion of social protection to new groups only if the governing parties expect such an expansion to improve their prospects of electoral success.

#### **3 Country Studies**

In the following sections, we frst describe the trajectories of the national pension systems in the four countries under study (see also the overviews in Kuhlmann and Nullmeier 2021 for Vietnam and Sri Lanka, and Kuhlmann and Nullmeier 2022 for South Korea and Malaysia). We then explain the development of these systems by referring to the causal efects and mechanisms that were introduced in the previous section.

#### **3.1 South Korea**

#### **Overview of the Pension System**

Although the president of the former military regime, Park Chung-hee, had announced the introduction of a compulsory national pension system as early as in 1973, a general pension insurance was introduced only in 1988, following the country's democratisation. It then expanded quite quickly. Since its inception, the South Korean pension scheme has been based on the principle of defned benefts and can be attributed to the Bismarckian type, which implies that contributions are paid by employers and employees. While compulsory insurance initially applied only to employees working in companies with more than ten employees, this changed in 1992 to also capture employees working in companies with more than fve employees. Between 2003 and 2006, the compulsory pension scheme was extended to additionally include employees working in companies with fewer than fve full-time workers. Te pension system was also opened to new groups: It was extended to the rural self-employed, farmers, and fshermen (2.1 million people; Yang 2017, 121) in 1995, and to the urban self-employed in 1999. Notably, it was not until 2008 that the pension scheme issued the frst full pension payments. Tis owed itself to a minimum insurance period of 20 years. Additional pension schemes exist for diferent occupational groups, such as civil servants (this scheme was established in 1960), military personnel (established in 1963), and private school teachers (established in 1974) (Hwang 2006; Kwon 2014). Moreover, large companies also have company pension schemes. Together with Germany, Austria, Slovakia, and the USA, Korea is one of fve OECD countries whose system of old-age security does not provide a minimum pension (frst tier) but only social assistance with or without special regulations for older people (OECD 2017, 88).

Since 1999, the country's pension law has stipulated that the entire population shall be included in an old-age provision scheme with a standard retirement age of 61, rising to 65 until 2034 at a current contribution rate of 9%. Diferent types of mandatory membership exist for workplace-based insured and individually insured persons. While the contributions of the former group are paid by both the employer and the employee, those in the latter group pay all contributions themselves. Moreover, individuals can be insured voluntarily and in some cases exempted from compulsory insurance (e.g. in the event of unemployment, business closure, temporary leave, livelihood difculties, and in the case of students in tertiary education). To increase the level of old-age pension coverage, subsidies for low-income earners were introduced in 2012 (Duru-Nuri programme). Current studies suggest that this has increased the number of insured people over the following four years by 2% (Yoo et al. 2016). Yet, given that coverage rates continue to be low, especially among people younger than 35 years of age, extending the subsidy to additional groups is a matter of debate (Young et al. 2016). Efective coverage rates are at 68.2% of the economically active population between 18 and 59 (Young et al. 2016), which can be considered a relatively high rate compared to other Asian pension systems (Hujo and Cook 2012).

#### **Mechanisms**

First and foremost, the expansion of the South Korean pension system can be explained by two of our mechanisms: Te gaining acceptance spiral mechanism, in the late phase of the dictatorship, and the outcompeting mechanism, which characterised party competition during the democratisation process. Te military dictatorship established a lifetime employment system and a seniority wage system. In the course of the country's democratisation process, the lifetime employment system gradually gave way to a more fexible labour market. Te South Korean labour market thus became more fexible, and workers were compensated for growing employment insecurity by introducing social policy measures (Peng 2012), specifcally by an expansion of health insurance and old-age pension insurance. However, despite the strengthening of civil-society organisations, labour unions, and the single-member district electoral system, there remained major obstacles to a vigorous social policy (Huber and Niedzwiecki 2015; Yang 2017).

Overall, however, the democratisation process gradually enhanced the importance of social policy legislation (Shim 2019). In the years that followed, pension insurance was extended to more and more groups of the population, especially to the conservative party's agricultural clientele. An attempt at a neo-liberal restructuring of old-age provision failed during the Asian crisis of 1996–1998. Te poor performance of the old government during the crisis also made it possible for the former opposition leader Kim Dae-jung to win the presidency. Only after parts of the middle classes, including the labour unions, protested the far-reaching reform plans of his successor, Roh Moo-hyun, was Dae-jung able to wage a successful anti-welfare election campaign. Yet, in the subsequent presidential election, the new conservative candidate, Park Geun-hye, returned with an aggressive pro-welfare programme that the centre-left candidate tried unsuccessfully to outbid. Since 1988 (with the exception of 2008), party competition in presidential elections has served as a driving force for the expansion of the welfare state, even though many electoral promises were not implemented (Yang 2017).

Te development of the pension system can also be explained by the double beneft mechanism and the ensuing ratchet efect: Creating a capital stock through a funded pension insurance scheme had already played an important role in South Korea's political decision-making processes before the pension scheme was actually introduced (Hwang 2006, 57–64), especially among executives (Kim and Choi 2014) who promoted a welfare state concept that focused more on economic than on social objectives (Kim 2008). Within this framework, the central aim of social policy was to promote the national economy (see also Holliday 2000). Te creation of a pension fund came with an enormous accumulation of capital, especially given the fact that a minimum insurance period ensured that no payments had to be made for 20 years. In essence, this resulted in a Bismarckian system that was based on capital funding instead of a pay-as-you-go system. Te fund was ultimately administered autonomously but in close coordination with the state (Kim and Stewart 2011; Yang 2017, 140). Today, the National Pension Fund is the third largest pension fund in the world, with a capital base equivalent to 516 billion euros. Between 1988 and 2020, it generated an average return on investment of 4.62%, with 34.4% currently invested in fnancial assets and development projects outside South Korea. Te fund owns shares in state-owned companies, is deeply involved in the fnancial sector, and is subscribed to a shareholder-activism philosophy (Choi et al. 2018; Nomura 2011). In the light of its sheer volume, the role of the pension fund for economic purposes can hardly be overestimated. It can support the country's national foreign and fnancial strategy even if social policy goals have gained signifcance over the years (Kuhlmann and Nullmeier 2022).

Te alarmed middle classes mechanism can explain the moderate but limited expansion of the system to new groups. In fact, the social policy reforms under the presidencies of Kim Dae-jung and Roh Moo-hyun came under increasing pressure from parts of their own clientele. After a short phase of opening up the pension scheme to include non-standard workers (Durazzi et al. 2018), this strategy towards greater inclusion soon lost the support of the labour unions (Yang 2017). A crucial factor was the employers' reaction to the reforms under Kim Dae-jung. To reduce labour costs, they massively expanded non-standard employment (Park 2008), which revived the major labour unions' focus on an insider strategy. Tis abandoning of a universalist policy at the beginning of the 2000s can only be explained by dualisation (Peng 2012) and the split within the working class (Yang 2017, 154–83). Te high-earning working class as a central component of the middle class defended its social privileges and was increasingly sceptical about extending social security to other groups of the population. Teir own economic situation was improved more markedly by occupational social benefts than by public benefts (Yang 2017, 213). Te situation was diferent at the beginning of the reform movement under Kim Dae-jung. Political plans to universalise social security had been supported by the working class in large companies because extending social protection complies with the norm of universal social rights and the norm to improve the situation of disadvantaged informal workers. However, detailed information about the redistributive efects of expanded coverage, especially the decrease in expected pension benefts for the middle classes, led the latter to a switch from norm compliance to self-interested rational choice:

Labor unions and civic groups did not openly question the legitimacy of the redistributive mechanism when they were involved in the making of the moderate pension reform bill. However, when it became apparent that the average income of the urban informal sector was so low that the pension benefts for those currently covered (i.e., corporate employees in the formal sector) would be signifcantly reduced as a result of the expansion, labor unions and middle-class NGOs turned their back on it. Te average income of urban self-employed and informal sector workers (i.e., workers at small businesses with four or less and day and casual workers) was about half the average income of formal sector corporate employees. Accordingly, coverage expansion entailed a signifcant reduction of average income for contributors, which would in turn reduce pension benefts. (Yang 2017, 141)

Under the presidency of Roh Moo-hyun (2003–2008), opposition to costly reforms in social policy intensifed, particularly among the unions of large companies. However, not just opposition by privileged workers but also the routine avoidance of contributions by the self-employed was part of the middle-class resistance against a universalist pension insurance with redistributive efects.

When it comes to the role of international organisations, South Korea is a special case, as it is one of three countries (the other two are Slovenia and Venezuela) that have resisted the World Bank's push since 1994 (and intensifed after the Asian crisis in 1997) to privatise old-age provision (Yang 2004; Orenstein 2008, 45). Orenstein (2008, 155–56) ascribes this to the presidency of Kim Dae-jung (1998–2003), who disagreed with the plans of his predecessor Kim Young-sam. Nevertheless, there is some evidence of the evasion mechanism being at work here: A privatisation plan was drawn up by the government in 1995, which moved in the direction of the World Bank's ideas in order to obtain a loan worth billions. But then these plans were modifed more and more within the government and fnally dropped. Tis refusal to privatise was successful, even though the literature usually classifes South Korea as a "productivist welfare state" (Holliday 2000; Rudra 2008; see also London 2018). Te Asian crisis, with its high unemployment and company collapses, even among the large ones, the *chaebols*, led to a policy shift that departed substantially from the privatisation blueprint to continue a unifed insurance system. Te government was able to pit the recommendations of the ILO (Hagemejer and Schmitt 2012) against the World Bank programmes and thus demonstrate compliance (thus creating an embellishment efect), albeit compliance with the ideas of a competing international organisation (Kim 2008).

#### **3.2 Vietnam**

#### **Overview of the Pension System**

Te Vietnamese social insurance system was established in 1961 and included a non-contributory defned-beneft pension scheme for state sector employees (Long 2012, 205–6). In the course of *doi moi*—the Vietnamese process towards economic liberalisation which had started in 1986—the Vietnamese welfare regime was modifed to a considerable degree (see London 2018). Te existing pension system was converted into a pay-as-you-go defned-beneft scheme in 1995 (Long 2012, 206) and became compulsory for all formal workers, thereby capturing employees that have a contract for at least three months and that work in frms with at least ten employees (ILO 1996b; Long 2012, 206). Vietnam Social Security (VSS) was established as a state agency subordinate to the Vietnamese government and tasked with managing the social insurance fund (VSS 2018). Te current pension system comprises persons who are insured under the pension system both before and after the 1995 reform (Long 2012, 206).

Te Social Insurance Law of 2006 specifed the foundations of both compulsory and voluntary social insurance, including old-age benefts (Socialist Republic of Vietnam 2006). It stipulated that social insurance was open to all employees who have a contract of at least three months' duration, irrespective of the size of their workplace. Te voluntary pension scheme was designed as a separate system for self-employed workers and people residing in the rural area, particularly farmers, and was implemented in 2008 (Nguyen and Chen 2017, 237). Te Social Insurance Law of 2014 laid down important revisions that sought to expand the number of insured persons (Castel and Pick 2018, 18). Tis became especially clear as from 2018 on workers with a contract of one month duration were included in the social insurance scheme as well. Moreover, the pension system was also opened to specifc groups of part-time workers and some foreign citizens legally employed in Vietnam (International Social Security Association 2019, 283).

Te current contribution rates are 8% of gross monthly earnings for employees and 14% for employers; for the self-employed, the contribution rate is 22% of their declared earnings. Te general retirement age is 60 for men and 55 for women. To receive a monthly old-age pension, people need to have paid contributions for at least 20 years (International Social Security Association 2019, 283); otherwise, they receive a lump sum. Early retirement with reduced pension payments is also possible (Socialist Republic of Vietnam 2014). Available data from the ILO suggests that both legal and efective coverage rates have remained at relatively low levels: While legal coverage for old age as a percentage of the working-age population is estimated at 33.1%, efective coverage is estimated at 20.6% (ILO 2017).

#### **Mechanisms**

Te expansion of the pension system to include employees working in the private sector can be explained by the gaining acceptance spiral mechanism in the context of economic liberalisation: In the 1990s, the government decided that the pension system should no longer be limited to state sector employees but expanded to private-sector employees as well, who played a key role in the country's economic liberalisation process (Goodkind et al. 1999, 147). Apart from the political rationale, fnancial and demographic considerations refecting system requirements also played a role, which can be explained by the crisis management by going further mechanism. In fact, extending the pension system to new groups was perceived as crucial to maintaining the system's fnancial sustainability in a society that was both ageing and shrinking (Goodkind et al. 1999, 147–48).

In addition to the social policy objectives associated with the expansion of the pension scheme, the double beneft mechanism can explain why expanding the system is considered an attractive political option (see also Kuhlmann and Nullmeier 2021). Here, it is important to note that the pension fund cannot be considered only as a fnancing source for oldage pensions. For the Vietnamese government, it also presents an opportunity to "obtain funds at the lowest cost", which thus "give[s] fscal space and may sustain growth" (World Bank 2012, 12). In fact, in 1998, three years after the pension reform, the fnancial management of the fund had been specifed so as to allow the VSS to invest in government bonds and bonds issued by commercial state-owned banks as well as in larger state projects and enterprises, given the government's approval (ILO 2000d, 5).

It can be assumed that using the money from the pension fund for other purposes was especially attractive in the frst years of the fund, when not many pensions were paid out due to the minimum insurance period of 20 years, which led to quite large reserves in the frst years. According to a report from the World Bank, in 2010, "VSS had accumulated US\$5.78 billion of reserves, equivalent to about 5 percent of GDP" (World Bank 2012, 7). Today, the social insurance fund is the biggest welfare fund in Vietnam. Most of the money from the fund is invested in government bonds, bonds by commercial banks owned by the state (VSS 2018) and in bank deposits. In 2018, 90% of the social insurance funds were invested in government bonds (Viêt Nam News 2018). While this is considered a safe investment strategy, it provides only limited returns (World Bank 2012, 11; VSS 2018). A solution to this problem that has been discussed is diversifying the investment strategy by investing in "national key industries" as well as in "infrastructure development, electricity, transportation, urban development" (VSS 2018). Tis would imply that the pension fund would not only fnance the government debt but also actively promote the country's economic development. Another variant of the double beneft mechanism can be identifed when it comes to early retirement: Especially in the 1990s, a popular option for state enterprises was to phase out older workers (ILO 2000d, 7). Tus, the pension fund can be seen as performing important stabilising functions in the area of labour-market policy to solve employment problems.

Finally, when it comes to the role of transnational factors in shaping the Vietnamese pension system, evidence from the ILO suggests that the evasion mechanism and the resulting embellishment efect plays a certain role (Kuhlmann and Nullmeier 2021). Te Social Protection Development and Training Project from 1995 to 1999 provides an early example of how the ILO was involved in Vietnamese social policy-making and how national reform trajectories difered from ILO recommendations. Te aim of the project was to create a sustainable social security system for all Vietnamese employees (ILO 2000b, 1). In the course of the project, the ILO made a number of recommendations on future reforms of the Vietnamese pension system, including harmonising the pension age for men and women at the level of that of men, abolishing early retirement subsidies and lump-sum benefts (unless conditions for receiving a regular pension are not fulflled), granting pension credits for maternity and childcare, maintaining adequate income for survivors, and keeping contribution rates low. Moreover, the ILO welcomed expanding the system to include new groups (including voluntary insurance) (ILO 2000d, 36–39).

Te project's executive summary to the government, however, suggests that Vietnamese policy actors did not follow ILO recommendations. Te ILO even criticised that "not only have […] recommendations not been adopted, except for maternity pension credits, but the opposite policies have been followed" (ILO 2000d, v). A case in point is that early retirement was further facilitated, which—together with a low female pension age—helped to keep people outside of the labour market (ILO 2000d, 30). ILO recommendations for the following years included the drafting of a social security act and pushing voluntary pension schemes in order to expand old-age provision to include more groups (ILO 2000b, 35).

Although the Vietnamese pension reforms difered from the reform path suggested by the ILO, both the ILO and Vietnamese policy actors expressed only minor criticism and stated that the project had been highly successful overall (ILO 2000c, 14–24). Several actors, among them the Ministry of Labour, Invalids and Social Afairs (MOLISA), stated that future assistance was necessary to improve the Vietnamese pension system (ILO 2000c, 15).

Some pension reforms that were adopted after this project are clearly in line with ILO recommendations, such as the frst comprehensive Social Insurance Law in 2006, which strengthened the role of social insurance within the country. Moreover, Vietnam established a voluntary pension scheme. On the other hand, many elements of the pension system that had been criticised by the ILO are still in place, such as early retirement and diferent pension ages for men and women. Lump sums are only available to retired persons with fewer than 20 contribution years, yet the majority of people currently receiving them are people who leave employment with social insurance and "cash out" their contributions (Castel and Pick 2018, 16). In fact, many employees do not necessarily consider their social insurance contributions as retirement provision but rather as unemployment protection. Moreover, trust levels towards the pension system are not very high to begin with, which is why opposition among formal workers to abolish lump-sum benefts is very strong and the ILO recommendation to abolish lump-sum benefts is currently considered politically unfeasible (Castel and Pick 2018, 17).

#### **3.3 Sri Lanka**

#### **Overview of the Pension System**

Te pension system in Sri Lanka has a dualised structure (Karunarathne and Goswami 2002, 95), mainly consisting of pay-as-you-go schemes (defned benefts) for public-sector workers and mandatory saving schemes (defned contributions) for employees in the private sector. A pension scheme for the public sector has been in place since 1901 (Public Service Pension Scheme, PSPS), and a provident fund for public-service workers who are not eligible for the pension scheme was established in 1942 (Public Service Provident Fund, PSPF). Te frst provident fund for workers in the private sector was established in 1958 (Employees' Provident Fund, EPF) as a mandatory defned-contribution scheme run by the state. It is based on employers' and employees' contributions at fxed rates, which are then accumulated in individual accounts.

While the EPF constitutes the largest social security scheme in Sri Lanka (Employees' Provident Fund 2021), additional funds and savings schemes exist as well. In 1981, the Employees' Trust Fund (ETF) was established for employees in the public sector, university employees, and private-sector employees (Karunarathne and Goswami 2002). Moreover, the self-employed, as well as migrant workers, can seek coverage on a voluntary basis (International Social Security Association 2019, 244). Te ETF is funded by employers' contributions only (with a contribution rate of 3%), with the initial aim to "promote employee ownership of equities" (Karunarathne and Goswami 2002, 98). Moreover, three additional voluntary schemes have been created for informal-sector workers. Tey are the Farmer's Pension and Social Security Beneft Scheme (1987), the Fisherman's Pension and Social Security Beneft Scheme (1990), and the Pension and Social Security Beneft Scheme for Self-Employed Persons (1996) (Rannan-Eliya and Eriyagama 2003b, 3).

Te current EPF contribution rates are 8% for the employee (with the possibility of additional contributions) and 12% for the employer (International Social Security Association 2019, 244). Employees are entitled to their benefts—which are paid out as lump sums—when they reach the statutory retirement age of 55 (men) and 50 (women), respectively. In certain cases, it is possible to withdraw benefts earlier; these include marriage (for women only), permanent disability, and migration (Employees' Provident Fund 2017, 220). Moreover, since 2015 members are permitted to withdraw money for housing loans or medical treatment. Overall, coverage rates are relatively low in Sri Lanka, with legal coverage at 32.9% of the working-age population and efective coverage rates at 18.9% (ILO 2017).

#### **Mechanisms**

Why do policy-makers in Sri Lanka stick to the provident fund? Evidence from the secondary literature suggests that the double beneft mechanism is very important when it comes to understanding this ratchet efect (see also Kuhlmann and Nullmeier 2021). Te EPF is Sri Lanka's largest savings stock. Whereas, as in many other former British colonies, it was initially established as a "second-best alternative to the national pension system already existing in the United Kingdom" (Rannan-Eliya and Eriyagama 2003a, vi), it also plays an important role in fnancing the government's structural fscal defcit (Karunarathne and Goswami 2002). In 2016, 93.1% of the EPF's investments were in government securities (Employees' Provident Fund 2017). Moreover, money from the EPF not only plays a role in old-age provision but also serves other (social) policy purposes, a fact that the fund explicitly acknowledges. A case in point is in particular pre-mature withdrawals for a housing loan. According to the EPF's self-description:

Te EPF is not only a helping hand or a shoulder to lean upon in the winter of life but a great partner throughout, for it will provide you with the option of obtaining a housing loan […] Tus, the EPF will help you realize your dream of a home before retirement. (Employees' Provident Fund 2021)

When it comes to transnational factors, there is also evidence of an embellishment efect, which can be explained by the evasion mechanism (Kuhlmann and Nullmeier 2021). In fact, the ILO has advocated for a contribution-based social insurance pension scheme in Sri Lanka at least since the 1980s (ILO 1980), as provident funds generally had some "intrinsic defciencies" (ILO 1980, 16) when compared to social insurance schemes as promoted by the ILO. Most importantly, lump sums instead of periodic payments provided only an "inadequate" level of social protection, and there was the "impossibility of maintaining the real value of the Provident Fund contributions" (ILO 1980, 16). Te ILO therefore recommended transforming the EPF into a contribution-based social insurance scheme with periodic payments. In 1982, a project on introducing a social insurance scheme was launched (ILO 1984). While the fnal report outlined several options for introducing such a pension scheme, a gradual conversion of the EPF into a pension scheme, thereby transferring all assets and liabilities, was considered to be the most fruitful option (ILO 1984, 45–48, 1991, 6). In 1989, the Sri Lankan government decided to replace the EPF with the Employees Pension Scheme and create the National Social Security System (ILO 1991, 1). Yet, the ILO's "Report on the Conversion of Provident Fund to Social Insurance System Prepared for the Government of Sri Lanka" (ILO 1991) indicates that the reform had been "delayed" (ILO 1991, 1) due to opposition by two domestic actors: While the Ministry of Labour and Vocational Training was concerned about the structure of the new benefts, the Ministry of Finance worried that the money collected through the EPF could no longer be used that easily for government fnancing, which again points to the role of the double beneft mechanism in the development of the pension scheme (ILO 1991, 1). Moreover, it is important to note that other domestic actors, such as trade unions and employer organisations, adopted a rather sceptical stance as well. Te ILO was asked to "review the situation and suggest a basis on which implementation of the cabinet decision can proceed" (ILO 1991, 1), which it laid out in its 1991 report. However, until today, the EPF continues to be the main instrument for old-age provision in Sri Lanka, and a pension scheme in line with ILO recommendations has yet to be implemented. Moreover,

the ILO no longer recommends converting the EPF into a pension scheme in ofcial documents but only to convert lump-sum benefts into periodic payments (ILO 2008, viii).

#### **3.4 Malaysia**

#### **Overview of the Pension System**

While in Sri Lanka, the Employees' Provident Fund (EPF) was introduced after independence, in Malaysia, it was established by the British colonial government in 1951 (thus making it one of the oldest provident funds in the world; see Dixon 1989). Te EPF is a mandatory defnedcontribution scheme for workers in the private sector as well as for some workers in the public sector that are not covered by a public-sector scheme. Generally speaking, the Malaysian pension system is highly fragmented. Apart from the EPF, there are a number of diferent old-age provision systems, such as the Civil Service Pension Scheme, stemming from colonial times (Darmaraj and Narayanan 2019), and the Armed Forces Fund, established in 1972 (Asher 2012). Te EPF can nevertheless be considered the central institution within the Malaysia pension system. In 2016, it was the 15th largest pension fund in the world (Price et al. 2018, 7). Over the years, the EPF was continuously expanded to cover all workers in the private sector, irrespective of the size of their workplace (Asher 2012, 106), and it was also opened to self-employed workers in 1977 (Employees Provident Fund 2019, 18). In recent years, one aim of the Malaysian government has been to extend the EPF to all Malaysian citizens, including housewives and other groups who do not belong to the group of formal workers. People who decide to contribute voluntarily to the EPF can receive an additional government subsidy to supplement their voluntary contributions (i-Saraan) (Employees Provident Fund 2019; Price et al. 2018).

Each individual member has two accounts comprising 70% (account 1) and 30% (account 2) of the contributions, respectively. At the age of 55, members can withdraw the accumulated money, and both accounts are merged. From the age of 50 on, members can withdraw the money accumulated in the second account. Earlier withdrawal from this account is moreover possible for purposes related to housing, education, health needs, or the Islamic pilgrimage to Mecca (hajj). For approved investments, it is also possible to withdraw money earlier from the frst account (Ramesh 2005, 193–94; Price et al. 2018, 15). It is not obligatory to retire at the age of 55 and withdraw the money from the EPF.

Te monthly contribution rate depends on employees' age and income. In 2018, the general employees' contribution rate was 11%. Te employers' contribution rate varied between 12% and 13% depending on the employees' income. For employees older than 60, the regular contribution rate is halved (Price et al. 2018, 13). Moreover, additional voluntary contributions to the frst account are possible. According to current data, roughly half of the Malaysian labour force contributes to the EPF, indicating more or less universal coverage of formal private-sector workers, while coverage gaps remain when it comes to self-employed workers and workers in the informal sector (Price et al. 2018, 42; see also ILO 2017).

#### **Mechanisms**

Despite international pressures to convert the EPF into a pay-as-you-go pension scheme, the EPF has remained in place, and coverage has been continuously expanded to the point of almost achieving universal coverage in the late 1980s—at least when it comes to formal workers (Ramesh 2005, 193). One reason for this ratchet efect of holding on to the EPF is that its functions go beyond old-age provision. In fact, the double beneft mechanism plays an important role (Kuhlmann and Nullmeier 2022).

Like in Sri Lanka, the provident fund was created as an alternative to social insurance schemes, the establishment of which was considered too ambitious at the time. As some authors have argued, the introduction of national provident funds during British colonial rule refects an acknowledgement by British colonial administrations that at least some social policy measures had to be introduced to address the social situation in the colonies, particularly with regard to the "increasingly urbanized and industrialized workforce" (Kaseke et al. 2011, 145; McKinnon et al. 1997; see also Chap. 6). Taking this into consideration, it is plausible to at least explain the introduction of the provident fund in Malaysia by means of the gaining acceptance spiral mechanism, although the immediate British infuence of course vanished after independence.

Policy-makers in Malaysia quickly realised the fund's great economic potential (Ramesh 2005, 192), using it as an "essentially bottomless source of long-term investment capital with which to fnance state-led industrial and infrastructural development" (McKinnon 1996, 50). In fact, EPF balances have grown considerably throughout the years and accounted for around 60% of GDP in 2016 (Price et al. 2018, 28). In the early days of the EPF, all funds had to be invested in government securities or bank deposits, and until the 1980s, 80–90% of the money from the EPF was in Malaysian government securities. Since the 1990s, the EPF's investment portfolio has become more diversifed, now also including privatisation programmes and joint-venture projects, equities, and debentures of public companies, as well as investments in the housing market (Ramesh 2005, 203). Importantly, this not only fuelled the Malaysian capital market but also enabled the government to fnance big infrastructure projects (Price et al. 2018, 25) that can be considered important for the country's economic development. Moreover, the EPF now also invests in foreign markets, which resulted in a share of foreign investments of 30% in 2017 (compared to only 1% in 2006). Tese overseas investments followed a solid investment strategy but they also refect the fact that the EPF had become too large for the domestic fnancial market (Price et al. 2018, 28; see also Asher and Bali 2015). In 2018, the EPF's rate of return on investment was 6.57% (Employees Provident Fund 2019, 16). Te EPF explicitly acknowledges not only its function for old-age provision but also its key role in promoting the country's economic development:

In its initial years, the EPF was able to park all of its assets locally in Malaysian Government Securities, Loans and Bonds, Equities, Money Market Instruments and Property. In so doing, it served to help fnance a large number of major government as well as private sector projects that have contributed towards shaping the nation's development. More recently, it has had to look at suitable investment options outside of the country, as the local capital and money markets have not kept pace with the growth of EPF's assets. (Employees Provident Fund 2013, 3)

Some authors suggest that the EPF's role in state fnancing is even more central than the EPF's role in old-age provision: "All in all, the achievements of the EPF and the CPF [the provident fund in Singapore, JK/FN] are considerable, despite the fact that they do little in the areas of income protection and health care. Tey played a signifcant role in fnancing economic development projects in the early stages of industrialization" (Ramesh 2005, 207). In this regard, the gaining acceptance spiral mechanism can also play a role in explaining the expansion of the EPF, as the country's economic success—to which the EPF contributes decisively—is crucial to the legitimacy of the Malaysian political system (McKinnon 1996).

For decades, the ILO has advised Malaysia to restructure the provident fund and to establish a pension scheme based on social insurance principles (McKinnon 1996, 48). A key reason was that the ILO perceived lump-sum benefts as inadequate and was concerned that many "will fritter their money away […] Clearly if it is accepted as desirable to compel workers to save for their old age, it is equally desirable to ensure that, when the time for retiring comes, the worker is protected against the rapid dissipation of his resources through misfortune, through his own folly or through the wiles of others" (ILO 1960, 62). Tere have been several projects—initiated by the government of Malaysia and involving Malaysian stakeholders—on how the EPF could be converted into a pension scheme (ILO 1996a, 2000a). An ILO report (ILO 2000a, 93–97) summarises that Malaysian stakeholders engaged with the suggestions for converting the EPF into a pension scheme; however, a summary of the discussion held at a national seminar also shows that Malaysian stakeholders stress the need for further studies and discussions with regard to key questions (such as retirement age or such a scheme's relation to the civil service pension scheme), all of which would require more time. In practice, this has resulted in a suspension of the suggested reform, which is in line with the evasion mechanism. In fact, especially given the EPF's economic importance, it is rather unlikely that policy actors will adopt recommendations by international organisations to convert their pension system:

EPF managers believe pragmatically that they have the best option. Te duality of the developmental role of the EPF is at the heart of this issue, and it is this more rounded appreciation of that institution's impact that suggests that the Malaysian government's rejection of the ILO's social security-focused advice will be repeated in the case of the World Bank's equally unbalanced concentration on issues relating to investment efciency. (McKinnon 1996, 48)

#### **4 Summary and Conclusion**

Te inherent problems of contribution-based pension systems are frequently discussed in the social policy literature on Eastern, South-Eastern, and Southern Asian countries. At frst glance, it might therefore seem quite noteworthy that, these well-known problems notwithstanding, many countries do not only stick to their contribution-based pension system but even undertake political eforts to expand them. Focusing on the cases of South Korea, Vietnam, Sri Lanka, and Malaysia, our analysis identifed similar causal efects and causal mechanisms. Taken together, they provide a modular explanation for this general observation of a maintenance and an expansion of contribution-based pension systems at the policy level, despite limitations when it comes to expanding their efective coverage. Notably, for diferent variants of contribution-based pension systems and for countries with diferent starting points as well as diferent political, societal, and economic backgrounds, the set of causal mechanisms that we identifed provides a useful and productive instrument for a modular explanation of the trajectories of these pension systems (Kuhlmann and Nullmeier 2021). Tis chapter identifed six key causal mechanisms that can explain these dynamics, although to a diferent extent. When it comes to the general political mechanisms, the analysis was able to identify the evasion mechanism in all four cases. By contrast, the outcompeting mechanism—which does not capture reform dynamics in autocratic regimes—could only be identifed in the South Korean case, while the gaining acceptance spiral mechanism played a role in the cases of South Korea, Vietnam, and, at least to a certain extent, Malaysia. With regard to the institution-specifc mechanisms, the double beneft mechanism fgured prominently in all four cases. While the analysis showed that all countries—albeit to diferent degrees—use the money from the pension fund for national investments, the Vietnamese case also presents an example in which money from the pension system is used to tackle problems in the labour market. In contrast, evidence for the crisis management by going further mechanism could be found solely in the Vietnamese case and the alarmed middle classes mechanism only in the South Korean case. Te fact that we fnd more cases for some mechanisms than for others means neither that these mechanisms are theoretically superior to the other mechanisms nor that these mechanisms are generally more likely to explain social policy developments. Te six mechanisms are a frst step in demonstrating how comparative studies can work with causal mechanisms. Examining individual cases allows us to develop a set of complex mechanisms that may then be employed to study other cases. However, whether the mechanisms derived from such case studies can explain additional cases is a matter that must be decided in empirical research.

#### **References**


#### **3 Causal Mechanisms in the Development of Contribution-Based…**

*for Older Persons. Social Pensions in Asia*, ed. Sri Wening Handayani and Babken Babajanian, 137–152. Mandaluyong City: Asian Development Bank.


———. 1984. *Social Security Pension Scheme SRL/82/004*. *Report Prepared for the Government of the Democratic Socialist Republic of Sri Lanka by the International Labour Organisation Acting as Executing Agency for the United Nations Development Programme.* Geneva.

———. 1991. *Report on Conversion of Provident Fund to Social Insurance System Prepared for the Government of Sri Lanka*. Geneva.

———. 1996a. *Malaysia. Report to the Government on the Development of Social Protection Policy and Strategy*. Geneva.

———. 1996b. *Viet Nam. Social Protection Development and Training Phase I—Assessment Report*. Geneva.

———. 2000a. *Malaysia. Report to the Government on the Employees Pension Scheme and the Social Security Scheme for the Self-Employed*. Geneva.

———. 2000b. *Social Protection Training and Development in Viet Nam: Project Terminal Report*. Geneva.

———. 2000c. *Viet Nam. Terminal Project Evaluation Report*. Geneva.

———. 2000d. *Viet Nam. Report to the Government on Social Protection Development and Training*. Geneva.

———. 2008. *Social Security in Sri Lanka. A Review of the Social Security System for the Ministry of Labour Relations and Manpower*. Geneva.


———. 2022. Te Formation of a National Capital Stock and the Pension Systems in South Korea and Malaysia. In *International Impacts on Social Policy: Short Histories in Global Perspective*, ed. Frank Nullmeier, Delia González de Reufels, and Herbert Obinger, 371–383. Cham: Palgrave Macmillan.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **4**

## **Causal Mechanisms in the Introduction and Development of Unemployment Insurance in Turkey**

**Kerem Gabriel Öktem**

## **1 Introduction**

Turkey created an unemployment insurance programme relatively late, in 1999, with the frst benefts only being paid out in 2002. Despite its short lifespan, the programme has already been transformed through successive reforms. While key statutes to protect employees in the case of job loss have remained intact, new policy instruments have been added to the programme. Collectively, these new instruments have changed the logic of the programme from passive labour market policies to active labour market policies and employment generation.

K. G. Öktem (\*)

University of Bremen, Bremen, Germany e-mail: oektem@uni-bremen.de

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

In this chapter, the causal dynamics behind the introduction and the transformation of Turkey's unemployment insurance programme up to 2019 are traced (for an overview of developments in 2020 see Öktem 2021). I identify four causal mechanisms in the process: *double beneft, business-led reform, transnational cooperation* and *outcompeting* (note that this list may not exhaust all relevant mechanisms).

First, the double beneft of unemployment insurance as both a social policy programme and a fnancing tool is crucial for understanding its development (double beneft mechanism). In the context of repeated failures to prudently design and manage social security schemes and against the backdrop of a deep economic crisis and the prospect of IMF intervention, unemployment insurance was designed so that beneft payments would in all likelihood never exceed contributions. Tis led to a structural surplus and ever-increasing fund reserves, possibly beyond what policymakers had envisioned when designing the policy. As a result, unemployment insurance had a "double beneft" for policymakers: it served as a social policy tool, but also as a device for fnancing the budget defcit (during a period of strict IMF supervision).

Tis structural surplus and the fund's ever-growing reserves, over time, generated pressures for reform. Business and labour unions demanded a lowering of contribution rates to a level that would be sufcient to pay benefts or increasing the quality of benefts. Yet, these demands were largely ignored. Instead, successive governments established a number of new instruments in the unemployment insurance programme. Tese new instruments were mainly active labour market policies (ALMP), such as vocational training courses, which had long been demanded by business. Terefore, these changes can be described as business-led reforms (business-led reform mechanism). ALMP also broadly ft into the agenda of a Europeanisation of labour market policies and mirrored developments in European countries, where an increasing emphasis on ALMP had been visible in the previous decades (Kenworthy 2010). In this sense, transnational cooperation with the European Union also shaped the direction of policy change (transnational cooperation mechanism). Furthermore, these instruments were often launched and expanded with a view on "outcompeting" the opposition. Governments aimed to maximise their prospects in elections by fexibly deploying tangible benefts to participants of these policies (outcompeting mechanism).

#### **4 Causal Mechanisms in the Introduction and Development…**

Te fndings add to the literature on causal mechanisms in the development of social insurance schemes and also resonate with the literature on the fscal politics of the welfare state (Koreh 2017a, 2017b). Tis literature argues that the expansion of social insurance may be driven by fscal concerns and that fscal arrangements may infuence the subsequent development of social insurance. Tus, social insurance programmes may have a double beneft, as social policies *and* as state-fnancing devices. Usually, this double beneft is most important for pension insurance (see also Chap. 3). However, the chapter shows that it can also apply to unemployment insurance. Yet, it is important to emphasise that the double beneft mechanism, by itself, does not explain the whole story. It may help to explain the initial policy design and why the programme was transformed, but it does not explain the further trajectory of the programme.

In terms of research method, the chapter builds on a combination of policy document analysis, archival research and elite interviews. With regard to policy documents, I analysed all primary legislation related to the programme. For the archival research, I surveyed parliamentary proceedings of key legislative changes, as well as government reports on the programme. Additionally, I scanned Turkish newspapers (primarily *Cumhuriyet*, *Milliyet* and *Dünya*) for relevant reports and looked at reports prepared and statements made by representatives of business and labour. I complemented this policy and archival research by interviewing two former Ministers of Labour and Social Security, who were in charge of the programme for nearly a decade.1

Te structure of the chapter is as follows. First, the literature on causal mechanisms in the development of unemployment insurance is surveyed. Ten, I describe the creation of the programme in Turkey. Here, I trace how a draft bill prepared in the early 1990s became the basis for the unemployment insurance legislation. Next, I explore the transformation of the programme from 1999 to 2019 and show how the accumulation of a massive surplus and pressures from business facilitated comprehensive policy changes towards more active labour market policies. Finally, I conclude by discussing the implications of the fndings for the comparative literature.

<sup>1</sup>Te interviews were conducted together with H. Tolga Bölükbaşı on 29 August 2018 in Izmir and on 10 January 2020 in Ankara.

## **2 Causal Mechanisms in the Development of Unemployment Insurance**

Unemployment insurance is a central institution of modern welfare states. It protects workers against job loss by paying regular cash benefts to the unemployed. Tis decommodifes workers to a certain degree (Esping-Andersen 1990), but also acts as an automatic stabiliser in case of economic recessions. It constitutes a signifcant intervention in the labour market, and this makes it one of the most contentious forms of social insurance (Sjöberg et al. 2010). Tis is refected in its incomplete and lagged difusion throughout the world. Compared to other forms of social insurance, insurance against unemployment has been adopted by far fewer countries. In terms of the adoption sequence, unemployment insurance typically comes last (ILO 2014, 4; Schmitt et al. 2015).

Nevertheless, around ninety countries in the world have adopted unemployment insurance. First implemented on a nationwide level in 1905 in France, the policy initially spread in Europe and was adopted by some non-European countries (e.g. South Africa, Uruguay) in the interwar years. In Southern Europe, the frst adoptions came after the First World War in Italy and Spain, while Greece, Cyprus and Portugal created programmes only after the Second World War. In the Middle East, the frst adoptions came after the Second World War (e.g. Israel and Iran). Compared to these countries, Turkey is a latecomer with legislation only being passed in 1999 (Sjöberg et al. 2010; SSA 2016).

Te comparative literature has identifed various drivers behind unemployment insurance. Te policy ofers tangible benefts to employees whose representatives—labour unions—are generally assumed to support the policy. In fact, in various cases, labour unions implemented unemployment insurance before the state did (Flora and Alber 1981, 152). Tat is why it is generally assumed that labour unions pushed for the introduction and expansion of unemployment insurance. However, the reality is more complicated. For various reasons, such as fear of cooptation, unions actually opposed unemployment insurance in some countries (Flora and Alber 1981, 153–54).

#### **4 Causal Mechanisms in the Introduction and Development…**

Business has many reasons to be lukewarm towards unemployment insurance. It not only increases non-wage labour costs, but also increases the bargaining power of labour through providing workers with a reservation wage (Paster 2013). However, the policy also has benefts for business. For instance, because workers can receive benefts in case of job loss, layofs can be easier to manage (Carter et al. 2013, 4). Hence, it may even serve de facto fexibility of the labour market. Furthermore, unemployment insurance increases productivity, which is benefcial for employers (ILO 2017, 41–42). Tese diferent advantages and disadvantages for business may translate into inter-sectoral confict among business, with high-risk sectors supporting and low-risk sectors opposing unemployment insurance (Mares 2003). Furthermore, business may interpret the policy depending on the prevailing ideational climate (Münnich 2010). Tus, although it is perceived as an arena of confict between labour and business, the policy also ofers an opportunity for "cross-class alliance" (Hellwig 2005).

In addition, even in cases where business opposes policy adoption, it might have an interest in sustaining or shaping the policy once it is implemented (Paster 2017). Terefore, it might propose reforms. Tis could involve programme downsizing through decreasing contribution rates and benefts, but also expansion of policy instruments from unemployment benefts to a broader set of labour market policies. In those cases where business is successful in convincing governments to adopt its proposals, one could speak of "business-led reforms".

From the perspective of the state, the issue is complex. Within competitive regimes, policy adoption may constitute a response by policymakers and/or bureaucrats to demands from the broader electorate or from workers (Hicks 1999). Alternatively, policymakers may support unemployment insurance as a result of party competition. However, the popularity of the policy may be limited if the unemployed are generally seen as undeserving and few people expect to beneft from the insurance. Particularly in authoritarian regimes, the policy may also be created in anticipation of future demands by workers. Furthermore, unemployment insurance may also be attractive from a statist perspective as it promises to defuse labour confict (Matsunaga 2017) and signifcantly increase state capacity in the realm of labour market policies.

Finally, unemployment insurance may not just increase the bureaucratic capacities of the state apparatus, but also strengthen state fnances. As with any type of social insurance, unemployment insurance has the potential to be of "double beneft" to the state, as an instrument of social policy and as a fnancing device for the state. Although this "double beneft" is most relevant for pensions due to the peculiarities of old-age pension systems, it may also be relevant for unemployment. First, if the unemployment insurance system builds up a surplus, the funds allocated to the programme may be used to fnance budget defcits. Often, social insurance (SI) entails regulations that funds may only be invested in certain areas, which in practice then translates into mostly buying public debt. Second, funds may be used to fnance other projects. Te comparative literature cites cases where an SI surplus is used to "fnance projects of state building, ofoading the costs of industrial restructuring and covering national debt" (Koreh 2017a, 117). Hence, although SI contributions are in principle "earmarked for specifc SI schemes", it would be wrong to understand "SI as a closed fnancial system" (Koreh 2017a, 117).

Tird, the generation of a sizeable surplus may also shape the development of social insurance, particularly within competitive regimes. It may lead, for instance, to demands for a decrease in contribution rates. Tese demands would likely be raised by employers and right-of-centre parties, which are both sensitive to the level of non-wage labour costs and cautious about perceived detrimental efects of "decommodifying" social security. Conversely, a surplus may also facilitate programmatic expansion. Te high level of contributions may create a need for "legitimation" which could underpin welfare state expansion. Alternatively, the surplus may also facilitate expansion as it gives policymakers the means to make social security more generous (the "surplus efect", Koreh 2017b).

In recent decades, unemployment insurance has been increasingly complemented by active labour market policies (ALMP), such as retraining, that focus on increasing the employability and productivity of the labour force (Kenworthy 2010; Weishaupt 2019). In institutional and administrative terms, ALMP may be closely linked to unemployment insurance. In some cases, ALMP is at least partially fnanced by contributions made to unemployment insurance and managed by public employment services (PESs), which also administer unemployment insurance programmes, and eligibility to ALMP is tied to requirements for unemployment insurance. In terms of the drivers behind ALMP, business appears to be more supportive than labour, although labour unions have been shown to support some instruments such as training and employment assistance (Tepe and Vanhuysse 2013).

With regard to the Turkish case, research on the causal drivers behind the development of unemployment insurance has been limited. Te arduous process of policy adoption has been analysed with a focus on how transnational cooperation with the International Labour Organization (ILO), the Organisation for Economic Co-operation and Development (OECD) and the World Bank shaped difusion (Özkan 2011) and with a focus on how domestic political obstacles prevented an earlier adoption (Öktem 2020a). Much less is known, however, about how the programme has been transformed. While it has been argued that the introduction of unemployment insurance failed to alter the broader dynamics of unemployment compensation (Özkan 2016), efective access to unemployment benefts did increase (Öktem 2020b). Moreover, some researchers observed an increase in ALMP (Gün 2013; Lordoğlu and Koçak 2015; Kapar 2015), which some read as signifying a neoliberal transformation (Gün 2016). Still, the causal drivers behind this process remain to be unearthed. In this chapter, I aim to fll this gap in the literature and explore which mechanisms drove the introduction and transformation of Turkey's unemployment insurance programme. I will particularly focus on how the policy was designed and how the policy design facilitated the programme's transformation.

#### **3 The Creation of Unemployment Insurance**

#### **3.1 A Failed Attempt to Introduce Unemployment Insurance in 1992**

Policymakers frst brought up the idea of creating an unemployment insurance programme in Turkey when they discussed the introduction of social insurance in 1935. From the late 1950s onwards, various governments prepared draft bills, often with the support of the ILO. Unemployment insurance became a bipartisan development goal with parties trying to "outcompete" each other with promises on this front but failing to live up to their promises when in power. In 1983, unemployment insurance legislation was even passed by parliament, but not implemented by the ruling military junta. In the context of a structural transformation of the economy from statism to neoliberalism in the 1980s, unemployment insurance was dropped from the government's agenda. Te centre-right Motherland Party (*Anavatan Partisi*, ANAP) governments of the 1980s opposed the policy (Öktem 2020a).

Te situation changed in the early 1990s when ANAP lost power to a centrist coalition government of the centre-right True Path Party (*Doğru Yol Partisi*, DYP) and the centre-left Social Democrat Populist Party (*Sosyal Demokrat Halkçı Parti*, SHP). Both parties had made unemployment insurance part of their election campaign. Te Minister of Labour, Mehmet Moğultay from the SHP, eagerly worked on drafting legislation. However, he immediately encountered resistance from business, which in the words of the former head of the Turkish Federation of Employer Associations (*Türkiye İşveren Sendikaları Konfederasyonu*, TISK), "went to war with the new government and the Minister of Labour Mehmet Moğultay over unemployment insurance" (Baydur 2006, 62).2 Te strategy of business was twofold. First, it insisted on a simultaneous cutback of severance payments, which had acquired the status of a functional equivalent of unemployment insurance (Başterzi 1995), as a precondition for introducing unemployment insurance—knowing full well that labour would not accept this. Second, it lobbied the centre-right DYP, which was the stronger party in the coalition. Eventually, this strategy succeeded.

However, Moğultay managed to prepare draft legislation, which became the standard on which subsequent drafts for legislation were based (Andaç 1999). Te draft bill was guided by ideas from the OECD and the World Bank that entailed concerns about making unemployment insurance not too generous (Özkan 2011). It linked beneft duration to the contribution period. Employees who had worked for

<sup>2</sup>Troughout the chapter, all Turkish quotes were translated.

twenty/thirty/thirty-six months of the last three years would receive benefts for six/seven/eight months. Tis was slightly reduced to four/six/ eight months in late 1992. Benefts were set at 45 per cent of the employee's previous net wage, with a beneft foor set at half the minimum wage. Financing was to be based on employer, employee and state contributions, with each contributing 2 per cent of the employee's gross wage (Öktem 2020a). Tis was equal to the contributions set in the ill-fated 1983 legislation that had not been implemented—even though the new proposal was markedly less generous than the 1983 legislation (Danışma Meclisi 1983).

Overall, the draft bill was geared towards labour market insiders, as people in intermittent employment could hardly fulfl contribution requirements, and benefts were not generous. Nevertheless, the plan encountered strong criticism from business, which maintained that severance pay reform should be a precondition for introducing unemployment insurance. Furthermore, business, represented by TISK, was afraid that unemployment insurance would raise non-wage labour costs and therefore demanded that the programme should not increase overall employer contributions. Similarly, labour unions, which generally supported the introduction of unemployment insurance but opposed severance pay reform, rejected any increase in employee contributions (Milliyet 1992b).

Te question of fnancing therefore posed the government with a conundrum. It needed to raise funds for the programme, but social partners were unwilling to contribute. Initial funding was to come from the state, but this alone could not maintain the programme. As a solution, the idea to divert money from other funds was proposed, and by 1993, Moğultay supported this demand. To ensure that unemployment insurance did not pose any additional burden for social partners, other contributions that employers and employees had to make were to be reduced (Öktem 2020a). For this purpose, two funds were identifed: the Homeownership Support (*Konut Edindirme Yardımı*, KEY) (Cumhuriyet 1993a) and the Mandatory Savings Fund (*Zorunlu Tasarruf Fonu*) (Milliyet 1993).

Both funds had been created by the previous ANAP government against the opposition of the DYP and SHP. Both policies had been devised in the heyday of neoliberalism in Turkey and aimed to achieve social security through fnancialisation—in an economic context of strong growth and high infation. Arguably, both policies had been designed with the "double beneft" in mind. Te idea behind KEY, legislated in 1986, was that workers would be supported when buying a house or fat (Resmi Gazete 1986). For this purpose, employers paid a fxed contribution for their employees into a fund. Tese contributions were invested mainly in public debt. After a contribution period, employees buying a home would receive a support payment based on their contributions plus interest. If employees did not buy such a home, their dependants would receive a payment on the worker's death. KEY soon ran into problems. Critics argued early on that a large amount of money was collected without workers really receiving benefts. Eventually, contributions were stopped in 1996, and KEY was abolished in 1999.

Te idea behind the Mandatory Savings Fund, legislated in 1988, was to create a provident fund-like savings account for employees (Resmi Gazete 1988). Tis fund was supposed to boost Turkey's notoriously low savings rate, although observers questioned the likelihood of achieving this goal from the outset (OECD Economic Survey 1988, 55–56). Te fund applied to private and public sector employees and complemented the public pension system. Akin to social security contributions, employers and employees paid contributions based on the employee's gross wage. Tese contributions were invested in various ways. Employees would receive their savings on retirement, or their dependants would receive the savings in case of the employee's death. If employees wanted to draw on their savings earlier, they could do so to a limited extent.

Initially, labour unions proposed to shift savings from KEY to unemployment insurance, and policymakers considered both KEY and the Savings Fund as potential resources for unemployment insurance. However, given that contributions to the Savings Fund were like social security contributions, and given that the government aimed to cancel the Savings Fund anyway (Milliyet 1992a), it made sense to simply repurpose the Savings Fund contributions for unemployment insurance (Cumhuriyet 1993b). Te Savings Fund contribution rates were close to what the government had envisioned for unemployment insurance, with only employer contributions being 1 per cent higher. Curiously, the increased contributions apparently did not lead to adjustments in the generosity of unemployment insurance in the draft legislation.

Te DYP-SHP government eventually failed to legislate on unemployment insurance. Instead of Moğultay's proposal, a programme limited to state-owned enterprises was implemented. Te privatisation of stateowned enterprises had been a central issue on the political agenda since the 1980s. Te World Bank, which pushed for privatisation, proposed providing unemployment benefts for workers afected by privatisation. Te SHP wanted this programme to be part of a comprehensive unemployment insurance policy, yet it failed to convince the DYP, and thus it was implemented as a separate programme.

While the DYP-SHP government did not implement unemployment insurance, it successfully brought the issue onto the political agenda. By the mid-1990s, nearly all parties promised to adopt the policy. Ministers of Labour in successive governments announced plans to introduce unemployment insurance, with all drafts circulated in the media remaining close to Moğultay's draft bill (Öktem 2020a). However, none of the proposals came close to being implemented.

#### **3.2 The 1999 Unemployment Insurance Legislation**

In 1999, a centrist coalition government led by veteran leftist politician Bülent Ecevit came to power in the midst of a deep economic crisis. Ecevit had championed the cause of unemployment insurance already during his terms as Minister of Labour in the 1960s and Prime Minister in the 1970s. Te new Minister of Labour and Social Security, Yaşar Okuyan, from the centre-right ANAP also strongly supported unemployment insurance. ANAP had a history of opposing the policy and was generally seen as a pro-business party, but Okuyan, who grew up in a *gecekondu* (shanty town), aimed to become a pro-labour minister.3 Within three months the government managed to push unemployment insurance legislation through parliament.

<sup>3</sup> Interview with a former Minister of Labour and Social Security, conducted on 29 August 2018 in Izmir.

Okuyan managed to pass the legislation by making it part of a comprehensive social security reform. In the 1990s, the social security system had started to run increasing defcits, mainly due to the maturing of a pension system with ample opportunities for early retirement and also due to the mismanagement of funds. Plans to ensure the sustainability of the system through raising the retirement age had been on the agenda of virtually all governments in that period. However, this met ferce resistance, especially from labour unions. A plan to link unemployment insurance to a comprehensive social security reform had been frst brought forward in 1995. Yet, the reforms did not materialise, and therefore the defcits continued to increase (Yentürk 2018). Tis was seen as a major risk for the economy as a whole, and thus the perceived need for reform became ever more pressing (Özkan 2011).

Te government immediately began to work on a comprehensive social security reform proposal. Initially, however, it did not include unemployment insurance in the bill. Te reform featured a sharp increase in the retirement age, and therefore strong opposition from labour unions was widely anticipated. Okuyan presented his plan to include unemployment insurance in the bill only after labour unions voiced their criticism and in order to soften their stance. However, he failed to change the minds of labour with this manoeuvre. Moreover, by attaching the policy to the reform, he drew the ire of business, which opposed the new programme, unless severance pay was retrenched at the same time (Özbek 2006, 350–351; Öktem 2020a).

Te unemployment insurance legislation that the new government prepared deviated slightly from earlier draft bills. It marginally reduced the contribution period by changing the requirement for continuous contributions over the last six months to four months. However, the focus on workers in "stable and continuous" employment remained (Öktem 2020a, 23). Te beneft level was set at 50 per cent of the previous net wage. Importantly, while earlier proposals contained a beneft foor that guaranteed higher replacement rates for low-income earners, Okuyan's proposal introduced a beneft cap. Tis cap was set at the net minimum wage. Hence, medium- and high-income earners would have a lower replacement rate. Te instrument of a beneft cap had been demanded by business groups as early as 1992 but had apparently not played a role in previous draft laws. Finally, as in earlier proposals (Andaç 1999) benefciaries would also be eligible for active labour market policies—a hitherto neglected policy area in Turkey.

Te legislation also retained the plan to cancel the Mandatory Savings Fund contributions and repurpose them for unemployment insurance. Te Savings Fund had continued despite government plans to abolish it. Te 2 per cent employee contribution and the 3 per cent employer contribution made to the Mandatory Savings Fund would now be channelled to unemployment insurance and be complemented with a 2 per cent contribution by the state. Due to mismanagement, the Savings Fund was no longer seen as credible and legitimate. Raising the same contributions for unemployment insurance, which was a key demand of labour and a long-standing bipartisan development goal, would serve to legitimise the contributions. Furthermore, in contrast to the Savings Fund, unemployment insurance was not limited to medium and large enterprises but would also cover small companies. Terefore, in efect it constituted a new contribution for companies with fewer than ten employees. Civil servants, however, would not be covered by unemployment insurance, and thus the contributions made by the state as their employer would end.

Both business and labour voiced their concern about the reform. Yet, instead of seeking consensus with employers and employees, the Ecevit government simply chose to push through the reform against all odds. Te draft legislation was debated and passed by parliament in August 1999. Te parliamentary debates mainly focused on pension reform, with opposition MPs accusing the government of bowing to IMF demands for retrenchment. With regard to unemployment insurance, the opposition parties, DYP and the Islamist Virtue Party (*Fazilet Partisi*), voiced various criticisms. Tey argued that the policy would only protect labour market insiders at the expense of labour market outsiders, that unemployment insurance had to be coupled with severance pay reform and with job security provisions (a long-standing demand of business), that the public employment service lacked the capacity to administer the programme and that beneft levels were too low (Öktem 2020a).

Interestingly, the opposition also criticised the way that unemployment insurance would be fnanced. MPs from both the DYP and the Virtue Party argued that the workings of the Unemployment Insurance Fund (UIF), which would collect and manage the contributions, were insufciently specifed and not transparent. While the legislation foresaw the inclusion of employers' representatives and employee unions in the management of the UIF, it left the question of how to invest the assets to a regulation to be issued later. Considering the negative experiences with the Mandatory Savings Fund and other policies, the MPs repeatedly warned that it was likely that the fund would go bankrupt and would have to be bailed out by the government (TBMM 1999a, 531, 1999b, 279, 1999c, 355, 372 and 390, 1999d, 493–494 and 504).

Te opposition's warnings were countered by government MPs who argued that contributions would be more than sufcient to pay for the benefts. Minister of State Fikret Ünlü argued that in the frst twenty months after coming into force, a signifcant amount of money would be accumulated as no one would be eligible for benefts in this period. After this accumulation period, new contributions would continue to exceed benefts, Ünlü argued. Under a normal scenario, contributions would be nearly one and a half times the level of benefts (TBMM 1999c, 339). It is noteworthy that these calculations were made during a severe economic crisis, in which social security revenues tend to fall.

Yet, these arguments did not alleviate the opposition's fears. Te experiences with KEY and the Savings Fund were still very vivid. Furthermore, one key reason for the social security reform, of which unemployment insurance was but a small part, was the previous mismanagement of social insurance funds. Hence, opposition lawmakers feared mismanagement and a fund failure much more than they expected the massive accumulation of a surplus that would actually characterise the UIF.

To conclude, policymakers were overly cautious in designing the fscal aspects of the programme (Alper 2019). Partly to alleviate opposition against the policy and partly as a reaction to the negative experiences with the management of social insurance and other funds, policymakers successively decreased the generosity of the unemployment insurance draft bills throughout the 1990s. At the same time, planned contribution rates were even increased as the Mandatory Savings Fund contributions were repurposed. A surplus was thus all but inevitable.

## **4 The Transformation of Unemployment Insurance (2000–2019)**

#### **4.1 Surplus Accumulation (2000–2007)**

Unemployment insurance started in June 2000. Given that eligibility for benefts required a minimum twenty months of contribution period, the frst benefts could be paid out only in early 2002. During this period, the UIF thus started to accumulate money, reaching 1 per cent of GDP. Tis was entirely in line with what policymakers and bureaucrats had envisioned for the accumulation period. Although no long-term actuarial planning documents are available for the early years, one can compare the actual surplus generated by the UIF to the expected surplus that was posted in ofcial documents, such as the government's annual programmes. Furthermore, letters of intent written to the IMF (which had a standby agreement with Turkey) in December 2000 and May 2001 provide information on how the government expected the UIF to develop before unemployment benefts started to be paid out (Önal and Erçel 2000; Bahçeli et al. 2001).

Figure 4.1 confrms that from the outset policymakers expected the fund to create a constant surplus. In fact, before the programme was rolled out, they underestimated the surplus that the fund would generate. Interestingly, there appear to be inconsistencies between the annual plans and the letters of intent to the IMF, with the latter being far more conservative in their outlook. Yet, both the letters to the IMF and the annual plans agreed that the UIF would create a sizeable surplus. Over time, this would lead to the accumulation of an enormous amount of money as Fig. 4.2 illustrates.

Tus, it appears that policymakers were very much aware that the fund would generate a signifcant surplus, even in times of high unemployment. Beneft levels were low and qualifcation conditions were quite demanding, so beneft payments could simply not surpass contributions. Before unemployment benefts were paid out in 2002, the government realised that the surplus would be higher than necessary. In the context of a renewed economic crisis, it decided to reduce contribution rates by one percentage

**Fig. 4.1** Annual surplus of the unemployment insurance fund as a share of GDP. (Source: "Plan" refers to annual programmes and illustrates estimates that were made in October of the previous year. So, the 2001 plan for instance, which was written in October 2000, foresaw an annual surplus of 0.9 per cent of GDP. The actual surplus in 2001 was 1 per cent of GDP. "Actual" refers to the revenue and expenditure posted in monthly reports on the UIF published by the PES. "IMF 12/2000" refers to the letter of intent sent by Turkey to the IMF on 18 December 2000, while "IMF 5/2001" refers to the letter of intent sent by Turkey to the IMF on 3 May 2001)

point for employers, employees and the state. Tus, overall contributions were decreased from 7 to 4 per cent of the gross wage with the aim of decreasing non-wage employment costs. Yet, as Figs. 4.1 and 4.2 illustrate and as policymakers quickly realised, even these lowered contributions were far above what was necessary to run the programme. Despite lower contributions, the fund continued to generate a massive surplus.

So, what happened to the money accumulated in the fund? Ofcial reports indicate that the surplus was mostly used to buy bonds issued by the state. Hence, the fund bought public debt (Alper 2019). Te accumulation of an ever-larger amount of money thus clearly served a fscal function for policymakers. Unemployment insurance had a double beneft. Comparing the fund's surplus to the size of the budget defcit would give an approximation of the fund's importance for the fnancing of public debt. Figure 4.3 visualises this comparison by showing the size of the

**Fig. 4.2** Unemployment insurance fund reserves as a share of GDP. (Source: See Fig. 4.1)

**Fig. 4.3** Surplus of the unemployment insurance fund as a share of the budget defcit. (Source: Öktem (2020b))

surplus as a share of the budget defcit. Note that this is just an illustrative comparison and does not refect the amount of public debt actually fnanced by the UIF.

Furthermore, it is also important to note that in the early 2000s, Turkey was under IMF supervision. Te primary surplus of the public sector was monitored—and the UIF was included in the calculation of this surplus. Hence, the surplus in the UIF helped Turkey to fulfl the requirements of a stringent IMF programme. Tis would suggest that "fscal concerns" (Koreh 2017b) played an important role in the development of unemployment insurance.

In any case, in the early 2000s, both the coalition government led by Bülent Ecevit and the centre-right Justice and Development Party (*Adalet ve Kalkinma Partisi*, AKP) government that succeeded it in late 2002 were far more concerned with ensuring that the new programme was properly implemented than they were with making substantive policy changes.4 Te Public Employment Service (PES) that had been put in charge of administering the programme was widely seen as lacking the capacity to do so. To ensure that the PES had the capacity to manage unemployment insurance, it was comprehensively reformed, receiving EU support in the process (Bölükbasi and Ertugal 2013).

In terms of policy innovation, the Labour Law passed in 2003 by the new AKP government foresaw the creation of a Short-Time Work Compensation (*Kısa Çalışma Ödeneği*) programme that would pay worker's wages when companies reduced working hours during an economic crisis and a Wage Guarantee Fund (*Ücret Garanti Fonu*) that would pay wages for several months when companies went bankrupt. Both instruments were implemented in 2004 and started to pay out benefts from 2005 onwards. However, until the global economic crisis, these instruments remained extremely marginal. Tus, these instruments did not decrease the annual surplus, and the UIF was expected to surpass 5 per cent of GDP in the late 2000s.5

However, the fund's growing reserves as well as its continued surplus increasingly caught the attention of policymakers towards the mid-2000s. Opposition parties criticised the government for not supporting the

<sup>4</sup> Interview with a former Minister of Labour and Social Security, conducted on 10 January 2020 in Ankara.

<sup>5</sup>To compare, after a decade of uninterrupted growth and strong labour market development, reserves for unemployment insurance in Germany reached around 0.75 per cent of GDP in 2019. Te *Konjunkturrücklage* was 25.8 billion Euro, with GDP being around 3.449 trillion Euro in 2019 (Bundesagentur für Arbeit 2020).

In contrast, reserves of Estonia's Unemployment Insurance Fund reached around 3 per cent of GDP, with reserves being 963 million Euro and GDP being around 31.39 billion Euro in 2019 (ERR 2019).

unemployed through the fund. Emphasising that 97 per cent of the fund was invested in public debt, one MP argued in parliament in 2004 that "instead of supporting the unemployed, the fund has become a support for the state" (TBMM 2004, 651). Similarly, business was complaining that despite the large amount of money in the UIF, contribution rates were not decreased (Milliyet 2006, 2007a). If the fund continued to grow, business representatives feared, the likelihood of governments misusing the fund would increase (Birgün 2005).

Te Minister of Labour and Social Security, Murat Başesgioğlu, agreed in principle that the programme needed reform, arguing that eligibility conditions were too strict and benefts were too low. Similar concerns were also voiced in the Second General Assembly of the PES (İŞKUR 2005b). Başesgioğlu repeatedly promised a reform that would have increased benefts and loosened eligibility criteria (Tan and Karslıoğlu 2010, 77–82). However, it was not made part of the government's social security reform agenda. Unemployment insurance reform was neither mentioned in the white paper on social security reform (T.C. Başbakanlık 2005), nor to be found in the fnal reform legislation. As a result, the policy was left largely untouched in the AKP's frst term that ended in 2007. Instead of reducing contribution rates further or increasing programme generosity, the government tackled the surplus by starting to tax income on interest earned by the fund from 2006 onwards. However, on its own this did not have a large efect, and this policy was cancelled (more or less by accident) in 2008.

Overall, the frst years of the unemployment insurance programme were thus shaped by a steady surplus accumulation that resulted in increasing fund reserves. In this way, the policy was shaped by the double beneft mechanism: it was not just a social policy tool but also served to fnance public debt (Öktem 2020b).

#### **4.2 Economic Crisis Triggers Policy Change (2008)**

In 2008, the AKP government launched a major reform of unemployment insurance that set the path for the future direction of the policy. While the core statutes of the programme were virtually left untouched, a number of new instruments were added to the programme that gradually changed the logic of the policy.

A number of factors came together that facilitated policy change in 2008. First, economic growth had slowed. Turkey had grown quite fast after the severe economic crisis of 2001, but unemployment had remained stubbornly high. As the economy appeared to reach the end of a growth cycle, fears of a surge in unemployment mounted. In response, demands by business to make use of the UIF for employment creation increased. In early 2007, the Union of Chambers and Commodity Exchanges of Turkey (*Türkiye Odalar ve Borsalar Birliği*, TOBB) proposed the introduction of massive employer subsidies. Te UIF should pay the social security contributions for all new employment generated for two years, TOBB demanded (Milliyet 2007c). TOBB had made the same demand in 2005, in response to a government call for more employment creation by the private sector. At that time, it had pointed to the treasury as the fnancing source (Milliyet 2005). Two years later, it asked for the UIF to fund the policy. TOBB's proposal was supported by other business groups (Milliyet 2007e) but opposed by labour unions who argued that this was against the purpose of the UIF (Milliyet 2007d). Te AKP government rejected TOBB's proposal (Milliyet 2007b). Nevertheless, the episode showed that demands to make use of the fund were increasing.

Second, the accumulated surplus had surpassed what was warranted in terms of fscal prudence. Te fund had reached close to 4 per cent of GDP by 2005. At the same time the budget defcit had decreased, so that by 2006 the fund's annual surplus was close to the annual budget defcit, as Fig. 4.3 illustrates. In theory, the fund could thus have been used to fnance nearly all new debt. Te decreasing budget defcit also relieved the IMF pressure on the government. Furthermore, the IMF programme was to expire in mid-2008. According to some sources, the IMF had previously opposed any substantial changes in the programme due to concerns about the budget defcit (Hürriyet 2008).6 In this sense, from a

<sup>6</sup>However, note that according to Duval and Loungani (2019, 29) the IMF advised Turkey to *expand* the programme in 2007. So, possibly the IMF might have simply been used by the government as a scapegoat for not expanding the programme.

fscal perspective, using the money accumulated in the UIF became much more feasible by 2008.

Unemployment insurance reform became part of the ofcial government agenda in the AKP's second term starting in 2007. Te AKP planned to make the policy more generous and accessible, as well as expanding ALMP measures, such as vocational training (Neziroğlu and Yılmaz 2013). Strengthening ALMP had long been demanded by business (Hürriyet 2003) and in the PES' General Assembly Decisions, where social partners voiced their views (İŞKUR 2005a, 2005b, 2007). Te draft bill presented by the government in 2008 proposed numerous changes to the programme. It envisioned making the policy more generous, expanding ALMP (including the creation of employment subsidies) and using the UIF to fund non-labour market policies. Of these three proposals only the latter two were eventually legislated. Still, the failed plan to make unemployment insurance benefts more generous deserves attention.

Te draft legislation set unemployment beneft at 50 per cent of the gross wage instead of 50 per cent of the net wage. Tis would have resulted in substantially increased replacement rates of up to two-thirds of the net wage (Öktem 2020b). Te draft bill had been submitted to parliament by the prime minister, and thus it was no surprise that the respective parliamentary committees, which were dominated by government lawmakers, accepted the draft. Yet, unexpectedly, government lawmakers decided to veto their own proposal for higher benefts in parliament. In its place they passed a watered-down version of the relevant article in the legislation that provided only a minor beneft increase. Government lawmakers justifed this decision by stating that the beneft increase in the draft bill was too high. Te increase should be more "reasonable", they argued, to ensure that "unemployment benefts were not made attractive" for workers (TBMM 2008, 945). In other words, benefts should be kept low to ensure that they do not have a decommodifying efect on workers. Te government thus shied away from making the programme more generous.

Instead of increasing unemployment benefts, the reform strongly expanded ALMP. Tis was done by expanding access to ALMP from just unemployment insurance benefciaries to all unemployed registered with the PES. As a result, participation in ALMPs, such as vocational training or on-the-job-training, strongly increased. Crucially, the reform also included the creation of employer subsidies for the employment of young workers and women. Tis was received very positively by business (Dünya 2008), which had earlier demanded using the UIF for employer subsidies, as described above. Overall, ALMP expansion was thus in line with the demands of business. Moreover, it also conformed to policy ideas promoted as part of the Europeanisation of labour market policy. ALMP was one component of "fexicurity" that may be seen as the key concept behind the EU's labour market policy (Bolukbasi and Ertugal 2013). Hence, the expansion of ALMP was driven by two causal mechanisms: the mechanism of business-led reform and the mechanism of transnational cooperation.

In addition to strengthening ALMP, the reform diverted money from the fund to a development project in southeast Turkey. Tis conforms to the double beneft mechanism, as the UIF was used for non-social policy purposes. In fact, these non-social policy-related expenditures surpassed the fund's regular expenditures for passive and active labour market policies from 2008 to 2010, as can be seen in Fig. 4.4. Yet, this move was

**Fig. 4.4** Share of unemployment insurance fund expenditure by category. (Source: Öktem (2020b))

opposed by some section of business, which feared that this would mean a return to the era of "populist" policies when dedicated public funds were spent for unrelated policies (Cumhuriyet 2008). Te move only received the open support of a government-aligned labour union (Dünya 2008).7 Tis might explain why government was more cautious in using the UIF for totally unrelated policies afterwards.

To conclude, the 2008 reform triggered by the economic crisis constituted a signifcant change in the structure of Turkey's unemployment insurance programme. Before the reform, the programme's focus was nearly exclusively on providing unemployment benefts. Trough the reform, new instruments were layered on top of the classic unemployment insurance programme. Moreover, the reform laid out the path for the following years in which ever-new instruments were added to the programme. Tese new instruments were often as much to the beneft of employers as they helped employees. Tis led to opposition criticism that workers' funds were being used for employers. However, the government was unimpressed by this view. In parliament, the Minister of Labour and Social Security Faruk Çelik outlined the underlying argument for the new policy direction, showing that the government basically accepted the idea that what benefts employers would beneft society:

Te argument has been made that "with this reform you [i.e. the government] give resource to the employers". Respected friends, there is no longer any diference between employers and employees. Employees and employers are seen as inseparable. Tis is the point reached in industrial relations. Tis is the point reached today after the antagonistic perspective of the nineteenth and early twentieth century. (TBMM 2008, 877)

#### **4.3 Frequent Policy Change (2009–2019)**

Following the 2008 reform, the unemployment insurance programme saw a decade of frequent policy changes. Tese changes mostly focused on expanding ALMP and employer subsidies. Furthermore, new policies

<sup>7</sup>Even this labour union changed its stance one year later when the government announced new transfers from the UIF to the development project (Bianet 2009).

were increasingly devised as temporary instruments that were made permanent through subsequent changes. Table 4.1 summarises key policy changes made between 2009 and 2019. Let us look at the maze of reforms in the realm of employer subsidies and active labour market policies in more detail.8

Active labour market policies, such as vocational training and on-thejob-training, were massively expanded following the 2008 reform. Tis was done through frequent and often temporary policy changes. In mid-2009, the amount that the UIF could allocate to ALMP (expenditure limit) was temporarily increased for 2009 and 2010. In 2011, a comprehensive reform broadened the aim of ALMPs and increased the expenditure limit permanently. In 2015, the expenditure limit was increased twice. In early 2017 and late 2018, this expenditure limit increase was temporarily extended. Tus, although temporary in nature, the expenditure increases in practice became permanent. As a result, by 2015 expenditure on ALMP had eclipsed spending on unemployment benefts (see Fig. 4.4).

ALMP expansion broadly followed the demands of business (TISK 2010, 114, 2013). Te government justifed this expansion by pointing to expected growth in employment levels, as well as employability and productivity of employees (TBMM 2011, 40). However, policy changes were also increasingly linked to the election cycle. For instance, the increases in the ALMP expenditure limit in 2015 coincided with general election campaigns. Tis suggests that the changes are also a case of the outcompeting mechanism. Particularly instruments such as public works could be fexibly deployed to ofer tangible benefts to select parts of the electorate. Looking at benefciary statistics confrms this assessment. For instance, monthly benefciary numbers in the public works programme, which provides temporary, low-skilled and low-paid labour (Gün 2013), peaked before the 2017 referendum, the 2018 general elections and the 2019 local elections. Tus, by all accounts ALMP expansion went beyond

<sup>8</sup>For other signifcant changes, such as the expansion of the Short-Time Work Compensation or the not-yet implemented creation of an unemployment insurance scheme for self-employed, see Öktem (2020b).


**Table 4.1** Signifcant policy changes to unemployment insurance (2009–2019)

(*continued*)


**Table 4.1** (continued)

Source: Own presentation

following business demands for increasing employment to serve as part of the government's election agenda.

In addition to classic ALMP, employer subsidies were massively expanded. Tis expansion also occurred through frequent and mostly temporary policy changes. Employer subsidies had been introduced in 2008 and consisted of reduced contribution rates for employers hiring women and young employees within the following one year. Tis instrument was extended for one more year in early 2009. In late 2009, new employment subsidies were launched that applied to broader parts of the unemployed. In early 2011, new (temporary) employment subsidies for unemployed were created, which were temporarily extended in 2015. In early 2015, special employment subsidies for participants in on-the-jobtrainings conducted by the PES until the end of 2016 were launched. Tese subsidies were temporarily extended in late 2016 and 2017. In early 2017, new and more generous employment subsidies for unemployed were devised that covered hirings made in 2017. Finally, in early 2018, new employment subsidies were launched that were even more generous and applied to hirings made until the end of 2020. Tus, what started as a small, temporary crisis response in 2008 to encourage employers to hire women and young workers became an increasingly permanent policy that covered an ever-greater share of the labour costs for more and more new hirings. Tellingly, by 2018 employment subsidies had become spending item number one, as can be seen in Fig. 4.4.

Again, the increase in employer subsidies responded to demands made by business. Business organisations actively demanded that "employers who create employment should be subsidised" (TISK 2010, 111; TISK 2013, 104). Labour unions, on the other hand, opposed using the UIF for such employment subsidies, even if some unions were sympathetic to the idea of employer subsidies in principle (Dünya 2015). In a way, with the recent massive increases in subsidies, the government fnally returned to the plan that had been frst outlined by TOBB in the mid-2000s. TOBB had unsuccessfully demanded that all additional employment should be subsidised by the UIF. However, in the late 2010s employer subsidies basically resembled what TOBB had wanted all along. In fact, the latest employer subsidies were even formally announced as a cooperation between the government and TOBB (TOBB 2019). Furthermore, similar to ALMP, employer subsidies were timed to coincide with election campaigns with the goal of reducing unemployment just before the elections. Terefore, both the business-led reform mechanism and the outcompeting mechanism apply to developments in this area.

Beyond the expansion of ALMP and employer subsidies, the programme was also reformed to include a number of new instruments, which were often beyond the scope of classic labour market policies. In 2014 and 2015, payments for mining workers afected by some highprofle mining accidents were made. In 2015, some supports for employees and employers active in dangerous professions were created. In 2016, support payments for internships were launched. In addition to these mostly temporary measures, a kind of parental leave beneft was created, also in 2016, as a part of unemployment insurance. Te Part-Time Compensation (*Yarım Çalışma Ödeneği*) provides benefts to parents of new-born children who decide to work part-time for up to six months. Yet, the compensation has had a very limited reach so far. Overall, the level of expenditure devoted to these non-labour market policies never reached the scale of the 2008–2011 period. Furthermore, these new policies were not as far away from the original function of the UIF as the development project fnanced in 2008–2011. It appears that the criticism of using the UIF to fnance a development project (launched not just by labour unions, but also by business) made the government more cautious in this regard. Te idea that the UIF should not be used to fnance nonlabour market policy and that social partners should always be consulted has become a mainstay of demands made by TISK (2013, 106) and has been voiced in the General Assembly of the PES (İŞKUR 2009, 2011, 2013, 2015).

To conclude, the second decade of the unemployment insurance programme was shaped by frequent reforms that established new policies. Tese reforms can generally be seen as driven mainly by three causal mechanisms: business-led reform, transnational cooperation and outcompeting.

#### **5 Conclusion**

In this chapter, I explored the introduction and transformation of unemployment insurance in Turkey with a focus on the causal mechanisms behind the developments. I focused in particular on the double beneft of social insurance as a social policy programme and as a fnancing device, as this double beneft mechanism shaped the early phase of the programme and facilitated subsequent reforms.

Te analysis has shown that policymakers designed unemployment insurance overly cautiously with restrictive eligibility criteria and modest benefts. Tis was due to negative experiences with previous social insurance policies and related funds, and due to the dire economic and fscal situation the country was in when unemployment insurance legislation was passed in 1999. As a result, the Unemployment Insurance Fund was bound to generate an ever-increasing surplus, which would in all likelihood be at least partly invested in public debt.

Soon after the frst contributions were collected in 2000, policymakers became aware that the surplus would exceed their expectations. Tey reacted by nearly halving contribution rates, but this was not enough to create an actuarial balance and so the fund kept growing. Te money accumulated in the fund was used to buy public debt, and this helped the country meet the IMF's stringent fscal criteria. In the early phase, unemployment insurance was thus as much a state-fnancing tool as it was a labour market policy.

However, the fund's increasing reserves also increased demands for policy change. Tese demands were frst brushed aside by the government, but as the state's fscal situation improved, the need for the UIF as a fnancing tool decreased. Tus, when the global economic crisis hit Turkey in 2008, the government comprehensively reformed the legislation. Responding to demands by business that promised increasing employment and productivity, the government strengthened ALMP and introduced employer subsidies. Furthermore, it used part of the money accumulated in the fund for a development project. Tus, in line with the double beneft mechanism, the fund was used for non-social policy purposes.

In the decade after the 2008 reform, the government frequently amended the unemployment insurance legislation. Te focus of these reforms lay on ALMP and employer subsidies. As a result of repeated changes, the weight of the programme shifted frst from UI to ALMP and then to employer subsidies. In terms of policy content, these changes mainly responded to demands made by business. In this regard, the changes are an example of the business-led reform mechanism. In addition, the expansion of ALMP is also part of the Europeanisation of labour market policy. Terefore, the transnational cooperation mechanism also explains the reforms. However, the diferent reforms were also increasingly tied to the election cycle. Especially insofar as they ofered the opportunity to provide tangible benefts to voters, reforms were announced and implemented during election campaigns so as to "outcompete" opposition parties.

Overall, the fndings on the signifcance of the double beneft mechanism resonate with the literature on the fscal politics of the welfare state. Both "legitimation" and "surplus efects" (Koreh 2017b) can be observed in the development of unemployment insurance in Turkey. When the policy was introduced, contributions for the troubled and highly contested Mandatory Savings Fund were channelled to the new programme. Tus, the state was able to continue levying the same contributions from employers and employees by introducing unemployment insurance. Te surplus efect is also clear: it was the very size of the UIF that allowed the government to attach new instruments, such as employer subsidies, to the programme.

Yet, it is important to emphasise that the double beneft mechanism, by itself, does not explain the whole story. In particular, the direction of policy change can be better explained by other mechanisms. Te infuence of business on government policy, the desire to outcompete the opposition in the context of competitive (even if increasingly unfair) elections and the cooperation with transnational actors all played a crucial role in shaping the development of unemployment insurance in Turkey.

**Acknowledgment** Te author would like to thank Cansu Erdoğan, Tolga Bölükbaşı, Lutz Leisering, Tobias Böger, Emine Bademci, Johanna Kuhlmann and Frank Nullmeier for very valuable feedback.

## **References**


———. 1993b. İşsizlik sigortası biçimleniyor [Unemployment Insurance Takes Shape]. November, 12.

———. 2008. Kaşıkla verip kepçeyle alma [Giving with a Spoon and Taking with a Ladle]. May, 28.


———. 2015. İstihdam teşvikleri iyi, ama işçinin kesesinden yapılması yanlış [Employment Subsidies Are Good, but it Is Wrong that Tey Are Financed through Workers' Funds]. April, 7.


———. 2016. Aktif İstihdam Siyasetinin Gelişimi Açısından İşsizlik Sigortasının Dönüşümü [Te Transformation of Unemployment Insurance in Light of the Development of Active Labour Market Policy]. *Çalışma ve Toplum* 2016 (3): 1295–1319.


———. 2007. 4. Genel Kurul Çalışma Raporu [Fourth General Assembly Report]. https://media.iskur.gov.tr/13581/4-genel-kurul-raporu.pdf. Accessed January 19, 2021.


Herbert Obinger, and Christopher Pierson, 435–447. Oxford: Oxford University Press.

Koreh, Michal. 2017a. Te Political Economy of Social Insurance: Towards a Fiscal-Centred Framework. *Social Policy & Administration* 51 (1): 114–132.

———. 2017b. Te Fiscal Politics of Welfare State Expansion: Te Case of Social Insurance in Israel. *Journal of European Social Policy* 27 (2): 158–172.


———. 2007e. TİSK: Ek istihdam için primler Fon'dan alınsın [TISK: Contribution Payments for New Employment Should Be Taken from the Fund]. March, 29.

Münnich, Sascha. 2010. *Interessen und Ideen. Die Entstehung der Arbeitslosenversicherung in Deutschland und den USA* [Interests and Ideas. Te Emergence of Unemployment Insurance in Germany and the USA]. Frankfurt, New York: Campus.

Neziroğlu, Irfan, and Tuncer Yılmaz, eds. 2013. *Hükümetler – Programları ve Genel Kurul Görüşmeleri* [Governments – Teir Programs and Plenary Session Debates]. Ankara: TBMM Basımevi.

OECD. 1988. *OECD Economic Surveys: Turkey 1988*. Paris: OECD Publishing.


———. 2021. Turkey's Social Policy Response to Covid-19: Labor Market Reforms to Protect Employment. *Covid-19 Social Policy Response Series* 12/2021, Bremen: CRC 1342.


———. 2016. Welfare Regime Change – or Lack of Change – in Unemployment Compensation. *International Journal of Social Welfare* 25 (2): 126–135.

Paster, Tomas. 2013. Business and Welfare State Development: Why Did Employers Accept Social Reforms? *World Politics* 65 (3): 416–451.

———. 2017. Die Rolle der Arbeitgeber in der Sozialpolitik [Te Role of Employers in Social Policy]. In *Handbuch Arbeitgeber- und Wirtschaftsverbände in Deutschland*, ed. Wolfgang Schroeder and Bernhard Weßels, 487–514. Wiesbaden: Springer VS.

Resmi Gazete. 1986. Memurlar ve İşçiler ile Bunların Emeklilerine Konut Edindirme Yardımı Yapılması Hakkında Kanun [Law on Homeownership Support for Civil Servants, Workers and Civil Servant and Worker Retirees]. November, 22.

———. 1988. Çalışanların Tasarrufa Teşvik Edilmesi ve Bu Tasarrufarın Değerlendirilmesine Dair Kanun [Law on Encouraging Employees to Save and on Investing Tese Savings]. March, 18.

Schmitt, Carina, Hanna Lierse, Herbert Obinger, and Laura Seelkopf. 2015. Te Global Emergence of Social Protection: Explaining Social Security Legislation 1820–2013. *Politics & Society* 43 (4): 503–524.


TBMM. 1999a. T.B.M.M. Tutanak Dergisi. Dönem 21, Cilt: 10, Yasama Yılı: 1, 48 inci Birleşim, 12.8.1999.

———. 1999b. T.B.M.M. Tutanak Dergisi. Dönem 21, Cilt: 10, Yasama Yılı: 1, 50 nci Birleşim, 14.8.1999.

———. 1999c. T.B.M.M. Tutanak Dergisi. Dönem 21, Cilt: 10, Yasama Yılı: 1, 56 nci Birleşim, 24.8.1999.

———. 1999d. T.B.M.M. Tutanak Dergisi. Dönem 21, Cilt: 10, Yasama Yılı: 1, 57 nci Birleşim, 25.8.1999.

———. 2004. T.B.M.M. Tutanak Dergisi. Dönem 22, Cilt: 57, Yasama Yılı: 2, 117 nci Birleşim, 16.7.2004.

———. 2008. T.B.M.M. Tutanak Dergisi. Dönem 23, Cilt: 20, Yasama Yılı: 2. 104 üncü Birleşim, 14.5.2008.

———. 2011. Bazı Alacakların Yeniden Yapılandırılması ile Sosyal Sigortalar ve Genel Sağlık Sigortası Kanunu ve Diğer Bazı Kanun ve Kanun Hükmünde Kararnamelerde Değişiklik Yapılmasına Dair Kanun Tasarısı [Draft of the Law on Rescheduling Some Debt Payments and on Making Amendments to the Law on Social Insurance and General Health Insurance and to Some Other Laws and Decrees with the Power of Law]. Dönem: 23, Yasama Yılı: 5, S. Sayısı: 606.


———. 2013. 25. Genel Kurul Çalışma Raporu [Twenty-ffth General Assembly Work Report]. https://www.tisk.org.tr/yayin/25\_genel-kurulraporu-2013.pdf. Accessed January 19, 2021.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **5**

## **Causal Mechanisms in the Introduction of Mandatory Health Insurance in the Post-Soviet Region**

**Andreas Heinrich, Gulnaz Isabekova, and Heiko Pleines**

## **1 Introduction**

Te Soviet healthcare system envisioned comprehensive, qualifed medical care available to everyone in the population free of charge and organised as a single, unifed service provided by the state. At the time of its introduction in the 1920s, it constituted substantial progress even in a global comparison; however, beginning in the 1970s, Soviet healthcare

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy," funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Projektnummer 374666841—SFB 1342.

A. Heinrich (\*) • G. Isabekova • H. Pleines

Research Centre for East European Studies, University of Bremen, Bremen, Germany e-mail: heinrich@uni-bremen.de; gulnaz@uni-bremen.de; pleines@uni-bremen.de

was increasingly falling behind Western standards because it was highly underfunded (Heinrich 2022). Te severe economic crisis accompanying the break-up of the Soviet Union in 1991 further worsened the problems of the healthcare system. Within fve years, the newly independent states on the territory of the former Soviet Union had on average lost over half of their budget revenues. Accordingly, they faced high reform pressure concerning healthcare fnancing. Starting from this initial state, complex processes involving healthcare fnancing reform began to unfold in all post-Soviet countries. Te aim of this chapter is to highlight some of the key mechanisms that were driving these processes.

In general, there are three ways to fnance health systems (Wendt et al. 2009; Rothgang et al. 2010; Isabekova 2019). Te money can come from the state budget, as it did in the Soviet Union, which means the system is tax-funded and usually takes the form of a state-run national health service. At the opposite end, healthcare can be ofered on a market basis by private for-proft providers with customers paying for services received. Alternatively, the state can arrange for a mandatory health insurance, which fnances healthcare through a payroll tax, that is through contributions related to employment status and salary level, not based on individual health risks and needs.1

Tis chapter will briefy describe the Soviet system as the common starting point of all reforms of healthcare fnance in the former Soviet Union, thus presenting the initial state that was decisive for the diferent mechanisms to unfold. Te focus of the following empirical analysis is the introduction of mandatory health insurance as the most popular reform model. First, the causal mechanism leading to its legal establishment is elaborated. Creating mandatory health insurance on paper can be explained by the *resistance avoidance mechanism*, indicating that—on the one hand—there were clear expectations from a majority of the population and no organised opposition to mandatory health insurance, and—on the other hand—there was no real interest in this model on the part of most policymakers. Actual implementation of mandatory health insurance

<sup>1</sup>Tese three forms of healthcare fnancing are usually linked to broader institutional arrangements in healthcare systems. Post-Soviet countries combine separate features of various healthcare systems, making a lean typology inapplicable to these countries. Tat is why our study focuses on health fnancing as such.

schemes, then, faced severe problems. We identify two causal mechanisms which were at the core of these problems: Te frst is the *fght for state funding mechanism*, which explains the political confict surrounding the sourcing of additional funding required for healthcare. Te second mechanism that we identify as being crucial for the implementation problem is the *informalisation mechanism*, which indicates that medical personnel and patients often agree on informal payments, leading to unrecorded service provision. In several countries, this resulted in a failure to adjust policies to changing circumstances. We argue that against this background (partially) successful introduction of mandatory health insurance was only possible if the *reform supporter mechanism* occurred, meaning that a strong supporter in favour of the new healthcare model was created in the form of a central Mandatory Health Insurance Fund (MHIF).

In all post-Soviet countries where mandatory health insurance was formally introduced, the resistance avoidance mechanism was crucial, leading irreversibly to implementation problems, which can be further specifed by the fght for state funding mechanism and the informalisation mechanism. However, those countries that created a centralised MHIF clearly outperformed those that did not. Tis reform supporter mechanism is illustrated for two cases with a central Fund, namely Kyrgyzstan and Moldova, and three cases with decentralised management of mandatory health insurance, namely Georgia, Kazakhstan and Russia. Te conclusion summarises the complex causal mechanisms that explain the outcomes of the introduction of mandatory health insurance.

## **2 Initial State and Reform Decisions: The Collapse of the Soviet Healthcare System**

In the Soviet Union, the healthcare system was an integral part of the planned economy. Tis system, named after Nikolai Semashko, the frst Soviet minister for public health who promoted its introduction in the 1920s, was fully fnanced and organised by the central government. Soviet fve-year plans for healthcare placed strong emphasis on quantitative targets based on input (i.e., number of physicians and hospital beds), while at the same time the health system was chronically underfunded, resulting in a poor quality of services. Tese inbuilt weaknesses of the Semashko system started becoming visible in the 1970s: health conditions deteriorated with stagnating life expectancy and high mortality rates, and disparities in health status and outcomes among the ffteen Soviet republics became striking (Heinrich 2022). In reaction to this situation, patients used informal relations, namely personal contacts and presents/bribes, to obtain preferential treatment in the healthcare system. Te provision of health services thus became a part of the Soviet Union's "economy of favours" (Ensor and Savelyeva 1998; Ledeneva 1998).

In the mid-1980s, the Soviet leadership started to discuss reforming the Semashko system through the introduction of quasi-market elements. As a result, several pilot regions introduced mandatory health insurance schemes in the late 1980s to improve the fnancial situation of their regional healthcare sectors. Te break-up of the Soviet Union in 1991, however, prevented the wider expansion of these pilot projects and eventually an introduction at the national level (Twigg 1998, 585–86).

Te ffteen newly independent states of the former Soviet Union thus inherited an unreformed Semashko system. Simultaneously, they faced a deep economic crisis, which led to steep declines in state revenues. Whereas this reform pressure favoured a leaner or privatised health system, the large majority of people in all these countries viewed a functioning health system as a key responsibility of state governance. At the same time, the population was used to employing informal relations and bribery in order to get access to state-run services. Tese conditions impacted not only the decision to reform, but also the implementation of reforms. Tis becomes clearly visible in a retrospective analysis.

Looking back at the longer trend of healthcare fnancing reform in the former Soviet Union, three groups of countries have emerged. Te frst group still runs a national health service. Four out of the ffteen post-Soviet countries (Azerbaijan, Belarus, Uzbekistan and Turkmenistan) made the conscious decision to keep the Soviet system largely intact. In a further two countries, Latvia and Ukraine, several reform attempts failed because of political deadlock, thus leaving these countries with a partly reformed national health service. In all six countries in this group, however, benefts were reduced drastically (if to varying degrees) due to a lack of fnancial resources. As a result, out-of-pocket payments contribute substantially to health fnancing in these countries. In Azerbaijan and Turkmenistan, the national health service exists mostly on paper, with citizens paying individually for over two-thirds of their health expenses, as Table 5.1 indicates.

In a small group, comprising just three countries, out-of-pocket payments ofcially dominate healthcare fnancing. However, only Georgia made a conscious political decision to introduce a market-based health system in the wake of neoliberal economic reforms in the early 2000s. In the other two countries, Armenia and Tajikistan, a deep economic crisis combined with armed conficts led to a sharp deterioration of the national health system. Consequently, these states restricted the promise of free healthcare to small groups of the population and a large share of health service provision is now ofcially fnanced directly by patients.

Te largest group of countries has introduced mandatory health insurance schemes. Actual coverage of services has to varying degrees been restricted by fnancial constraints, again often leading to a substantial role for out-of-pocket payments. In addition to the fve countries which introduced mandatory health insurance permanently in the 1990s (Estonia, Kyrgyzstan, Lithuania, Moldova and Russia), Georgia and Kazakhstan also introduced mandatory health insurance but later cancelled it. Kazakhstan returned to mandatory health insurance in the mid-2010s. Te relevance of this health insurance scheme in post-Soviet reform debates is also demonstrated by the fact that in the 2010s, Armenia, Azerbaijan, Latvia, Tajikistan and Ukraine were discussing its introduction.



2005–09; in Kyrgyzstan for 5% in 2000–04, 9% in 2005–09, 6% in 2010–14 as well as 2015–17; in Moldova for 5% in 2005–09, 10% in 2010–14 and 5% in 2015–17; in Tajikistan for 5% in 2000–04, 11% in 2005–09, 9% in 2010–14 and 7% in 2015–17. Voluntary private health insurance is only of very limited relevance in the post-Soviet region (Dieleman et al. 2017). Source: Own calculations based on World Bank "World Development Indicators," (Indicators used: Domestic general government health expenditure (% of current health expenditure), Out-of-pocket expenditure (% of current health expenditure), External health expenditure (% of current health expenditure)) available at: < https://databank.worldbank.org/

source/world-development-indicators#> (last updated: 15 October 2020, accessed December 12, 2020).

5% are listed) in Armenia it accounted for 10% in 2000–04 and 9% in 2005–09; in Georgia for 7% in 2000–04 and 6% in

### **3 Introducing Mandatory Health Insurance: The Resistance Avoidance Mechanism**

Te creation of new state structures after the break-up of the Soviet Union included the need for the formal reorganisation of the inherited Semashko system on the territory of the newly independent states. After the end of the socialist planned economies in Central and Eastern Europe, neoliberal concepts, summarised in the so-called Washington Consensus, and established models from OECD countries were often treated as standard solutions. Pension reform in the post-Soviet region is a case in point: International organisations, most prominently the World Bank, gained importance on this issue because they presented a clear model as a "blueprint" for reform (Orenstein 2008; Kaasch 2013). It has been argued that part of the popularity of this approach was a process of "competitive signalling," in which numerous post-Soviet countries used the adoption of sometimes extreme neoliberal economic reforms to impress foreign investors (Appel and Orenstein 2016, 2018).

In the area of healthcare reform, however, international organisations did not provide such a "blueprint": "Many international, bilateral governmental and non-governmental agencies and groups provide technical and fnancial assistance to the health sector in Eastern Europe. Most of these focus on specifc projects; few address broader health policies and strategies" (Nelson 2001, 259). Accordingly, the reform of healthcare fnancing in the post-Soviet region remained in the realm of domestic politics.

In order to understand the domestic politics of healthcare reforms in the former Soviet Union, it is important to consider the diference between policies by design and efects of reform blockades. In all three groups of countries described above, there are those where coherent reforms have been implemented. Belarus is a clear case of a conscious decision to preserve the Soviet national health system as far as possible, based on the logic of the honouring one's legacy mechanism and fnanced with economic subsidies from Russia (Cook 2007, 204–06). Ukraine, on the other hand, still runs a national health system because repeated attempts at reform have failed. While Georgia opted for a market-based healthcare system as part of radical neoliberal reforms aiming at "competitive signalling" to foreign investors in the 2000s (Appel and Orenstein 2018), the governments of Armenia and Tajikistan simply failed to fnance comprehensive healthcare, forcing the population to rely on its own fnancial resources (i.e., out-of-pocket payments).

Tis diference is also visible in the third group of countries, which is analysed here: Estonia adopted a radical reform of the healthcare system in the frst half of the 1990s when mandatory health insurance was introduced, with only incremental change since then (Jesse 2008). Kazakhstan, on the other hand, has been moving back and forth, introducing country-wide mandatory health insurance already for the second time in the mid-2010s.

At the same time, the dominant reform model in the post-Soviet region was clearly payroll-based mandatory health insurance. In the 1990s, only four countries in the region (Azerbaijan, Belarus, Uzbekistan and Turkmenistan) made an explicit political decision in favour of an alternative model. Te resistance avoidance mechanism is able to explain the introduction of mandatory health insurance in all post-Soviet countries that adopted this model. Te relative attractiveness of the mandatory health insurance model was mainly due to two popular perceptions.

First, the drastic decline in state revenues in combination with the bad image of the planned economy, of which the Semashko system had been an integral part, created pressure to move away from the national health service model. At the same time, mandatory health insurance provides the chance to obtain additional funding that is earmarked for health. Te thinking went that a payroll tax would generate a stable source of funding because in the past the payroll of state enterprises had been large and easy to tax (Ensor and Tompson 1998, 208–09).

Second, socialisation in the Soviet system had created strong preferences among the populace for a comprehensive welfare state. In the representative Life in Transition Surveys, which capture public opinion in the post-Soviet states in 2006 and 2010, when asked about the frst priority for additional state expenditure, healthcare was the most frequent answer; on average, over a third of respondents gave this answer. If the second priority is considered as well, over two-thirds of respondents were in favour of increased state spending on healthcare.2 Consequently, purely market-based solutions were highly unpopular in the post-Soviet region.

With the Semashko system discredited and private insurance being very unpopular, mandatory health insurance was an obvious choice as a reform model.3 Where this model gained the support of key political decision-makers, it was adopted swiftly. Tis logic worked especially well in Central Eastern Europe including two Baltic States (Estonia and Lithuania),4 where universal mandatory health insurance was adopted early on, often with strong support from policy entrepreneurs and interest groups with backgrounds in the medical professions (see Preker et al. 2002 and Chap. 9).

However, in the post-Soviet region (with the exception of Estonia and Lithuania) there were no strong forces promoting the introduction of mandatory health insurance. As Cook (2007, 55–56) elaborates on the case of Russia:

Te welfare state that Russia inherited was a top-down creation, established under conditions of society's political exclusion. […] Trade unions served mainly as administrators rather than demanders of welfare. Te professional groups that provided social services—doctors, educators, and so forth—belonged to mandatory associations but had no experience with organizational autonomy and little with interest articulation.

Instead, policymakers in the state executive and legislature simply adopted related laws to get the healthcare fnance reform of the table. Tis development has to be understood in the context of the former Soviet Union in the early 1990s. Many policymakers were inexperienced or had a background in the Communist Party. Tey had to pass fundamental reforms addressing nearly all aspects of political, economic and societal afairs. At the same time, they were used to "policies of

<sup>2</sup>Data available at https://www.ebrd.com/what-we-do/economic-research-and-data/data/lits.html (accessed May 30, 2020).

<sup>3</sup>A concise summary of this reasoning accompanied by a critical assessment is ofered in Ensor and Tompson (1998).

<sup>4</sup>Latvia is an exception as reforms to introduce mandatory health insurance were not supported by a stable majority in parliament.

promises" from Soviet propaganda as well as overly optimistic neoliberal reform expectations. As a result, there was a tendency to adopt laws frst and check their content later. Cook (2007, 77) quotes "one of the architects" of the introduction of mandatory health insurance in Russia with the statement that "there were only about 12 people at the time who really understood the implications [of the health insurance law,] so it was easy to get it through the legislature." Adopting a law on mandatory health insurance was thus the path of least resistance.

However, this resistance avoidance mechanism, which explains the formal introduction of mandatory health insurance, has important implications for its actual implementation. As no strong political actor promoted its introduction, none was really interested in making it a success. In Russia, for example, parliament did not pass any legislation to regulate the private medical practices which had been introduced with the mandatory health insurance law, thus leaving them in legal limbo (Cook 2007, 27). In an even stronger case of initial neglect, Moldova passed the law on mandatory health insurance already in 1998, but only created the MHIF in 2001 and fnally implemented the health insurance system in 2004. Kazakhstan created, abolished and reintroduced mandatory health insurance within two decades.

## **4 (Not) Running Mandatory Health Insurance: The Fight for State Funding Mechanism and the Informalisation Mechanism**

Te following analysis of the actual implementation will focus on fve countries which introduced mandatory health insurance in the 1990s, namely Georgia, Kazakhstan, Kyrgyzstan, Moldova and Russia. Tey faced substantial challenges. Two major problems were related to, frst, the fght for funding between diferent state and public actors and, second, the practice of healthcare personnel and patients to engage in informal interactions, often related to corruption.

#### **5 Causal Mechanisms in the Introduction of Mandatory Health…**

We argue later in this chapter that partial success of mandatory health insurance was only possible in cases where a strong supporter in favour of the new healthcare model was created in the form of a central MHIF. Te decision about centralised or decentralised fnancial management thus turns out to have enormous implications for the further development of healthcare reform. Only in countries which opted for a strong central MHIF, namely Kyrgyzstan and Moldova, did the mandatory health insurance reform "lead to demonstrable positive achievements" (Kutzin et al. 2009, 295).

#### **4.1 Fight for State Funding Mechanism**

With the shift to a mandatory health insurance scheme, insurance contributions became linked to employment (Dixon et al. 2004, 59). However, the transition in the former Soviet Union was accompanied by a severe economic recession which led to a huge increase in unemployment and rising informal employment, which is difcult to tax. Countries with little formal employment and a large informal economy, like all post-Soviet countries apart from the three Baltic states, were faced with the problem that insurance contributions5 were not viable. Tus, general state revenues often continued to play a signifcant role in healthcare funding, despite the switch to mandatory health insurance. Tis led to political confict about the sourcing of this additional funding.

All former Soviet Union countries that enacted mandatory health insurance in the 1990s initially opted for a decentralised system, which was meant to counter the over-centralisation of the planned economy and to bring the organisation of healthcare closer to the citizens. At the same time, many national governments found it hard to resist regional and local aspirations. Russia's President Boris Yeltsin, for example, famously remarked that the regions should "take as much sovereignty as

<sup>5</sup>Despite inter-country variation, there are some broad similarities: Most post-Soviet countries collected contributions to mandatory health insurance from enterprises in the form of a 2–4% payroll tax. Contributions were mostly shared between employer and employees, although in Russia only the employer paid (Ensor and Tompson 1998, 206).

they can swallow" (Teague 2002, 207). Tis also relieved fnancial pressure on the national budgets of the respective states.

Tus, in Russia, regional and local governments were supposed to contribute to the mandatory health insurance on behalf of the non-working population. "Initially, however, no national norms for these budgetary contributions were adopted. Tis led to wide variation in practices and, by 1997, twenty-seven out of the country's eighty-eight regions were making no compulsory health insurance contributions from their budgets" (Sheiman et al. 2010, 102). Contrary to reform plans, local governments rarely redirected all of their health revenues to the territorial MHIFs, instead continuing to fnance their healthcare facilities directly. In reaction to this problem, payroll tax collection for the mandatory health insurance was shifted in 2001 from the federal and territorial MHIFs to the general tax authority (Sheiman et al. 2010, 109). However, this was not enough to solve the related problems. A decade later, implementation of mandatory health insurance still difered strongly across Russia, suggesting a high dependence on the commitment of regional authorities (Popovich et al. 2011, 146).

Georgia had a similar experience. In 1995, it amended the constitution and in a "devolution type of reform" consigned the responsibility for healthcare to the regional authorities (Sehngelia et al. 2016, 350). In 1997, the Law on Local Self-governance "set the responsibility for funding health care facilities, as well as planning and implementing local health care programmes at the municipal level" (Chanturidze et al. 2009, 18). To ensure common standards across the country, the Georgian government designed a Basic Beneft Package to cover standard healthcare needs of the population, jointly funded by the mandatory health insurance, municipality health funds and the Ministry of Health (Chanturidze et al. 2009, 16–17; Sehngelia et al. 2016, 350).

However,

the government was unable to meet its revenue and expenditure targets, which led to (1) across-the-board expenditure cuts in the 1998, 1999 and 2000 budgets; (2) the continued accumulation of large arrears in reimbursements for health facilities, wages and pensions; and (3) increasing reliance on private out-of-pocket payments to fnance health care. (Chanturidze et al. 2009, 17)

In 2003, revenues generated from mandatory payroll contributions accounted for only 5% of total health expenditure. Due to these shortcomings, the Georgian mandatory health insurance system was abandoned in 2004 (Chanturidze et al. 2009, 17). In a radical shift towards a market-based healthcare system, the government decided to recentralise the system and to develop a private health insurance scheme along with the privatisation of public healthcare facilities (Sehngelia et al. 2016, 349–50).

Te fght for healthcare funding developed along the same lines in Kazakhstan, which introduced mandatory health insurance in 1996. Some regions withheld almost all their required transfers for healthcare (Ensor and Tompson 1998, 214). Tere was also a lack of coordination with the local government authorities, as they withdrew from health system funding. As a result, the mandatory health insurance system was cancelled by the end of 1998 after failing to meet fnancial commitments to healthcare providers (Sheiman et al. 2010, 103–05).

In summary, the fght for state funding mechanism is relevant for countries where a high level of unemployment and/or informal employment render a payroll tax insufcient to fnance a developed healthcare system. In such a situation, the social insurance-based system needs substantial fnancial support from the state budget. If the system is decentralised and local or regional authorities are able to reject or boycott such support, the results are arrears, underfnancing and potentially—as in the cases of Georgia and Kazakhstan—the breakdown of the system of mandatory health insurance.

#### **4.2 Informalisation Mechanism**

Te lack of sufcient healthcare funding in all post-Soviet countries has provoked a discussion about reducing coverage and limiting benefts packages. However, in the regional context both citizens and politicians "see comprehensive and free health care as a right, and are not ready to accept cuts in benefts. Providers, who depend on the income, similarly oppose it" (Dixon et al. 2004, 66). As a result, healthcare systems are based on promises which cannot be realised in practice.

#### **154 A. Heinrich et al.**

Concerning coverage, a major issue is people who do not directly make insurance payments. For instance,

anecdotal reports from Kazakhstan […] indicate that those who do not pay insurance contributions directly (and there are signifcant numbers in the region, such as the self-employed, those in small informal businesses, farmers, the unemployed, students and pensioners) are treated as "uninsured." […] Tis highlights the importance of distinguishing between being theoretically insured and "functionally insured" (that is, actually having physical and fnancial access to needed care at an afordable cost). (Dixon et al. 2004, 64)

Concerning benefts, the World Bank has encouraged the development of a basic package of services which excludes the bulk of inpatient treatments provided by the health systems of the post-Soviet region. While most governments have recognised the need to defne basic services, few have been able to do so. In Kazakhstan, "a basic package, funded by insurance, and a guaranteed package, funded from the budget were developed. But together they account for almost all services that could be provided. Some countries, such as Russia, have excluded cosmetic surgery and dentistry" (Ensor and Tompson 1998, 212).

As a result, many formally free healthcare services are only available after some additional, usually informal out-of-pocket payments have been made. Tese payments are often perceived as bribes by those involved as well as by the legal regulations in the respective countries. In contrast to presents, these out-of-pocket payments have to be made in advance and are directly linked to the performance of specifc services (Lewis 2002; Stepurko et al. 2015). In this context, Cook (2007, 10) speaks of "informalized welfare states that were to a signifcant extent governed neither by state authorities nor by market principles."6

A representative study of corruption in everyday life in Russia, conducted in 2001 and 2005, found healthcare to be the most corrupt area with 80% (2001) and 62% (2005) of the respondents stating their willingness to ofer bribes. Te authors of the study estimated that this

<sup>6</sup>For a systematic discussion of the impact of diferent forms of corruption on the healthcare system, see Vian 2020.

amounted to annual informal payments of about US\$600mn and US\$400mn, respectively (INDEM 2005).

A coordinated attempt to reduce informal out-of-pocket payments has been made in Kyrgyzstan. Te country has introduced a State Guaranteed Beneft Package (SGBP). Te SGBP clearly defnes the rights and obligations of patients and the state with regard to provision of health services and clarifes the entitlements of diferent population groups. (Ibraimova et al. 2011, 15; World Bank 2007, 43–44, 292; World Bank 2014, 3–4). As a result, "according to the Kyrgyz Integrated Household Surveys between 2000 and 2009, the proportion of the population that reported that they needed health care but did not seek it because it was too expensive or too far away fell signifcantly, from 11.2% in 2000 to 4.4% in 2009" (Ibraimova et al. 2011, 140). Tough this is substantial progress, it also highlights the persistence of informality.

In summary, the informalisation mechanism unfolds similarly in most post-Soviet countries, though with varying degrees of importance. As formal funding is insufcient to provide the legally guaranteed coverage and benefts (and as some people lack access to additional healthcare services they desire), medical personnel and patients often agree on informal payments, leading to unrecorded service provision. Tese out-of-pocket payments are partly used to fund actual services, while another part, more similar to a bribe, increases the income of individual medical professionals. As these informal payments provide patients with additional services and medical personnel with additional income, they are hard to abolish. At the same time, they decrease reform pressure on the formal healthcare system, as many patients can fnd a way to get along without it. However, they also increase social inequality as some patients cannot aford out-ofpocket payments and receive reduced, delayed or inferior health services. Moreover, they make the system less reliable as real costs and quality of health provision are hard to predict and compare.

## **5 Making Mandatory Health Insurance Work: The Reform Supporter Mechanism**

A key feature of social insurance is the existence of institutions with a certain organisational autonomy towards the state, including fnances that are managed separately from the general state budget (see also Chap. 14). Accordingly, together with the introduction of mandatory health insurance an MHIF is established to administer the insurance scheme. Te decision about the competencies and powers granted to the MHIF is crucial for the implementation of mandatory health insurance. Tis can be illustrated by a comparison of the countries that introduced a central Fund, namely Kyrgyzstan and Moldova, and those that opted for decentralised management of mandatory health insurance, like Georgia, Kazakhstan and Russia.

Where the MHIF is weak, it has been torn apart, as the fght for state funding mechanism outlined above explains. Russia, for example, opted for a decentralised MHIF in line with its federal political order. In the case of healthcare reform, this meant that the switch to mandatory health insurance was to a large degree left to the regions, thus becoming dependent on the attitudes of regional authorities. As a result, by the mid-1990s "the federal ministry of health had almost lost control over the health care system" (Ensor and Tompson 1998, 215). Regional fragmentation produced remarkable variation in access to healthcare (Borisova 2011). As of 2018, the average life expectancy at the regional level ranged from 64 years in Chukotia to 78 years in Moscow, while the country-wide average stood at 73 years. In summary, "despite nearly 30 years of post-Soviet reforms, Russia has not developed an efective system of mandatory social insurance" (Gontmakher 2019, 447).

In Georgia and Kazakhstan, similar problems even led to the cancellation of mandatory health insurance reforms, as described above. It has been argued that

a central factor in the failure of reforms in some countries has been the lack of capacity of health ministries to adopt [related] new functions. Two key contributory factors to this failure are the rapid turnover of public sector employees migrating to better paid jobs in the private sector, and the chaotic decentralization of authority to health insurance agencies and/or regions that has left ministries with accountability for implementation but little authority or capacity to drive reforms forward. (Figueras et al. 2004, 30)

While these often-cited factors are important elements of failure, we argue that they are not the root cause. In Kyrgyzstan, which has been categorised as a fragile state due to violent power struggles and "revolutions," and which is marked by exceptionally high staf turnover in the Ministry of Health (Isabekova and Pleines 2021), mandatory health insurance has been a relative success story.

Te core reason for the failure of mandatory health insurance in the post-Soviet region is the lack of any relevant political force or interest group strongly supporting it. As elaborated above, the resistance avoidance mechanism can explain the formal decision to introduce mandatory health insurance, but does not explain to what extent mandatory health insurance is also implemented. Successful implementation of mandatory health insurance is only possible if a strong supporter for the new system emerges.

As the cases of Georgia, Kazakhstan and Russia show, implementation of mandatory health insurance is politically controversial, while public expectations are unrealistically high. It is, therefore, highly unlikely that any political force will "burn its fngers" by supporting it. At the same time, there are no relevant interest groups that would beneft from a health system with general rules that is chronically underfunded (as opposed to individualised informal bargaining). For an MHIF, on the other hand, supporting the health system is the only reason for its existence. A strong MHIF can thus become the motor for successful mandatory health insurance implementation, as can be illustrated by the cases of Kyrgyzstan and Moldova.

When Kyrgyzstan introduced mandatory health insurance in 1997, it started with a decentralised MHIF like Georgia, Kazakhstan and Russia. Initially, the MHIF in Kyrgyzstan did not collect revenues itself; instead, it received transfers from the Social Fund and the federal budget on behalf of defned categories of "insured" persons. "Owing to chronic defcits across the social security system, however, these transfers were in many cases not made or delayed, and substantial arrears built up. Health insurance funds were often obliged to provide health services to the whole population, despite the lack of contributory income" (Dixon et al. 2004, 61).

In this situation, Kyrgyzstan opted for a substantial reform. In the early 2000s, the country shifted to a single-purchaser model by integrating general budget revenues and mandatory health insurance contributions (Dixon et al. 2004, 75). In 2001, the pooling of budget funds and insurance contributions was established in two pilot regions.7 Each year, additional regions established regional health fnancing pools until the entire country was covered by 2005. Te regional pools were merged at the national level in 2006, ending both the fragmentation of health resources and the duplication of services (Giufrida et al. 2013, 5; World Bank 2014, 4).

Te single-payer reform in 2001 changed the role of local governments "from direct funders and controllers of health facilities to funding sources for the regional MHIFs" (Sheiman et al. 2010, 107). Te frst result was similar to developments in Russia and Kazakhstan. Local authorities feared a loss of control over their local health system, which led them to divert funds for other uses. Tis, in turn, caused a decline in budget allocations for the health system (Sheiman et al. 2010, 107). As a result, opponents "managed to block key activities, most importantly the restructuring of tertiary hospital facilities in the two major cities" (World Bank 2008, xii). Resistance to eforts to rationalise the health sector came also from medical staf fearing job losses and from patients in catchment areas of facilities being closed (World Bank 2008, 10).

In this situation, it can clearly be seen that the MHIF had established itself as an important political player:

During 2002, the situation became so grave (with under-payment of providers and growing informal payments) that the MHIF—backed by international partners supporting the health reform process—pushed the government to amend an existing agreement with the IMF so that, begin-

<sup>7</sup>Te idea was to introduce the reform gradually, expanding its territorial spread step by step, not to test diferent concepts. Te concept was implemented as originally designed without any reference to "experimental competition."

ning in October 2002, the Social Fund was required to remain current in its cash transfers to the MHIF (i.e., no new arrears would be allowed). (…) In the Kyrgyz context, it was only possible to overcome this problem by engaging powerful external agencies. (Sheiman et al. 2010, 106)

If Kyrgyzstan's MHIF had not engaged external agencies, funding practices would not have changed.

Tus, the creation of a strong, centralised MHIF put Kyrgyzstan's healthcare system on a diferent path. Te MHIF was established as a parastatal organisation under the Ministry of Health outside the core public bureaucracy. As a parastatal entity, it was freed from the Soviet-era input-based budgeting mechanisms. Tis created incentives for downsizing and savings (Jakab and Manjieva 2008, 297). Tis development led to a more equal distribution of funds between regions, improved fnancial protection and laid the foundation for a considerable reduction of informal payments, as outlined above. In summary, "the new institutional structure created a platform for profound reforms in pooling and purchasing that led to demonstrable gains in equity and efciency in the Kyrgyz health system" (Sheiman et al. 2010, 107).

Similarly, in Moldova the MHIF was the key instrument in the health fnancing reform. It became "the single institution responsible for the pooling and managing of funds; it purchases services by contracting with autonomous health care providers. In this way, the 2004 introduction of mandatory health insurance represents a fundamental shift away from the fnancing mechanisms used in the Semashko system" (Turcanu et al. 2012, 16–17).

In summary, the creation of a strong reform supporter in the form of an MHIF can at least partly neutralise the implementation problems related to the fght for state funding mechanism and the informalisation mechanism. In Kyrgyzstan and Moldova, the reform supporter mechanism led to a more equal distribution of funds between regions. Moreover, it allowed for the consolidation of healthcare fnances.

## **6 Conclusion**

For the ffteen states which became independent after the dissolution of the Soviet Union in 1991, the creation of new state structures included the need for the formal reorganisation of the national part of the inherited Semashko healthcare system. Te drastic decline in state revenues in combination with the bad image of the planned economy, of which the Semashko system had been an integral part, created pressure to move away from the national health service model. At the same time, socialisation in the Soviet system had created strong preferences for a comprehensive welfare state among the population. As a result, purely market-based solutions were highly unpopular.

With the Semashko system discredited and private insurance being very unpopular, mandatory health insurance was an obvious choice as a reform model. As a result, in most post-Soviet countries mandatory health insurance was not adopted because it was promoted by policy entrepreneurs or interest groups, but simply because it looked like the least controversial option. Te resistance avoidance mechanism can explain why nine of the ffteen former Soviet Union countries have introduced mandatory health insurance at some point after independence.

However, as there was no infuential promoter of mandatory health insurance, implementation did not receive much attention in these countries. In some of them, governments did not bother at all about the implementation of mandatory health insurance schemes for several years. In others, responsibility was passed on to the regional and local levels. Tis was not enough to overcome the substantial challenges of implementation, as the fght for state funding mechanism and the informalisation mechanism show.

We argue that implementation is only successful when a fourth mechanism unfolds, namely the reform supporter mechanism. If the national government continues along the path of least resistance, the mandatory health insurance is not able to function properly. In countries with weak governance systems, like many in the post-Soviet region, this means that the national level largely loses its coordinative function. Instead, regional governments take over responsibility for public healthcare. Tis immediately leads to diferences in the level of healthcare fnancing. In the longer run, it also leads to divergent institutional reforms and (especially in larger countries with stronger regional socio-economic disparities) to considerable diferences in healthcare provision and health outcomes. Moreover, in the case of increasing fnancial defcits, it may lead to the failure and abolishment of mandatory health insurance.

If, on the other hand, the government makes a signifcant attempt at reform and manages to (re)centralise healthcare competencies into a coherent national MHIF, this creates a strong supporter for country-wide healthcare development. As the MHIF is not a direct part of the government, but a parastatal entity, it has more leeway and can build strategic coalitions. As occurred in Kyrgyzstan and Moldova, the MHIF then ensures a more equal distribution of funding and, thus, healthcare provision. In the post-Soviet situation of severe underfunding of the healthcare sector, the MHIF is also likely to promote reforms to increase efciency in order to cut costs. Healthcare in Kyrgyzstan and Moldova has for this reason performed relatively well, although both countries have—in a regional comparison—low levels of GDP per capita, weak governance capacities and regular political crises.

#### **References**


#### **5 Causal Mechanisms in the Introduction of Mandatory Health…**


#### **5 Causal Mechanisms in the Introduction of Mandatory Health…**

———. 2008. Project Performance Assessment Report: Kyrgyz Republic Health Sector Reform Project (CREDIT 2860-KG), Second Health Sector Reform Project (CREDIT 3506-KG). Washington, DC: Te World Bank (Report No.: 44513).

———. 2014. Kyrgyz Republic: Public Expenditure Review Policy Notes— Health. Washington, DC: Te World Bank (Report No. 88979).

**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **Part III**

**Causal Mechanisms and Social Policies in African Countries**

## **6**

## **Appropriating the Colonial State: The Emergence of Social Insurance in Tunisia and Uganda**

**Kressen Thyen and Klaus Schlichte**

## **1 Introduction**

In most African countries, social insurance has played a limited role in the provision of social protection.1 Less than 10 per cent of the economically active population in Sub-Saharan Africa is covered by a social insurance scheme, most of which consist of old-age pensions and access

K. Tyen (\*) • K. Schlichte

e-mail: thyen@uni-bremen.de; kschlich@uni-bremen.de

<sup>1</sup>Tis chapter adopts a defnition of social insurance which includes both "Bismarckian" insurance systems as well as provident funds granting lump-sum payments at retirement. We further avoid the term "social security" for analytical purposes since its use varies across countries, sometimes referring to contribution-fnanced cash beneft schemes only or, as in the United Kingdom, including means-tested and non-contributory benefts (Walker 2005, 4).

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

University of Bremen, Bremen, Germany

to healthcare (OECD 2017, 11). While coverage rates in North Africa are higher, large parts of the population still remain excluded. Te limited reach of social insurance is closely linked to low levels of industrialisation and high levels of informality that have persisted in African countries since the colonial period (Barchiesi 2019, 48). Te bifurcation of labour markets led to the creation of two distinct segments of society with unequal social rights and entitlement to public welfare. In the light of these unfavourable economic and labour market conditions, what caused the introduction of social insurance in African countries?

In recent years, a certain consensus has emerged regarding the infuence of the former colonial powers in introducing and shaping social protection schemes across Africa (Midgley and Piachaud 2011; Schmitt 2015; Künzler 2016). Taking this as a departure point, we argue that the introduction of the frst social insurance programmes in African countries is however not the result of a simple extension of social policies from European countries to their colonies. Te frst old-age pension schemes in the form of provident funds emerged as a result of *imperial stafng* of public administration and distinguished between European and "indigenous" civil servants in colonial parlance. It was only with the new political and social arrangements that emerged during the historical period of decolonialisation that social insurance schemes turned into political reality. Specifcally, we fnd their introduction to have been caused by *appropriation* of the colonial state by the formerly colonial subjects. As the latter gained in infuence, they used and moulded colonial structures and policies along their domestic political constellation of forces. In this process, the pension schemes for domestic civil servants included more and more people as their relative number grew, while new insurance schemes were invented to cover permanent public employees as well as additional pressure groups.

Moreover, our analysis suggests the existence of two additional mechanisms that set in after independence. In industrialising countries with strong labour union movements, such as Tunisia, the new governments introduced broad social insurance schemes for private sector workers as a means of *labour incorporation*. In mainly agrarian societies with low levels of labour activism, such as Uganda, economic stagnation and low tax incomes seem to have motivated the top-down creation of pension funds for a small group of formal employees as a means of *public resource accumulation*, for example the mobilisation of capital that could be used for the purpose of state-directed development (cf. Kohli 2004). Overall, our hypothesised mechanisms underline the crucial importance of local agency, revealing both an adjustment to changing circumstances and clear intentions.

To demonstrate the above-mentioned causal mechanisms, we compare the most diferent cases of Tunisia and Uganda. Both countries have social insurance systems extending into the private sector, whereby more industrialised Tunisia has a diversifed system providing old-age, disability and survivor pensions, healthcare and, for some professional groups, family, maternity, disease and death allowances, and additional cash benefts for survivors and in the case of work accidents (IILS 2011, 72). In 2019, 58 per cent of the Tunisian population were covered by at least old-age and health insurance, public and private sectors combined (CRES 2019). In agrarian Uganda, social insurance is limited to old-age protection only.2 Here, 12.6 per cent of the estimated active labour force and 5.6 per cent of the total population were covered by a public provident fund in 2018 (Munyambonera et al. 2018).3 As political entities, the two countries have little in common except a period of colonisation by a European power, during which political, economic and social structures developed that left their imprint on both as post-colonial states. Tunisia was a former French protectorate and a settler colony, economically relatively diversifed (trade, agriculture, mining) and did not go through any major armed confict following independence. Uganda, a former British colony, relied on enforced cash-crop production, which created a gentry of export-oriented farmers, contrasting with far less prosperous northern and eastern provinces. Huge diferences in political traditions and the distribution of wealth paved Uganda's way into an extended period of

<sup>2</sup> In the case of work accidents, maternity and sickness, employers are required to pay the total cost or provide benefts directly to the insured, if the person is formally employed (ISSA 2019, 30). Despite this legal situation, employers often defy any obligation (Interview with construction worker, Kampala, 10 November 2018). Tis lack of efective labour regulation can be attributed to the high barrier costs of legal counselling (cf. Ayok 2016).

<sup>3</sup>Tis number drops to 4 per cent when excluding the civil service pension scheme, which is noncontributory (authors' own calculation based on data provided in Munyambonera et al. 2018).

political violence between 1971 and 1987. In that period, practically all public institutions decayed.

Te empirical analysis is based on a close reading of policy reports, ofcial statistics and secondary literature, as well as information obtained during research stays in Tunisia (March–May 2019) and Uganda (November–December 2018 and October 2019). Our detailed study addresses two major stalemates when it comes to social policy research in African countries: frst, its contextual focus on the more developed parts of the world, and second, the severe lack of case studies on the country level (Hickey et al. 2018, 6). We hereby hope to contribute to advancing theorising about social policy, which to this day remains largely based on West European and North American cases.

Te remainder of this chapter is structured as follows: the next section discusses existing research on the introduction of social insurance with a specifc focus on African countries. Ten the causal mechanisms underlying the introduction and extension of social insurance will be specifed. Te subsequent case studies present the emergence of social insurance schemes in decolonising and early independent Tunisia and Uganda. Finally, the conclusion considers broader implications of the study for social protection in Africa.

## **2 The Emergence of Social Insurance in African Countries**

Tere is general agreement today that the former colonial powers shaped the development of social insurance schemes in Global South countries. Specifcally, it has been shown that the former French colonies introduced social insurance schemes resembling those in France, while the former British colonies adopted central provident funds as a transition to a pay-as-you-go insurance scheme (Hu and Manning 2010, 143). With the British Colonial Development and Welfare Act of 1939, the creation of the *Fonds d'Investissement pour le Developpement Economique et Sociale* in 1946 and the gradual extension of French legislation, including the Social Security Act of 1945, to French overseas departments after 1946, it indeed appears as if the introduction of public social protection in African countries was just a prolongation of politics in the metropoles. However, this perspective neglects that colonial expansion was initially a response to the social crises in the metropoles (Elwitt 1967) and that the frst social protection schemes in the colonies were "for whites only" (Eckert 2019, 153).4 Moreover, colonial arrangements and social legislation were far from homogeneous, even within empires. Te French empire, often described as highly centralised and having "conceived colonies as extensions from France" (Künzler 2016, 4), spanned from French overseas departments to formally autonomous protectorates where French law did not directly apply. Many studies further tend to ignore the politics behind the introduction of social protection for the "indigenous populations". Indeed, more detailed country studies have shown that European settlers used their political weight to veto more inclusive policies emanating from central government and that colonial ofcials boycotted or delayed implementation in the hope of saving the old power alliances (Lewis 2000; Eckert 2004; Künzler 2020). Overall, most "difusion" studies underestimate the efects of local conditions and the impact of national political forces in shaping social protection, including the introduction and design of social insurance schemes (also see Seekings 2020).

In this regard, it seems useful to note that across the world, pensions and health benefts were frst introduced for members of the military and civil servants, and only gradually extended to include more and more people, and to cover a wider range of eventualities (see, e.g., Hannah 1986; Turner et al. 2020). In their work on Latin American countries, Collier and Collier (1991) have convincingly linked this expansion to rising levels of industrialisation and labour market formalisation, which created pressure to integrate labour as a political and economic actor. Part of the process of labour incorporation was the expansion of social insurance schemes to the urban working class (Huber and Stephens 2005, 620). In his work on decolonialisation and the labour question in Africa,

<sup>4</sup>Tis however did not mean that white settlers benefted from the same degree of social protection as in the metropoles, as many countries lacked a strong white working class or white urban poor that could have pressured for comparable benefts (see, e.g., Seekings 2005).

Cooper (1996) explicitly emphasises the role of social protection in the development and social integration of the new, "modern" urban African worker. Despite the proliferation of related ideas across the empires, more comprehensive social insurance schemes for workers emerged only after independence. Tis was not least due to the powerful combination of European employers, government, armed forces and white workers that hindered the organisation of African workers (Orr 1966, 73–74). Indeed, colonial governments ultimately considered more encompassing social protection too expensive (Eckert 2004, 475). Basic social services in health and education, for example, were delegated to a patchwork of religious institutions, in British colonies predominantly Christian missions (Scully and Jawad 2019, 557). Tese fndings suggest that the introduction of more encompassing social insurance schemes depends not only on labour mobilisation itself, but also on the growing political infuence of national labour unions after independence.

Still, the existing literature has difculties explaining why social insurance emerged in countries with largely agrarian economies where labour never emerged as a pressure group. Particularly south of the Sahara, public social protection and social insurance achieve low levels of coverage because of the limited importance of formal, institutionalised labour markets (Eckert 2004, 468). Tis is particularly the case in the countryside, where most people continue to rely on informal and kinship-based forms of social protection. Scholars have insisted on the key role of the International Labour Organization (ILO) in spreading norms of full employment and universal social insurance (Hu and Manning 2010; Schmitt et al. 2015). Yet, related studies often remain silent on what made African governments invest in the introduction of social insurance, given that many agrarian countries were not pressed by the same social question as their industrialised counterparts.5 Here, evidence from Sub-Saharan Africa and South-East Asia ofers a promising perspective, which links the introduction or expansion of pension funds to their potential

<sup>5</sup> It has been speculated that emulation may be at the source, for example the desire "to improve their international image and status as modern states" (Hu and Manning 2010, 130). However, given the manifold tasks and pressures that African governments faced after independence, we are not convinced that this actually caused the introduction of social insurance (but potentially infuenced their design).

for increasing the capital available to the state (Gerdes 1971; Kuhlmann and Nullmeier, 2021).

In sum, the existing literature suggests that late-colonial actor constellations, including colonial authorities, domestic elites and emergent pressure groups, are crucial for the emergence of social insurance in African countries. However, related claims often remain restricted to a general level. Terefore, the next section will specify a number of causal mechanisms behind the introduction of social insurance schemes in decolonising Africa, which we derived from the literature and refned in the course of our case comparison.

### **3 Mechanisms Behind Social Insurance in African Countries: Imperial Staffng, Appropriation, Labour Incorporation, Public Resource Accumulation**

In most African countries, social insurance was introduced in the global historical period of decolonialisation. Te term usually refers to the three decades after 1945, but frst signs of the declining legitimacy of the colonial project, including anticolonial unrest, the internationalisation of colonialism, and projects and expectations of reform, already emerged with World War I (Jansen and Osterhammel 2017, 38–42). It is therefore in this larger context that we situate our mechanisms.

While the course of decolonialisation was specifc in each case, there is general consensus in the literature that the historical events of the time formed a mutually sustaining context for each single colony (see, e.g., Albertini 1971; Cooper 1996; Fieldhouse 1986). Since World War I, the League of Nations and later the United Nations had promoted selfgovernment and controversies about the empires abounded, both outside and inside the colonies.6 In the early 1950s, the idea of "self-rule" and "decolonialisation" as an engineered transfer of power to "'trustworthy' indigenous leaders" (Jansen and Osterhammel 2017, 4) gained ground

<sup>6</sup> India's independence (1947) and the Chinese revolution (1949) further added pressure to end colonial rule.

among imperial powers. Tis was not least because of US pressure in the Cold War context, as Soviet solidarity with the colonies and the Chinese revolution threatened to tip the balance between East and West. Conversely, African populations felt on the verge of achieving independence, and national liberation movements began to negotiate with the colonial governments throughout the continent. Teir negotiating position improved as wars defending the status quo turned out to be ever costlier. Tis was the case in French Indochina (1946–1954), Kenya (1953–56), Algeria (1954–1962) and Cameroon (1955–1960). Following formal independence, domestic elites gradually took over while colonial ofcials, public employees and businesses prepared their departure. By explicitly situating our analysis in this period, we emphasise the presence of the past in the development of social policies in each country case but also on the international level.

#### **3.1 Imperial Staffng Mechanism**

Quite obviously, the imperial projects would have been impossible to realise without "indigenous" staf in the colonial administration, who were not only needed in terms of human resources but also as links and communicators between the imperial centre and the local populations (Darwin 2012, chapter 7). Te World Wars increased the need for bureaucratic control and thus additional staf in the colonies. At the same time, anticolonial unrest and expectations of reform were on the rise, with French and English colonies—especially in North Africa—worrying about pro-German sympathies and the loyalty of their "indigenous" staf (Jansen and Osterhammel 2017, chapter 2). Tis shift in power relations motivated the colonial governments to make a number of concessions to strategically relevant groups, including the entitlement to pensions for African civil servants and soldiers fghting in the imperial armies. While these frst pensions were accessible to only a small part of the respective populations and did not yet take the form of social insurance, they form the starting point for the introduction of the ensuing social insurance systems. Note that while in North Africa, nationalist movements and reform demands emerged by the First World War, many Sub-Saharan countries saw similar developments only shortly before or during World War II.

#### **3.2 Appropriation Mechanism**

Several authors have argued that the process of decolonisation must be understood in terms of an appropriation of colonial structures and policies, not a rupture (Bayart 1989; Cooper 1996; Bayart et al. 2007; see also Fanon 1961). Indeed, appropriation of governance forms facilitated cohabitation between colonial authorities and "indigenous" elites and, following independence, prevented violent resistance to the new order (Rinke et al. 2012). With regard to the introduction of social insurance, we fnd appropriation to have been triggered by the eroding legitimacy of the colonial projects, which motivated late-colonial governments to invest in the political and social integration of domestic populations into the colonial order. Tis involved partial autonomy and the "indigenisation" of the public sector, for example the inclusion of ever more "natives" into civil service positions. Domestic elites used the opportunities ofered by the new structures, forming them to their own ends. Tis implied the inclusion of more and more people in the colonial pension schemes for public ofcials, the invention of social insurance for additional pressure groups and, in many cases, coverage of a wider range of eventualities. As a result, the emerging insurance systems were highly refective of the social power relations in the late-colonial period.

Note that while the mechanism of appropriation may be generalised across diferent cases, the design of the resulting insurance schemes can vary. Both decolonialisation and the emergence of socialist governments in Africa produced considerable national variation within an imagined global pattern, hitherto dominated by the Bismarckian system of national and compulsory insurance (Hu and Manning 2010, 148). In the following, we propose two further mechanisms that we suspect to have infuenced the introduction of social insurance beyond the public sector.

#### **3.3 Labour Incorporation Mechanism**

A crucial factor determining the introduction of social insurance for private sector workers is the weight of labour, itself a result of specifc historical developments in the colonies, the degree of world market integration and capitalist diferentiation (Collier and Collier 1991). Accordingly, we fnd the introduction of broad social insurance schemes in African countries to be related to the increased political infuence of labour unions at independence. During the colonial period, African unions had begun to direct claims for social justice at the colonial authorities, which—given their repressive capacities—had nonetheless minimised their concessions to maintain cheap labour. Consequently, unions became important supporters of independence movements and a "standard component of African nationalism" (Orr 1966, 68). Once the new, independent governments were formally in charge, they needed to integrate their support base to maintain power. Tey entered negotiations with the unions, introducing comprehensive social insurance for private sector workers in exchange for political support. Note that this did not imply an extension of social insurance to the many informal workers without a lobby or mobilising power.

#### **3.4 Public Resource Accumulation Mechanism**

Many African countries experienced an economic downturn at independence, in addition to having been depleted of their resources for decades. Even today, the results of colonial exploitation remain visible in the overall low and inconsistent levels of capital accumulation (Nkurunziza 2019). At the same time, generalised expectations regarding improved well-being loomed high. Combined, these conditions appear to have triggered the introduction of social insurance funds, which not only projected the image of the aspired ideal of social progress, but also provided the government with a means of relatively rapid resource accumulation. Te investment of capital from insurance funds in development projects or even defence eforts has been shown for both North and Sub-Saharan African countries (see, e.g., Gerdes 1971; Eibl 2020, chapter 6). Public resource accumulation can occur in combination with labour incorporation but can also unfold on its own: where not tied to the presence of a specifc pressure group, the introduction and/or further extension of social insurance schemes are likely to have been driven by considerations of proftability.

## **4 The Expansion of Social Insurance in Decolonising Tunisia and Uganda**

In the following, we will illustrate the above mechanisms with the most diferent cases of Tunisia and Uganda which, despite very diferent preconditions, have some forms of social insurance systems. As we will show, the mechanisms of imperial stafng and appropriation are common to Tunisia and Uganda. Labour incorporation paved the way for a more encompassing social insurance system in Tunisia, confrming the importance of labour history. Te introduction of a limited social insurance scheme in Uganda is better explained by the need for public resource accumulation. Te two case studies start with a general overview of the historical context and actor constellations. We will then sketch the emergence and expansion of social insurance in both countries with a view to the mechanisms outlined above.

#### **4.1 The Step-by-Step Development of Social Insurance in Tunisia: From National Liberation to Labour Incorporation**

Having been home to European settlers and transnational trade for decades, Tunisia became a French protectorate in 1881. Weakened by internal power struggles, state bankruptcy and pressure from European consuls, the regent (*bey*) was forced to accord the French responsibility over Tunisia's foreign afairs and, in 1883, internal afairs. Te country remained a de jure regency throughout the colonial period, but beylical authority was de facto exercised by the French resident general who relied on a highly centralised administration run by French bureaucrats (Nelson 1986, 33), supported by colonial intelligence (Saf 2020). Because the Regency of Tunisia was never formally abolished, French (social) legislation did not directly apply.7 However, a number of French laws were subsequently promulgated via beylical decree. French colonial economic policy promoted trade, agriculture and mineral extraction, although the industrial sector remained underdeveloped in order not to create competition with the metropole. Politics and the economy were dominated by the French who settled and built their businesses particularly in the coastal areas. While only French citizens were represented in elected councils, a small Tunisian elite maintained sizeable infuence. Te protectorate status implied the coexistence and collaboration of French and Tunisian state ofcials, of which the latter were the frst to beneft from colonial old-age pension funds. As decolonialisation set in, appropriation of the colonial state and its social policies occurred through two major groups: the nationalist conservative elites which also dominated the civil service and, following independence, the new leadership around Habib Bourguiba, head of the national liberation movement and frst Tunisian president after the dispossession of the Bey in 1956.

Te roots of Tunisia's social insurance system, which formed throughout this process, can be traced back to the frst decades of the occupation. In 1898, the colonial government charged a provident society with the management of retirement schemes for the French civil servants working in the colonial administration (Chaabane 2002, 3). During World War I, in 1915, imperial stafng prompted the creation of a similar pension fund for Tunisian civil servants (Destremeau 2010, 131), which was complemented by a system of family allowances in 1918 (Chaabane 2002, 3). Tese developments are closely linked to the enlistment of Tunisian soldiers in the imperial army. In addition to veterans' pensions, they became eligible to family allowances, though inferior to those of French soldiers, after 1915 (Arnoulet 1984, 54–59). Upon their return, over a 1000 veterans obtained civil posts, including lower positions in the colonial administration. At the same time, the introduction of pensions

<sup>7</sup>Tunisia had been an Ottoman province until French occupation, but the beys had ruled the country relatively autonomously during the nineteenth century. In 1861, Tunisia received the frst written constitution in the Arab world.

did not mean equal treatment for French and Tunisian state ofcials. Not only were the schemes for Tunisians less generous, but in 1918, the introduction of the *tiers colonial* accorded one-third more pay to French nationals. Tis decision, alongside mounting calls for self-determination, triggered the creation of Tunisia's frst nationalist party, Destour, in 1920. Its programme demanded the restoration of the 1861 constitution (*destour*) as well as equal pay for equal work. At the same time, the party's aims were far from revolutionary. Conservative and elitist in orientation, it limited its demands for equal treatment to civil servants and religious authorities, hereby setting itself apart from manual workers (Beinin 2016, 14). Tis increased its bargaining power vis-à-vis the French authorities, but also split and thus weakened the national movement.

In response to Destour's position and its collaboration with French interests, a group of radical nationalists formed Neo-Destour in 1934, which combined demands for national sovereignty with social rights, hereby gaining the support of the labour movement. Te colonial authorities responded through repression, jailing and exiling its leaders. At the same time, the short period of the *Front Populaire* in France led to the proliferation of ideas of "colonial socialism" throughout the empire. In 1936, parallel to the Matignon reforms in France, workers in individual industries were granted paid leave and a forty-hour week. In 1943, the authorities proclaimed a minimum wage for mine workers.8 Family allowances for workers in the private sector, which had been demanded by the unions since 1932, were fnally accorded in 1944, but even then, the Tunisian *Caisses d'Allocations Familiales* functioned diferently to the one in France and its other colonies (Guelmami 1996, 99–102). Overall, social legislation remained far behind the workers' demands.

Following World War II, growing aspirations for independence coupled with the proliferation of social rights gave rise to what Guelmami (1996) has termed a "colonial welfare state" (ibid., 77–131).9 Tis implied a shift in colonial social policy, which now aimed at the social integration

<sup>8</sup>Tis decision was impacted by the economic consequences of the war, which increased the demand for mine workers. Tese, however, were mainly Italians and Tunisians, who had welcomed German occupation in the hope of independence (Wagner 1951).

<sup>9</sup> In contrast to World War I, returning soldiers played an inferior role. Only 22,000 Tunisian soldiers fought in World War II, and of those who returned, only 280 received posts in the colonial

of the "indigenous" population into the colonial order.10 Congruently, ideas of a reformist Tunisian government and stepwise internal autonomy began to take hold. In 1947 and 1951, France instated Tunisian governments with "indigenous participation" under the oversight of the resident general. Tis included, amongst others, the creation of a Ministry of Labour and Social Security under a Tunisian minister. Already in 1948, the new minister announced the creation of the *Caisse Nationale de Retraite* (CNP), a contribution-based pension fund for permanent employees in the public sector and concessionary enterprises (electricity, gas and transport). Te scheme was extended to the banking and insurance sector in 1949 (Chaabane 2002). In 1951, a health scheme covering long-term sickness and surgery for permanent public employees was added via the *Caisse de Prévoyance Sociale* (CPS) (Ladhari 1996; Chaabane 2002; Ministère des Afaires Sociales 2021).

As a result, three categories of employees emerged (Guelmami 1996, 100): frst, a superior category consisting of employees of the state, public services and municipalities, concessionary services, as well as employees in the banking and insurance sectors, who all benefted from social insurance. Many of these were, de facto, Europeans. Te intermediate category regrouped private sector employees, mostly mine workers who were granted a minimum wage and could receive family allocations. Yet, even here, few Tunisians benefted compared to the total population: by 1952, there were 48,000 Tunisians formally employed in mining and industry (Murphy 1999, 81), and even among these, many did not beneft from their legal rights. Finally, the inferior category comprised agricultural workers, peasants and craftsmen who were not eligible for social benefts. Obviously, the late-colonial interest coalition consisting of French authorities, French-dominated business and a small Tunisian elite gave birth to a highly exclusive social insurance system.

Despite French eforts to hold on to Tunisia via political autonomy and expanded social protection, independence became inexorable.

administration (Giudice 2013). Moreover, many were directly deployed in Indochina, returning to Tunisia only after 1954.

<sup>10</sup>Tis shift in social policy during World War II, which included increased spending on health and education, was also observable in Lebanon and Syria under the "Free French" after 1941 (Tompson 2000).

Neo-Destour grew into a powerful opposition, and in 1951, the political compromise between the French and the Tunisian authorities for reform instead of independence triggered a guerrilla war led by the armed wing of the liberation movement. Shortly after, Bourguiba entered negotiations with the French government who frst granted internal autonomy (1955) and then independence (1956). Te independence treaty of 1956 initially foresaw the instauration of a constitutional monarchy, preserving French and Tunisian elite interests. Following the frst elections, from which Neo-Destour emerged as the single power, the new power-holders dethroned the Bey and proclaimed a Republic in 1957.

Under Neo-Destour, appropriation continued, though now accompanied by labour incorporation. First, the independent government took over the state apparatus and its personnel. One of the frst steps consisted in the "Tunisifaction" of the state and the economy. Tis meant, on the one hand, replacing 12,000 French civil servants with Tunisians. Because of the shortage of qualifed personnel, the Tunisian government agreed to keep 3500 French civil servants for a transition period of unspecifed duration; this number had already dropped to 2000 by 1961 (Carter 1965, 26). In the light of overall low qualifcation levels, the new government relied on the Tunisian civil servants who had served in the colonial administration and who had not necessarily welcomed the abolition of the short-lived monarchy. To unify the state bureaucracy and secure support from public sector employees, Neo-Destour expanded social insurance benefts for the public sector, specifcally by adding health insurance in 1959.11

Second, the government needed to respond to the demands of labour, who had been crucial in achieving independence and bringing Neo-Destour to power. To understand why labour incorporation set in at this specifc point in history, a more detailed look at the role of unions before

<sup>11</sup>Note that the *Caisse Nationale de Retraite* and the *Caisse de Prévoyance Social* were merged into the *Caisse Nationale de Retraite et de Prévoyance Sociale (CNRPS)* on 30 December 1975. Under Law 85-12 of 5 March 1985, coverage was extended to the entire public sector, including temporary employees of the state, local government, public enterprises and national corporations and a special regime for members of the central government, deputies and governors (Chaabane 2002; Ministère des Afaires Sociales 2021). Contributions to the general regime correspond to roughly 24 per cent of the salary in 2020, for the special regime covering government ofcials, governors and members of parliament, which is co-funded by the state, 34 per cent.

and after independence is warranted. Already during the colonial period, Tunisia's mining sector had produced an increasingly activist labour movement. Te *Compagnie des Phosphates et des Chemins de Fer de Gafsa* (CPCFG) alone employed 20,000 miners in 1920, of which 3600 were Europeans (Beinin 2016, 13). In 1919, French and Italian workers established Tunisia's frst trade union, which also had a branch for Tunisian workers. Initially, the movement demanded that their working conditions be adapted to French standards. Joint mobilisation of European and Tunisian workers led to the introduction of the frst social protection programmes, including work accident indemnity for mine workers (1921) and agricultural workers (1924). At the same time, the labour movement remained divided, as the French branch did not support wage equalisation demanded by their Tunisian co-workers. Tis prompted major strikes and led to the creation of the frst Tunisian national trade union federation in 1924, the *Confédération Générale Tunisienne du Travail* (CGTT), that claimed to liberate Tunisian workers from the "union protectorate (*protectorate syndical*)" (Guelmami 1996, 85, own translation). Viewing this mobilisation as an anticolonial uprising, the French authorities banned the union and exiled its leaders.

Tis opened the door for the Neo-Destour party to emerge as a defender of the labour movement. In 1944, Ferhat Hached and Ahmed Tlili, both close to Neo-Destour, formed two regional federations of Tunisian trade unions with the aim of organising workers on a national basis. In 1946, the two federations merged into the *Union Generale des Travailleurs Tunisiens* (UGTT). Te UGTT defned itself as a federator of national aspirations with the task of communicating "a progressive message aimed at the destruction of the colonial system and the foundation of a new society based on social justice and labour" (cited after Guelmami 1996, 87, own translation). Legalised in 1947, the UGTT grew to 80,000 members by 1952, henceforth constituting the most infuential civic organisation and the main mobilising base of Neo-Destour. Te establishment of an encompassing social insurance system for all employees, including health insurance and old-age pensions, fgured among the core demands of its 1949 congress (ibid.).

Once independence was achieved and Neo-Destour took formal power, the new government was faced with the challenging task of fulflling the high expectations held by its constituency and outcompeting the more radical socialist and dissident wings of the labour movement (Cameau and Geisser 2003). Social progress and economic development became the founding principles of the new independent state and its proclaimed "joie de vivre" (*farhat el-hayat*) (Catusse and Destremeau 2010). Important resources went into state-led industrialisation, involving capital-intensive projects such as steel mills, an oil refnery, a paper plant, several textile factories and an automobile assembly plant (Ayadi and Mattoussi 2014, 3). In parallel, the government negotiated the form and scope of the future social insurance system with the UGTT. Contrary to the draft proposal, the initial scheme did however not include pensions as the government sought not to overburden employers. Further, it frst remained limited to employees in the non-agricultural private sector.

Te implementation of the National Social Security Fund (CNSS), established by Law 60-30 of 14 December 1960, was supported by the ILO, which Tunisia had joined in the very year of its independence. In 1974, the scheme was fnally complemented by pension benefts covering retirement, disability and survivors. Mirroring the UGTT's socialist orientation, the resulting social insurance system featured aspects from the Soviet model insofar as it covered all the major social risks of injury, sickness and old age at once instead of introducing them separately (Hu and Manning 2010). At the same time, it followed the Bismarckian model promoted by the ILO as it bet on the expansion of paid labour in an industrialising economy. Mirroring the power constellations of the time, it was thus the "*travailleur-citoyen*" and not the "poor" that was put at the centre of the new social insurance system that took on paternalist, bureaucratic and corporatist structures (Catusse and Destremeau 2010). In exchange for the concessions made to workers, the UGTT was incorporated in a Single-Party, Single-Union system (Bellin 2002).

Te CNSS, like the CNRPS for public sector employees, came under the auspices of a tripartite board consisting of the state, employers and employees (Chaabane 2002, 5). Tis meant that the social insurance funds were outside direct government control, attributing the Tunisian social insurance system a "purely social character" (Eibl 2020, 185). Te insurance's surplus was initially invested in bonds of banks and public infrastructure. Only after 1973 were these funds tapped to expand social welfare benefts, for example to cover low-income contributors and to fnance social housing (Guelmami 1996, 159). Te increase in public savings for development purposes occurred via alternative saving funds such as the *Caisse Nationale d'Epargne* and the *Société Nationale d'Investissment*. In this, Tunisia difered from Egypt, for example, where public resource accumulation was a major motivation (Eibl 2020, chapter 6).

Troughout the 1960s, the UGTT successfully pressured for the inclusion of further professional groups into the social insurance system. In 1965, it was extended to students (health insurance and family allocations only). Agricultural workers permanently employed for at least 180 days per year gained partial access in 1970; this was extended to agricultural workers employed for at least forty-fve days per trimester in 1981. Independent workers have been included since 1982. Tunisian workers abroad gained access in 1989. Since 2002, the social insurance system also includes formally employed domestic workers, contractual workers, small-scale fshermen and farmers, ranchers, labourers and sharecroppers, as well as artists, and intellectuals, craftsmen and designers.12 Today, roughly 25 per cent of the salary goes into the insurance fund (ISSA 2019, 29). While social insurance for all the above groups is managed by the CNSS, diferent schemes apply. Tese are the so-called general regime for non-agricultural workers (RSNA), two diferent schemes for agricultural workers (RSA and RSAA), a scheme for selfemployed workers (RTNS) as well as a scheme for Tunisian workers abroad (RTTE).13 In addition, several reforms of the various schemes occurred over time (see ISSA 2019). Te stepwise inclusion of professional groups via diferent schemes resulted in a highly segmented social insurance system. Tere are, for example, eleven diferent pension schemes depending on the branch and nature of activity (IILS 2011, 74). Benefts vary not only between the public and private sectors, but also within the private sector (Ben Romdhane 2005, 65–68).

<sup>12</sup>For more detail see Chaabane 2002; Destremeau 2010; Ministère des Afaires Sociales 2021; Sghari 2018.

<sup>13</sup>Te diferent social insurance schemes have diferent contribution rates. For a detailed overview see Ben Rhomdane (2005) and IILS (2011).

#### **6 Appropriating the Colonial State: The Emergence of Social…**

Overall, the Tunisian social protection system, today consisting of social insurance and social assistance, continues to privilege urban employees at the expense of those involved in small-scale agricultural or independent labour. Apart from general food subsidies, Tunisia's many poor began to directly beneft from the proclaimed national solidarity relatively late. Te Program for Aid to Needy Families (PAFN), created in 1986, constituted an important step. However, over 17 per cent of Tunisians are still estimated to be excluded from any form of public social protection (CRES 2019), which is not least due to problems in the PAFN's selection process, which for a long time was guided by party patronage.14 Further, even among those who are formally covered, problems of accessing adequate health services prevail, particularly in the country's interior regions. Consequently, the social question remains highly politicised and potentially destabilising to this day (Tyen 2019).

#### **4.2 How Ugandan Elites Appropriated the Colonial State and Its Old-Age Protection Institutions**

Uganda's colonial period began in 1888, when the British East Africa Company set up in the then Kingdom of Buganda after having helped to reinstate the *Kabaka* (king) after an internal uprising. It transferred its administration rights to the British government in 1894, and two years later, protectorate control was extended to Bunyoro, Ankole and Toro to cover what roughly corresponds to present-day Uganda. Te colony had to be integrated into the Empire's division of labour, with enforced cotton production as the frst scheme for funding the costs of colonial rule without creating competition to British interests. Cotton production for export began in 1904 and was complemented, in the 1920s, by commercial production of cofee and sugar. In contrast to settler colonies, half of the land was reserved for the Bugandan gentry and could not be acquired by non-Africans. Tis was very much to the beneft of Bugandan chiefs who secured the most fertile land and would furthermore serve as tax

<sup>14</sup> Interview with Tunisian social scientist, Tunis, 26 March 2019.

collectors. Tis constituted an important diference to settler colonies such as Tunisia, where Europeans came to dominate the economy.

In this particular context, the British colonial state of the "Protectorate Uganda" was appropriated mainly by two powerful groups (Mamdani 1996, chapters 5–6). One was the group of chiefs and large-scale landowners of the colony's core, the Kingdom of Buganda, which the British used as a textbook case of indirect rule. Large-scale Bugandan cotton and cofee producers benefted from the colonial economic policy and maintained considerable bargaining power throughout the period of colonial rule (1900–1962). However, formally, African actors gained access to colonial decision-making very late. Legislative and Executive Councils that were created already in the 1920s were opened up to African representatives only in the 1950s.

Te other group that appropriated the colonial state was the African staf in the public sector, growing in number as the colonial state grew in size and extending into ever more spheres of life. When Uganda gained independence, the formal empowerment of these two groups was the main means by which Ugandans took over the state. Tis applied equally to the formal institutions of social protection that had emerged during colonial rule. Te rudimentary system of free healthcare was massively expanded during the 1960s, as was the education system. Te colonial state had already invested in both since the 1940s, but churches had remained the main players in both felds. Of course, in both felds, inclusion and output remained very low during colonial rule.

Te mechanism of appropriation, however, becomes most visible in the feld of old-age protection. Tis history can be told in three parts, with the most important background condition being that up until today about 85 per cent of the active labour force has never been involved in any formally institutionalised pension system. Uganda is still a largely agrarian society, with the majority of the workforce living from smallscale farming and informal economic activities.

Te frst part of this history started in 1921 when the frst pension scheme was created for British citizens working for the colonial service in Uganda. In 1927, this tax-funded scheme was further extended to Asian employees of the colonial state. Tousands of South Asian migrant workers had been hired for the construction of railways in the early colonial period, and many of them decided to stay, forming an intermediary group between an African peasantry and a dominating white colonial class. In 1929, the growing number of African civil servants were included in this pension scheme (Bukuluki and Mubiri 2014, 37), which might be explained by the imperial stafng mechanism elucidated above. Formally, the year 1929 thus marks the introduction of a non-racially discriminating old-age protection system; yet for a long time it included only a tiny minority of the colonially subdued population.

With the extension of public service, more African Ugandans became benefciaries of this pension scheme. In 1939, with the beginning of World War II, an Armed Forces Pension Scheme was created for the more than 70,000 African Ugandans who fought for the British Empire. Tis number amounted to a tenth of the male workforce of the time. Yet formally, these were pensions, so that the only real social insurance that came about during colonial times started in 1941 when African employees, including the growing number of teachers in public schools and clerks in the local administration, became benefciaries of the Government Employees Provident Fund (Barya 2011, 9).

With the end of World War II, the days of colonial rule were numbered. In Uganda, several political parties mobilised through general strikes in 1945 and 1949, to which the colonial government reacted by repressing trade unions. By 1957, the Railway African Union was the only one left with more than 500 members (Mamdani 1996, 191). On the political level, colonial rule opened up slowly in the 1950s, including African representatives into the Legislative Council, who would make up 50 per cent by 1955. Te Executive Council was, similar to developments in other colonies, developed into a Ministerial Council. Tis and the massive increase in African personnel in public service (see Table 6.1) indicated the encroaching appropriation of the colonial state by Ugandans.

In contrast to other colonies where independence was achieved through violent struggle, the transition was comparatively smooth in Uganda. In 1961, one year prior to independence, general elections were held to determine the new government. Benedicto Kiwanuka's Democratic Party led Uganda into independence but was defeated by Milton Obote's Uganda People's Congress (UPC) in the frst general elections in 1962.


**Table 6.1** Africanisation of *public service personnel* in Uganda 1956–1959

Source: Uganda Argus, 5 April 1960, quoted after Apter (1961, 417)

Te king of Buganda became the non-executive president. Up to this point, Uganda did not have any contribution-based social insurance apart from the provident fund for local government employees, of which the vast majority were teachers. And even this had been integrated into the tax-based pension scheme in 1953 for civil service. It was only in 1967, fve years after gaining independence, that Uganda would have a social insurance system outside the public sector ofering old-age insurance for a small group of private sector employees in the urban centre(s).

Part of the explanation for the long-term absence and limited signifcance of social insurance schemes for private sector employees is the mode of colonial world market integration. Right from its start, colonial economic policy aimed at the production of cotton and cofee, which were processed elsewhere, in the case of cotton, for example, in Great Britain. In Uganda, both major export goods were produced by farmers, who depended on kinship solidarity for their social protection. Te political economy of the colonial state relied on rents from exporting agrarian production while industrialisation was avoided in order to protect industries in the UK metropole. Tis meant a lack of wage earners in the Ugandan economy (Elkan 1961). While Ugandan labour unions formed relatively early in comparison to other African countries, the labour movement thus never developed the strength and size it did in more industrialised countries, so that the labour incorporation mechanism cannot explain the Ugandan case.

Apart from a small urban service economy, wages and salaries were thus restricted to the public sector for which the tax-funded pension scheme mentioned above was entrenched. It was only in the late stage of colonialism of the 1950s when a stronger "development" orientation in colonial policies allowed for the creation of textile production in Uganda itself. Industrial production, however, never grew to an extent that organised urban labour became a powerful political force that could call and push for social insurance legislation. Te actual birth of Uganda's social insurance system was thus a top-down initiative in the form of the Social Security Act of 1967. In contrast to the development in Tunisia and other cases on the African continent, this innovation was thus not the result of social confict.

Uganda, like other successor states of former British East Africa,15 introduced a National Social Security Fund (NSSF) for private sector employees and non-pensionable public servants. It consisted of a savings scheme based on earnings-related contributions by workers and employers. Te creation of this social insurance scheme can be seen as part of the social service expansion of the early independence period with its huge investments in health and education (Reid 2017, 244–67). Its introduction may also be explained by the government's need for capital and its aspiration to build a modern public infrastructure with a strong welfare element (Gerdes 1971).

Te fund was planned to cover an estimated 200,000 workers, whereby 15 per cent of their salary should go into the NSSF. One-third of the contribution was to be paid by the respective employee, and the other two-thirds by the employer.16 Te accumulated reserves of the fund were to be used by the Ugandan government to accelerate Uganda's economic development (Nyakundi 2009). While the reporting US embassy saw at the time "few administrative difculties" ahead (US Department of Labor 1967, 23), the fund was, however, entirely plundered during the rule of General Idi Amin (1971–1979), mostly for prestigious projects, for example, a conference centre for the OAU summit in 1975. Tis dissolution was part of a larger process of institutional decay in Uganda. During the ensuing years of political turmoil and civil war, almost all public institutions stopped working. It was only after 1987, under the enormous impact of donor initiatives and policy prescriptions, that state agencies slowly recovered.

<sup>15</sup>Te Nigerian Provident Fund was the oldest, followed by Tanzania, Ghana, Kenya, Zambia and Uganda (Gerdes 1971, 573). Today, Nigeria and Uganda are the only countries that have not moved to social insurance systems (see country profles in ISSA 2019).

<sup>16</sup>Te initial contribution rates have been maintained to this day (ISSA 2019, 30).

Te NSSF was reconstituted by a law in 1985 and started working again in 1987. Since then, it has been a relative success story, even if many of its directors were sued for embezzlement and mismanagement. In the view of contributors at least, it is of now "fairly working".17 In 2014, it covered about 1.45 million members of which 500,000 were working. Yet the fund is not really a solidarity fund as each contributor maintains her or his own account of assets on which an annual interest rate of 10 per cent is paid. It covers the average infation rate in Uganda over the last ffteen years. Again, like in 1967, the NSSF is also the largest domestic capital formation institution. It has invested its USD 3.5 bn in government bonds and in a host of Ugandan enterprises (Kamukama 2019). Its emergence and revitalisation might, again, be explained along the public resource accumulation mechanism. But case literature does not exist, so a thorough historical reconstruction still has to be written. Its revitalisation was certainly driven by other needs than social protection, namely the perceived need for capital formation and the fnancial demands of the Ugandan state. Causally though, this growth has become possible only since the Ugandan workforce has grown considerably in services like construction, tourism, trade and transport. Te business of development, with a multitude of NGOs, and the boom in private education are also part of this change in the occupational structure. Te decisive pressure to create the NSSF, we assume, thus rather stemmed from the interaction between the government that came into power in 1986, starting with empty state accounts, and the International Financial Institutions aiming at reducing the share of grants and loans in the government budget (Schlichte 2021).

Te system of old-age protection is currently generally politicised. Pensions for public employees as well as the special funds for armed forces and the police are perceived as socially unjust, because the pensions of former state employees are a tax burden for a largely informally employed or self-sustaining population. Pensioners have been mostly better educated and are rather urban males, so that the pension system is much less benefcial for rural areas and the less educated. Te vast majority however,

<sup>17</sup>Conversation with Ugandan NGO employee, Kampala, 13 November 2018; Interview with Ugandan sociologist, Berlin, 17 September 2018.

about 70 per cent of the workforce, does not beneft at all from these schemes. It consists of subsistence farmers who sell a little surplus on the market (Munyambonera et al. 2018, 14). In this regard, Uganda's old-age protection system displays a continuity since colonial times: since the last decade of colonial rule, public employees have been criticised for their relative privileges and for demanding Western standard salaries in a society of poor peasants (Ehrlich 1963, 264).

#### **5 Conclusion**

As former colonial subjects took over the colonial state, they also appropriated its social policies, including the early pension and healthcare schemes that had been introduced for "indigenous" civil servants as a means of promoting imperial stafng. During this process, the frst social insurance was granted to public employees and additional professional groups that had gained infuence in the course of decolonialisation. Following independence, pressures for labour incorporation and/or public resource accumulation led to the invention of further insurance schemes beyond the public sector. By unravelling these causal relationships, we show that the introduction of social insurance in African countries is not simply a result of colonial powers extending their policies into their colonies, but strongly related to political contentions that emerged in the context of waning empires and persisted beyond formal independence. Specifcally, in contexts where organised labour played a crucial role in national liberation, as in Tunisia, their increased negotiating power after independence led to a broad extension of social insurance into the private sector. Te socio-economic trajectory of colonial Uganda, however, did not create a strong demand for social insurance and led to a top-down decision to extend social insurance with the motive of creating a public investment fund.

More generally, our analysis shows that the limited coverage and segmentation of present-day social insurance can be directly related back to colonial structures and divide-and-rule strategies. In Tunisia, social protection was—since its inception—organised via a discriminatory system that ofered little protection to the larger population (Catusse and Destremeau 2010). Te gradual, but unequal extension of social insurance mirrors experiences from Latin America, where the stepwise expansion also resulted in highly fragmented and generally unequal social insurance systems (Huber and Stephens 2005, 620). Needless to say, social insurance systems in Africa cover only the formally employed, thus excluding many, if not the majority of workers in the informal economy. Especially in the countryside, public social protection rarely exists. While this situation has been exacerbated by the cutbacks in public sector employment since the 1980s (Eckert 2019, 163), this particular constellation of the "social question" in Africa has a long history that goes back to early colonial times (Veit et al. 2017) and is particularly remindful of Mamdani's famous binary of "citizens" and "subjects" (Mamdani 1996).

Our comparison shows quite clearly that the realm of possible "difusion" of social policies, including the introduction of social insurance for broader parts of the working population, was heavily infuenced by the structures set by the mode of colonial world market integration. In the case of Uganda, it has turned out to be extremely difcult to diferentiate the colonially inherited economy as other countries had already occupied prospective niches in the world market. During the 1960s, this was a direct outfow of the colonial division of labour. Due to a period of civil strife and economic recession in the 1970s and 1980s, Uganda's polity deinstitutionalised generally. Nor did the process of internationalised state formation after 1990, with the neoliberal age in full swing, ofer a chance for economic diversifcation (Obwona et al. 2014). In contrast, Tunisia mobilised enormous resources for state-sponsored industrialisation after independence. Private sector development followed in the 1970s, whereby the Tunisian economy diversifed into new sectors in manufacturing and services. Tese investments meant an expansion of formal labour, ofering the base for a more comprehensive social insurance system that today covers more than half of the population. Te map of how social insurance spread in space and time might thus ultimately mirror the history of the global division of labour and national political constellations.

#### **References**


———. 2019. From Poverty to Informality? Te Social Question in Africa in a Historical Perspective. In *Te Social Question in the Twenty-First Century. A Global View*, ed. Jan Breman, Keaven Harris, Ching Kwan Lee, and Marcel van der Linden, 152–169. Berkeley: University of California Press.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **7**

## **Policy Pollination as a Causal Mechanism Explaining Social Protection Adoption in Africa**

**Stephen Devereux**

## **1 Introduction1**

Since the late 1990s, a particular form of social protection has become entrenched in the social policy agenda across the world. At the global level, social protection was not mentioned in the Millennium Development Goals in 2000, but it features in three Sustainable Development Goals (United Nations 2015). At the continental level, no African country had a national social protection policy in 2000, but thirty-fve of ffty-fve had produced one by 2019. Millions of Africans

S. Devereux (\*)

<sup>1</sup>Tis chapter is an amended version of a working paper co-published by the SOCIUM Research Centre on Inequality and Social Policy and the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", University of Bremen, and the UK Institute of Development Studies (Devereux 2020).

Centre for Social Protection, Institute of Development Studies, Brighton, UK e-mail: s.devereux@ids.ac.uk

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_7

who had no access to social assistance twenty years ago now receive social cash transfers (SCTs) from their governments every month (UNDP 2019).

Te literature on social protection has also grown and evolved during this period. Initially (as discussed below), empirical research focused on impact evaluations to establish the efectiveness of social cash transfers and of alternative design choices (e.g. conditional versus unconditional cash transfers). More recently, academic interest has shifted to trying to understand the causal mechanisms driving the widespread adoption and expansion of social protection as a policy sector, especially in Africa. Most relevant for this chapter is an emerging literature that analyses the difusion of social protection policies, specifcally cash transfers (von Gliszczynski and Leisering 2016; Leisering 2019), with a particular focus on sub-Saharan Africa (Adésínà 2011; Hickey et al. 2019). Tis literature resonates with theoretical and empirical literatures on social welfare in OECD countries (Esping-Andersen 1990; Schmitt and Starke 2012), on welfare provision in the Global South during and since the colonial period (Gough and Wood 2004; Midgley and Piachaud 2011; Schmitt 2015), and on cross-country policy transfer processes (Dolowitz and Marsh 2000; Dobbin et al. 2007; Obinger et al. 2013).

Policy transfer can be understood as a process whereby "knowledge about policies, administrative arrangements and ideas in one political system (past or present) is used in the development of policies' administrative arrangements, institutions and ideas in another political system" (Dolowitz and Marsh 2000, 5). Causal mechanisms for policy transfer include learning, competition, and emulation—a tendency for countries to mimic policies such as free primary education that have been socially constructed as desirable at the global level.

A fourth causal mechanism in the policy transfer literature is coercion, which involves the use of power (e.g. fnancial leverage) by transnational actors to induce policy change by national governments, especially those that are "reliant on those entities for trade, foreign direct investment, aid, grants, loans" (Dobbin et al. 2007, 454). Tis mechanism has been labelled coercive difusion. "An often-mentioned case of coercive difusion is fnancial aid linked to certain domestic reforms defned by donor countries and international institutions such as the International Monetary Fund or the World Bank" (Obinger et al. 2013, 115).

#### **7 Policy Pollination as a Causal Mechanism Explaining Social…**

Tis chapter identifes a ffth causal mechanism, here labelled *policy pollination*, that carries elements of all four mechanisms in the policy transfer literature but is most closely aligned to coercive difusion and, as argued below, persuasively explains the spread of social protection in sub-Saharan Africa. In nature, cross-pollination is the process by which pollen is carried between fowers by bees, birds, and other pollinators. For this analogy, fowers are African countries, pollen is social protection policies, and pollinators are agents (staf, consultants) of international development agencies, fying from country to country to propagate social protection. Fertilisation is achieved when countries adopt and fnance social protection policies and programmes, as advised by their pollinators.

Te next section introduces the notion of policy pollination as a causal mechanism, drawing on Africa's experiences with structural adjustment in the 1980s and 1990s, Poverty Reduction Strategy Papers in the early 2000s, and social protection more recently. Next, development agencies are identifed as the dominant pollinators in the social protection policy transfer process, led by ILO, UNICEF, and the World Bank. Five strategies for policy pollination are then discussed: (1) evidence-building; (2) fnancing; (3) capacity strengthening; (4) policy formulation; and (5) domestication of international law. Tere follows a brief discussion of why some African governments have resisted consistent pressure and incentives to adopt the specifc variant of social protection advocated by transnational actors.

#### **2 Policy Pollination as a Causal Mechanism**

International development agencies2 have intervened in social policy formulation in Africa since the colonial period and continue to do so today, decades after former colonies achieved independence (Schmitt 2020). Tis has been achieved primarily through coercive difusion, leveraging

<sup>2</sup> International development agencies refer to a range of transnational actors, including bilateral donors (e.g. DFID, GIZ, Irish Aid, USAID), multilateral agencies (e.g. European Union), United Nations organisations (e.g. ILO, UNICEF, WFP), international fnancial institutions (e.g. IMF, World Bank), international NGOs (e.g. ActionAid, Concern Worldwide, Oxfam, Save the

the hard and soft power conferred by development fnance—ofcial development assistance (ODA), concessional lending, and humanitarian relief. In the 1980s, for instance, international fnancial institutions imposed conditions on structural adjustment loans that required African governments to, among other things, remove food price subsidies and impose user fees on poor citizens for health and education services. A striking feature of these Washington consensus (Williamson 2000) policy prescriptions was their uniformity. A simple economic model for market liberalisation and state withdrawal, devised by a powerful group of transnational actors led by the World Bank and International Monetary Fund (IMF), was implemented reluctantly by many African governments, with consequences for poor Africans that were ambivalent at best (Easterly 2006; Moyo 2009).

In the early 2000s a wave of Poverty Reduction Strategy Papers (PRSPs) spread across the Global South, more than half in Africa. In the year 2000 alone, twenty-fve African governments prepared a PRSP document (IMF 2016). Again, this policy process was driven by the World Bank and IMF. Preparing a PRSP was rewarded with concessional loans or debt relief (Ejolu 2008). All PRSPs embodied the same pillars, framed within a neoliberal ideology: good governance, pro-poor economic growth, investment in human capital, and in some cases, social safety nets to protect the poorest. Te striking similarities in these documents across diverse countries reveal the extent to which they were designed by external agencies acting as "policy pollinators," rather than emerging organically from domestic deliberative processes about optimal strategies for reducing poverty in each country context.

PRSPs were criticised as "a primary policy device of international development institutions (that) restrict practical and political options, while exacting heavy establishment and compliance costs" (Craig and Porter 2003, 53). Compliance was enforced through coercive application of conditionalities on loans by the World Bank and IMF. However, their roll-out to dozens of low- and middle-income countries was represented by these agencies as participatory and democratic. Te IMF described the

Children), and research institutes or think-tanks (e.g. the Institute of Development Studies, International Food Policy Research Institute (IFPRI)).

adoption of PRSPs as a country-driven process, "promoting national ownership of strategies through broad-based participation of civil society" (quoted by Ejolu 2008, 22). Although PRSPs were implemented in more than thirty-fve African countries, in recent years they have almost disappeared, confrming how shallowly grounded in domestic policy processes they actually were.

As PRSPs faded into history or were absorbed into national development strategies, another wave of policy documents started spreading across Africa. Since 2010, thirty African countries have published a National Social Protection Policy (NSPP) or National Social Protection Strategy (NSPS). Tese documents share many common features. Notably, they all include promises that the government will deliver social assistance to specifed target groups: children, older persons, persons with disabilities, and "the poor" (UNDP 2019, 56). As discussed later, these policy statements have been heavily infuenced—fnanced, commissioned, even (co-)authored—by transnational policy pollinators: development agencies and agents working for them.

A statement by two United Nations Special Rapporteurs highlights the disproportionate power that international agencies exercise over social policy formulation in poor countries, and the contradictions between the Washington consensus policies that they imposed in the 1980s (e.g. making poor people pay for education and healthcare) and the expansionist social protection policies (notably giving free cash transfers to poor people) that were advocated by the same agencies in the 2000s.

In the past, major international institutions have pushed States to lower government spending and programming in favour of economic development, opening markets and reducing the size of the State. In the last decade however, many international institutions have begun to address the benefts of social protection systems to development and to promote their adoption. (de Schutter and Sepúlveda 2012, 3)

As with PRSPs, the story of social protection's rapid rise in Africa is interpreted diferently by diferent observers, some seeing it as evidence of progressive thinking by African governments, others as evidence of the power of international development actors to impose their will on governments they advise and support. Tis can be represented as two stylised narratives.


Support for both positions can be found in the social protection literature. In 2010 a book titled *Just Give Money to the Poor* presented social cash transfers in its subtitle as *Te Development Revolution from the Global South* (Hanlon et al. 2010), implying that the idea is indigenous rather than imported. By contrast, an article published in the *Journal of Social Policy* in 2016 was subtitled "How international organisations defned the feld of social cash transfers in the 2000s" (von Gliszczynski and Leisering 2016). Te next section builds the case in support of this second position, showing how external actors pollinated the ideas of social protection and cash transfers throughout much of Africa.

### **3 International Development Agencies as Policy Pollinators**

Social protection as a development discourse emerged out of social safety nets, a limited set of policy instruments that straddles the gap between social assistance and humanitarian relief. Te 1990 World Development Report set out a two-pronged strategy for reducing poverty: labourintensive economic growth and provision of basic social services to the poor. Te report also identifed the need for safety net programmes as a complementary third element.

Even if this basic two-part strategy is adopted, many of the world's poor the sick, the old, those who live in resource-poor regions, and others will continue to experience severe deprivation. Many others will sufer temporary setbacks owing to seasonal variations in income, loss of the family breadwinner, famine, or adverse macroeconomic shocks. A comprehensive approach to poverty reduction, therefore, calls for a program of *welltargeted transfers and safety nets* as an essential complement to the basic strategy. (World Bank 1990, 3; emphasis added)

Tis is possibly the frst articulation in the development policy discourse of what later became the two building blocks of social protection systems in the Global South: targeted transfers or *social assistance* to the chronically poor (who "experience severe deprivation") and categorical vulnerable groups ("the sick, the old"), as well as safety nets or *social insurance* for those experiencing livelihood shocks ("temporary setbacks").3 In the late 1990s the World Bank developed the safety net component into its Social Risk Management framework (World Bank 2001), and it continues to use the term safety nets even though it is out of favour with

<sup>3</sup> In the global social policy discourse, these building blocks have their origins in the Universal Declaration of Human Rights (United Nations 1948) and the Social Security (Minimum Standards) Convention (ILO 1952).

other agencies. As recently as 2018 the World Bank published the third edition of "Te State of Social Safety Nets" (World Bank 2018).

In 2000, social protection efectively did not yet exist in the development policy discourse. Te period since has been framed by two global position statements about development policy: the Millennium Development Goals (MDGs) and the Sustainable Development Goals (SDGs). As noted earlier, the MDGs did not mention social protection at all. Just ffteen years later, social protection consolidated its entrenchment in the international development policy discourse by being named in three of the seventeen SDGs, as an instrument for ending poverty (goal 1), enhancing gender equality and empowerment (goal 5), and reducing inequality (goal 10).4

Te year 2012 was another watershed for social protection. Tree international development agencies—the ILO, UNICEF, and World Bank—had emerged as the leading actors and policy pollinators in this sector in the Global South. It was increasingly clear that social protection was not a development policy experiment or fad that would disappear after a few years. Coincidentally, these agencies each released defnitive statements of their approach to social protection in the same year. Tese documents give operational and programming content to the aspirations of the international community to pollinate distinct visions of social protection in aid-recipient countries, and the types of interventions each agency chooses to pollinate, refecting their mandates and ideological positions as much as technical assessments of social needs and contextspecifc policy options.

In June 2012, the International Labour Conference endorsed a "Recommendation concerning National Floors of Social Protection," which presented an explicit rights-based approach to social protection, building on the Universal Declaration of Human Rights of 1948. Te Recommendation defned social protection foors as "nationally defned

<sup>4</sup>Goal 1, target 3: End poverty: Implement nationally appropriate social protection systems and measures for all, including (social protection) foors; Goal 5, target 4: Gender equality and empowerment: Recognise and value unpaid care and domestic work through the provision of public services and social protection policies; Goal 10, target 4: Reduce inequality: Adopt policies, especially fscal, wage, and social protection policies, and progressively achieve greater equality (United Nations 2015).

sets of basic social security guarantees which secure protection aimed at preventing or alleviating poverty, vulnerability and social exclusion" (ILO 2012). "Te guarantees should ensure at a minimum that, over the life cycle, all in need have access to essential health care and to basic income security" (ILO 2012). Te four guarantees are:


In efect, this set of provisions elaborates and gives rights-based efect to the underdeveloped social assistance component of social protection systems, since the Social Security (Minimum Standards) Convention sixty years earlier had focused mainly on social insurance (ILO 1952).

Also released in 2012, UNICEF's "Social Protection Strategic Framework" (UNICEF 2012) identifed three principles for "integrated social protection systems": progressive realisation of universal coverage, nationally owned systems and national leadership, and inclusive social protection—tackling social exclusion. Te specifc focus of UNICEF's engagement with social protection is revealed in the subtitle: "Enhancing equity for children." While this refects UNICEF's mandate as the United Nations agency for children, it also leads UNICEF to favour a certain set of policy instruments and to prioritise some vulnerable or "at risk" groups above others.

UNICEF's vision of a social protection system has four components (UNICEF 2012):


4. *Legislation and policy reform*: Changing policies and legislation to remove inequalities in access to services and/or economic opportunities, addressing issues of discrimination and exclusion.

Te World Bank's "Social Protection and Labor Strategy" (World Bank 2012) links social protection directly to labour markets: "social protection and labour systems … help people and families fnd jobs, improve their productivity, cope with shocks, and invest in the health, education, and well-being of their children." Te subtitle—"Resilience, Equity, and Opportunity"—reveals that the World Bank sees social protection as performing instrumental functions, beyond being a right or entitlement. It should contribute to building resilience and generate opportunities to escape from poverty, such that benefciaries will become self-reliant and resilient and ultimately no longer need social assistance.

Te World Bank strategy has three overarching goals:


Te strategy envisages a linear progression to building social protection and labour systems, starting with small uncoordinated projects towards fully connected, well-functioning, and efciently run national programmes. "Te strategic direction is to help developing countries move from fragmented approaches to more harmonized systems for social protection and labour." With this framing, the main prongs of the World Bank's strategy are social assistance, social insurance, and active labour market programmes (Robalino et al. 2012):


3. *Active labour market programmes*: intermediation, counselling, job search and matching, vocational skills training, wage subsidies to frms

While social assistance programmes have been efectively pollinated throughout Africa (UNDP 2019), many with World Bank fnancial and technical support, social insurance has received less attention and active labour market programmes have often failed. A recent review of evaluations concluded that "active labour market policies are much less efective than policymakers typically assume. Many of these evaluations fnd no signifcant impacts on either employment or earnings" (McKenzie 2017).

Tese position statements by three of the most infuential international development agencies are important because they defne the focus and parameters of each agency's engagement with, and support for, social protection policies and programmes in countries where they operate—where they exercise substantial infuence over policymaking processes, using policy pollination strategies that are discussed next.

## **4 Policy Pollination Strategies**

Te international development community has invested heavily in advocating for social protection in Africa since 2000, using various strategies. Five such policy pollination strategies are discussed here: (1) building the evidence base for social protection impacts, (2) fnancing social protection programmes and systems, (3) strengthening government technical and operational capacity, (4) instigating national social protection policies, and (5) domesticating international law.

#### **4.1 Building the Evidence Base**

In the early 2000s the international development community selected social cash transfers (SCTs) as their preferred instrument for social protection in low- and middle-income countries. Te UK's Department for International Development (DFID) set out the theoretical case for cash transfers.

While poverty is multi-dimensional, low and variable income is central to the problem. Modest but regular income from cash transfers helps households to smooth consumption and sustain spending on food, schooling and healthcare … Over time, transfer income can help households to build human capital (and) save up to buy productive assets … Cash transfers can thus both *protect* living standards (alleviating destitution) and *promote* wealth creation. (DFID 2011, i)

However, the evidence base at frst was thin. Tere were few unconditional cash transfer programmes in low-income countries, and even fewer that had been rigorously evaluated. Development partners therefore decided, for strategic reasons, to invest in building the evidence base that cash transfers can protect and promote living standards. Te implicit theory of change was that policy adoption is evidence driven, so proving that cash transfers can contribute to desired outcomes such as poverty reduction would convince governments to implement and pay for their own national programmes.

From the very frst cash transfer pilot projects in Africa, development agencies allocated substantial funds to monitoring and evaluation (M&E) of projects that they instigated and fnanced. Te technical justifcation for conducting impact evaluations is to confrm that a policy intervention is achieving its objectives and to generate learning for improved design or delivery. Pilot projects also allowed experimentation with different design modalities. In Lesotho's Cash and Food Transfers Pilot Project, some benefciaries received food packages, some were given cash, while others received both cash and food. In Kenya's Hunger Safety Net Programme, three targeting mechanisms determined eligibility in diferent communities—high household dependency ratios, community-based wealth ranking, and older persons—and their targeting accuracy was compared (Sabates-Wheeler et al. 2015).

Development agencies also commission evaluations of interventions they support for advocacy reasons, aimed at two politically important audiences. First, donor governments need to justify their spending on development programmes to their domestic constituencies—taxpayers in high-income countries. Showing positive impacts in terms of, say, poverty reduction in Africa is strategically necessary to keep the pipeline of development assistance fowing. Cash transfer projects have immediate demonstrable benefts, not least because giving money to poor people makes them less poor, by defnition, so it was relatively easy to quantify these outcomes as attributable achievements of development spending.

Second, as noted, development agencies commission impact evaluations of their projects for advocacy purposes in countries where they work, as one policy pollination strategy. Findings of positive outcomes from pilot SCTs were intended to convince African governments to adopt this idea and scale it up. Te UK development agency explicitly acknowledged the link between building the evidence base and building political support. "Robust monitoring and evaluation are crucial both for programme performance and political sustainability. Te rapid spread of cash transfers in MICs in recent years has been in large part due to just such high-quality analysis and M&E" (DFID 2011, vii). For this reason, even small-scale pilot projects received disproportionately large evaluation budgets.

Te frst decade of this century saw a rapid accumulation of evidence that social protection in Africa, specifcally cash transfers, "works." But early evaluations were not always methodologically rigorous. A GTZfunded evaluation of the infuential Pilot Social Cash Transfer Scheme in Kalomo District, Zambia, found improvements for benefciary households in several indicators relative to baseline, including food security and asset ownership (MCDSS 2007). However, this study was criticised for not including a control group, which meant that fndings of positive changes in benefciaries' wellbeing could not be unambiguously attributed to the SCT. Following these frst-generation social protection initiatives, a more sophisticated wave of impact evaluations was commissioned of larger programmes such as the Productive Safety Net Programme (PSNP) in Ethiopia, the Hunger Safety Net Programme (HSNP) in Kenya, and Concern Worldwide's "Graduation model" projects in Burundi and Rwanda. Multi-year multi-round household surveys were designed (baseline, midline, endline, sometimes also follow-up) with treatment and control groups, following rigorous quasi-experimental randomised control trial (RCT) protocols.

Numerous publications have synthesised the accumulating experience and evidence with various forms of social protection programmes across African countries. Notable contributions include:


efects of cash transfers on individuals and households through a rigorous review of the literature of ffteen years, from 2000 to 2015" (Bastagli et al. 2016, 5). Summarising the impacts reported in 165 evaluations of ffty-six cash transfer programmes in low- and middleincome countries, the authors found "strong evidence" that cash transfers are associated with increases in household expenditure (a proxy for reductions in monetary poverty), food expenditure and dietary diversity (indicators of food security), school attendance, use of health services, household savings, and women's decisionmaking power.


None of these publications are products of "blue skies" academic research; they were written or commissioned by international development agencies for advocacy purposes, to promote specifc policy positions. Te Foreword to "From Evidence to Action" states: "Tese pages also document the ways in which the Transfer Project has infuenced the policy debate in each of the eight countries" (Davis et al. 2016, vi). While the primary intended impact of cash transfer *projects* is to improve the wellbeing of benefciaries, the primary intended impact of cash transfer *evaluations* is to persuade governments to implement and fnance national cash transfer programmes.

#### **4.2 Financing Social Protection Programmes**

Many African governments were reluctant to introduce social protection programmes, despite pressure and fnancial incentives from international agencies to do so. One major reason is that these governments see social protection as unafordable, given fscal constraints in low-income contexts and competing priorities for public spending (Seekings 2017). In particular, cash transfers at scale are regarded as too expensive, especially if they involve regular transfers of meaningful amounts of cash to all people defned as poor in the country every month, not only for a year or two but indefnitely.

One response to the reality of limited budgets and low prioritisation of social protection in much of Africa was for external actors to provide the seed funding, especially for technical inputs such as design and systemsbuilding, but also in many cases for the cash transfers themselves. Tis was intended to be an interim measure until governments assumed responsibility for running and fnancing these projects as national programmes. Te anticipated shift in fnancing from external to domestic sources can be illustrated as a "funding seesaw." It is striking that the countries where social protection is fully funded by external actors are some of the world's poorest and most aid dependent. Conversely, all the countries where social protection is fully funded from domestic resources are middle or high income (World Bank 2018, 18).5 Tis provides empirical evidence for a familiar dilemma—that countries with the greatest need for social protection have the least resources to deliver it—and

<sup>5</sup> https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and- lending-groups (accessed July 24, 2020).

justifes external actors stepping in to fll the fnancing gap. Sub-Saharan Africa has "the smallest proportion of working-age population in the world" (ILO 2017, 130) and high levels of economic informality, therefore a very small tax base. Te funding seesaw predicts that as national incomes rise, governments will collect more taxes and allocate more fscal resources to social protection, which also allows them to take more control over social protection design and delivery. Until then, development agencies will continue to ofer full or partial funding of social protection in low-income African countries, and they will continue to exploit this leverage by shaping social protection in these countries in line with their own mandates and objectives.

A second response by external actors to claims that social protection is unafordable in low-income countries was to develop the "business case" for "investment" in social protection. Tis required demonstrating that public spending on social protection generates economic returns to individuals, local economies, and the national economy. Pathways from cash transfers to economic growth include human capital formation (this explains why the World Bank promotes conditional cash transfers that require benefciaries to send their children to school and health clinics) and local income multipliers (because cash transfers stimulate demand for goods and services). FAO simulated the multiplier efects of cash transfer programmes using a methodology called Local Economy-Wide Impact Evaluation (Tome et al. 2016). Te Economic Policy Research Institute (EPRI), a think-tank, used a microsimulation model showing how investment in social protection contributes to economic growth, reducing poverty and requiring less social protection in the future (Samson 2005). Te Australian and UK aid agencies each commissioned reviews of the efects of social protection on economic growth (Mathers and Slater 2014; Barrientos and Scott 2008; respectively). While these studies might be characterised as contributions to the evidence-building agenda rather than the fnancing debate, the primary purpose was advocacy—challenging government perceptions that social protection is wasteful or unproductive expenditure that low-income countries cannot aford.

Related to this was work demonstrating that social protection can generate "value for money" for governments as well as agencies, in terms of cost-efciency and cost-efectiveness against policy objectives such as poverty reduction. DFID commissioned studies of "measuring and maximising value for money" in social transfer programmes and social protection systems (White et al. 2013, 2015). Tese manuals ofer guidance on how to assess whether social protection programmes and systems actually are reducing poverty, and how to achieve this more cost-efciently and more cost-efectively.

A third response by external actors was to point out that public spending decisions are political choices and that even the poorest countries can aford some spending on social protection. Te World Social Protection Report 2017–19 reveals that spending on social protection as a percentage of GDP averages just 4.5 per cent in sub-Saharan Africa, but 18 per cent in Western Europe (ILO 2017). Te ILO argues that African countries should allocate more to social protection, because poverty is lowest in countries that spend the most on social protection, and they produced guidelines setting out how governments can create more fscal space. Options include increasing tax revenues, reallocating public expenditure, expanding contributory social security coverage, lobbying for international aid, reducing illicit fnancial fows, drawing on foreign exchange reserves, and borrowing or restructuring government debt (Ortiz et al. 2015). ILO's research draws on empirical data, modelling, and country case studies, but it was undertaken for explicit advocacy purposes, to persuade governments that they can (and should) spend more on social protection than they do already.

Despite these eforts, the question of who fnances social protection programmes remains a contested issue. As Seekings (2017, ii) explains, "international organizations have generally failed to convince national policymaking elites to raise and to allocate scarce domestic resources to social protection programmes. Te result is an 'afordability gap' between what is advocated for African countries and what these countries' governments are willing to spend."

#### **4.3 Strengthening Capacities to Deliver Social Protection**

International agencies have invested in building African government capacity to design and implement social protection programmes and to strengthen social protection systems. Strategies include embedding expatriate consultants as advisors within government ministries, arranging study tours for politicians and technical staf to observe social protection practices in other countries, and facilitating high-level dialogues with parliamentarians or regional bodies such as the African Union.

Te most popular capacity-building mechanism is training workshops, where government ofcials convene for periods between one day and two weeks, to learn from "experts" about social protection theory and practice. Te workshop setting allows agencies to transfer their preferred approaches to social protection to participants, through positions taken by trainers in relation to specifc design choices. Agencies have fundamentally diferent visions of social protection. For instance, while the World Bank favours poverty targeting using proxy means tests, UNICEF prefers categorical targeting of vulnerable groups, while ILO advocates instead for universal rights-based programmes.

Some agencies run their own training workshops. Te World Bank has run a course in Washington every year since the early 2000s, now called the "Social Safety Nets and Delivery Core Course," for policymakers, analysts, and operational staf from international agencies. Tis course "aims to provide participants with an in-depth understanding of the conceptual and practical issues on safety nets or social assistance as part of broader social protection systems."6 One session in the 2019 course was called "Making the case for social safety nets," confrming that the pedagogical objective is policy pollination as much as being purely educational. Te ILO runs an annual two-week Academy on Social Security at its International Training Centre in Turin. Tis course ofers "a diversifed

<sup>6</sup> www.worldbank.org/en/events/2019/03/18/social-protection-and-jobs-core-courses-2019 (accessed August 6, 2020).

training package on governance and fnancing, reforms and extension of social protection systems."7

Online training is also available through TRANSFORM, an interagency initiative led by ILO, UNICEF, and UNDP, and hosted by the Virtual Campus of socialprotection.org. TRANSFORM—Leadership & Transformation Curriculum on Building and Managing Social Protection Floors in Africa—explicitly promotes the ILO's preferred approach. "By the end of this course, you should be able to understand why and how a Social Protection Floor is benefcial to your specifc country context and how it can assist social and economic development."8

Training courses convey specifc information and technical skills, but they also ofer an unparalleled opportunity for policy pollination, by giving trainers who are employed or contracted by international agencies a platform and a captive audience of social protection policymakers and stakeholders. Social protection has many defnitions and is characterised by numerous design dilemmas—cash or food? conditional or unconditional? targeted or universal?—that are not technical problems but policy choices that refect competing ideological positions. Consciously or unconsciously, trainers inevitably communicate their personal biases and those of their agencies—to course participants, and this causally infuences social protection policy processes when participants return home to their ofces.

#### **4.4 Instigating National Social Protection Policies or Strategies**

Te frst National Social Protection Policy (NSPP) in Africa was promulgated by Mali in 2002, and only four more countries had followed by 2010. Ten in 2011, fve more countries published their social protection policies, and by 2019 the number had risen to thirty-fve, almost twothirds of the ffty-fve countries in Africa (Fig. 7.1).

<sup>7</sup>www.ilo.org/secsoc/information-resources/meetings-and-events/WCMS\_207266/ (accessed August 6, 2020).

<sup>8</sup>https://virtualcampus.socialprotection.org/moodle/mod/page/view.php?id=57 (accessed August 6, 2020).

**Fig. 7.1** New national social protection policies or strategies in Africa by year, 2000–2019. (Source: Author's compilation)

Nonetheless, it does not necessarily follow that this was a nationally owned process of developing and institutionalising a domestic policy agenda in every case. Te short period in which so many countries published these documents suggests that a deliberate policy transfer process was underway. As with the wave of Poverty Reduction Strategy Papers (PRSPs) that spread across the Global South in the early 2000s (as discussed earlier), external actors have been heavily involved in guiding or driving the process of drafting social protection policies in Africa since 2002. Most of these policy documents were drafted or commissioned by international agencies and their expatriate consultants, with varying degrees of involvement of national consultants and relevant government ofcials. Twenty-two of the thirty-fve African countries that have an NSPP are low income and twelve are lower-middle income, all with high levels of involvement of international development partners in domestic policy formulation.

From 2010 to 2012 the World Bank led a process of developing a National Social Protection Policy and Strategy for Togo. Te Acknowledgements explain that: "Tis report was prepared by a World Bank team with support from the Government of Togo and (…) various donors, including the World Bank, UNICEF, ILO, and UNDP (…) Tis fnal version incorporates the Government's comments" (World Bank 2012, i).

Te covers of some African social protection policy documents display the institutional logos of international agencies that supported the development of the policy, alongside the country's coat of arms, clearly signifying that these policies were at best co-produced and co-owned. Examples include the Gambia (UNDP and UNICEF) and Liberia (African Development Fund, European Union, UNICEF, World Bank, World Food Programme, and Concern Worldwide). Te content of these policies further confrms that they refect the ideas, ideologies, and priorities of these agencies, rather than those of national governments. At least twenty-fve NSPPs are organised around four imported conceptual frameworks: the life-cycle approach, social protection foor, social risk management, and transformative social protection. Tese frameworks derive, respectively, from UNICEF, ILO, the World Bank, and the Institute of Development Studies (Devereux and Kapingidza 2020).

#### **4.5 Domestication of International Law**

Another plausible explanation for the adoption of social protection by so many African governments almost simultaneously is social construction, which asserts that globalisation has been associated with a growing consensus across the world about appropriate societal goals, and the actors and means to achieve them (Dobbin et al. 2007). A case in point is the rise since World War II in the recognition of individual human rights (as a societal goal), the establishment of the United Nations and the International Criminal Court (as actors), and the overthrow of repressive dictatorships and their replacement with democratic institutions like regular elections and a free press (as means).

Seen through a constructivist lens, the rapid difusion of social protection throughout Africa could be interpreted as refecting the voluntary incorporation of globally constructed human rights-based agendas into domestic policy processes by African governments. After all, they are active participants in global policy forums (notably the United Nations General Assembly) where international standards are debated, resolutions are adopted, and conventions are ratifed. Support for this view might be found by analysing the application of international law in specifc country contexts.

In 1948 the Universal Declaration of Human Rights asserted that "everyone, as a member of society, has the right to social security" (United Nations 1948). In 1952 the ILO's Social Security (Minimum Standards) Convention 102 established the basis of modern social protection systems with nine branches of social security,9 including family beneft, healthcare, unemployment, and old age. In 1966 the International Covenant on Economic, Social and Cultural Rights "recognise(d) the right of everyone to social security" (Mpedi and Nyenti 2015).

Other international conventions afrm the right to social protection or social security for specifc vulnerable groups, notably the Convention on the Elimination of All Forms of Discrimination Against Women (1979), the Convention on the Rights of the Child (1989), and the Convention on the Rights of Persons with Disabilities (2006). Most African countries have ratifed these three conventions. More recently and most pertinently, in 2012 all member states of the International Labour Conference voted to adopt the Social Protection Floors Recommendation (R202), which advocates for a rights-based package that guarantees access to healthcare for all as well as income security for children, people of working age, and older persons (ILO 2012).

Te United Nations appoints Special Rapporteurs who function efectively as policy pollinators, by drafting policy statements that clarify the commitments governments have made under international law and travelling to countries to verify that member states are fulflling their obligations. In 2012 the United Nations Special Rapporteurs on the Right to Food and on Extreme Poverty and Human Rights co-authored a proposal for a "Global Fund for Social Protection" (de Schutter and Sepúlveda

<sup>9</sup>Note that the ILO uses the terms social protection and social security interchangeably. "*Social protection, or social security*, is a human right and is defned as the set of policies and programmes designed to reduce and prevent poverty and vulnerability throughout the life cycle" (ILO 2017, xxix; emphasis added).

2012). Tis document clarifes the content of the right to social protection in international law and the obligations of all member states to "respect, protect, and fulfl the right to social protection" (de Schutter and Sepúlveda 2012, 5) as comprehensively and expeditiously as possible, including by passing appropriate domestic laws.

Under the International Covenant on Economic, Social and Cultural Rights (ICESCR), States must devote their *maximum available resources* to the fulflment of economic and social rights, including through the establishment of social protection systems. As recognised under the ICESCR, some dimensions of economic and social rights can only be achieved progressively over time. However, this cannot be invoked as a pretext for delaying action. (de Schutter and Sepúlveda 2012, 6)

An implicit theory of change explains how the globally afrmed human right to social protection might "cascade" down to the realisation of this right by individuals living in specifc country contexts (Devereux 2017). First, a government representative signs or ratifes relevant instruments in global (e.g. Union Nations) or regional (e.g. African Union) forums. Next, the provisions of these global and regional instruments are codifed in national position statements such as the Constitution or National Development Plan. Tird, policies, programmes, and projects are designed and implemented to give efect to these provisions. Fourth, legislation is passed that gives these social protection interventions the status of a justiciable right. Finally, local civil society organisations campaign to hold the government accountable to deliver on the right to social protection.

In reality, this process plays out in very few countries. Te road from ratifcation to implementation to enforcement is very long—in some cases, non-existent. Most global Declarations, Conventions, Covenants, and Recommendations are not legally binding. Te United Nations Special Rapporteurs have no legal authority to enforce UN resolutions. No government has yet been prosecuted at the International Criminal Court for violating the right to social protection. At the national level, a right to social security is specifed in about half of the constitutions in Africa (fourteen out of thirty examined by Fombad 2013). But in most cases, this right is not justiciable—it cannot be enforced. In South Africa, civil society has taken the government to court to uphold or extend the right to social protection. For example, an alliance of local civil society organisations successfully lobbied Parliament and the High Court to extend the age of eligibility for the Child Support Grant from seven to eighteen years, on the basis that eighteen years is the legal defnition of a child in South Africa (Proudlock 2009).

But South Africa is an exception to the rule, for reasons related to its unique history (Devereux 2011). In many other African countries, civil society activities are curtailed and strictly regulated, and they have no freedom to campaign for economic, social, and cultural rights. Tis illustrates a critical point: that national governments are accountable to their domestic constituencies and are responsive to local political imperatives, rather than to declarations signed in global forums.

### **5 Understanding Resistance: Why Some Governments Say "No"**

Political self-interest drives government policy processes, and this is the lens through which all decisions about national social protection policies should ultimately be analysed. Teories of change asserting that policy choices are evidence based, or refect the domestication of international law, are not necessarily aligned with political realities. Governments need to be convinced that introducing a new policy and committing resources to new or scaled-up programmes will help them to win votes—in other words, policymaking is driven more by "what's popular" with key political constituencies than by "what works" for the poor.

In this context, social protection is susceptible to both positive and negative politicisation (Devereux and White 2010). In democratic regimes with accountable governments, delivering benefts to poor people makes governments popular and earns them votes—a positive outcome for poor people and their governments. Negative politicisation occurs if governments manipulate targeting and eligibility criteria to ensure that benefts are disbursed not on the basis of need, but to their own supporters as a reward for their loyalty, or to opposition supporters to induce them to switch their vote in future elections. An example of positive politicisation comes from an election campaign in Lesotho, when opposition parties made a manifesto commitment to double the amount paid to pensioners and the ruling party responded by promising to raise the social pension payment, which they did after winning the election (Croome et al. 2007). On the negative politicisation side, donor agencies in Zambia who argued for rolling out the SCT from poorest to least poor districts, based on poverty headcount rates, faced pressure from politicians who wanted the programme to be launched in their districts frst, irrespective of poverty rankings (Harland 2014).

African governments are located at the intersection between external pressure from international development agencies to implement specifc policies and programmes in certain ways—exerted through hard or soft conditions on development aid—and internal pressure from constituents and local civil society lobbying on behalf of specifc groups of citizens. In this context, the imperative driving policy choices are not necessarily how to reduce poverty rapidly, but political survival. One strategy that governments deploy to balance these competing priorities is to use "development partners" to fnance programmes favoured by these agencies—such as social cash transfers—and to commit government resources to interventions that are more popular domestically. Malawi is a case in point. For many years international agencies advocated to scale up social cash transfers, building on a positively evaluated UNICEF-funded pilot project in Mchinji District (Miller et al. 2011). But the government preferred to allocate its funds to supporting farmers with subsidised fertiliser and seed, regarding this as a productive investment in a politically infuential constituency. Eventually the social cash transfer did scale up to cover all districts, but mainly through funding provided by development agencies. In 2019, the World Bank, European Union, Germany, and Ireland supported the SCT programme in twenty-seven districts, while the government fnanced the SCT in just one out of twenty-eight districts (Government of Malawi 2019).

Te Malawi case illuminates two important asymmetries between African governments and their development partners. One is political: donor agencies believe (and have invested resources to prove) that cash transfer programmes uplift poor people and that governments should therefore allocate their own resources to such spending. But governments do not necessarily regard marginalised groups among the poor as politically important constituencies. Teir political interests lie elsewhere—in this case, with supporting farmers who grow the country's food—a national strategic priority—and vote. Te second asymmetry is about the choice of instruments. Despite years of policy advocacy for social cash transfers by the international development community, many governments remain unconvinced. Pejorative attitudes exist towards the poor and towards social cash transfers everywhere, not only in Africa. Free "handouts" from the state are believed to create "dependency syndrome" (Shepherd et al. 2011), laziness, and wasteful spending (e.g. on alcohol) among benefciaries, who should instead be given support for productive livelihoods (e.g. agricultural inputs).

It follows that the reluctance of some African governments to implement cash transfers refects careful political calculations, as well as genuine disagreements between national actors and international development agencies, about the optimal allocation of scarce public resources. From the perspective of policy pollination, these cases reveal that the eforts of international agencies to propagate social protection do not always succeed, even if they invest heavily in all the pollination strategies described above.

### **6 Conclusion**

Many governments in sub-Saharan Africa have implemented social cash transfer programmes for their poor and vulnerable citizens during the last twenty years. Tis process has been actively promoted by international development agencies that have ofered substantial fnancial and technical support, tied to the adoption of this specifc form of social protection. Several plausible explanations for this policy difusion process have been considered in this chapter, which complement rather than contradict each other.

Te frst explanation is that this is an *evidence-driven policy process*. Social protection has proven—through rigorous evaluations of cash transfer schemes—to achieve signifcant positive impacts on poverty and vulnerability. Tis accumulation of evidence convinced many African governments to implement, scale up, and ultimately pay for their national social protection programmes. A second possibility is that it is a *political choice*. Politicians adopt policies that are electorally popular and consolidate their power. Given this context, social protection is expanding throughout Africa because delivering cash transfers is amenable to both positive and negative politicisation. A third factor is that social protection is well established in *international law*, from the Universal Declaration of Human Rights in 1948 to the Social Protection Floors Recommendation in 2012. Te adoption by African governments of national social protection policies, programmes, and laws might simply refect the domestication of their legal commitments under international law.10

Four causal mechanisms are commonly discussed in the policy difusion literature. In three of these—learning, competition, and emulation—policies are voluntarily chosen by adopting governments, while in the fourth—coercion—policies are imposed on reluctant governments by external actors, using their hard or soft power. Tis chapter has focused on a mechanism closely linked to coercion, which I call policy pollination. Te argument is that social protection has been *pollinated* throughout sub-Saharan Africa, especially in low-income countries, by international development agencies, using their fnancial leverage and technical expertise to incentivise policy adoption by African governments and to institutionalise social protection in national policy structures.

Te extent to which this policy process was coercively imposed on African governments, and the extent to which it was autonomously chosen with support solicited by these governments from their development partners, varies from country to country. Devereux and Kapingidza (2020, 298) propose that a social protection policy process can be characterised as being externally induced or pollinated by international actors, rather than being domestically driven and nationally owned, when the following conditions apply:

<sup>10</sup>Tis framing builds on an earlier article that identifed three drivers of social protection in Africa—evidence ("what works"), politics ("what's popular"), and ideology ("what's right") (Devereux and White 2010).


It could be argued that the mechanisms of difusion of social protection policies and programmes across Africa are of secondary concern, and what matters is the outcome: that millions of poor and vulnerable Africans now have access to social protection, which is their human right. However, the way that social protection is introduced into a national policy discourse matters—which instrument? who benefts? who decides? who pays? In some countries, external actors appear to be more committed to social protection than national actors. Te test for sustainability of social protection in such contexts will come when development agencies withdraw their fnancial and technical assistance, leaving each government to decide whether it is politically benefcial and fscally cost-efective to continue prioritising this specifc set of social policy instruments.

**Acknowledgements** Te ideas in this chapter were presented by the author at seminars at the Universities of Bath, Maastricht, Southern Denmark, and the Institute of Development Studies (IDS); also, at the SOCIUM conference "Building Social Protection Systems in the Global South" (Bremen, 2018), the Development Studies Association Annual Conference (UK, 2019), the 11th International NordWel Summer School Conference (Bremen, 2019), and the conference on "Social Policy in Africa" (South Africa, 2019). Te author thanks two referees, Rachel Sabates-Wheeler at IDS and Carina Schmitt at the University of Bremen. Tis work is supported by the National Research Foundation of South Africa (Grant Number: 98411) and the Newton Fund, administered by the British Council, as well as a Mercator Fellowship with the CRC 1342 based at SOCIUM Research Centre on Inequality and Social Policy at the University of Bremen, Germany.

## **References**


———. 2017. Te Right to Social Protection in Africa: From CCCCDFPRTs to CLSPPPs. *Law in Africa* 20 (1): 11–32.

———. 2020. Policy Pollination: A Brief History of Social Protection's Brief History in Africa. *SOCIUM SFB 1342 Working Papers* 8. Bremen: SOCIUM Research Centre on Inequality and Social Policy.


<sup>———. 2015.</sup> *Transforming Our World: Te 2030 Agenda for Sustainable Development*. New York: United Nations General Assembly.


———. 2015. *Measuring and Maximising Value for Money in Social Protection Systems*. London: Department for International Development (DFID).


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **Part IV**

**Causal Mechanisms and Social Policies in European Countries**

## **8**

## **Anti-communist Backlash in the Croatian Healthcare System**

**Ante Malinar**

## **1 Introduction**

During the transition, in 1993, the Croatian healthcare system underwent profound changes demarking it from the system established in the communist period. Reforms took place in all three dimensions of healthcare: fnancing, regulation and provision. Tis chapter focuses only on reforms in the fnancing dimension and its regulation (Rothgang et al. 2010). Healthcare fnancing in the communist period was characterised by formally established social ownership, self-management and decentralisation based on Bismarckian principles. Te reforms in the early 1990s tackled the defciencies of the communist system. Tey departed radically from its

A. Malinar (\*)

University of Bremen, Bremen, Germany e-mail: malinar@uni-bremen.de

© Te Author(s) 2022 **239**

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_8

principles and established a mix of fnancing policies stemming from the experiences of other Western and Central and East European countries, providing evidence for horizontal policy transfer and learning (Dolowitz and Marsh 2000; Klein 1997). Tis policy mix established a hybrid system of healthcare fnancing combining Beveridgean, Bismarckian and market principles as well as radically changing the role of the state. Schmid et al. (2010, 455) argue that healthcare systems are becoming hybrid. Hybridisation of healthcare policies occurs through policy transfer, "the cross-national difusion of ideas about policy concepts and instruments" (Schmid et al. 2010, 460). Although their research includes only OECD countries, the Croatian case confrms that the trend of hybridisation is present even in non-OECD countries.

Croatia is a case worth investigating for several reasons. First, it is an example of one of the few countries in Central and Eastern Europe (CEE) which had a Bismarckian system during the communist regime. Healthcare fnancing was decentralised and involved multiple social health insurance (SHI) funds managed by providers and users (Džakula et al. 2014; Parmelee 1985; Šarić and Rodwin 1993). Second, while the idea of implementing SHI difused across CEE, the reforms in Croatia took the opposite direction. SHI was retained, but it was centralised with the creation of one national health fund closely controlled by the government. At the same time, formal out-of-pocket payments (co-payments and self-medication payments) (Kaminska and Wulfgramm 2019) and two diferent forms of private insurance were instituted. Tis has efectively created a hybrid fnancing system (Chen and Mastilica 1998, 1157; Vončina et al. 2007; Kovačić and Šošić 1998). And third, transition accompanied by war has made Croatia an extremely vulnerable country in dire need of international aid where, consequently, one would expect international organisations to have greater bargaining power and leverage. However, the direction and ideas for reform were defned by domestic actors, while international organisations had only a minor infuence or none at all.

Tis chapter answers two research questions: (1) How and why did reforms in the fnancing of Croatian healthcare move towards a hybrid system in the 1990–1993 period? And (2) what was the role of policy transfer processes in those reforms? Te chapter answers the research questions by referring to the elementary and complex causal mechanisms adopted in this volume (Chap. 1). Elementary causal mechanisms explain the production of activities by individual or collective actors that comprise only one causal step, while complex causal mechanisms contain several elementary causal mechanisms which form a causal chain of several causal steps (Chap. 1). Tus, process tracing is used to illuminate the black box between independent and dependent variables (George and Bennett 2005, 206) while holding "the promise of a rich account of how a complex political phenomenon like public policy emerges" (Baker and Kay 2015, 2). In this case, the independent variable is a high level of attention to the issue stemming from the mismatch in healthcare policies and the changing environment (economic and political crisis and its culmination in the early 1990s), while the dependent variable is the 1993 health policy output.

Tree complex causal mechanisms are identifed: the *doctors enter politics mechanism*, the *old system departure mechanism* and the *seeking solutions abroad mechanism*. Te frst mechanism explains how doctors occupied key political positions in the healthcare policymaking domain. Te second mechanism explains how negative perceptions of the communist healthcare system and intermediary policy solutions led to a departure from the old system. Te third mechanism explains how, during the reform process, Croatia scanned policy solutions and drew on lessons learnt abroad. Combined, these three mechanisms form the complex causal mechanism of *anti-communist backlash*. It explains how the prevailing dissatisfaction with the communist healthcare system, particularly among medical doctors, pushed the reforms in a new direction towards hybridisation of fnancing. What is more, in all four complex causal mechanisms, diferent elementary causal mechanisms are identifed, such as an emotional orientation, a rational orientation and a comparative orientation.

To reconstruct the reform process, the chapter analyses qualitative data such as media publications, medical journals, laws and parliamentary minutes.1 Furthermore, 13 interviews were conducted with experts

<sup>1</sup>Te majority of the newspaper articles analysed come from the most infuential newspapers at that time: *Vjesnik*, *Večernji list*, *Slobodna Dalmacija* and *Glas Slavonije* (Ramet 2013), but the analysis also included newspaper articles from non-mainstream media such as *Radničke novine* and *Sindikalna javnost*. In total, 125 newspaper articles were collected. In addition, 48 medical journals (*Liječnički vjesnik* and *Liječničke novine*), 10 healthcare laws and minutes of 26 parliamentary debates have been collected and analysed.

knowledgeable of the subject, such as ministers and their assistants, healthcare administration staf from the institutions relevant to policy (Ministry of Health, Croatian Institute for Health Insurance and Croatian Institute for Public Health), politicians involved in the healthcare system, but also academics, experts and journalists. Te research started with a review of all available secondary literature and then proceeded with the interviews. Tis chapter is based on an analysis of the collected data using a qualitative content analysis methodology. Te mechanisms derived in this chapter were inspired by existing theories and concepts from policy process research, most notably the concept of epistemic communities, the policy transfer literature and punctuated equilibrium theory (Baumgartner and Jones 2009; Haas 1992; Dolowitz and Marsh 2000).

Te chapter starts with a literature review on healthcare reforms in CEE countries and Croatia, while also providing an explanation of the role of policy transfer processes in these countries. Subsequently, in Sects. 3 and 4 the chapter explains the context of and the trigger for the reforms. Section 5 identifes the actors involved, their interaction and the activities that produced the reforms. Moreover, it ofers explanations for how the reforms produced a hybrid system of healthcare fnancing in Croatia. Te concluding section relates the mechanisms that were identifed in this chapter to policy process research and draws some broader conclusions.

## **2 Healthcare Reforms in CEE and the Role of Policy Transfer**

Te transition of CEE2 countries from the communist regime, which started in 1989, had a profound impact on their health systems. During the communist regime, the majority of CEE countries had a centralised healthcare system named after Semashko. It was characterised by state control of the system, tax-based fnancing, universal coverage, free

<sup>2</sup> In this chapter, the CEE region is understood as post-communist countries, excluding the post-Soviet ones (except the Baltic states).

provision of health services at the point of use, informal payments, overreliance on hospital services and underfunding (Lawson and Nemec 2008). CEE countries underwent a major paradigm shift and introduced market and liberal reforms into their systems, such as decentralisation by establishing social health insurance (SHI) funds, allowing a free choice of practitioners, privatisation and introducing out-of-pocket payments, while also reducing the hospital sector and developing primary care (Lawson and Nemec 2008; Rechel and McKee 2009).

Although there were many similarities with other CEE countries, the so-called Yugoslav *Štampar* model was organised on distinctly Yugoslav communist values of self-management and Bismarckian principles whose formal goals were to establish democratic governance through decentralisation and devolution of authority in the decision-making process. Te reforms in the early 1990s followed similar principles as in other CEE countries, although Croatia centralised the system and retained the SHI.

In general, the literature on policy transfer processes in CEE countries is inconclusive, suggesting that the ideas for reforms were either exogenous or endogenous. A frst strand of literature explains that healthcare reforms in CEE were a result of the infuence and pressures of international organisations. It suggests that early transition created a policy vacuum in the CEE countries and that international organisations such as the World Bank (WB), the World Health Organization (WHO), the International Monetary Fund (IMF) or the European Union (EU) seized the opportunity to exert their infuence through fnancing and policy advice (Deacon et al. 1997; Cerami 2006; Kaasch 2015). For instance, Nemec and Kolisnichenko (2006, 15) argue that the World Bank and the IMF were instrumental in providing ideas on marketisation reforms in CEE healthcare systems.

On the other hand, it is suggested that the reform processes were mostly endogenous and that other factors beyond the infuence of international organisations accounted for the policy changes (Rechel and McKee 2009; Sitek 2008; Radin 2003; Roberts 2009). Sitek (2008) argues that the direction of change was infuenced by the interaction of political institutions, party politics and in some cases professional organisations such as medical chambers. Moreover, the majority of CEE countries wanted to move away from communist policies and looked to Western systems. Tis suggests horizontal interdependencies in that CEE countries wanted to "emulate the apparent success of models used in Germany and Austria" (Rechel and McKee 2009, 1187) by moving towards "a Western-style insurance system" (Jacoby 2004, 48).

## **3 The Political and Economic Context of Croatia**

After the end of the Second World War, Croatia became a federal republic within Yugoslavia governed by the Communist regime led by Tito. In 1948, after Tito–Stalin split and its culmination in 1952, Yugoslavia embarked on a diferent path of socialism than that of the other countries in the Eastern bloc—the path of self-management socialism (Ramet 2002). Self-management introduced social ownership and self-managing interest communities (SIZ—*samoupravne interesne zajednice*) in which workers could participate in the management of their enterprises. Tis type of governance was also refected in public services such as healthcare, education and social welfare (Ramet 2002). However, there was a discrepancy between the formal authority of SIZs and the authority of provincial and republic governments. Tis "democratic centralism" (Sunić 1995, 67) "in which two parallel structures exercised jurisdiction in the same area, was mocked as SIZ-ophrenia" (Ramet 2002, 9).

After Tito's death in 1980, diferent ideas emerged on the future organisation of Yugoslavia. Serbia wanted to maintain its hegemony in a centralised federation, while Croatia and Slovenia were favouring a loose confederation or independence (Žižmond 1992). Moreover, a severe economic crisis plagued Yugoslavia during the 1980s and already in 1981, foreign debt amounted to \$19.2 billion (Ramet 2002, 10). Te economic crisis came to a head at the end of the decade. GNP and labour productivity decreased, while unemployment increased and hyperinfation ensued (Žižmond 1992). As the pressures of political and economic crisis were mounting, the Communist Party was losing its legitimacy and was pushed into holding the frst multiparty elections in 1990. Te elections established a new party in power, the Croatian Democratic Union (*Hrvatska demokratska zajednica*, HDZ). HDZ was a pro-reform rightwing party oriented towards breaking with the communist past, independence for Croatia, democratisation, liberalism, pluralism and transition to a market economy (Milanović 2011; Dunatov 2010).

Due to the perceived political and economic crisis, HDZ opted for the introduction of a semi-presidential political system with a strong presidential fgure (Boban 2008). After coming to power, HDZ purged policy venues and appointed people loyal to them in public administration and the judiciary, facilitating even more control over policy processes at the time (Ramet 2013, 37). Terefore, the transition to liberal democracy was severely limited. In the 1990s, Croatia was a defected democracy with limited pluralism, dominance of the president and widespread corruption (Ramet 2010, 259).

Te ground was set for the dissolution of Yugoslavia when Croatia and Slovenia proclaimed independence in 1991. However, the dissolution was followed by the war in 1991 which lasted until 1995. At the end of 1991, Croatia lost control of 30% of its territory, 40% of industry was destroyed, income from tourism dropped by about 80% and infation and unemployment increased (Ramet 2013, 38). Between 1991 and 1993, Croatia experienced a decline of about 31% in GDP (World Bank 1995). Moreover, a large population of refugees from occupied Croatian and Bosnian territories came to Croatia (Hebrang et al. 2007). For all these reasons, the fnancial revenue of the healthcare system dropped considerably. Compared to 1991, the revenue dropped by 62% in 1992 (Hebrang 2015).

### **4 Financing of Healthcare in Croatia: Historical Background**

Te history of SHI in Croatia can be traced back to the times when Croatia was part of the Austro-Hungarian empire. A frst form of SHI was introduced already in 1891, although its coverage was very limited. Afterwards, SHI went through a number of changes in 1907, 1922 and 1937, mostly expanding the coverage to a wider range of workers and including more health services (Zrinščak 2003). In Yugoslavia, Croatia followed a similar path to other CEE countries by abolishing SHI and instituting tax fnancing from local, district, republic and federal levels (Parmelee 1985, 720). However, SHI was reintroduced, thus supporting a new "third way" of self-management socialism.

Te model of social ownership and self-management was most explicitly defned after the passage of the 1974 Constitution, which was followed by the 1976 and 1980 Healthcare Acts (Ivčić et al. 2017). During that time, healthcare was heavily decentralised and inefcient. It was organised on socialist principles of expanding health services, "free" healthcare and solidarity (Šarić and Rodwin 1993; Džakula et al. 2012). Tese formal goals could not be achieved with the fnancial organisation of a system governed by 113 self-managing interest communities which acted as SHIs, collecting funds according to the Bismarckian model of payroll taxes (Parmelee 1985). In theory, "every local and republican selfgoverning medical unit managed its own afairs, with a high level of fnancial independence" (Džakula et al. 2012, 69). However, there was a discrepancy between the self-management component and the infuence of the Communist Party.

Parmelee (1985, 725) notes that "SIZ professional administrative stafs are almost constantly accused of usurping the decision-making prerogatives of the self-managed SIZ assemblies, and acting little better than the state bureaucrats they were meant to replace". Tere was

a fairly established practice of political interfering in the internal organisation of health care institutions and in particular with personnel policy. It has often been the case that these authorities impose administrative staf who have been unable to fnd employment elsewhere. (Popović and Škrbić 1968, 89)3

Te power and infuence of the Communist Party permeated all levels of governance, be it local, republic or federal with only a few exceptions. Tis kind of system was also refected in the position of medical professionals in the healthcare system. Following the establishment of the

<sup>3</sup>Troughout the chapter, all Croatian quotes were translated.

Yugoslav state, all professional associations (except lawyers) were disbanded due to ideological reasons. As a representative of the professional interests of doctors, the Croatian Medical Chamber was banned and its properties seized already in 1946 (Ivaničević 2015). Te position of the doctor was equated with the position of a regular worker in service to society, the regime and the economy. With the expansion of selfmanagement in the late 1960s and early 1970s, medical professionals could exert more infuence in medical facilities and SIZs, mostly because of the asymmetry of information between users and providers (Ivčić et al. 2017), but again no major changes could have been made without the approval of the Communist Party.

We [*the doctors*; my emphasis] assessed this attitude towards the tendency to develop self-government as quickly as possible in our country as negative, and still believe that the responsibility for healthcare management often goes hand-in-hand with a certain monopoly which ignores the growing demands of reducing the dominance of public administration in decision-making. (Ferber and Knežević 1969, 133)

No one could work in SIZ, there were some exceptions, if one was not a member of the Party because everything was conducted through the Party so that there would be no rebellion or any protests. One received a position along the Party line from which one had some material gain and had to be *silent regarding the issues* (emphasis mine). On paper … you said it yourself, patients, the population was electing people to those governing bodies, but the list had only Communist Party members. (Hebrang, interview 2019)

Doctors voiced concerns about low salaries, difculty to fnd employment despite the lack of doctors in the system, emigration of doctors, accusations of taking bribes and difculties of retired doctors (Ferber and Knežević 1969). It was emphasised that "doctors should be more involved in health policy, especially in relation to stafng and funding, which, in the current framework, makes full self-management impossible" (Ferber and Knežević 1969, 142). Te result was a disoriented system which suffered from disorganisation and bordered on anarchy (Džakula, interview 2019; Mastilica, interview 2019). Moreover, there were huge discrepancies between diferent regions in terms of fnancing, quality and access to services (Šarić and Rodwin 1993; Chen and Mastilica 1998). With the culmination of the economic and political crisis these problems became highly apparent and a radical reorganisation of the healthcare system followed in the 1990–1993 period. Prevailing dissatisfaction among medical professionals with their position within the healthcare system pushed them to become involved in politics where they fgured prominently, being key actors during the transition period and healthcare reforms of the early 1990s.

## **5 The Croatian Healthcare Reform Process 1990–1993**

In the 1980s, the economic crisis severely decreased the pooling of funds for healthcare. During this time, the discourse in healthcare slowly started to change. Several publications started to introduce new terms into healthcare discourse which also permeated the media, such as "copayments", "supply and demand", "efciency" or "cost benefts" (Ivčić et al. 2017; Ivčić and Vračar, interview 2020). Te period of the late 1980s and early 1990s marked a turning point in Croatian healthcare for which only 3.6% of GDP was allocated (Hebrang 1990a, 10). In 1989, a new communist government led by Ante Marković initiated economic reforms to curb growing infation and national debt while at the same time introducing aspects of a market economy and limited privatisation. Te new minister of health, Mladen Radković, started to prepare a major reorganisation of healthcare and initiated a programme called Basis (*Osnova*). Te idea was to introduce individual responsibility for health through an expansion of co-payments and to reorganise the fnancing of the system by establishing a two-tier insurance. Te frst tier would provide fnancing through a general government budget for basic health services and a second tier would act as a worker's additional health insurance which was to be covered by the employer (Radković 1990, 7).

Although the discourse was changing and gaining traction, the reform proposal from the new communist government encountered many obstacles. "Te reorganisation of SIZs has been under discussion for two years and all these attempts have shown the egoism of the municipalities expressed in the demand for having their own SIZ, that much energy, paper and money has been spent on elaborating the new organisation" (Cvitkušić 1990, 1). Yet, nothing was implemented after the new government came into power (Cvitkušić 1990). Individuals who were benefting from the existing organisation of the system and who were organised around the Communist Party at diferent government levels blocked the reform attempts. Despite the mounting problem pressures and an obvious need to reform the system, it was not possible to overcome the resistance, thus preventing any large-scale reorganisation of the system. Once HDZ won the elections and came to power, the new government had diferent ideas on how to reform the system. Te *Osnova* programme was discontinued and sufered a dismal fate.

#### **5.1 The Doctors Enter Politics Mechanism**

Te elections and breakdown of the Communist Party and its infuence served as a trigger for the whole causal mechanism underpinning the changes. It created a space for new actors and ideas in Croatian policymaking. Te doctors enter politics mechanism explains how doctors entered the political stage and occupied key political positions in the healthcare domain. It consists of several elementary causal mechanisms, more specifcally an emotional and a rational orientation. On the one hand, the emotional orientation explains that the dominant feelings of dissatisfaction, frustration and marginalisation of medical professionals drove them to become involved in politics.

As Poljak notes in an interview:

Since the war [Second World War], the Croatian Medical Association has had to serve the ruling politics and even work against the interests of its members. One received support from above [Te Communist Party] in proportion to one's obedience … Tanks to the new political opportunities, the Association now has the freedom to oppose and criticise. (Šimunić 1990, 5)

On the other hand, the rational orientation explains the strategic action in which medical professionals seized their opportunity, joined the HDZ and consequently occupied powerful positions in the government, enabling them to exercise infuence on health policy. "Although unprepared for the nuances of politics and governance, doctors flled the political vacuum by replacing the ousted lawyers in the new government" (Blaskovich 1997, 81). A considerable number of medical professionals occupied positions in the government, parliament and municipal councils. Medical professionals not only began to be involved in the formal decision-making venues, but also infuenced policymaking through the Medical Association and a newly formed Croatian doctor's union which was established in 1990 (M. V. 1990, 9). Te decision-making process in healthcare shifted from being dominated by the Communist Party to being dominated by doctors. Tus, the doctors had assumed the most important political positions and had a signifcant infuence on healthcare decision-making, for example, in the Ministry of Health and later in the national health fund. Besides occupying political positions, doctors acted as an epistemic community (Haas 1992) which had the knowledge and competence to deliberate on the healthcare system, its problems and policy solutions.

A large number of doctors ran for the state parliament and for the municipal councils, parliaments. Why? Te doctor is ahead of the great majority in his social environment and understands what is going on and how. Many got involved in it and I encouraged them strongly. (Hebrang, interview 2019)

Count up how many HDZ ministers are medical doctors, all right. From Mate Granić, I mean the medical lobby and by medical lobby I mean top surgeons … professorial level doctors. (Stubbs, interview 2019)

It is no coincidence that in the political life of Croatia, and I do not know if it is the same in other transition countries, but it seems to me that there are many medics, doctors. Even at the highest functions. You see, there was Foreign Minister Mate Granić, a doctor, a professor of the faculty and today Reiner … doctors were very much involved as ministers. (Mastilica, interview 2019)

#### **5.2 The Old System Departure Mechanism**

Although many doctors were involved, the most important one was the Minister of Health Andrija Hebrang, as his ideas were largely implemented. Hebrang was a person with a turbulent history and a strong resentment towards the communist regime.4 Te old system departure mechanism explains the frst changes which were introduced in the health system. It consists of an emotional and a rational orientation of policy actors. At the time, negative feelings towards and perceptions of the communist healthcare system were prevalent in the government, parliament and media outlets. Tis meant that the policies under consideration were quite dissimilar to those of the communist system (Anonymous expert, interview 2019; Mastilica, interview 2019). Te media and policymakers discussed the policy failures of the communist system, evaluating its policies not only through emotional, but also through rational appeals, such as lack of accountability, coordination and expenditure controls, heavy involvement of politics or corruption. Te following quotes illustrate this:

Te self-management dislocated way of fnancing and decision-making, as well as the incompetence of the staf, is the cause of anarchy in the management of the Croatian healthcare system, which is why we did not achieve an adequate health standard. (Hebrang 1990b, 6)

Te legacy of the old system is still in people's minds. Te term used in this law proposal is actually reminiscent of the term healthcare worker. And that sounds just like a port worker, a railroad worker, a foreman and so on and so forth … I ask that doctors should not fall under that Bolshevik phrase. (Štanfel 1993, 6)

Moreover, the crisis in healthcare meant that the policies had to be based on rational solutions that would alleviate its defciencies, most importantly contain its costs. Te mix of the rational and the emotional orientation thus produced policies dissimilar to the communist system

<sup>4</sup>Hebrang's father Andrija Hebrang had advocated Croatian national interests and was accused of Croatocentrism (Banac 1988, 90). He was thus perceived as a threat to Tito who ordered his imprisonment and assassination.

which were at the same time based on the rational deliberation of the situation in which

the money to cover health services and accumulated debts was non-existent. Rather, the Communists had funded healthcare by issuing government bonds which had no real value. Te value was given by the legitimacy of the Party, but as soon as the Party lost legitimacy it became obvious that debts had to be paid with sound money for the system to survive. (Hebrang, interview 2019)

According to Hebrang, the only rational and fast solution for resolving the situation and departing from the previous system was to introduce centralisation which "goes in terms of fnancial control, primarily fnancial because we have come into the situation that without it, we would not have a chance for better days" (Hebrang 1990a, 10). Terefore, the authority of SIZs over fnancing was disbanded, users and providers no longer had a say in the way fnancing of healthcare was to be conducted in their municipalities and the responsibility for fnancing was taken over by one fund called Te Republic Health Fund managed by another physician, Mate Granić. Former SIZ authorities now worked under strict control of the SHI fund which established an equal contribution rate5 for payroll taxes across the whole country.

SHI lost its Bismarckian principles of decentralised multiple funds and corporatist governance (however limited it was in Yugoslavia) and the government had strict control of the SHI fund by appointing its directors and board of directors; at the same time, it had the authority to dismiss them (Vončina et al. 2007, 147; Pezo, interview 2019). Individuals in SIZ assemblies could no longer rely on the power and legitimacy of the Communist Party to push their agenda and block the reforms. Coupled with the government's heavy determination towards reforming the sector, the context of policy processes changed and enabled a fast-sweeping reform. A new law was passed after only four months after new government came into power.

<sup>5</sup> Set at 11.6% in 1991, changed to 20% in 1992 and 16% in 1999 (Croatian Parliament 1991, 1992; WHO 1999).

#### **5.3 The Seeking Solutions Abroad Mechanism**

Centralisation and rationalisation of fnancing was only a stepping stone which prevented the collapse of the system. Once this burning issue was resolved, policymakers oriented themselves towards introducing new policies which would bring the healthcare system into line with the perceived successes of Western European countries.6 Here the seeking solutions abroad mechanism, consisting of an emotional, a rational and a comparative orientation, can provide an explanation. Te mechanism explains how policymakers wanted to move away from the communist policies by scanning the policy solutions abroad and implementing those that were best suited for the Croatian context. Te emotional orientation explains the role of the media and the policymakers in propagating dissatisfaction with the communist system and a need to further depart from it. It can be said that such a perception of the communist system was prominent in all three causal steps.

Moreover, the comparative orientation refects the position of policymakers to emulate the perceived success of the West, look for policy solutions elsewhere and draw lessons from other countries. Here it is important to note that the Minister of Health (Andrija Hebrang) had, as a physician in Yugoslavia, travelled to Western countries to attend medical conferences. During his time abroad, he learned about the organisation of healthcare in these countries, most notably the USA and Germany (Hebrang, interview 2019). Learning from his experiences abroad, the minister favoured a radically diferent organisation of healthcare fnancing which would introduce neoliberal7 policies based on market principles which had already been introduced in many European countries (Šimunić 1990, 7).

<sup>6</sup>A common theme in the newspapers at the time and in the interviews with relevant policy actors was the perception of Western European healthcare systems as a success story and a role model to which Croatia has to strive, which entailed introducing policies that would bring Croatia closer to these countries.

<sup>7</sup>Neoliberal policies in healthcare include commodifcation of health services, individual responsibility for health, introduction of market principles such as privatisation, competition and private insurances, deregulation and decentralisation (McGregor 2001; Rotarou and Sakellariou 2017). Obviously not all neoliberal ideas were implemented in Croatia.

As other CEE countries were also in the process of transition, the Ministry of Health had set up an ofce for evaluating and comparing the policies which were being introduced in countries facing similar problems. Tis facilitated the process of gathering experiences from other countries. Tus, in the third step, the rational orientation explains how diferent policy solutions were being evaluated according to their costs and benefts and how they could ft into the Croatian context. Te transfer of policy experiences from other countries initiated by the Ministry of Health sought to, on the one hand, avoid the negative consequences of the reforms implemented in other countries and, on the other hand, implement the reforms that proved useful, thus taking into consideration the consequences of alternative courses of action.

Te minister's experiences abroad and the designated ofce in the Ministry of Health enabled the process of horizontal policy transfer by evaluating policies in other countries, be it CEE or Western ones, in order to draw positive and negative lessons. For example, the experiences and adverse efects in the Czech Republic (Hebrang et al. 2007, 2), which established a "pluralistic semi-competitive insurance-based system" (Earl-Slater 1996, 16), and the USA, where "small business employees are completely unprotected from the negative side of the healthcare market" (Hebrang 1993, 7), have served as a lesson to approach marketisation and privatisation policies carefully. On the other hand, positive learning stemmed from Western Europe.

Among the many organisational forms that are possible in the fnancing of the health system, we have selected those which have the most favourable ratings in the world based on the experience of others. (Hebrang et al. 2007, 3)

In Europe, there is a sensible combination of the state or SHI funds and private initiatives. Why is that important? Because it brings competition while keeping solidarity. Tat is the most delicate balance a healthcare system should have. Competition increases quality, lowers service prices, and at the same time you have to keep that social component. (Hebrang, interview 2019)

Individual responsibility for health, co-payments and private insurance schemes appeared on the agenda. Despite heavy criticism in the media, a law which increased co-payments up to 10% for selected healthcare services (Kovačić and Šošić 1998, 4) was supported by the government and was passed by parliament in 1991. Te law established co-payments for drugs, visits to primary care, specialists, diagnostic and hospital treatments, among others, while parts of the population such as children or the elderly were exempted (Kovačić and Šošić 1998, 4). Such policies increased the trend of commodifcation of healthcare. A survey conducted by Mastilica and Chen (1998) shows that over half of the respondents had great or very great problems with out-of-pocket expenses. At the same time, the introduction of private insurance schemes was postponed due to the ever-increasing political crisis and the expected dangers of war. Te healthcare system was turned into an integrated military-civilian system. Many doctors were mobilised into the army or were required to serve in reoriented war hospitals. At the same time, the system had to provide healthcare services to civilians in areas not afected by war (Hebrang 2015).

Once the Serbian aggression and advance subsided in the 1992–1993 period, the ground was laid for the formulation of two encompassing healthcare laws. Tese laws established the Croatian Institute for Health Insurance with 21 regional branches, replacing the former Republican Fund and completely eradicating former SIZ bodies. Te perception was that "the state insurance principle … has given the best results in Western countries" (Hebrang et al. 2007, 3). Moreover, a limited space for private insurance market operations was introduced. One form of private insurance was supplementary, covering additional and better quality of health services on top of the mandatory SHI, for which citizens could apply in the private insurance market. Te second form, according to the German model, was entirely private health insurance. Eligibility for coverage of citizens was determined by an income above a threshold specifed by the Minister of Health and entailed opting out from the mandatory SHI.

Tus, two types of private insurance were reserved for people with high incomes who wanted to have a better standard and coverage of health services. Te government introduced market policies and limited competition in the private insurance market while at the same time preserving solidarity and access to healthcare for a majority of population which was insured by mandatory SHI operating under the Croatian Institute for Health Insurance. Again, both the comparative and the rational orientation is evident as the Ministry of Health tried to avoid "the traps of sudden privatisation, which has yielded very poor results in the healthcare of some post-communist countries" (Hebrang et al. 2007, 3). Te goal was to slowly expand the private health insurance market which would have an increased role in the years to come.

My idea was to make eighty to ninety percent of the system social, and ten to twenty percent which would go to the market in order to level it all together, and that's why supplementary insurance and co-payments were the frst attempts, and it worked until the 2000s when we lost the election to the leftist parties. (Hebrang, interview 2019)

Te goal was never achieved and the private insurance market was only used by few people. For instance, almost ten years after the reforms, "in 2002, private health insurers reported annual revenues of HRK 962 million (EUR 130 million) or roughly 6% of total health expenditure" (Vončina et al. 2007, 151). However, the institution of a single SHI fund managed to curb healthcare expenditures, pool additional funds and save the system from collapsing. "Te debts of the previous system have been eliminated and in 1995 a surplus was accumulated to pay for new capital equipment" (WHO 1999, 46).

Apart from implementing horizontal policy transfer, the Ministry of Health sought help from international organisations as well. However, due to the severe political crisis the presence and infuence of international organisations were limited.8 Tese organisations were extremely careful not to engage in healthcare projects in a country that was at war and not recognised by the international community (Stubbs and Zrinščak 2007; Hebrang, interview 2020). Tis was especially the case with the World Bank, IMF and the EU PHARE programme because they ofered fnancial aid for development projects. Ironically enough, this extreme

<sup>8</sup>However, it is important to note that a lot of international organisations and NGOs were indeed present in the early years of transition, such as the UN, Red Cross and alike which were very active in ofering humanitarian aid (Stubbs and Zrinščak 2007).

vulnerability actually inhibited the presence of international organisations in the country.

Nevertheless, the minister wanted advice for the initiated reforms and the only possible venue he could turn to was the WHO because its expertise is not reliant on fnancial aid or strong conditionalities (Kaasch 2015; Deacon 2007). Te WHO and its general message of abolishing health inequalities, achieving universal coverage, solidarity and risk pooling was a perfect match since the minister did not want to risk solidarity with the introduction of market and private initiatives. Te problem was that the WHO recognised only Yugoslavia as a partner and not Croatia. Tus, the minister used informal connections to meet with the WHO president of the European Regional Ofce, Jo Asvall, who agreed to set up a small ofce in the Croatian Ministry of Health in 1991.

[Asvall] tasked one of his men to communicate with us: "I said I don't need you for money, I only need you for advice. I'll tell you a problem, you give me advice. Whether or not I will listen to it depends on the situation." And this man was phenomenal, coming once a month for two to three days. I would invite my co-workers, we talked and fltered out a lot of our uncertainties. (Hebrang, interview 2019)

A working group consisting of domestic and WHO experts was established to work on the formulation of new healthcare laws which were passed by parliament in 1993. Other international organisations were not present in the reform process, while the World Bank only became involved in 1995, supporting the initiated reforms with fnancial aid and a healthcare project. Te World Bank and the WHO praised the introduced reforms and even admitted they could serve as a model for other Eastern European countries (Hebrang et al. 2007; World Bank 1995).

#### **5.4 The Anti-communist Backlash Mechanism**

To recapitulate, the dynamics of the reform process can be explained by three complex causal mechanisms: the doctors enter politics mechanism, the old system departure mechanism and the seeking solutions abroad mechanism. Te frst mechanism explains how the prevailing dissatisfaction towards communist policies drove doctors to get involved in the reform process by joining the new ruling party and occupying key political positions in the healthcare domain. Te second mechanism explains the frst part of the reform process in which the newly introduced policies were as dissimilar as possible to the communist system while at the same time providing a rational foundation for resolving the defciencies of the healthcare system. Te third mechanism explains the second part of the reform process, namely, the departure from the negatively perceived communist legacies by looking to Western Europe, drawing lessons from other countries and implementing policies which suited the Croatian context according to their perceived costs and benefts. Tese three complex causal mechanisms form the combined mechanism of anticommunist backlash. It explains how the prevailing dissatisfaction with the communist healthcare system, particularly among medical doctors, pushed reforms in the opposite direction and initiated a search for "noncommunist" policies abroad.

#### **5.5 Hybridisation of Healthcare Financing as the Output of the Reform Process**

Te output of the described reform process was a hybrid model of healthcare fnancing and its regulation. SHI insurance accounted for most of the revenue in healthcare (93% in 1994), more than in 1980 (74%). A negligible number of people signed up for the two types of private insurance, while co-payments remained constant at around 2%. However, the revenue from co-payments is understated and "not necessarily included in the national accounts" (WHO 1999, 13). Other sources were subsidised by the state budget (prevention, education, statistics, etc.) and county budgets (special programmes and healthcare for elderly peasants) (Kovačić and Šošić 1998, 4). Table 8.1 represents the sources of income for Croatian healthcare during the 1990s.

Although the revenue from SHI increased, there are arguments to be made that Croatia actually moved away from the SHI model,


**Table 8.1** Financing sources for the Croatian healthcare system

Source: WHO (1999, 13)

introducing Beveridgean and neoliberal principles. First, self-governance where users and providers negotiated the scope and price of healthcare services through SIZ assemblies was abolished. Second and inextricably linked to the frst argument, the decentralised and fragmented health system consisting of 113 health funds was abolished in favour of one national SHI fund, the Croatian Institute for Health Insurance, which holds a monopoly in the SHI market. In theory, the Croatian Institute for Health Insurance established a form of corporatist governance as the managerial board consisted of representatives of employers, medical professionals and patients (Croatian Parliament 1993).

However, professional organisations or unions in Croatia are largely underdeveloped and their infuence in decision-making is marginal. Te medical professionals' organisations are an exception (Škaričić, interview 2019; Rukavina, interview 2019; Radin, interview 2019; Belina, interview 2019). Tus, corporatism is severely limited and in practice the Croatian Institute for Health Insurance is just an extension of government politics as the fund implements policies already agreed upon at the governmental level (Vončina, interview 2019; Anonymous expert, interview 2019). "Te observation that everything happens in one place is only partially correct. Everything is happening in one place, which is the government, i.e., the Ministry of Health. Te Croatian Institute for Health Insurance only implements a specifc policy" (Hebrang 1996, 4).

Tird, although the system was stabilised, once the debts started accumulating again, the government has, more often than not, covered these debts through government budget transactions.

Every year we are 2 to 3 billion kuna [HRK, Croatian currency] in defcit and while I was a minister, I always covered it from the budget so I made this mixed Beveridge model … and then I would come to the Government session: now look people I have rationalised this, introduced controls and records. I did everything and cannot go any further. Expenditures are higher because prices are expensive … Every year I have managed to transfer 2 to 2.5 billion HRK from the budget to the health fund and this is how we covered the debts. (Hebrang, interview 2019)

Terefore, incentives to rationalise and curb expenditures are lacking (Anonymous expert, interview 2019) since the government eventually pays for the accumulated debts from the general budget—a form of retrospective tax fnancing (Mossialos et al. 2002, 69). Tus, the "principle of stable contribution rate" (Giaimo 2001, 351) in an SHI, where equalising revenue and expenditures should fgure prominently, is actually non-existent.

Fourth, an explicit basket of services to which insurees are entitled was not defned (Vehovec, interview 2019). Rather similarly to general taxation systems, the Ministry of Health and the Croatian Institute for Health Insurance produced an annual health plan containing "regulations on health insurance entitlements" which had to be approved by parliament (WHO 1999, 10). Although limited, neoliberal principles were also introduced in the healthcare fnancing dimension, such as a move towards individual responsibility for health, setting up healthcare services as a commodity by instituting co-payments and self-medication payments to private providers and opening a private insurance market.

Tus, a hybrid system of healthcare fnancing incorporating Beveridgean, Bismarckian and neoliberal principles was created. Practically, the only diference between a true Beveridge system was that the funds were mainly collected by payroll taxes and pooled into an extrabudgetary SHI fund. Te arguments mentioned above confrm Stefen's (2010) conclusions which show that categorisations of healthcare systems are rather difcult and that every country has its own specifc policies borrowed from various healthcare models.

### **6 Discussion and Conclusions**

Tree causal mechanisms explain the perceptions and interpretations of key policy actors and consequently their action orientation towards reforming the health system in Croatia: the doctors enter politics mechanism, the old system departure mechanism and the seeking solutions abroad mechanism, which together form the combined causal mechanism of anti-communist backlash. It explains the dominant perception of communist policies as a failure, particularly by medical doctors, and searching for new policy solutions abroad. Te mechanisms are composed of diferent elementary causal mechanisms, namely, emotional orientation (prevailing feelings of dissatisfaction towards the communist regime and its policies), rational orientation (strategic actions of doctors, cost and beneft analysis of new policy solutions) and comparative orientation (emulating the perceived success of Western European countries, avoiding policy failures of other CEE countries).

Taken together, this has created a hybrid system of healthcare fnancing and its regulation based on Bismarckian, Beveridgean and neoliberal principles. In sum, analysing the causal process by dividing it into several causal mechanisms and linking them together into a complex causal mechanism proves to be a useful tool for tracing and explaining the reform process. Moreover, the causal mechanism approach can work across and link diferent theoretical traditions which aim to explain policy changes, thereby contributing to the existing literature.

Te mechanisms in this chapter were inspired by theories and concepts such as punctuated equilibrium theory, policy transfer and epistemic communities. Despite the external pressures of a growing economic crisis during the 1980s, healthcare policy in Croatia remained stable, leading to the accumulation of policy errors which created a mismatch or friction between a changing environment and unchanging policy (Zehavi 2012; Baumgartner and Jones 2009). Baumgartner and Jones (2009, 25, 31) argue that increasing policy failures and increasing problem pressures modify policy images or shared "public understandings of policy problems" which are then coupled to policy venues or "institutions or groups in society" that "have the authority to make decisions". Tis process leads to positive feedback which punctuates the equilibrium of policymaking and results in major policy change. In Croatia the same process can be observed.

Policy change in Croatian healthcare was only possible once the policy monopoly (Baumgartner and Jones 2009) of the Communist Party at diferent levels of government was dissolved. After the communists lost the elections in 1990, a causal mechanism which underpins the changes was triggered. Doctors who were unsatisfed with their position joined the new ruling party and occupied relevant policy venues, while veto actors in the municipalities lost their legitimacy and could no longer exert their infuence and block the reforms. Te exogenous crisis and endogenous problem pressures within the healthcare system were only a sufcient condition for change, while elections and the installation of a new party in the power structures provided a necessary condition for change. Indeed, Walgrave and Varone (2008, 370) argue that "if parties adopt a new policy image and control the new institutional venue, then it will translate into a major policy change". After the elections, the negative policy image was able to be coupled with new policy venues which pushed for a reorganisation of the healthcare system.

Doctors fgured prominently in the reforms, acting like an epistemic community, "a network of professionals with recognized expertise and competence in a particular domain and an authoritative claim to policy relevant knowledge within that domain or issue area" (Haas 1992, 3). Te infuence of such an epistemic community was obvious as doctors did not have any competition in the policymaking feld. Moreover, doctors occupied key political positions and were crucial actors in formulating new healthcare policies. As they wanted to move away from communist legacies and towards the perceived successes of Western policies, policymakers were involved in policy transfer, a "process by which knowledge about policies, administrative arrangements, institutions and ideas in one political system (past or present) is used in the development of policies, administrative arrangements, institutions and ideas in another political system" (Dolowitz and Marsh 2000, 5). It can be either voluntary or coercive, revolving around a "continuum that runs from lesson-drawing to the direct imposition of a program, policy or institutional arrangement on one political system by another" (Dolowitz and Marsh 2000, 13).

Besides setting up an ofce within the Ministry of Health tasked with drawing policy lessons from other CEE countries, the Minister of Health already had knowledge about Western policies acquired during his travels abroad. Te transfer of policies was completely voluntary and consisted of mostly horizontal lesson drawing from other countries. It was both negative (e.g., CEE, USA) and positive (Western Europe), which served to either avoid the mistakes of others or to add potential policy tools to the repertoire (Klein 1997, 1270). Moreover, it was also a "symbolic act whereby politicians seek to enhance their status, credibility or modernity" (Stone 2017, 61). Terefore, contrary to the literature which states that international organisations exploited the policy vacuum in CEE countries, the Croatian case ofers evidence that international organisations were not crucial for the reform process at all. Although the WHO was involved in the formulation of new healthcare laws by providing advice, it has not enforced any conditionalities. Te World Bank, however, was involved in the reform process only after the new laws were passed and agreed on health projects with the government in 1995 in order to support the new government agenda.

Such a comparative orientation among key policy actors established a mix of policies, thus radically changing health fnancing and its regulation from self-management and Bismarckian principles towards hybrid policy. Te Croatian case has indeed demonstrated that policy transfer plays a major role in hybridisation of healthcare systems (Schmid et al. 2010; Stefen 2010). As policymakers respond to problem pressures by searching for compatible solutions elsewhere, they develop "distinct policy responses" and "new elements that are not system specifc" (Schmid et al. 2010, 460). Croatian policymakers considered policy solutions irrespective of their ideological background except for dismissing anything resembling communism. Terefore, a hybrid model consisting of Bismarckian, Beveridgean and neoliberal principles of fnancing was introduced. Although fnancing from SHI contributions has expanded, its regulation was heavily in line with the Beveridgean system. Te state has assumed a major role in the regulation of fnancing, controlling the SHI fund. At the same time, neoliberal principles were introduced. Individual responsibility for health was introduced by establishing formal co-payments and self-medication payments (Kaminska and Wulfgramm 2019), while market principles were introduced in the health insurance feld by establishing two forms of private insurance.

To conclude, the Croatian case provides insights on the dynamics of reform in healthcare policy in one CEE country by using process tracing as a method and establishing causal mechanisms underpinning the changes. It provides evidence that key actors in the reform process were domestic doctors, similarly to other CEE countries (Kaminska et al. 2021), thereby supporting the literature which states that healthcare reforms in CEE were mostly endogenous (Rechel and Mckee 2009; Sitek 2008; Radin 2003; Roberts 2009). On the other hand, it contradicts the literature which claims that exogenous actors such as international organisations were instrumental in directing the reform processes in CEE countries (Deacon et al. 1997; Cerami 2006; Kaasch 2015; see Nemec and Kolisnichenko 2006). Nevertheless, the WHO and the World Bank were present at diferent stages of the reform process, mostly supporting the initiated reforms rather than initiating themselves. Although the impetus came from the inside, policy ideas were infuenced by the experiences of other countries, be it CEE ones or Western ones like Germany, thus supporting the literature which states that CEE countries wanted to emulate policy models used in Western countries (Rechel and Mckee 2009; Jacoby 2004), Terefore, the chapter contributes to the literature on healthcare policy changes in CEE during the transition period and provides a piece of the puzzle which helps us understand how and why changes in healthcare happened in those countries. Furthermore, it contributes to future research on international interdependencies in social policymaking in CEE countries.

### **References**


———. 2015. *Hrvatski sanitet tijekom srpsko-crnogorske agresije na Republiku Hrvatsku 1990–1995* [Croatian Medical Corps During the Serbian-Montenegrin Aggression on the Republic of Croatia 1990–1995]. Zagreb: Medicinska naklada: Udruga hrvatskih liječnika dragovoljaca 1990–1991.


*i Razgradnja*, ed. Chiara Bonfglioli and Boris Koroman, 99–128. Pula-Zagreb: Srednja Europa.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **9**

## **Against All Odds: Introducing Social Health Insurance in Albania. An Actor-Centred Approach to Causal Mechanisms**

**Ertila Druga**

## **1 Introduction**

Te World Bank is among the most infuential organisations in international development. Research is dominated by a view of the World Bank as a coercive actor, constraining developing countries to accept and adopt its prescribed policies, and much of it emphasises the World Bank's infuence in the form of conditional structural adjustment loans (Easterly 2003; Larmour 2002). Less academic literature focuses on its noncoercive side, in persuading countries, recipients of development

E. Druga (\*)

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

University of Bremen, Bremen, Germany e-mail: ertila.druga@uni-bremen.de

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_9

assistance, to adopt particular policies (Bazbauers 2018, 239). Tis chapter attempts to fll this gap by investigating the introduction of Social Health Insurance (SHI) in Albania.

Albania was the last communist country in Central and Eastern Europe (CEE) to embark on the road of transition. In need of fnancial aid and expertise to overcome the crisis, stabilise the macroeconomic situation and build market institutions, the Albanian government turned to the World Bank Group and the International Monetary Fund (IMF). After conducting a dedicated mission in healthcare, in March 1992 the World Bank (WB)1 delivered a report presenting its recommendations for the future of the Albanian healthcare system. In response to the domestic reform agenda that was dominated by a liberal approach, advocating privatisation of healthcare services and the introduction of health insurance,2 the WB suggested keeping the existing model based on general tax revenues and only introducing a scheme for social health insurance at a later date (World Bank 1992a). Notably, the Albanian government did not accept this recommendation and, in June 1993, presented its strategy for the healthcare sector, aiming, among other reforms, at the "introduction of a scheme of health insurance" (Nuri 2002). Despite the WB's early reservations, the Albanian parliament adopted the SHI law in October 1994. In the years 1993–1994, the World Bank became involved in policy dialogue and analytical work with the government (Shehu 2012; Shehu, interview 20193 ; Nuri, interview 20194 ), and it sought, with some success, to infuence the reform process. Tus, the WB, in the role of a facilitator, was able to convince the government to follow a simpler SHI model than the one previously intended—with lower contribution rates and restricted coverage of healthcare services—and assisted in preparing the draft legislation. Considering this background, this chapter

<sup>1</sup>Troughout this period (1991–1994) the WB and the IMF missions in Albania worked together closely.

<sup>2</sup>Tere was a general perception that the introduction of social health insurance itself would favour the liberalisation and marketisation of the health sector.

<sup>3</sup>Tritan Shehu, Former Minister of Health of Albania, April 1992–January 1994. Face-to-face interview conducted on 14 October 2019.

<sup>4</sup>Besim Nuri, Former Deputy Minister of Health of Albania, May 1992–August 1994. Face-to-face interview conducted on 13 June 2019.

addresses the following questions: How did the World Bank infuence the policymaking process, and how much was it able to achieve?

Te presence of various external actors in the post-communist CEE region makes it an ideal location to test and develop better explanations for external infuences (Jacoby 2005, 623). However, there is a rationale behind the selection of the Albanian case. It contradicts the usual perception of the international organisation's conditionality and its alleged coercion of countries to accept its policy prescriptions. Albania was the least developed country in the region and depended on the WB's fnancial leverage. It had already accepted all policy prescriptions and future reforms proposed by the WB and the IMF (Bezemer 2001, 1–2), except for the one in health fnancing.

To throw light on the role of the World Bank in the Albanian reform process, I rely on actor-centred institutionalism (Mayntz and Scharpf 1995) and investigate "the interactions among purposeful actors" (Scharpf 1997, 1). Te actor-centred institutionalism framework has been previously used by scholarship on welfare state reforms (Adascalitei 2012; Cook 2007; Aidukaite 2009) and healthcare reforms in the CEE region (Ovseiko 2009; Sitek 2010). Te analysis focuses on the dynamics of the actors' interaction and aims at identifying the causal mechanisms behind the process. In examining actors' behaviour, this approach emphasises the causal role played by their interpretations, ideas, and beliefs.

In this chapter, I focus on the mechanism of *transnational cooperation* to elucidate the role of the World Bank as a transnational actor in introducing SHI in Albania. In a nutshell, the mechanism explains the successful cooperation during the formulation of the law after the non-successful cooperation during the agenda setting stage. In contrast to the often researched "coercive" aspect of transnational cooperation, this study aims at expanding the understanding of its "non-coercive" form. In doing so, I draw on insights from international development studies in general and the non-coercive international development assistance of the World Bank in particular. Tough not abundant (Bazbauers 2018), they ofer this study a rich explanatory account on this more discursive understanding of the World Bank's infuence (Escobar 1988; Smith 2008; Bazbauers 2018).

Te fndings show that the WB, in the role of an epistemic community (Haas 1992), pushed for its own prescribed policy in the health fnancing reform in Albania. But it could not change the domestic agenda. So, the WB stayed in the reform game and sought through a set of tools, such as survey missions and technical assistance, to shape the fnal law's formulation. Tis research also highlights the strategy used by the WB in the reform process: its "keep trying" strategy explains the World Bank's attempts to stay in the reform game even though the reform was not in line with its preferences, and after it failed to convince the government of its preferred choice. In the end, the "keep trying" strategy helped the WB to induce the government to accept some of its recommendations in formulating the law.

In brief, the core of my argument in this chapter is that the interaction between the World Bank and the Albanian government was characterised by the dual dynamic of the persuasion power of the former and its inability to impose a specifc policy model. Te outcome of this interaction is explained by a complex causal mechanism (Chap. 1) of transnational cooperation that comprises three steps: (1) Pressures emanating from the economic and political environment forced the Albanian government to anticipate SHI reform in the healthcare sector (elementary causal mechanism of *rational orientation*). (2) Te World Bank, invited by the Albanian government to support the country fnancially and technically, proposed the draft on future healthcare reforms (elementary causal mechanism of *normatively embedded calculatory orientation*). (3) Even though the government did not accept the advice of the WB in preserving the old fnancing model, the WB stayed in the reform game adapting its preference to the domestic choice of policy model (elementary causal mechanism of *refective orientation*) and shaping to some extent the fnal policy formulation.

Tis study contributes to the scholarship on international development practice and the use of the World Bank's non-coercive instruments, enriching the scarce literature on the role of the World Bank as a transnational actor in the processes of healthcare fnancing reforms in the CEE region. Importantly, this chapter adds to the scholarship on Albanian healthcare reforms and more broadly to the scholarship on healthcare politics in post-communist countries of CEE.

In the following section, I introduce the literature review. Next, the chapter proceeds with the section on methods and data, an overview of the politics of the early transition in Albania, and the case study. It concludes with a discussion.

### **2 The Role of Actors in Welfare State Reforms in Post-Communist CEE: Domestic Actors Versus the World Bank**

Te combination of economic crises and the democratic transition in the post-communist region ofered the opportunity to introduce policy changes and led to a wave of health policy reforms (Preker et al. 2002; Kornai and Eggleston 2001). Following the market-oriented recommendations of the World Bank and the International Monetary Fund, the CEE countries took steps to implement reforms to improve their health systems' efciency and productivity (World Bank 1993a; Haggard and Kaufman 2008, 2018). As a result, introducing SHI, as a distinct marketoriented move (Nemec and Kolisnichenko 2006), was a fundamental step in reforming the fnancing dimension (see Wendt et al. 2009) of their healthcare systems.

Research indicates that welfare state reforms in post-communist CEE countries followed diferent trajectories, while remaining silent about the mechanisms that contributed to the emergence of social policy confgurations (Adascalitei 2012). In healthcare, several factors explain policy outputs. Te most obvious one is the infuence of International Organisations (IOs), notably the World Bank and the International Monetary Fund, which endorsed the neoliberal ideas of marketisation, liberalisation, privatisation, and decentralisation and called for a reduced role of the state in healthcare (World Bank 1993a). Pursuing this argument, the review in this section draws mostly on the literature on the reforms of the social sector in post-communist CEE countries, highlighting the role of domestic and transnational actors, and their mode of interaction as explanatory factors to dynamics of reforms. Te existent literature on health fnancing reform and SHI in the CEE is not yet systematic, for that reason the scholarship on healthcare reform in the region and from other regions complements the review.

#### **2.1 The Role of the World Bank in Social Health Insurance Reform**

Te existent scholarship on the role of the WB in SHI reform remains contested, thus implying that new research on the topic continues to be of much interest. One group of scholars posits that the introduction of SHI was one of the main interests of the WB (Cerami 2006) and that the latter was able to infuence the process through conditionalities, material leverage (fnancial aid, loans), and ideas (technical assistance and expertise) (Kaasch 2015; Radin 2003; Cerami 2006). Other scholars (Kaminska et al. 2021; Roberts 2009; Sabbat 2010) contend that in some countries, such as the Czech and the Slovak Republics, Latvia, Poland, and Albania, it was domestic actors who shaped the health reform agenda while pushing for their preferred policy option—the introduction of SHI—and frmly rejecting the World Bank's recommendation on retaining the general taxation model. While systematic research on the reform of SHI in the post-communist CEE region is still lacking, other literature ofers complementary insights. For instance, scholarship on pension and social sector reforms in the region amply confrms the World Bank's infuence (Müller 2001; Orenstein et al. 2008) in these processes. Te WB played the role of a "proposal actor" infuencing the information, interests, values, and policy preferences of domestic "veto players" to achieve policy change and used multiple strategies such as inspiring and recruiting reformers to pursue pension privatisation, forming coalitions with domestic supporters to win battles in favour of reform, convincing domestic opponents to support reform, and devising strategies to neutralise opposition to reform (Orenstein 2009, 18, 129). Importantly, the most systematic analysis of the infuential role of the WB and the IMF in healthcare reforms comes from another region, Latin America. In his book on policy difusion of social sector reform in Latin America, Weyland (2006) explains that while domestic policy actors hold formal, institutional veto power, transnational actors provide the legitimate, wellelaborated policy ideas and proposals that domestic actors sometimes lack. He also observes that the WB and the IMF usually try to win domestic support and that they have the capacity to "induce countries to move toward the principles they advocate" when they cannot "impose specifc blueprints and models" (Weyland 2006, 18, 179).

#### **2.2 The Domestic Rationale for the Social Health Insurance Reform**

At the start of transition, countries in the CEE region possessed a strong political will to initiate and implement reforms and limited time to do so (Ovseiko 2009, 25). Most switched from a tax-based to an insurancebased healthcare fnancing model (Marrée and Groenewegen 1997). Arguments about plurality, independence, and competition (Lawson and Nemec 2003), efciency improvement (Rechel and Mckee 2009, 1187; Medved et al. 2005, 75; Deppe and Oreskovic 1996), and distrust of governments rooted in the communist period (Rechel and McKee 2009; Oreskovic 1998) underline this change.

Although research on this matter covers only a few countries in the region, the existent scholarship provides useful information. In the case of the Czech Republic, Nemec and Lawson (2008, 29) explain that "a mixture of considerations" was behind "the rationale for such [a] switch". According to them, the frst consideration was the administrative argument that earmarking taxes for health and sub-contracting their administration to an independent institution (insurance fund) would ofer the advantage of decentralisation and distancing the government from a contentious area of public policy. Tus, separating the new insurance companies from the government administration would improve the quality of the new administration, and the purchaser-provider split would increase the quality of healthcare services. Te second consideration was that of economic efciency which assumed that the switch to an insurance system would have the advantage of reducing pressure on general budgets (Medved et al. 2005, 75), though there was less concern about where the extra resources for health would come from. Last but not least, there were political considerations. Nemec and Lawson (2008, 29) described these as the desire to signal a break with the old regime and considered them to be the most important factor behind the change.

Other scholars posit very similar fndings. Vlădescu et al. (2005) explain that, in Romania, opting for a Bismarckian system was determined by arguments connected with politics, ideology, and technical aspects. After the changes of 1989, everything associated with the old regime was challenged, and "the Soviet-imported Semashko system of healthcare could not keep itself of the chopping block" (Vlădescu et al. 2005, 467). Ideologically, the SHI system was positioned between the free market and government planning, thus it was acceptable to the supporters of both of these options (ibid., 468). Te technical aspect is similar to the rationale of efciency and administrative benefts in providing "more resources allotted to health, increased earnings for health professionals, greater fnancial independence, an increase in transparency, a better match between patients' needs and the services provided, improved quality of care, and an increase in service-provider accountability" (ibid., 488).

To summarise, frst, fndings on the WB's infuence in SHI reform in the region are controversial. Second, because of the diferences in the World Bank's position, scholarship on the pension reform in the CEE region is of little help to explain the SHI reform. For the pension reform, the Bank proposed replacing traditional social security with individual pension savings accounts. In several countries in the region, such as the Czech Republic, Poland, and Albania, the WB proposed no alternative model of fnancing health but suggested preserving the status quo or the previous communist model and objected to the choice made by the governments—the introduction of SHI. Te fndings in Kaminska et al. (2021) convincingly show that the WB could not infuence the agenda setting of SHI introduction in Albania (and other CEE countries). Tis chapter focuses on the subsequent step of policymaking and investigates the policy formulation process, which developed after the decision to introduce SHI was settled. Te fndings demonstrate that the Bank was able to some extent to infuence that phase of the policymaking process (see also Druga 2022).

#### **3 Methods and Data**

Tis research employs a qualitative design and uses the case study method and process tracing to describe and explain the process. Te period under investigation considers the events that took place from 1990 until 1994. Te analysis started with a desk review of documents in order to construct a chronology of events. Tey included those from the Albanian government, parliament, legislation, and the World Bank, such as strategies, projects, and reports. Next, to gain a deeper understanding of the crucial steps in the policy process and the motivations of diferent policy actors involved, the research proceeded with interviews, which supplemented the data gathered from the written sources. Between June 2019 and November 2020, I conducted seven semi-structured interviews with health policymakers and national and international experts.5 Tree interviews were face to face, the other four took place remotely. It is important to note that all interviewees were directly involved in the policymaking process. In particular, the expert from the World Bank, Goldstein, and both the Minister of Health, Shehu, and the Deputy Minister of Health, Nuri, were part of the dialogue process. Te other Albanian experts were involved in the legislation drafting process and later also in the process of implementation. I supplemented the empirical data with material from other written resources, such as two Albanian newspapers, *Voice of the People*, the ofcial newspaper of the Labour Party of Albania (later the Socialist Party of Albania), and *Te Democratic Renaissance*, the ofcial newspaper of the Democratic Party of Albania. Both newspapers were highly infuential during the period and were used by the respective political parties as channels of ofcial information and communication.

<sup>5</sup>All except one of the interviews were conducted in Albanian. Terefore, quotes from these interviews have been translated into English.

## **4 The Political Context of Social Health Insurance Reform**

In view of the literature on health policy analysis, which underlines the inseparability of policy and political context (Collins et al. 1999), this section focuses on the context in which the Albanian healthcare reform was conducted.

Until 1990, the totalitarian regime that prevailed for forty-fve years in Albania was rigid and based on Stalinist economic ideology (Schnytzer 1982). All domestic and foreign capital was nationalised or confscated. Market mechanisms disappeared, and economic decisions shaped by a highly centralised one-party state were implemented through fve-year plans. After some initial growth in the post-war years, the system ended in "substantially decelerated economic growth" (Blejer et al. 1992, 3) and stagnation. Limited reforms with the paradoxical name New Economic Mechanism, borrowed from the very early Hungarian reform experience, started at the end of the 1980s. Tey were partial and did not represent a coherent or radical programme (Pashko 1993, 908). As a result, in 1990, industrial production virtually ceased because of the shortage of inputs, and mass unemployment, rampant infation, and commodity shortages were drawing Albania into chaos. Te internal crisis was coupled with the external pressure of the demise of communism in CEE, while Albania remained the last country to still resist the change.

In December 1990, the Student Movement forced the communist president Alia to decree political pluralism (Rama 2019) and eventually announce the frst multi-party elections to be held on 31 March 1991. Te newly established Democratic Party of Albania (DPA) started preparing for the process. Te electoral programmes of both parties, the excommunist Labour Party of Albania (LPA) (Programi Elektoral i Partisë së Punës së Shqipërisë 1991) and the DPA (Platforma Elektorale e Partisë Demokratike të Shqipërisë 1991), were ideologically not diferent. Tey both concentrated on economic reforms, advocating a market economy, though the ex-Communists were for a gradual approach to economic and social transition (Buf 2015), while the Democrats supported rapid and radical measures known as "shock therapy" (Pashko 1993, 917).

#### **9 Against All Odds: Introducing Social Health Insurance in Albania…**

Te LPA survived the elections. On 11 May 1991, the Nano government presented its reform programme to parliament, addressing issues of privatisation, the establishment of the rule of law, and the market economy.6 Tat was the frst phase of the Albanian economic reform (until the frst non-communist government of April 1992), accompanied by confusion, half measures, and, above all, high political instability. In June 1991, a Government of National Stability, formed as a coalition of Communists and non-Communists from fve major parties (Socialist, Democratic, Republican, Social-Democratic, and Agrarian), endeavoured to steer the country through the difcult period until the new elections planned for spring 1992. Democrats' representatives in the Government of National Stability undertook several economic reforms. After the abrogation of the entire 1976 (communist) Constitution, at the end of October 1991 parliament approved a series of laws about price liberalisation, investment deregulation, macroeconomic intervention, such as a tight monetary policy and budgetary austerity, land distribution, and small-scale privatisation (Pashko 1993, 911).7 Another political crisis slowed down the reform process when in December 1991 the DPA walked out of the ruling coalition. Te prime minister resigned and the communist President Alia set the date for the next general election to be held in March 1992. A caretaker government replaced the multi-party coalition government, intending to prepare the new upcoming general election.

Te economic situation deteriorated further during the winter of 1991–1992. Te unemployment rate increased from twenty-nine per cent in 1990 to eighty-eight per cent in 1992, industrial production fell by over sixty per cent compared to 1991 (Tarifa 1995, 155), and ffty per cent of the urban labour force was out of work (Pashko 1996, 70). Since liberalisation, prices had risen by up to 500 per cent, while wages remained fxed. Infation reached an annual rate of almost 300 per cent in the frst quarter of 1992 (Pashko 1993, 913).

Te DPA won the majority in the general election of 22 March 1992 and formed a stable government in April 1992. Hesitant reforms,

<sup>6</sup>Te programme of the Nano government (11 May 1991–4 June 1991). https://shtetiweb. org/2015/09/13/programi-qeverise-se-fatos-nano-11-maj-4-qershor-1991/. Accessed January 28, 2021.

<sup>7</sup>For a detailed description of this programme, see Pashko (1993).

introduced in 1991, were reinforced politically in 1992 as the Democrats transferred the government's economic policies to the IMF (Tarifa 1995). Terefore, a macroeconomic stabilisation and restructuring programme under an IMF standby agreement was approved in August 1992, aiming at a short-term stabilisation and launching of structural reform in the country (World Bank 1993b, 2).

## **5 Introducing Social Health Insurance in Albania**

#### **5.1 The Early Albanian Government Efforts for Reform**

During the communist period, Albania developed a healthcare system like those of other countries in the CEE region. Management decisions were highly centralised and resource allocation was centrally planned and based on historical budgets rather than on an output-based formula (Albanian Council of Ministers 2001). As a result, the system ofered few incentives for quality of care, efciency, and cost control. Some aspects of the healthcare system though were even more problematic than in other CEE countries: the quality of care was the lowest in the CEE region (Davis 2010, 46), health personnel were poorly paid, and because of the lack of investment in the sector, the average age of medical equipment was twenty-fve years.8

Early eforts for reforming the fnancing model in the country started in mid-March 1991 when a group of experts from the Ministry of Health gathered to discuss how to respond to the request of the Albanian Council of Ministers to adjust the old healthcare legislation to the new dynamics in the country. After reviewing several laws and decrees, the experts proposed a list of recommendations.9 First—article number one in their report—they suggested an amendment of the law that had established

<sup>8</sup>Compared to an average age of 7.5 years for equipment in Western Europe. Te data are from a survey of medical equipment made by WHO in July/August 1991 (see World Bank 1992a, 23).

<sup>9</sup>Archives of the Ministry of Health, "New ideas on legislation improvements", 12 March 1991.

the concept of universal free healthcare in the country.10 Te experts argued that introducing fees and co-payments and the privatisation of several health services would require a complete revision of the law. According to them, the privatisation of health services would call for "fundamental changes" in the role of the state in health fnancing and, as a result, would demand the introduction of a "state social insurance law". In addition—article number four in their report—they proposed the revision of the existent budgeting law.11 Tat would then facilitate the introduction of new fnancing models in healthcare institutions, through the new social health insurance or a combination of funds from the general budget and the social health insurance. Terefore, the Ministry of Health promised to immediately initiate drafting the new law.

Te political events during the rest of 1991 (see section above) hindered the early phase of the health reform and the overall economic reform, leaving many issues unresolved. Terefore, the presentation of the draft bill on (social) health insurance to the Council of Ministers (on 15 December 1991) and parliament a few days later (on 20 December 1991) was cancelled.12 Te reform agenda, however, remained unchanged and work on the draft law was postponed until the frst trimester of 1992.13

After the elections of 22 March 1992, the DPA established a stable government and started rapid economic reforms aiming at privatisation, liberalisation, and decentralisation.14 Te government's programme in health was based on the DPA party's electoral programmes of March 1991 and March 1992, both promising the introduction of social (health) insurance in the healthcare reform package (Programi Elektoral i Partisë së Punës së Shqipërisë 1991; Platforma Elektorale e Partisë Demokratike të Shqipërisë 1991).

<sup>10</sup>Law no. 3766, dated 17/12/1963 "On Health Care in the Popular Republic of Albania".

<sup>11</sup>Law no. 6803, dated 29/06/1983 "On Budget Drafting and Implementing".

<sup>12</sup>Archives of the Ministry of Health, "Planning for the new legislation"—22 November 1991.

<sup>13</sup>Archives of the Ministry of Health, "Te Programme of the Ministry of Health of the frst Trimester of 1992"—6 January 1992.

<sup>14</sup>Te speech of the Prime Minister, Mr. Aleksandër Meksi in Albanian Parliament on 13 April 1992. http://shtetiweb.org/2015/09/13/programi-qeverise-aleksander-meksi-1992-1996/. Accessed January 28, 2021.

#### **5.2 The World Bank's 1992 Strategy for the Albanian Healthcare Sector**

Te International Monetary Fund carried out its frst mission to Albania in February 1991, shortly after the country sent its request for membership to the World Bank Group. Until October 1991, when the country received ofcial acceptance, three missions visited the country, aiming to get an understanding of the situation and prepare for future engagement. Recalling that period, the head of the IMF mission, Mario Blejer, states that

[n]ot much [was] known about the country's economy but the picture that emerge[d] [was] one of an economy in the midst of a very serious and profound crisis that [was] probably deeper than that experienced by other former socialist countries. (Bank of Albania 2005, 176)

During the mission conducted in October and November 1991, a team of experts from the World Bank concentrated on healthcare issues, collecting and documenting data and information for further analysis and preparation of a strategy proposal. Alongside the examination of the problems inherited from the communist period, the Bank's experts analysed the recent reform attempts in the sector and reviewed the new body of legislation together with the drafts of the National Health Law, the Privatisation of Medical Practice and Health Insurance (World Bank 1992a, 1).

In February 1992,15 the WB presented its insights in a report and a strategy proposal "For the survival and the long-term development of the Albanian healthcare system in the context of the larger economic transformation of the country", which was ofcially delivered on 23 March 1992. In this report, the WB raised the concern that even though the health status of the population was "impressive for a country at [its] stage of economic development", however "the threat [of deterioration in health outcomes] was real" (World Bank 1992a, i). Te Bank assessed

<sup>15</sup> In some of the WB's reports, it can be found under the title "Te health sector strategy paper of the WB of February 1992" (World Bank 1992b, 7).

that the Albanian healthcare system was sufering from both short-term fnancial difculties and fundamental structural weaknesses in service provision and pharmaceuticals which impeded its efective functioning in the long run. Terefore, the proposed strategy rested on three main pillars: "past achievements, improve quality, contain costs" (ibid., 29). As it was more broadly explained,

[T]he proposed strategy call[ed] for a reform of the system that would capitalize on its strengths, by maintaining and further promoting the primary care system, improving the input mix and the quality of inputs, and improving services, and correct its structural weaknesses, by streamlining the hospital system, improving the input mix and the quality of inputs, and improving management and planning systems. (Ibid., 1)

Alongside the inherited problems, the Bank's experts also analysed the recent reform attempts in the sector, namely, "the privatisation of health services" and "the introduction of the health insurance scheme". Te Albanian government was hoping to reduce the inefciencies in the healthcare system in the same way as in the other sectors of the economy, through privatisation and liberalisation measures. Tis raised concerns at the Bank, which was preoccupied with the social objective of ensuring equal access to services and argued that "equal access would be threatened by a rapid shift to privatisation, particularly given the low-income levels of the population and the expected increase in income disparity" (ibid., 1). Te Bank provided a thorough argument on the issue while stating that

[T]here are strong economic (efciency) and equity arguments for keeping health services in the public domain even when other sectors in the economy are being liberalised, and the state must fnd an appropriate role for itself—in the fnancing, providing, regulating health services—to ensure that efciency and equity are both maintained. Although the extent and nature of public sector involvement vary considerably among diferent countries, there is a growing consensus that public fnancing is important if only to ensure equitable access to services. (Ibid., iv)

As a result, the Bank recommended "gradual private sector involvement in the provision of health services", but "continued public sector fnancing, until at least more stable economic conditions permit the introduction of [a] social health insurance scheme" (ibid., 2). Consequently, the WB stated that the drafts of the National Health Law, the Privatisation of Medical Practice and Health Insurance were not ready for adoption because of serious shortcomings in both substance and formulation. Te Bank clearly articulated that "while new legislation is essential to the reform, it would probably be best to delay fnalisation of these laws until after the policies and strategies they refect are better developed" (ibid., vi).

#### **5.3 The Road to Comprehensive Healthcare Reforms**

Healthcare reforms were initiated in May 1992, soon after the appointment of the new Minister of Health, Shehu.16 One month later, in June 1992, the government adopted the proposed WB health sector strategy (World Bank 1992b, 7), though still insisting on the idea of introducing SHI. However, Albania lacked previous institutional experience with a contribution-based health fnancing model. In addition, the lack of experts on market economies, democratised institutions, and health policy was a further problem in Albania, as in other countries in the CEE region. Terefore, an expert, the director of budgeting at the Ministry of Health, was invited to attend a German-funded summer school on healthcare fnancing in CEE at ILO Turin Centre in Italy (Hobdari, interview 2020; Mano, interview 2020)17 (ILO 1995, 108). Later, the Ministry of Health organised and sent a group of experts on study tours to Germany and France (Kadiu, interview 2020; Jani, interview 2020; Mano, interview 2020).18 Te visit to Cologne in October 1992 was the

<sup>16</sup>Archives of the Ministry of Health, Protocol No. 371 dated 6 February 1993 "Report on the activity of the Ministry of Health during 1992".

<sup>17</sup>Flora Hobdari, expert at the Ministry of Health of Albania during the period 1992–2003 and a member of the group drafting SHI law. Interviewed via video connection on 2 September 2020; Laureta Mano, expert at the Ministry of Health of Albania during the period 1992–1994 and member of the group drafting SHI law. Later she held the position of Head of Legal Department at the Health Insurance Institute (1995–2019). Interviewed via video connection on 26 October 2020.

<sup>18</sup> Saimir Kadiu, expert at the Ministry of Health of Albania from 1992 until now, responsible for the World Bank projects in health and expert in the group drafting SHI law. Face-to-face interview

only one to Germany. Albanian experts and doctors were not familiar with the German language, which seems to have been a communication barrier (Kadiu, interview 2020; Jani, interview 2020). So, the study tours continued in France, as French was "the language of elite" in Albania (Kadiu, interview 2020; Jani, interview 2020). Old ties with France and the French healthcare system19 and the personal relationship between Minister of Health Shehu and President Berisha, two well-known doctors and professors at the Medical University in Tirana who both became politicians and joined the DPA, contributed to the future of such developments. As a result, the study tours and training with the French experts took place in Tirana and Paris in 1993 and 1994 (Kadiu, interview 2020; Mano, interview 2020; Jani, interview 2020; Hobdari, interview 2020). Even though the French assistance was useful, the lack of previous institutional experience and the diferences in economic and social development between Albania and the Western European countries made it difcult to fnd the right starting path (Kadiu, interview 2020; Jani, interview 2020).

#### **5.4 The Reform Game: "Keep Trying" (Until You Succeed)**

Despite the resistance from the WB, the government kept insisting on the idea of introducing SHI (Shehu, interview 2019; Nuri, interview 2019; Kadiu, interview 2020). Te WB task manager for the healthcare project, Ellen Goldstein, explained that "there was a huge push by the government [of Albania]. [It] wanted to start right away with the insurance that would cover health services and drugs" (Goldstein, interview 2020).20 Minister of Health Shehu confrmed the same conficting position with the Bank adding that the WB "considered us [the Albanian

conducted on 27 October 2020; Alqi Jani, expert at the Ministry of Health of Albania since 1991 and expert in the group drafting SHI law. Interviewed via video connection on 17 November 2020. 19For more on the historical relations between Albania and France, see the case of Albania in Kaminska et al. (2021).

<sup>20</sup>Ellen Goldstein, World Bank expert and Albania Health Project Task Manager. Interviewed via Skype on 19 June 2020.

government] incapable of governing such a model" (Shehu, interview 2019). In fact, the Bank was presenting the same arguments as in the 1992 report when suggesting that introducing SHI should be postponed to a later date because positive projections for the economy would increase individuals' incomes and improve prospects for the new policy (World Bank 1992a). Because of these recommendations, the government discourse transformed into "[we are] exploring the opportunity for introducing SHI until a new level of economic growth, and sectoral development will allow its viability".21 Te 1992 WB report also provided a list of instructions or "some preliminary steps leading to the development of SHI".22 Tese steps included establishing accounting and fnancial management systems at all health facilities to enable estimations of unit costs, the development of payment/reimbursement schemes, and their gradual application in the system to replace the ex-ante budgeting method. Te World Bank believed that this period would take several years, and for this reason, introducing SHI a decade later seemed to be the right-time projection (World Bank 1992a). Ergo, SHI became an issue "of further analyses".23

Tere was no further progress until 1993. Since joining the World Bank group in 1991, the government agreed on the future strategy and lending programme the Bank would ofer to the country. Based on the Critical Imports Project signed in June 1992, emergency issues like the provision of essential goods became the Bank's number one priority. In 1993, the Bank planned to implement the Employment and Social Protection project, followed by the Health Sector project in 1994 (World Bank 1993c, 18). However, in July 1992, the government abolished the guaranteed payments (sheltered wages because of the lack of raw materials and the halting of production), thus raising the unemployment level and increasing the pressure on future social assistance payments, though

<sup>21</sup>Archives of the Ministry of Health, no date, year 1992, signed by the Deputy Minister of Health Mr. Besim Nuri "Brief Information on Health Sector Reforms and Emergency Needs in the Sector".

<sup>22</sup>For more details, see Druga (2022).

<sup>23</sup>Archives of the Ministry of Health, no date, year 1992, signed by the Deputy Minister of Health Mr. Besim Nuri "Brief Information on Health Sector Reforms and Emergency Needs in the Sector".

the latter had to be ready when the unemployment benefts would expire (in 1993, one year later). Te existing social insurance was itself another problem; the budget subsidised it. In 1991, budget subsidies accounted for 28 per cent of social insurance expenditure (World Bank 1993c, 5). In 1992, the situation became even worse as contributions were declining sharply due to the rise in unemployment and the dissolution of cooperatives. Contributions from the latter accounted for forty-two per cent of the insured in 1991, a considerable component in the social insurance system. Furthermore, the unemployment compensation programme was designed as a separate contributory insurance scheme and again depended on general revenues to meet the defcit, even though employers were paying a further six per cent tax on payroll. Terefore, the Social Safety Net project planned in two components, (1) social insurance and (2) social assistance, became a pressing problem.

Pursuing the intention to introduce reforms to reduce the state's involvement in social insurance and increase individuals' and employers' responsibility for the fnancing of the benefts, the government appointed a standing committee to coordinate the process of social insurance reform. Two international experts engaged by the World Bank, Jon Eivind Kolberg, a Finnish expert, and Igor Tomes, a Czech expert, assisted in drafting the New Social Insurance Act which was adopted by parliament on 11 May 1993 (World Bank 1993c, 14–15; 67).

After settling the Bank's number one priority of reforming the social sector, the Bank and the government entered into an extensive dialogue to fnd a way forward with the SHI reform. As the WB's task manager explained in an interview, the WB "had a problem in Albania" because "payroll taxes [pension contributions] were already extremely high" and "more contributions would raise the cost of labour". According to the Bank's expert, the competitiveness of the Albanian economy, a lowincome country, was dependent on the cost of labour—since low-cost labour could turn it into "the perfect place for manufacturing". Te concern about labour costs was not, however, the only problem the Bank perceived. Another structural issue would potentially harm the viability of the future contributory system. Albania was not a highly industrialised country and "large parts of the population were not wage earners". Terefore, choosing a fnancing model based on payroll contributions while "not all the people are on the payroll" was clearly a problem (Goldstein, interview 2020).

Importantly, the dialogue between the Bank and the government was about building an understanding of what it meant to transition from one system to the next. Terefore, the Bank aimed at raising awareness of the cost that the transition period and other measures taken to stabilise the fnancial system would have on the Albanian healthcare system.

Tose who were advocating the quick shift to health insurance probably had not fully appreciated how much the collapse of the economy would have infuenced the ability to even maintain the level of healthcare that (Albania) had before. Suddenly you have half the money you had before. Instead of the quality of the services being better or more responsive, the quality of the services would get worse, and that was something that no government could aford politically. (Goldstein, interview 2020)

Finally, after discussions both parties agreed on a contribution rate of two per cent for social health insurance. Tat was the green light for the government to start drafting the SHI law. Te SHI initially started rather as a scheme (Nuri 2002), but with its introduction, the government fulflled its wish to introduce the new fnancing policy in the country.

So, we argued, and I think probably persuasively that the worst possible scenario would be to move in some big way toward the health insurance scheme (…) We [WB] carefully discussed and shifted opinion toward starting slowly with very smaller change (…) So, we advised to take a cautious approach and that is what was done. (Goldstein, interview 2020)

In April 1993, the government presented its medium-term strategy to the donors' community: "A new policy for the healthcare sector in Albania", articulating "the introduction of market elements into healthcare fnancing" and "a careful introduction of a scheme of health insurance" among the proposed reforms (Nuri 2002, 70). By the end of 1993, an inter-ministerial group of experts (from the Ministry of Health, the Ministry of Finance, the Ministry of Labour, Emigration, Social Protection and Ex-Politically Persecuted People, the Ministry of Agriculture, and the Social Insurance Institute) were appointed to work on the draft formulation. Te Czech expert, Igor Tomes, joined the group from the start and played a prominent role in policy formulation (Kadiu, interview 2020). Te Albanian Council of Ministers consulted on the draft in summer 1994 and the bill was passed by parliament on 13 October 1994 (Albanian Parliament 1994).

## **6 Opening the Black Box: Transnational Cooperation**

Te mechanism of transnational cooperation, which unfolds in a noncoercive way, explains the role of a transnational actor, namely, the World Bank, in introducing SHI in post-communist Albania. It elucidates how the World Bank, even though it was not able to infuence the policy choice, namely, the introduction of SHI, could however infuence the process during subsequent policy formulation. Te complex causal mechanism of transnational cooperation consists of three elementary causal mechanisms (see Chap. 1): First, the *calculatory orientation* of the government in ideating the introduction of SHI; its action is driven by the logic of efciency and quality improvement; second, the *normatively embedded calculatory orientation* of the World Bank manifested in its report from 1992; and third, the *refective orientation* of the Bank in the dialogue process with the Albanian government and, as a result, in the change of its position towards the policy choice.

1. Te government's calculatory orientation

Te Albanian post-communist healthcare system heavily depended on state budget resources and inadequate fnancial resources "were a fundamental problem" (Shehu, interview 2019; Nuri, interview 2019). Te performance of the (dwindling) national economy further constrained the government's fnancial resources for health, which were even lower than those in the other countries of the CEE region (Goldstein, interview 2020). All actors (doctors, politicians, patients) were aiming at extending the scope and scale of health services. As a result, Albania had to cope with the discrepancy between needs and resources (Goldstein, interview 2020; Shehu, interview 2019).

In 1991, the proposal for introducing SHI was a result of attempts to reform the healthcare system, based on the New Economic Mechanism model.24 Te government perceived this model to be the remedy for the economy after decades of inefciency and economic mismanagement in the country. Terefore, introducing SHI was driven by the economic logic of efciency because of tight budgets and fnancial limitations and quality improvement in the system. In 1992, the government's healthcare strategy refected the preferences of the new political elite that was advocating a restricted role for the state in the economy (Shehu, interview 2019). It must be stressed that despite the diferences in the reform's speed, all governments, either before the election in 1992 or after, were in favour of a limited role of the state in health fnancing. To them, SHI meant individual responsibility through payroll contributions, and the introduction of SHI would lead to a smaller role of the state in the healthcare system.

Terefore, the behaviour of the Albanian government in the process was driven by an opportunistic logic—a logic based on time-specifc conditions—and a self-interest seeking one, driven by efciency and quality improvement.

2. Te normatively embedded calculatory orientation of the World Bank

Te 1992 World Bank's report for Albania's future reform of the health sector was a product of the norm compliance orientation of the WB combined with rational calculation. Public documents produced by the Bank and other data support this argument. As for the norm compliance orientation, the 1992 WB report refects the content of an earlier WB report, published in 1987 and titled "Financing Health Services in Developing Countries: An Agenda for Reform" (World Bank 1987), where the Bank proposed an alternative approach to fnancing healthcare in developing countries through four measures: frst, introducing user fees in health; second, providing insurance or other risk coverage and encouraging well-designed health insurance programmes to help mobilise resources for health; third, using non-government resources

<sup>24</sup>Archives of the Ministry of Health, "Planning for the new legislation"—22 November 1991.

efectively; and fourth, decentralisation (World Bank 1987, 6). Tese four measures are refected in the Bank's normative position in Albania in 1992. As for the rational calculation behind the 1992 WB report, it is visible in the Bank's arguments of efciency25 and equity and the recommendation on preserving the tax-based model of health fnancing. Te equity argument is related to the Bank's concern about ensuring equal access to healthcare. According to the Bank, this issue was particularly relevant during a transition period marked by infation and unemployment and by growing income disparities, like in Albania (indeed, one year later, the "1993 World Development Report: Investing in Health" would advocate for a synergy between equity-enhancing and efciencyoriented change with the government's role more on the regulatory side (World Bank 1993a, 7)).

To sum up, the logic underlying the 1992 WB report can be explicated by the norms derived from the WB's report "Financing Health Services in Developing Countries: An Agenda for Reform" (World Bank 1987) coupled with the rational arguments of efciency in healthcare provision and equity in healthcare access, relevant for a transition country.

#### 3. Te refective orientation of the World Bank

Te consensus on introducing SHI in Albania derived from a refection process. Tough both the WB and the Albanian government changed their positions from their initial stance, the WB did so considerably from opposing the government's choice to a negotiated decision. Te World Bank had to re-evaluate its recommendation on the future health fnancing model in Albania and follow the government's wishes.

Pro and contra arguments exploring the costs and benefts of the policy choice were part of the dialogue process. Te most important

<sup>25</sup>Te efcient use of health services is related to the product (health services) which lacks two key features of an efcient free market good. Te frst element is the lack of full information about the product on the part of the consumer (patient), who is not able to sufciently keep up with medical development to make informed decisions on what or how much to consume (i.e., pays for). Te second element is the complete independence in decision-making between the consumer and the supplier (health provider), since the latter determines in large part the nature and quantity of the product that the consumer demands (i.e., pays for). Tese two features tend to result in inefcient use.

argument was related to higher labour costs, because the (additional) health contributions would harm the country's strategy developed by the World Bank for future economic growth. Next, the WB considered the health sector to be diferent to the other economic sectors, and as a result, it was concerned about the fulflment of two principles: efciency and equity in healthcare provision. For these reasons, the Bank advised that the state must fnd an appropriate role for itself—in fnancing, providing, regulating, and/or setting policy for health services—so that it could ensure that efciency and equity are both maintained. Finally, the Bank provided helpful technical recommendations for setting up an SHI scheme. Tus, it was able to convince the government to follow a simpler SHI model than the one previously intended and so infuenced the policy formulation process (see Druga 2022). Importantly, the WB remained in the process and was able to exert pressure on the government regarding the formulation of the new policy.

## **7 Discussions and Conclusions**

Tis chapter addressed the challenge of explaining the role of the World Bank in the social health insurance reform in Albania. Focussing on the strategic interaction between the Albanian government and the World Bank and relying on an actor-centred perspective (Scharpf 1997) to frame it, the analysis of the chain of events revealed the crucial role of the mechanism of transnational cooperation to understand the process and underlined the non-coercive form it took.

Te fndings show that the World Bank had no preference for introducing SHI in Albania. Next, they illustrate that the WB did not force the government to accept its policy prescription. Notwithstanding, the WB used a strategy, which I name the "keep trying" strategy, to challenge the government. Tis strategy expresses the World Bank's attempts to stay in the reform game despite its course not being in line with its preferences. Interestingly, the "keep trying" strategy adds to the list of the strategies that transnational actors, the WB and the IMF in particular, have employed to shape health and social reforms in the countries of the Global South (Orenstein 2009; Weyland 2006). Finally, the case demonstrates that the World Bank took a non-coercive approach throughout the process, with successful cooperation during the formulation stage after it had failed during agenda setting.

Te interaction between the WB and the Albanian government is characterised by the dual dynamic of the persuasion power of the former and its lack of ability to impose a specifc policy model. Te WB's policy prescription was but maintaining the status quo, asking the government to preserve the same model of health fnancing as during the communist period. Further, the WB refected on its policy prescription, and even though SHI was not its favourite policy choice, it assisted the government in preparing the draft bill. In the end, the WB was able to convince the government to follow a simpler SHI model than the one previously intended, with lower contribution rates and restricted coverage of healthcare services. As a result, even though it was not able to impose its policy choice on the Albanian government, the World Bank could induce it to move towards the principles of the strategy it advocated.26

Te study reveals the government's rationale for choosing SHI as the new model for health fnancing. Consistent with the fndings from previous studies on the post-communist CEE region (Medved et al. 2005; Vlădescu et al. 2005), the government was driven by the rationale of efciency and quality improvement. From the other side, alongside the actors' preferences, the study explicitly also evinces their position. Tus, as the government did not change its position towards its policy choice, introducing SHI, I would consider that the government possessed the attributes of a "veto" and "proposal" actor. Te domestic actors indeed held the veto role (Tsebelis 2002), but the role of a proposal actor has previously been attributed to the WB in pension reform in the CEE region (see Orenstein 2008). In contrast to this view, I suggest that here the WB played the role of a "refective" actor because of the change of its position on introducing SHI.

Finally, the chapter sheds light on the role of the WB in development assistance in the post-communist CEE region, emphasising its noncoercive form. In line with the previous scholarship on this more discursive understanding of the WB's infuence in developing countries, the

<sup>26</sup> See "Bank Group Country Assistance Strategy and Operations" (World Bank 1992b, 6).

case of Albania confrms that development assistance, which is concerned with selling ideas and practices from the international to the domestic stage through the production and circulation of discourses, is indeed a form of exercising power (Escobar 1988, 430). Further, the case sheds light on the persuasive side of development assistance, which stands in contrast to the imposed conditions attached to loans. Rather than dictating policy solutions, the recipient of development assistance can internalise new ways of thinking because development assistance convinces via implied technical legitimacy and is premised upon dialogue, collaboration, and consent (Smith 2008, 238–39). Ultimately, development assistance is a social and political construction. As such, it does not in itself possess legitimacy. Forces that grant or deny legitimacy are those embedded in the political and social interactions between technical assistance provider and recipient (Bazbauers 2018, 240). Tereby, conferring that legitimacy is of the same great importance as the norms the WB brings to the domestic policy arena is highly relevant and has been spelled out in the case of Albania.

I have focused my analysis on the Albanian government. Further research should focus on the elites, particularly the medical elite, that managed to place introducing SHI on the agenda and won the resources to implement it. Tat way, the analysis might unravel another rationale for the policy choice. Furthermore, placing the medical elite at the centre of investigation provides valuable insights on its role in enacting healthcare reforms in the post-communist CEE region compared to the doctors' veto role in blocking reforms, as the existent scholarship from the Global North countries emphasises (Roberts 2009).

### **References**


Dixon, Josep Figueras, and Joe Kutzin, 80–108. Buckingham: Open University Press.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

# **10**

## **Migrants to the Rescue? Care Workforce Migrantisation on the Example of Elder Care in Germany**

**Anna Safuta, Kristin Noack, Karin Gottschall, and Heinz Rothgang**

## **1 Introduction**

Te number of older citizens has been growing dramatically across the world, not only in absolute terms, but also in relation to the working-age population that can potentially provide care (OECD 2020, 16). At the same time, women—the traditional providers of care within the family have been entering the paid labour force in larger numbers than before. Tus, they have much less time and drive to care. Due to the

University of Bremen, Bremen, Germany

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

A. Safuta (\*) • K. Noack • K. Gottschall • H. Rothgang

e-mail: anna.safuta@uni-bremen.de; knoack@uni-bremen.de; karin.gottschall@uni-bremen.de; rothgang@uni-bremen.de

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_10

combination of these demographic and societal factors, the need for additional long-term care (hereafter LTC) is intensifying across Europe.

At the same time, recruiting enough workers in LTC remains a challenge in most OECD countries (OECD 2020, 3). Across the world, the demand for LTC workers is increasingly fulflled by hiring workers "with a migratory background", including migrants1 (Da Roit and Weicht 2013; Lightman 2020; Williams 2012). Care is thus increasingly becoming a "migrantised" occupation, not only in the West, but all across the world. Migrantisation2 is here defned as the process of incorporating migrant workers into the formal and informal care workforces of a country. Formal care is provided within residential settings or in the receiver's home by native and migrant workers with regular contracts. Informal care encompasses care provided not only by relatives and friends, but also by migrant workers employed with or without a regular contract and often residing in the receiver's home (this type of care is thus often referred to as "live-in"). A sector is undergoing migrantisation when the share of migrant workers therein has been increasing over time.

Despite similar trends in terms of overall LTC migrantisation, countries difer in the ways in which migrant workers are involved in national care regimes (Da Roit and Weicht 2013). Previous research identifed several modes (real types) of migrantisation, including the "migrant in the family" and "migrant in formal care" outcomes, which we defne in the next section. Te emergence of each mode is explained by referring to a country-specifc intersection of regimes3 of care, migration and employment (Williams 2012). However, analyses of the concrete mechanisms

<sup>1</sup> In this chapter, "migrant worker" refers to individuals who (temporarily or permanently) migrated to another country than their country of origin *as adults*. Te more encompassing category of "worker with a migratory background" refers to the German statistical category of people "with a migratory background" (*Migrationshintergrund*). Te Federal Statistical Ofce of Germany defnes persons with a migratory background "as (1) persons who have immigrated to the Federal Republic of Germany (FRG) since 1949; (2) foreign citizens born in the FRG; and (3) all German citizens born in the FRG with at least one parent who either immigrated to the FRG after 1949 or was born in Germany as a foreign citizen" (Elrick and Farah Schwartzman 2015, 1543).

<sup>2</sup> "Ethnicisation" is a similar, although broader, concept. It describes the process through which the formal and informal care workforces in a country are made up of people with ethnic backgrounds which are not dominant in the overall labour force of the country (see Ranci et al. 2019).

<sup>3</sup>A "regime" is here understood as a combination of policies, institutions and practices pertaining to a policy domain (Williams 2012, 374). See also Frisina Doetter et al. (2021).

through which such intersections produce diferential modes of migrantisation are still missing. Tis chapter aims to address this gap. It does so by uncovering causal mechanisms driving migrantisation processes in the elder care domain.

As the second country in the world to do so after the Netherlands, Germany introduced social insurance against the "new" social risk of LTC dependency (hereafter long-term care insurance, LTCI) in 1995–1996. Current political and societal dynamics are shaped by earlier policy choices (Streeck and Telen 2005), and this also applies to migrantisation. In this chapter, we identify the elements of the German LTC regime, regulated chiefy by the LTCI, which shaped migrantisation in Germany.

We demonstrate that the co-existence of two parallel "circuits" of family and formal care generates two relatively distinct migrantisation processes, which result in two distinct outcomes ("migrant in the family" on the one hand and "migrant in formal care" on the other). Second, we go beyond existing studies of migrantisation by identifying fve complex causal mechanisms that drive the process.


In order to identify the causal mechanisms behind the emergence of the "migrant in the family" and "migrant in formal care" parallel outcomes in Germany, we use process tracing (Beach and Pedersen 2019). We follow an actor-centred approach of causal mechanisms (Chap. 1). Tis means that we collected information on the causally productive activities of a variety of actors, both individual (migrants and households) and collective (state institutions, federal policymakers, stakeholder organisations, for-proft companies, etc.). As sources, we used seventeen original expert interviews with specialists, stakeholders and policymakers; academic and grey literature; and own calculations based on ofcial statistics.

We start with a brief review of the literature on migrantisation in LTC, before describing the dual outcome that will be explained, namely, the migrantisation processes in formal and family care in Germany. We then reconstruct two causal chains and identify the mechanisms driving their progression from a trigger to the outcome, and subsequent policy feedback(s). Te article ends with conclusions discussing the wider implications of our fndings.

## **2 Migrantisation in the Literature**

Most studies notice substantial diferences in the way migrants are involved in care work in host countries. Previous research identifed two main modes (real types) of "migrant workers" incorporation into national LTC systems ("migrant in the family" and "migrant in formal care"). Tese modes are shaped not only by LTC-specifc policies, but also by the host country's care, migration, employment and gender regimes (van Hooren 2012, 135; Williams 2012, 363).

Te "migrant in the family" model has been identifed in familialistic care regimes, including Southern European countries and Austria. It refers to the private employment of migrant care workers by individual households, with or without a legal employment contract (Bettio et al. 2006). Te "migrant in formal care" model refers to the presence of migrants with regular employment status within both for-proft companies and non-proft organisations providing residential or home-based care. Tere are two distinct varieties of the "migrant in formal care" model: the frst variety developed in countries with a high incidence of for-proft care providers (such as the UK), while the second evolved in countries with high LTC expenditures, low levels of undocumented migration and the segregation of migrants in low-skilled jobs (such as France, the Netherlands, Norway or Sweden) (Da Roit and Weicht 2013, 477). Countries with an overall bigger (for-proft and/or non-proft) formal care sector register a higher demand for migrant workers than those with limited formal provision (Da Roit and Weicht 2013, 471). Additionally, larger for-proft provision is "expected to increase opportunities for migrant care workers to enter formal care and to crowd out (informal) employment in the domestic sector" (Da Roit and Weicht 2013, 471).

Te closest that the care literature comes to outlining the concrete mechanism through which growing for-proft provision leads to migrantisation comes from labour market economics. In the absence of public policies guaranteeing and fnancially supporting wage levels and quality standards in the care sector (or in the case of insufcient wages and labour standards), care provision through the market involves decreases in wages and employment standards. Te sector thus becomes less attractive to a "native" workforce, which results in its migrantisation (Ranci et al. 2019, 4). However, labour market regulation in coordinated capitalist societies (such as Germany) limits the efects of care marketisation on migrantisation (Ranci et al. 2019, 4).

"Migrant in the family" care is said to emerge in countries with limited formal care services (Bettio et al. 2006, 278), a strong ideal of family care (Böcker et al. 2017, 228), cash-for-care subsidies that users can spend freely (Ungerson 2004), and a tendency to segregate migrant workers into low-paid jobs (Da Roit and Weicht 2013, 479).4 According to Da Roit and Weicht, it is the combination of these causes that stimulated the development of a "migrant in the family" model in Germany and in other countries, such as Italy (2013, 479). As a driver of "migrant in the family"

<sup>4</sup> It has already been demonstrated that migrant workers are generally segregated into low-status jobs characterised by harsh or unpleasant working conditions and with limited chances of job mobility (Piore 1979, 17–19). Because care jobs are generally characterised by disadvantageous employment and working conditions (low wages, long working hours, night shifts, etc.), migrants are thus often concentrated in this sector to the detriment of other, more prestigious and better paid occupations (Da Roit and Weicht 2013, 471). A recent study by Khalil et al. (2020) demonstrates that migrants in Germany work in low-paid and precarious positions, especially in so-called system-relevant occupations such as healthcare.

care specifc to Germany, the literature cites the fact that the LTCI covers only a part of actual elder care costs (Böcker et al. 2017, 228).

Within informal care, demand for care workers in Western European households is primarily aimed at white, Christian and female candidates, to the detriment of migrants with other ethnic, religious and gender backgrounds (Safuta 2018). Tese preferences combine with unfavourable socio-economic conditions in the countries of origin to encourage the employment of "peripherally white"5 migrant carers from Central and Eastern Europe (hereafter CEE).

Among the "push factors" specifc to care emigration from formerly communist CEE countries, the literature mentions the mass unemployment that followed the post-1989 political and economic transformations in that part of Europe. Unemployment afected women more intensely because of the restructuring of many female-dominated sectors, such as health (Robert 2006, 161–63). In short, the activation (or "occupational empowerment") of women in Western Europe combined with the deactivation of their counterparts from CEE to stimulate infows of female migrants from CEE to Western Europe, including Germany (Kniejska 2018, 479). In accordance with the neoclassical theory of migration, (perceived) income diferentials between CEE countries and their Western European neighbours also play a crucial role in encouraging migration (Cyrus and Vogel 2006, 81).

## **3 Migrantisation of Elder Care Work in Germany**

### **3.1 Elder Care Shortages in Light of the LTCI**

Te elder care system in Germany is strongly dual, due to the Bismarckian organisational principle of subsidiarity, derived from Catholic social doctrine. As with other welfare functions, subsidiarity afrms the family and

<sup>5</sup> "Peripheral whiteness" is a subject position of simultaneous privilege and subordination experienced by white migrants originating from non-Western (semi-)peripheral countries (here CEE). For more on "peripheral whiteness", see Safuta (2018).

non-profts (rather than the state itself) as the most appropriate providers of elder care (Gottschall and Dingeldey 2016). Tis principle explains why, up to the introduction of the LTCI, public support for elder care in Germany was residual and was available only if families could not provide care themselves or aford the costs of externally provided care (Götze and Rothgang 2014, 64). Te LTCI created a right to services in kind and to unregulated cash benefts paid directly to benefciaries. In practice, this means that citizens with a recognised need for care can choose between regulated benefts that can only be spent on formal care (homebased care services or residential care) and unregulated cash payments. A combination of in-kind and cash benefts is also possible (Götze and Rothgang 2014, 82).

Te crucial argument behind the introduction of cash benefts was the idea that such payments would be an efective way to acknowledge, support and activate family-based care, which was to remain the main modality of elder care provision in Germany. Together with "pension credit points", cash benefts were described as an incentive "particularly for women with low qualifcations, to take over care responsibilities" (Teobald and Hampel 2013, 15). For their part, civil society organisations (such as pensioners and disability groups) supported cash benefts as a way to increase benefciaries' autonomy in choosing their preferred mode of care provision (Teobald and Hampel 2013, 10). Te overarching argument behind direct cash payments was however that they are a less costly way of supporting family care than benefts in kind (Teobald and Hampel 2013, 10).

Te capacity/willingness of families to provide elder care has however been eroding since the 1960s, due to a combination of demographic and social factors, including population ageing and women's increased labour market participation (Götze and Rothgang 2014, 70; Sopp and Wagner 2013, 2016). Furthermore, formal home-based care in Germany is designed as selective relief for family carers, focusing mainly on medical and nursing tasks (e.g. administering medication or wound treatment) (Böcker et al. 2017, 237).

Beyond supporting family care, the LTCI also aimed to increase the efciency and supply of formal care (Teobald and Hampel 2013, 13). In line with this objective, LTCI legislation introduced regulated competition between non-proft, for-proft and (the rare) public providers. Before the introduction of LTCI, non-proft charity organisations had priority over for-proft providers: local governments had to contract charity organisations frst and were allowed to "hire" for-proft providers only if charities were unable to fulfl municipalities' demands (Götze and Rothgang 2014, 70). Regulated competition was thus perceived as a way to ofer benefciaries more choice between care providers (Teobald and Hampel 2013, 17). Another declared aim of provider competition was to create job opportunities for the female labour force. However, the LTCI scheme did not improve the attractiveness of the sector in terms of wages, working conditions and professional status (Ostner 1998, 128).

Increasing numbers of care-dependent people, unfavourable working conditions and comparatively low salaries make it difcult for elder care providers to recruit and retain workers. Besides the ageing of care personnel, residential care facilities are also afected by the ffty per cent quota of highly skilled personnel (*Fachkraftquote*). Te measure was introduced in 1993 to safeguard the quality of care, as part of the stafng regulation of residential care facilities (*Heimpersonalverordnung*).6 Strongly contested by (for-proft) care providers, but supported by trade unions, this controversial measure forces care facilities that do not fulfl the quota to close certain wards. Formal care providers are thus confronted with severe staf shortages and the care sector is marked by a high share of part-time work and high turnover rates (DGB and ver.di 2018). It currently takes an average of 205 days to fll a geriatric nurse vacancy (Bundesagentur für Arbeit 2020), and the need for skilled elder care workers is projected to further increase in the future (Flake et al. 2018, 34).

#### **3.2 The Migrantisation of Family Care**

Since the 1990s, care for dependent elders within high- and middleincome German households has been increasingly provided by migrants (Lutz and Palenga-Möllenbeck 2010; Kniejska 2016). Research refers to this type of care provision as the "migrant in the family" model, because

<sup>6</sup>Federal states can deviate from this regulation since 2006, although most kept to the original quota of ffty per cent of skilled care personnel in residential care facilities (Stoy 2016, 348).

employing households perceive migrants' services as a replacement for family-provided care (Kniejska 2016, 84–85). Tis type of care is mostly provided on a live-in basis: migrants live with the care receiver and are available to them almost 24/7.

Most recent estimates suggest that there are 500,000 live-in migrant care workers ("live-ins") in Germany (Benazha and Lutz 2019). Estimating the actual numbers of live-ins is difcult because of the mostly irregular character of this type of employment. Te biggest association representing brokering agencies in Germany, VHBP, estimates that ninety per cent of all migrant care workers in German households work irregularly (Petermann et al. 2017, 4).

In a comparative perspective, Germany ranks in-between countries such as Italy or Spain, where four to six people out of every hundred aged sixty-fve or over are cared for at home by a migrant worker, and France, the Netherlands, Sweden or the UK, where the phenomenon has a very low incidence or is nearly absent (Da Roit and Weicht 2013, 473–74). Böcker et al. (2017, 228) are much less moderate when they conclude that "the employment of migrant care workers in private households has become a mass phenomenon [in Germany]". Lutz and Palenga-Möllenbeck even conclude that the German care regime would collapse without live-in carers from abroad (Lutz and Palenga-Möllenbeck 2011, 349).

Most live-in care workers are women over ffty who come from CEE (Karakayali 2010, 291–93) without formal elder care qualifcations (Böcker et al. 2017, 230). Live-ins used to be recruited through informal migratory networks, which are now increasingly superseded by private brokering agencies (Leiber et al. 2019, 366). Estimates of the number of such agencies active in Germany and recruiting across CEE have substantially increased: from 50 in 2014 (Krawietz 2014) to 274 in 2018 (Leiber et al. 2019, 375). Tey target almost exclusively households in Western Germany, probably due to higher average incomes in this part of the country (Krawietz 2014).

Most live-ins in Germany come from Poland, although the share of workers from Bulgaria, Romania and Ukraine is increasing (Emunds 2016, 190). In the 1990s, Poles were at an advantage compared with other non-EU nationals, as they were exempted from the German visa obligation. However, possibilities for employment in Germany were mostly limited to irregular employment, which meant that female migrants from CEE were segregated into domestic and care work within private households. Te 2004 enlargement of the EU to eight CEE countries7 did not immediately grant citizens of those countries access to regular employment in Germany. Until May 2011, they could only work as self-employed service providers or as workers posted by a company based outside of Germany. Tis meant that they could not work within formal care settings without obtaining a work permit and hence concentrated in informal live-in care. Similarly, after their countries joined the EU in 2007, citizens of Bulgaria and Romania had to wait until January 2014 to access the German labour market without restrictions. Non-EU citizens are less likely to work on a live-in basis because of residence and work permits, although Ukrainians often work in Germany as posted workers, on the basis of a contract with a Polish employing agency.8

#### **3.3 The Migrantisation of Formal Care**

Formal care in Germany consists of home care, day care and residential care provision. Formal care has a comparatively well-qualifed workforce thanks to high-standard occupational training programmes. Formal care workers are mostly employees of registered non-proft and for-proft organisations providing care, although within home-based care they can also work on a self-employed basis (Auth 2017, 338).

In line with the literature on migrantisation, in this chapter we are primarily interested in migrants in the sense of individuals who migrated (temporarily or permanently) to another country as adults. However, it is difcult to obtain numbers on people who migrated to Germany as adults, as most ofcial statistics in the country refer either to people with foreign citizenship or to "persons with a migratory background" (see footnote 1). Statistics showing the percentage of elder care workers with foreign citizenship (including those born in Germany) or "with a

<sup>7</sup>Te Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia.

<sup>8</sup>Expert interview, Consumer protection association, 18 November 2019; Expert interview, Counselling centre for migrant workers, 27 November 2019.

migratory background" are, therefore, only an approximation of the extent to which the elder care workforce in Germany is migrantised.

On average, foreign-born workers make up over twenty per cent of the LTC workforce in the OECD (OECD 2020, 44). In a comparative perspective, migrant care workers are less predominant in the formal care workforce in Germany than in other European countries, but the numbers show a clear tendency towards migrantisation. According to ofcial statistics from the Federal Employment Agency, the number of formal elder care workers with foreign citizenship9 doubled from 6.8 per cent in 2013 to 13.6 per cent in 2019.10 Te share of workers with foreign citizenship in elder care is slightly higher than their average share in the labour market overall (12.5 per cent in 2019).11 Statistics on care workers with foreign citizenship show that the number of those employed as lowskilled care assistants was twice as high as of those employed as skilled care workers.12 Based on self-identifcation, the Socio-Economic Panel (SOEP) shows an even higher proportion of foreign-born workers: in 2018, 26.3 per cent of all employees in elder care stated that they were born abroad compared to 17.7 per cent in the overall labour market (Khalil et al. 2020, 5).

Most foreign care workers are women aged between twenty-fve and ffty-fve. In 2019, the top countries of origin of foreign care workers employed in Germany were Poland, Bosnia-Herzegovina, Turkey and Romania.13 Te formal care sector in Germany draws from a migrant workforce with permanent work and residence rights, including "ethnic Germans"14 from Kazakhstan, Russia and Ukraine, who have the possibility to obtain German citizenship (Teobald 2017, 222). Besides spontaneous labour migration, German authorities have also concluded specifc

<sup>9</sup>Data on people with foreign citizenship are the most comprehensive ofcial data regarding migrant workers in formal care.

<sup>10</sup>Own calculations based on unpublished data on socially insured employment that we obtained from the Federal Employment Agency in March 2020, including the occupational category (821) elder care professions.

<sup>11</sup>Own calculations based on unpublished data mentioned in footnote 10 and Bundesagentur für Arbeit (2019) for overall labour market shares.

<sup>12</sup> See footnote 10.

<sup>13</sup> See footnote 10.

<sup>14</sup>Representatives of the German diaspora.

recruitment agreements with several non-EU countries.15 With highskilled care migration on the rise in recent years, placement agencies have proliferated to broker between migrants and organisations providing care (Mosuela 2020).

## **4 Actor-Centred Approach to Migrantisation Processes**

Te previous section introduced the care shortages faced by households (as providers of family care) and by organisations providing formal care. Migrantisation is, to a large extent, the result of households and formal care providers meeting their care needs by hiring workers from abroad. Recruitment abroad is now facilitated by the emergence of a new type of actor—brokering agencies. Tese agencies proft from and further enable care migrantisation by matching demand in Germany with labour supply from CEE and other regions with substantial wage diferentials compared to Germany.

Tis section shows that the type of migrantisation ("migrant in the family" or "migrant in formal care") is, to a large extent, structured by three principles governing the country's elder care regime: cash benefts, provider competition and the skilled worker quota. Adopting an actorcentred approach to migrantisation allows us to show the way households and employers "make use of" migrant workers as a solution to the care shortages they encounter. While remedying these care shortages, households and care providers trigger causal chains involving other social and political actors, such as brokering agencies, decision-makers, trade unions and employers' associations. Tose causal chains are examined here along with the mechanisms that drive them (see Figs. 10.1 and 10.2).

<sup>15</sup>For example, several hundred qualifed nurses have been recruited within the framework of the "Triple Win" project inaugurated in 2013 and led by the International Placement Services (*Zentrale Auslands- und Fachvermittlung*, ZAV) of the Federal Employment Agency and the German Society for International Cooperation (*Deutsche Gesellschaft für Internationale Zusammenarbeit*, GIZ). Te countries involved include Bosnia-Herzegovina, the Philippines, Serbia and Tunisia. Similar agreements have been concluded with China, Mexico and Vietnam.

**Fig. 10.1** Causal chain behind the entrenchment of the "Migrant in the Family Model". (Source: Own representation)

**Fig. 10.2** Causal chain illustrating the rise of the "Migrant in Formal Care" model. (Source: Own representation)

#### **4.1 The "Migrant in the Family" Causal Chain with Its Mechanisms**

Part one. Individuals and households in Germany are confronted with care needs they are unable or unwilling to fulfl themselves. At the same time, female migrants from CEE were searching for employment in Germany, following the economic and political transformation of that part of Europe after 1989, due to rising unemployment, early retirement, bankruptcy, debt or low wages (Dietz 2007, 32).

Part two. Te infows of migrants from CEE (especially Poland) to Germany created the opportunity structure for German households to fulfl their care needs by privately hiring migrant live-in carers. Tey were encouraged to do so by the unregulated cash benefts ofered by the LTCI: the employment of live-in migrant carers increased rapidly after the introduction of the LTCI, with its cash benefts (Böcker et al. 2017, 229). Furthermore, families can cover only parts of necessary home care services from LTCI benefts—additional services must be paid entirely out of pocket. Private co-payments in residential care are increasing constantly, as nominally fxed beneft caps are not adjusted, while fees are ever increasing (Rothgang and Müller 2018, 86–87). In consequence, increases in the price of formal care might encourage relatives to provide care informally (themselves and/or by hiring a migrant care worker) and opt for cash payments instead of benefts in kind (Götze and Rothgang 2014, 84). Tis mechanism could be described as a turn to the market: when relatives cannot or do not want to carry out social functions (in this case elder care) traditionally fulflled by the family, and if they do not perceive state or non-proft alternatives as sufciently accessible/afordable/good-quality, relatives make use of market solutions. Te likelihood of a turn to the market is increased by the combination of unregulated cash benefts with the availability of a cheap migrant labour force (Ungerson 2004). Te spread of "migrant in the family" employment creates a transnational market for live-in care between Germany and CEE countries.

Part three. Te spread of "migrant in the family" care encourages a new type of actor to proft from the transnational care market. Brokering and employing agencies took over previously informal networks in facilitating movement across borders and matching labour supply with demand. Tey make use of the posted workers directive (Directive 96/71/EC concerning the posting of workers) to hire care workers under more advantageous country of origin terms, which in practice often means circumventing German working hours and wage regulations (Leiber et al. 2019). Some agencies located in Germany however aspire to be recognised as legitimate care providers under the LTCI scheme (Leiber et al. 2019, 383). Tis mechanism could be described as market-driven formalisation: when the market takes over a social function previously carried out by the family, supply and demand initially connect through informal actors and networks. With time however, the matching of supply and demand is taken over by formal commercial intermediaries, some even developing a preference for regularising their activities and the market they participate in. Non-proft care providers Caritas and Diakonie for their part introduced alternatives to irregular "migrant in the family" employment (CariFair and FairCare). Tese initiatives are not signifcant in terms of numbers of hired migrant care workers, but they mark a move towards the formalisation of "migrant in the family" employment (Emunds and Habel 2020, 114).

German authorities tolerate irregular live-in care work, but do not regulate it. Te country did not follow the example of its smaller neighbour Austria, which introduced a legal framework regularising the grey market of migrant live-in care in 2007 (Österle and Bauer 2012). Until 2002, migrant workers could not even obtain a work permit in Germany on the basis of care work in private households. In 2002, German authorities introduced limited work permits for "household helpers", later extended to migrants providing home-based basic nursing care. Permits were granted only to workers with a valid work contract, which are still a minority among all migrant domestic workers. Te take-up of this scheme remained very low, as related administrative procedures were very complex (Karakayali 2010, 118).

Even though Germany ratifed the ILO's Convention No. 189 safeguarding basic labour rights for domestic workers, the government allows exceptions for agencies and households that apply to live-in employment, irrespective of the Convention's working time regulations. Experts, opposition parties and NGOs strongly criticise this course of action (Jaehrling and Weinkopf 2020). Te spread of "migrant in the family" care can be explained by a laissez-faire mechanism, combined with limited and superfcial formalisation: authorities neither regulate nor outlaw this type of care. Tis type of care sustains family care, which was entrenched by the LTCI legislation as a central pillar of the German elder care system. Authorities' reluctance to regulate "migrant in the family care" is often explained by the constitutional right to the inviolability of the home (Auth 2017, 345). At the same time, fragmented and superfcial formalisation might protect decision-makers against accusations of complicity in migrant care workers' exploitation/rights violations and of neglect towards care receivers' and their families' needs.

Outcome. Supported by positive media representation, "migrant in the family" care provision is now a cheap and socially acceptable option for households to individually solve their care needs (Storath 2019). Te entrenchment of this model further maintains the centrality of informal care in the German elder care system. Te migrantisation of family care is mainly driven by the market, as the state does not really support families and agencies in solving care shortages. Reforms of the LTCI framework do not address this development and focus instead on expanding the range of benefciaries, introducing new types of benefts (some targeted at family care givers) and increasing the value of monthly benefts (Nadash et al. 2018, 592).

Benefts remain too low to cover all the costs of formal home or residential care, which contributes to the comparative attractiveness of "migrant in the family" care: too low to cover the total costs of formal care, LTCI benefts are however sufcient to fnance the salary of an irregularly employed migrant care worker (Kniejska 2018, 479). Reforms expanding the range of LTCI benefciaries result in less time per caredependent person within formal care. Some households thus perceive live-in migrant care not only as a cheaper alternative to formal care, but also as a better-quality solution (Kniejska 2018, 479). Formal care is illadapted to provide support with tasks which cannot be scheduled and when there is a need for nearly constant supervision (in cases of dementia for example or when there is a high risk of falls). "Migrant in the family" care does not have such limitations and is often supplemented by formal home-based care (Böcker et al. 2017, 235).

Policy feedback. From trade unions (Böning and Stefen 2014) to care providers (bpa 2020), stakeholders criticise the undeclared character of "migrant in the family" care provision and the fact that it does not adhere to labour law, especially with regard to working time, resting periods and so on.16 Care providers do not however agree regarding the regularisation of such arrangements: while Caritas advocates for it, the Federal Association of Private Social Service Providers (*bpa*) was in the past reluctant towards formalisation (Scheiwe and Krawietz 2010, 145). Stakeholders acknowledging "migrant in the family" care and positioning themselves triggered a stakeholder pressure mechanism, resulting in this mode of care provision fnally arriving on the policy agenda.

After ffteen to twenty years of "semi-compliance", the Federal Ministry of Health and the Ministry of Labour are now discussing a possible regularisation of live-in care provision, together with ways to enforce labour law compliance in the sector.17 Te Health Ministry proposed a reform of the LTCI, which would allow (under certain conditions) recipients to spend up to forty per cent of their benefts for home care on live-in (migrant) care.

#### **4.2 The "Migrant in Formal Care" Causal Chain with Its Mechanisms**

Part one. Te quota for skilled personnel introduced in 1993 entrenches the existing segmentation of the care labour market into skilled and unskilled jobs. It also sustains a high demand for skilled care workers. Brought about by the introduction of the LTCI, provider competition further stimulates the demand for care workers, due to the multiplication of care providers that ensued. Te number of for-proft and non-proft home care providers rose from 10,820 in 1999 to 14,688 in 2019, while

<sup>16</sup>A recent judgement in the Regional Labour Court Berlin-Brandenburg could however set a precedent, as a Bulgarian live-in care worker obtained compensation from her former employer, a Bulgarian agency, because she was not paid for on-call hours (Dribbusch 2020).

<sup>17</sup>Expert interview, German Federal Health Ministry, 26 February 2020.

the number of institutions providing residential care rose from 8859 to 15,380 (Mätzke and Wiß 2017, 131–32, Statistisches Bundesamt 2020). Care provider competition did not substantially improve wages or employment and working conditions: up until today elder care workers express that they are not paid appropriately and experience extreme time pressure at work (DGB and ver.di 2018). Tese evolutions did not contribute to the attractiveness of the care sector; rather care providers face continuous recruitment difculties.

Part two. Faced with recruitment difculties, care providers recruit among vulnerable sections of the German labour market, including migrants already present in Germany (Afentakis and Maier 2013). Te active recruitment of workers from abroad only starts in the 2010s.

Part three. At frst, decision-makers do not directly address those workforce shortages in subsequent reforms of the LTCI. Reforms lead instead to an extension of benefciaries by including people with low to moderate care needs and expanding services (e.g. *Pfegestärkungsgesetze I-III*), providing higher benefts for people with dementia (*Pfegeneuausrichtungsgesetz*) and modifying entitlement rules ensuring that specifc needs are assessed properly (*Pfegestärkungsgesetz II*). Combining with the ageing of care personnel and the skilled worker quota, these measures further intensify the demand for care workers, particularly skilled ones.

In the absence of substantial policy reforms addressing recruitment difculties in the formal care sector, a similar turn to the market mechanism occurs as in family care: in the 2010s, German care providers start actively recruiting high-skilled workers abroad. Tey initially establish recruitment schemes with Southern European countries that were particularly hit by the economic crisis, such as Spain and Greece (Braeseke and Bonin 2016, 252). As many care workers hired within the framework of those schemes returned home after short periods of work in Germany or moved to the better-paid hospital sector, those initiatives are considered unsuccessful (Sell 2019, 93–94). Unsuccessful attempts with EEA18 workers prompt care providers to recruit outside of the EEA and better prepare the integration of employees hired this way into the German labour market.

<sup>18</sup>Te European Economic Area consists of the European Union, Iceland, Lichtenstein and Norway.

Part four. Recruitment outside of the EEA is made possible by the 2005 major reform of the German migration regime,19 which opens the German labour market to migrants with non-academic vocational credentials in shortage occupations such as care work (Braeseke and Bonin 2016). Te Federal Employment Agency supports care providers' foreign recruitment eforts from 2012 onwards. Te EURES (European Employment Services) network facilitates the recruitment of skilled care workers from Southern European EU member states (Krawietz and Visel 2016, 188). Moderately successful, these recruitment schemes within the EEA were a learning experience both for care providers and for public authorities. Unlike dynamics in family care, formal care is characterised by a mechanism of state-supported migrantisation, as decision-makers and state actors facilitate (through ad hoc policy changes) and actively participate in the foreign recruitment of care workers.

Part fve. Many employers "abandon their own recruitment activities in favour of outsourcing them to private sector or state-like actors (such as the German Corporation for International Cooperation, GIZ)", while others develop their own business model, combining the role of service provider and labour market intermediary (Kordes et al. 2020, 9). Similar to dynamics in family care, a mechanism of market-driven formalisation also occurs in formal care: commercial or state-like agencies act as intermediaries between workers and employers, reducing transaction costs and easing cultural brokerage after arrival (Pütz et al. 2019, 28).

Outcome. Te "migrant in formal care" model of care provision is not as entrenched in terms of prevalence and normalisation as its "migrant in the family" counterpart. Although formal elder care provision is not yet as dependent on migration as family care, a recent representative survey found that every third care provider hires personnel abroad (Evangelische Heimstiftung GmbH, COGITARIS GmbH et al. 2020). Te

<sup>19</sup>Previously restrictive, the German migration regime underwent a major reform in 2005. Te reform adapted migration policies to the needs of the German labour market. Since 1973, German law prohibited the recruitment of foreign workers. Te 2005 reform opened the door for some exceptions to this prohibition, most notably for high-skilled care professionals. Since then, the exceptions have been gradually expanded. For example, the transposition of the Blue Card directive 2009/50/EC into German law in 2012 created a more favourable framework for high-skilled labour migration. Additionally, a law facilitating the recruitment of high-skilled migrants (*Fachkräfteeinwanderungsgesetz*) was passed in 2020.

migrantisation process that we observe in the case of formal care is hybrid: market-driven and, more recently, also supported by the state.

Policy feedback. German authorities conclude several bilateral agreements with non-EEA countries targeted both at skilled care workers and at candidates seeking a care apprenticeship in Germany (Krawietz and Visel 2016). Te German Agency for Health and Nursing Professions (DeFa) dedicated to the transnational recruitment of skilled health and care workers is established in 2020. Tose evolutions result, at least in part, from a stakeholder pressure mechanism, as some private care providers advocated with federal authorities to support their eforts to recruit abroad.20

Policymakers generally assume that increasing the attractiveness of care work for native workers will decrease the sector's dependence on a migrant labour force (van Hooren 2012, 144). Te 2018 reform of the training system for nurses and care professionals (*Pfegeberufereformgesetz*) stems, at least in part, from such a concern with improving the sector's attractiveness (Bundesregierung 2018). Tis reform makes the training of care workers similar to the system that already exists in Germany for (maledominated) industrial professions. While previous school-based training programmes for nurses required an unpaid internship, candidates will now be paid while in training. According to a representative of the Federal Health Ministry, a positive side efect of this reform is that foreign credentials are now more easily recognised,21 which is likely to stimulate further foreign infows.

#### **5 Conclusions**

Tis chapter explains the migrantisation of elder care in Germany. Migrantisation is a dual process, in line with the strong dualism of the German elder care system. On the one hand, "migrant in the family" care has been an entrenched pillar of the German care system since the 1990s. On the other hand, formal care providers and public authorities have

<sup>20</sup>Expert interview, Federal Association of Private Social Service Providers (bpa), 13 January 2020.

<sup>21</sup>Expert interview, German Federal Health Ministry, 26 February 2020.

been exploring active recruitment abroad for less than a decade. "Migrant in formal care" is thus an emerging solution to care workforce shortages, although not yet as entrenched as "migrant in the family" provision. Te duality of the German care system creates a bifurcation of recruitment profles in the family and formal care settings. Most households currently recruit CEE live-in migrants without corresponding qualifcations (more rarely, with unrecognised care qualifcations from their country of origin), while formal care providers actively recruit mostly skilled non-EEA migrants.

Both migrantisation outcomes stem from the bottom-up eforts of individual actors (households and care providers) to solve the care shortages they are confronted with. As "migrant in the family" and "migrant in formal care" spread, brokering agencies emerge, formalising recruitment processes. Tose agencies are a new type of actor in the care domain, appearing precisely due to migrantisation. Public authorities get involved only at later stages—once "migrant in the family" is a widespread and socially accepted mode of care provision (to potentially regulate it), and once foreign recruitment becomes a generalised practice in formal care (to further facilitate and support it).

By adopting an actor-centred approach to migrantisation, the article identifed a new dimension of migrantisation, so far unexplored in the literature: which actors drive migrantisation? We show that migrantisation is originally driven by households themselves and the market. However, once (and if) the state acknowledges care shortages and opts to support recruitment abroad, migrantisation is then driven both by the market and by the state. We show that the mechanisms driving elder care migrantisation are also the mechanisms that maintained the basic structures of the LTCI programme in a relatively unchanged form over the last twenty-fve years.

Te migrantisation of elder care in Germany contradicts the previously held idea that "migrant in the family" provision develops in countries with low levels of state support for elder care. While Germany has comparatively high levels of social support for elder care provision, its elder care system focuses frst on upholding care by relatives, particularly through unregulated cash benefts. Tis resulted in the development of "migrant in the family" care, so far characterising less generous care regimes, but in this context manifesting in a system with substantial care benefts and supply of in-kind care.

Te development of "migrant in the family" care in Germany also disproves the expectation that larger for-proft provision crowds out informal care provision by migrant workers (Da Roit and Weicht 2013, 471). Indeed, the multiplication of for-proft care providers in Germany following the introduction of provider competition did not reduce the growth of informal live-in care. Tis is due to a strong ideal of homebased family care and capped formal care benefts.

Te German variety of formal care migrantisation also complicates the two types of "migrant in formal care" provision identifed in the literature. Te recruitment of care workers from abroad is still less pronounced in Germany than in liberal welfare regimes with more deregulated care markets, characterised by a high incidence of for-proft providers (such as the UK). Regulatory measures such as the skilled workers' quota steer recruitment eforts towards skilled migrants. Nor does the German model ft in with the channelling of migrants into low-skilled care jobs identifed in countries with high public care expenditure, low levels of undocumented migration and a strong overall segregation of migrants into low-skilled jobs (such as France or Sweden). On the contrary, the German skilled workers' quota incentivises the recruitment of skilled foreign care workers.

Foreign recruitment in Germany is still less pronounced than in the more deregulated labour markets of liberal care regimes. However, labour regulation measures (such as the skilled workers' quota) and professionalisation eforts (such as the training reforms) do not necessarily slow down migrantisation through improving the sector's overall attractiveness. On the contrary, the German case shows that regulatory policies can intensify shortages of skilled care workers, encouraging care providers and public authorities to recruit abroad.

As wage levels between Western and CEE countries are likely to gradually equalise, the importance of CEE as a source region for the German elder care system should decrease, to the beneft of countries further East and South. Such a shift is already ongoing, as the share of live-ins from Bulgaria, Romania and Ukraine increases (to the detriment of Poland), while formal care recruitment targets non-EEA countries. It is however unlikely that the reliance of the German elder care system on migrant workers will end any time soon, as care needs will most probably intensify due to demographic and societal dynamics. Tus, migrantisation in Germany contributes to existing or emerging care gaps in migrants' countries of origin. Tese gaps in turn stimulate the formation of transnational care chains, such as the one already linking Ukraine and Poland (Safuta et al. 2016).

## **References**


Restrictions. In *Illegal Immigration in Europe. Beyond Control?* ed. Franck Düvell, 75–105. Houndmills, Basingstoke: Palgrave Macmillan.


*in Deutschland* [Te German Elder Care Barometer: Impressions from Our Sector in Germany]. Vincentz Network. https://www.vincentz.de/wp-content/ uploads/2020/10/Deutsches-Altenpfegebarometer-20200929.pdf. Accessed January 21, 2020.


<sup>———. 2018.</sup> In-Between-Carer: Towards a New Type of Elder Care Worker? Te Example of Polish Migrant Care Workers in Germany. *International Journal of Care and Caring* 2 (4): 477–498.


———. 2011. Care, Gender and Migration: Towards a Teory of Transnational Domestic Work Migration in Europe. *Journal of Contemporary European Studies* 19 (3): 349–364.


*Betreuung*, ed. Bernhard Weicht and August Österle, 221–244. Münster: LIT Verlag.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **Part V**

**Causal Mechanisms and Social Policies in Latin American Countries**

# **11**

## **Transnational Events and National Health Reform: The Latin American Medical Congresses and the Legitimisation of Public Health Reforms in Chile and Uruguay in the Early Twentieth Century**

**Delia González de Reufels and Teresa Huhle**

## **1 Introduction**

Te introduction of discrete public health policies was accomplished in Latin America by a range of diferent actors who gained importance in the late nineteenth and early twentieth centuries. Amongst them were workers' unions, the women's movement, and medical professionals, in

University of Bremen, Bremen, Germany e-mail: dgr@uni-bremen.de

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

D. González de Reufels (\*)

particular. Tis case study on Chile and Uruguay puts the latter at the centre of attention. It analyses how medical professionals from two countries that were pioneers in public health used transnational venues to exchange concepts and ideas on public health policies and sought to strengthen their domestic political position. Te Latin American Medical Congresses are highly important as venues which enabled doctors to address, among others, questions of power and resources while they also highlighted their monopoly on questions of health and disease prevention. Furthermore, the Latin American Medical Congresses also served to legitimise a new and highly ambitious epistemic community which sought to validate its activities. At the same time, these congresses ofered the various national communities of doctors the opportunity to show their skills and the superiority of their training. Tis chapter therefore focusses on four Latin American Medical Congresses that took place between 1901 and 1909. Tese were important forums for the exchange of medical knowledge and ideas during the early twentieth century that have largely been ignored by historical research (de Almeida 2006) and political science. Te congresses show how two mechanisms—*legitimisation* and *competitive cooperation*—were at play in a process that positioned the medical profession at the centre of state policies in health and served to cement its leadership in a new feld of state action.

Te analysis of the two mechanisms, their interplay and dynamics also draws attention to the timing of the congresses. Tese took place at a very remarkable time in Latin American history because back then, none of the countries had yet established any social security system. Te medical profession was still young, and at an early stage of professionalisation, and it was breaking ground in a new feld which was still in the making: public health. Tus, this period ofered these historical actors a unique window of opportunity for the transnational debate on health and social policy introduction. Te fact that the Latin American nations had begun to consolidate since the second half of the nineteenth century only adds

T. Huhle

University of Cologne, Cologne, Germany e-mail: teresa.huhle@uni-koeln.de

importance to the congresses. Te new republics had just left behind violent conficts over independence and were undergoing an unprecedented transformation whose results were difcult to predict (Sábato 2018); yet, by the last quarter of the nineteenth century, it was clear that the independent Latin American nations were ready to embark on a new phase of their existence that would redefne their relationship with each other. Te congresses were only one manifestation of this change.

Te timing of the congresses is also relevant in a further respect. During the frst decades of the twentieth century, nation states in the Americas increasingly perceived the health of their citizens to be endangered by endemic and epidemic diseases. Tis, in turn, increased the attention they paid to hygiene and disease prevention. Particularly the latter demanded more cooperation among Latin American countries as epidemics did not stop at national borders at a time when trade and migration were increasing. Historians of public health in Latin America have furthermore highlighted that this was when "cooperation, including in the sanitary realm, became a genuine possibility" (Birn et al. 2017, 20). Tus, the congresses highlight the existence of an "inter-state collaboration in South America" which resonates with current Latin American debates on how to control the COVID-19 pandemic and how to act together (Birn 2020, 357).

Te willingness to cooperate and discuss the future of public health and the medical profession had certainly made the Latin American Medical Congresses possible in the frst place, but this cooperation also went hand in hand with competition. And this competition was twofold: it referred to the Latin American countries that sent delegations as well as to Europe, the example medical doctors sought frst to emulate and then to surpass. European doctors and Latin American doctors who had been trained overseas were excluded from the congresses; nevertheless, Europe was ever present at these events: it provided the medical profession in the Americas with a model, it provided a road-map to professionalisation and set the standards for public health development as well as for social security systems. While the latter were not debated at the Latin American Medical Congresses, the state of European medical science and public health inevitably were. In sum, the Latin American Medical Congresses provided an ideal space for competitive cooperation and for legitimisation. To analyse it, this chapter focuses on two countries that were not only Latin American pioneers in the feld of public health, but also pace setters at these congresses: Chile and Uruguay.

In both countries, the early professionalisation of medical doctors had contributed to the formation of an important epistemic community whose relevance to the development of public health is part of the history of the medical congresses studied here. Te congresses also fostered the careers of individual doctors who presented important papers at these events and commended themselves to their respective national states as advisors and partners in public health development. While the medical profession was an important driving force behind national public health policies, it also was an interested party in the process of policy development. In this way, the path towards public health legislation evidences "the close alliance of medical and state interests" that historian Julia Rodríguez has observed for Argentina (Rodríguez 2006, 40).

Tis alliance is particularly valid for the cases of Chile and Uruguay, which also confrm that while in Latin America medicine and public health were "(…) acutely political, (…) even their most nationalist expressions were internationalist and cosmopolitan in origin, orientation, and networks" (Cueto and Palmer 2015, 59). Te Latin American Medical Congresses thus call attention to the transnational dimension of national health policies and underscore the interlinkage between "nationalism and internationalism" of health that political scientist Evelyn Huber identifed (Huber 2006, 458).

Chilean and Uruguayan doctors were very aware of this interlinkage. Tey shared an interest in changing national public health policies and at the same time pushed for public health agendas for all of Latin America. Overall, both nations performed well at the Latin American Medical Congresses; however, the dynamics and rhythm of public health policy implementation varied signifcantly in the two countries. Even more, although Chile and Uruguay regularly sent large delegations to the Latin American Medical Congresses and, in turn, hosted one of the events, these events meant diferent things for the respective epistemic community. While Chilean doctors were intent on fostering cooperation within the Southern Cone and competing with respect to medical progress in the region, Uruguayans took special pride in challenging Europe's leading role, claiming to be on a par or even outrivalling it. Tus, the congresses were about bringing the leading Latin American physicians together, while they also were about putting Europe and its achievements into perspective.

Tis chapter will trace the mechanism of competitive cooperation and the mechanism of legitimisation that can explain the activities of physicians who successfully lobbied for increased state involvement in public health. Taken together, both complex causal mechanisms explain the dynamics and the impact of the events on public health ideas. Particularly, the competitive cooperation mechanism highlights the importance of transnational factors for understanding national social policy developments that are increasingly gaining importance among social policy researchers (Leisering 2019).

By drawing attention to both complex causal mechanisms, this analysis furthermore hopes to understand the limitations of the community's "authoritative claim", analysing both the potential and the limitations of epistemic communities at the time. In this way, this chapter accounts for the success as well as for the failure of one specifc epistemic community to promote and impose discrete policy measures within the national context.

Te empirical research is based on the close reading and analysis of proceedings of the frst four Latin American Medical Congresses which took place between 1901 and 1909. Teir proceedings fll various volumes, and this material is complemented with other contemporary publications in the feld such as books on public health and its implications for state action, personal memoirs, and medical journals.

Tis chapter starts with a recapitulation of the concept of an epistemic community, a frst examination of the two complex causal mechanisms, and an overview of the congresses which then will be followed by two case studies, starting with Chile which initiated and hosted the frst congress in 1901. Te importance of this congress lies not only in the topics it covered but also in that it became the model for all future congresses. While it was deemed a huge triumph, which added to national prestige, unlike other members of the epistemic community, Chilean doctors frequently failed at imposing their policy preferences. Chilean physicians had succeeded in legitimising themselves through transnational events, but at the beginning of the twentieth century they still did not wield enough power to change national social policies or establish a ministry of health. Tis is the more revealing as this level of institutionalisation was then one of the main goals of the profession in Chile. When the *Ministerio de Higiene, Asistencia y Previsión Social* was eventually founded in 1924, its frst minister was one of the doctors who had been involved with the Latin American Medical Congresses, as we shall see. Nevertheless, the Chilean experience difered signifcantly from the Uruguayan.

Te second case study analyses Uruguay at the frst four congresses through a specifc group of medical doctors, focusing in particular on public health expert and renown medical doctor José Scosería. Tis case study also assesses interlinkages between these international events and national public health debates about the foundation of a new central institution, the *Asistencia Pública Nacional* (1910). In 1932, that is 11 years after the last congress had taken place, it fnally merged into the Uruguayan Ministry of Health. A conclusion sums up the fndings with regard to the two complex causal mechanisms and assesses the diferences between the two cases with regard to the political use they could make of these events.

## **2 The Epistemic Community of Medical Doctors and the Mechanisms of Competitive Cooperation and Legitimisation at the Latin American Medical Congresses, 1901–1909**

We conceptualise the role of the medical profession in Chile and Uruguay at the time of the medical congresses as a well-established epistemic community. Te mechanisms of competitive cooperation and legitimisation can explain how they carved out a position for themselves within the newly established feld of public health. With respect to the concept of epistemic community, we follow the defnition proposed by political scientist Peter Haas who defned an epistemic community as a "network of professionals with recognized expertise and competence in a particular domain and an authoritative claim to policy-relevant knowledge within that domain" (Haas 1992, 3).

Te Latin American Medical Congresses served to expand and strengthen the connections within this network. We argue that crucial aspects of this development can be understood by using a mechanismbased approach to identify specifc sequences within these interconnected processes and to explore the potential of comparative analysis (Kuhlmann and Nullmeier 2021). While historical studies have studied the establishment and development of expert networks and their infuence on decision- and policymakers in countries such as China (Greenhalgh 2008, 195), recent research in political science has begun to focus on epistemic communities in connection with complex causal mechanisms to understand their infuence on decision-makers and the role of policy entrepreneurs (Löblová 2018; Safuta 2021). In this way, it is possible to elucidate the question why some epistemic communities have been more successful than others in bringing about policy change in domestic contexts. Ultimately, by studying smaller sequences, larger processes of infuencing policy decisions can be disaggregated to really explain how "knowledge and evidence make its way into policy" (Löblová 2018, 161).

Te legitimisation mechanism helps us to understand how the epistemic community of Latin American doctors sought to claim its role and position that in the future would enable it to infuence decisions. Te competitive cooperation mechanism, on the other hand, asserted their position within transnational and international contexts.

Te actors involved were medical doctors who had been trained at one of the new national schools of medicine, and they epitomised the epistemic community which soon enough would encompass nurses and other providers of healthcare. At the time of the Latin American Medical Congresses, these groups were not yet included and thus not invited to the congresses.

Te Latin American Medical Congresses were events which brought the epistemic community together, and as they are important for this chapter, some preliminary understanding of the congresses is required. While they were prestigious and important at their time, they have overall met with little scholarly attention. Brazilian historian of science Marta de Almeida is a notable exception, and she has identifed and analysed the congresses as important venues for the exchange of knowledge and ideas. Te congresses also fostered the professionalisation of medicine and the legitimisation of physicians as political actors (de Almeida 2004, 2006). Others have only been interested in the congresses to trace the rise of competing epidemiological theories (Caponi 2002), to indicate the growing rapprochement between Latin American academic medical communities (González Leandri 2013), and to illustrate how public health development in one country, in this case Argentina, was showcased (Veronelli and Veronelli Correch 2004).

In order to situate the congresses in time and space, it is important to note that during the time of the Latin American Medical Congresses under study here, medical congresses had become more numerous and more important and awareness of the spread of diseases had grown, in particular in Europe (Zylberman 2006). Also, within the contemporary landscape of sanitary and medical congresses, the Latin American Medical Congresses managed to fll a gap. Until then, Latin American physicians and hygienists had mainly met at three diferent types of venues: international sanitary conferences in South America (1873, 1878, 1888, see Chaves 2013), the Pan-American Medical Congresses that had been inaugurated in the United States in 1893, and the Latin American Scientifc Congresses (Bastias Saavedra and Plaza Armijo 2016). Te latter had been taking place since 1898 and counted with large medical sections.

Te Latin American Medical Congresses, on the other hand, focused exclusively on medicine and health, including sanitary questions related to epidemics, and were meant to be free of infuence from the United States. Te latter had begun to claim a leading role in matters as diverse as scientifc development, medical progress, and public health (García 1981). Envisioned and organised by Chilean doctors, the frst Latin American Medical Congress was an important innovation that can also be viewed as a continuation of the international medical congresses that had been taking place in European cities since 1870 with mainly European participants. Following the example of European medical doctors, the Latin American organisers of their continental medical congress aspired not only to showcase the medical state of the art, but also to make themselves heard in matters of public health and hygiene.

Te frst four Latin American Medical Congresses were hosted exclusively by countries of the Southern Cone: the initial congress took place in Santiago de Chile in 1901, and the following congresses were held in Buenos Aires, Argentina, in 1904, in Montevideo, Uruguay, in 1907, and in Rio de Janeiro, Brazil, in 1909. Tese events emphasise the pioneering role of these four countries of the Southern Cone which, in many ways, were closely connected to each other and functioned as a "corridor of ideas" (Biagini 2000; Kuhlmann et al. 2020). When the congress reassembled for the ffth time in Lima, Peru, in 1913, this marked an important shift of focus: on the occasion of this event, the Latin American Medical Congress merged with the Pan-American Medical Congress (de Almeida 2006, 741). Finally, the sixth and last Latin American Medical Congress took place in Havana, Cuba, in 1922.

Overall, the Latin American Medical Congresses started of as and can be characterised as events of the Southern Cone: in particular, the frst four congresses were dominated by physicians from the region whose large delegations were tirelessly delivering the largest number of papers. As a direct outcome and a mirror of medical professional consolidation, these congresses ofered the epistemic community an occasion to celebrate itself. On the occasion of the congresses, the epistemic community of Latin American medical doctors for the frst time defned itself by its sites of training and thus legitimised its aspiration to debate and decide public health matters. Chilean medical doctors already had excluded medical doctors who had been trained outside of Latin America in the context of their frst national congress in 1889 (see Sect. 3 of this chapter). Also, it is noteworthy that all these congresses were cost intensive and needed the backing of the national governments. Te role of the host required a budget for the opening and closing ceremonies, the publication of the conference proceedings and other related costs, which were considerable. Te conference proceedings show that these congresses clearly put fnancial strains on national governments; also, it was expensive to send delegations to the congresses, which is why the conference calls usually asked Latin American governments to help the epistemic community with travel expenses ("Circular enviada" 1902). Te fact that these calls were answered, at least in the cases studied here, reinforces the argument of the great importance these events had. It also points to a successful rise of medicine as a profession which could claim public funding for its events.

## **3 Calling for Cooperation While Excelling in Competition: The Chilean Epistemic Community at the Medical Congress of 1901**

### **3.1 The Historical National Background**

At the beginning of the twentieth century, the Republic of Chile was in the midst of preparations for the celebrations on the occasion of its frst 100 years of independence from Spain. Tis was an opportunity to assess national development, and Chilean medical doctors in particular found much to criticise: unlike its hemispheric neighbour Uruguay, Chile was not undertaking broad public health reforms. After independence, the state had mainly invested in education (Mac-Clure 2012). It had been reluctant to intervene in health, yet public health was much debated in the last third of the nineteenth century. Real progress was only made from 1918 onwards and in particular during the 1920s under presidents Alessandri and Ibañez del Campo. Also, the new constitution of 1925 was another important milestone (Rengifo 2017). Welcomed with high hopes for a better future, this normative text for the frst time in Chilean history explicitly included health as a feld of state intervention which Ibañez del Campo would later defne as "not only the absence of illness but (…) the plentifulness of life" (Góngora 1981, 85).1 Both politicians left an important mark on the structure of the Chilean national state, as Góngora has rightly pointed out, because they coincided in stressing that it was the state's responsibility to take care of all the diferent groups and classes of society, including their health (Góngora 1981, 88). Te new and all-encompassing concept of health which, among others, concerned living conditions and the workplace was further expanded until the

<sup>1</sup>All translation from the Spanish original were made by the authors.

military coup in 1973 and system change set a new agenda (Cruz-Coke Madrid 1988).

In 1900, however, the Chilean state was still in its formational phase. It had not fully assumed its interventionist, albeit paternalistic role, and social unrest was revealing the contradictions and limitations of modernisation (Rinke 2002), while the "social question" remained unresolved as labour unions and protests were calling into question Chile's path for economic and social development of Chile. Economic growth served only a few because, in spite of its wealth, Chile had failed to address widespread poverty and social inequality which also impacted public health. In comparison to Uruguay, the Chilean state was lagging behind: while it did institutionalise health and established a ministry of health in 1924, it only implemented the *Servicio Nacional de Salud Pública* in 1931, long after the Uruguayan *Asistencia Pública* had been established.

Te epistemic community of Chilean medical doctors, however, had come a long way comparatively fast: within three decades, it had shed the image of a dirty and unprofessional feld that members of the elite would shy away from (Cruz-Coke Madrid 1995). At the end of the nineteenth and early twentieth centuries, it showed all the "ideal observable manifestations" Olga Löblova calls indispensable (Löblová 2018, 165) to speak of an epistemic community. Chilean physicians counted with a medical society, a scientifc journal, and members who were well known within the community and beyond. It also had successfully sought the support of other important groups such as the armed forces; it had promoted discrete policies such as physical education at schools and advocated for vaccination against smallpox which eventually would be implemented by law (González de Reufels 2020). Finally, a new medical school and an innovative national conference whose proceedings were published fve years after the event had taken place attested to its success, while it clearly ranked among the leading Latin American nations in medicine and was proud of it (Maira 1893).

Nevertheless, the epistemic community was still in need of legitimisation to be able to consolidate its bureaucratic power. So far, it had aimed at political accommodation of its interests by styling itself as a resource of the state, rendering relevant services as articles in the journal *Revista Médica de Chile* show. Te journal explicitly promoted public health and emphasised that it was meant to be a publication that ofered policyrelevant knowledge and data (Schneider 1872). Nevertheless, the political reach of Chilean medical doctors remained limited, which is remarkable at a point in history defned by economic expansion and increased exchanges of people, goods, and disease (Harrison 2012). Te epistemic community was not yet able to push through all its propositions for public health, and it had not reached the level of institutionalisation it wanted: a ministry of health was still necessary to turn doctors into members of national government and increase their resources, guaranteeing access to the inner circle of national politics. Chilean doctors felt that they had come far, but not far enough.

When it came to representing the nation at European events, Chilean decision-makers relied on the epistemic community. For example, the government sent of doctors to the Paris exposition of 1889 which certainly was not only a "market of wares of nineteenth-century industrial capitalism" (Rodgers 1998, 8), but also a venue to discuss hygiene and social assistance with European peers and to compete with them. Tus, renown Chilean medical doctors were called upon to impress the Europeans with their achievements and report on the progress of the country to, as one physician wrote, make "us [i.e. Chile] known abroad" (Murillo 1889, VII). Tus, the ability to excel in competition with European experts emphasised the importance of the community that was meant to put Chile on the world map with respect to medical progress, even though at home the epistemic community continued to struggle for legitimisation and access to political power. Like their US counterparts, Chilean physicians had upheld their authority to "defne and interpret the standards and the understandings that govern medical work" (Starr 1982, 421), but their calls for new legislation such as a national sanitary code and new agencies would not be heard until 1918 and 1925, respectively (Cruz-Coke Madrid 1995, 412). Also, it took a military coup and a short-term regime change to create the national ministry concerned with hygiene, public assistance, and social provision. It was led by Alejandro del Río Soto-Aguilar, who had been trained as a doctor in Chile, then received a fellowship to spend three years in Europe and returned to Chile to become involved in the medical journal and one of the most infuential medical doctors in the country (Cruz-Coke Madrid 1995, 479). He had also been a regular participant at the Latin American Medical Congresses, which emphasises the nexus between the congresses and nascent national health institutions in Chile. Te connections between medical doctors and national politics were manifold because as early as the nineteenth century, medical doctors had joined the Chilean National Congress to increase funds and the political infuence of the epistemic community (Cruz-Coke Madrid 1988), but health reforms were stalling at the turn of the century because the community's reach remained limited.

When Chilean doctors began to organise the transnational event in 1900, it was based on the concept of their frst national medical congress and drew on this experience. Te Latin American event featured opportunities for cooperation within the hemispheric region and promised to compete with leading European venues while also showing of the reach of the epistemic community in Chile itself, which can be explained by the mechanism of competitive cooperation. Te mechanism of legitimisation can be identifed in the attempts to increase the epistemic community's leverage on national Chilean politics and to promote public health as the basis of a future healthcare system. Also, by 1900 high infant mortality rates such as Chile's were considered alarming as they had become synonymous with a lack in modernity and progress in public welfare and hygiene (Chávez Zúñiga 2019). To succeed in public health became even more pressing as population growth was stalling, and cases of highly infectious diseases such as syphilis and tuberculosis soared. Tey were considered especially dangerous as there was not yet any reliable cure for them, and they damaged the health of future generations.

#### **3.2 Chile at the First Medical Congress in 1901**

Te frst Latin American Medical Congress was held in Santiago de Chile in 1901 to underscore the leading role Chilean medical doctors claimed amongst their hemispheric and European peers. Tis event did not include any female doctors although in 1886 Eloisa Díaz, the frst Chilean female physician, had graduated from the national medical school. Her graduation was especially applauded by the epistemic community and in retrospect highlights its ability to change over time to include new historic actors (Maira 1893). It is important to note that the congress of 1901 set the example of all congresses until 1922, and in this way, the Chilean epistemic community left its mark on a transnational event. Tis was particularly the case because the event furthermore built directly on the frst Chilean medical congress of 1889 that was characterised by contemporaries as the "seed" of the Latin American congress (Pérez Canto 1910). Just like this national congress, the frst Latin American Medical Congress excluded medical doctors who had been trained in Europe and only admitted younger and nationally trained medical doctors (Primer Congreso 1901). Tey represented the future of the epistemic community that could now compete with European peers. Still, the congress followed the blueprint of European medical congresses and copied their organisation, sections, and publications. In spite of this act of copying European events, the mechanism of competitive cooperation explains why the congress expressly excluded "Europeans" ("Bases del congreso" 1901, xiii), as did all the following congresses. Tis also shows how the mechanism of competitive cooperation is connected to the legitimation mechanism, because the new "national quality" legitimised the claims of the Chilean epistemic community and drew attention to the equally "national quality" of the public health challenges it was addressing.

Also, the Congress of 1901 put the epistemic community and its achievements in medical science and public hygiene on display. Again, this feature of the congress brought Latin American doctors together to learn from each other, as the calls to solidarity and cooperation emphasised which reminds us that cooperation was in fact crucial. Te advances in medical science and the enormity of the task that all republics had to take on in public health were at the centre of the frst congress of 1901. Also, hygiene loomed large at a special exhibit which was huge and had been organised since 1900 and for which manufacturers and experts had been approached in time ("Crónica" 1900, 439). It was to inform the general public and medical doctors about the state of the art in sanitation and healthcare, thus the *Exposición de Higiene* and the congress of 1901 functioned as interconnected spaces of policy debate and policy presentation. Furthermore, this was a marketplace for medical equipment, while it also was a forum to display diferent national social policy measures. Hemispheric neighbour Argentina, for example, asserted its superiority and competitiveness as well as its willingness to cooperate in public health by providing material which, as one visitor would later on write, "illustrated very much the discussion of the problems of health and hygiene which are impressively current amongst us" ("Revista de la exposición de hygiene" 1901, 58), that is the region as a whole. Tis was a moment to study, copy, or reinvent the measures other nations had implemented which also explains why the Chilean *Instituto de Higiene*, founded in 1892 (Cruz-Coke Madrid 1995, 414), was part of the *Exposición de Higiene*. It used this exhibit to display its services and eforts to improve the sanitary conditions of Santiago while skilfully drawing attention to the shortcomings of national hygienic infrastructure: drawings of the sewage systems of the German city of Berlin reminded visitors of the lack of water management in the capital and elsewhere in the country. In this way, the institute's exhibit also gave ample proof of how important its work was at a time when Santiago obviously was losing the competition with Europe.

But, as stated before, this congress was also about cooperation: the congress of 1901 invited medical doctors to work together on solutions for recurrent health problems in the hemisphere and argued for the establishment of networks of colleagues: acclaimed as an event whose most valuable result was that it had "created enthusiasm for scientifc collaboration" ("Crónica" 1901, 128), this event called for "fraternity and union under the banner of science" ("Nuestras felicitaciones" 1901, 7). Both, fraternity and union, would serve to legitimise the epistemic community in its national context, while this also asserted a new identity as a pioneer in the Southern Cone. In spite of this, the papers presented at the congress in 1901 also sought to assess Chile's progress in comparison to Europe where dentistry, for example, had become a feld in its own right and vaccination campaigns were held widely.

In the same vein as this congress, all other Latin American Medical Congresses that were to follow rendered cooperation with the neighbouring republics and competition amongst Latin American nations and with Europe recurrent themes. Still, the congress of 1901 was special in that the largest group present was made up of Chileans who used this event to show of their capacity in front of the Chilean government. Although this was considered an international event that counted with acclaimed Latin American physicians, such as the Argentinian doctor Emilio Coni, the congress largely functioned as a national event that was held for a national audience and served to legitimise the epistemic community. Also, some of the foreign attendants were not medical doctors but diplomats and representatives of the Latin American republics who resided in Santiago and could hardly interfere with the agenda the Chilean epistemic community had set for the congress. Terefore, it is not surprising that the event picked up on national Chilean debates on, for example, alcoholism, tuberculosis, or venereal diseases such as syphilis which had long been identifed as endemic and a burden to Chile (Murillo 1869). While these diseases were also discussed in the neighbouring countries, they were considered to be particularly rampant in Chile and especially detrimental to demographic growth (Chávez Zúñiga 2019). Given the absence of so many of their Latin American peers, the Chilean doctors in the end debated many issues largely amongst themselves. Current worries were voiced, for example, about the spread of leprosy, which was debated at a session during the frst day of the congress, and there were calls to create common standards in the medical disciplines across all of Latin America ("Congreso Médico Latino-Americano" 1901, 18–19). Infectious diseases claimed special attention as did epidemics whose prevention required cooperation between the states as smallpox and bubonic plague had only recently returned to Argentina ("La peste bubónica" 1900). Now, this was an opportunity to show that Chileans were competent in fghting such an outbreak. After all, the epistemic community was aware that, as the medical journal put it, epidemics did not stop at national borders and measures had to be taken to stop the advent of the plague ("La peste bubónica" 1900). Competitive cooperation among Latin American doctors amplifed the eforts and reach of Chilean doctors who hoped to achieve national reform, for example, in the feld of child health and wished to see progress in related felds much faster.

Last but not least, the congress of 1901 was a social event which added lustre to the epistemic community and thus legitimised its current position and its claims for increased infuence in the near future. Its opening took place at the theatre in Santiago where plays and concerts would usually take place. As the authors of the conference proceedings have pointed out, the event counted with the massive presence of the Chilean government, turning this congress into a political event which enabled physicians to openly ask for political support and funds ("Sesión de apertura" 1901). Furthermore, the debates at this congress included, amongst others, housing problems and the risks of unhealthy living conditions, which were especially evident in Santiago and other large Chilean cities. Tese debates echoed national debates which Chilean medical doctors had unsuccessfully pushed for some time, such as sanitary and decent housing for the masses. Although it is difcult to pinpoint the development of Chilean legislation to the frst Latin American Medical Congress, the *Ley de Habitaciones Populares* fnally was passed in 1906. Equally important was the question of how to advance the treatment of syphilis and tuberculosis, and because the Chilean doctors had lamented the policymakers' inertia, the resolution of the Latin American Medical Congress on the formation of a Commission on Tuberculosis was very much applauded.

Te newly won awareness of the need to take action in the feld of health served the interests of the epistemic community. It chose to cooperate within the hemisphere to be up to the competition with Europe and to thus advance its standing. Nevertheless, Chilean doctors only succeeded much later at implementing policy measures debated at the Latin American Medical Congress at the national level because their propositions met with resistance on all levels of the Chilean state. Political actors in the Chilean national congress and the Chilean senate, on the municipal level and within state administrations, were not convinced of innovation in the feld of public health (Cruz-Coke Madrid 1995; Laborde Duronea 2002, 40) and thus did not support policy changes. Te medical congresses of 1904, 1907, and 1909, in turn, raised important issues such as the establishment of systematic vaccination schemes and specialised hospitals for children that required active political support the Chilean doctors could not yet count on: compulsory smallpox vaccination and vaccination schemes in general would only be achieved in 1918 when the Código Sanitario and the law 3385 were approved. Still, hospitals for children only came into existence in the 1920s. Terefore, when Chile celebrated its frst 100 years of nationhood in 1910, Dr. Pérez Canto vented his anger and wrote that all progress made in matters relating to health was due to the initiatives of the professional society and not to the government (Pérez Canto 1910). Although it had successfully performed at the frst Latin American Medical Congress and would continue to do so over the next years, the epistemic community of medical doctors in Chile was left to its own devices when it attended the medical congresses studied here.

## **4 Envisioning Cooperation While Winning the Competition: Uruguayan Policy Entrepreneurs at the Latin American Medical Congresses**

#### **4.1 The Historical Background in Uruguay**

Te Latin American Medical Congresses coincided with a major restructuring and expansion of the public provision of health, sanitation, and hygienic infrastructure in Uruguay. Changes in the public health system were closely related to a broader transformation of the Uruguayan state that was stimulated by the *batllistas*, a fraction of the liberal Colorado Party that owes its name to two-time president José Batlle y Ordoñez (1903–1907, 1911–1915). A frst modernising reform era in the 1880s had seen the establishment of a comprehensive public education system (Hentschke 2016). After decades of civil war, the *batllistas* sought to radically break with the past to construct a new country from scratch (Caetano 2000, 16). Tis was enabled by the expansion of state institutions, the creation of state enterprises, and the co-optation of organised workers through labour and social reforms. Here, the key to the expansion of the state was the reconstruction and foundation of public institutions.

In the feld of public health, the creation of the umbrella institution *Asistencia Pública Nacional* in 1910 was the most important and visible embodiment of the "model country" Batlle y Ordoñez aimed to create. Te history of this institution was closely linked to the career of physician Dr. José Scosería. Born in 1861, Scosería belonged to an infuential generation of Uruguayan medical pioneers and *higienistas* who had started their careers when "everything was still to be done" (Soiza Larrosa 2010), both within the academic and the political sphere. In many ways, these men were pioneers and convinced of the public relevance of their medical knowledge and succeeded in forming an infuential epistemic community. Many physicians within this community gained individual fame both within and outside of Uruguay for their medical expertise, but especially for their role in creating and shaping Uruguay's public health institution. Among the most prominent fgures are Luis Morquio, who conducted research on child mortality and directed infant welfare services, Rafael Schiafno, who specialised on school hygiene, and Augusto Turenne, an obstetrician who would lead the feld of maternal health. José Scosería, on the other hand, did not primarily gain fame for a specifc feld of medicine or public health, but quickly climbed up from one directing position to the next. As most of his peers, he had studied medicine at the Faculty of Medicine of Uruguay's public university in Montevideo. Scosería enrolled in 1880, only four years after the faculty's foundation. From 1898 to 1904 he served as the faculty's dean, while his public health career started in the *Comisión Nacional de Caridad y Benefciencia* in 1903, of which he became the director in 1905. From within this institution, he worked to reform and transform it into the *Asistencia Pública Nacional*, becoming its frst director in 1910. Te *Asistencia* strengthened and centralised state control over public hospitals, orphanages, asylums, and so on and, most importantly, followed through with the secularisation of these institutions. Te religious congregations that had been taking care of the sick and dependent since the institutions' founding in the mid-nineteenth century were expulsed as part of a power struggle between diferent generations and groups of actors. Here, the secular and liberal *batllista* physicians that Scosería belonged to, and the more conservative and Catholic practitioners that had dominated the *Comisión de Caridad* until Scosería became its director in 1905, had been at odds. Although Scosería succeeded in implementing reforms from 1905 onwards, his actions were far from uncontested: heatedly debated in congress and the national public, the congress fnally adopted the *Ley de Asistencia Pública Nacional* on November 7, 1910.

With this background and time frame in mind, this case study argues that the Latin American Medical Congresses from 1901 or rather 1904 to 1909 served as an important forum for exchange and cooperation for the Uruguayan policy entrepreneurs, while they were also a venue to promote, gain support, and legitimise this national reform process. Tis case study thus traces the mechanisms of competitive cooperation and legitimisation that were driving forces of Uruguayan reform ideas at the congresses. Te congresses further point to how the mechanism of legitimisation worked for José Scosería and the medical doctors around him to favourably infuence public opinion and politicians in Uruguay. Te Uruguayan case furthermore complements the analysis of the two complex causal mechanisms with classical studies on the role of policy entrepreneurs as social policymakers (Heclo 1974). Just like their British and Swedish counterparts, Uruguayan reformers were trained as physicians, but when it came to the feld of public health they merely were "talented amateurs" (Heclo 1974, 309).

#### **4.2 Uruguay at the Latin American Medical Congresses, 1901–1909**

Te frst Latin American Medical Congress in 1901 counted with very little Uruguayan presence.2 One possible explanation is that Uruguay was to host the Second Latin American Scientifc Congress in Montevideo two months later, and its organising committee was mainly made up of physicians with a focus on public health (Segunda Reunión del Congreso Científco Latino-Americano 1901, III).

But from 1904 onwards, the who is who of Uruguayan public health attended the Latin American Medical Congresses, starting with over 100 individual and a dozen institutional participants from Uruguay at the congress in Buenos Aires. A similar number arrived from Brazil and even more from Chile ("Nómina de Miembros Adherentes" 1904, 133–146). By 1904, José Batlle y Ordoñez had just come into ofce for the frst time, and José Scosería had begun to work for the *Comisión Nacional de Caridad y Benefciencia*, a position from where he could promote his idea to replace charity with social rights. Tanks to this position, Uruguay's

<sup>2</sup>From the published proceedings of the conference, we know that the Uruguayan minister to Chile served as Uruguay's ofcial delegate and that three papers by Uruguayan participants were presented: two from the same Uruguayan pharmacist and one by a physician. Te proceedings do not include a list of participants, so it is not clear whether more Uruguayans attended the congress.

participation at the Second Medical Congress was already in his hands: he presided over the Uruguayan delegation and was elected as honorary president of the congress and as head of the commission to organise the next congress that was to take place in Montevideo in 1907. Te existence of national delegations and preparatory committees within all participating countries underlines the strong role of the nation state in this endeavour to stimulate international and transnational cooperation in the region. In the Uruguayan case, the national committees were instituted and fnanced by the government. Tese committees were dominated by public health professionals who elaborated strategies on how to best present national progress that was always framed in a competitive manner. Furthermore, a letter from Scosería, written as president of the preparatory committee to the Uruguayan Minister of the Interior in the context of the congress in Rio in 1909, illustrates his desire to subordinate all the national participants and exhibitors to the committee's master plan (Scosería and Etchepare 1909). He was thus a policy entrepreneur who not only interacted with all parties involved in the fnancing and preparation of the congress, but who could also use his ofcial position to exert infuence on all Uruguayan participation at the exchange venue according to his own ideas.

Te third Latin American Medical Congress in Montevideo in 1907 came at the perfect time for the Uruguayan policy entrepreneurs: as hosts of the congress, they could showcase before an international audience the progress of Uruguay's public health institutions in general as well as of the commission that Scosería had directed since 1905 in particular. And they intended to channel international praise for that progress to strengthen their position in the national debates about the restructuring of the *Comisión de Caridad* to the *Asistencia Pública Nacional*. As at the other congresses, Uruguayans gave several presentations that refected the joint interests of the transnational epistemic community. Luis Morquio stood out among the Uruguayan speakers with two lengthy presentations on school hygiene and infant mortality. As hosts, the Uruguayan delegation further had the largest number of exhibitors and ofered visits to many public health institutions in the capital and its surroundings. Te conference proceedings, as always written and edited by the hosts, listed all these excursions, the applause Uruguay had received from its distinguished foreign visitors and some of the speeches delivered (Pou Orfla 1908, 73–87). Tis is how we know, for example, that the Director of Public Health of the city of Montevideo meant to "show our true aim for progress" when he mentioned that Uruguayan public institutions had started to serve pasteurised milk exclusively from cows that had previously received a vaccination against tuberculosis. According to him, this measure had preceded the recommendations of the International Congress on Tuberculosis in 1905 by nine years ("Discurso del Dr. Enrique Figari, Director de Salubridad" 1908, 80). Obviously, public health achievements could best be emphasised by pointing to a successful competition with Europe, and to enact a measure much earlier than any European counterpart was what could be termed the highest "currency" of progress.

Publications which had been specifcally prepared for distribution among the participants also served to showcase the country's progress. Te high-end publication with over 500 pages of text, photographs, and fgures about the accomplishments of the *Comisión Nacional de Caridad y Benefciencia Pública* up to 1905 (Comisión Nacional de Caridad y Benefciencia Pública 1907) pursued an additional and more specifc goal: although the book had been compiled by the commission's former president, his successor in ofce, José Scosería, seized the opportunity to add a preface. In it, he thanked his predecessor for the work and progress of all the institutions under the commission's umbrella but emphasised that important changes had started under his, Scosería's, directorship. Tese changes refected the "new orientation of the philosophic ideas of the majority of the commission", and more specifcally, the abolition of all religious symbols in the health and charity institutions, and the employment of secular teachers in all institutions for the protection of children (Comisión Nacional de Caridad y Benefciencia Pública 1907, ix–x). By the time the fourth congress took place in 1909 in Rio de Janeiro, the project to secularise public health in Uruguay had taken further important steps.

In 1908, a commission that drafted the law for the new *Asistencia Pública Nacional* had started its work under Scosería's direction, and parliamentary discussion of this proposal was about to begin. According to the report that Scosería together with fve other ofcial delegates to the congress in Rio sent to the Interior Minister of Uruguay, their colleagues from the neighbouring countries had praised this legislative project in the making. Tey reported that Scosería had presented a paper on "Te Intervention of the State on Matters of Public Assistance" and that whenever matters of public health and assistance were discussed, the congress attendants had spoken very highly of Uruguay, acknowledging that it was becoming "maybe the frst among the American countries" to "incorporate the recognition of the right to assistance to its positive legislation" (Scosería et al. 1909, 456). Here the mechanism of competitive cooperation becomes very visible as the epistemic community referred to an alleged advantage over the neighbouring countries with whom it nevertheless was cooperating to infuence policy reforms. In fact, the whole report recollected the praise Uruguay had received on the occasion, summing up its high rank in comparison with other countries: apparently, this was the best approach to present the success of the delegation and their national institutions to the minister. Te report also includes explicit references to the national discussions of the drafted law which underscores the importance of the competitive cooperation mechanism: transnational success was cited to favourably infuence political and public opinion within Uruguay (Scosería et al. 1909). Similar to these fndings, Anne-Emanuelle Birn has argued in her research on Uruguayan child health policies that the "international interchanges" of Uruguayan physician Joaquín de Salterain were "most efective (…) to leverage increased attention and resources at home" (Birn 2006, 41).

In the years to come, Scosería continued to participate in the Latin American Medical Congresses and used his presentations, for instance, to document his country's progress in the prevention of tuberculosis among children, illustrating the institution's success with photographs that formed part of the hygiene exhibit in Rio. Furthermore, he pointed out that this was one of only three existing programmes in Latin America, and at the end, the self-praise tied in with a resolution calling all Leagues against Tuberculosis in Latin America to initiate preventive programmes for children focused on "living in fresh air" and the hope that these private leagues would encourage state institutions to follow (Scosería 1909). Te resolution was one of many passed at the congress, and it points to how an approach focused on cooperation, in this case the preventive measures against tuberculosis, fostered national policies and was intertwined with the competitive character of the congresses.

When the Latin American Medical Congress met for the ffth time in 1913, the Uruguayan *Asistencia Pública Nacional* had been founded, and President Batlle y Ordoñez was in the middle of his second term. Tus, the congress in Lima presented the unique opportunity to showcase the success of this new public health institution, which led to the publication of yet another comprehensive book (República Oriental del Uruguay 1913). It is noteworthy that although Uruguayan public health experts participated in international congresses all over the world during the 1910s and 1920s, the two major publications on Uruguayan public health were produced for Latin American Medical Congresses. Tis points to their particularly high status as venues for exchange, cooperation, and competition both with Europe and within the continent.

### **5 Conclusion**

In the time frame under study here, the new nation states welcomed scientifc cooperation and intensifed processes of exchange and support within Latin America. Here, public health claimed special interest. While Europe remained the point of reference and centre of medical modernity, it clearly now served as a model to be surpassed. Te transnational events of the Latin American Medical Congresses ofered ample opportunity for cooperative competition and legitimisation of the epistemic community and their national reform agendas.

Te Chilean case underscores that while doctors could claim that medicine was indispensable for societal progress and build impressive careers on this assertion, this claim did not sufce to promote the epistemic community's preferred public health policies when their introduction to national policy was discussed. Even though Chilean doctors mustered the support of their hemispheric peers at the periodically reoccurring event, the national context and discussions usually neutralised the transnational impulses in the feld of public health, which in most of the cases would have an impact only much later. Our analysis of Chilean discussions at the frst Latin American Medical Congress in 1901 also points to the Chilean state's reluctance to intervene in the feld of public health and assume full responsibility for the well-being of the citizens on the one hand. On the other hand, this also points to the still limited power on the side of the doctors; the congresses which followed until 1909 reinforce this result as the Chilean governments continued to be sceptical of the doctor's proposals for public health reform. Te Chilean case thus shows that the mechanisms of competitive cooperation and legitimisation certainly worked within the framework of the congresses but were of limited reach. Although transnational venues certainly were important to shape and exchange concepts and ideas, it was the national framework which decided on the translation of these impulses into public health policies. None of the mechanisms studied here ultimately fostered the transmission of these ideas into the realm of social policy decision-making. Tis would also explain why the corresponding national agencies were not established although the epistemic community demanded them. But, when the Chilean political framework changed, the transnational impulses and the legitimacy gained by the Chilean medical community through transnational competitive cooperation eventually led to policy implementation. Tus, for instance, the Chilean sanitary code of 1918 preceded the Pan-American Sanitary Code of 1924 (Cueto 2007, 63–68). By 1924, that is, two years after the last Latin American Medical Congress, Chile counted with a Ministry of Hygiene and Social Welfare and a Mandatory Workers Insurance Fund which enabled the state to cover illness, disability, old age, and death (Rengifo 2017).

However, the importance of the national framework for the success of an epistemic community and the value of the competitive cooperation mechanism is reinforced by the Uruguayan case study. Here, the receptive national framework served to amplify the eforts of the community to learn and to cooperate, and to compete with cutting-edge policies at transnational venues. National public health policies refected transnational impulses, while they also served to improve the nation's international standing and increase its visibility as a leader in public health and in social policy development.

Te analysis of the Uruguayan case study further points out that cooperation and competition at transnational events served to strengthen the participants' position and added to their national standing. Te restructuring of the major Uruguayan public health institution was led by the *batllista* physician José Scosería from 1903 to 1910, the years when he and his colleagues used the Latin American Medical Congress to gain international support for the national reform process. Tis again stresses the importance of dynamics that lie beyond the nation for national developments, which is why *batllista* physicians were able to use the support of the epistemic community to strengthen their position in national debates. Also, these physicians did not miss the chance to report the favourable reception of Uruguay's progress. Here, the competitive cooperation and legitimisation mechanisms worked for the epistemic community and those medical doctors who drove national reforms. Tey successfully linked the medical congresses to national public health debates.

Still, contrary to Chilean developments, Uruguay would not see the creation of its Ministry of Health until 1933 and of statutory health insurance until four decades later. Interestingly, it was José Scosería of all people who lamented this fact in the 1920s. His career within Uruguay's public health bureaucracy had continued after his directorship of the *Asistencia Pública Nacional*, a task he assumed from 1910 to 1917. More than ten years later, he went on to direct the *Consejo Nacional de Higiene*, a post he held from 1928 to 1931. Tis state institution was responsible for developing, implementing, and monitoring sanitary and public health jurisdiction in Uruguay and would be merged with the *Asistencia Pública Nacional* into the Ministry of Health in 1933, a process that Scosería claimed to have initiated ("Discurso del Doctor José Scosería" 1936, 49). In his position as director of the *Consejo Nacional de Higiene*, he also entered an important international realm of exchange: in 1928, he took on the task to study Uruguay's social security landscape in order to collaborate with a joint commission on health insurance and public health by the League of Nations' Health Organization (LNHO) and the International Labour Organization (ILO) (Borowy 2009, 222, 371). We can assume that the mechanism of competitive cooperation was equally important in this context. Scosería prepared a lengthy report on social security in Uruguay, taking a critical stance on the low coverage, fragmented character, and dispersed institutionalisation of social security systems in his home country. He especially lamented the absence of health insurance and tried to push for its introduction. His ideas were well received in Geneva, but this time this reaction did little to promote his agenda at home: public health insurance would only be introduced in Uruguay in 1975 (Soiza Larrosa 2010).

#### **References**


Congress. I Organization and General Results of the Congress]. 1901. Montevideo: Tip. y Enc. "Al Libro Inglés".


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

# **12**

## **The Politics of Universal Health Coverage: Mechanisms in the Process of Healthcare Reform in Bolivia**

**Sebastian Sirén**

## **1 Introduction**

Te quest for Universal Health Coverage (UHC) has been an enduring theme in discourses about global social development during the last decades (UN 2019; WHO 2010), and in recent years, many countries have adopted this goal as a national aspiration (Reich et al. 2016). Globally, important progress has been made in this respect, but healthcare reforms have on many occasions been inadequate, or sometimes even increased the fragmentation of health systems, with persistent inequities in health as a consequence (Vega 2013).

Bolivia is an example of a country that has long sought to recast a fragmented and exclusionary health system in order to universalise access and overcome stark inequities. Te introduction of a universal public healthcare system, the *Sistema Único de Salud* (SUS), in 2017, thus represented

S. Sirén (\*)

Swedish Institute for Social Research, Stockholm, Sweden e-mail: sebastian.siren@sof.su.se

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_12

a milestone in the struggle for UHC in Bolivia, and at least a partial departure from the historically prevailing institutional logic of Bismarckian social insurance. But the reform has been long in the making and the path towards its realisation has been winding and riddled with complications. Te challenges partly echo those of other countries, rich and poor, that have adopted UHC as an aspiration for national policy (Roberts et al. 2003; Reich et al. 2016) or that have otherwise sought to overcome a "truncated" social insurance system (Holland 2018).

Drawing on the revived interest in causal mechanisms to which this book is a testament, this chapter employs a mechanism-based approach to explain the winding processes of healthcare reform in Bolivia. Te purpose of the study is, beyond proposing building blocks for a modular explanation of this specifc case, also to contribute to the identifcation of mechanisms of broader relevance for the comparative literature on welfare state politics in the Global South (Dorlach 2021; Lavers and Hickey 2016; Sirén 2021). Te empirical analysis elaborates on and updates previous work (Sirén 2011), drawing on a combination of sources, including semi-structured interviews, newspaper articles, ofcial documents and research reports. Teoretically, the chapter draws on comparative welfare state research (see, e.g. Gough 2008), while recognising the need for an approach to policy analysis that is "sensitive to the sometimes complex historical interplay of forces" (Hacker 1998, 58; cf. Falleti and Lynch 2009).

Six diferent mechanisms emerge from the exploration of health system reform in Bolivia. Firstly, policy emulation by *expert theorisation*, whereby transnational epistemic communities infuence the social construction of which policies are considered appropriate, was decisive for the emergence of an orientation of health policy that focuses on health as a citizenship right and a duty of the state. Secondly, *class-based mobilisation*, coloured by the protagonism of the indigenous peoples' movements in the political coalition that came to dominate Bolivian politics from 2005 onwards, was decisive for turning these ideas into government policy. Tirdly, *social movement–state interaction*, involving relatively autonomous public ofcials engaging in dialogue with the social movements making up the support coalition of the leftist government, shaped the orientation of the reform initiatives and gave the process sustained force.

#### **12 The Politics of Universal Health Coverage: Mechanisms…**

A fourth mechanism, the *alarmed middle classes*, makes reference to the role of formal sector workers covered by the existing social health insurance funds and organised through the main trade union confederation, as a pivotal interest group whose contingent consent proved to be crucial for the advancement of a more transformative reform. A similar logic, whereby previous policies generate groups with interests in the preservation of existing institutional arrangements, fostered resistance also among employees of the health insurance funds, through a ffth mechanism labelled *provider resistance*. Lastly, healthcare systems rely on a powerful profession keen to act in defence of its autonomy. Indeed, the medical profession reacted to proposals for increased government regulation of their profession as such measures were perceived as a threat to their privileges and autonomy, giving rise to a specifc mechanism here referred to as *professional autonomy*.

Te analytical approach of the study is further elaborated in the next section, introducing how theories from comparative-historical research on welfare state development have been used to identify potentially relevant modular causal mechanisms. Te following sections then delineate the process of healthcare reform in Bolivia and highlight sequences of events of particular relevance for the progress towards UHC. Te chapter concludes with a summary of mechanisms making up the building blocks of the provided explanation, along with a few preliminary notes on how to further improve our understanding of the workings of the identifed mechanisms.

### **2 Explaining Healthcare Reform: Causes and Mechanisms**

Te right to health can be seen as one dimension of a broader concept of social citizenship (Marshall 1950), and healthcare systems are fruitfully understood as embedded within the larger organisation of the welfare state (Quadagno 2010). Te idea of social citizenship implies universalistic claims and is realised through the extension of social rights to every member of society, whereby the state is committed to its responsibility for the welfare of its citizens. Within social policy research, universalism is moreover often understood as a specifc characteristic of welfare state institutions that cater to the needs of all citizens, including the poor and more afuent workers within the same programmes (cf. Leisering 2019). In the same vein, universalism is a key concept in the literature on healthcare reform, describing an aspiration of health policy as well as a characteristic of health systems (see, e.g. Quadagno 2010).

Comparative analyses of healthcare systems typically focus on how countries compare with regard to the provision, regulation and fnancing of health services. Refecting systematic diferences between countries on these dimensions, two basic models of national health programmes stand out (Wendt et al. 2009; Navarro 1989). Within a *National Health Service* (NHS) or Beveridge system, the state is the main provider of healthcare through tax-fnanced services, while in *insurance-based* systems, the role of the state is more limited and restricted to regulating the function of insurance agencies whose services are funded by fees. Universality and social solidarity are core principles within NHS systems, but also systems based on compulsory social insurance have fostered virtually universal coverage in advanced capitalist economies. Tese systems thus represent parallel historical routes towards the expansion of health coverage, but with diferent causes and distributional outcomes (Navarro 1989; Quadagno 2010).

Previous research on the causes of institutional change in health policy refects theoretical developments in the broader feld of comparative welfare state research (this is clearly refected in, e.g. Hacker 1998; Quadagno 2010; Béland 2010). A central perspective within this literature, power resource theory (Korpi 1983), has established the micro-foundations of a class-based mobilisation mechanism to explain welfare state development. Tis approach assumes purposive and reasoning actors, intentionally seeking to modify the market distribution of welfare. In particular for the working class, the efcacy of such action is supposedly enhanced by collective organisation, leading to mobilisation through trade unions and social democratic parties. Democracy and the extension of political citizenship moreover hold the promise of a more level playing feld between social actors with asymmetric power in the marketplace, including on the labour market. Tis makes using the political arena to expand and enrich social citizenship a central strategy of social democratic and other leftwing parties with their main constituencies in the working class. Te extension of encompassing social policies is accordingly a core goal of working-class mobilisation, and the emergence of a comprehensive government health programme would accordingly be directly related to "the strength of the working class and its political and economic instruments" (Navarro 1989, 897). Beyond the strife to shift distributional struggles from the market towards the realm of democratic politics, trade unions and social democrats have also sought ways to resist or transform institutions fomenting the segmentation of social rights, including corporatist social insurance (Korpi 2001). Similar frameworks, emphasising the role of political parties on the Left, are now also being employed in explanations of the expansion of welfare policy in the Global South (Sirén 2021), not least with regard to Latin America (Huber and Stephens 2012).

Institutionally oriented scholars have in turn argued for an appreciation of the role of historical sequence, timing and contingency (Hacker 1998), emphasising the importance of policy feedback and institutional veto points (Telen 1999; Pierson 1996). Tese insights are arguably important in order to understand the emergence of other organised interests or "power groups" (Navarro 1989) engaging in struggles over the organisation of health systems. Institutionalists have highlighted that welfare state expansion in itself produces new interest groups composed of benefciaries and providers, who are swiftly mobilised in face of perceived threats to the policies from which they beneft (Pierson 1996). Te character of existing social insurance institutions is thus likely to infuence the mobilisation of current benefciaries and providers of healthcare services.

Te particular resilience of state corporatist social insurance institutions has moreover been attributed to the involvement of trade unions in the management of these programmes, making them resemble "wellorganized regiments that can be mobilized with short notice" (Korpi and Palme 2003, 442). When such counter-mobilisation happens in reaction to government attempts to universalise access to contributory social insurance, we could speak of an alarmed middle classes mechanism, as it is generally not the most marginalised groups that beneft from social insurance systems in the Global South, but a less deprived and better organised "middle class" of formal sector workers (cf. Rudra and Tobin 2017). When resistance comes from those providing the services, seeking to maintain their position, we can instead speak of a provider resistance mechanism.

Te role of the middle classes can also be understood in light of scholarship focusing on social capital, and generalised trust in particular. Research has underlined the importance of trust for the possibility of acquiring contingent consent from the middle classes to universal social policy, while also stressing the institutional origins of such trust (Rothstein 1998; Levi 1993). In such a view, well-functioning and impartial government institutions increase generalised trust, making it more likely that ambivalent middle classes will consent to universal public policies. Social insurance systems of the Bismarckian kind, on the other hand, are prone to foster a "compartmentalised" form of trust reserved for one's own group, while preventing the development of generalised trust (Rothstein and Stolle 2003). Given the legacies of widespread corruption and ineffective institutions, in combination with the legacies of social insurance, it is accordingly likely that universalistic reforms are perceived as threats by groups that are privileged by the segmented social insurance system. Te quality of government institutions thus emerges as a contextual factor with potential relevance for shaping the alarmed middle classes mechanism.

Teories about the impact of states on social policies have moreover highlighted that the relatively autonomous "collectivities of administrative ofcials can have pervasive efects on the content and development of major government policies" (Skocpol 1985, 12). Tis perspective encourages scholars to appreciate state bureaucracies as independent actors and to assess how state capacities to formulate and implement policies impact on processes of institutional change. However, state ofcials may not only have interests stemming from their position within the administration, but may additionally have ideological motivations. Tulia Falleti argued that subversive actors within the Brazilian healthcare system were able to push for more universal policies following the "infltration of the state by reformist elements of society" (Falleti 2009, 40). Analogous arguments can be found in the literature on state feminism (Stetson and Mazur 1995), highlighting how political activists, through their engagement with existing state institutions, create new opportunity structures through which they can pursue their aims (Chappell 2000). Along these lines we might thus additionally think of a mechanism of social movement–state interaction comprising, in a frst step, activists with a reformist agenda occupying positions in the state bureaucracy, and subsequently relatively autonomous activism on the part of these public ofcials, interacting with social movements in shaping public policy rather than merely responding to societal pressures.

In addition to the abovementioned perspectives, there is an emerging literature emphasising the role of ideas, and particularly their transnational difusion, for developments in social protection policy, not least with reference to the Global South (Béland 2010; Weyland 2006; Lavers and Hickey 2016). Research on social policies has commonly highlighted the role of International Organisations (IOs) for the spread of particular policies across countries (Leisering 2019), and recent scholarship has also emphasised the importance of South–South exchange in this regard (Stone et al. 2020). While diferent mechanisms have been implied by this literature, including coercion, competition and learning, sociological research has mostly been concerned with the mechanism of emulation, referring to the social construction of appropriate policies (cf. Leisering 2019). Weyland (2006), for example, found emulation to be of relevance for the establishment of the universal right to health for all as an overarching norm among policymakers across Latin America. Unlike the other commonly specifed mechanisms of policy difusion, emulation does not assume rational assessments of policy consequences, focusing instead on the adoption of appropriate policies conforming to prevailing norms (Gilardi and Wasserfallen 2019). A mechanism that can generate such difusion processes is through expert theorisation within epistemic communities (Haas 1992), but emulation may also result from countries following the lead of a perceived role model or by policymakers imitating policies pursued in countries perceived as close or similar (Dobbin et al. 2007; Weyland 2006).

Te role of the medical profession has moreover been a salient theme in health systems research (Immergut 1992; Wendt et al. 2009). As noted by Jacob Hacker, in no Western country have physicians wholeheartedly welcomed the extension of government control over the healthcare sector (Hacker 1998, 66). Tis refects the fact that the introduction of national health insurance inescapably involves an inherent confict of interest between governments and doctors as the buyers and sellers of medical services, respectively (Immergut 1992, 58). Still, the medical profession has not everywhere been equally resistant to government intervention, arguably refecting the diferential structure of healthcare markets and government institutions, with consequences for the preferences, orientations and relative strengths of professional organisations (Hacker 1998). One would accordingly expect a mechanism of professional autonomy, shaped by the historical legacies of the existing healthcare system, to unfold in the face of government attempts to increase the scope of its control and regulatory capacities.

In the next section, the development of the Bolivian healthcare system is explored in light of these theoretical insights. Te narrative starts with an introduction of the historical context of healthcare reform, before reviewing the various attempts at realising the promise of healthcare for all in Bolivia, underscoring the identifable causal mechanisms as they impinge on this process.

## **3 The Progress Towards Universal Health Coverage in Bolivia**

#### **3.1 The Emergence of Social Health Insurance**

Te origins of social health insurance in Bolivia can arguably be described as a confuence of the mechanisms of class-based mobilisation with a form of policy emulation driven by the adoption of policy scripts advocated by IOs (primarily the International Labour Organization, ILO) and the inclination of the Bolivian government to follow the examples of countries perceived as pioneers in social protection.

Te national revolution in 1952 comprises a major branching point in the history of Bolivia, with fundamental impacts on state–society relations. Te trade union movement, organised through the encompassing trade union confederation, the *Central Obrera Boliviana* (COB), and centred around the mining workers' union, was instrumental for the revolution and gained considerable infuence over the frst post-revolutionary governments of the *Movimiento Nacionalista Revolucionario* (MNR). Te other leg of the labour movement, the mainly indigenous peasants' movement, organised primarily through the *Federation Sindical Unica de Trabajadores Campesinos de Bolivia* (CSUTCB), also allied itself with the revolution and managed to push through extensions of sufrage and agrarian reform. In the post-revolutionary period, class was made the dominant form of expression of social identity in state–society relations, subsuming the indigenousness of the peasant movement to worker-based leftist ideologies (Postero 2010; Regalsky 2010).

It was in this context that social health insurance was introduced in Bolivia, with the adoption of the Social Security Law of 1956. Te legislation built on the law on Obligatory Social Insurance from 1951, drafted with support from the Spanish government and infuenced by the Iberian-American Social Security Organization (OISS). Te 1956 legislation was in turn a result of a collaboration of representatives from the International Labour Organization (ILO), national specialists and a committee of deputies (Bocangel Peñaranda 2004). It included sickness, maternity, old-age, work injury and family benefts as well as social housing and was to be funded by tripartite contributions from employers, employees and the government. Tis system was based on a Bismarckian social insurance logic with the intention that, as industrialisation progressed and formal employment grew, universal coverage would eventually be achieved.

Social insurance did grow to become the most important subsystem in the Bolivian national health system, but the hopes for universal coverage failed to materialise as the Bolivian economy is still today characterised by very high levels of informality, even compared to other Latin American economies (Medina and Schneider 2018). Starting at a coverage rate of around thirteen percent of the population at the time of the introduction of social security in 1956, the rate gradually increased to a peak at around twenty-seven percent in the early 1980s. Tis rate then declined slightly during the 1980s, partly because of layofs resulting from the structural adjustment policies implemented at the time. Coverage then started to grow slowly again over the course of the following decades, standing at around thirty percent in the mid-2000s (see Fig. 12.1). Accordingly, despite the progress made during previous decades, a majority of Bolivians has lacked efective health coverage, with particularly high rates of exclusion among the poor, rural and indigenous population (Chacón and Valverde 2009).

Moreover, fragmentation in the Bolivian social health insurance system has increased over time, as the number of social health insurance funds (*Cajas de salud*) has grown over the years, from three originally to eight today. Te system, indebted to the alliance between the frst postrevolutionary government and the COB, moreover instituted the unions as important veto players by granting them infuence over the management of the healthcare funds. However, in combination with political interference, this system has made the *cajas* major sites of patronage, nepotism and corruption (Bocangel Peñaranda 2004, 69), arguably setting the scene for the unfolding of the alarmed middle classes mechanism in response to the reforms that were eventually attempted.

#### **3.2 Democratisation, Decentralisation and Neoliberalism**

Democratisation after 1982 was accompanied by severe economic crises and hyperinfation. With support from the IMF, the government soon adopted a neoliberal structural adjustment programme, under the banner of the New Economic Policy in 1985. While seemingly curbing hyperinfation and stabilising the economy, reductions in public employment, fexibilisation of labour markets and austere budget policies also led to harsh social costs and came to weaken the class dimension in Bolivian politics, especially the infuence of the COB.1 Instead, the re-installation of democracy marked the beginning of the gradually increasing importance of other social movements, most prominently the peasant workers' unions, indigenous movements and women's organisations. Te 1990s

<sup>1</sup>Te privatisation of the mining industry, and the subsequent "relocation" of workers towards the peri-urban areas around El Alto in the highlands and the coca growing region of Chapare, proved to unintendedly foment the emergence of new social movements in the following decade (Postero 2010, 21–22).

saw a growing salience of indigenous groups in the political landscape, primarily the *Confederation Indigena del Oriente Boliviano* (CIDOB) as well as the emergence of "identity politics" and multiculturalism as salient political themes (Postero 2010; Regalsky 2010).

While the social health insurance system was left largely untouched by the government during the decades following the return of democracy, a public system was gradually rolled out alongside. Te frst government after the return of democracy, headed by Hernán Siles Suazo of the MNR, came to introduce the concept of health as a human right and launched several public programmes, including health brigades in rural areas, popular health councils, vaccination campaigns and free medical care in relation to childbirth. Te orientation of public health policy at this time was arguably shaped by the international infuences regarding what was considered appropriate health policy, thus fostering policy emulation. In particular, crucial inspiration was provided by the International Conference on Primary Health Care held in Alma-Ata (WHO 1978). Te efects of this stronger emphasis on public health can be seen in a stronger focus on issues related to maternal and infant health and communicable diseases, visible in the statistics on the share of births attended by skilled health staf and immunisation rates depicted in Fig. 12.1.

Although economic crisis, hyperinfation and the subsequent cuts to government expenditure hampered the expansion of a more comprehensive public healthcare system at the time, these initiatives marked the entry of the principles of primary care in Bolivian healthcare politics and would come to infuence the future development of public programmes. Not least, by introducing and deepening public programmes in rural areas, a space was created where alliances between medical professionals and social and community organisations could be cultivated, an interaction which would prove to be crucial for later developments.

#### **3.3 Transnational Infuences and New Repertoires of the Social Movements**

Refecting a discursive change among international fnancial institutions towards a recognition of the instrumental value of social infrastructure for the promotion of sustained economic growth and a fertile investment climate, the 1990s were characterised by structural adjustment policies simultaneously promoting ideas about indigenous citizenship, cultural recognition, decentralisation and popular participation, on the one hand, and privatisation and "capitalisation", on the other (Kohl 2002). Te failure of these political reforms to actually revamp exclusionary social structures and ameliorate poverty frustrated many of the indigenous and less afuent segments in Bolivian society, thereby promoting the search for new strategies for political intervention among these groups (Postero 2010; Haarstad and Andersson 2009).

Te policies of decentralisation and popular participation pursued during the 1990s had several unforeseen consequences. One such consequence was increasing opportunities for interaction between medical doctors and social movements in indigenous and rural communities, following the municipalisation of health policy. A strong current within the medical profession was already at that time heavily infuenced by the Latin American social medicine tradition, which had emerged in the 1970s, promoting the importance of economic, political and social determinants for understanding the reproduction of health inequalities and advocating for healthcare as a social right and responsibility of the state (Tajer 2003). Organisationally, this current was articulated by the Latin American Association of Social Medicine (ALAMES). Te intermingling of this epistemic community with the emerging social movements at the local level seems to have caused an amalgamation of the social medicine agenda with the demands for social justice coming out of the rural and indigenous communities (cf. Falleti 2009).

A two-stage process of policy difusion is accordingly distinguishable. Firstly, there was an instance of difusion through a regional network with the function of a transnational epistemic community. Secondly, in connection to their work in marginalised communities, these ideologically motivated health professionals came to associate themselves with the emerging social movements, thus transferring ideas and knowledge from the transnational context to the site of local struggles for redistribution and recognition. Trough this process, these ideas were intertwined with the demands voiced by the emerging social movements, who accordingly came to adopt the understandings and policy prescriptions advocated by the social medicine tradition as their own. Arguably, expert theorisation, with the support of regional epistemic communities related to health policy, was thus an important mechanism for shaping ideas about what constituted appropriate health policy among the agents that would later assume responsibility for the direction of state policy.

Another consequence of the particular blend of pro-market structural adjustments policy and a decentralisation process with multicultural underpinnings was a growing frustration with the efects of neoliberal economic policy, especially in indigenous communities. Tis frustration was increasingly channelled towards ofcial institutions at the local level, in particular following the municipalisation of politics implied by the Law on Popular Participation from 1994. In this context, a coalition of peasants' unions, spearheaded by the CSUTCB, formed what they called a Political Instrument for the Sovereignty of the People in the mid-1990s. Te "instrument" eventually took over the name and legal identity of a pre-existing political party, the *Movimiento al Socialismo* (MAS) (Postero 2010). Te parallel processes of precarisation and democratisation thus seem to have set in motion a chain of events refecting the mechanism of class-based mobilisation. Tis sequence displays clear analogies with the circumstances and strategies employed by labour movements in Western Europe around the turn of the last century, when social democratic parties were founded as the political instrument of labour unions as the political arena became the focal point of struggles for political recognition and social citizenship (cf. Korpi 1983).

#### **3.4 Mobilising for the Right to Health**

As the neoliberal agenda of privatisation and capitalisation was continued over the course of the 1990s and early 2000s, reoccurring massive protests erupted, particularly against the government's plans to privatise water and natural gas. Te "water war" in the year 2000 marked the beginning of a cycle of popular protests that resulted in the demise of two presidents and the virtual collapse of the traditional party system in Bolivia (Faguet 2019). During this time, MAS emerged as the main opposition party with a variety of associations, including the unions of teachers, street vendors and factory workers, as well as urban neighbourhood associations and left-leaning intellectuals allying with the party, thus broadening the party's appeal also among the *mestizo* urban middle class (Madrid 2011). In the elections following the resignation of Carlos Mesa in 2005, MAS's candidate Evo Morales, the former leader of the coca growers' union, gained the presidency with ffty-four percent of the votes, and MAS managed to win a majority in the chamber of deputies, but fell short of a majority in the senate.

Te electoral programme for the run-up to the 2005 elections is a testament to the infuence of the social medicine tradition on the party's health policy, stating that the health system should be based on a social conception of health and characterised by "the identifcation and transformation of the determinants of health-illness and not only addressing its efects" (MAS-IPSP 2005, 113, author's translation). With MAS in government, positions within the Ministry of Health and Sports (MSD) were being occupied by ofcials with close ties to ALAMES, as well as with the social movements and various left-wing parties. Not least was the strong emphasis on social medicine represented by the frst minister of health appointed by Morales, Nila Heredia.2 Tis not only corresponds to the frst stage in the social movement–state interaction mechanism defned earlier, but also builds on the difusion of ideas associated with the Latin American social medicine tradition, thus pointing to the importance of recognising the political circumstances causing expert theorisation to foster actual policy change.

Following demands from the social movements, the formation of a constituent assembly was the most central promise of Morales' candidature in 2005 and became the main political project following his installation as president. Te MSD was actively engaged in the discussion about the inclusion of the right to health within the working groups set up to draft the new constitutional text. Te ministry organised discussions on healthcare policy through departmental assemblies as well as a national pre-constituent assembly on health, with representatives from civil society, the social movements and the medical profession. Central topics

<sup>2</sup>Heredia, Bolivian physician with a master's degree in public health and who participated in the leftist resistance movement under dictator Hugo Banzer in the 1970s, served as Minister of Health in 2006–07, and again in 2010–11, and as coordinator of ALAMES 2009–14.

arising from these deliberations were the recognition of health as a fundamental right, and the responsibility of the state to guarantee access to healthcare for all, but also the incorporation of communitarian and indigenous notions of healthcare (Johnson 2010). Tese themes eventually found their way into the constitutional text adopted in 2009, which includes formulations about the state's obligation to guarantee the universality of the right to healthcare, as well as the necessity of having a unitary health system to which access should be free of charge (Constitución Política del Estado 2009). Tis sequence of events thus represents a telling example of the social movement–state interaction mechanism, with public ofcials actively shaping the political agenda through iterative communication with social movements, rather than merely responding to calls from societal actors.

Te initial institutional strategy suggested by MAS in order to reach the goal of universal health coverage was to extend access to the existing public insurance programmes, primarily the public scheme for mothers and infants as well as the old-age health insurance, to the entire population not insured by the social health insurance funds (MAS-IPSP 2005). Te early orientation of health policy in the new administration is described in the National Development Plan from 2006, which specifcally contemplates universal access to healthcare through the implementation of a new intercultural, community-based family health system. Tis was also refected in the frst legislative proposal of the Morales government in this respect, which envisioned the introduction of a new universal health insurance programme, *SuSalud*, that would incorporate all persons up to the age of twenty-one within a public insurance system. Te government envisioned this as a frst step towards reaching UHC through a gradual extension of these rights to the entire population, pending availability of sufcient resources to fund such an expansion (Bill 005/2007).

Funding was obviously a fundamental obstacle for a truly transformative reform. Tis issue was at the time intimately connected with the question of "nationalising" the country's hydrocarbons sector. Bolivia has one of the largest deposits of natural gas in the region, and stronger state control as well as redistribution of revenues from the sector had been central demands from the social movements in the run-up to the 2005 elections. Te Morales government soon took action to strengthen the regulatory role of the state in this sector, to renegotiating concessions with the foreign companies, and introduced new fees and taxes on hydrocarbon extraction. A cornerstone of this policy was the introduction of the *Impuesto Directo a los Hidrocarburos* (IDH), which was channelled directly to the departments, municipalities, as well as universities and indigenous communities (Kaup 2010, 131). Eventually, the escalating regional and ethnic tensions during the negotiations over the new constitution resulted in a political deal being struck in which the government conceded far-reaching autonomy to the departments in managing their revenues from natural resources, in order to appease the oppositional forces based in the eastern lowlands (Postero 2010).

Te newly introduced IDH was envisioned as the main source of funding for the *SuSalud* reform (Bill 005/2007). However, the centralisation of departmental resources implied by the proposal would mean signifcant redistribution from the wealthier lowlands to the more populous highlands. Consequently, four oppositional eastern departments opposed *SuSalud* (Santa Cruz 2010), arguing in favour of using the funds from the IDH for autonomous public health insurance programmes at the departmental level (see, e.g. Galindo Soza 2010). Accordingly, although the government proposition on *SuSalud* was in a frst instance passed by the Chamber of Deputies, where MAS held the majority of seats, resistance from the opposition however resulted in it never being discussed in the Senate, which at the time was dominated by the opposition.

Tis sequence of events illustrates the mechanism of class-based mobilisation, with resistance to reform coming from traditional and economic elites. Te constitution of MAS as a "political instrument" of the social movements representing the less afuent classes gave sustained presence to these demands in the political arena. Te electoral success of MAS, as well as the sustained mobilisations of the social movements during the frst years of Morales' presidency, arguably helped bringing the issue of universal health coverage to the fore. MAS's electoral success also brought ideologically driven actors into central positions in the Ministry of Health, setting in motion another mechanism, that of social movement– state interaction. Te confguration of the political institutions, especially the devolution of decision-making power to the departments, constitutes a pivotal contextual factor activating regional cleavages and providing oppositional forces allied with the country's traditional elites with veto points where the government's initiatives could be blocked.

#### **3.5 Reactions from Labour and the Medical Profession**

In 2010, after the approval of the new constitution, Morales was reelected as president by a broad margin, and MAS won a majority in both chambers of parliament. Still, recognising the difculties in moving forward with a reform that would require consent from the departments regarding the use of the funds raised through the IDH, the Ministry of Health assumed a new strategy, now proposing an entirely public system, the *Sistema Único de Salud* (SUS), based on the NHS model. Policymakers saw breaking with the Bismarckian social insurance logic of the existing system as a way to overcome fragmentation and segmentation, perceived as particularly problematic in the context of a highly informal Bolivian labour market.

A frst step in this direction was taken in the budgetary law for 2011, which demanded that the health insurance funds' fnancial resources are deposited in accounts supervised by the Ministry of Finance, with the ambition to strengthen the government's ability to monitor the system (Law 062/2010). Tis move by the government caused immediate reactions, which can be explained by the alarmed middle classes mechanism. Benefciaries and employees of the health insurance funds, together with the La Paz branch of the union of medical doctors employed in the social insurance system (*Sindicato Médico y Ramas Afnes*, SIMRA), immediately signed a joint statement with the workers' confederation, the COB, denouncing this move by the government. While claiming to support the idea of universal access, the organisations clearly stated their disapproval of reduced independence of the *cajas*, fearing that the social contributions paid into the funds would be diverted in order to fnance the expansion of the public system (Sirén 2011, 34).

#### **12 The Politics of Universal Health Coverage: Mechanisms…**

Te ministry later organised a series of departmental health congresses where the new proposals for healthcare reform were debated and where delegates were elected to a national health congress held in January 2011. Here, the proposals from the MSD were debated and agreed upon, in the presence of representatives from the health sector and social movements (MSD 2011), again highlighting the relatively autonomous role of state actors in structuring the debate over the reform along the lines of the social movement–state interaction mechanism. And, as a manifestation of their discontent with the government's ambition to intervene in the management and fnancing of the *cajas*, the COB refused to participate in the congress (e.g. Los Tiempos 2011), thus continuously acting in accordance with what the alarmed middle classes mechanism would suggest.

Interviews with representatives from the COB reveal a critical aspect of how the alarmed middle classes mechanism unfolds in this particular case. Te unions' resistance towards authoritarian governments in the 1970s and early 1980s, as well as towards the neoliberal economic policies pursued during the decades following the return of democracy, seemingly constitute a historical legacy provoking an anti-statist posture of the labour confederation, and a reluctance to cede control over social health insurance, won through previous struggles and now perceived as "patrimony of the workers" (Sirén 2011, 34). Moreover, unions accused the government of not complying with its obligations as an employer in funding social health insurance, raising doubts about the commitment of the state to support the new system fnancially. Tese observations underscore the possibly pivotal role of trust for the proclivity of ambivalent middling segments of the population to consent to universal social policy solutions and thus for the alarmed middle classes mechanism.

Te ambition to incorporate the health insurance funds into a fully public system continued to spur resistance, in a situation where the popularity of the government was already seriously compromised due to a recent decision to discontinue a number of fuel subsidies, causing a wave of protests referred to as the *gasolinazo* (Deheza 2012). In March 2011, the medical doctors' association (*Colegio Médico*) launched a strike opposing a new law regulating the medical profession, while also voicing their disapproval of the incorporation of the social insurance funds into the public system. Tis can be explained by both the professional autonomy mechanism and the provider resistance mechanism, pointing to the broader interests of physicians as a profession, as well as the more specifc views of those employed by the social insurance funds. Te confict over the future of the social insurance funds also coincided with a strike for higher salaries for healthcare employees, which further fuelled the ongoing mobilisations against government intervention along the same logic, an efect that can thus also be explained by the provider resistance mechanism. In response, the government aired the idea to use its position as employer to move all public employees currently insured with the *cajas* to the proposed public scheme, thereby debilitating the funds to the point where these would have to merge with the public system out of selfinterest. Simultaneously, several indigenous organisations, allied with the government and supportive of the reform proposals, came out in defence of the plans to implement a unifed NHS-type system (Sirén 2011).

Te chain of events described above also illustrates how the institutional context contributes to structuring conficts over health reform by the way in which defensive strategies are activated. In the conficts over regulation and funding, COB and to some extent also the professional organisations in the health sector emerged as veto players, despite the government's power to take legislative action following MAS's landslide victory in the 2009 elections, which gave the party a majority in both chambers of parliament. Tis observation underscores how the alarmed middle classes mechanism is underpinned by the presence of corporatist social insurance in the existing health system, as it both structured the interests of these actors and provided them with opportunities to act as de facto veto players due to their position within the system.

#### **3.6 Reassembling a Reform Coalition**

While the plans to unify the healthcare system were not realised at the time, the government eventually managed to get less contentious legislation passed that integrated the two existing public insurance schemes, the infant and maternal health programme and the old-age health insurance scheme, into one integrated programme with stable sources of state funding, while also extending the range of services covered (Law 475/2013). With this legislation in place, growth in public provision and fnancing of health services accelerated, refected for example in the number of health consultations conducted by the public sector as well as in the government's share of total health expenditure, and mirrored in a falling rate of out-of-pocket expenditures (see Fig. 12.2).

Against this backdrop of a more extensive public subsystem, 2018 marked the beginning of a new cycle of discussions about rolling out a universal healthcare system. In September, Evo Morales announced that the government would set aside USD 200 million during the next year in order to implement Free Universal Health Insurance. Te president also emphasised that only resources from the National Treasury would be used and that the social insurance funds would be left untouched (MSD 2018a). Tis announcement was followed by an initiative from the MSD to organise a new series of meetings at diferent sites around Bolivia to discuss the new plans to once again implement the *Sistema Único de Salud*, SUS (MSD 2018b).

Tese meetings were swiftly followed up with the passing of a new law that replaced the legislation from 2013 and extended access to a defned package of free healthcare services to the entire population not covered by social insurance. Te law moreover envisions a gradual extension of the number of services to be included. Funding was to come from the budget of the central government, as well as from the municipalities' own resources in the case of primary and secondary care for the beneft of their own residents (Law 1152/2019).

Te government's new orientation seemed to have appeased resistance from the COB, which now came out as willing supporters of the new policies, with the alarmed middle classes mechanism thus apparently no longer playing a key role. Tis was not least made clear by the leader of the COB, Juan Carlos Huarachi, as he spoke at the ceremonial signing of the new law, stating that "Te COB will defend [universal healthcare] because it is for my brother, my brother-in-law, my nephew, even my friend, we should all fght to defend its implementation, no longer will health be commercialised or private" (Opinión 2019, author's translation). Te new system was ofcially inaugurated on 1 March 2019,

b) Health expenditure (% of current health

expenditure), by source

a) External consultations (millions), by subsector

**Fig. 12.2** Healthcare system indicators, 2003–2019. (**a**) External consultations (millions), by subsector. (**b**) Health expenditure (% of current health expenditure), by source. (Sources: (**a**) UDAPE 2020. (**b**) World Bank: World Development Indicators)

giving all Bolivians the right to free medical care, in the form of a package comprising 300 defned treatments. Te legislation was followed up with an agreement between the COB and the government about a legislative initiative supporting the development of the main social insurance fund, the *Caja Nacional de Salud* (CNS), through the construction of a large number of health centres and hospitals, using funds accumulated in the CNS.

Still, medical doctors employed in the CNS continued to protest also against these policies, referring to their exclusion from the negotiations, illustrating the continued relevance of a provider resistance mechanism. In this vein, representatives of SIMRA pointed to the risk that the new health infrastructure would primarily serve non-insured patients and that the plans therefore constituted a threat to the fnancial stability of the CNS (La Razón 2019). Meanwhile, the main coalition of indigenous organisations, *Pacto de Unidad*, came out in defence of the initiatives of the government and the COB, jointly proclaiming themselves "vanguard in the defence of the [SUS]", fercely denouncing the strike called by the medical doctors (Pacto de Unidad 2019, author's translation), accordingly demonstrating class-based mobilisation as a mechanism continuously supporting universalisation.

#### **3.7 Mechanisms in Bolivian Healthcare Reform**

Drawing on insights from comparative social policy research, this exploration of healthcare reform in Bolivia identifes a set of complex causal mechanisms with modular qualities. Firstly, the immersion of healthcare ofcials within regional networks oriented towards social medicine is a salient feature, refecting a mechanism of emulation through expert theorisation. Te infuences from IOs, through coercion as well as emulation, are apparent in earlier decades, but the formulation of proposals for a universal health system was infuenced by the social construction of appropriate health policy. Here IOs were relevant for putting the issue of the universal right to health on the policy agenda. But the idea of the state as a guarantor of this right, and the particular orientation of health policy under the presidency of Evo Morales, seems more indebted to horizontal transmission of ideas through the Latin American Association of Social Medicine (ALAMES), clearly representing an example of an epistemic community with a shared belief in the universal right to health, an understanding of ill-health as refecting broader social inequalities, a shared focus on improving public health and combating exclusions in healthcare, as well as a common policy goal of promoting UHC through state policy (cf. Haas 1992). Te case study shows how the ideas of this community were in a frst stage intertwined with the broader agenda of the social movements and how they later guided ofcial policy when afliates to ALAMES and members of the social movements took up central positions within the Ministry of Health.

Secondly, the progress towards UHC seems closely linked to a mechanism of class-based mobilisation. In particular, the mobilisation and organisation of the indigenous majority, through peasants' unions and neighbourhood organisations engaging in demonstrations, strikes and popular assemblies, emerges as a central driver of the reform process. Te movement-based organisation of the *Movimiento al Socialismo*, upheld by the incorporation of workers', women's and indigenous peoples' organisations in its social base, is arguably key to explaining the orientation of the party's health policy. Tis mechanism, which enacts a chain of events resembling the experience of early welfare state development in Western Europe, seems to be indebted to the parallel processes of democratisation and precarisation of labour unfolding over the course of the 1980s and 1990s.

Tirdly, pervasive efects of a social movement–state interaction mechanism on the content and development of healthcare reform, following a pattern resembling the "infltration of the state" observed by Falleti (2009) with regard to the developments in Brazil, are salient. In the Bolivian case, this mechanism moreover corresponds to a broader shift in the relationship between state and society tied to the overarching quest to "re-found" the nation (Artaraz 2012). In this process, public ofcials arguably acted in a relatively autonomous fashion, not merely responding to the pressures from organised interests but actively engaging in formulating policies, arguing for their appropriateness and bargaining for their enactment.

Fourthly, medical doctors were resistant to the stricter regulation that the government sought to impose on the profession in order to strengthen the state's ability to monitor and coordinate the provision of healthcare services. We might think of this as a mechanism of professional autonomy, as it activated defensive strategies broadly within the profession. As a ffth mechanism, physicians employed by the social health insurance funds were particularly resistant to the proposed changes, fearing that universalisation would bring with it a deterioration of their working conditions and threaten their privileged position within the system. Tis illustrates the tendency for welfare state expansion to generate constituencies with strong interests in the preservation of current institutions. Service providers constitute such a category, and we might accordingly conceive of this mechanism as provider resistance.

Lastly, unions representing formal sector workers, who were already covered by social health insurance, emerge as instigators of resistance to the proposals for universalisation, in line with the alarmed middle classes mechanism. While part of the coalition that brought Morales and MAS to power, the central trade union confederation, in which the miners play a particularly central role, feared that the incorporation of these funds into a universal system would result a dilution of the funds' resources, and accordingly acted to halt the government's reform initiatives on several occasions. Conversely, the support of the same organisation at a subsequent stage was arguably pivotal for the introduction of legislation that formally universalised healthcare coverage. On this later occasion, the ability of the government to negotiate consent from the labour movement seems to have been a crucial factor for neutralising the alarmed middle classes mechanism, making possible the passing of the fnal legislation introducing the Unifed Health System.

Tese six mechanisms can jointly serve as building blocks in a modular explanation of the winding path towards a universalisation of health coverage in Bolivia, and more particularly of the winding progress towards a unifed government-funded system. In order to explain the challenges to the reform process and its intermittent advancement, the last three of these mechanisms are of particular relevance, sharing their origin in the feedback efects of institutions on actors' perceived interests and identities and the activation of particular social cleavages.

#### **4 Final Remarks**

Tis study has been motivated by the ambition to unveil the mechanisms underpinning the politics of universal health coverage in Bolivia. Te analysis presented above has highlighted expert theorisation, class-based mobilisation, social movement–state interaction, alarmed middle classes, provider resistance and professional autonomy as mechanisms that can be used as building blocks in a modular explanation of the present case, but which arguably also have the potential to travel to other instances of social insurance reform in the Global South.

While this chapter has identifed these mechanisms as relevant for the political processes surrounding institutional changes in health policy, understanding the triggers of these mechanisms, as well as the contextual factors shaping their outcomes, should be the subject of further inquiry. Te theoretical framework presented at the outset of this chapter gives us some clues regarding some of the potentially relevant factors in this respect. Te precarisation of labour is arguably a central trigger of the class-based mobilisation mechanism, causing the less privileged to join forces and act to fght against perceived injustices. However, for these actions to be directed towards public policy reform, there needs to be at least a hope on the part of these groups that they will be able to exercise sufcient control over the state. Democratisation, or at least the prospect thereof, would thus arguably comprise a necessary contextual condition.

Any mechanism underpinned by a reaction to a perceived threat to current privileges, like the alarmed middle classes mechanism, is moreover more likely to unfold in a context of low trust, both towards one's fellow citizens and towards public institutions. Indeed, the low level of trust in the benevolence and impartiality of the Bolivian government voiced by trade unions is surely a common issue across many attempts at universalising access to social protection throughout the Global South and should motivate further attention to the interplay between social policy, trust and opinions. Moreover, the Bolivian experience illustrates how political institutions, and decentralisation in particular, contributed to inspire collective action aimed at changing public policy, but also had a conservative efect by giving oppositional forces associated with traditional elites the possibility to veto subsequent reform initiatives. Meanwhile, the social insurance system provided actors with opportunities to veto proposals when conficts emerged over regulation and funding of the system.

Lastly, the concession made towards the trade unions to not interfere with social insurance was arguably made possible partly as a result of the gradual strengthening of the public subsystem that had taken place since the frst proposal to introduce a unifed health system. Tis strengthening of state capacity unfolded without the clashes and grand gestures of the frst years of the Morales administration. In a sense, the gradual implementation here seems to have been a prerequisite for legislation, rather than the opposite. Tis insight seemingly points to the potential of exploring the interplay between high-level policymaking and less conspicuous institutional evolution, in the spirit of institutional theory highlighting the potential of gradual changes to have transformative outcomes (Mahoney and Telen 2009). At the end of the day, gradualism with regard to the means might prove to be the key to reaching more transformative ends.

#### **References**


Participate in the National Congress Convened by the Ministry of Health]. January 20, 2011.


———. 2018a. Gobierno nacional destinará 200 millones de dólares para el seguro universal de salud gratuito [National Government Destines 200 Millions of Dollars for Free Universal Health Insurance]. September 5, 2018. https://www.minsalud.gob.bo/3446-gobierno-200mm-sus. Accessed May 20, 2021.

———. 2018b. Primer encuentro por la salud y la vida se iniciará el lunes 24 en la ciudad de Oruro [First Meeting for Health and Life Commences on Monday 24 in the City of Oruro]. September 23, 2018. https://www.minsalud.gob.bo/3477-encuentros-salud-oruro. Accessed May 20, 2021.


org/2019/09/02/vanguardia-social-en-defensa-del-sistema-unico-de-saludpacto-de-unidad-defendiendo-el-derecho-a-la-salud/. Accessed May 20, 2021.


Rueschemeyer, and Teda Skocpol, 3–38. Cambridge: Cambridge University Press.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

# **13**

## **Political Responses of Conditional Income Transfer Recipients: A Mechanism Approach**

**Armando Barrientos**

## **1 Introduction**

Te emergence of social assistance, and particularly large-scale conditional income transfers, is the dominant factor in the recent dynamics of welfare institutions in low and middle-income countries (Barrientos 2013; Fiszbein and Schady 2009). A longstanding fnding in the poverty literature is that disadvantaged groups face restrictions in their capacity to infuence political decisions, which reinforces their disadvantage over time. Limited political inclusion is a key dimension of their disadvantage. Tis chapter studies whether participation in conditional income transfer programmes in Latin America generates observable political responses

© Te Author(s) 2022 **403**

A. Barrientos (\*)

Global Development Institute, Te University of Manchester, Manchester, UK e-mail: Armando.Barrientos@manchester.ac.uk

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_13

and whether these responses imply improvements in the political inclusion of participants.1

Studies on welfare institutions in high-income countries suggest that social insurance institutions mobilise participants to support and protect their institutions (Baldwin 1990), although research on attitudes to the welfare state have mixed fndings (Svallfors 2012; Laenen et al. 2020). Studies on social assistance, on the other hand, suggest that programme participation has net negative efects on political participation. Bruch et al. (2012) compare welfare programmes in the USA and fnd that "policy designs can have signifcant efects on civic and political engagement among the poor; the feedback efects of meanstested programs can be positive as well as negative; and such efects tend to be more positive when a policy's authority structure refects democratic rather than paternalist principles" (Bruch et al. 2012, 210).

In Latin America, the growth of large-scale social assistance this century raises the issue of whether it leads to political outcomes likely to result in the political inclusion of participants.2 Within social assistance programmes, conditional income transfers3 have attracted particular attention due to the availability of impact evaluation and administrative data which make it possible to study this issue in a quasi-experimental context. Variation in programme implementation, a staggered territorial implementation for example, helps shape treatment and control groups. Comparisons across these groups support reliable identifcation of programme efects. Researchers have applied this methodology to the study of electoral efects stemming from conditional income transfer programme participation. Studies fnd higher electoral registration, voting,

<sup>1</sup>Tis is also relevant to the efectiveness of anti-poverty policies. In the medium and longer term, efective poverty reduction requires improvements in the productive capacity of disadvantaged groups and improvements in their political inclusion.

<sup>2</sup>Tere is a large body of literature on the socio-economic efects of conditional income transfers, based on the analysis of programme evaluation data. Quasi-experimental methods dominate this literature (Deaton 2020; Dufo and Kremer 2003). Meta-studies show positive efects on recipients' consumption and on schooling and health (Bastagli et al. 2016; Davis et al. 2016; Cecchini and Madariaga 2011).

<sup>3</sup>Conditional income transfer programmes provide a regular subsidy and services to families in poverty, conditional on children attending school, and on regular access to primary healthcare (Barrientos 2019).

and political participation among participants compared to control groups (Nupia 2011; De la O 2013, 2015; Baez et al. 2012; Linos 2013).4 Another group of studies adopts a quasi-experimental approach to attitudinal data in order to identify political efects associated with participation in conditional income transfer programmes (Layton et al. 2017; Layton and Smith 2011, 2015; Zucco 2008, 2013).

In the context of assessing political inclusion, however, this black box approach has important limitations. With few exceptions, existing studies have focused on programme participation outputs such as registration and voting, but less attention has been paid to political outcomes such as political inclusion. Distinguishing between political outputs and political outcomes from programme participation is important. Electoral registration is an output of conditional income transfers when this is required for programme entry, while political inclusion is an outcome when it refects greater political engagement among programme participants.5 Tis chapter adopts a causal mechanism approach to study political responses to transfer receipt. A mechanism approach considers how interventions come to have observed efects and in doing so it pays attention to outputs and outcomes. In the context of this chapter, a mechanism approach seeks to cast light on the causal links existing between transfer receipt and political inclusion outcomes.

Te analysis in this chapter reviews the existing literature on political responses to conditional income transfer programme participation with a view to identify potential causal mechanisms which might shed light on the issue at hand. Te analysis focuses primarily on recipients and on their responses to the receipt of the transfers.6 After identifying three

<sup>4</sup>Manacorda et al. (2011) employ a regression discontinuity approach to study the electoral efects of an integrated anti-poverty programme in Uruguay; Baez et al. (2012) also employ a regression discontinuity approach.

<sup>5</sup>A similar issue arises in the context of the schooling conditions in conditional transfer programmes. Studies have considered whether required school attendance actually improves human capital if classes are overcrowded, teachers are absent or unmotivated, or if programme participants crowd out non-participants from already stretched public schools (Reimers et al. 2006). Expected programme outputs might not be a direct guide to programme outcomes.

<sup>6</sup>Te literature on the politics of social assistance expansion in low and middle-income countries has paid special attention to policymakers, their motivations, and calculations (see de la O (2013, 2015) and the literature reviewed there). Te focus on elites refects a concern with clientelism as a motivation for the adoption and implementation of anti-poverty transfers.

causal mechanisms, they are assessed against attitudinal data from Latin America, from the AmericasBarometer (Latin American Public Opinion Project (LAPOP) 2020) for the period 2010–2019. Most Latin American countries have implemented large-scale conditional income transfers, justifying a regional focus.

Tis chapter is organised as follows: Sect. 2 justifes the application of a causal mechanism approach and considers its main elements; Sect. 3 reviews the available literature on political responses to conditional income transfer receipt in the region; Sect. 4 identifes alternative causal mechanisms explaining political responses; Sect. 5 assesses the presence of these mechanisms using data from LAPOP; a fnal section concludes.

## **2 Why a Causal Mechanism Approach?**

Interest in causal mechanisms as an approach to studying social relationships refects a concern with the limitations of "black box" statistical/ correlation approaches to explanation in the social sciences (Elster 1998; Hedström and Ylikoski 2010). A focus on causal mechanisms is motivated by the need to study the generative processes underpinning the social relationships of interest. Identifying the mechanisms at work helps trace causal links existing between initial conditions, behavioural responses, and particular outcomes.

Adapting Coleman's boat (Coleman 1990) in Fig. 13.1, the dashed line represents the observed correlation, while the efect of the transfer on recipients and their eventual political response is represented by the solid arrows. Te generating mechanism has three main elements (themselves commonly referred to as mechanisms in the literature). Transfer Programme stands for the policy implemented, which signals a change in

**Fig. 13.1** Coleman's boat adapted. (Source: Own presentation)

resource distribution with a subsidy fowing to selected participants (S). Recipients are expected to react to transfers with specifc behavioural change or action (A), in our particular context, increased electoral participation for example. Recipients' behavioural change leads to a transformation of the social situation (T), in this case political inclusion.7 It is helpful to consider the generating process as consisting of inputs, outputs, and outcomes, matching the elements of the mechanism described above. Te causal mechanism is therefore a representation of the generative causal links existing between inputs, outputs, and outcomes. Tis is often missing in statistical/correlation approaches or is only mentioned in passing by reference to simplistic behavioural models.

It should be acknowledged that the literature on causal mechanisms is variegated and a number of distinct approaches to their use are available (Hedström and Ylikoski 2010). Hedström and Ylikoski note that a

mechanism based explanation describes the causal processes selectively. It does not aim at an exhaustive account of all details but seeks to capture the crucial elements of the process by abstracting away the relevant details. Te relevance of entities, their properties, and their interactions is determined by their ability to make a relevant diference to the outcome of interest. (Hedström and Ylikoski 2010, 53)

Tis applies directly to the analysis in this chapter.

Some approaches to causal mechanisms avoid the use of statistical analysis (Mayntz 2003; Morgan and Winship 2015; Mayntz 2020). But, as a matter of fact, the processes under study here are probabilistic in the aggregate, given the large number of infuencing and contextual variables afecting observed heterogeneity in political responses (Elster 1998). Analysis of empirical counterparts will necessarily involve statistical analysis. Te main objective of the analysis in this chapter is not restricted to fnding out how the efects are feasible, nor solely to establishing the conditions under which the efects might be possible, rather it is to seek to establish whether the mechanisms are in actual fact present. Tis is in line with Morgan and Winship who "take the position that genuine

<sup>7</sup> See Hédoin (2013) for a discussion of the sources of the three main elements and the implications for mechanism-based explanations.

casual depth must be secured by empirical analysis" (Morgan and Winship 2015, 346).

Applying a causal mechanism approach to the study of the political responses to conditional income transfers could make an important contribution. One feature of research on the expansion of social assistance programmes in low and middle-income countries has been the use of quasi-experimental methods to identify the efects of participation in transfer programmes. Tis applies particularly to conditional income transfer programmes (Barrientos and Villa 2015). Often, anti-poverty policies meet considerable political resistance from elites. Impact evaluation studies have provided a means of protecting anti-poverty policies by showing how efective they are (Levy 2006). Multilateral organisations and bilateral donors have supported impact evaluations in order to show aid efectiveness to distant electorates (Barrientos and Villa 2015). Impact evaluation studies identify a treatment and a control group and compare their outcomes before and after the implementation of the programme (Ravallion 2005). Normally evaluation data focus on the particular objectives of the conditional income transfer programmes, for example improvements in consumption and utilisation of basic services, health, and education.8

Researchers have made use of evaluation data to study political efects of programme participation. Tis literature exploits information available from survey questionnaires and, where possible, matches evaluation and electoral data. Studies of conditional income transfer programmes in Latin America fnd higher electoral registration, voting, and political participation among participants (the treatment group) compared to nonparticipants (the control group) (Nupia 2011; De la O 2013, 2015; Baez et al. 2012). Tese fndings have been confrmed with data collected independently (Schober 2019).

Te fndings from these studies are very informative and provide a precise measure of causal efects, but they shed limited light on how these efects are generated. For the purposes of assessing the capacity of conditional income transfers to generate political inclusion, information on the generating processes or causal mechanisms is essential.

<sup>8</sup>Ludwig et al. (2011) have championed the use of mechanism experiments as a cheaper alternative or as a complement to social policy evaluations.

### **3 Findings from the Latin American Literature**

In this section we review the literature on political responses to participation in conditional income transfer programmes with the aim of extracting information and hypotheses on potential causal mechanisms. Tis literature is concerned with identifying the efects of programme participation on a range of political variables, but it seldom tracks the processes by which these efects come to be. Its interest is the "what" rather than the "how". Causal mechanisms are often implicit in the postulated relationships, which the review will attempt to highlight.

Hunter and Sugiyama (2014) provide an excellent perspective on political responses to participation in conditional income transfer programmes based on extensive research in the Northeast of Brazil. Relying on interviews and focus group data they investigate whether *Bolsa Família* receipt enhances citizenship. Teir main hypothesis is described as follows: "Social benefts, acquired through procedures that are judged to be reasonable, fair and transparent can only deepen poor people's appreciation of their newfound political rights" (Hunter and Sugiyama 2014, 829).

Tey discuss two main pathways to improved citizenship, which will be interpreted for our purposes as causal mechanisms. First, transfer receipt enhances citizenship through social inclusion. Transfers signal to recipients that governments "recognise all members of the national community as worthy enough to have their basic needs met and life chances lifted" (Hunter and Sugiyama 2014, 830). Second, transfer receipt strengthens the agency of recipient households in as far as they "foster a sense of recognition, fairness, and rights [that] facilitate the exercise of 'voice', which helps citizens hold their government accountable for meeting basic needs" (Hunter and Sugiyama 2014, 830). Teir research confrms the presence of both causal mechanisms. Te strengthening of agency shows up among respondents who highlight greater economic independence associated with transfer receipt. Voice and accountability arise strongly in the context of recipients' perceptions of the rules-based nature of the programme and their preparedness to engage in political activity to support the programme and resist attempts to dismantle it.

Schober (2019) is interested in the question whether conditional income transfer receipt leads to increasing political participation. Using a dedicated survey of a representative sample of adults in three municipalities in Mexico with high levels of poverty, he explores political participation outcomes and pathways, which will be interpreted as causal mechanisms. He then extends his analysis to other Latin American countries participating in LAPOP in 2014.9 His study contrasts political participation efects of conditional and unconditional income transfers.

His underlying theoretical framework is a rational choice model of political participation emphasising the role of benefts and costs. Transfers infuence political participation by changing the balance of costs and benefts. Conditional income transfers raise the benefts of political participation because of the likelihood that participation will protect beneft receipt in the future. Additionally, he argues that conditions have positive net efects on political participation. On the one hand compliance with conditions will reduce the time available to transfer recipients to engage in political activity, say mothers taking over housework duties previously performed by their daughters. On the other hand, conditions involve greater engagement with other recipients and ofcials leading to improved organisational and communication skills. He describes these as "nonmaterial resources", which act to reduce the costs of political participation. For my purposes, the rational choice model can be dropped in order to focus attention on the observable causal mechanisms identifed here.10

Te study fnds that conditional transfer recipients are more likely to engage in a greater number of political activities (voting, contacting public ofcials, community activism, and civil society engagement) compared with unconditional transfer recipients in the sample who are only likely to engage in a single activity (voting).11 Looking further into

<sup>9</sup>LAPOP is a project collecting attitudinal data from Latin American countries. Tese data will be described in detail in later sections.

<sup>10</sup> In fact, basic rational choice models like the one sketched in the paper are normally estimated with reduced form latent statistical models dropping causal mechanisms as a matter of course.

<sup>11</sup> "In line with my expectations, there is a positive and statistically signifcant relationship between CCTs [*conditional income transfers*, AB] and several modes of political participation. CCTs are positively associated with contacting public ofcials, community activism, civil society engagement, and voting. For the average respondent, the estimated efect of CCT participation corresponds to an increase of 26 percent in contacting public ofcials, an increase of 28 percent in community

possible pathways, the study fnds that conditional transfer "participation is positively associated with each measure of civic skills, whereas UCT [*unconditional income transfer*, AB] participation is only positively associated with attending meetings. In terms of the substantive efect, it is estimated that CCTs [*conditional income transfers*, AB] lead to a 59 percent increase in the exercising of civic skills for the average respondent, while UCTs only lead to an increase of 29 percent" (Schober 2019, 598–99). He does not rule out the possibility of alternative pathways or causal mechanisms but fnds no supporting evidence for alternative causal mechanisms put forward in the literature.12

Analysis of LAPOP cross-country data for 2014 confrms that for conditional income transfers, "participation is positively and signifcantly associated with several modes of political participation, including campaign activism, contacting public ofcials, civil society engagement, and protest" (Schober 2019, 602).

Layton and Smith (2011) study the electoral outcomes from transfer receipt using 2012 LAPOP data for twenty-three countries. Like Schober, they start with a basic rational choice model in which transfers infuence the balance of costs and benefts. As they put it, "by altering benefciaries' pocketbook calculations and directly linking recipients' well-being with state actors and policies, social assistance has the obvious potential to alter voting behaviour" (Layton and Smith 2011, 855). Tey pay particular attention to programme conditions as a source of political efects.

Te paper correlates conditional income receipt with voter choice and turnout preferences. Four of the institutional conditions infuencing preferences are modelled directly: enforced compulsory voting, programme conditionality, ideological leftist president, and programme independence. In terms of causal mechanisms, meaning how these efects are generated, the paper only provides a basic discussion. It makes

activism, an increase of 21 percent in civil society engagement, and an increase of 6 percent in voting" (Schober 2019, 597).

<sup>12</sup> Supplementary analysis of his sample of Mexican households fails to confrm an association between conditional transfer receipt on the one hand and: (i) proxies for inclusion (whether respondents believe governments are interested in them or whether they believe their vote matters); (ii) agency (whether respondents understand political issues or are informed about local government or are able to help community groups); or (iii) interest in politics.

reference to the possibility that: "Familiarization with state bureaucracies could also reduce the psychological or cognitive costs of turning out at the polls for marginalized families with very few prior positive experiences with the state" (Layton and Smith 2011, 859). It also suggests that "benefts may increase psychological attachment to the state and national politics … Once they discover that the state can directly meet their needs, assistance recipients gain a new stake in political contests" (Layton and Smith 2011, 859).

Layton et al. (2017) examine whether the observed political efects of participation in *Bolsa Família* constitute reciprocal support for incumbents (transfers for votes) or refect broader positive attitudes to state legitimacy. Te study relies on LAPOP data for Brazil for the period 2007–2014. Methodologically, the study compares attitudes among *Bolsa Família* recipients with a control group constructed using propensity score matching techniques.13 Te authors adopt a "thick" notion of state legitimacy requiring "attachment to a political object for its own sake … independent of system outputs" (Layton et al. 2017, 102), also referred to as difuse forms of legitimacy. Teir main fnding is that "the immediate political consequences of CCTs are not limited to positive electoral results from incumbents …. [but] we fnd little support for the hypothesis that Bolsa Família is associated with difuse dimensions of legitimacy … Brazil's CCT recipients are not any less supportive of diffuse elements of the political regime than nonrecipients" (Layton et al. 2017, 113).

Te operationalisation of state legitimacy in the context of the LAPOP is of particular interest for our purposes. An index of political legitimacy combines six dimensions: support for incumbent; support for local government; support for core political institutions; support for core regime principles; regime economic performance; and sense of political community. Tis is a complex index with some dimensions themselves an index

<sup>13</sup>Propensity score matching is a statistical technique matching treated observational units with non-treated units on selected observables. It helps identify non-treated units that have similar profles as the treated units. Its main drawback is that it necessarily relies on observed characteristics and therefore assumes unobserved characteristics are the same across treated and matched units. Tis is often problematic in selective interventions.

of specifc survey responses. Placed against a causal mechanism framework, the legitimacy difusion index has the disadvantage of mixing inputs, outputs, and outcomes. For example, incumbent support is measured additively in terms of the perceived efcacy of incumbents on three issues: whether they combat corruption, improve citizen security, or manage the economy well. Support for core political institutions is operationalised through an additive index of eight survey responses, including trust in state institutions, respect for political institutions, and support for the political system. Tis dimension is closest to describing an outcome measure of a causal mechanism. Teir estimation fnds that "across all survey years, Bolsa Família recipients report signifcantly higher levels of support for core political institutions, trust in local government, and support for political actors than their matched nonrecipient peers" (Layton et al. 2017, 109).

Te review of this literature shows that most studies reference causal mechanisms, even if only implicitly. Where referenced, causal mechanisms are mainly accessories to the estimation of correlates and their interpretation. Te review drew attention to four *potential* mechanisms structuring recipients' political responses to transfer receipt: (i) *reciprocal support for incumbent mechanism* (or votes for transfers); (ii) a *support for redistribution mechanism* encouraging electoral participation in support of incumbent redistributive policies; (iii) a *bureaucratic mechanism encouraging political engagement* by transfer recipients through the implementation features of the transfer programmes, for example birth and electoral registration, compliance with programme conditions, interactions with programme agencies; and (iv) a *cognitive change mechanism* in which transfers facilitate a cognitive change in recipients as regards their societal and political inclusion.14

<sup>14</sup>A case could be made to consider a ffth potential causal mechanism in which transfers encourage recipients' engagement in local or community politics. Unfortunately, data restrictions preclude empirical investigation of a mechanism along these lines. However, this is studied by Schober (2019) using a dedicated survey of three Mexican municipalities. He fnds supporting evidence that participants in the *Prospera* programme engage in community politics to a greater extent than recipients of unconditional income transfers in Mexico.

## **4 Causal Mechanisms Structuring Political Responses to Transfer Receipt**

Tis section discusses in more detail the four potential mechanisms identifed in the previous section. In fact, applying the causal mechanism approach rules out reciprocal support for incumbents as a causal mechanism. Te discussion starts with this point.

#### **4.1 Reciprocal Support for Incumbents Is not a Causal Mechanism**

Te literature on political responses to transfer receipt often fnds relatively higher support for the incumbent among participants in conditional income transfer programmes (when compared to non-participants) (De la O 2015; Zucco 2013; Díaz-Cayeros et al. 2016; Baez et al. 2012). From the perspective of political coalitions, this fnding appears to confrm the view that conditional income transfers are primarily an electoral strategy open to policymakers. One advantage of approaching political participation from the perspective of recipients is to raise alternative interpretations for this fnding.15 A pocketbook approach to electoral politics views politicians as entrepreneurs placing a "retail ofer" of policies to voters. Voters rank the policy portfolios in line with their pocketbook and vote accordingly. In the context of social assistance transfers, participants can be seen to reward incumbents that provide them with public transfers. A condition for this to work is that transfer recipients associate the transfer with a particular incumbent. Tis condition is easily met in conditional income transfers. An alternative interpretation is that recipients vote for incumbents in order to protect the programmes that provide benefts to them. Tey are concerned that if an opposition candidate is elected, they will drop or reform the transfer programme. Tis was mentioned in Hunter and Sugiyama's research (Hunter and Sugiyama

<sup>15</sup>Tilly (2001) approvingly cites Collier's summing up of political research on democracy to the efect that "recent analyses have concentrated excessively on deliberate elite decisions at the expense of social processes and popular actors" (Collier 1999).

**Fig. 13.2** Reciprocal support for incumbent. (Source: Own presentation)

2014). Electoral competition in the context of re-election motivates political leaders to search for additional votes.

Te postulated relations do not amount to a mechanism. Tis follows from the description of causality in directed graphs. Causality requires, inter alia, that directed graphs are acyclic. In directed acyclic graphs, "no directed path [the direction of the arrows] emanating from a causal variable also terminates at the same variable … ruling out representations of simultaneous causation and feedback loops", see Morgan and Winship (2015, 80). Figure 13.2 depicts the links. Arrows show causal links. Tey indicate that recipients vote for incumbents in order to preserve the transfer programme or as a reward for the benefts, while politicians run the programme as a means of eliciting electoral support from participants. Note the circular nature of the relationships implying that no causal direction could be identifed by relating transfer receipt and higher incumbent support after the programme has started running. Te transformation component that is present in Fig. 13.1 is missing in Fig. 13.2 as the original social situation is reinforced by recipients' actions in response to the transfer.

#### **4.2 Support for Redistribution Mechanism**

A causal mechanism can be proposed when recipients vote for the incumbent because they acknowledge, qua citizens, that the incumbent is committed to poverty reduction and social justice. Tis mechanism entails several conditions: that recipients can identify the relevant politicians or coalitions reasonably well and can be reassured that voting for incumbents does not generate negative trade-ofs with other policy felds, say economic policy for example. Ideological or partisan loyalties with the governing coalition could help strengthen this mechanism.

**Fig. 13.3** Support for redistribution mechanism. (Source: Own presentation)

Figure 13.3 shows the causal links. Te dashed line describes the fnding from correlation studies on the presence of a transfer programme efect on support for incumbents. Te fgure shows in more detail the main elements of the causal mechanism at work. Recipients are shown to vote for incumbents supporting transfer programmes because of their commitment to the reduction of poverty and inequality. Isolating the links existing between recipients, incumbent support, and support for redistributive policies now shows acyclic causal links, compared to the cyclical relationships in Fig. 13.2. Note that the support for redistribution mechanism rejects the proposition that recipients vote for incumbents solely because of the transfer, as in the reciprocal support hypothesis discussed above. Recipients are unlikely to vote for politicians who might run conditional income transfer programmes as a pure electoral strategy but who otherwise fail to support wider redistribution policies.

#### **4.3 Bureaucratic Mechanism Encouraging Political Engagement**

Te review of the literature on political responses to conditional transfer programme participation suggests that, in the context of restricted citizenship for disadvantaged groups well established in the literature, conditional income transfers work to raise political engagement and extend citizenship (Hunter and Sugiyama 2014). Te predicted outcome is increased support for political institutions among low income groups.

**Fig. 13.4** Bureaucratic mechanism encouraging political engagement. (Source: Own presentation)

Political engagement associated with programme participation might work through alternative channels. A bureaucratic channel operates through the requirements for birth and electoral registration embedded in the implementation of the transfer programmes (Baez et al. 2012; De la O 2015; Hunter and Sugiyama 2018). In particular, electoral registration is expected to generate improvements in turnout and engagement in associated electoral activities. Figure 13.4 depicts the set of relationships posited by this mechanism.

#### **5 Cognitive Change Mechanism**

A diferent channel identifed in the literature works through cognitive changes associated with programme participation. Tis is related to recipients' engagement with programme agencies and local politicians involved in the implementation of the programme (Schober 2019). An ideational channel works through the fact that transfers strengthen perceptions among recipients that government policy acknowledges their disadvantage and is committed to addressing it. In so doing recipients are recognised as full members of a polity, as citizens, with associated rights (Hunter and Sugiyama 2014). An outcome of this mechanism is that recipients are likely to show support for political institutions (compared to non-recipients). Figure 13.5 depicts the relationships involved in this mechanism.

In the next section, the three mechanisms will be assessed with attitudinal data from Latin America. Before doing so, it will be helpful to underline the main contribution of the causal mechanism approach. Applying a mechanism perspective helps to clarify the causal links existing between conditional income transfer programmes, behavioural responses, and outcomes. Te discussion in this section points to three important contributions.

First, while acknowledging the valuable insights from the existing literature on the political efects of large-scale transfer programmes, an explicit focus on causal mechanisms rectifes two weaknesses in this literature: (i) its one-sided focus on policymakers; and (ii) its inattention to the transformational components of social assistance. Shifting the focus onto transfer recipients and paying attention to outcomes extends the causal links in productive ways. Te existing literature is perhaps too focused on correlating inputs and outputs, much less so on outcomes.

Second, applying a causal mechanism framework shows that the most common fnding in the literature, recipients' reciprocal electoral support for incumbents, does not in itself support causal inference. A causal mechanism approach shows that the transformational component is missing.

**Fig. 13.5** Cognitive change mechanism. (Source: Own presentation)

Tird, paying attention to how the transfers may infuence recipients' political engagement is essential to our understanding of the broader impact of conditional income transfers on the (political) inclusion of disadvantaged groups.

## **6 Interrogating LAPOP Data**

Te causal mechanisms identifed in the preceding section will be examined empirically using data from the AmericasBarometer (Latin American Public Opinion Project (LAPOP) 2020). LAPOP carries out regular opinion surveys across all the countries in the Americas. LAPOP remains the only regionally consistent dataset that identifes social assistance respondents. An additional advantage of LAPOP surveys for our purposes is their extensive coverage of political attitudes. During the 2000s, selected country surveys asked respondents whether they received any support from conditional income transfers. See Table 13.1 for a listing of


**Table 13.1** Conditional income transfer recipients in LAPOP by wave and country

Source: Own calculation from LAPOP data.

the surveys including identifers for conditional income transfer participation. Identifcation of conditional income transfer recipients is very reliable because respondents were presented with an explicit reference to the main transfer programme. It asked: *Now referring specifcally to [a name of the country's fagship programme followed] are you or someone in your household a benefciary from this programme?* Tis question was applied in countries with at least one large-scale conditional income transfer programme but only in selected years for selected countries. In some cases, country samples were divided in two and this question was included only for one half of the sample.16

Te empirical work reported below interrogates pooled data for the period 2000–2019. It does not consider the extent to which particular country conditions infuence political responses. Tis is an important issue which will be the subject of further work, but the focus here is on identifying core causal mechanisms. It is important to keep in mind that LAPOP collects a similar number of respondents from each of the countries covered, say 1000 interviews from Brazil and 1000 interviews from Paraguay, so that the pooled data is not representative of the population in Latin America taken as a whole.17 Te analysis will compare political responses and outcomes between conditional income transfer participants and non-participants. Tis is justifed because the objective is to examine whether participation in conditional income transfer programmes improves the political inclusion of disadvantaged groups.18

Table 13.2 reports on the odds ratios of the particular variables of interest from a logistic regression of the pooled data. Te three causal mechanisms identifed in the previous section are tested separately. Te

<sup>16</sup>Tey also included a question on whether they received any support from the state which could be interpreted to refer to social assistance as a whole. As the focus in this chapter is on conditional income transfers, the analysis will rely on the question specifc to them. It enables a precise identifcation of recipients. But, in fact, most empirical results presented here carry through to the social assistance question.

<sup>17</sup>LAPOP data is representative at the country level, and within each country at the rural or urban levels.

<sup>18</sup>Are there self-selection issues capable of confounding the results? Tey would be present if, for example, conditional income transfer recipients had been selected in the frst place for their particular support for redistribution or for the political system. Confounding could also occur when selecting recipients on socio-economic grounds replicates these biases. Te consensus in the literature indicates these conditions are unlikely to be present in Latin America (Kaufman 2009).


**Table 13.2** Odds ratios using LAPOP 2010–2019 data

\*\*\* p<0.01, \*\* p<0.05, \* p<0.1

Source: own calculations.

table reports on the main variables of interest. Te controls include a wealth index, and country and year dummies.

Beginning with the support for redistribution mechanism, results are reported in column (1) in the table. Tis causal mechanism links transfer receipt to voting and then to preferences for redistribution policies: transfer recipients are more likely to vote for incumbents in order to support redistributive policies. Transfers are the input, voting for incumbents is the output, and support for redistributive policies is the outcome. Te latter refects the transformational component. Te empirical counterparts are as follows. Recipients are directly identifed in the survey. Te survey asked respondents whether, if presidential elections were to be held next week, they would be likely to abstain from voting, spoil their vote, vote for the incumbent or vote for the opposition. To simplify matters the frst two options are combined and are the reference category in the regression (referred to below as "not voting"). Te surveys also asked respondents whether they agree with the statement: *Te state should*  *implement strong policies to reduce income inequality between rich and poor* by selecting a value on a scale from one to seven, where one is "strongly disagree" and seven is "strongly agree". A "supports redistribution" binary variable was generated by combining categories six and seven, those most strongly agreeing with the statement.

As can be seen from the frst column of Table 13.2, conditional transfer recipients are 1.08 times more likely to support redistribution than non-recipients, while those planning to vote for the incumbent were 1.26 times more likely to support redistribution than those not voting. Figure 13.6 shows linear combinations of the odds ratio of the variables of interest. As can be seen there, transfer recipients planning to vote for the incumbent are 1.36 times more likely to support redistribution than non-recipients not planning on voting. Tis is consistent with the support for redistribution mechanism, indicating that transfers encourage recipients to register and vote, and to vote for incumbents with the objective of supporting redistribution policies.

Turning to the bureaucratic mechanism encouraging political engagement, the results can be found in column (2) of Table 13.2. Tis

**Fig. 13.6** Odds ratios linear combinations of variable estimates. (Source: Own elaboration)

mechanism links transfer receipt to bureaucratic requirements associated with participation, electoral registration in particular, themselves leading to greater political engagement and support for the political system. To capture support for the political system the analysis relies on a binary transformation of responses to the question: *On a scale of one to seven, where one is "none" and seven is "a lot", to what extent you believe the political system should be supported?* Te binary transformation coded support for the political system by combining responses six and seven on the scale. When examined, LAPOP data show no signifcant diference in the rates of registration among recipients and non-recipients. If participation in conditional income transfer programmes required voter registration at the start of programme participation, this efect is bound to decline after the implementation of the programme as it applies to new entrants only.19 Registration is not included as a separate variable in the regression analysis but voting for the incumbent or the opposition can be used as a proxy.

Te estimates in Table 13.2 show that transfer recipients are 1.4 times more likely to support the political system than non-recipients. Respondents planning to vote for the incumbent are 1.2 times more likely to support the political system than respondents not voting. Te estimates for respondents planning to vote for the opposition are lower but positive and signifcant. Figure 13.6 shows the linear combinations of the odds ratio of a transfer recipient who plans to vote for the incumbent. It shows that recipients are 1.7 times more likely to support redistribution than a non-recipient not voting. Tis result is consistent with the bureaucratic mechanism encouraging political engagement.

Column (3) of Table 13.2 reports on the cognitive change mechanism. Tis mechanism stresses cognitive change associated with transfer recipients' awareness of their fuller membership in their polity, leading to the same transformational element as the previous mechanism, namely a strengthening of their support for the political system. To capture this particular mechanism, the analysis relies on a survey question asking respondents whether they agree with the following statement: *Political leaders in this country are interested in what people like you think*, again on

<sup>19</sup> See the discussion in Layton and Smith (2015) underlining the fact that voting and voter registration are compulsory in most countries in the region.

a Likert scale with one denoting "strongly disagree" and seven denoting "strongly agree".

Te odds ratios reported in column (3) in the table indicate that transfer recipients are 1.2 times more likely to support the political system. Respondents agreeing that political leaders listen to people like them are 1.7 times more likely to support the political system. Figure 13.6 shows that transfer recipients who believe the government listens to people like them are 2.1 times more likely to support the political system. Again, these fndings are in line with the cognitive change mechanism proposed above.

To sum up, the analysis compared political responses and outcomes for conditional income transfer recipients and non-recipients. Analysis of the LAPOP 2010–2019 data fails to reject the presence of all three causal mechanisms proposed in the previous section.

## **7 Conclusions**

Te chapter tackled the issue whether conditional income transfer receipt generates political responses associated with improvements in political inclusion. Tis is central to an assessment of the emerging new welfare institutions in low and middle-income countries. Political inclusion is key to sustainable and efective poverty reduction.

Te chapter is related to the wider literature examining the political implications of conditional income transfer programmes. It contributes to this literature in several ways. First, a causal mechanism approach opens up the black box of quasi-experimental evaluation studies, paying attention to the "how" as a complement to the "what" and the "why". Second, and this is perhaps another way to make the frst point, a causal mechanism approach clarifes the linkages between interventions, responses, and outcome dimensions of welfare institutions. An assessment of whether participation in conditional income transfer programmes improves political inclusion requires that we pay attention to political outcomes (redistribution, strength of the political system), not just to actions or outputs (voting, registration). A causal mechanism approach also ofers essential clues as to transformational components of social assistance. Te focus on participants in transfer programmes is essential in this context. Tird, the analysis of LAPOP data confrms that the mechanisms proposed in the literature can be productively studied with empirical counterparts.

Te analysis in the chapter identifes three main causal mechanisms linking conditional income transfer receipt to political inclusion. Tey include a support for redistribution mechanism, a bureaucratic mechanism encouraging political engagement and a cognitive change mechanism. Analysis of empirical counterparts in the LAPOP data provides some support for their relevance to our understanding of political responses among participants.

Te discussion raises several issues for further research. Te approach and fndings in this chapter aimed at explanation as opposed to prediction. As noted above, pooled LAPOP data is not representative for Latin America as a whole. Te empirical results are indicative of the direction of causal links, not necessarily of their particular weight. Te causal mechanism approach does not provide clear guidance on how to assess the relative relevance or complementarity of competing mechanisms. Further research will need to be undertaken in order to clarify under what conditions causal mechanisms can be strong or weak, for example by examining causal mechanisms across political regime conditions at country level. Te same applies to the need to study changes in political regimes over time. Finally, there is only scant literature that considers the role of programme design in strengthening or weakening causal mechanisms and outcomes (Schober 2019; Bruch et al. 2012). Future research tackling comparisons across social assistance programmes might prove to be informative.

#### **References**


Barrientos, Armando. 2013. *Social Assistance in Developing Countries*. Cambridge: Cambridge University Press.

———. 2019. Conditional Income Transfers, Social Policy and Development. In *Handbook of Social Policy and Development*, ed. James Midgley, Rebecca Surender, and Laura Alfers, 373–392. Cheltenham: Edward Elgar Publishing.


———. 2015. *Crafting Policies to End Poverty in Latin America. Te Quiet Transformation*. Cambridge: Cambridge University Press.

Deaton, Angus. 2020. Randomization in the Tropics Revisited: A Teme and Eleven Variations. *NBER Working Paper Series*, Working Paper 27600. Cambridge: National Bureau of Economic Research.


66. Vanderbilt: LAPOP Vanderbilt University. https://www.vanderbilt.edu/ lapop/insights/IO866en.pdf. Accessed January 18, 2021.

———. 2015. Incorporating Marginal Citizens and Voters: Te Conditional Electoral Efects of Targeted Social Assistance in Latin America. *Comparative Political Studies* 48 (7): 854–881.


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.

## **Part VI**

**Concluding Section**

## **14**

## **Conclusion and Outlook: Towards a Systematisation of the Mechanism-Based Approach in Social Policy Research**

**Johanna Kuhlmann and Frank Nullmeier**

## **1 Introduction**

Tis edited volume seeks to explore new ways of explaining social policy by adopting a mechanism-based approach. Te aim of the chapters has been to identify causal mechanisms that can explain the development of one or several social policy programmes. Te chapters thus demonstrate how individual case studies can use the tools of a mechanism-based analysis. In this concluding chapter, we aim to show how causal mechanisms

Tis chapter is a product of the research conducted in the Collaborative Research Centre 1342 "Global Dynamics of Social Policy", funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation)—Projektnummer 374666841—SFB 1342.

J. Kuhlmann (\*) • F. Nullmeier

University of Bremen, Bremen, Germany e-mail: johanna.kuhlmann@uni-bremen.de; frank.nullmeier@uni-bremen.de

J. Kuhlmann, F. Nullmeier (eds.), *Causal Mechanisms in the Global Development of Social Policies*, Global Dynamics of Social Policy, https://doi.org/10.1007/978-3-030-91088-4\_14

can also contribute to cross-case study research. Mechanism-based comparative social policy research aims to achieve an analytical level at which insights from single case studies and comparative analyses can mutually inform and complement each other. Tis can be achieved through modularisation, that is, by combining several mechanisms to explain a single policy process. Accordingly, individual mechanisms can serve as modules to explain social policy developments in a wide range of cases, but they cannot explain the entire process of introducing a new social policy programme or reforms of existing policies. Mechanisms are thus analytical elements that can explain a particular case when combined with each other. In fact, this is the standard approach in the preceding chapters. In most studies, a historical episode or period is explained through several interacting mechanisms. Some contributions, however, proceed diferently. Tey focus on one or two mechanisms and clarify their specifc role in a broader course of development, while not aiming to explain the whole process (see Chaps. 7, 9, and 11). What is more, some contributions also try to break down the complex causal mechanisms into elementary causal mechanisms as elaborated in the Introduction to this edited volume (see Chaps. 8 and 9). Tese case studies aim for an even more detailed analysis by shedding light on the reasons for individual actors' actions.

Modularisation facilitates the detection of links between studies on quite diferent areas of social policy in diferent countries and at diferent times. For example, the *alarmed middle classes mechanism—*which is, like the other mechanisms mentioned here, explained in Sects. 2 and 3—can be found to be just as relevant for pension policy in South Korea (see Chap. 3) as for health policy in Bolivia (see Chap. 12). Likewise, the *outcompeting mechanism* was identifed in the cases of South Korea (see Chap. 3) and Turkey (see Chap. 4), and the *transnational cooperation mechanism* can capture developments when it comes to unemployment insurance in Turkey (see Chap. 4) and health policy in Albania (see Chap. 9). Moreover, there are proximities or similarities between mechanisms, such as the *double beneft mechanism* for the cases of South Korea, Vietnam, Sri Lanka, and Malaysia (see Chap. 3) and the mechanism of *public resource accumulation* for the case of Uganda (see Chap. 6).

Notably, across all contributions, there are some mechanisms that focus on the same aspect when explaining a particular development in social policy, such as transnational infuences on national social policies. However, the individual studies highlight how diverse the interaction between national and transnational actors can be. Te analyses throughout this volume therefore show that it is not enough to argue that single factors like transnational infuences can, to a certain extent, explain social policy developments. Rather, it is important to understand how this factor actually plays out and, accordingly, to explain it through diferent causal mechanisms.

Finally, it also turns out that the number of mechanisms identifed throughout this volume is not so limited that we could derive a simple and general theory of social policy. Rather, we fnd that the diversity of social policy developments is refected in a plurality of mechanisms that have been identifed as relevant. Terefore, as a next analytical step, it is necessary to categorise the mechanisms and to bring them into a systematic order. Tis concluding chapter will therefore present a structured compilation of the complex causal mechanisms that were identifed by the authors in this edited volume and indicate perspectives for further mechanism-based social policy research.

Te complex causal mechanisms compiled in this volume can be distinguished, frst, according to whether or not they are mechanisms that are closely linked to a specifc social policy sector or to a particular social policy institution, for example social insurance (see also Chap. 3). Tis distinction into special and more general mechanisms aims at stressing diferences in the scope of a mechanism. Can a mechanism occur in all felds of social policy (or even in felds that are not related to social policy), or is it tied to a specifc institutional arrangement which, if absent, also indicates that the mechanism cannot be present? Tis type of clustering is based on how the mechanisms have been defned in the individual chapters of this volume. If they are formulated in such a way that they are tied to the preconditions of a specifc social policy feld or to a specifc type of social policy institution, we classify them as *policy-specifc mechanisms*; if not, we classify them as *general mechanisms.* Of course, this classifcation remains to a certain extent provisional, as it is only based on the analyses that were conducted throughout this volume. Follow-up studies can put these particular mechanisms to the test to analyse whether they have a broader scope.

What is even more interesting is how a systematic clustering of the mechanisms mentioned in the individual chapters could be achieved. We distinguish two options. First, the mechanisms can be classifed depending on which actors are the reference point of the respective mechanism. Tis is in line with the actor-centred approach pursued throughout this volume. Second, we can focus on the thematic proximity between the mechanisms and on well-known theoretical approaches within the feld of comparative social policy as a way of classifying the mechanisms.

Following the frst option of clustering via actors, we can structure the general mechanisms according to diferent groups of actors that have been found to be decisive for social policy development, which are (a) transnational actors, (b) policymakers and administrators, (c) political parties, (d) voters, and (e) interest groups and social movements. However, some chapters also identify mechanisms that we cannot directly assign to a single actor or type of actor, such as the mechanism of *anti-communist backlash* (see Chap. 8). Terefore, an actor-centred classifcation of mechanisms does not cover the full set of mechanisms that were identifed throughout this volume. Moreover, a classifcation that is only actorcentred is not linked to the established approaches in social policy research and therefore not able to show how the identifed mechanisms relate to existing theoretical approaches.

Given this background, as a second criterion of clustering, diferent mechanisms can be linked to established approaches in comparative social policy research. Here, a distinction between (a) the socio-economic theory, (b) the power resource approach, (c) the parties matter approach, approaches focusing on (d) political institutions, (e) on globalisation and policy difusion, and (f) on policy heritage and path dependency as core elements of explanation can serve as inspiration (e.g. Obinger and Schmidt 2019). Such a clustering has two analytical advantages: First, while applications of these theories often tend to leave processes uninvestigated, our mechanisms help to illuminate the black box of what is happening between x and y. For example, instead of arguing that parties matter for expanding social benefts, causal mechanisms are able to explain what the causally relevant steps in such a process are, that is, precisely how parties and their interaction with other relevant actors lead to the expansion of social benefts. Second, it also becomes clear from such a clustering that the mechanisms in this volume are by no means meant to replace existing theories and approaches in comparative social policy. Rather, they ofer building blocks for explaining social policies with clear links to existing approaches. Importantly, sometimes these mechanisms also allude to causal relationships between the elements in a theoretical framework that tend to be overlooked in empirical applications, but that clearly belong to the core of the respective theoretical approaches. Te risk of clustering solely on the basis of established theoretical approaches is to reproduce these approaches in the language of mechanisms. Yet the modularised approach ofers many opportunities to identify causal relationships that also indicate gaps in or between these established approaches.

In the following, we therefore work tentatively with a combination of both forms of clustering, that is, actor-centred and theory-centred clustering. Our approach is facilitated by the fact that many of the established approaches also place specifc actors at the centre of their explanatory approach. Still, some caveats to our combined actor-centred and theory-based clustering should be mentioned: First, not all theoretical approaches outlined above can serve as a reference point for our classifcation of mechanisms. We exclude the socio-economic theory (e.g. Obinger 2019) due to its diferent methodological background: It focuses on macro-level causal efects and not on the causal links between independent and dependent variables, which is at the core of mechanismbased analysis. Moreover, the socio-economic theory does not theorise on actors and is therefore not compatible with our actor-centred approach. Second, with regard to the actor-centred clustering, it is well known that social policy developments can rarely be explained by the actions of a single actor (Heclo 2010). Although the clustering via actors thus highlights single actors, most mechanisms focus on the interaction of diferent actors—for example, transnational actor mechanisms trace the interaction between transnational actors and national actors. Moreover, our modular approach can capture how diferent actors can be involved in a particular process, as studies can combine, for example, transnational actor mechanisms and interest group mechanisms. Tird, some chapters in this volume also refer to mechanisms that have been discussed in the social policy literature, but without the emergence of a separate theoretical approach. In these cases, the criterion of actor-centred clustering can function as a way of systematically integrating such mechanisms as well. Taken together, this type of clustering thus serves the main purpose of this book, namely to complement, expand, deepen, and possibly also correct existing approaches to social policy (see Chap. 1).

In the following, we collate the mechanisms that were identifed throughout this volume and complement them with additional mechanisms that can be theorised from the literature, to arrive at a comprehensive list of causal mechanisms that can contribute to explanations of social policy developments, while not claiming that this list is exhaustive. We begin by presenting the general mechanisms and focus afterwards on the policy-specifc mechanisms.

### **2 General Mechanisms**

#### **2.1 Transnational Actor Mechanisms**

A vast body of literature acknowledges the important role that developments beyond the national level play for social policy-making. Scholars focusing on globalisation have studied how this very process shapes the level of national social benefts (Starke and Tosun 2019). Scholars focusing on policy difusion have studied to what extent social policies that are adopted in one country have been infuenced by social policies from other countries or International Organisations (IOs), distinguishing coercion, competition, emulation, and learning as types of difusion (Maggetti and Gilardi 2016; Obinger et al. 2013). However, this literature has often been criticised for not acknowledging processes and the active role that diferent actors play within them (Kuhlmann et al. 2020). A more systematic search for causal mechanisms can contribute to a refnement and modifcation of the causal relations that shape difusion processes.

In his chapter on the adoption of social protection policies in Africa, Devereux (Chap. 7) introduces the mechanism of *policy pollination* and traces the process in which single actors from international development agencies travel to diferent countries to promote their preferred policies, thereby applying diferent strategies. Focusing more on the social construction of social policies as they have, for example, also been discussed in the literature on policy emulation, Sirén in his chapter on healthcare reform processes in Bolivia (Chap. 12) traces how transnational epistemic communities shaped an understanding of health as a citizenship right through a mechanism of *expert theorisation*.

However, it would be misleading to neglect the role that national actors play in such processes. IOs make use of diferent strategies to convince national actors of their preferred policies, ranging from loans or technical assistance to expert exchange through workshops or publications (Orenstein 2008). National actors might then acknowledge the promoted policy as a suitable option. Yet, they are not passive actors that are just doing what IOs ask them to. Rather, they have considerable leeway in shaping and implementing IO suggestions (Kuhlmann and ten Brink 2021; Leisering et al. 2017). Te mechanism of *transnational cooperation* captures this interaction, focusing on how IOs and national actors build alliances to push through a social policy reform. For example, Druga (Chap. 9) traces this mechanism in the case of social policymaking in post-communist Albania, showing how the World Bank and the Albanian government worked together on the introduction of social health insurance. What is interesting here is that the World Bank, which did *not* favour the Albanian decision for an insurance scheme, nevertheless continued to cooperate with the national government, hoping to be able to infuence the fnal law. Druga's study is therefore a remarkable example of how national actors play a crucial role in transnational cooperation, keeping a clear eye on national considerations. Öktem (Chap. 4) also refers to the mechanism of transnational cooperation with regard to the role of the European Union in shaping the development of unemployment insurance in Turkey.

Another way in which the interaction between transnational and national actors can also clearly follow the ideas and preferences of national actors is stressed in the *evasion mechanism*, which captures how national actors highlight the compatibility of IO suggestions and national policy proposals, while actually pursuing their own agendas. Tis has, for example, been observed with regard to pension policy in some Eastern, South-Eastern, and Southern Asian countries (Chap. 3). Finally, the *seeking solutions abroad mechanism* (Chap. 8) captures how national policymakers actively look for policies developed in other countries or possibly also IOs, which again stresses the proactive role that the national level can play in transnational interactions.

#### **2.2 National Policymaker and Administrator Mechanisms**

Tis group of mechanisms focuses on actors within the national political arena. We conceptualise policymakers as actors within the political system who have authority to make policy decisions. Alongside mechanisms focusing on policymakers, we consider mechanisms that highlight the role of actors within the administration, who have often been found to play a crucial role not only when it comes to policy formulation, but also in the agenda-setting phase (Klenk 2019). To begin with, existing analyses on the role of administrators might be translated into a *bureaucratic inspiration mechanism*: Decision-making processes are strongly driven by individual actors within the administration. Individual ofcials often have decades of experience in a policy feld. Because of their central positions in ministries, they can pursue long-term goals in social policy and enjoy professional recognition within their organisation and among the political elites (Hassenteufel et al. 2010). Tus, they might push for introducing or expanding a particular social policy inside their organisation.

Ten Brink, Müller, and Liu (Chap. 2) provide an example of how mechanisms that stress the role of policymakers and administrators interact and can, when combined, explain the introduction of social insurance schemes for urban areas in China. Te *policy experimentation mechanism* explains how administrative units within the state or semi-state administrations set out to fnd new solutions to perceived social and economic developments, thereby making use of the internal diferentiation of the state apparatus and its federal multi-level structure. In the case of China, instead of implementing one policy solution, policymakers tested diferent variants of social insurance schemes at the regional level, which were inspired by domestic and international experiences. More specifcally, ten Brink, Müller, and Liu distinguish a *strategic policy experimentation mechanism* and a more *neutral policy experimentation mechanism*. While in the former policy experimentation is a means to push particular policies in a context of political confict, in the latter policy experimentation serves to provide information on how diferent social policies work. Moreover, ten Brink, Müller, and Liu identify an *elite cooperation mechanism*, which explains how decisions within the governing party-state elite were made, either building on a broad consensus between the involved actors or refecting a compromise. Finally, the *top-leader intervention mechanism* explains how hierarchical decisions are made by single powerful actors within the administration.

Tyen and Schlichte (Chap. 6) explore the role of policymakers and administrators in a colonial context, focusing on Tunisia and Uganda. Te *imperial stafng mechanism* explains how colonial governments create pensions for specifc staf such as civil servants and soldiers to ensure the loyalty of groups that they perceive as important. Moreover, the authors identify an *appropriation mechanism* in the process of decolonisation, referring to existing colonial structures in the feld of social policies that governments now relied on for their own aims. To maintain or increase legitimacy, governments expanded colonial pension schemes for public ofcials to greater and greater parts of the population and introduced social insurance schemes for additional groups.

Political and economic expectations within the broader population are also highly relevant for some mechanisms. A mechanism that we can theorise but that was not identifed in our chapters is the *anticipatory reforming mechanism*, which centres on how governments anticipate possible future problems. For example, if a socio-economic situation is experienced as upheaval or if fundamental changes are taking place and these could also shape political conficts in the future, parties can advocate for or push the introduction or expansion of social policies, even if there is no pressure from other political actors and especially other parties (Rimlinger 1971). Heinrich, Isabekova, and Pleines (Chap. 5) explain the introduction of mandatory health insurance schemes in the post-Soviet region by what they defne as a *resistance avoidance mechanism*. Te mechanism explains how policymakers introduced social insurance although it was not their preferred policy solution. Yet, the adoption can be traced back to social policy expectations within the population and no clear opposition to the reform. In their chapter on the migrantisation of long-term care in Germany, Safuta, Noack, Gottschall, and Rothgang (Chap. 10) show how policymakers within the political arena promote the recruitment of care workers from other countries to prevent a lack of care workers, which they describe as *state-supported migrantisation*. Finally, in his chapter on healthcare reform in Bolivia (Chap. 12), Sirén identifes what he names a *social movement–state interaction mechanism*, stressing that actors within the administration can also be comprised of people with a social movement background: Activists from social movements conquer positions in the state bureaucracy and continue to cooperate—now as public ofcials—with these movements, thus taking up certain demands.

#### **2.3 Political Party Mechanisms**

Most scholars in comparative social policy would probably agree that political parties play an important role for the development of social policy, not least given that national governments—who formulate and implement social policies—are predominantly constituted through political parties (Zohlnhöfer 2019). Classical studies focused, for example, on how left- and right-wing parties difer in their social policies, with leftwing parties being more in favour of redistribution than right-wing parties, which could then be ascribed to the diferent constituencies of both parties (Häusermann et al. 2013). Te question if parties matter for social policy development has often been studied in quantitative analyses. In these studies, the policy process is often highly simplifed. Newer research on partisan politics takes a closer look at the constituencies of diferent parties, acknowledging that links between voters and parties have changed (Häusermann et al. 2013; Gingrich and Häusermann 2015). Yet also this line of research remains very much focused on voters and their preferences and therefore to a certain extent remains centred on partisan politics as transmission belts. In contrast to these accounts, some authors have stressed that the agency of political parties—expressed through both party members and party elites—should not be neglected (Wenzelburger and Zohlnhöfer 2021).

Given this general background, two chapters in this volume identify an *outcompeting mechanism*, in which the development of social policy, most notably its introduction and expansion, is explained as a result of party competition. Political parties try to win elections or compete with one another by making promises on social policies, which could be traced for the case of pension policy in South Korea in the countries' democratisation process (Chap. 3) and for the case of unemployment insurance in Turkey (Chap. 4). While the outcompeting mechanism is focused on democratic settings, we can add the *gaining acceptance spiral mechanism* as another political party mechanism. Although the mechanism focuses on autocratic regimes, these can also be constituted by political parties who introduce or expand social policies to increase their legitimacy within the population. In this book, examples of such processes could be identifed for the introduction or expansion of pension policies in South Korea before democratisation, in Vietnam during the economic liberalisation process, and plausibly also for Malaysia in the late colonial period (Chap. 3).

Importantly, in this understanding political parties are not conceptualised as actors that aggregate voter preferences, although the mechanisms that are highlighted here focus on how political parties respond to such actual or perceived voter preferences. Rather, in our mechanism-based approach we see political parties as corporate actors who make decisions. Still, we can identify voter-oriented mechanisms as a subgroup of our political party mechanisms. In his chapter on political responses of conditional income transfer recipients, for example, Barrientos (Chap. 13) identifes a *support for redistribution mechanism,* capturing that people who receive conditional income transfers vote for the incumbent who is perceived as a politician committed to values like poverty reduction and social justice.

#### **2.4 Interest Group Mechanisms**

Interest groups are another key actor in the comparative social policy literature. Since the confict between labour and capital has been found to be crucial for explaining the introduction and expansion of social policies, trade unions (Korpi 1983) and employers' associations (Hall and Soskice 2001) are particularly important. Similar to other types of mechanisms introduced in this book, the mechanisms here are also based on the interaction between two groups of actors. On the one hand, these are interest groups or social movements, NGOs, or other civil society actors. On the other hand, these are political elites, governing parties, or other actors within the executive. For example, living and working conditions might be experienced as problematic, or demands for more social rights might arise due to fairness considerations. Consequently, demands for introducing or reforming social security systems increase in some parts of society. Interest groups, social movements, NGOs, or other associations take up and strengthen these demands (Mesa-Lago 1978; Korpi 1983). A mobilisation involving many parts of society emerges, which is expressed in demonstrations, strikes, and other forms of protest.

A key assumption of Korpi's power resource theory (1983) is that labour unions, who represent the interests of the working class, mobilise their constituencies to push their demands on social policies. Sirén (Chap. 12), in his study on healthcare reform in Bolivia, relies on this literature to identify a mechanism of *class-based mobilisation*. While the traditional power resource literature focuses especially on trade unions in the industrial sector, for the Bolivian case Sirén argues that the mobilisation of the indigenous majority through peasants' unions and neighbourhood organisations was the driver of social policy reforms. Tyen and Schlichte (Chap. 6) illuminate the role of interest groups from a diferent angle. With regard to social insurance in the case of Tunisia, they identify a mechanism of *labour incorporation*. It focuses on trade unions who had supported their country's aspirations for independence. Shortly after national independence, the working-class groups mobilised by these unions were then rewarded with a comprehensive social insurance programme. While the previous mechanisms rather focus on the role of labour, the *business-led reform mechanism* that Öktem identifes in his analysis of unemployment insurance in Turkey stresses how business actors can also convince policymakers to adopt their policy proposals (Chap. 4). In this particular case, pressure from business actors contributed to a more prominent role for active labour market policies within the system.

Finally, a mechanism that we can theorise but that was not identifed in our case studies is the mechanism of *cross-class mobilisation*. In market economies, the world of labour is predominantly or at least partly determined by wage-dependent formal employment. Tis usually results in the formation of employer and employee organisations. If more stable, informal, or institutionalised forms of cooperation between employers and the working class emerge due to a certain rational compatibility of interests, this can also lead to a welfare state coalition that jointly supports the introduction or the expansion of social policies (Hall and Soskice 2001). Employee organisations advocate for social security for their core clientele. Employer organisations might favour such policies because they protect workers, for example in cases of sickness, and therefore in the long run contribute to productivity, or because of normative convictions that value "decent" working conditions. In this coalition mechanism, however, the course of social policy expansion will be limited as the interests of the organised core groups (large companies, sectors with a high degree of unionisation) will be in the foreground (Palier and Telen 2010; Yang 2017).

#### **2.5 Political Heritage Mechanisms**

As already previously highlighted, there are several mechanisms that do not focus on a specifc actor. Some of them can be classifed as mechanisms that are in line with what is often called the political legacy approach. Especially in historical institutionalism, which pays special attention to the temporal dimensions of politics (Telen and Mahoney 2015), the concept of path dependence plays a prominent role. In a broad understanding, it simply implies that within a temporal sequence, previous sequences are relevant (Pierson 2000). Many studies take the notion of path dependence as their theoretical point of departure, stressing that social policies are shaped by previous developments. However, it is especially in the historical institutionalist tradition that the role of actors and their interactions within historical processes is detailed (e.g. Streeck and Telen 2005). Against this background, historical institutionalist approaches are highly compatible with actor-centred mechanism-based approaches (see also Chap. 1).

In our volume, the *old system departure mechanism* as identifed by Malinar (Chap. 8) can be named as an example of a mechanism that is focused on the political heritage. Policy actors perceive the old, longstanding system as so bad, delegitimised, and inefective that they will fnd any political solution better than continuing with the present system. Statements such as "Tings cannot go on like this" serve as evidence for this mechanism, which demands a turning away from the past, but does not provide a new policy solution. As Malinar shows for the case of the Croatian healthcare system, the combination of this mechanism with two other mechanisms, the actor-oriented *doctors enter politics mechanism*, and another transnational mechanism, the *seeking solutions abroad mechanism*, can be combined into the mechanism of *anti-communist backlash*, which now contains an idea about the policy solution to be adopted. Tis solution is no longer just a matter of creating something new. It must not resemble the old system—in this case the communist system—, quite on the contrary, it should be as far away from the previous system as possible. Political elites will therefore avoid any similarity between new policies and the old system because there is too much resistance to the old system, mostly for emotional reasons. Rejecting old policies enables all political actors to profle a solution that has to meet one key requirement, namely, not to have any similarity with the old regime. Tis opens considerable scope for action and creates room for fundamental institutional ruptures and the overcoming of path dependencies.

While not represented in this volume with a case study, a positive reference to historical heritage is also possible. A situation of regime or system change can also lead to the rejection of certain new ideas that will destroy what was worth preserving in the previous regime. Tus, despite the dictatorial and authoritarian constitution of a regime, in the society there can be a prevailing perception that certain areas or outcomes, such as pensions, were good and should be maintained because they provided social security. Consequently, national actors might oppose privatisation programmes in the wake of the system transformation from socialism to democratic market economies. Te new political elites are not able to implement certain radical alternatives but must follow previous paths despite the fundamental transformation of state and society. Te *honouring one's legacy mechanism* explains how an old institutional system is protected and defended against political attempts to break with this tradition.

#### **3 Policy-Specifc Mechanisms**

Te mechanisms listed in the preceding section can be relevant in all felds of social policy and for diferent institutional arrangements of the welfare state; in fact, some might even be traced in cases that are not connected to social policy at all. In addition, the following mechanisms are linked to specifc institutional arrangements of the welfare state or to specifc policy sectors within the broad spectrum of social policy. One approach to clustering this set of mechanisms is to proceed according to these institutional characteristics and to distinguish, for example, mechanisms that can only become relevant either in Bismarckian or Beveridge systems, or that can only occur in health policy, but not in labour market and old-age security policy. However, it is also possible to apply the type of clustering that we used for the general political mechanisms, that is, a combined actor-centred and theory-based clustering.

#### **3.1 Medical Profession Mechanisms**

Social policy does not only depend on actors who advocate for or introduce policies. It also depends on occupational groups who deliver social policies. Te expansion of social services leads to the growth and increased importance of specifc professional groups in the respective policy feld (Pierson 1994). Notably, throughout our edited volume, the medical profession turned out to be a particularly important group, that has been found to often have a special say when it comes to health decisions (Hassenteufel and Genieys 2021). However, precisely how the medical profession infuences social policy decisions can be very diferent.

In situations of regime or system change, the upheaval in the political elite can be so fundamental that new actors enter the political stage. Te *doctors enter politics mechanism* captures how the medical profession becomes a key political actor and can thus infuence the health system in line with its interests (see Chap. 8). If, on the other hand, the political elites and the government are dominated by other political forces, the physicians' position can also be captured in what Sirén terms the *professional autonomy mechanism* in the case of Bolivian healthcare reform (Chap. 12). Here, too, the medical profession is an important actor, but rather when it comes to defending the medical profession's privileges.

Focusing on Chile and Uruguay in the early twentieth century, González de Reufels and Huhle (Chap. 11) show how governments and the medical profession worked together in a kind of modernisation alliance to legitimise each other. By supporting the medical profession, the state becomes more modern in terms of public healthcare institutions and well-equipped hospitals, while the medical profession receives political support for the further expansion of the healthcare system. Te *legitimisation mechanism* enables the further professionalisation of the medical profession and the acceptance of physicians as political actors. Tis legitimisation is supported when the medical profession organises on a transnational level and can thus also expand its legitimacy from one country to the entire region. A second mechanism that González de Reufels and Huhle focus on in this regard is the *competitive cooperation mechanism*. Te transregional cooperation of the medical profession also drives competition, as nations compare themselves with each other and compete for the position of regional leader. At the same time, doctors and states in the region perceive a strong competition with other regions—in their case, with Europe. Taken together, this supports a form of competition that does not amount to a zero-sum game, but to an overall higher level of modernity in health policy.

#### **3.2 Social Service Provider Mechanisms**

Te literature on comparative social policy has also stressed the important role that social service providers can play for social policy (e.g. Noordegraaf and Steijn 2013). Tis becomes also relevant in some chapters of this book. Sirén in his chapter (Chap. 12) traces a *provider resistance mechanism*, focusing on the interests of the people who are employed in the Bolivian health insurance funds. On a more general level, providers are public, private for-proft, or private non-proft institutions, including not only insurance funds but also hospitals or pharmaceutical manufacturers. Tey can also perceive themselves as key actors in the health system and try to resist certain reforms that they consider unfavourable. Te *stakeholder pressure mechanism* that Safuta, Noack, Gottschall, and Rothgang identify in their chapter (Chap. 10) points in a similar direction. Here, it is private care providers who advocate a certain approach to care and migration policy to be able to maintain their business model. What is characteristic for both cases is a rational pursuit of interests, but the actor constellations are diferent in the health system than in long-term care, which can result in diferent strategies and political alliances.

#### **3.3 Status Group/Social Insurance Mechanisms**

Another group of mechanisms that we can identify refers to the characteristics of social insurance, including its focus on employees instead of the entire population, on fnancing through social contributions instead of taxes, on relative autonomy from the state instead of administrative subordination, and on a performance-based understanding of social justice (e.g. Klenk et al. 2012). One mechanism that belongs to this group is the *alarmed middle classes mechanism*, which captures how fears among groups already incorporated into the social insurance system about either needing to pay more or receiving fewer benefts leads to an only modest expansion of social policies to new groups. Kuhlmann and Nullmeier (Chap. 3) identify this mechanism in the pension reform process in South Korea. Sirén (Chap. 12) identifes the same mechanism when it comes to formal sector workers who are integrated into the health insurance funds in Bolivia.

Another mechanism, which was not identifed in a case study, can be found among those groups that have *not* yet been included in a social security system. If a social insurance scheme was initially introduced for only a limited group of employees and if this social security programme is successful, whether this group is privileged will soon become a contested issue. Demands arise that other groups, namely those with higher and lower incomes than the group that is already included in the social insurance system, should also beneft from these services or even receive a better form of social security. Other groups (e.g. white-collar workers, farmers, self-employed, fshermen) develop comparative preferences, which can be explained by the *status group competition mechanism*. Tese groups also want to be included in the social insurance scheme. Tis competition between socio-economic status groups can lead to an extension of social insurance coverage if there are no massive disadvantages (e.g. fnancial burdens) for the groups that are already included.

#### **3.4 Public Budget/Public Actor Mechanisms**

We can also identify a group of mechanisms that focuses on the fnancing of social security systems and the management of budgets in the various and often highly fragmented welfare state institutions. A striking example here is the *double beneft mechanism*, as Kuhlmann and Nullmeier (Chap. 3) show for the case of pension policy in South Korea, Vietnam, Sri Lanka, and Malaysia, in which governments make sure that there is always enough money in the pension fund because this money is used for other purposes, most importantly to fnance the public debt or to boost the country's economic development (see also Holliday 2000; Koreh 2017). By identifying a mechanism of *public resource accumulation*, Tyen and Schlichte (Chap. 6) refer to a very similar process in Uganda. In his chapter on Turkey, Öktem shows how the double beneft mechanism can also explain processes in the feld of labour market policy, as money from unemployment insurance was also used by national governments to fnance budget defcits (Chap. 4). As the case of Vietnam shows, policymakers also use money from the pension system to fulfl other social policy-related tasks, such as facilitating early retirement to ease a tight job market (Chap. 3), which can be explained by a variant of the double beneft mechanism. Moreover, the Vietnamese case also serves as an example of the *crisis management by going further mechanism*, in which policymakers expand a social insurance system to new groups so that the system remains fnancially sustainable in times of economic upheaval.

Te *fght for state funding mechanism* can be considered as a counterpart to this mechanism. It refers to a confict about the fnancing of social security systems within the public sector, as the chapter by Heinrich, Isabekova, and Pleines shows (Chap. 5). Decentralisation, fragmentation, or multi-level interdependence are the initial conditions for such fnancial conficts between diferent public actors. If a social security system is signifcantly underfunded, for example due to fxed social contribution rates at the federal level, substantial fnancial support is required, which can only be obtained directly from public budgets at the state level, regional level, or municipal level for the decentralised institutions. It then depends entirely on these authorities whether state subsidies are provided or not.

Heinrich, Isabekova, and Pleines (Chap. 5) also identify a mechanism in which an institution is established in the social security system that acts as a driving force for reforms and fnancial stabilisation. Tey describe this as the *reform supporter mechanism*. It is precisely when social groups, parties, and associations have no interest in shaping the extremely complex social policy (in their case: health policy) that the creation of a separate central public institution can appear as a decisive factor in an otherwise shunned and uncontested feld. Te public institution then becomes the motor for reform and a self-sustaining state-based reform dynamic emerges.

#### **3.5 User Mechanisms**

Finally, we focus on the clients, users, or benefciaries in social security systems who are afected by social policy at the implementation level. First of all, the role of informal practices comes to the fore here, which has been described as an important characteristic in social policy-making (see also Niedzwiecki 2018). In their analysis of mandatory health insurance in the post-Soviet region, Heinrich, Isabekova, and Pleines identify an *informalisation mechanism* (Chap. 5), which captures a special form of user behaviour that occurs in combination with doctors' economic interests when the health system is underfunded. As regular funding is insuffcient, medical personnel and patients agree on informal payments, leading to unregistered service provision. As these informal payments provide patients with additional services and medical personnel with additional income, a stable alliance emerges that unfolds an intermediate form of public and private service provision, since services are mostly provided within the framework of the public institutions. Another mechanism that can be classifed as a user mechanism is identifed by Safuta, Noack, Gottschall, and Rothgang in their analysis of long-term care in Germany (Chap. 10). Tey trace a *turn to the market mechanism*, which explains how existing structures within the policy sector, such as unregulated cash benefts to benefciaries that were established with the introduction of long-term care insurance, encourage relatives of elderly people who are in need of care to consider market solutions. In both mechanisms, a calculatory orientation is very pronounced.

### **4 Future Perspectives for a Mechanism-Based Approach in Social Policy Research**

Te clustering of mechanisms presented in this concluding chapter is an initial attempt to compile a list of complex causal mechanisms that can further stimulate research on social policy developments, especially in a comparative perspective. Yet, this can only be the frst step of a more encompassing research agenda on mechanism-based analysis in social policy. In further refning this approach to social policy, we suggest three future directions, which can mutually reinforce each other.

First, by clustering the mechanisms identifed in this volume, we have emphasised existing links to established theories of comparative social policy research. However, mechanism-based analysis should engage more profoundly with these approaches. Until now, two debates have been at the fore: Either established theories are understood as holistic approaches, and the design and orientation of the case studies that apply these theories is also entirely determined by this approach. Or these theories are interpreted as implying relevant explanatory factors that are then built into statistical models, for example in macro-quantitative research, to be able to assess their explanatory power. A mechanism-based approach as introduced in this volume now ofers a diferent option, as case studies that were based on single theoretical approaches can be subjected to a reanalysis. Tis means that existing studies can be decomposed to fnd out which causal mechanisms are included in these studies, although a mechanism-based approach was not explicitly pronounced in these studies (see also Starke 2021). Instead of assuming one unifed theory, such an approach presupposes that several mechanisms are referred to within one theoretical approach. Established approaches in social policy research can thus serve as the reference point for this frst direction of future research on causal mechanisms in comparative social policy.

Second, instead of analysing political developments over a longer period of time, research can focus on individual mechanisms. Already in the studies contained in this volume, authors were able to identify the same mechanisms in very diferent countries and policy felds, as well as at very diferent times. Such cases can thus provide an opportunity for a more detailed analysis of a particular mechanism: Is it really the same mechanism that is at play in these cases? Or do we see variations that rather lead to identifying two or more diferent mechanisms? Tese questions should be answered in follow-up analyses of already existing case studies or by bringing in new case studies based on the expectation that a particular mechanism might be relevant there. Additionally, we encourage more in-depth analysis of single complex causal mechanisms by identifying the elementary mechanisms that underlie them (see also Chaps. 8 and 9). Such an approach can rely on both theoretical work and empirical analysis and decompose the identifed complex causal mechanism step by step. Te reference point for such an approach is the elementary causal mechanisms that have been identifed in other research traditions (see Chap. 1).

#### **454 J. Kuhlmann and F. Nullmeier**

Tird, this technique of decomposition can also be applied to the analysis of individual decision-making processes in social policy. Tis presupposes that a study focuses on a limited period of time and can analyse the processes that are taking place in great detail. Such studies require a systematic application of process tracing and the methodology of qualitative social research to identify causal pathways and mechanisms (see, e.g. Beach and Pedersen 2019; Nullmeier 2021). However, not all guidelines for process tracing use the actor-centred and modular conception of mechanisms as suggested throughout this volume. Along with the precise exploration of individual processes, the tools and concepts for such micro-analyses still need to be refned.

Taken together, these three future directions of mechanism-based research can pave the way for case study-centred social policy research to develop a set of instruments that also open up broader comparative perspectives. Tis can also prevent comparative social policy research from producing fndings that barely relate to each other or using case studies to confrm or reject entire theoretical traditions. Modularity at the level of complex causal mechanisms enables case studies on diferent countries, felds of social policy, and historical periods to speak more profoundly to each other.

#### **References**


**Open Access** Tis chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/ licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence and indicate if changes were made.

Te images or other third party material in this chapter are included in the chapter's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.