## **SpringerBriefs in Economics**

**Fakhri J. Hasanov · Frederick L. Joutz · Jeyhun I. Mikayilov · Muhammad Javid**

**A Macroeconometric Modeling for Saudi Arabia** A Case Study on the World's Largest Oil Exporter

SpringerBriefs in Economics

SpringerBriefs present concise summaries of cutting-edge research and practical applications across a wide spectrum of fields. Featuring compact volumes of 50 to 125 pages, the series covers a range of content from professional to academic. Typical topics might include:


SpringerBriefs in Economics showcase emerging theory, empirical research, and practical application in microeconomics, macroeconomics, economic policy, public finance, econometrics, regional science, and related fields, from a global author community.

Briefs are characterized by fast, global electronic dissemination, standard publishing contracts, standardized manuscript preparation and formatting guidelines, and expedited production schedules.

Fakhri J. Hasanov • Frederick L. Joutz Jeyhun I. Mikayilov • Muhammad Javid

# A Macroeconometric Model for Saudi Arabia

A Case Study on the World's Largest Oil Exporter

Fakhri J. Hasanov Energy Systems and Macroeconomics King Abdullah Petroleum Studies and Research Center Riyadh, Saudi Arabia

Jeyhun I. Mikayilov Energy Systems and Macroeconomics King Abdullah Petroleum Studies and Research Center Riyadh, Saudi Arabia

Frederick L. Joutz Energy Systems and Macroeconomics King Abdullah Petroleum Studies and Research Center Riyadh, Saudi Arabia

Muhammad Javid Energy Systems and Macroeconomics King Abdullah Petroleum Studies and Research Center Riyadh, Saudi Arabia

ISSN 2191-5504 ISSN 2191-5512 (electronic) SpringerBriefs in Economics ISBN 978-3-031-12274-3 ISBN 978-3-031-12275-0 (eBook) https://doi.org/10.1007/978-3-031-12275-0

© The Author(s) 2023. This book is an open access publication.

Open Access This 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 license and indicate if changes were made.

The images or other third party material in this book are included in the book's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the book'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.

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

The 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. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

This book is dedicated to the memory of Frederick L. Joutz.

## KGEMM: A Macroeconometric Model for Saudi Arabia

#### Extended Abstract

The KAPSARC Global Energy Macroeconometric Model (KGEMM) is a policy analysis tool for examining the impacts of domestic policy measures and global changes on economic, energy, and environmental variables in Saudi Arabia at both aggregate and disaggregate levels. There are nine blocks (real sector; fiscal; monetary; external sector; domestic prices; labor and wages; energy; emission; and population and age cohorts) that interact with each other to represent the Kingdom's economic, energy, and emission linkages. KGEMM can be used to analyze the current status and future paths of aggregate and sectoral economic as well as energy and environmental indicators. It can also be used to evaluate the effects of different policy options on the Kingdom. KGEMM is a hybrid model in two main senses: it combines the data-driven approach with the theory-guided approach (New Keynesian demand-side features "anchored" to medium-run equilibrium and Neoclassical long-run supply-side representations), and it incorporates input-output table representations into the macroeconometric framework.

It has flexibility and simultaneity in evaluating multiple research and policy questions. It can also be easily customized for different research and policy questions. Unlike in many other macro-models for the Kingdom, the energy (oil, natural gas, and electricity) sector is not exogenous in KGEMM as it is interlinked to non-oil activity.

KGEMM has been validated using different validation tests, such as the statistical significance and theoretical consistency of the estimated parameters, post-estimation tests for the residuals of the estimated equations, in-sample performance for the approximation of historical data, and out-of-sample performance for policy analysis. In-sample and out-of-sample simulations show that KGEMM has robust predictive and policy analysis capabilities. Indeed, the model has been used extensively since December 2015 in multi-stakeholder projects to evaluate the macroeconomic effects of the Kingdom's domestic energy and fiscal reforms, the key initiatives in Saudi Vision 2030's Fiscal Balance Program.

This book introduces the fifth version of KGEMM and briefly describes developments from the first version till this one. It also provides a detailed survey of macroeconomic modeling in Saudi Arabia. We hope that this book will be a useful information source for researchers, scholars, and practitioners as it presents econometrically estimated elasticities and other coefficients of the behavioral relationships for economic, energy, and environmental dimensions of Saudi Arabia. We also hope that this book can further enhance cooperation and collaboration between modelers, researchers, and practitioners from various government agencies and research institutions.

Riyadh, Saudi Arabia Fakhri J. Hasanov Frederick L. Joutz Jeyhun I. Mikayilov Muhammad Javid

## Contents



#### Contents xi


## Chapter 1 Executive Summary

The objective of this book is to introduce the KAPSARC Global Energy Macroeconometric Model (KGEMM) and to conduct a detailed survey of the existing macroeconomic models for Saudi Arabia, discussing their strengths and weaknesses. KGEMM is a policy analysis tool for examining the impacts of domestic policies and changes in global economy including energy markets on the Saudi Arabian economic, energy, and environmental variables at both aggregate and disaggregate levels.

Macroeconometric models inform policymakers about the dynamic relationships among economic indicators, how economies have performed in the past, and how their current and future behaviors might differ. Models are simplifications of actual relationships that provide analytical and empirical frameworks concentrating on important aspects of the relationships. Macroeconometric models can explain, project, and evaluate economic processes. They can be especially useful in helping policymakers to evaluate the impact of alternative policy choices. In this regard, macroeconometric models provide analytical and evidence-based foundations for policy decisions, contributing to the increased likelihood of successful policies.

A macroeconometric model for Saudi Arabia is particularly important for the country's policymakers. The Kingdom faces numerous economic, demographic, and social changes as it diversifies its economy away from oil toward a greater role for the non-oil private sector and a more efficient public sector with enhanced services. The Kingdom's economy is as responsive to international energy market dynamics as any other oil exporting economy. The future of the country's economy is based on a transformation process, outlined in the Kingdom's strategic roadmap, Saudi Vision 2030 (SV2030). SV2030 involves important initiatives such as an energy price reform to induce more rational use of the Kingdom's energy resources, fiscal reforms for more effective government spending and increasing non-oil revenues, investment projects for supporting economic growth and diversification. The role that oil exports and revenues play in Saudi Arabia's fiscal policy and economy will certainly continue during the transition. Projections and assessments by international agencies suggest that global energy markets will see slower oil demand growth. Obviously, renewable transitions and environmental protection policies play a central role in this slowdown. Reforming the Kingdom's investment environment and legislation attracts more foreign direct and domestic investments. Such institutional and structural reforms could, over time, bring about changes in the country's fiscal and monetary policies and in financial markets, as highlighted in SV2030 realization programs, such as the Fiscal Sustainability Program, the National Transformation Program, and the Financial Sector Development Program. Labor market reforms are also necessary to absorb the current and coming young male and female Saudis entering the labor force. Expatriate labor's role in the workforce will change. Quantifying the effects of the above-mentioned domestic and international developments on Saudi Arabia's economic energy, and environmental dimensions requires well-designed models.

Policymakers need to know how the implementations of the above-mentioned initiatives of SV2030 coupled with global changes could influence the country's economic, energy, and environmental outlook. Macroeconometric models augmented with energy and environmental representations could provide insights about these. They could also show how the economy could be restructured and how sectoral and macroeconomic indicators should be changed to help achieve SV2030 targets.

No publicly available macroeconometric models can address the points above to the best of our knowledge. The KGEMM research project is an attempt to fill this gap and provide insights into economic-energy-environmental related policy options for the Kingdom's decision makers.

The model has nine blocks (real, fiscal, monetary, external, domestic prices, labor and wages, energy, CO2 emissions, population and age cohorts) that interact with each other to represent the Kingdom's economic (macro and sectoral), energy, and environmental linkages.

KGEMM can address two broad sets of policies and areas of research: domestic economic-energy-environmental relationships and the effects of the rest of the world, particularly global energy markets.

The following characteristics of KGEMM show its contribution to the existing literature on macroeconomic modelling in Saudi Arabia:


#### 1 Executive Summary 3


It can also contribute to economic (macro and sectoral) modeling for other open economies that are rich in natural resources, as there are similarities among these countries.

KGEMM can be used to analyze the current stance and project the future paths of economic (macro and sectoral) and energy indicators as well as CO2 emissions. The model can also be used to evaluate the effects of different policy options (e.g., SV2030 initiatives and targets). It can further enhance effective cooperation between modelers, researchers, and practitioners from various government agencies and research institutions as it represents different aspects of the economy, energy, and environment in the Kingdom.

KGEMM has been validated using different validation tests, such as statistical significance and theoretical consistency of the estimated parameters, post-estimation tests for the residuals of the estimated equations, in-sample performance for the approximation of historical data, and out-of-sample performance for policy analysis. In this book, we run the model for an in-sample forecast to approximate the historical time path of the endogenous variables as well as an out-of-sample simulations to assess the impact of Arabian light crude oil's international price, foreign direct investments inflow, and renewable deployments on Saudi Arabia's economic, energy, and environmental indicators. In-sample and out-of-sample simulations show that KGEMM has robust predictive and policy analysis capabilities. Indeed, KGEMM has been used extensively in multi-stakeholder projects to assess the macroeconomic and sectoral effects of the Kingdom's domestic energy price and fiscal reforms, the key initiatives in SV2030's Fiscal Balance Program since December 2015.

KGEMM is a hybrid model in two senses. First, it incorporates Input-Output Table representations of intermediate, final, and total demand by economic activity sector into a typical macroeconometric modeling framework. Second, it nests the theory-guided modeling approach with the data-driven approach. The KGEMM version presented in this book has a number of developments since its third version presented in Hasanov et al. (2020). The main developments are the representations of the petrochemical sector, disaggregated imports, and CO2 emissions. The latter one makes KGEMM a type of E3ME model (Energy-Environment-Economy Macro-Econometric model) and being similar to SEEEM (Sectoral Energy-Economic Econometric Model) and PANTA RHEI, the models that cover energy-economic-environmental dimensions.

Open Access This 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.

The 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.

## Chapter 2 Literature Review

The history of macroeconometric model-building is comprehensively documented in Fair (1984, 1994), Bodkin et al. (1991), Hendry and Mizon (2000), Favero (2001), Pagan (2003a, b), Bårdsen et al. (2004, 2005), Valadkhani (2004), Hendry and Muellbauer (2018), Jelić and Ravnik (2021) inter alia. Also, history and macroeconometric modeling activities over the world and their classification are documented in Welfe (2013).

This section reviews only general equilibrium macroeconomic models that have been built for the Saudi Arabian economy. In other words, we do not review either partial equilibrium models built for Saudi Arabia (e.g., see Mohaddes et al. 2020) or general equilibrium models built for other resource-rich economies. The former ones are not in line with our objectives, and the latter ones are out of the scope of this book and have been reviewed by Welfe (2013) and Hasanov and Joutz (2013) to some extent, among others. Our review here is limited to models that are publicly available or available to us.<sup>1</sup> Table 2.1 documents these models.

As the strengths and weaknesses of each model are documented in Table 2.1, we do not discuss them again here. However, it is worth mentioning that their strengths and weaknesses are also determined by their type that they belong to among other factors. In general, structural, that is theory-guided models, such as computable general equilibrium (CGE) and dynamic stochastic general equilibrium (DSGE) models have the main strengths of being strongly consistent with textbook economic theory, useful for long-term projections and analyzing the effects of changes in policy variables. The studies listed below discuss that these models have the following main weaknesses: using micro-foundations strictly as theoretical foundations and not allowing data 'to speak freely'; they do not incorporate information about behavioral economics and information economics; they are calibrated to

<sup>1</sup> Of course, we are unable to review the models that are not publicly available, including those built and used by government agencies, international institutions, academia and research centers, and private companies. We also do not review master or dissertation theses such as Tawi (1984), Taher (1987), Aljerayed (1993) and Al-Teraiki (1999).


 2.1Macroeconomic models for Saudi Arabia

Table



Table 2.1 (continued)



Table 2.1





Table 2.1 (continued)



Notes: MEM ¼ macroeconometric model, IOT ¼ input output table, NR ¼ not reported, NA ¼ not applicable, ECM ¼ error correction model, CPI ¼ consumer price index, GDP ¼ gross domestic product, OEGEM ¼ Oxford economics' global economic model, GE ¼ general equilibrium, AGE ¼ applied general equilibrium, KEM-SA ¼ KAPSARC energy model for Saudi Arabia. Calibrated models include computable general equilibrium (CGE), dynamic stochastic general equilibrium (DSGE), and hybrid, among others

Table 2.1

capture only equilibrium positions with none to limited information about short-run dynamics, and they do not provide information about the errors that they make in

their representations and simulations; they rely on many assumptions, restrictions, parametrizations that are not always true in reality (see Romer 2016; Stiglitz 2018; Blanchard 2017, 2018; Hendry and Muellbauer 2018; Wren-Lewis 2018; Fair 2019; Colander et al. 2008; Colander 2006; Hara et al. 2009; Pagan 2003a; Gürkaynak and Tille 2017; Crump et al. 2021; Wickens 1995 inter alia). Additionally, Giacomini (2015), Gürkaynak et al. (2013), among others, show that DSGEs, pure structural models produce very poor forecasting performance compared to econometric models in the empirical analyses. Moreover, Wickens (1995), Pesaran and Smith (2011), Blanchard (2017), inter alia, discuss that for DSGE models to survive in the future, they should account for data and hence switch from calibration of the deterministic relationships to estimation of the stochastic specifications, they should estimate well-specified long-run relationships rather than trend approximations and consider more dynamic short-run specifications to possibly account for habits, expectational errors, learning, and the costs and frictions of search and matching, and they should relax the assumptions made, such as optimal behavior, homogenous agent, symmetric information about market conditions, etc. Furthermore, Nikas et al. (2019, p. 37–38) discuss that standard structural models assume that markets clear in the short-run and, hence, they ignore disequilibrium and short-run relationships. For example, they usually assume that there is no unemployment in their representation of an economy. This obviously is not a relevant assumption even in the long-run and, hence, leads to drawbacks in their performances. Most likely due to the abovementioned issues, the government agencies such as central banks recently prefer hybrid type macroeconometric models, which are built using equilibrium correction equations, in their policy analyses, forecasting, and projections. Because hybrid macroeconometric models perform better than purely theory-based models (e.g., CGE, DSGE, optimal growth models) and purely data-based models (e.g., unrestricted vector autoregression models) since they are the combination of theory-guided and data-driven approaches as the literature discusses (see discussions in Ballantyne et al. 2020, Cusbert and Kendall 2018, Hendry 2018, Hendry and Muellbauer 2018, Bulligan et al. 2017; Jelić and Ravnik 2021; Giacomini 2015; Pagan 2019; Gervais and Gosselin 2014). Moreover, the behavioral representations of economic agents in the macroeconometric models are based on their historical evolution, whereas in the theory-guided models, they are usually based on the optimization of a representative agent, imposed parameters, and calibration using data from a single year or an average of years (e.g., see Lutz 2011; Lehr et al. 2012).

Jelić and Ravnik (2021) and Pagan (2019), among other studies, discuss four generations of macroeconometric models that are coexisted for the last more than 80 years and recent hybrid models incorporate the insights derived from the thirdand fourth-generation models into the second-generation models. The main strengths of the hybrid types of macroeconometric models (MEMs) over the other types of macroeconomic models are that they have theoretical coherence to represent long-run equilibrium relationships (like CGE and DSGE models and unlike VAR models). They also possess empirical coherence, i.e., they allow the data 'to speak freely' (unlike CGE and DSGE models and like VAR models) to represent short-run dynamics and disequilibrium. In other words, they bring together 'theory-guided' and 'data-driven' approaches (e.g., see Hendry 2018). They can represent the behavioral aspects of economic relationships based on the statistical time series properties of national data. Other advantages of MEMs are that they can be modified or customized to accommodate different policy questions and various simulations can be done in one model simultaneously, making them user-friendly for policy analyses. Their main weaknesses are, as mentioned in the Table 2.1, being datadependent, data updates and revision issues require a reconsideration of all the behavioral equations, require a large team for data and model maintenance and update. For detailed strengths and weaknesses of different kinds of models, interested readers can refer to the above-listed references as well as Ackerman (2002), Pagan (2003b), Hoover et al. (2008), Herbst et al. (2012), Arora (2013), Hurtado (2014), and Oxford Economics (2022).

KGEMM is a hybrid model, i.e., it combines an economic theory-guided modeling approach with empirical data-driven evidence.<sup>2</sup> This is performed through statistical estimations and testing, not by imposing theory on the model. Practically, it attempts to adjust for econometric weaknesses in earlier models built for Saudi Arabia. KGEMM also incorporates detailed demand-side representations and CO2 emissions of the main energy products by customer type. In this regard, KGEMM is a type of E3ME model (Energy-Environment-Economy Macro-Econometric model, see Econometrics, Cambridge 2019; Nikas et al. 2019; Gramkow and Anger-Kraavi 2019; Lee et al. 2018; Dagoumas and Barker 2010, inter alia). And it is similar to SEEEM (Sectoral Energy-Economic Econometric Model, see Blazejczak et al. 2014a, b) and PANTA RHEI (see Lutz et al. 2014a, b; Flaute et al. 2017; Lehr and Lutz 2016; Lehr et al. 2012; Lutz 2011), which both cover energy-economicenvironmental representations.

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

The 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.

<sup>2</sup> In the view of Pagan (2019) classification, KGEMM is a type I hybrid model, i.e., the long run paths are not articulated, leaving equilibrium correction mechanisms to ensure convergence.

## Chapter 3 Theoretical Framework and Stylized Facts

In a very broad sense, KGEMM is a demand-side macroeconometric model augmented with several supply-side representations. The fifth version of KGEMM, presented in this book, has more supply-side augmentation compared to earlier versions. Welfe (2011), among others discuss that for macroeconometric models to be used for policy analysis and projections they should represent both the demandand supply-side relationships.

The demand-side relationships are mainly represented using Keynesian and new-Keynesian schools of thought. This is true for many relationships modeled in the blocks of the model that we discuss in Chaps. 6 and 7. Therefore, we do not discuss equation-by-equation application of the theory here.

Supply-side representations are mainly in the real and price blocks of the model using Neoclassical theory. For example, modeling the supply side of the economic activities using the production function framework from the Neoclassical theory. Supply- and demand-side determinants have also been considered in modeling the price indexes for the consumer basket sub-groups, GDP deflators for sectoral economic activities, and non-oil exports. In general, the model brings together demand-side factors, such as consumption, investment, exports and imports, and supply-side factors, such as potential output, capital stock, employment, and prices.

As mentioned earlier, one of the characteristics of KGEMM is that it takes into consideration stylized facts of the Saudi Arabian economy. The stylized facts originate from a number of characteristics of Saudi Arabia, some of which are listed below:

Saudi Arabia is an oil-based economy. In 1970–2019, oil constituted on average 59% of the total economy, 79% of budget revenues, and 93% of total exports (SAMA 2020).

Saudi Arabia, like other Gulf Cooperation Council countries, has a substantial foreign labor force. Foreign nationals, particularly from East and Southeast Asian countries, account for 37% of its total population and more than 81% of the country's private sector employment (e.g., see Hasanov et al. 2021).

Historically, policymakers did not consider taxes as the main sources of fiscal revenue and, thus, they did not use taxes as key fiscal policy variables in economic adjustments (e.g., see Looney 1988).

The country is the custodian of the two holiest cities of the Islamic world, Makkah Al-Mukarramah and Al-Madīnah Al-Munawwarah. As such, there is great potential for religious tourism to be one of the main sources of the country's fiscal revenues (SV2030).

Open Access This 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.

The 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.

## Chapter 4 KGEMM Methodology

This chapter briefly describes the methodological framework KGEMM uses. KGEMM is a hybrid model, i.e., it brings together theoretical and empirical coherences at some degree. Put differently, KGEMM nests "theory-driven" and "datadriven" approaches suggested by Hendry (2018), among others, and employed by many modelers in building semi-structural, that is, hybrid macroeconometric models (e.g., see Jelić and Ravnik 2021; Gervais and Gosselin 2014; Bulligan et al. 2017). For this purpose, it uses an equilibrium correction modeling (ECM) framework, in which the long-run relationships follow economic theories, and the short-run relationships are mainly data-driven (see Pagan 2003a, b inter alia). Hara et al. (2009) and Yoshida (1990), among others, note that ECM-based MEMs provide realistic results as their equilibrium correction mechanisms help stabilize long-term projections and capture short-term fluctuations more than other models while Engle et al. (1989) find the forecast performance of ECM more accurate.

KGEMM's methodological framework for estimating the behavioral equations is based on three pillars: cointegration and ECM, the general-to-specific modeling strategy (Gets) with Autometrics, a machine-learning econometric modeling method (Ericsson 2021), and the encompassing test (Fig. 4.1). This chapter briefs the methodological framework to save space and details of it will be described later in Appendix A.

The econometric methods are employed (i) to estimate behavioral equations, which represent behavioral aspects of the economic and energy linkages of Saudi Arabia and (ii) to test the existence of relationships and hypotheses. The following is the "road map" that we use in our empirical estimations and testing.

Because we use annual time series data, the first step is to check the stochastic properties of the data using unit-root tests. For the unit-root analysis, we use the conventional tests, that is, Augmented Dickey–Fuller (ADF) (Dickey and Fuller 1981), Phillips–Perron (PP) (Phillips and Perron 1988), and Kwiatkowski et al. (1992). Additionally, we use unit root tests with structural breaks where it seems reasonable to do so based on the nature of the data. We employ the ADF with a structural break (ADFBP hereafter) developed by Perron (1989), Perron and

Fig. 4.1 KGEMM's methodological framework

Vogelsang (1992a, b), and Vogelsang and Perron (1998). We also use the Fourier ADF developed by Enders and Lee (2012a, b) and extended by Furuoka (2017), where there are multiple breaks in a given series and the conventional tests do not produce commonly accepted results. We do not describe these tests here as they are widely used in the literature. Readers interested in these tests can refer to the abovegiven references as well as Enders (2015), Perron (2006), Zivot and Andrews (1992), and Banerjee (1992).

If the variables are non-stationary, we perform cointegration tests to check whether they are cointegrated. For this purpose, we use Johansen's (1988) trace and maximum eigenvalue tests, the Pesaran's bounds test (Pesaran and Shin 1999; Pesaran et al. 2001), Engle and Granger (1987) test, Phillips–Ouliaris test (Phillips and Ouliaris 1990), and variable addition test by Park (1990). If more than two variables are involved in the analyses, which is mostly the case, then we first apply Johansen's cointegration test since it can reveal the number of cointegrating relationships if there is more than one, while the other tests above assume only one or no cointegrating relationship. We also use Hansen's (2000) cointegration test, which considers the break in the cointegration relationship. If there is a need to include level shift or trend break dummies in the Johansen cointegration test procedure, then we conduct our analysis in OxMetrics, as this software automatically calculates critical values that account for dummy variables.

If cointegration exists between the variables, then we estimate numerical parameters such as long-run coefficients. For this, we employ the following estimation methods to get robust results. Vector error correction (VEC) maximum likelihood estimation (Johansen 1988; Johansen and Juselius 1990), autoregressive distributed lags (ADL) (Hendry et al. 1984a, b; Pesaran and Shin 1999), fully modified ordinary least squares (FMOLS) (Phillips and Hansen 1990), dynamic ordinary least squares (DOLS) (Saikkonen 1992; Stock and Watson 1993), and canonical cointegration regression (CCR) (Park 1992) methods. After the estimation, we perform postestimation tests, such as residuals diagnostics for serial correlation, non-normality, heteroscedasticity, parameter stability, misspecification, and other tests where possible.

In the last part of the chain, we employ an ECM to conduct a short-run analysis, that is, estimating short-run coefficients, including the speed of adjustment. We utilize the Gets with Autometrics, following the London School of Economics, or the Hendry, modeling approach (Ericsson 2021). Gets first includes contemporaneous and lagged values of all the relevant variables, based on the related economic theory and evidence of modeler to the specification called the general unrestricted model (GUM). Then it chooses the final specification based on a range of econometric tests for diagnostics, stability, and misspecification. Further details of the Gets can be found in Davidson et al. (1978), Hendry et al. (1984a, b), Ericsson et al. (1990), de Brouwer and Ericsson (1995), and Campos et al. (2005), among others. We usually perform Gets using Autometrics in the OxMetrics software (Doornik 2009; Doornik and Hendry 2018). Autometrics is a cutting-edge machine-learning multi-block search algorithm of modern econometrics. It performs Gets automatically to select a final specification from a GUM using the tests indicated above. One of the advantages of Autometrics is that it can also account for structural breaks and other extraordinary changes observed in data using the impulse indicator saturation technique (e.g., see Doornik 2009; Hendry and Doornik 2009, 2014). Another key advantage of Autometrics is that it addresses the time invariance of the estimated coefficients and hence, super exogeneity properties of variables remained in the final ECM specification can be tested (Castle et al. 2021; Hendry and Santos 2010). These features of Autometrics allows to address the so-called Lucas critique (Ericsson and Irons 1995). It is shown that Autometrics outperforms other model selection methods, such as, Stepwise regression, the least absolute shrinkage and selection operator (LASSO), and the adaptive LASSO (e.g., see Epprecht et al. 2021; Desboulets 2018; Castle et al. 2011 inter alia).

The short-run growth equation is estimated if there is no cointegrating relationship between the variables under consideration. The procedure is the same as in the ECM analysis above, but the equilibrium correction term (ECT) is absent as the variables are not cointegrated. Gets with Autometrics is also applied to growth equations.

We use encompassing tests to compare, choose, and combine different estimated specifications for analysis and forecasting purposes. The encompassing tests compare the predictive ability of alternative specifications and select the best one. Although this is part of our econometric methodology, we have not used this test frequently, primarily because there are not enough previously estimated specifications for Saudi Arabian energy-macroeconomic relations to compare with ours. Details of the tests can be found in Mizon (1984), Mizon and Richard (1986), Harvey et al. (1998), Harvey and Newbold (2000), Ericsson (1992, 1993), Bontemps and Mizon (2008), and Clements and Hendry (2011).

As part of the KGEMM methodology, we also use various tests to validate the estimated behavioral equations and the entire model as a whole (including in-sample and out-of-sample testing for predictive ability). These include post-estimation tests for the residuals of the estimated equations, testing the statistical significance and theoretical consistency of the estimated parameters, as well as in-sample performance test and out-of-sample performance test for the entire model. Detailed discussions of these methods can be found in Fair (1984), Klein et al. (1999), Fair (2004), Bardsen and Nymoen (2008), Clements and Hendry (2011), Hendry and Mizon (2014), Beenstock et al. (1986), Calzolari and Corsi (1977), and Welfe (2011).

The KGEMM model has been built in the EViews software package, as it provides a number of advanced features for building and simulating MEMs compared to other programs. Different stages of the empirical estimations and testing are conducted in EViews and OxMetrics, which includes Autometrics. The final specifications of the equations estimated in OxMetrics are then transferred to EViews to include in the model.

Interested readers can refer to Appendix A for a detailed discussion of KGEMM's methodological framework and philosophy, including the use of Gets and Autometrics. Appendix A also discusses addressing the endogeneity issue and the Lucas critique using invariance and super exogeneity tests.

Open Access This 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.

The 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.

## Chapter 5 Database

One of the heaviest resource-consuming tasks of KGEMM, as with all MEMs, is the collection, update, revision, and maintenance of data. In econometric modeling, data are the key elements in determining the statistical properties of relationships. In this regard, data availability plays an important role in establishing linkages between the variables in time series-based MEMs. As discussed in the literature review, MEMs are heavily data-intensive, and obtaining comprehensive results is conditional upon the accuracy and time span of the data. MEMs are also data-dependent, with data updates and revisions resulting in re-estimation of the behavioral equations.

The fifth version of KGEMM has 828 annual time-series variables. In total, 397 of them are endogenous, expressed by behavioral equations and identities. There are 96 behavioral equations, and the rest endogenous variables are represented by identities. The endogenous variables are those on which we are interested in examining the impacts of other variables including domestic policy variables, as well as variables from the rest of the world. There are two main types of identities across the blocks of the model: System of National Accounting identities (e.g., total demand is the sum of private and government consumption, investments, and net exports) and definitional identities (e.g., nominal value added is obtained by multiplying real value added and the respective price deflator). The other 431 variables are exogenous in KGEMM. Many of the exogenous variables are dummy variables that capture permanent and temporary changes in relationships (that cannot be explained by the data)<sup>1</sup> and discrepancy or error terms that are used to balance relationships. The rest of the world variables, which provide a comprehensive picture of the global economic and energy ties of Saudi Arabia, are also treated as exogenous variables.<sup>2</sup> The remaining exogenous variables are policy-related variables and energy prices.

<sup>1</sup> For numerical interpretation of different types of dummy variables, see Roberto (2013), Kennedy (1981), and Halvorsen and Palmquist (1980).

<sup>2</sup> For example, the identity for the world trade index for refined oil demand variable (WTREF) contains 45 countries' demand for refined oil products and thus 45 exogenous variables (see identity # 228 in Sect. 7.4).

F. J. Hasanov et al., A Macroeconometric Model for Saudi Arabia,

SpringerBriefs in Economics, https://doi.org/10.1007/978-3-031-12275-0\_5

The data were collected from various domestic and external sources. Most of the domestic data come from the General Authority of Statistics (GaStat), formerly the Central Department of Statistics (CDSI) and the Saudi Arabian Monetary Agency (SAMA). These two sources provide a crucial portion of the country's data. Some domestic data are collected from the Ministry of Energy (MoE), and Saudi Aramco, the Ministry of Economy and Planning (MEP), the Ministry of Finance (MoF). External data mainly come from the databases of Oxford Economics Global Economic Model, the World Bank, the United Nations, the International Monetary Fund, and the International Energy Agency. The KGEMM database includes aggregated and disaggregated sector-level data. The KGEMM database contains nominal, real (usually at 2010 prices), index, ratio, and other user-calculated variables data for the real, monetary, fiscal, external, energy sectors, as well as consumer and producer prices, labor market, and population. The mnemonics and descriptions of the variables used in the fifth version of KGEMM are documented in Appendix B.

Open Access This 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.

The 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.

## Chapter 6 A Brief History and Structure of KGEMM

It is useful to provide a brief overview of the development stages of KGEMM, as these stages shaped the structure of the current version of the model. As mentioned previously, KGEMM has been developed to have a better representation of the Saudi Arabian economic (sectoral and macro) and energy relationships. The main motivation for developing it was that there was no available model (including subscription based) that properly represented the Saudi Arabian economy and could comprehensively inform the policy decision-making process.

The model presented in this book is the fifth version of KGEMM. The first version of the model was built in early 2014 by Frederick L. Joutz, Fakhri J. Hasanov, and John Qualls, researchers of the KGEMM team at KAPSARC. It was built upon the Saudi Arabian module of the Oxford Economics Global Economic Model (OEGEM). KGEMM enhanced the OEGEM's Saudi Arabian module by addressing its key limitations, including its oversimplified representation of the Saudi economy, which would prevent it from being used to comprehensively inform the policymaking process.<sup>1</sup> KGEMM differs from the Saudi Arabian module of

<sup>1</sup> The details of the oversimplification of the module were documented by the KGEMM team in 2014–2015 and are available on request. The oversimplification mainly resulted from shortcomings such as numerical unification across the sectors, its limitation to non-theoretical underpinning, and the inconsistency between its reported data and estimated parameters. This is understandable from the perspective of Oxford Economics as they serve more than 1500 clients of international corporations, financial institutions, government organizations, and universities (https://www. oxfordeconomics.com/about-us). The Saudi Arabia module is just one out of 80 countries' and 13 regional modules in OEGEM and enhancing modules require additional time and resources. The KGEMM team at KAPSARC cooperated with Oxford Economics in 2017–2018 and communicated the shortcomings of OEGEM Saudi Arabia module and the developed characteristics of KGEMM so that Oxford Economics could take them (shortcomings and developed characteristics) into account in their revision of the Saudi Arabian module. For this, Oxford Economics acknowledges KAPSARC's cooperation in their each release.

OEGEM considerably.<sup>2</sup> The main feature of the second version of KGEMM was the development of a detailed energy block representing 14 energy demand relationships by energy type and customer. The third version of KGEMM had estimated production function relationships for economic activity sectors, and thus derived sectoral output gaps feeding into consumer price index (CPI) equations for 12 household consumption basket items. It also had employment demand equations for economic activity sectors estimated as a function of output and wage (see Hasanov et al. 2021). Finally, it had more detailed external sector representations. For example, it linked the export of Saudi Arabian refinery oil to 45 individual countries' refinery oil demand, which offers the opportunity to simulate the impact of the global demand for oil and thus, environmental implementations and energy transitions on the Saudi economy. The third version of KGEMM was published as Hasanov (2020). In the fourth version of KGEMM, sectoral wage relationships were econometrically estimated as a function of the labor productivity and output price mainly. Additionally, sectoral investment relationships were developed using econometrically estimated investment demand equations, as a function of output, interest rate, and exchange rate (see Javid et al. 2022). That version also considered substitution effects in energy demand equations. Finally, the fifth version of KGEMM has a number of developments that considerably differentiates it from the previous versions. These developments include but are not limited to the following: CO2 block, which assesses carbon emissions of the energy block, that is, each of 15 energy products; more detailed external sector represented by enhanced non-oil export equation (Hasanov et al. 2022c), developed oil refinery export equation, and outflow remittances equation (Javid and Hasanov 2022); representations of the supply-side of energy, that is, fuel mix components for electricity generation as well as renewable energy represented by solar energy (Elshurafa et al. 2022; Hasanov et al. 2022a); representations of the petrochemical sector through mainly estimated output, investment, employment, energy demand, and feedstock (ethane, methane, naphtha, liquified petroleum gases) demand equations; imports of goods disaggregated into capital, intermediate, and consumer goods, each estimated econometrically as a function of domestic demand and real exchange rate. Details of the developments in the fifth version of KGEMM can be noticed in the description of each block below in this section as well as in behavioral equations and identities given in the following section. Figure 6.1 illustrates the structure of the fifth version of KGEMM.

<sup>2</sup> Since the first version of KGEMM has been built on the OEGEM's Saudi Arabia module to overcome limitations of the module, the notations of the variables in these two models are quite similar. The main points that differentiate KGEMM from the OEGEM' Saudi Arabia module are the well-established theoretical foundation and consistency between data and the estimated parameters. Moreover, KGEMM has a detailed energy block, CO2 block, a number of newly developed sectoral and aggregated relationships, and uses cutting edge econometric tools, such as Autometrics, a machine-learning econometric methodology.

Fig. 6.1 KGEMM structure

The model has nine blocks interacting with each other to represent the Saudi Arabian energy–economic–environment relationships. What follows is a brief block-by-block description of KGEMM's structure.

#### 6.1 Real Block

This block can be broadly divided into demand-side and supply-side representations.

### 6.1.1 Real Block: Demand Side

Conventional MEMs, many of them for the Saudi Arabian economy, treat demand on an aggregate level. However, as mentioned earlier, aggregate demand in KGEMM is broken down into intermediate demand, final demand, and total demand for 13 economic activity sectors. The intermediate demand is modeled as the demand of all these economic activities for each other's goods and services using coefficients derived from the input–output table for Saudi Arabia. Also, the final demand components of gross domestic product (GDP), such as private consumption, government consumption, investment, and exports are disaggregated into the economic activities using the coefficients derived from the input–output table. In addition, the investments are further broken into government, oil sector, and non-oil sector investments. While exports are represented by non-oil, oil, and service exports. The total demand for a given economic activity sector is the sum of its intermediate demand and final demand. Such a detailed framework makes the model able to distinguish intermediate, final, and total demand effects in different sectors of the economy. For example, the effects of government investment in the construction and transportation sectors are not identical, but sector-specific, each represented by its own coefficient. This allows a modeler to quantify sector-specific effects of the government policies.

Private consumption is econometrically estimated as a function of private disposable income, interest rate, and wealth using cointegration and ECM methods by Hasanov et al. (2022b) and incorporated into KGEMM. Note that the private disposable income data is not available from official agencies and hence has been constructed by the authors using the System of National Accounts framework.

Investment is the sum of oil and non-oil private and government investments as mentioned above. Non-oil private investment is the sum of domestic and foreign private investments. The latter one is the sum of foreign direct, foreign portfolio, and foreign other investments—all coming from the external block. The domestic private investments are the sum of private investment in eight non-oil economic activity sectors. Sectoral private investments are econometrically modeled as a function of sectoral output, interest rate, and exchange rate by Javid et al. (2022) over the period 1989–2017. Later, the KGEMM team updated estimations till 2019 and expanded the estimation coverage by modeling petrochemical sector investment.

The other final demand components, i.e., government consumption, exports and imports, will be discussed in their respective blocks later. These indicators link the real block to the fiscal and external blocks.

The economic activities are econometrically estimated, using the demand-side approach, where demand for a given economic activity sector is a function of total demand for this sector and the sector's demand for energy. This approach comes from the input–output framework and additionally includes energy demand. This is very similar to the modeling approach taken by Bradley et al. (1995) for the European countries. The purpose of including energy demand variable in the estimations is to measure the explicit effects of energy on economic activities. This makes it possible to use the model to analyze the impact of the domestic energy sector-related reforms on various economic activities. These estimations link the real sector to the energy and price blocks.

### 6.1.2 Real Block: Supply Side

The supply side of the real block mainly contains production functions for the economic activities, which estimate the potential outputs of the activities as functions of capital stocks and employment mainly, alongside technological change proxied by the time trend and other explanatory variables. This is consistent with the theory of production (Cobb and Douglas 1928; Douglas 1976). We use the Cobb–Douglas type as the form of the production function because it has been recommended and widely used in macroeconometric modeling (e.g., see Welfe 2011).

The capital stock for a given economic activity sector has been constructed using investment in the sector, the sector-specific depreciation rate, and assumed initial capital stock in the perpetual inventory method framework (Collins et al. 1996; Nehru and Dhareshwar 1993; Hall and Jones 1999; Arezki and Cherif 2010). The relationships representing employment for economic activity sectors, used in the production function estimations, will be discussed in the wage and labor block.

The output gaps of the economic activities have been calculated using the identities, which express the differences between the actual outputs and potential outputs coming from the estimated production functions. The gaps feed into the behavioral equations for the CPIs of the different household consumption basket items, as discussed in the price block.

Demand-side and supply-side breakdowns of the entire economy into economic activity sectors can provide useful information about sectoral compositions and changes in the structure of the economy. This is very important for diversification and local content purposes, key targets of SV2030, as articulated in the National Transformation Program (SV2030 2019b).

Section 7.1 provides additionally definitional relationships for disposable income and wealth of households, total final expenditure, and domestic demand. The sub-section also presents the nominal value added of the economic activity sectors and the aggregation of the economic activity sectors into large sectors, such as the service sector, the industry sector, the oil sector, and the non-oil sector.

#### 6.2 Fiscal Block

Total government expenditure is the sum of government capital and current expenditure in KGEMM. The former is a function of the one-period lagged capital expenditure, the relative increase in government investment, and the remainder of capital expenditure. The government current expenditure is the sum of the government's five current spending items, namely, wages, salaries, and allowances; administrative expenses; maintenance and operational costs; transfers to the private sector; and other current expenses. Government consumption is the sum of the first three current spending items above. Each of these five current spending items alongside government investment spending is econometrically estimated using government revenues as an explanatory variable.

Total government revenue is the sum of the government's oil and non-oil revenues. The former is linked to oil exports from the external block. The government's non-oil revenues are the sum of revenues from energy sales, value added tax (VAT), the expatriate levy, the Umrah and Hajj visa fees, other visa fees, tax on international trade and transactions (which is effectively customs duty fees on imports as Saudi Arabia does not apply tax or fee on exports), taxes on income, profits, and capital gains, and other non-oil revenues. The latter is the balancing item and includes collections such as traffic fines, the idle land tax, the luxury good tax, government investment returns, other non-oil sector related fees, tariffs, and collections. Energy sales are linked to the economy's total energy consumption in monetary measure coming from the energy block. VAT revenues is formed by the final consumption, VAT rate, and VAT collection efficiency ratio. The last two are policy variables for the fiscal authority. The Umrah and Hajj visa fees and other visa fees are linked to developments in the exports travel service, while the expatriate levy is tied to developments in the migrant population. The taxes on income, profits, and capital gains are linked to the changes in the non-oil sector's activity. The total government revenue in the model has been structured according to the stylized facts of the Saudi Arabian economy. This allows us to model the impacts of oil market changes, which are mostly exogenous, as well as the impacts of internal policy decisions for the energy prices reform, VAT, expatriate fees, and other revenue components—the key initiatives of the Fiscal Balance Program (FBP) of SV2030 (2019a)—on the Kingdom's economy.

Lastly, the block contains the government's budget balance: the difference between its total revenues and total expenditures. It also contains the government's non-oil budget balance: the difference between the government's non-oil revenues and its total expenditure. The latter is another stylized fact of the economy, and its purpose in the model is to show to what extent non-oil sector revenues can finance government spending. The government can then take measures to increase the efficiency of both its spending and revenue collection and fiscal consolidation the key objectives of the FBP (SV2030 2019a). The block also has general government debt, which is econometrically modeled as being dependent on government balance. General government gross debt is the sum of its past value and general government debt. Section 7.2 expresses the relationship described above.

#### 6.3 Monetary Block

The monetary block of the model is fairly small, mainly because of the nature of monetary policy in Saudi Arabia, another stylized fact of the economy. The exchange rate of the Saudi riyal (SAR) is pegged to the US dollar . Hence, Saudi Arabia's interest rate follows the US Federal Reserve interest rate and, therefore, monetary policy does not have much to do with economic growth, employment, and price stability.

The block contains definitional identities for all monetary aggregates, i.e., cash held outside banks, cash with demand deposits, a broad monetary aggregate, and a broad monetary aggregate with foreign currency deposits. It also has identities for interest rates, namely, the interest rate on lending, the effective interest rate on external debt, the interest rate on 10-year government bonds, and spread between the foreign and domestic interest rates. The block uses the definitional identity for the real exchange rate the real effective exchange rate. The latter one is determined by the ratio of domestic and foreign prices, together with the nominal effective SAR exchange rate against a basket of currencies of its main trading partners. This identity links the monetary block to the external and prices blocks. Real effective exchange rates are considered as a measurement of international price competitiveness in many theoretical and empirical studies. Therefore, we have econometrically modeled the impacts of the key theoretically predicted and Saudi-specific determinants, such as relative productivity in the oil and non-oil sectors, government consumption, net foreign assets on the real effective exchange rate, which allows us to calculate the equilibrium trajectory of the real effective exchange rate. (This equation is the slightly modified version of the equation estimated in Hasanov and Razek 2022) In this regard, the model can simulate how changes in domestic and foreign markets can shape Saudi Arabia's international competitiveness. Such simulations might be important for policy making as SV2030 targets to increase Saudi Arabia's position up to among the 15 most competitive economies over the world by 2030. Finally, the block contains another behavioral equation for broad money demand, M2 aggregate, which estimates it in real terms as a function of the interest rate differential between the domestic and foreign interest rates, GDP from the real sector, real oil price from the external sector, real effective exchange rate, and financial innovation proxied by time trend (This equation was borrowed from Hasanov et al. (2022) and slightly modified.) Section 7.3 provides details of the relationships of the monetary block.

#### 6.4 External Block

The external sector in KGEMM is classified into exports, imports, and other balance of payment components.

Total exports are the sum of exports of goods and services. The exports of goods are broken down into oil exports and non-oil goods exports as stylized facts of the Saudi economy. The former is the sum of the exports of crude oil and oil refinery products. Crude oil exports are represented by an identity, which links it to domestic oil production minus domestic oil use, multiplied by the global price of Arabian light. It is one of the most important relationships in the block, as it links the external block to the real block, the energy block, and to the rest of the world. Thus, one can simulate the model to see how energy price reform could increase crude oil export revenues by lowering domestic oil use and freeing up more oil for export. Alternatively, one can look at the impact of international oil price dynamics and the impact of OPEC oil production agreements and exports on Saudi Arabia's oil export revenues. The behavioral equation for the exports of refined oil products includes, alongside other variables, the world trade index for refined oil products. This index combines 45 individual countries' demand for refined oil products. This enables the model to simulate how changes in the demand for refined oil products in a given country caused by renewable energy transitions or environmental policies influence Saudi Arabia's refinery oil exports. The Kingdom's oil exports feed into the government's oil revenues in the fiscal block. The non-oil goods exports are estimated as being dependent on the real effective exchange rate, domestic production capacity, and demand from the rest of the world. Put differently, the equation brings together demand and supply-side factors and links the external sector to the real sector, and the monetary sector and the rest of the world. The total exports of services are the sum of the exports of services from the following nine activities: oil, investment income, other services, transportation, travel, communications, freight and insurance, financial services, and government. This detailed breakdown allows a modeler to simulate each service activity's role in total exports and in overall economic performance.

Total imports are the sum of the imports of goods and services, broken down into the same categories as total exports. The imports of goods in turn broken into three categories, imports of capital goods, imports of intermediate goods, and imports of consumer goods. Each category is econometrically modeled as a function of domestic demand and the real effective exchange rate, which reflects the nominal effective exchange rate and differences in domestic and foreign prices. The imports of services are also estimated as a function of domestic demand and the real exchange rate. Such breakdown allows a modeler to examine the role of various imported goods and services in domestic demand which may provide insights about import substitutions and local content implications being important for the diversification of the Saudi economy. The block also contains overall and non-oil trade balance both determined as identities, that is, the difference between exports and imports as well as non-oil exports and imports, respectively.

As with other balance of payment components in the model, we have identities for foreign direct investments outflows and net foreign direct investment. Outflows are linked to the development of the Saudi economy and the net is the difference between the inflows and outflows. As mentioned above, the world trade index for refined oil products is represented by an identity combining 45 individual countries' demand for refined oil products. KSA tourism demand indicator is also represented by an identity, which reflects the kingdom's tourism demand for 10 countries, including developed and developing economies. Additionally, the block has an outflow of remittances econometrically estimated as a function of domestic economic activity measured by Saudi GDP, employed foreigners, living costs in Saudi Arabia measured by domestic price level, and expatriate levy. (The equation is borrowed from Javid and Hasanov 2022 and modified slightly.) This equation can provide insights about the main determinants of the outflow remittances, which is a leakage from the Kingdom and thereby diminishes the magnitude of its fiscal multipliers. Such insights are quite important given that fiscal policy-related economic development initiatives are dominant in the Saudi economy, and the Kingdom is one of the top countries globally in terms of outflow remittances. Lastly, the KGEMM team is examining the inflow of foreign direct investment using its theoretically predicted determinants such as productivity, macroeconomic stability, openness, and business costs, including the unit labor cost, infrastructure, and institutional development. This is the work under progress and not completed yet. Section 7.4 details the external sector relationships.

#### 6.5 Domestic Prices Block

This block comprises three sub-blocks in a broader classification: consumer price indexes (CPIs), GDP deflators or producer price indexes, and aggregated energy prices. We discuss them briefly below.

The model has all 12 consumer basket prices as given by GaStat, estimated using behavioral equations. In forming the specifications for the CPIs we considered mainly supply-side factors using the markup and purchasing power parity approaches (e.g., see Brouwer and Ericsson 1998; Juselius 1992). This is because we assume that Saudi inflation is primarily cost-push inflation. The markup approach factors we considered were unit labor cost or just wages, domestic energy prices, producer prices, and VAT rate. In this framework, a modeler to simulate the effects of policy interested variables (such as VAT rate, domestic energy prices) and foreign prices on domestic consumer prices. As the Saudi economy undergoes a transformation process in line with SV2030, future versions of the KGEMM could also incorporate the money market and output gap approaches in inflation modeling. In addition to the 12 estimated equations, the sub-block contains an identity for overall CPI, which is a weighted average of the 12 CPI components. The weight variable of each of 12 components contains two values, old weights from 1970 to 2012 and new weights for 2013–2019 as documented in Appendix E.

GDP deflators, considered as the producer prices, sub-block, estimates GDP deflators for economic activity sectors, considering the respective overall price (i.e., deflators for non-oil GDP, oil GDP, and GDP) as a catch-up factor and domestic energy prices, foreign/import prices among other control variables. This setup allows a modeler to simulate the model to investigate the impact of the ongoing energy price reforms on the production costs, and thus the competitiveness of the economic activities.

The aggregated energy prices sub-block mainly contains weighted average domestic prices of energy products by customer type for several economic activity sectors, if a given sector uses more than one energy product. For example, the aggregate energy price for the utility sector is the weighted average price of natural gas, crude oil, diesel, and heavy fuel oil. The weighted average domestic energy price was calculated for the following economic activity sector as well: distribution, agriculture, financial and banking services, other services, transportation and communication, construction, and government services. Such a weighted average price allows us to simulate the model for both the price and demand effects of different energy products on a given economic activity sector under consideration. It should be noted, however, that with the exception of the utility sector (and electricity consumption to some extent), we do not have data on the consumption of energy products in the economic activity sectors. Therefore, the calculated weighted average prices for the economic activity sectors other than the utility sector are only rough approximations. A detailed description of the domestic prices block is documented in Sect. 7.5.

#### 6.6 Labor and Wages Block

This block contains relationships for employment, wages, and unit labor cost for economic activity sectors. The block also comprises definitional identities for labor force, unemployment, and its rate.

Total employment is broken into oil sector and non-oil sector employment. Employment in the oil sector is the sum of employment of oil mining and oil refinery. Employment in the non-oil sector is the sum of employment in 11 non-oil economic activities. Employment of each non-oil economic activity is econometrically estimated as a function of wage and output in a given activity sector (see Hasanov et al. 2021). We also have employment in the private sector and it is the sum of nationals and foreigners.

The wage equations for some economic activity sectors are econometrically estimated using output price and labor productivity as main determinants. However, they have not yet been completed for all sectors at the time of writing and, hence, cover seven economic activity sectors. Employment and wage equations establish links with the real block and domestic prices block.

The block contains identities that represent the unit labor costs for 12 economic activity sectors (and three aggregated sectors, that is, service, oil and non-oil). The identities use the conventional definition for unit labor cost, that is, each sector's unit labor cost is the sector-specific average wage rate multiplied by the sector-specific employment divided by the sector-specific value added. Also, the block has economy-wide labor compensation. This is the sum of labor compensation (wage rate times employment) in 11 non-oil activities and two oil activities. This variable links this block to the real block of the model as the variable feeds into the identity for disposable income. In addition, it provides the capability to simulate the model for the impact of different wage rates and employment in various sectors on households' disposable income.

Lastly, considering the stylized facts of the Saudi economy, the labor force is linked to working age population groups of Saudis and non-Saudis, both broken into the sum of males and females coming from the population and age cohorts block. This allows one to differentiate the role of each group and their male and female components in the labor force and, thus, also in the unemployment. A detailed description of the labor market and wages block is documented in Sect. 7.6.

#### 6.7 Energy Block

The energy block differentiates KGEMM from conventional (semi-)structural macroeconomic models. The block comprises demand for different energy products in volume and value (monetary) measures as well as the supply of electricity.

The block has econometrically estimated 15 behavioral equations for energy demand in volume terms. As a volume term, we use million tons of oil equivalent (MTOE) for all the energy products to make them comparable with each other. There are nine energy products (crude oil, diesel, heavy fuel oil [HFO], natural gas, electricity, liquefied petroleum gas [LPG], kerosene, gasoline, and other oil products) and six customer types (residential, industry, commercial, government, transport, agriculture and forestry). In addition to this, the first four energy products are used in the utility sector, but they are not econometrically estimated as the sector is mostly government-owned, that is, exogenous. However, not all six customers consume all 15 energy products (e.g., electricity is consumed by all customers except transport, whereas only industry and utility consume HFO and crude oil). The equations have been estimated using the conventional energy demand framework, where demand for a given energy product is a function of its own price and customer-specific income, both in real terms. For some energy products, we extended the conventional framework with other explanatory variables as deemed reasonable. For instance, we had real prices of substitutable energy products, where they became theoretically interpretable and statistically significant alongside own price and income in the estimations (see estimated equations for the industrial demand for crude oil, HFO, natural gas; for the transport demand for diesel; for the residential demand for kerosene and LPG). Also, we included cooling degree days in the residential electricity demand equation and accounted for the population effect. The latter effect was also accounted for in the gasoline demand estimation. Moreover, we had a working age group population in the industrial electricity demand estimations following the theoretical framework developed in Hasanov et al. (2021). Furthermore, we found that the investments are statistically significant and have a negative sign, indicating efficiency gains in the electricity demand equations for commercial and agricultural sectors. The customer-specific price deflators have been used to calculate energy prices in real terms, as suggested by the energy demand literature. The estimated equations mainly link this block to the domestic prices block and real block. Recall that energy consumption also feeds into the demand-side estimation of the economic activities in the real block.

The block also represents the above-mentioned 15 energy products consumption in value terms, that is, their volumes in MTOE are multiplied by their prices in Saudi riyals for per TOE. This monetary representation is for the purpose of calculating government energy sale revenues by product, customer, and by total economy. Recall that total energy consumption in monetary terms feeds into the government's non-oil revenues in the fiscal block.

The supply of electricity is broken into two generation sources: fossil fuels and renewable. Fossil fuels-based generation accounts for four main energy products used in the electricity generation in Saudi Arabia, namely, crude oil, diesel, HFO, and natural gas. The sum of these fossil fuels is multiplied by the efficiency ratio to calculate the amount of electricity generated. The second part of the electricity generation is coming from solar energy sources. We did not consider renewable energy sources other than solar, as they are very negligible historically. Even solar power generation has an average share of 0.1% of total electricity generation over the 2010–2021 period. The good news is that this share has increased significantly in recent years. Saudi Arabia plans to increase the share of renewables in total electricity generation demanded to 50% by 2030, with the other 50% coming from natural gas. This implies excluding the other fossil fuels from the power generation. Having total electricity generation coming from fossil fuels and solar sources makes KGEMM unique in simulating the economic, energy, and environmental effects of different scenarios for displacing fuels with renewables as it was done in Elshurafa et al. (2022) and Hasanov et al. (2022a). The above discussed behavioral equations and identities in addition to others are represented in Sect. 7.7.

### 6.8 CO2 Emissions Block

This is another block that makes KGEMM different from conventional (semi-) structural macroeconometric models. The literature on environmental pollution shows that usually, energy-related CO2 emissions are around 90% of the total CO2 emissions. Therefore, this block was constructed using the energy block above. It is calculated as volume of a given energy product consumed (e.g., crude oil) multiplied by the product-specific conversion factor to reach up to the amount of CO2 emissions. We used each product-specific emission conversion factor in the calculation. The conversion factors are retrieved from various reputed sources such as the International Energy Agency, US Energy Information Administration, US Environmental Protection Agency. Appendix E documents conversion factors and their sources while Sect. 7.8 reports calculations, i.e., constructed identities. We grouped product-based CO2 emissions by customer type—industry, transport, residential, commercial, government, agriculture, but other classifications can be considered as well. As mentioned at the beginning of this section, CO2 emissions are one of the new developments in the fifth version of KGEMM meaning that it has room for extension. We plan to expand the block by incorporating other emission indicators into it. For example, a carbon price or carbon tax is one of the considerations for future work. The block allows a modeler to assess the impacts of different CO2 emissions reduction options on energy consumption and economic indicators.

#### 6.9 Population and Age Cohorts Block

The total population in the block is the sum of the 12 age groups: 0–14, 15–19, 20–24, 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65 and above. Each age cohort is determined as the sum of males and females. The block represents the total Saudi population as the sum of Saudi males and females. The same formula is applied to the non-Saudi population. The working age population is the sum of males and females. Both male and female working age population groups are the sum of age cohorts 15–19 through 60–64. The population of Saudis and non-Saudis feeds into the labor force in the wages and labor block as mentioned above. One can simulate the model to assess the impacts of changes in each of the 12 age cohorts broken into females and males on the on various relationships and the total economy, as well as the Saudi and non-Saudi impacts on the labor force and unemployment. Details of the block can be found in Sect. 7.9

Open Access This 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.

The 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.

## Chapter 7 KGEMM Behavioral Equations and Identities

This chapter reports the estimated long-run equations and identities in Sects. 7.1, 7.2, 7.3, 7.4, 7.5, 7.6, 7.7, 7.8, and 7.9 while the estimated short-run equations, i.e., final ECM specifications associated with the long-run equations are reported in Appendix B to save space in the main text.1 Note that the long-run and ECM equations are estimated till 2019 in the fifth version of KGEMM. Starting years of the estimations range from the 1970s to the 1990s dictated by the data availability. For the readers ease, we describe one of the long-run equations below and the rest equations here follow the same context. As an example, we select the first appeared long-run equation, i.e., Eq. (7.43).

LOGðIFDISÞ ¼ 1:00- LOG ðGVADISÞ 0:03-RRLEND1 2:33 þ ECT IFDIS

Where LOG indicates the natural logarithmic transformation of a given variable. IFDIS, GVADIS, RRLEND1 are the variables (see Appendix B for the definitions and notations of the variables). ECT\_IFDIS is the residuals of the long-run equation of the retail, wholesale, hotels, and catering sector's investment, i.e., equilibrium correction term. In general, ECT\_X denotes the equilibrium correction term for the variable X (Its one period lagged series enters the short run, i.e., ECM equation of the variable X as reported in Appendix B).

The long-run equation shows that private investment in the retail, wholesale, hotels, and catering sector (IFDIS) increases by 1% when the sector's output (GVADIS) increases by 1%, while it decreases by 3% if the real lending rate (RRLEND1) rises by 1 percentage point, holding other factors constant. A detailed

<sup>1</sup> The long-run and short-run relationships among the variables are estimated using the cointegration and ECM frameworks, respectively. Therefore, there are two versions of the model: the long- and short-run versions. The former one uses the estimated long-run (cointegrated) equations (e.g., such as Musila, 2002; Fair 1979, 1993), while the latter one uses the estimated final ECM equations (e.g., such as Buenafe and Reyes 2001; Welfe 2013) alongside identities. A detailed description of both, short-run and long-run, versions of the model are available from the authors upon request.

discussion of this and other estimated long- and short-run equations of investments for economic activity sectors can be found in Javid et al. (2022).

To save space, we do not report the significance levels of the estimated coefficients as well as pre-estimation (e.g., unit root and cointegration) and post-estimation (e.g., diagnostics, stability) test results for the long- and short-run equations because they number 192. They are available from the authors on request. Note, however, that all the estimated coefficients (reported below in the tables and in Appendix B) are statistically significant and the cointegration tests indicate long-run relationships among the variables at different conventional significance levels.

### 7.1 Real Block Equations and Identities (Demand-Side and Supply-Side)

7.1.1 Demand-Side

#### 7.1.1.1 Identities for Intermediate, Final and Total Demand by Economic Activity Sector

Intermediate Demand

$$\begin{aligned} \text{IDAGR} &= \text{Al}^{\circ} \text{GVAOL} + \text{A2}^{\circ} \text{GVAMINOH} + \text{A3}^{\circ} \text{GVAEPCH} \\ &+ \text{A4}^{\circ} \text{GVAMINNOL} + \text{A5}^{\circ} \text{GVAO} + \text{A6}^{\circ} \text{GVAONON} \\ &+ \text{A7}^{\circ} \text{GVAON} + \text{A8}^{\circ} (\text{GVATRACOM} - \text{GVATRAPPE}) \\ &+ \text{A9}^{\circ} \text{GVAAGR} + \text{A10}^{\circ} \text{GVAIFBU} + \text{A11}^{\circ} \text{GVAOV} \\ &+ \text{A12}^{\circ} \text{GVAOTHS} + \text{D5}\_{\text{D6AG}} \\ \text{IDCNON} &= \text{BI}^{\circ} \text{GVAOL} + \text{B2}^{\circ} \text{GVAONITH} + \text{B3}^{\circ} \text{GVAPICH} \\ &+ \text{B44}^{\circ} \text{GVAONLOI} + \text{B58}^{\circ} (\text{GVAPI+BOC} - \text{GVATRAPPE}) \\ &+ \text{B97}^{\circ} \text{GVAOSR} + \text{B84}^{\circ} (\text{GVAPIB} - \text{GVATRAPPE}) \\ &+ \text{B12}^{\circ} \text{GVAOSR} + \text{B13}^{\circ} \text{GVAPIB} + \text{B111}^{\circ} \text{GVAOF} \\ \text{IDISIS} &= \text{Cl}^{\circ} \text{GVAOL} + \text{C2}^{\circ} \text{GVAAMNOT} + \text{C3}^{\circ} \text{GVAONE} \\ &+ \text{C4}^{\circ} \text{GVAONL} + \text{C8}^{\circ} \text{GVATACOM} - \text{GVATORCOV}$$


Þ Þ Þ Þ þ þ ð Þ IDOILREF¼ I1- GVAAGRþI2- GVACONþI3- GVADIS þ I4- GVAFIBUþI5- GVAFIBUOTHþI6- GVAREAL þ I7- GVAMANNOþI8- GVAMINOTHþI9- GVAOILMIN þ I10- GVAOILREFþI11- GVAOTHSþI12- GVAGOV þ I13- GVATRACOMþI14- GVAU ð7:9Þ IDOTHS ¼ J1- GVAOIL þ J2- GVAMINOTH þ J3- GVAPETCH þ J4- GVAMANNOLPC þ J5- GVAU þ J6- GVACON þ J7- GVADIS þ J8- ðGVATRACOM GVATRAPIPE þ J9- GVAAGR þ J10- GVAFIBU þ J11- GVAGOV þ J12- GVAOTHS þ DIS IDOTHS ð7:10Þ IDPETCH ¼ K1- GVAOIL þ K2- GVAMINOTH þ K3- GVAPETCH þ K4- GVAMANNOLPC þ K5- GVAU þ K6- GVACON þ K7- GVADIS þ K8- ðGVATRACOM GVATRAPIPE þ K9- GVAAGR þ K10- GVAFIBU þ K11- GVAGOV þ K12- GVAOTHS þ DIS IDPETCH ð7:11Þ IDTRACOM¼ L1- GVAOILþL2- GVAMINOTHþL3- GVAPETCH þ L4- GVAMANNOLPCþL5- GVAUþL6- GVACON þ L7- GVADISþL8- ðGVATRACOMGVATRAPIPE þ L9- GVAAGRþL10- GVAFIBUþL11- GVAGOV þ L12- GVAOTHSþDIS IDTRACOM ð7:12Þ IDU ¼ M1- GVAOIL þ M2- GVAMINOTH þ M3- GVAPETCH þ M4- GVAMANNOLPC þ M5- GVAU þ M6- GVACON þ M7- GVADIS þ M8- ðGVATRACOM GVATRAPIPE þ M9- GVAAGR þ M10- GVAFIBU þ M11- GVAGOV M12-GVAOTHS DIS IDU 7:13

Here, the coefficients from A1 to M12 are Input-Output coefficients. We do not report numerical values of them due to the data confidentiality issue as they obtained from OEGEM.

Final Demand

þ þ þ þ ð Þ þ þ þ þ ð Þ þ þ þ þ þ þ þ þ ð Þ ð Þ ¼ þ þ þ þ ð Þ ¼ þ þ þ þ ð Þ þ ð Þ FDAGR¼ N1- CONSþN2- GCþN3- IFOILþN4- IFNOILPþN5- GI þ N6- XGOILþN7- XGNOILþN8- XSþDIS FDAGR ð7:14Þ FDCON¼ O1- CONSþO2- GCþO3- IFOILþO4- IFNOILPþO5- GI O6- XGOIL O7 XGNOIL O8- XS DIS FDCON 7:15 FDDIS ¼ P1- CONS þ P2- GC þ P3- IFOIL þ P4- IFNOILP þ P5- GI P6- XGOIL P7- XGNOIL P8- XS DIS FDDIS 7:16 FDFIBU¼ Q1- CONSþQ2- GCþQ3- IFOILþQ4- IFNOILPþQ5- GI Q6- XGOIL Q7 XGNOIL Q8- XS DIS FDFIBU ð7:17Þ FDGOV¼ R1- CONSþR2- GCþR3- IFOILþR4- IFNOILPþR5- GI R6- XGOIL R7- XGNOIL R8- XS DIS FDGOV 7:18 FDMANNOLPC¼ S1- CONSþS2- GCþS3- IFOILþS4- IFNOILP þ S5- GIþS6- XGOILþS7- XGNOILþS8- XS þ DIS FDMANNOLPC 7:19 FDMINOTH T1- CONS T2- GC T3- IF T4- X T5- IS 7:20 FDOIL¼ U1- CONSþU2- GCþU3- IFOILþU4- IFNOILPþU5- GI þ U6- XGOILþU7- XGNOILþU8- XS ð7:21Þ FDOILREF V1- CONS V2- GC V3- IF V4- X V5- IS 7:22 FDOTHS ¼ W1- CONS þ W2- GC þ W3- IFOIL þ W4- IFNOILP þ W5- GI þ W6- XGOIL þ W7- XGNOIL þ W8- XS DIS FDOTHS 7:23 FDPETCH ¼ X1- CONS þ X2- GC þ X3- IFOIL þ X4- IFNOILP þ X5- GI þ X6- XGOIL þ X7 - XGNOIL þ X8- XS þ DIS FDPETCH ð7:24Þ FDTRACOM ¼ Y1- CONS þ Y2- GC þ Y3- IFOIL þ Y4- IFNOILP þ Y5- GI þ Y6- XGOIL þ Y7- XGNOIL þ Y8- XS þ DIS FDTRACOM ð7:25Þ

$$\text{FDU} = \text{Z1}^\* \text{CONS} + \text{Z2}^\* \text{GC} + \text{Z3}^\* \text{IFOIL} + \text{Z4}^\* \text{IFNOILP} + \text{Z5}^\* \text{GI}$$

$$+ \text{Z6}^\* \text{XGOIL} + \text{Z7}^\* \text{XGNOL} + \text{Z8}^\* \text{XS} + \text{DIS} \\_\text{FDU} \tag{7.26}$$

Here, the coefficients from N1 to Z8 are Input-Output coefficients. We do not report numerical values of them due to the data confidentiality issue as they obtained from OEGEM.

#### Total Demand


#### 7.1.1.2 Identities for Total Final Expenditure and Domestic Demand

$$\text{TFE} = \text{CONS} + \text{IF} + \text{GC} + \text{IS} + \text{X} \tag{7.40}$$

$$\text{DOMD} = \text{CONS} + \text{IF} + \text{IS} + \text{GC} \tag{7.41}$$

$$\text{DOMD\\_Z} = \text{CONS\\_Z} + \text{IF\\_Z} + \text{IS\\_Z} + \text{GC\\_Z} \tag{7.42}$$

Equations for Investments by Economic Activity Sector

$$\begin{array}{l} \text{LOG (IFDIS)} = 1.00^{\circ} \text{ LOG (GVADIS)} - 0.03^{\circ} \text{ RRLEND1} - 2.33\\ + \text{ECT\\_IFDIS} \end{array} \tag{7.43}$$

$$\begin{array}{l} \text{LOG (IFCON)} = 0.82^{\circ} \text{ LOG (GVAONN)} - 0.02^{\circ} \text{ RRLEND1} + 0.30\\ + \text{ECT\\_IFCON} \end{array} \tag{7.44}$$

$$\text{LOG (IFFIBU)} = 2.32^{\circ}\text{LOG}(\text{GVAFIBU}) + 0.94^{\circ}\text{ LOG (REER)} - 21.98$$

$$-0.54^{\circ}\text{ DSH2010} + \text{ECT\\_IFFIBU} \tag{7.45}$$

LOGðIFMANNOLPCÞ ¼ 0:69- LOGðGVAMANNOLPCÞ0:03- RRLEND1 þ 0:86-LOG RER ð Þþ1:40þECT IFMANNOLPC

$$\text{LOG} \left( \text{IFOTHS} \right) = 2.65^{\circ} \text{LOG} \left( \text{GVAOTHS} \right) - 0.13^{\circ} \text{RRLEND1} \tag{7.46}$$

$$-2.75^{\circ} \text{LOG} \left( \text{RER} \right) - 14.26 + \text{ECT\\_IFOTHS} \tag{7.47}$$

$$\text{LOG (IFPETCH)} = 2.72^{\circ}\text{LOG (GVAPETCH)} + 2.24^{\circ}\text{ LOG (RER)}$$

$$-0.13^{\circ}\text{ RRLEND1} - 20.72 + \text{ECT }\underline{\text{IFPETCH}}\tag{7.48}$$

$$\begin{array}{l} \text{LOG (IFTRACOX)} = 0.81^{\circ} \text{LOG (GVATRACOM)} - 0.06^{\circ} \text{ RRLEND} \\ + 1.09 + \text{ECT } \underline{\text{IFTRACOM}} \end{array} \tag{7.49}$$

$$\begin{array}{l} \text{LOG (IFU)} = 0.79^{\circ} \text{LOG} (\text{GVAU}) - 0.05^{\circ} \text{RRLENDI} - 0.37^{\circ} \text{LOG} (\text{RER})\\ + 2.54 + \text{ECT\\_IFU} \end{array} \tag{7.50}$$

$$\text{LOG (IFAGN)} = 3.25^\circ \text{LOG(GVAGR)} + 3.26^\circ \text{LOG (RER)} - 33.17$$

$$+ 2.54^\circ \text{DST1012} + \text{ECT } \text{IFAGB} \tag{7.51}$$

#### 7.1.1.3 Identities for Investments

$$\text{IF} = \text{IFOIL} + \text{IFNOIL} + \text{DIS} \,\text{IF} \tag{7.52}$$

$$\text{IF } \underline{\mathbf{Z}} = \text{IF}^\* \text{ PIF}/100 \tag{7.53}$$

$$\text{IFNOIL} = \text{IFNOILP} + \text{GI} + \text{ISP} \tag{7.54}$$

IFNOILP¼ IFDOMPþ

$$100^\* \left( \left( \text{(FI\\$IN\\_Z^\* \\$XD)} \right) / \left( \text{WPMF\\$\\_WLD} / \text{111.4992}^\* \, 100 \right) \right)$$
 
$$+ \text{DIS\\_IFNOILP} \tag{7.55}$$

$$\text{FISIN\\_Z} = \text{FDI\'lIN\\_Z} + \text{FPI\'lIN\\_Z} + \text{FOI\'lIN\\_Z} \tag{7.56}$$

$$\text{IFDOMP} = \text{IFAGR} + \text{IFCON} + \text{IFDIS} + \text{IFFIBU} + \text{IFMANNO}$$

$$\text{+ IFMINOTH} + \text{IFOTHS} + \text{IFTRACOX} + \text{IFU}$$

$$+\text{DIS\\_IFDM}\tag{7.57}$$

$$\text{IFMANNO} = \text{IFMANNOLPC} + \text{IFPETCH} \tag{7.58}$$

$$\text{IFNOIL\\_Z} = \text{IFNOILP\\_Z} + \text{GI\\_Z} \tag{7.59}$$

$$\text{GI} = \text{GI\\_Z}/\text{PIF}^\*100 + \text{DIS\\_GI} \tag{7.60}$$

Equations for Gross Value Added by Economic Activity Sector

$$\begin{aligned} \text{LGG}(\text{GVAGR}) &= 0.14^{\circ}\text{LGG}(\text{DELL-AGR}) + 0.07^{\circ}\text{LGG}(\text{TDAR}) \\ &+ 0.08^{\circ}\text{LGG}(\text{DIDL-NID}) + 9.35 + 0.01^{\circ}\text{sr}\text{D720D} \\ &- 0.07^{\circ}\text{D720D} - 0.11^{\circ}\text{D720D} + 0.01^{\circ}\text{D720D} \\ &+ \text{ECT }\text{GVAGR} \\ \text{LGG}(\text{GVAON}) &= 0.28^{\circ}\text{LGG}(\text{TIDCON}) + 0.63^{\circ}\text{LGG}(\text{DELL-COMM}) \\ &+ \text{DEL-TOT-TRA}) + 5.68 + \text{ECT }\text{GVAON} \\ \text{LGG}(\text{GVABM}) &= 1.22^{\circ}\text{LGG}(\text{DIDL-S}) + 0.18^{\circ}\text{LGG}(\text{DELL-COMM}) \\ &- 3.07 + \text{ECT }\text{GVABM} \\ \text{LGG}(\text{GVAFHB}) &= 0.16^{\circ}\text{LGG}(\text{DELL-COMM}) + 0.71^{\circ}\text{LGG}(\text{TBFBIB}) \\ &+ 3.57 + \text{ECT }\text{GVAFHB} \end{aligned} \tag{7.64}$$

$$\begin{aligned} \text{LGG}(\text{GVAGW}) &= 0.25^{\circ}\text{LGG}(\text{TIDCO}) + 0.29^{\circ}\text{LGG}(\text{DELL-GON}) + 0.13^{\circ} \\ &- 3.07 + \text{LGG}(\text{TOL} + \text{TOL}) + 0.25^{\circ}\text{LGG}(\text{DELL-GON}) + 0.13^{\circ} \\ &- 3.07 + \text{LGG}$$

$$\begin{array}{c} \text{LOG}(\text{GVAOILMIN}) = 0.98^\* \text{LOG}(\text{OILMBD}) + 11.44 + 0.002^\*@ \text{TREND} \\ + \text{ECT } \underline{\text{GVAOILMIN}} \end{array} \tag{7.67}$$

$$\text{LOG} \left( \text{GVAOILREF} \right) = 1.01^\* \text{LOG} \left( \text{TDOL} \right) + 0.56^\* \text{LOG} \left( \text{OILUSE}^\* \ \text{365}^\* \right)$$

$$0.1486 - \text{DCOIL\\_U}) - 0.36^\* \text{LOG} \, (\text{DNGA\\_IND})$$

$$\text{+ DNGA\\_IND\\_NEU} + \text{DNGA\\_EOU} \left( \text{-4.41} \right)$$

$$-0.21^\* \text{DP2013} + \text{ECT\\_GVAOILREF} \tag{7.68}$$

$$\text{LOG(GVAOTHS)} = 0.47^{\circ} \text{LOG(TDOTHS)} + 0.19^{\circ} \text{LOG(DELE\\_COOH)}$$

$$+5.43 + 0.03^\* \text{DBT2015} + \text{ECT\\_GVAOTHS} \tag{7.69}$$

LOGðGVAPETCHÞ ¼ 0:31-LOGðTDPETCHÞ

$$+0.54^\* \text{ LOG} (\text{DETH}\_{-} \text{IND}\_{-} \text{NEU})$$

$$+ \text{DLPG\\_IND\\_NEU} + \text{DNAP\\_IND\\_NEU}$$

$$\text{+ DNGA\\_IND\\_NEU}) + 0.38^\* \text{ LOG}(\text{DELE\\_IND})$$

$$1 + 4.70 - 0.20^\* \text{ DSH2008} + \text{ECT\\_GVAPETCH} \qquad (7.70)$$

Þ LOG DETH IND NEU ð Þ¼ 0:14- LOG PETH IND NEU=PGDPPETCHð 100

$$+0.71^\* \text{LOG}(\text{GVAPETCH}) - 4.19^\*$$

$$-0.21^\* \text{ DSH000102} - 0.19^\* \text{ DP2008}$$

þ ECT DETH IND NEU ð7:71Þ

Þ LOG DLPG IND ð Þ NEU ¼ 0:14- LOG PLPG IND NEU=PGDPPETCHð 100

$$+0.53^\* \text{LOG}(\text{GVAPETCH}) - 2.93^\*$$

þ 0:17- DTB95010:47-DSH2003

$$+\text{ECT\\_DLPG\\_IND\\_NEU} \tag{7.72}$$

Þ LOG Dð Þ NAP IND NEU ¼ 0:31- LOG PNAP IND NEU=PGDPPETCHð 100

$$+3.47^\* \text{LOG}(\text{GVAETCH}) - 24.81^\*$$

$$-0.14^\* @ \text{TREND} - 1.35^\* \text{DP1991}$$

$$+\text{ECT\\_DNAP\\_IND\\_NEU} \tag{7.73}$$

$$\text{LOG(DNGA\\_IND\\_NEU)} = -0.08^\* \, \text{LOG(PNGA\\_IND\\_NEU/PGDPPETCH}^\* \, 100)$$

$$+0.3\mathbf{\hat{S}}^{\ast}\mathbf{L}\mathbf{O}\mathbf{G}(\mathbf{G}\mathbf{V}\mathbf{A}\mathbf{P}\mathbf{E}\mathbf{T}\mathbf{C}\mathbf{H}) - 1.76$$

$$+0.06^\* \text{ TI} 2009 + 0.12^\* \text{ S1} 2001$$

þ ECT DNGA IND NEU ð7:74Þ

$$\text{LOG}(\text{GVATRACOM}) = 1.40^{\circ} \text{LOG}(\text{TOTRACOM})$$

$$+ 0.38^{\circ} \text{LOG}(\text{DELE\\_COMM})$$

$$+ 0.65^{\circ} \text{LOG}(\text{DGAS\\_TRA})$$

$$+ 0.14^{\circ} \text{LOG}(\text{DEOTH\\_TRA}) - 4.21$$

$$- 0.06^{\circ} \text{@TREND} + 0.07^{\circ} \text{DBT2016}$$

$$+ \text{ECT\\_GVATRACOM} \tag{7.75}$$

$$\text{LOG}(\text{GVAU}) = 0.14^{\circ} \text{LOG}(\text{TDU}) + 0.46^{\circ} \text{LOG}(\text{DNGA\\_U})$$

$$+0.17^\* \text{LOG}(\text{DCOIL\\_U}) + 0.49^\* \text{LOG}(\text{DDIS\\_U} + \text{DHFO\\_U})$$

$$+5.16-0.07^\* \text{DP2008} + \text{ECT\\_GVAU} \tag{7.76}$$

#### 7.1.1.4 Identities for Gross Value Added

Sectoral Aggregations


#### Value Added in Nominal Terms by Economic Activity Sector


PGDPFISIM=100 ð7:103Þ

GVAFISIM Z ¼ GVAFISIM-


Disposable Income, Private Consumption and Wealth

$$\begin{aligned} \text{DI\\_T\\_Z} &= \text{NNSA\\_Z} + \text{CONS\\_Z} + \text{GC\\_Z} & (7.108) \\\\ \text{DI\\_Z} &= \text{LABCOMP} + \text{GCGPE} - \text{REMF} + \text{DIS\\_DI\\_Z} & (7.109) \\\\ \text{DI} &= \text{DI\\_Z/CP\'100} & (7.110) \\\\ \text{LOG(CONS)} &= 1.00^\circ \text{LOG(DI)} - 0.03^\circ (\text{RCB} - \text{@PCH}(\text{CPI}) \* 100) \\ & - 0.54^\circ \text{LOG} (\text{WEALTH}) + 5.97 + \text{ECT\\_CONS} & (7.111) \\\\ \text{PCONS} &= \text{CONS\\_Z/CONS\'100} & (7.112) \end{aligned}$$

$$\text{UEALTH} = ((\text{M3} - \text{M0}) - \text{LLABP}) / \text{CPI}^\* \ 100\tag{7.113}$$

#### 7.1.2 Supply-Side

#### 7.1.2.1 Identities for Capital Stocks by Economic Activity Sector

$$\text{CAPAGR} = \text{CAPAGR} (-1)^{\*} \, 0.95 + \text{IFAGR} + \text{DIS\\_CAPAGR} \tag{7.114}$$

$$\text{CAPCON} = \text{CAPCON}(-1)^{\*} \, 0.95 + \text{IFCON} + \text{DIS\\_CAPCON} \qquad (7.115)$$

$$\text{CAPDIIS} = \text{CAPDIIS}(-1)^{\*}\ 0.95 + \text{IFDIIS} + \text{DIS\\_CAPDIIS} \tag{7.116}$$

$$\text{CAPFIBU} = \text{CAPFIBU}(-1)^{\circ}.0.95 + \text{IFFIBU} + \text{DIS\\_CAPFIBU} \qquad (7.117)$$

$$\text{CAPGOV} = \text{CAPGOV}(-1)^{\ast} \, 0.95 + \text{GI} + \text{DIS\\_CAPGOV} \tag{7.118}$$

$$\begin{aligned} \text{CAPMANNOLPC} &= \text{CAPMANNOLPC} (-1)^{\ast} \ 0.95 + \text{IFAMANOLPC} \\ &+ \text{DIS\\_CAPMANNOLPC} \end{aligned} \tag{7.119}$$

$$\text{CAPNOIL} = \text{CAPNOIL} (-1)^{\text{\textquotedblleft}} 0.95 + \text{IFNOIL} + \text{DIS\\_CAPNOIL} \qquad (7.120)$$

$$\text{CAPOILREF} = \text{CAPOILREF}(-1)^{\circ} \ 0.95 + \text{IFOIL} + \text{DIS\\_CAPOILREF} \ (7.121)$$

$$\text{CAPOTHS} = \text{CAPOTHS}(-1)^{\*}.0.95 + \text{IFOTHS} + \text{DIS\\_CAPOTHS} \quad (7.122)$$

$$\begin{array}{l} \text{CAPPETCH} = \text{CAPPETCH} (-1)^{\*} \ 0.788 + \text{IFPETCH} \\ + \text{DIS\\_CAPPETCH} \end{array} \tag{7.123}$$

$$\text{CAPU} = \text{CAPU}(-1)^{\*}\ 0.95 + \text{IFU} + \text{DIS\\_CAPU} \tag{7.124}$$

$$\text{KOLREF} = \text{IFREF} + \text{KOLREF} (-1)^{\text{\textquotedblleft}} 0.9500000 + \text{DIS\\_KOLREF} \quad (7.125)$$

#### 7.1.2.2 Identities for the Estimated Potential Output Equations by Economic Activity Sector

$$\text{POT\\_GVAGR} = \text{EXP} \left( 0.21^\circ \text{ LOG} (\text{CAPAGR}) + 0.19^\circ \text{ LOG} (\text{ETAGR}) \right)$$

$$+ 0.56^\circ \text{ LOG} (\text{AGRLANDSH}) + 4.82$$

$$+ 0.003^\circ \text{ DSH2010} (\text{°REND}) \text{ } \tag{7.126}$$

$$\text{POT\\_GVACON} = \text{EXP} \left( 0.28^\* \text{ LOG} (\text{CAPON}) + 0.72^\* \text{ LOG} (\text{ETCON}) \right)$$

$$+ 3.09 + 0.10^\* \text{ DP2011} \right) \tag{7}$$

$$+3.09 + 0.10^{\circ} \text{ DP2011} \Big) \tag{7.127}$$

$$\text{POT\\_GVADIS} = \text{EXP} \Big( 0.73^{\circ} \text{ LOG(CAPDIS)} + 0.56^{\circ} \text{ LOG(ETDIS)} - 0.98 \Big)$$

$$\left(-0.14^{\circ} \text{ DP2003} - 0.07^{\circ} \text{ DP2011}\right) \tag{7.128}$$

$$\text{POT\\_GVAFIBU} = \text{EXP} \left( 0.36^\circ \text{LOG(CAPFIBU)} + 0.51^\circ \text{LOG(ETFIBU)} \right)$$

$$\left(+4.96 + 0.41^\* \text{ DP2013}\right)\tag{7.129}$$

$$\text{POT\\_GVAGOV} = \text{EXP} \left( 0.09^{\text{\*}} \text{ LOG} \left( \text{CAPGOV} \right) \right)$$

$$+ 0.66^{\text{\*}} \text{ LOG} \left( \text{ETGOV} \right) + 6.74 \right) \tag{7.130}$$

$$\text{POT\\_GVAMANNOLPC} = \text{EXP} \left( 0.82^\circ \text{LOG} (\text{CAPMANNOLPC}) \right) \qquad (7.131)$$

$$+ 0.46^\circ \text{LOG} (\text{ETMANNO}) - 0.24 \right)$$

$$\text{POT\\_GVANOIL} = \text{EXP}(0.40^\* \text{ LOG}(\text{CAPNOIL}) + 0.99^\* \text{ LOG}(\text{ETNOIL}) + 2.51) \tag{7.132}$$

$$\text{POT\\_GVAOILREF} = \text{EXP}\left(0.86^\* \text{ LOG}(\text{CAPOHILEF})$$

$$+ 0.44^\* \text{ LOG}(\text{ETOILREF}) + 1.23\right) \tag{7.133}$$

$$\text{POT\\_GVAOTHS} = \text{EXP}\left(0.11^\* \text{ LOG}(\text{CAPOHHS}) + 0.71^\* \text{ LOG}(\text{ETOTHHS})\right)$$

$$\text{POT\\_GVAOTHS} = \text{EXP}\left(0.11^\* \text{ LOG(CAPOTHS)} + 0.71^\* \text{ LOG(ETOTHS)}\right)$$

$$+ 4.19 + 0.04^\* \text{ DBT2015}\right) \tag{7.134}$$

$$\text{POT\\_GVAPETCH} = \text{EXP} \left( 0.43^\* \text{ LOG} (\text{CAPPETCH}) \right)$$

$$+ 0.31^\* \text{ LOG} (\text{ETPETCH}) + 3.96 + 0.24^\* \text{ DST0312}$$

$$- 0.29^\* \text{ DP2012} \right) \tag{7.135}$$

$$\text{POT\\_GVATRACOM} = \text{EXP} \left( 1.23^\* \text{ LOG} (\text{CAPNOIL} - \text{CAPMANOO}) \right)$$

$$+ 0.40^\* \text{ LOG} (\text{ETTRACOM}) - 8.93 \right) \tag{7.136}$$

$$\text{POT\\_GVAU} = \text{EXP} \left( 0.78^{\circ} \text{ LOG} (\text{CAPU}) + 0.35^{\circ} \text{ LOG} (\text{ETU}) - 1.11 \right) \tag{7.137}$$

POT GVAMANNO ¼ POT GVAMANNOLPC þ POT GVAPETCH ð7:138Þ

#### 7.1.2.3 Identities for Output Gaps by Economic Activity Sector

$$\text{GAP\\_GVAGR} = \text{GVAGR} - \text{POT\\_GVAGR} \tag{7.139}$$

$$\text{GAP\\_GVACON} = \text{GVACON} - \text{POT\\_GVACON} \tag{7.140}$$

$$\text{GAP\\_GVADIS} = \text{GVADIS} - \text{POT\\_GVADIS} \tag{7.141}$$

$$\text{GAP\\_GVAFIBU} = \text{GVAFIBU} - \text{POT\\_GVAFIBU} \tag{7.142}$$

$$\text{GAP\\_GVAGOV} = \text{GVAGOV} - \text{POT\\_GVAGOV} \tag{7.143}$$

$$\begin{array}{cc} \text{GAP\\_GVAMANNOLPC} = \text{GVAMANNOLPC} \\ \text{ } & \text{POT\\_GVAANNOLPC} \\\\ \text{GAP\\_GVANOL} = \text{GVANOL} - \text{POT\\_GVANOL} \\ \end{array} \tag{7.144}$$

$$\text{GAP\\_GVAOILREF} = \text{GVAOILREF} - \text{POT\\_GVAOILREF} \tag{7.146}$$

$$\text{GAP\\_GVAOTHS} = \text{GVAOTHS} - \text{POT\\_GVAOTHS} \tag{7.147}$$

$$\text{GAP\\_GVAPETCH} = \text{GVAPETCH} - \text{POT\\_GVAPETCH} \tag{7.148}$$

$$\text{GAP\\_GVATRACOM} = \text{GVATRACOM} - \text{POT\\_GVATRACOM} \qquad (7.149)$$

$$\text{GAP\\_GVAU} = \text{GVAU} - \text{POT\\_GVAU} \tag{7.150}$$

### 7.2 Fiscal Block Behavioral Equations and Identities

#### 7.2.1 Equations for Government Expenditure Items

$$\text{LOG(GWSA\\_Z)} = 0.86^\circ \text{LOG(GREV)} + 1.18 + \text{ECT\\_GWSA\\_Z} \qquad (7.151)$$

$$\text{LOG}(\text{GAE\\_Z}) = 0.87^\* \text{ LOG}(\text{GEF}) - 0.96 + \text{ECT\\_GAE\\_Z} \tag{7.152}$$

$$\text{LOG}(\text{GMO\\_Z}) = 0.68^\* \, \text{LOG}(\text{GREV}) + 2.05 + \text{ECT\\_GMO\\_Z} \tag{7.153}$$

$$\begin{array}{c} \text{LOG}(\text{GCGPE}) = 0.72 \, ^\circ \text{LOG}(\text{GREV}) - 5.57 + 6.21 \, ^\circ \text{DSH1981} \\ + \text{ECT\\_GCGPE} \end{array} \tag{7.154}$$

$$\text{LOG}(\text{GC\\_Z\\_OTH}) = 1.03^\* \text{ LOG}(\text{GREV}) - 2.28 + \text{ECT\\_GC\\_Z\\_OTH} \tag{7.155}$$

$$\text{LOG}(\text{GI\\_Z}) = 0.81^\circ \text{LOG}(\text{GREV}) + 0.80 + \text{ECT\\_GI\\_Z} \tag{7.156}$$

#### 7.2.2 Identities for Government Expenditures

$$\text{GEXP} = \text{PSCE} + \text{PSCAPE} \tag{7.157}$$

$$\text{PSCE} = \text{GWSA\\_Z} + \text{GAE\\_Z} + \text{GMO\\_Z} + \text{GCGPE} + \text{PSCE\\_OTH} \quad (7.158)$$

#### LOG PSCE OTH ð Þ¼ 0:45- LOGðGREVOILÞ þ 0:65-LOGðGREVNOILÞ

$$-1.70-2.29^\* \text{ DP1986} + \text{ECT\\_PCE\\_OTH} \tag{7.159}$$

$$\text{PSCAPPE} = \text{PSCAPE} (-1)^{\ast} \text{GI\\_Z/GI\\_Z} (-1) + \text{DIS\\_PSCAPE} \tag{7.160}$$

$$\text{GC\\_Z} = \text{GWSA\\_Z} + \text{GAE\\_Z} + \text{GMO\\_Z} + \text{GC\\_Z\\_OTH} \tag{7.161}$$

$$\text{GC} = \text{GC\\_Z}/\text{PGC}^\* \ 100\tag{7.162}$$

#### 7.2.3 Identities for Government Revenues

$$\text{GREV} = \text{GREVOL} + \text{GREVNOIL} \tag{7.163}$$

$$\text{GREVOIL} = 0.80^{\circ} \text{ XGOIL} \sum \text{RXD} + \text{DIS\\_GREVOIL} \tag{7.164}$$

$$\text{GREVNOIL} = 0.85^{\circ}\text{CEN\\_TOT\\_KSA} + \text{VAT\\_REV} + \text{EXPL} + \text{HUVF}$$

$$+ \text{OVF} + \text{TOTIT} + \text{TOIPC} + \text{DIS\\_GREVNOIL} \tag{7.165}$$

$$\text{VAT\\_REV} = \text{VAT\\_RATE} / 100^\circ \text{ C\\_RATIO}^\* \left( 0.90^\circ \text{ GC\\_Z} \right)$$

$$+ 0.90^\circ (\text{CONS}^\circ \text{ PCONS} / 100) \Big|^\ast \text{VAT\\_REV\\_DUUMMY}$$

$$+ \text{DIS\\_VAT\\_REV} \tag{7.166}$$

$$\text{EXPL} = \text{EXPL}(-1)^{\circ} \text{ POPNS} / \text{POPNS}(-1) + \text{DIS\\_EXPL} \tag{7.167}$$

$$\text{HUVF} = \text{HUVF}(-1)^{\circ} \text{ XSTRAV\\_Z/XSTRAV\\_Z}(-1) + \text{DIS\\_HUVF} \quad (7.168)$$

$$\text{OVF} = \text{OVF}(-1)^{\text{\textquotedbl{}X}} \text{XSTRAV\\_Z} / \text{XSTRAV\\_Z}(-1) + \text{DIS\\_OVF} \qquad (7.169)$$

$$\text{TOTIT} = \text{TOTIT}(-1)^{\*} \,\text{M} \,\underline{Z}/\text{M} \,\underline{Z}(-1) + \text{DIS} \,\underline{T} \,\text{TOT} \tag{7.170}$$

$$\text{TOIPC} = \text{TOIPC}(-1)^{\*} \text{ GVANOL} / \text{GVANOIL}(-1) + \text{DIS\\_TOIPC} \quad (7.171)$$

#### 7.2.4 Identities for Total and Non-oil Budget Balance

$$\mathbf{GB} = \mathbf{GREV} - \mathbf{GEXP} \tag{7.172}$$

$$\text{GBNOIL} = \text{GB} - \text{GREVOIL} \tag{7.173}$$

$$\begin{array}{c} \text{DEBT\\_GOV} = -0.44 \, ^\circ \text{GB} + 17290.59 + 225653.47 \, ^\circ \text{DP2008} \\ + \text{ECT\\_DEBT\\_GOV} \end{array} \tag{7.174}$$

$$\text{DEBTG\\_GOV} = \text{DEBTG\\_GOV}(-1) + \text{DEBT\\_GOV} \tag{7.175}$$

### 7.3 Monetary Block Equations and Identities

M0 ¼ M2 DTS DD ð7:177Þ M1 ¼ M2 DTS ð7:178Þ M3 ¼ M2 þ DQM ð7:179Þ Þ þ Þ ð7:180Þ RLEND ¼ RLENDð Þþ 1 ð Þð RSH RSHð Þ 1 7:181Þ RRLEND ¼ RLEND DLOG CPI ð Þ- 100 ð7:182Þ RLG ¼ RLGð Þþ 1 ð Þ RSH RSHð Þ 1 þ DIS RLG ð7:183Þ IRD ¼ IR UK RLEND ð7:184Þ RDEBT ¼ RDEBTð Þþ 1 RLEND RLENDðÞ ð 1 7:185Þ RRXD ¼ ðRXD=3:75-100ÞðCPI USA=CPIÞ ð7:186Þ REER ¼ NEER- CPI=CPI USA þ DIS REER ð7:187Þ LOG M2 ð Þ¼ 1:00- LOG PGDP ð Þþ 0:82- LOG GDP ð Þ 0:02- IRD þ 0:10- LOG WPO AL R ð Þþ 0:624835- LOG REER ð Þ þ0:03- @TREND 50 ð Þ 3:21 þ ECT MD UR ð7:176Þ LIABP ¼ LIABPð Þ 1 - ð ð Þ DD þ DTS =ðDDð Þþ 1 DTSð Þ 1 DIS LIABP REERE ¼ EXPð1:28- LOGðððGVANOIL=RXDÞ=ðPOP- 10<sup>3</sup> ÞÞ= ðGDPPC WLDÞ - 100Þ þ 0:20- LOGðððGVAOIL=RXDÞ= ðPOP- 10<sup>3</sup> ÞÞ=ðGDPPC WLDÞ - 100Þ 0:24- LOGðNFA=

Þ þ ð Þ Þ ð Þ GDP Z- 100 0:68- LOG GC Z=GDP Z-100 3:28 7:188

$$\text{PRODIN} = \left( \left( \text{GVANOIL} / \text{RXD} \right) / \left( \text{POP}^\* 10^3 \right) \right) / \left( \text{GDPPC\\_WLD} \right) \* 100 \ (7.189)$$

$$\text{PRODDO} = \left( \left( \text{GVAOL} / \text{RXD} \right) / \left( \text{POP}^\* 10^3 \right) \right) / \left( \text{GDPPC} \\_ \text{WLD} \right) \* 100 \quad (7.190)$$

### 7.4 External Block Equations and Identities

#### 7.4.1 Exports Related Equations and Identities

$$\mathbf{X} = \mathbf{X}\mathbf{G} + \mathbf{X}\mathbf{S} + \text{DIS\\_X} \tag{7.191}$$

$$\text{XG} = \text{XGNOIL} + \text{XGOL} \tag{7.192}$$

$$\begin{aligned} \text{LOG (XGNOIL)} &= -1.17^\ast \text{ LOG (REER)} \\ &+ 0.82^\ast \text{ LOG (GDP\\_MNA^\* \ RXD)} \\ &+ 1.08^\ast \text{ LOG (GVANOIL)} - 10.30 \\ &+ \text{ECT } \text{XGNOIL} \\ \end{aligned} \tag{7.193}$$

$$\text{LOG(XOLREF)} = 2.68^\* \text{ LOG(WITEF)} + 1.49^\* \text{ LOG(GVAILREF)}$$

$$-0.18^\* \text{ LOG(WPO\\_AL\\_R)} - 20.52 - 0.12^\* @ \text{TREND}$$

$$-0.21^\* \text{ DP2001} + \text{ECT\\_XOILREF} \tag{7.194}$$

$$\text{WPO\\_AL\\_R} = \text{WPO\\_AL/CPI\\_USA}^\* \left[ 100 + \text{DIS\\_WPO\\_AL\\_R} \right] \qquad (7.195)$$

$$\text{XGOIL} = \text{XGOIL\\_Z/PGDPOIL}^\* \ 100\tag{7.196}$$

$$\text{XGOIL\\_Z} = \text{XGOIL\\_Z}^\* \text{ RXD} \tag{7.197}$$

þ DIS XGOIL\$ Z ð7:198Þ XGOIL\$ Z ¼ XOILC- 365- WPO AL þ XOILREF- 1:2-WPO AL

$$\text{XOILC} = \text{OILMBD} - \text{OILUSE} \tag{7.199}$$

XGNOIL Z ¼ XGNOIL-PGDPNOIL=100 ð7:200Þ

$$\mathbf{X} \,\underline{\mathbf{Z}} = \mathbf{X}^\* \,\mathbf{P} \mathbf{X} / 100 \,\tag{7.201}$$

$$\mathbf{X}\underline{\\$}\underline{\mathbf{Z}} = \mathbf{X}\mathbf{G}\underline{\\$}\underline{\mathbf{Z}} + \mathbf{X}\mathbf{S}\underline{\\$}\underline{\mathbf{Z}}\tag{7.202}$$

$$\text{XG\\$\\_Z} = \text{XGOL\\_Z} + \text{XGNOIL\\_Z} \tag{7.203}$$

$$\text{XGNOILS\\_Z} = \text{XGNOIL\\_Z/RXD} \tag{7.204}$$

$$\mathbf{X} \mathbf{S} \underline{\mathbf{S}} = \mathbf{X} \mathbf{S} \underline{\mathbf{Z}} / \mathbf{R} \mathbf{X} \mathbf{D} \tag{7.205}$$

$$\mathbf{XS} = \mathbf{XS} \\_ \mathbf{Z} / \mathbf{P} \mathbf{X}^\* \ 100 \tag{7.206}$$

XS Z¼ ðXSOIL ZþXSII ZþXSTRAN ZþXSTRAV Z

þ XSIP ZþXSFIN ZþXSCOM ZþXSOBS ZþXSGOV ZÞ ð7:207Þ

#### 7.4.2 Imports Related Equations and Identities

$$\mathbf{M} = \mathbf{M}\mathbf{G} + \mathbf{M}\mathbf{S} + \mathbf{D}\mathbf{I}\mathbf{S}\\_\mathbf{M} \tag{7.208}$$

$$\text{MG} = \text{MGCAP} + \text{MGCONS} + \text{MGINTER} + \text{DIS\\_MG} \tag{7.209}$$

$$\text{LOG(MGCAP)} = 1.03^{\circ} \text{ LOG(DOMD)} + 0.84^{\circ} \text{ LOG(NEER)}$$

$$-0.31^{\circ} \text{ LOG(PGDP\\_US/PGDP)} - 7.16$$

$$1 + 0.37^\* \text{ DSH2003} - 0.15^\* \text{ DBT2018} + \text{ECT\\_MGCAP} \quad (7.210)$$

$$\text{LOG(MGCONS)} = 0.86^\circ \text{ LOG(DOMD)} + 1.08^\circ \text{ LOG(REER)} - 7.07$$

$$+ 0.02^\circ \text{@TREND} - 0.12 \ast \text{ DBT2018} + 0.32^\circ \text{ DSH2003}$$

$$+ \text{ECT\\_MGCONS} \tag{7.211}$$

$$\text{LOG(MGINTER)} = 2.56^\* \text{ LOG(GVANOIL} + \text{GVAOL)}$$

$$+0.56^\* \text{ LOG(NEER)} - 0.86^\* \text{LOG(PGDP\\_US/PGDP)}$$

$$-25.62 - 0.04^\*@ \text{TRED} - 0.09^\* \text{ DBT2018}$$

$$-0.14^\*\text{ DSH2010} + \text{ECT\\_MGilbertR} \tag{7.212}$$

$$\text{MG\\_Z} = \text{MGCAP\\_Z} + \text{MGCONS\\_Z} + \text{MGINTER\\_Z} + \text{DIS\\_MG\\_Z} \quad (7.213)$$

$$\text{MGCAP}\\_Z = \text{MGCAP}^\* \text{PMG} \tag{7.214}$$

$$\text{MGCONS}\_{\square} Z = \text{MGCONS}^\* \text{ PMG} \tag{7.215}$$

$$\text{MGINTER\\_Z} = \text{MGINTER}^\* \text{PMG} \tag{7.216}$$

$$\begin{array}{l} \text{LOG(MS)} = 0.44^{\circ} \text{LOG(DOMD)} - 1.61^{\circ} \text{LOG(RRXD)} + 12.60\\ + \text{ECT\\_MS} \end{array} \tag{7.217}$$

$$\begin{aligned} \text{I} \text{ } \text{MOLREF} &= \text{DOLREF\\_T} + \text{XOILREF}^\* \, 0.14 - \text{QOILREF} \\ &+ \text{DIS\\_MOLREF} \end{aligned} \tag{7.218}$$

$$\mathbf{M}\\_\mathbf{Z} = \mathbf{M}^\* \text{ PM} / 100\tag{7.219}$$

#### 7.4.3 Overall and Non-oil Trade Balance

$$\mathbf{T}\mathbf{B} = \mathbf{X} - \mathbf{M} \tag{7.220}$$

$$\text{TBNOIL} = \text{XGNOIL} - \text{M} \tag{7.221}$$

#### 7.4.4 Other BOP Related Equations and Identities

$$\text{LOG}(\text{l00\* }\text{REMOF\* }\text{RXD}/\text{PGDP}) = 1.70\*\text{ LOG}(\text{GDP})$$

$$+ 1.00\*\text{ LOG}(\text{ETNS})$$

$$+ 1.43\*\text{ LOG}(\text{PGDP})$$

$$-0.09\*\text{ LOG}(\text{(EXPL/PGDP\* }\text{100)}+\text{l)}$$

$$+ \text{ECT }\text{REMOF} \tag{7.222}$$

$$\begin{array}{c} \text{FDI\\$OUT\\_Z} = \text{FDI\\$OUT\\_Z}(-1)^\* \left( \text{GDP\\$\\_Z/\text{GDP\\$\\_Z}(-1) \right) \\ + \text{DIS\\_FDI\\$OUT} \end{array} \tag{7.223}$$

$$\text{FDI\ $} = \text{FDI\$ } \text{IN\\_Z} - \text{FDI\ $} \text{OUT\\_Z} + \text{DIS\\_FDI\$ } \tag{7.224}$$

$$\begin{aligned} \text{WTOUR} &= (\text{WTOUR}(-1) \, ^\circ (\text{AA'} \, ^\circ \text{MS} \, \text{ZAF} / \text{MS} \, \text{ZAF}(-1) + \\ \text{AA2}^\* \, \text{MS} \, \text{USA} / \text{MS} \, \text{USA} (-1) + \\ \text{AA4}^\* \, (\text{MS} \, \text{WIE}) / (\text{MS} \, \text{WIE}(-1)) + \text{AA8}^\* \, \text{JIN} / \text{MS} \, \text{JIN} \, \text{JIN} (-1) + \\ \text{AA6}^\* \, (\text{MS} \, \text{TUR}) / (\text{MS} \, \text{TUR}(-1)) + \text{AA7}^\* \, \text{MS} \, \text{DEU} / \text{MS} \, \text{DEU} (-1) + \\ \text{AA8}^\* \, \text{MS} \, \text{JSS} \, \text{FRA} (-1) + \text{AA9}^\* \, \text{MS} \, \text{TIA} / \text{MS} \, \text{TIA} (-1) + \\ \text{AA10}^\* \, \text{MS} \, \text{GBR} / \text{MS} \, \text{GBR} (-1)) \, \text{)} + \text{DS} \, \text{WTOUR} \end{aligned}$$

ð7:225Þ

WTREF ¼ 100 ðAB1- DOILREF ARGENTIN=AB2 þ AB3- DOILREF AUSTRALI=AB4 þ AB5- DOILREF AUSTRIA=AB6 þ AB7- DOILREF BELGIUM=AB8 þ AB9- DOILREF BRAZIL=AB10 þAB11- DOILREF BULGARIA=AB12 þ AB13- DOILREF CANADA=AB14 þ AB15- DOILREF CHILE=AB16 þ AB17- DOILREF CHINA=AB18 þ AB19- DOILREF CROATIA=AB20 þ AB21- DOILREF CZECH=AB22 þ AB23- DOILREF DENMARK=AB24 þ AB25- DOILREF FINLAND=AB26 þ AB27- DOILREF FRANCE=AB28 þ AB29- DOILREF GERMANY=AB30 þ AB31- DOILREF GREECE=AB32 þ AB33- DOILREF HK=AB34 þAB35- DOILREF HUNGARY=AB36 þ AB37- DOILREF INDIA=AB38 þ AB39- DOILREF INDONESI=AB40 þ AB41- DOILREF IRELAND=AB42 þ AB43- DOILREF ITALY=AB44 þ AB45- DOILREF JAPAN=AB46 þ AB47- DOILREF KOREA=AB48 þ AB49- DOILREF MALAYSIA=AB50 þ AB51- DOILREF MEXICO=AB52 þ AB53- DOILREF NETH=AB54 þ AB55- DOILREF NORWAY=AB56 þ AB57- DOILREF PHILIPPI=AB58 þ AB59- DOILREF POLAND=AB60 þ AB61- DOILREF PORTUGAL=AB62 þ AB63- DOILREF ROMANIA=AB64 þ AB65-DOILREF RUSSIA=AB66

þ AB67- DOILREF SAFRICA=AB68 þ AB69- DOILREF SINGPORE=AB70 þ AB71- DOILREF SLOVAKIA=AB72 þ AB73- DOILREF SPAIN=AB74 þ AB75- DOILREF SWEDEN=AB76 þ AB77- DOILREF SWITZ=AB78 þ AB79- DOILREF TAIWAN=AB80 þ AB81- DOILREF THAILAND=AB82 þ AB83- DOILREF TURKEY=AB84 þ AB85- DOILREF UAEMOD=AB86 þ AB87- DOILREF UK=AB88 þAB89-DOILREF US=AB90Þ ð7:226Þ

Here, the coefficients from AA1 to AA10 and from AB1 to AB90 are obtained from OEGEM. We do not report numerical values of them due to the data confidentiality issue.

### 7.5 Domestic Prices Block Equations and Identities

#### 7.5.1 Consumer Prices

#### 7.5.1.1 Equations for Sectoral CPI

$$\begin{aligned} \text{LOG (CPU)} &= 0.44^{\circ} \text{ LOG (PGDPREAL)} + 0.13^{\circ} \text{ LOG (PE\_RES)} \\ &+ 0.69^{\circ} \text{ LOG (PGDPES)} - 1.47 - 0.20^{\circ} \text{DP} 2018 \\ &- 0.26^{\circ} \text{ DP2019} + \text{LOG ((VAT\_RATE + 100)/100)} + \text{ECT\\_CPU} \\\\ \text{LOG (CPIFOD)} &= 1.67^{\circ} \text{ LOG (PGDPARGR)} + 0.50^{\circ} \text{ LOG (PMG)} \\ &+ 0.16^{\circ} \text{ LOG (WDS)} + 0.08^{\circ} \text{LOG (PEL\\_COMM)} \\ &- 7.55 + \text{LOG ((VAT\\_RATE + 100)/100)} + \text{ECT\\_CPU} \text{PHO} \\ \text{LOG (CPITRA)} &= 0.18^{\circ} \text{ LOG (WSER)} + 0.19^{\circ} \text{ LOG (PE\\_RACEOM)} \\ &+ 0.30^{\circ} \text{ LOG (PMG)} + 0.25 + \text{LOG ((VAT\\_RATE + 100)/100)} \quad (7.229) \end{aligned}$$

$$+ \text{ECT\\_CPURA\\_ULCC}$$

$$\begin{aligned} \text{LOG (CPHH)} &= 0.21^\circ \text{LOG (WDS)} + 0.71^\circ \text{LOG (PGDPMANNO)} \\ &+ 0.25^\circ \text{LOG (PM)} - 1.20 - 0.01^\circ \text{GTREND} \\ &- 0.08^\circ \text{DP2008} + \text{LOG ((VAT\\_RATE + 100)/100)} \\ &+ \text{ECT\\_CPHHH\\_ULC} \\\\ \text{LOG (CPICOMM)} &= 0.25^\circ \text{LOG (WTRACOM)} + 0.87^\circ \text{LOG (PGDPTRACOM)} \end{aligned} \tag{7.230}$$

$$\overset{\cdot}{+} + 0.49^\* \text{ LOG (PM)} - 0.03^\* (\text{@TREND})$$

$$+ \text{ LOG } ((\text{VAT\\_RATE} + 100)/100) + \text{ECT\\_CPUCOM} \text{ ULC} \quad (7.231)$$

$$\begin{array}{l} \text{LOG} \left( \text{CPHTL} \right) = 1.36 + 0.63^{\circ} \text{ LOG} \left( \text{PGDPDIS} \right) + 0.06^{\circ} \text{ LOG} \left( \text{PMS} \right) \\ + \text{LOG} \left( \left( \text{VAT\\_RATE} + 100 \right) / 100 \right) \\ + \text{ECT\\_CPHTL\\_ULC} \end{array} \tag{7.232}$$

þ 0:14- LOG ðWDISÞ 0:02- @TREND þ 0:09- DP2016 þ 1:58 þ LOG ððVAT RATE þ 100Þ=100Þ þ ECT CPICLOTH ULC ð7:233Þ LOG ðCPICLOTHÞ ¼ 0:50- LOG ðPGDPMANNOÞ þ 0:16-LOG PMð Þ

$$\text{LOG } (\text{CPIMSC}) = 0.52^\circ \text{ LOG } (\text{PGDPMANOO}) + 0.34^\circ \text{ LOG } (\text{PGDPSER})$$

$$\begin{aligned} &+0.21^\* \text{LOG} \left( \text{PMG} \right) - 0.07 - 0.004^\* \text{@TREDD} \\ &+ \text{LOG} (\text{(VAT\\_RATE + 100)/100}) + \text{ECT\\_CPIMISC\\_ULC} \end{aligned} \tag{7.234}$$

$$\begin{aligned} \text{LOG (CPIEDU)} &= 0.12^{\circ} \text{ LOG (W)} + 0.24^{\circ} \text{ LOG (PGDPSER)} \\ &+ 0.05^{\circ} \text{ LOG (PMS)} + 2.13 + \text{LOG ((VAT\\_RATE + 100)/100)} \\ &+ \text{ECT\\_CPIEDU\\_ULC} \\ \end{aligned} \tag{7.235}$$

$$\text{LOG (CPIART)} = 0.60^{\circ} \text{LOG (PGDPSER)} + 0.51^{\circ} \text{LOG (WSER)}$$

$$+ 0.55^{\circ} \text{LOG (PM)} - 0.05^{\circ}(@ \text{TREND})$$

$$+ \text{LOG}((\text{VAT\\_RATE} + 100)/100) + \text{ECT \_CPUART\\_ULLC} \quad (7.236)$$

$$\text{LOG} \left( \text{CPIMEAL} \right) = 0.16^{\circ} \text{LOG} \left( \text{PGMPS} \right) + 0.09^{\circ} \text{LOG} \left( \text{PMS} \right) + 3.4 \\$$$

$$+ 0.05^{\circ} \text{DP2003} + 0.05 \star \text{DB1617} + \text{LOG} (\left( \text{VAT\\_RATE} + 100 \right) / 100)$$

$$+ \text{ECT\\_CPHEAD\\_ULC} \tag{7.237}$$

$$\text{LOG (CPT)} \text{(CPT)} \text{BC} = 1.61^{\circ} \text{ LOG (PGDPAMNNO)} + 0.71^{\circ} \text{ LOG (WDIS)} + 0.15^{\circ}$$

$$\text{LOG (PELE \\_COMM)} + 0.01^{\circ} \text{ (@TRENID)} + \text{LOG ((VAT \\_RATE + 100)/100)}$$

$$+ \text{ECT \_CPT} \text{COBC\\_ULC}$$

#### 7.5.1.2 Identity for Total CPI

$$\begin{aligned} \text{CPI} &= \text{CPIU\\_W^\*} \text{ CPIU} + \text{CPIFOOD\\_W^\*} \text{ CPIFOOD} \\ &+ \text{CPITRA\\_W^\*} \text{ CPITRA} + \text{CPIHH} \text{ W'CPHH} \\ &+ \text{CPICOOH\\_W^\*} \text{ CPICOOH} + \text{CPIHTL\\_W^\*} \text{ CPIHTL} \\ &+ \text{CPICLOTH} \text{ W'} \text{ CPILOTH} + \text{CPIMSC} \text{ W'} \text{ CPIMSIC} \\ &+ \text{CPIEDU\\_W^\*} \text{ CPIEDU} + \text{CPIART} \text{ W'} \text{ CPIART} \\ &+ \text{CPIHEAL\\_W^\*} \text{ CPHEAL} + \text{CPITOBC\\_W^\*} \text{ CPITOBC} \\ &+ \text{DIS\\_CPI} \\ &+ \text{DIS\\_CPI} \end{aligned} (7.239)$$

#### 7.5.2 Producer Prices

#### 7.5.2.1 Equations for Sectoral Producer Prices

$$\text{LOG}(\text{PGDPAGR}) = 0.05^\circ \text{LOG}(\text{WAGR}) + 0.24^\circ \text{LOG}(\text{PELE} \text{ AGB})$$

$$+ 0.34^\circ \text{LOG}(\text{PMG}) + 0.82 + 0.09^\circ \text{ DP2017}$$

$$+ \text{ECT\\_PGDPARGR} \tag{7.240}$$

$$\text{LOG(PGDPCON)} = 0.09^\* \text{LOG(ULCCON)} + 0.09^\* \text{LOG(PE\_{CON})}$$

$$+ 0.59^\* \text{LOG(PM)} + 0.20 + 0.01^\* \text{@TRENID}$$

$$+ \text{ECT\_PGDPCON} \tag{7.241}$$

$$\text{LOG(PGDPDIS)} = 0.22^\circ \text{LOG(WDIS)} + 0.50^\circ \text{LOG(PMG)} - 0.67$$

$$+ 0.01^\circ \text{@TREND} + 0.16^\circ \text{DP2017} + \text{ECT\\_PGDPDIS} \tag{7.242}$$

$$\dots \quad . \qquad . \qquad . \qquad . \qquad . \qquad . \qquad . \qquad . \qquad . \qquad . \qquad . \qquad .$$

$$\text{LOG(PGDPFBU)} = 0.14^\circ \text{LOG(ULCFIBU)} + 0.63^\circ \text{LOG(PELE\\_GOAM)}$$

$$+ 0.51^\circ \text{LOG(PMG)} - 3.08 + \text{ECT\\_PGDPFIBU}$$

$$\text{LOG(PGDPGOV)} = 0.30^\circ \text{LOG(ULCGOV)} + 0.12^\circ \text{LOG(PE\\_GOV)}$$

$$+ 1.74^\circ \text{LOG(PM)} - 5.65 + \text{ECT\\_PGDPGOV} \quad (7.244)$$

$$\text{LOG}(\text{PGDPMANNO}) = 0.04^\circ \text{LOG}(\text{WMAN}) + 0.14^\circ \text{LOG}(\text{PE\\_MANNO})$$

$$+ 0.97^\circ \text{LOG}(\text{PM}) - 1.08 + \text{ECT\\_PGDPPMANNO}$$

$$(7.245)$$

$$\begin{aligned} \text{LOG(PGDPOLREF)} &= 0.11^\* \text{LOG(ULCOIL)} + 0.75^\* \text{LOG(PGDI\\_IND)} \\ &+ 0.69^\* \text{LOG(PGDPOL)} + 0.75^\* \text{LOG(PGDI\\_IND)} \\ &- 2.52 + \text{ECT\\_PGDPOLREF} \\ \text{LOG(PGDPOTHS)} &= 0.12^\* \text{LOG(ULCSER)} + 0.04^\* \text{LOG(PE\\_OTHS)} \\ &+ 0.62^\* \text{LOG(PM)} + 0.96 + \text{ECT\\_PGDPOTHS} \\ &+ 0.07^\* \text{LOG(ULCTRAOM)} \\ &+ 0.07^\* \text{LOG(PE\\_TRAOM)} + 0.32^\* \text{LOG(PM)} \\ &+ 2.04 + \text{ECT\\_PGDPITACOM} \\ \text{LOG(PGDPSER)} &= 0.07^\* \text{LOG(WSER)} + 0.75^\* \text{LOG(PM)} \\ &+ 0.12^\* \text{LOG(PELE\\_CONFIG)} - 1.27 + 0.01^\* \text{GTREND} \\ &- 0.06^\* \text{DPA008} + \text{ECT\\_PGDPSER} \\ \text{LOG(PGDPU)} &= 0.22^\* \text{LOG(LUCU)} + 0.80^\* \text{LOG(PE\\_U)} \\ &+ 0.92^\* \text{LOG(PDG)} - 4.73 - 0.20^\* \text{DPA008} + \text{ECT\\_PGDPID} \end{aligned}$$

ð7:250Þ

#### 7.5.2.2 Identities for Producer Prices

$$\begin{aligned} \text{PGDP} &= \text{(GVANOIL/GDP)^\* PGDPNOIL} + \text{(GVAOL/GDP)^\* PGDPOIL} \\ &+ \text{(GVANT/GDP)^\* PGDPNOIT} + \text{DIS}\_P \text{PGDP} \\ \text{PGDPNOIL} &= \text{(GVAAGR/GVANOIL)^\* PGDPAGR} \\ &+ \text{(GVAAMNOTI/GVANOL)^\* PGDPMICOOH} \\ &+ \text{(GVAAMNON/GVANOL)^\* PGDPMIANNOI} \\ &+ \text{(GVAU/GVANOIL)^\* PGDPON} \\ &+ \text{(GVAON/GVANOL)^\* PGDPON} \\ &+ \text{(GVAON/GVANOL)^\* PGDPINGS} \\ &+ \text{(GVAFRAOM/GVANOL)^\* PGDPTRAPON} \\ &+ \text{(GVAFIBU/GVANOL)^\* PGDPFIBUM} \\ &+ \text{(GVAOTIS/GVANOL)^\* PGDPFIBU} \\ &+ \text{(GVAOTIS/GVANOL)^\* PGDPITHS} \end{aligned}$$

þ ð Þ GVAGOV=GVANOIL -PGDPGOVþDIS PGDPNOIL ð7:252Þ

#### 7.5.2.3 Identities for Aggregated Energy Prices

$$\begin{aligned} \text{PE}\_{\text{A}}\text{AGR} &= (\text{REE}\_{\text{L}}\text{AG})\langle\text{ODE}\_{\text{L}}\text{AS} + \text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA}) \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA}) \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DZ}\_{\text{L}}\text{TA}) \text{TA} + \text{DOS}\_{\text{L}}\text{TA} \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DZ}\_{\text{L}}\text{TA} + \text{DOS}\_{\text{L}}\text{TA} + \text{RSA}\_{\text{L}}\text{TA}) \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DZ}\_{\text{L}}\text{TA} + \text{DZ}\_{\text{L}}\text{TA}) \\ &+ (\text{DOS}\_{\text{L}}\text{TA} + \text{DZ}\_{\text{L}}\text{TA} + \text{RSA}\_{\text{L}}\text{TA} + \text{RSA}\_{\text{L}}\text{TA}) \\ &+ (\text{D}$$

$$\begin{aligned} \text{T\_{AC}}\text{COV} &= (\text{DELE}\_{\text{C}}\text{CO}/(\text{DELE}\_{\text{C}}\text{CO} + \text{D}\_{\text{C}}\text{R}\_{\text{A}} + \text{D}\_{\text{D}}\text{R}\_{\text{T}}) \\ &+ (\text{DAS}\_{\text{T}}\text{TR}\_{\text{A}}/(\text{DELE}\_{\text{C}}\text{O} + \text{D}\_{\text{S}}\text{R}\_{\text{T}}) \\ &+ \text{DDS}\_{\text{T}}\text{TR}\_{\text{A}} + (\text{DELE}\_{\text{C}}\text{R}\_{\text{T}})^{\text{T}}\text{R}\_{\text{S}} + \text{T}\text{R}\_{\text{S}} \\ &+ (\text{DDS}\_{\text{T}}\text{TR}\_{\text{A}} + \text{D}\_{\text{D}}\text{R}\_{\text{T}})^{\text{T}}\text{R}\_{\text{S}} + \text{T}\text{R}\_{\text{S}} \\ &+ (\text{DDS}\_{\text{T}}\text{TR}\_{\text{A}} + \text{D}\_{\text{D}}\text{R}\_{\text{T}})^{\text{T}}\text{R}\_{\text{T}}\text{R}\_{\text{T}} \\ &+ (\text{DDS}\_{\text{T}}\text{TR}\_{\text{A}} + \text{D}\_{\text{D}}\text{R}\_{\text{T}}\text{R}\_{\text{T}})^{\text{T}}\text{R}\_{\text{T}}\text{R}\_{\text{T}} \\ &+ \text{DDS}\_{\text{T}}\text{TR}\_{\text{A}} + \text{D}\_{\text{E}}\text{R}\_{\text{T}}\text{R}\_{\text{T}} + \text{D}\_{\text{S}}\text{R}\_{\text{T}} \\ &+ (\text{DDS}\_{\text{T}}\text{$$

ÞÞ - PNGA IND þ ðDCOIL U=ðDNGA U þ DCOIL U þ DDIS U - þð ð þ þDHFO IND þ DDIS IND þ DCOIL IND þ DELE IND þ þ þ ÞÞ ð7:261Þ PE OILREF ¼ ðDCOIL IND=ðDCOIL IND þ DELE EOU þ DNGA EOUÞÞ- PCOIL IND þðDELE EOU=ðDCOIL IND þ DELE EOU þ DNGA EOUÞÞ- PELE IND PE U ¼ ðDNGA U=ðDNGA U þ DCOIL U þ DDIS U þ DHFO U þ DHFO UÞÞ PCOIL IND þ ðDDIS U=ð DNGA U þ DCOIL U þ DDIS U þ DHFO UÞÞ - PDIS IND þðDHFO U=ðDNGA U þ DCOIL U þ DDIS U þ DHFO UÞÞ - PHFO IND ð7:260Þ PE MANNO ¼ ðDNGA IND=ðDNGA IND þ DNGA IND NEU þDHFO IND þ DDIS IND þ DCOIL IND þDELE IND þ DOTH INDÞÞ - PNGA IND DNGA IND NEU= DNGA IND DNGA IND NEU þDOTH INDÞÞ - PNGA IND þ ðDDIS IND=ðDNGA IND þDNGA IND NEU þ DHFO IND þ DDIS IND þDCOIL IND þ DELE IND þ DOTH INDÞÞ - PDIS IND þðDHFO IND=ð DNGA IND þ DNGA IND NEU þDHFO IND þ DDIS IND þ DCOIL IND þ DELE IND þDOTH INDÞÞ - PHFO IND þ ðDCOIL IND=ðDNGA IND þ DNGA IND NEU þ DHFO IND þ DDIS IND DCOIL IND DELE IND DOTH IND - PCOIL IND þ ðDELE IND=ðDNGA IND þ DNGA IND NEU þDHFO IND þ DDIS IND þ DCOIL IND þ DELE IND þ DOTH INDÞÞ - PELE IND þ ðDOTH IND=ðDNGA IND þ DNGA IND NEU þ DHFO IND þ DDIS IND þDCOIL IND þ DELE IND þ DOTH INDÞÞ -POTH IND

þðDNGA EOU=ðDCOIL IND þ DELE EOU

þ DNGA EOUÞÞ -PNGA IND ð7:262Þ

$$\begin{aligned} \text{PE\\_AGR} &= (\text{DELE\\_AGR}/(\text{DELE\\_AGR} + \text{DGAS\\_TRA} + \text{DDS\\_TRA} \\ &+ \text{DKER\\_TRA}))^+ \text{PELE\\_AGR} \\ &+ (\text{DGS\\_TRA}/(\text{DELE\\_AGR} + \text{DGS\\_TRA} \\ &+ \text{DDS\\_TRA} + \text{DKER\\_TRA}))^+ \text{PGAS\\_TRA} \\ &+ (\text{DDS\\_TRA}/(\text{DELE\\_AGR} + \text{DGS\\_TRA} \\ &+ \text{DDS\\_TRA} + \text{DKER\\_TRA}))^+ \text{DIS\\_TRA} \\ &+ (\text{DKER\\_TRA}/(\text{DELE\\_AGR} + \text{DGS\\_TRA} \\ &+ \text{DDIS\\_TRA} + \text{DKER\\_TRA}))^+ \text{PKER\\_TRA} \end{aligned} \tag{7.263}$$

### 7.6 Labor and Wages Block Equations and Identities

#### 7.6.1 Equations for Sectorial Employment

$$\begin{aligned} \text{LGG (ETAGG)} &= 0.92^\circ \text{LGG (GVAGRR)} - 0.86^\circ \text{LGG (WaRGR/PGDPARC^\* 100)} \\ &+ 1.36 + 0.04^\circ \text{(TREEND + 0.18}^\circ \text{DP2018)} \\ &+ \text{ECT\\_ETAGR} \end{aligned} \tag{7.264}$$

$$\begin{aligned} \text{LGG (ETCON)} &= 0.94^\circ \text{ (GVAON)} - 0.21^\circ \text{ LGG (WCON/PGDPCC^\* 100)} \\ &- 1.68 - 0.24 \text{ 1} \text{ DP200708} + \text{ECT\\_ETCON} \end{aligned} \tag{7.265}$$

$$\begin{aligned} \text{LGG (ETDIS)} &= 0.13^\circ \text{ 1} \text{ } \text{LGG (GVADIS)} - 0.72^\circ \text{ LGG (WDIS/PGGDIS^\* 100)} \\ &+ 12.87 + 0.13^\circ \text{DBT2016} - 0.13^\circ \text{ DP2003} + \text{ECT\\_ETDIS} \end{aligned} \tag{7.266}$$

$$\begin{aligned} \text{LGG (ETHIBU)} &= 0.93^\circ \text{ } \text{LGG (GVAPIBU)} \\ &- 0.16^\circ \text{ } \text{LGG (WIFBUJ)} \text{ } \text{ } \text{T}^2 \text{U} \end{aligned} \tag{7.267}$$

$$\begin{aligned} -3.50 - 0.35 &+ \text{DP201314} + 0.38^\circ \text{ DP201718} \\ &+ \text{ECT\\_ETFIBU} \end{aligned} \tag{7.267}$$

$$\begin{aligned} \text{LGG}(\text{ETGOV}) &= 1.12^{\text{t}} \text{LGG}(\text{GVAGO}) \\ &- 0.14^{\text{t}} \text{LGG}(\text{GVASA}\_{\text{Z}}/\text{PGRGOW}^{\text{t}} \ \ ^{100} \\ &- 5.42 + \text{ET}\_{\text{ET}} \text{ETOW} \ \text{or} \\ &- 5.42 + \text{ET}\_{\text{ET}} \text{ETOW} \ \text{(} \\ &+ 0.13^{\text{t}} \text{LGG}(\text{WAMAN}) \\ &- 6.21 - 0.36^{\text{t}} \text{D} \ \text{LGG} \ 100 \text{/RODPM} \ \text{(} \\ &- 6.21 - 0.36^{\text{t}} \text{D} \ \text{LGG} \ 100 + \text{ET}\_{\text{ET}} \text{EMONNO} \ \text{(} \\ \text{LogG(ETMNOOH)} &= 0.28^{\text{t}} \text{LGG}(\text{GVAAMNOHT}^{\text{t}} \ \ ^{100}) + 9.21 \\ &- 0.15^{\text{t}} \text{D} \ \text{LGG}(\text{WAMNHOH}^{\text{t}} \ \text{H}^{100}) \\ &- 0.15^{\text{t}} \text{D} \ \text{LGG}(\text{WAMNHHS}^{\text{t}} \ \ \text{(} \\ &- 0.3^{\text{t}} \text{LGG}(\text{GVAOMHS}^{\text{t}}) \\ &- 0.3^{\text{t}} \text{LGG}(\text{W} \ \text{GVAHOHS}^{\text{t}} \ \ \text{100}) \\ &+ 1.94 + 0.16^{\text{t}} \text{D} \ \text{LGG} \ 10-0.09^{\text{t}} \text{D} \ \text{LGG}^{\text{0$$

#### 7.6.1.1 Identities for Labor Market

$$\text{ET} = \text{ETNOIL} + \text{ETOIL} + \text{DIS\\_ET} \tag{7.275}$$

$$\text{ETNOIL} = \text{ETAGR} + \text{ETCON} + \text{ETDIS} + \text{ETFIBU} + \text{ETGOV}$$

$$+ \text{ETMANNO} + \text{ETPETCH} + \text{ETMINDH} + \text{ETOTHS}$$

$$+ \text{ETTRACOM} + \text{ETU} + \text{DIS\\_ETNOIL} \tag{7.276}$$

$$\text{ETOIL} = \text{ETOILREF} + \text{ETOILMIN} \tag{7.277}$$

$$\text{ETMIN} = \text{ETMINOTH} + \text{ETOLMIN} \tag{7.278}$$

$$\text{ETP} = \text{ETPS} + \text{ETPNS} \tag{7.279}$$

ETSER ¼ ETDIS þ ETTRACOM þ ETFIBU þ ETOTHS þ ETGOV ð7:280Þ

$$\text{LF} = \text{PART}/100.00^{\circ} \ (\text{POPS} + \text{POPS} - \text{POP}014) + \text{DIS\\_LF} \qquad (7.281)$$

$$\mathbf{U} = \mathbf{L}\mathbf{F} - \mathbf{E}\mathbf{T} + \mathbf{D}\mathbf{I}\mathbf{S}\\_\mathbf{U} \tag{7.282}$$

$$\mathbf{UR} = \mathbf{U}/\mathbf{LF}^\* \ 100\tag{7.283}$$

$$\mathbf{U}\mathbf{R}\\_\mathbf{C} = \mathbf{U}\mathbf{R} - \mathbf{U}\mathbf{R}\\_\mathbf{N} \tag{7.284}$$

#### 7.6.2 Equations for Sectoral Wages

$$\begin{aligned} \text{LOG(WAGE)} &= 0.77^{\circ} \text{ LOG}(\text{GVAGR/ETAGB}) \\ &+ 2.20^{\circ} \text{ LOG}(\text{PGDPAGR}) - 4.63 + \text{ECT\\_WAGN} \quad (7.285) \\ \text{LOG}(\text{WCON}) &= 0.81^{\circ} \text{ LOG}(\text{GVACON/ETCON}) + 0.21^{\circ} \text{ LOG}(\text{PGDPCON}) \\ &+ 5.12 - 0.41^{\circ} \text{DST} 0308 - 0.14^{\circ} \text{ DB0910} \\ &+ 0.16^{\circ} \text{ DST9803} + \text{ECT\\_WCON} \end{aligned}$$

$$\begin{aligned} \text{LOG}(\text{WFIBU}) &= 0.37^{\circ} \text{ LOG}(\text{GVAFIBU}/\text{ETFIBU}) \\ &+ 0.80^{\circ} \text{ LOG}(\text{PGDPFIBU}) + 4.68 + \text{ECT\\_WFIBU} \quad (7.287) \end{aligned}$$

$$\begin{aligned} \text{LOG(WMAN)} &= 0.95^\* \text{ LOG(GVAMAN/ETMAN)} \\ &+ 0.50^\* \text{ LOG(PGDPMANNO)} + 2.95 + 0.58^\* \text{DP2011} \\ &+ 0.40^\* \text{ DP2012} + \text{ECT } \text{WMAN} \end{aligned} \tag{7.288}$$

$$\text{LOG(WMIN)} = 0.17^\* \text{ LOG(GVAMIN/ETMIN)} + 0.17^\* \text{ LOG(PGDPMIN)}$$

$$\begin{aligned} \text{H}\_{2}\text{O}\_{2} &= \text{O}\_{2} + \text{O}\_{2} + \text{O}\_{2} + \text{O}\_{2} + \text{O}\_{2} + \text{O}\_{2} + \text{O}\_{2} \\ &+ 9.85 + 0.31^{\circ} \text{DP2009} - 0.21^{\circ} \text{DP2011} - 0.17^{\circ} \text{DP2016} \\ &+ \text{ECT\\_WMN} \end{aligned}$$

ð7:289Þ

$$\begin{aligned} \text{LOG}(\text{WTRACOM}) &= 0.14^\circ \text{LOG}(\text{GVATRACOM}/\text{ETTRACOM}) \\ &+ 0.77^\circ \text{ LOG}(\text{PGDPRACOM}) + 6.36 \\ &- 0.32^\circ \text{ DSH2003} - 0.22^\circ \text{ DP2011} + \text{ECT\\_WTRACOM} \end{aligned} \tag{7.290}$$

$$\begin{aligned} \text{LOG}(\text{WU}) &= 1.61^\circ \text{ LOG}(\text{GVAU}/\text{ETU}) + 0.95^\circ \text{ LOG}(\text{PGDPU}) + 1.09 \\ &- 0.05^\circ \text{@TREND} - 0.57 \ast \text{ DST1012} + \text{ECT\\_WU} \end{aligned} \tag{7.291}$$

#### 7.6.2.1 Identities for Unit Labor Cost by Sector

$$\text{ULCAGR} = \text{WAGR} \ast \text{ETAGB} / \text{GVAGR} \tag{7.292}$$

$$\text{ULCCON} = \text{WCON} \ast \text{ETCON} / \text{GVACON} \tag{7.293}$$

$$\text{ULCDIS} = \text{WDIS} \ast \text{ETDIS} / \text{GVADIS} \tag{7.294}$$

$$\text{ULCFIBU} = \text{WFIBU} \ast \text{ETFBU} / \text{GVAFIBU} \tag{7.295}$$

¼ ð Þ ULCFIBUOTH WFIBU ETFIBU=GVAFIBUOTH 7:296

$$\text{ULCGOV} = \text{GWSA}\,\angle \ast \,\text{ETGOV} / \text{GVAGOV} \,\tag{7.297}$$

$$\text{ULCMANNO} = \text{WMAN} \ast \text{ETMANNO} / \text{GVAAMNOO} \tag{7.298}$$

$$\text{ULCOILREF} = \text{WMAN} \ast \text{ETOILREF} / \text{GVAOILREF} \tag{7.299}$$

$$\text{ULCOTHS} = \mathbf{W} \ast \text{ETOTHS} / \text{GVAOTHS} \tag{7.300}$$

¼ ð Þ ULCTRACOM WTRACOM ETTRACOM=GVATRACOM 7:301

$$\text{ULCU} = \text{WU} \ast \text{ETU} / \text{GVAU} \tag{7.302}$$

$$\text{ULCSER} = \text{W.OLD} \* \text{ETSER} / \text{GVAER} \tag{7.303}$$

$$\text{ULCSER} = \text{W.OLD} \ast \text{ETNOIL} / \text{GVANOIL} \tag{7.304}$$

$$\text{ULCOIL} = \text{W} \, \text{OLD} \, \* \, \text{ETOIL}/\text{GVAOL} \, \tag{7.305}$$

#### 7.6.2.2 An Identity for Labor Compensation

$$\begin{aligned} \text{LABCOMP} &= \left( \text{ETAGG}^\* \,\text{WAGE} + \text{ETCON}^\* \,\text{WCON} + \text{ETDIS}^\* \,\text{WDI} \right) \\ &+ \text{ETIBIU}^\* \,\text{WFIR} + \text{ETOOV}^\* \\ &\quad \left( \left( \text{GWSa} \,\text{L}^\* \,\text{10}^6 \right) / \left( \text{ETGOV}^\* \,\text{10}^3 \right) \right) \\ &+ \text{ETMANN}^\* \,\text{WMAR} + \text{ETPETCH}^\* \,\text{WPETCH} \\ &+ \text{ETMONO}^\* \,\text{W} \,\text{CEIC} + \text{ETOTHS}^\* \,\text{W} \,\text{CEC} \\ &+ \text{ETRACOM}^\* \,\text{WTRACOM} + \text{ETU}^\* \,\text{WU} \\ &+ \text{ETOIL} \,\text{REF}^\* \,\text{W} \,\text{CEC} + \text{ETOILMIN}^\* \,\text{W} \,\text{CECIC} \,\text{} \,\text{/} \,\text{1000} \\ &+ \text{DIS} \,\text{LABCOMP} \end{aligned}$$

### 7.7 Energy Block Equations and Identities

#### 7.7.1 Energy Demand Equations

#### 7.7.1.1 Industry

$$\begin{aligned} \text{LOG} \left( \text{DCOIL\\_IND} \right) &= 1.32^\* \text{LOG} \left( \text{GVAIND} - \text{GVAU} \right) \\ &- 0.65^\* \text{LOG} \left( \text{PCOIL\\_IND} / \text{PGDPIND}^\* \ 100 \right) \\ &+ 1.76^\* \text{ LOG} \left( \text{PDIS\\_IND} / \text{PGDPIND}^\* \ 100 \right) \\ &- 25.77 + \text{ECT\\_DGOIL\\_IND} \ &(7.307) \\ \text{LOG} \left( \text{DDIS\\_IND} \right) &= 0.21^\* \text{LOG} \left( \text{GVAAMANNO} \right) \\ &- 0.13^\* \text{ LOG} \left( \text{PDIS\\_IND} / \text{PGDPMANNO}^\* \ 100 \right) \\ &- 2.06 + 0.03^\* \text{@TREND} \\ &- 0.17^\* \text{ DBI} \mathbf{72016} - 0.16^\* \text{ DPI} \mathbf{986} + \text{ECT\\_DDS} \ \text{IND} \end{aligned} (7.30)$$

ð7:308Þ

LOG DELE ð Þ IND ¼ 0:61- LOGðGVAMANNOÞ Þ þ 0:41- DP2006 þ ECT DELE IND Þ Þ 11:22 0:42- DP2009 þ ECT DHFO IND ð7:310Þ Þ þ ð Þ þ 0:47- LOG NG ð Þ PRO 3:34 þ 0:17- DP2015 0:09- LOG POTH IND=PGDPMANNO ð Þ 100 þ 1:14- LOG NGL ð Þ 9:68 þ ECT DOTH IND ð Þ 0:31- LOG PELE IND ð 100=PGDPMANNO þ 0:94- LOG POP1564 ð Þ 13:12 þ 0:61- DSH9005 ð7:309Þ LOG DHFO IND ð Þ¼ 0:94- LOGðGVAMANNOÞ 0:13- LOG PHFO IND=PGDPIND ð 100 þ 0:59- LOG PNGA IND=PGDPIND ð 100 LOG DNGA IND ð Þ¼ 0:44- LOGðGVAMANNOÞ 0:43- LOG PNGA IND=PGDPIND ð 100 0:08- LOG PHFO IND=PGDPIND- 100 0:14- DP1997 þ ECT DNGA IND ð7:311Þ LOG DOTH ð Þ¼ IND 0:06- LOGðGVAMANNOÞ 7:312

#### 7.7.1.2 Transport

$$\begin{aligned} \text{LOG(DDIS\ TRA)} &= 0.81^\* \text{ LOG(GVANOIL)} \\ &- 0.22^\* \text{ LOG(PDIS\ TRA/PGDPNOIL^\* 100)} \\ &+ 0.08^\* \text{ LOG(PGAS\ TRA/PGDPNOIL^\* 100)} - 7.56 \\ &+ \text{ECT\\_DDIS\\_TRA} \\ \text{LOG(DGAS\ TRA)} &= 0.20^\* \text{ LOG(GVANOIL)} \\ &- 0.21^\* \text{ LOG(PGAS\ TRA/CPI^\* 100)} \\ &+ 1.23^\* \text{LOG(POP)} - 10.97 + \text{ECT\\_DGAAS\ TRA} \end{aligned} \tag{7.314}$$

$$\begin{aligned} \text{LOG}(\text{DKER\\_TRA}) &= 0.31^\* \, \text{LOG}(\text{GVANOIL}) \\ &- 0.10^\* \, \text{LOG}(\text{PKER\\_TRA}/\text{PGDPTRACOM}^\* \, 100) \\ &- 3.83 + \text{ECT\\_DKRR\\_TRA} \end{aligned} \tag{7.315}$$

#### 7.7.1.3 Residential

$$\begin{aligned} \text{LOG}(\text{DELE\\_RES/POP}) &= 0.39^\* \text{ LOG}(\text{DI/POP}) \\ &- 0.32^\* \text{ LOG}(\text{PELE\\_RES/CONs/CP1}^\* \ \ \ \text{100}) \\ &+ 0.001^\* \text{ CDD}-8.79 - 0.26^\* \text{ DST1998} \\ &+ 0.25^\* \text{ DP1997} + \text{ECT\\_DELE\\_RES} \\ \text{LOG}(\text{DKER\\_RES/POP}) &= 0.26^\* \text{ LOG}(\text{DI/POP}) \\ &- 0.25^\* \text{ LOG}(\text{PKE\\_RES/CPI}^\* \ \ \ \text{100}) \\ &+ 0.29^\* \text{ LOG}(\text{PELE\\_RES/CPI}^\* \ \ \ \text{100}) \\ &- 12.70 + 0.72^\* \text{ DPI991} + \text{ECT\\_DERR\\_RES} \\ \text{LOG}(\text{DLPG\\_RES}) &= 0.28^\* \text{ LOG}(\text{GDP}) - 0.13^\* \text{ LOG}(\text{PLPG\\_RES/PGP1}^\* \ \ \text{100}) \end{aligned} (7.317)$$

$$\text{LOG(DLPG\\_RES)} = 0.28^\circ \text{LOG(GDP)} - 0.13^\circ \text{ LOG(PLPG\\_RES/PGDP 100)}$$

$$+ 0.28^\circ \text{LOG(PELE\\_RES\\_CONS/CPI}^\* 100) - 4.95$$

$$+ 0.35^\circ \text{DP2011} + \text{ECT\\_DLPG\\_RES} \tag{7.318}$$

#### 7.7.2 Commercial, Government and Agriculture

$$\text{LOG(DELE\\_COMM)} = 0.44^\circ \text{LOG(GVADIS + GVATRACOM + GVAFIBU}$$

$$+ \text{GVAOTHS} + \text{GVACON}$$

$$-0.15^\circ \text{LOG(PELE\\_COMM/CPI^\circ 100)}$$

$$-0.17^\circ \text{LOG(FDIS + IFRACOX + IFIBU)}$$

$$+ \text{IFOTHS} + \text{IFCON}) - 6.93 + 0.08^\circ \text{OTPEND}$$

$$-0.11^\circ \text{ DBT2017} + \text{ECT\\_DLEE\\_COMM}$$

$$\begin{aligned} \text{LOG(DELE\\_GOV)} &= 0.57^{\circ} \text{LOG(GVAGOV)} \\ &- 0.12^{\circ} \text{LOG(PELE\\_GOV/PGDPGOV^{\circ} \ 100)} \\ &- 7.34 + 0.03^{\circ} \text{(OTREND)} + 0.169704273369^{\circ} \text{ DP2018} \\ &+ \text{ECT\\_DLE\\_GOV} \\ \text{LOG(DELE\\_AGR)} &= 1.86^{\circ} \text{LOG(GVAGR)} \\ &- 0.25^{\circ} \text{LOG(PELE\\_AGR/PGDPGR}^{\circ} \ 100) \\ &- 0.09^{\circ} \text{ LOG(FAGB)} \\ &+ 0.34^{\circ} \text{ LOG(WAGR/PGDPGR^{\circ} \ 100)} \\ &- 22.01 + 0.29^{\circ} \text{DPI2009} + \text{ECT\\_DLE\\_AGR} \ (7.321) \end{aligned}$$

#### 7.7.2.1 An Identify for Electricity Supply

$$\text{ELE\\_STOT\\_KSA} = (\text{DCOIL\\_U} + \text{DDIS\\_U} + \text{DHYO\\_U})$$

$$+ \text{DNGA\\_U})^\* \text{ELE\\_EF} + \text{ELE\\_SS}$$

$$+ \text{DIS\\_ELE\\_STOT\\_KSA} \tag{7.322}$$

### 7.7.3 Identities to Calculate Total Energy Demand in TOE for Customer Types and the Kingdom

DEN TOT IND ¼ DCOIL IND þ DDIS IND þ DHFO IND

$$\text{+ DOTH} \underline{\text{IND}} + \text{DNGA} \underline{\text{IND}} + \text{DELE} \underline{\text{IND}} \qquad (7.323)$$

DEN TOT TRA ¼ DGAS TRA þ DDIS TRA þ DKER TRA ð7:324Þ

$$\text{DEOTH\\_TRA} = \text{DDIS\\_TRA} + \text{DKER\\_TRA} \tag{7.325}$$

$$\text{DEN\\_TOT\\_RES} = \text{DELE\\_RES} + \text{DLPG\\_RES} + \text{DKER\\_RES} \qquad (7.326)$$

$$\text{DEN\\_TOT\\_CGA} = \text{DELE\\_COMM} + \text{DELE\\_GOV} + \text{DELE\\_AGR} \quad (7.327)$$

$$\begin{array}{c} \text{DEN\\_TOT\\_KSA} = \text{DEN\\_TOT\\_IND} + \text{DEN\\_TOT\\_TRA} \\ + \text{DEN\\_TOT\\_RES} + \text{DEN\\_TOT\\_CGA} \\ \end{array} \tag{7.328}$$

$$\text{DELE\\_TOT\\_KSA} = \text{DELE\\_RES} + \text{DELE\\_IND} + \text{DELE\\_COMP} \quad (7.329)$$

$$+ \text{DELE\\_GOV} + \text{DELE\\_AGR} + \text{DELE\\_OTH}$$

### 7.7.4 Identities to Calculate Energy Demand in Million SAR by Customer Type

#### 7.7.4.1 Industry

$$\text{'CCOL\\_IND} = \text{DCOIL\\_IND^\* } \text{'PCOIL\\_IND} \tag{7.330}$$

$$\text{'CDIS\\_IND} = \text{DDIS\\_IND}^\* \text{PDIS\\_IND} \tag{7.331}$$

$$\text{CHFO\\_IND} = \text{DHFO\\_IND^\*} \,\text{PHFO\\_IND} \,\tag{7.332}$$

$$\text{TOTH}\\_\text{IND} = \text{DOTH}\\_\text{IND}^\* \text{ POTH}\\_\text{IND} \tag{7.333}$$

$$\text{CNGA\\_IND} = \text{DNGA\\_IND}^\* \text{ PNGA\\_IND} \tag{7.334}$$

$$\text{CELE\\_IND} = \text{DELE\\_IND}^\* \text{PELE\\_IND} \tag{7.335}$$

$$\text{CEN\\_TOT\\_IND} = \text{CCOL\\_IND} + \text{CDIS\\_IND} + \text{CHFO\\_IND}$$

$$+\text{COOH}\,\underline{\text{IND}} + \text{CNGA}\,\underline{\text{IND}} + \text{CELE}\,\underline{\text{IND}} \qquad (7.336)$$

#### 7.7.4.2 Residential

$$\text{LLPG\\_RES} = \text{DLPG\\_RES}^\* \text{PLPG\\_RES} \tag{7.337}$$

$$\text{CKER\\_RES} = \text{DKER\\_RES}^\* \text{ PKER\\_RES} \tag{7.338}$$

$$\text{CELE\\_RES} = \text{DELE\\_RES}^\* \text{PELE\\_RES} \tag{7.339}$$

$$\text{CEN\\_TOT\\_RES} = \text{CLPG\\_RES} + \text{CKER\\_RES} + \text{CELE\\_RES} \qquad (7.340)$$

7.7.4.3 Transport

$$\text{CGAS\\_TRA} = \text{DGAS\\_TRA^\* \gets PGAS\\_TRA} \tag{7.341}$$

$$\text{CDIS\\_TRA} = \text{DDIS\\_TRA}^\* \text{PDIS\\_TRA} \tag{7.342}$$

$$\text{CKER\\_TRA} = \text{DKER\\_TRA\\_PKRR\\_TRA} \tag{7.343}$$

$$\text{CEN\\_TOT\\_TRA} = \text{CGAS\\_TRA} + \text{CDIS\\_TRA} + \text{CKER\\_TRA} \qquad (7.344)$$

### 7.7.5 Commercial and Public Services and Agriculture and Forestry

$$\text{CELE\\_COMM} = \text{DELE\\_COMM\* } \underline{\text{PELE\\_COMM}} \qquad (7.345)$$

$$\text{CELE\\_GOV} = \text{DELE\\_GOV}^\* \text{PELE\\_GOV} \tag{7.346}$$

$$\text{CELE\\_AGR} = \text{DELE\\_AGR\*PELE\\_AGR} \tag{7.347}$$

$$\text{CEN\\_TOT\\_CGA} = \text{CELE\\_COMM} + \text{CELE\\_GOV} + \text{CELE\\_AGR} \quad (7.348)$$

### 7.7.6 Identity to Calculate Energy Demand in Million SAR for the Kingdom


### 7.8 CO2 Emissions Block Equations and Identities

#### 7.8.1 Industry

$$\text{'}\text{'CO}\_2\text{'}\text{'}\text{'IND\\_IND} = \left(\text{'}\text{'COIL\\_IND}^\*10^{6^\circ}7.33\right) \* 0.43\tag{7.353}$$

$$\text{'CO}\_2\text{-IDS\\_IND} = \left(\text{DDIS\\_IND}^\*10^{6^\circ}0.99^\circ \text{7.5}^\*42\right) \* 0.01\tag{7.354}$$

$$\text{CO}\_2\\_\text{HFO}\\_\text{IND} = \left(\text{DHFO}\\_\text{IND}^\*10^6\,6.7\right) \ast 0.43\tag{7.355}$$

$$\text{CO}\_2\text{ NGA\\_IND} = \left(\text{DNGA\\_IND}^\*10^{6^\circ}\text{ 39.2}\right) \* 0.05\tag{7.356}$$

$$\text{'CO}\_2\text{ ELE}\\_\text{IND} = \left(\text{DELE}\\_\text{IND}^\*10^6\,11.63\right) \* 0.65\tag{7.357}$$

#### 7.8.2 Transport

$$\text{TCO}\_2\text{ DIS\\_TRA} = \left(\text{DDIS\\_TRA}^\* 10^6 \text{ $0.99^\*7.5^4$ }\right) \* 0.01\tag{7.358}$$

$$\text{CO}\_2\text{-GAS}\_4\text{TRA} = \left(\text{DGA}\\_\text{TRA}^\*10^6 \, 0.95^\* 8.5^\* 42\right) \* 0.01\tag{7.359}$$

$$\text{TCO}\_2\\_\text{KER}\\_\text{TRA} = \left(\text{DKER}\\_\text{TRA}^\*10^6 \, 0.95^\* \, 7.8^\* 42\right) \* 0.01\tag{7.360}$$

#### 7.8.3 Residential

$$\text{CO}\_2\text{ ELE\\_RES} = \left(\text{DELE\\_RES}^\* 10^{6^\circ} 11.63\right) \* 0.645\tag{7.361}$$

$$\text{GCO}\_2\text{HER\\_RES} = \left(\text{DKER\\_RES}^\*10^6 \text{0.95}^\*7.8^\*42\right) \* 0.01\tag{7.362}$$

$$\text{'CO}\_2\text{ LPG\\_RES} = \left(\text{DLPG\\_RES}^\*10^{6^\circ}0.89^\*11.6^\*42\right) \* 0.24 \tag{7.363}$$

### 7.8.4 CO2 from Commercial, Government, Agriculture, and Other Electricity Use

$$\text{CO}\_2\text{ ELE }\underline{\text{COMM}} = \left(\text{DELE }\underline{\text{COMM}}^\*10^6\,11.63\right) \ast 0.65\tag{7.364}$$

$$\text{CO}\_2\text{ELE\\_GOV} = \left(\text{DELE\\_GOV}^\* 10^{6^\circ} 11.63\right) \ast 0.65\tag{7.365}$$

$$\text{'CO}\_2\text{ ELE}\\_\text{AGR} = \left(\text{DELE}\\_\text{AGR}^\*10^6 \, 11.63\right) \* 0.65\tag{7.366}$$

$$\text{CO}\_2\text{ELE\\_OTH} = \left(\text{DELE\\_OTH}^\*10^{6^\circ}11.63\right) \* 0.65\tag{7.367}$$

### on the Fuel Mix Components 7.8.5 CO2 from Total Electricity Generated and Based

$$\text{'} \text{'} \text{CO}\_2 \text{'} \text{LE} \text{'} \text{'} \text{TOT\\_KSA} = \left( \text{'} \text{LE} \text{'} \text{'} \text{TOT\\_KSA}^\* 10^6 \text{'} 11.63 \right) \* 0.65 \qquad (7.368)$$

$$\text{CO}\_2\text{-COIL\\_U} = \left(\text{DCOIL\\_U}^\* 7.33^\* 10^6\right) \* 0.43\tag{7.369}$$

$$\rm CO\_2\\_DIS\\_U = \left(\rm DDIS\\_U^\*10^6\,0.99^\*7.5^42\right) \* 0.01\tag{7.370}$$

$$\text{'CO}\_2\text{-HFO}\_\text{'}\text{U}=\left(\text{DHFO}\\_\text{U}^\*10^{6^\circ}1.04^\*6.7\right)\*0.43\tag{7.371}$$

$$\text{'CO}\_2\text{-NGA}\_\text{-U} = \left(\text{DNGA}\_\text{-U}^\* 10^{6^\circ} 39.2\right) \* 0.05\tag{7.372}$$

$$\begin{array}{c} \text{CO}\_2\text{-}\underline{\text{E}}\text{.} \underline{\text{K}}\text{SA\\_PFM} = \text{CO}\_2\underline{\text{COIL\\_U}} + \text{CO}\_2\underline{\text{DIS\\_U}}\\ + \text{CO}\_2\underline{\text{HFO\\_U}} + \text{CO}\_2\underline{\text{NGA\\_U}} \end{array} \tag{7.373}$$

### 7.8.6 Total CO2 Emissions by Sector

$$\begin{aligned} \text{CO}\_{2}\text{EN\\_TOT\\_IND} &= \text{CO}\_{2}\text{CONL\\_IND} + \text{CO}\_{2}\text{DIS\\_IND} \\ &+ \text{CO}\_{2}\text{HFO\\_IND} + \text{CO}\_{2}\text{NGA\\_IND} + \text{CO}\_{2}\text{ELE\\_IND} \end{aligned} \tag{7.374}$$

$$\text{CO}\_2\text{EN\\_TOT\\_TRA} = \text{CO}\_2\\_\text{GAS\\_TRA} + \text{CO}\_2\\_\text{DIS\\_TRA} + \text{CO}\_2\\_\text{KER\\_TRA} \tag{7.375}$$

$$\text{CO}\_2\text{EN\\_TOT\\_RES} = \text{CO}\_2\text{ELE\\_RES} + \text{CO}\_2\text{LPG\\_RES} + \text{CO}\_2\text{KER\\_RES} \tag{7.376}$$

$$\text{CO}\_2\text{EN\\_TOT\\_CGAO} = \text{CO}\_2\text{ELE\\_COMM} + \text{CO}\_2\text{ELE\\_GOV} \quad (7.377)$$

$$+ \text{CO}\_2\text{ELE\\_AGR} + \text{CO}\_2\text{ELE\\_OTH}$$

### 7.8.7 Total CO2 Emissions from Oil Use and for the Kingdom

$$\text{CO}\_2\text{OILUSE} = \left(\text{OILUSE}^\* \mathbf{365}^\* 10^6\right) \* 0.43\tag{7.378}$$

$$\text{CO}\_2\text{EN\\_TOT\\_KSA} = \text{CO}\_2\text{EN\\_TOT\\_IND} + \text{CO}\_2\text{EN\\_TOT\\_TRA}$$

$$+ \text{CO}\_2\text{EN\\_TOT\\_RES} + \text{CO}\_2\text{EN\\_TOT\\_CGAO}$$

### 7.9 Population and Age Cohorts Block Equations and Identities

#### 7.9.1 Identities for Saudis and Non-Saudis

$$\text{POPS} = \text{POPSM} + \text{POPSF} \tag{7.380}$$

$$\text{POPNS} = (\text{POP} - \text{POPS}) + \text{DIS\\_POPNS} \tag{7.381}$$

#### 7.9.2 Identities for Age Cohorts

$$\text{POP014} = \text{POPM014} + \text{POPF014} \tag{7.382}$$

$$\text{POP1519} = \text{POPM1519} + \text{POPF1519} \tag{7.383}$$

$$\text{POP2024} = \text{POPM2024} + \text{POPF2024} \tag{7.384}$$

### POP2529 ¼ POPM2529 þ POPF2529 ð7:385Þ


$$\text{POP3539} = \text{POP3539} + \text{POPF3539} \tag{7.387}$$

$$\text{POP4044} = \text{POPM4044} + \text{POPF4044} \tag{7.388}$$


$$\text{POP6064} = \text{POPM6064} + \text{POPF6064} \tag{7.392}$$

$$\text{POP65A} = \text{POPM65A} + \text{POPF65A} \tag{7.393}$$

¼ þ þ þ þ þ þPOP4044 þ POP4549 þ POP5054 þ POP5559 þ POP6064 þPOP65A þ DIS POP POP POP014 POP1519 POP2024 POP2529 POP3034 POP3539

> ð Þ 7:394

#### 7.9.3 Identities for Working Age Group

POPW ¼ POPWF þ POPWM ð7:395Þ

¼ þ þ þ þ þ POPF4044 þ POPF4549 þ POPF5054 þ POPF5559 þ POPF6064 ð7:396Þ POPWF POPF1519 POPF2024 POPF2529 POPF3034 POPF3539

POPWM ¼ POPM1519 þ POPM2024 þ POPM2529 þ POPM3034þ þ þ þ þ POPM5559 þ POPM6064 ð7:397Þ POPM3539 POPM4044 POPM4549 POPM5054

Open Access This 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, credit to the original author(s) and the source, provide a link to the Creative Commons license and adaptation, distribution and reproduction in any medium or format, as long as you give appropriate indicate if changes were made.

The 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.

## Chapter 8 KGEMM Simulations

As mentioned earlier, MEMs are evaluated and validated using in-sample and outof-sample simulations, and policy analysis, among other validation methods. In this section, we run KGEMM for in-sample forecasting and out-of-sample policy analysis to evaluate its predictive ability.<sup>1</sup> Hasanov and Joutz (2013) provide an overview of the literature that covers in-sample and out-of-sample forecasts and other methods for evaluating the predictive ability of MEMs. This includes Calzolari and Corsi (1977), Beenstock et al. (1986), Klein et al. (1999), Fair (1984, 1994, 2004), Bardsen and Nymoen (2008).

MEMs can be run using either the static simulation method, in which the model takes historical lagged values, or the dynamic simulation method, in which the model takes predicted lagged values. The predicted lagged values are the combination of historical lagged values and any errors in the model's predictions. The smaller the prediction errors of the model, the better the model can perform/approximate in simulations. Therefore, it would be advisable to run the model with dynamic simulation to see how significant or insignificant its prediction errors are in simulating the in-sample or out-of-sample values of the endogenous variables. Static, dynamic, deterministic, and stochastic MEM simulation methods have been comprehensively discussed in Klein et al. (1999) and Fair (1984, 1994, 2004), among others. This section employs the dynamic simulation method.

<sup>1</sup> The literature discusses using both long- and short-run equations/models for forecasting and projections purposes: Hara et al. (2009) and Yoshida (1990), among others, note that ECM-based MEMs provide realistic projections. Engle et al. (1989) compare forecasts from short-, long, and ECM models. Hendry et al. (2019) discuss that both level- and difference-based models should be considered in forecasting/projections. Fanchon and Wendel (1992) finds VAR in level outperforming VEC in first difference. Engle and Yoo (1987) finds the same for the short-horizon forecasting. We use the long-run version of KGEMM because our out-of-sample simulations span 9 years and because of the discussion in Appendix A.5.1. Note that Weyerstrass et al. (2018); Khan and ud Din (2011); Weyerstrass and Neck (2007), Musila (2002), Fair (1979, 1993), among others also used long-run version of their macroeconometric models in their policy analyses and simulations.

### 8.1 In-Sample Simulation

As mentioned previously, an in-sample simulation exercise is an evaluation/validation method to check how well a model can approximate historical data. We run KGEMM for the period 1999–2019 to check its in-sample predictive ability of approximating the Saudi Arabian historical data.2

The results of the in-sample simulations are plotted in Appendix D. Figures D.1, D.2, D.3, D.4, D.5, and D.6 in the appendix illustrate selected key endogenous variables from each block. Interested readers can refer to the details of the in-sample forecasting for each variable in Appendix D. The figures show that the model closely approximates the historical time path of the endogenous variables under consideration. It performs especially well in capturing the historical turning points and sudden changes in data.

It can be concluded that KGEMM's in-sample predictive ability for the historical values of the endogenous variables is quite high. The literature suggests that if a model successfully approximates historical time paths of the variables, then there is a high probability that the model will also perform well in out-of-sample simulations or policy analyses.

### 8.2 Out-of-Sample Simulations

This sub-section simulates KGEMM to evaluate the economic, energy, and environmental effects of domestic and global changes. It should first be stated that these simulations aim to illustrate the model's ability in addressing domestic and global changes through the linkages across its blocks. Two things have to be noted in this regard: (i) for economic, energy and CO2 emissions variables, input values are just the authors' calculations and output values are the results of the KGEMM simulations both based on many assumptions (which might not be adequate representations of the real life). Therefore, either input or output values in the simulations do not represent any official and or policy views at all. (ii) The simulations conducted here do not aim to evaluate any policy options. SV2030 provides a number of targets and initiatives to consider, related to fiscal stance, energy transitions (e.g., domestic consumption of fossil fuels and renewable deployments), competitiveness, investments, non-oil export expansion, and diversification among others. Hasanov et al. (2020) simulated KGEMM to assess the effects of domestic energy price reform (and also fiscal reform at some extent). Also, Hasanov et al. (2022c) simulated KGEMM to assess the non-oil exports effects of the expansions in the non-oil tradable and non-tradable sectors. Moreover, Hasanov and Razek (2022) assessed the positive impacts of the Public Investment Fund's new investment strategy for 2021–2025 on the Saudi competitiveness by using KGEMM. Furthermore, Elshurafa et al. (2022)

<sup>2</sup> The reason for starting in 1999 is that data for some variables in the model are only available from that year.

used KGEMM to quantify the macroeconomic and sectoral effects of diesel displacement from the agriculture sector. Lastly, Hasanov et al. (2022a) couple KGEMM with power generation models to evaluate economic, energy, and environmental effects of renewable deployments at the distributed generation and utility farm scales.

Given the above-mentioned KGEMM simulations done already, we do not want to repeat the same or very similar exercises here. Rather to provide readers with new insights here, we simulate KGEMM to assess the effects of the following scenarios: changes in (i) global oil prices, (ii) foreign direct investments to Saudi Arabia, and (iii) renewable penetration in the electricity generation. The scenario analyses cover the period 2022–2030. We simulate two scenarios in each out-of-sample exercise business as usual (BaU) scenario and one of the three scenarios (denoted by S1, S2, and S3) mentioned above. The simulations end in 2030 to be in line with the time span of the SV2030 although the model can be solved till 2035. In the BaU, it is assumed that the Saudi Arabian economy moves forward as it did in 2021. Precisely, the BaU scenario includes the fiscal reform items (in particular, the implementations of expat levy in 2017, and 5% and 15% VAT rates in 2018 and 2020, respectively)<sup>3</sup> and domestic energy price increases in 2016 and 2018, and does not include the goal of having 50% of renewable and 50% of natural gas in power generation by 2030 because we simulate this here as a scenario (S3) in 8.2.3. Thus, the differences between the BaU scenario and given scenario will be stemmed only from the inputs in a given scenario. As an output of each exercise, we consider not only economic indicators but also energy and environmental indicators. The idea is to assess the effects of given changes from the sustainable development perspectives by considering economic-energy-environmental dimensions. To this end, we consider non-oil value added (GVANOIL), non-oil exports (XGNOIL), non-oil government budget revenues (GREVNOIL), households' disposable income (DI), total energy consumption (DEN\_TOT\_KSA), and total CO2 emissions from energy consumption in the kingdom (CO2\_EN\_TOT\_KSA). The rationale behind this selection would be as follows: the first three indicators show developments in the non-oil sector, its exports and revenue collection, which are the main economic streamlines in the SV2030 and its realization programs. A wellbeing of the nation and hence increasing disposable income of the households is the end goal of any economic policies. Lastly, energy consumption and associated emissions are considered from the United Nations sustainable development goals standpoint. The description of each variable is given in Appendix B.<sup>4</sup> We discuss policy background, inputs, and outputs of each scenario in the next sub-sections. We keep our discussions very brief, since we have three scenarios to examine and since out-of-sample simulations are not the entire objective of this book.

<sup>3</sup> BaU scenario also accounts for other fiscal implementations in 2017 such as Umrah and Hajj visa fees and other visa fees.

<sup>4</sup> As Appendix B documents, GVANOIL, XGNOIL, GREVNOIL, DI, DEN\_TOT\_KSA are measured in million scale while CO2\_EN\_TOT\_KSA is measured in metric ton. For readers' convenience, we scaled up these measures to billions in the graphical illustrations.

### 8.2.1 The Effects of Global Oil Price Changes

As in any oil-exporting country, oil prices and related revenues play an important role in economic activities including non-oil sector in Saudi Arabia. Oil constitutes large shares in total exports, budget revenues, and total economy in the Saudi economy as mentioned earlier in this book. There is a consensus in the literature that oil-related revenues are important for the development of the non-oil sector mainly through fiscal spending. Given this, we simulate KGEMM to examine effects of the international price of Arabian light crude oil (WPO\_AL) on economic, energy, and environmental relationships of the Kingdom. As an input, the international price of Arab light crude oil in Scenario 1 (S1) is increased by 25% in each year of the 2022–2030 period compared to the values in the BaU scenario, as shown in Fig. 8.1. 5

Figure 8.2 documents the effect of this increase on the selected indicators.

The graphs in the figure demonstrate increases in the selected indicators in S1 compared to BaU when the Arabian light crude oil international price increases. This is pretty much expected given the nature of the Saudi economy as discussed above. High oil price increases oil income, which is channeled into the rest economy through increased oil sector's demand and increased government demand for other

Fig. 8.1 Arabian light crude oil international price, US\$ per barrel

<sup>5</sup> The reason we considered 25% is because under this setup, Arabian light price reaches up to US \$100 per barrel in 2030.

Fig. 8.2 Projections of the selected indicators in S1 and BaU scenarios, 2021–2030

sectors' goods and services as Eqs. (7.1)–(7.26) describe. This increases total demand and sectoral economic activities (see Eqs. (7.27)–(7.39) and (7.61)– (7.76)). This is the first round and demand-side effect, and it translates into the supply-side effect over time mainly through investment-capital accumulation as the Eqs. (7.43)–(7.51) and (7.114)–(7.138) explains. XGNOIL increases because domestic production measured by GVANOIL increases (see Eq. (7.193). An expansion in economic activities leads to more employment and wages and resultantly, the disposable income of the household's raises (see Eqs. (7.264)–(7.274) and (7.285)–(7.291), and (7.306) and (7.109)). Expanded economic activities demand more energy (see Eqs. (7.61)–(7.76) and (7.307)–(7.321)) in S1 compared to BaU. Numerically, calculated implied elasticities for the average of 2022–2030 show that a 1% increase in the international price of Arabian light crude oil leads to a 0.4%, 0.4%, 0.2%, and 0.14% increase in GVANOIL, XGNOIL, GREVNOIL, and DI, respectively, while DEN\_TOT\_KSA and CO2\_EN\_TOT\_KSA only increase by 0.2% each.<sup>6</sup>

<sup>6</sup> Following the macroeconometric modeling literature, period average implied elasticity (e) is calculated as the ratio of mean percentage change deviation of an output variable to the mean

percentage change deviation of an input variable. Precisely, <sup>e</sup> <sup>¼</sup> ð1 <sup>T</sup> - PT t OV St OV BaUt - <sup>100</sup> <sup>100</sup>Þ=ð<sup>1</sup> <sup>T</sup> - PT t IV St IV BaUt -<sup>100</sup> <sup>100</sup><sup>Þ</sup> . Where, OV \_ St and OV \_ BaUt are

¼ the values of the output variable (say GVANOIL) in a given scenario (say S1) and in the BaU scenario, respectively; IV \_ St and IV \_ BaUt are the values of the input variable (say WPO\_AL) in a given scenario (say S1) and in the BaU scenario, respectively; t denotes time, that is year; T indicates the total number of years. In our case, t changes from 2022 to 2030 and T 9.

### 8.2.2 The Effects of Foreign Direct Investments Inflow to Saudi Arabia

The National Investment Strategy (NIS) has been announced in October, 2021. The strategy highlights a significant increases in foreign direct investments (FDI) inflow and domestic investments in the coming years. Precisely, the cumulative total (FDI inflow+Domestic) investment of 12.4 SAR trillion in KSA during 2021–2030 (Jadwa Investment 2021). 388 SAR billion of FDIs inflow and 1.65 SAR trillion of domestic investment and hence the total of 2 SAR trillion in 2030 have been indicated.7 Table 8.1 records the shares of the sources in the total investment by 2030.

The strategy targets to raise FDI inflow from SAR 17 billion in 2019 to SAR 388 billion in 2030—this is about 23 times increase in 10 years. Likewise, it is targeted to increase domestic investments and overall investments by 2.6 times and 3.1 times, respectively, during 2019–2030.

Obviously, the NIS and associated domestic and foreign investment targets have large policy implications regarding their impacts on the development of Saudi economy. They also have implications for energy and environmental dimensions of the Kingdom. It was also discussed that accomplishing the above-mentioned targets necessitates well-designed development measures determined by the NIS and this is based on four main milestones, namely, enhancing investment opportunities, targeting different investor types, diversifying financing options, and improving competitiveness.8 We do not discuss more policy implications of the NIS here to save space, but vividly this deserves a scenario analysis to quantify its impact on the Kingdom. To do so, we focused on the FDI inflow aspect of the NIS. This is because FDI inflow is exogenous to the Saudi economy, and it is considered so in the KGEMM framework. Thus, it is more realistic to simulate the impact of FDI inflow, as a global factor, on Saudi Arabia. As the input for scenario 2 (S2), we obtained nominal values of FDI inflow for 2022–2030 in SAR billion from the online media


Table 8.1 Breakdown of the total investment by stakeholder by 2030

Source: Modified from Jadwa Investment 2021

<sup>7</sup> https://www.spa.gov.sa/viewfullstory.php?lang¼en&newsid¼2294440#2294440, https://www. spa.gov.sa/viewfullstory.php?lang¼ ¼ en&newsid <sup>2295200</sup> <sup>8</sup>

¼ ¼ See https://www.spa.gov.sa/viewfullstory.php?lang en&newsid 2295200

Fig. 8.3 Projected values of FDIs inflow in the S2 and BaU scenarios, US\$ billions

source of https://www.argaam.com/en/article/articledetail/id/1503922 and then converted them to US\$ billions using the exchange rate.<sup>9</sup> Figure 8.3 illustrates the FDI inflow values in the S2 scenario (FDI\$IN\_Z S2) and those in the BaU scenario (FDI\$IN\_Z BaU) over the simulation period.

The figure shows that values for the FDIs inflow are significantly large in S2 scenario compared to the values in the BaU scenario. The difference between the FDIs inflow values in two scenarios increases from about 3 times in 2022 to 10 times in 2030; these are large numbers and should have sizeable growth effects. Figure 8.4 reports the results for the selected indicators.

The graphs in the figure convey heterogeneous information. In other words, for GVANOIL, XGNOIL, DEN\_TOT\_KSA, and CO2\_EN\_TOT\_KSA, the deviations of the S2 scenario values from the BaU scenario values are considerable, whereas that for the GREVNOIL and DI are not. Numerically, in 2030, the percentage change deviations of the S2 values from the BaU values for GVANOIL, XGNOIL, DEN\_TOT\_KSA, and CO2\_EN\_TOT\_KSA are 23%, 25%, 14%, and 14%, respectively. While the deviations for GREVNOIL and DI are 4% and 1%, respectively. At the first glance, small deviations in the case of latter variables might be seen puzzling as the increases in FDIs inflow in the S2 scenario are quite large compared to that of the BaU scenario. However, a closer look reveals out a few points that are worth considering. First, historically, the size of FDIs inflow in total domestic investments

<sup>9</sup> Alternatively, we can calculate FDI inflow values for each year of the simulation period using the announced value of SAR 388 billion in 2030 and the historical value of SAR 17 billion in 2019, and extrapolate different development paths. However, we chose to use the values already calculated by www.argaam.com because we believe this media source has more information content for its calculations.

Fig. 8.4 Projections of the selected indicators in S2 and BaU scenarios, 2021–2030

and thus its role in the economy was quite limited and even diminishing over time. In terms of numbers, the size rose from 19% in 2005 to a peak of 33% in 2008 and has been steadily declining since then, falling to just below 3% in 2019. A simple cointegration analysis shows that the magnitude of the impact of the FDIs inflow on the non-oil activities was very small compared to that of the domestic investments.<sup>10</sup> Second, if we calculate the implied elasticities of the most largely impacted variables, i.e., XGNOIL and GVANOIL, with respect to FDIs inflow, they would be as small as 0.0216 and 0.0197, respectively, for 2022–2030, pointing to minor growth effects of FDIs inflow for the non-oil activities. Third, another reason why GREVNOIL and FDI are less affected by the increase in FDIs inflows may be due to outflow remittances. Eq. (7.109) illustrates that net national disposable income declines when outflow remittances (REMOF) increase. In this regard, in 2030, percentage deviation of REMOF value in S2 from the value in BaU is 25%, indicating a quite large amount of leakage. Fourth, the growth in the other component of the net national disposable income, namely, labor compensation (LABCOMP), which positively affects it, is very small. Numerically, the percentage deviation of LABCOMP value in S2 from the value in BaU is as small as 0.2% in 2022 and it grows to 1.5% only in 2030. This is not surprising because of the two reasons: the expansion in the economic activity in this scenario is investment-driven and the labor elasticity of output is smaller than that of capital elasticity for a number of the non-oil activity sectors, namely, manufacturing less petrochemical; transportation and communication; petrochemical; agriculture; utility; retail, wholesale, hotels, and catering (see identities for the estimated potential output equations by economic activity sector in Sect. 7.1). Both reasons indicate a limited role of employment in non-oil economic activities in this scenario. In addition, the increase

<sup>10</sup>A long-term, cointegrated, regression for the period 2007–2019 was estimated as LOG (GVANOIL) ¼ 0.036\*LOG(FDI\$IN\_Z\*RXD/PIF\*100) + 0.281\*LOG((IFNOIL)-(FDI \$IN\_Z\*RXD/PIF\*100)) + 7.857 + 0.034\*@TREND.

in non-oil sector employment (ETNOIL) in the S2 scenario compared to the BaU scenario is 0.6% in 2022 and 4.8% in 2030. As a result, limited role of employment coupled with its limited increase makes LABCOMP to increase with a minor magnitude in the S2 scenario. Moreover, the third component of DI, namely, government transfers to households (GCGPE) does not increase in the S2 scenario compared to the BaU scenario and even gets slightly small due to the decrease in government oil revenues (GREVOIL) as explained below. Fifth, low growth in employment and wages and thus in net national disposable income and household consumption makes GREVNOIL to grow less too as the VAT collections is one of its main components. Sixth, another reason for a small increase in GREVNOIL would be none to negative growth rates of government expenditure (GEXP). Numerically, its growth in S2 compared to BaU is on average 0.4%. The main reason for it is that an expansion in economic activity demands more energy to be consumed domestically (see changes in DEN\_TOT\_KSA), and this results in less amount of oil being available for export (XOILC) since the oil production (OILMBD) in KGEMM is treated as exogenous due to the global oil market conditions such as OPEC+ agreements (see Eq. (7.199)). Consequently, government oil revenues (GREVOIL) in S2 decrease by an average of 2.5% over 2022–2030 compared to BaU. Since the share of GREVOIL in total government revenues (GREV) is considerably large (the average of 1969–2019 was almost 80%), the latter declines slightly in S2 compared to BaU although GREVNOIL increases in S2 as figure above shows. As a result, GEXP declines and it is one of the key drivers of economic growth including non-oil economic activity, which is the base for GREVNOIL collections.

In the conclusion of the scenario 2 exercise, some insights can be derived from the simulation. The FDIs inflow has a positive impact on the economy, but historical structure and business environment in the economy should be changed to make the magnitude of this impact larger. The simulation results here support the four main milestones (enhancing investment opportunities, targeting different investor types, diversifying financing options, and improving competitiveness) that are already considered in the NIS framework as the main policy measures to make FDIs inflow more impactful in the economy. Second, the authorities may wish to think about measures to further increase the labor contribution to output in some non-oil economic activity sectors. Investments in research and development, education, trainings, and other human capital enhancing activities might be fostered in this regard. Third, the authorities also may wish to think about measures to reduce leakages, such as outflow remittances, which would further increase aggregate demand and domestic economic activities. Increasing the expat levy rate does not seem a best measure in this regard because it might encourage foreign workers to consider another Gulf countries for work. One measure, which seems reasonable, is to further facilitate business and investment opportunities, as well as ownership and property rights for foreigners. In this regard, it is acknowledged that the government successfully implemented reforms, measures, and initiatives in line with Saudi Vision 2030 recently.<sup>11</sup> This would encourage foreign workers to invest, establish, or expand their business and own properties in Saudi Arabia. The other measure might be further development of service sectors including entertainment, such as cinema, sport games, so that foreigners spend their money domestically, which would boost economic activities through spillover effects. Fourth, none to very small growth in GEXP in the S2 scenario brings up two policy insights: (i) the share of renewable in total domestic energy consumption (DEN\_TOT\_KSA) should be increased. The point here is that increased economic activity will demand more energy, resulting in less oil to export. Increasing the share of renewable to meet domestic energy demand will displace more oil from domestic use. The government has already announced its strategy of completely displacing liquid fuels in the electricity generation and making it based on renewable and natural gas only with the shares of 50% and 50% by 2030. This would allow to save more oil that can be either exported as a crude or refined domestically for export purpose, which would bring more foreign exchange reserves, which can be used for covering the government debt or put in human capital, research and development, innovation, and other long-run drivers of productivity. Renewable deployments also will reduce CO2 emissions, which would help to achieve environmental targets. (ii) weaken the role of government spending in the development of the non-oil sector activities and putting more emphasis on the private sector. Note that this is one of the key targets of SV2030—to increase the private sector's contribution to GDP from 40% to 65% by 2030.

### 8.2.3 The Effects of Raising the Share of Renewables in Power Generation Mix to 50% by 2030

Saudi Arabian government has very comprehensive strategies aiming at increasing the share of renewables in energy consumption. Energy and Sustainability strategy is one of the important streamlines of SV2030, a roadmap for the development of the Kingdom. The National Renewable Energy Program (NREP) is an important initiative under this strategy with the purpose of increasing the share of renewable energy production, achieve a balance in the mix of local energy sources, and fulfill Saudi Arabia's obligations toward reducing CO2 emissions.<sup>12</sup> One of the key targets of NREP is achieving optimum energy mix to produce electricity.<sup>13</sup> It targets removing liquid fuel from the mix to make it comprised of renewables and gas, each with a share of 50% by 2030.<sup>14</sup> Obviously, the targeted figures make it important to assess

<sup>11</sup>https://www.argaam.com/en/article/articledetail/id/1467828

<sup>12</sup>Energy & Sustainability – Vision 2030. https://www.vision2030.gov.sa/thekingdom/explore/ energy/?msclkid¼0d24312cb65211ecb7e408e39a812fec 13https://www.moenergy.gov.sa/en/OurPrograms/Pages/default.aspx

<sup>14</sup>https://www.moenergy.gov.sa/en/OurPrograms/EnergyMix/Pages/default.aspx

their potential economic, energy, and environmental implications. Such a green energy transition brings two main benefits at least: reducing environmental pollution and gaining extra revenues from exporting displaced fossil fuels.

Given that the share of natural gas in the power mix was almost 50% in 2019 (see source in footnote 19), the target of the optimum energy mix is to increase the share of renewable to 50% by 2030. Put differently, the use of crude oil, diesel, and HFO in the electricity generation mix will be replaced by natural gas over time, so that the shares of these liquid fuels in the mix will be zero by 2030. This opens a great avenue to export this replaced liquid fuels to make more foreign exchange reserves or use them domestically (for example, crude can be used to produce refined products). These are the two options to deal with the displaced liquid fuels from the power mix, but the first option might be seen more attractive.

Obviously, increasing the share of renewables to 50% by 2030 requires a large amount of investments among other efforts. Although we have not came across any announced investment figures for this purpose, some media resources mention investing 380 billion SAR by 2030.<sup>15</sup>

In this scenario (S3), we simulate KGEMM to assess economic, energy, and environmental effects of the above given renewable initiative. First, we calculated the needed amount of renewable energy for the power mix. We considered solar given the fact that it is the main renewable energy source in Saudi Arabia so far, as discussed earlier in this book. Solar energy was 0.063 MTOE and 0.075 MTOE in 2019 and 2020, respectively, the Kingdom according to IEA (2021). We forecasted it to be 0.09 MTOE in 2021 assuming the same growth rate of 2020. We also forecasted total electricity production to be 37.74 MTOE in 2030 using different factors including efficiency of fossil fuels and contribution of renewable, i.e., solar energy. Given that it is targeted the half of the total generation will be solar (and the other half will be generated from natural gas) and considering solar energy forecasted for 2021, we extrapolated solar energy between 2022 and 2030 in S3 scenario. Next, we calculated crude oil equivalent of the projected solar energy using the power generation efficiency factor of crude oil in Saudi Arabia from the electricityrelated agencies for S3 scenario.<sup>16</sup> Finally, we extrapolated the announced investment figure for renewable energy and deflated the resulted values by investment deflator for 2022–2030. In scenario S3, we assume that domestic oil use (OILUSE) will be decreased by the crude oil values to be displaced and utility investment will be increased by the calculated real investment values. These two are the inputs that differentiates S3 from BaU. Note that we included these two inputs in the S3 as add factors (named OILUSE\_ADD and IFU\_ADD) rather treating OILUSE and IFU

<sup>15</sup>https://www.reuters.com/world/middle-east/saudi-arabia-plans-100-bln-renewables-investmentsays-minister-2021-12-13/. https://www.zawya.com/en/projects/saudi-arabia-plans-100bln-renew ables-investment-says-minister-ex3w7c6i

<sup>16</sup>The resulted crude oil values, to be displaced, in MTOE, were converted to million barrel per day using the conversion factor of (0.1486\*365). This is because we will reduce domestic oil use (OILUSE) by the calculated crude oil values, which will be displaced, and OILUSE in the model is measured in million barrel per day.

Fig. 8.5 Projected OILUSE\_ADD in million barrel per day and IFU\_ADD in SAR 2010 billion in S3 scenario, 2021–2030

Fig. 8.6 Projections of the selected indicators in S3 and BaU scenarios, 2021–2030

exogenous. This is because both variables are the endogenous variables and hence should change in line with economic activity performance in S3 scenario, but treating them exogenous would ignore this reality. Figure 8.5 illustrates calculated crude oil to be displaced and subtracted from domestic crude oil use, and renewable investment to be added into the utility sector investment.

The figure demonstrates that both variables are zero in 2021. This is to show that our scenario simulation starts in 2022 and this does not mean that either OILUSE or IFU are zero in 2021. The displaced crude oil in S3 (OILUSE\_ADD S3) is on average 15% of OILUSE in BaU in 2022–2030. In the same token, additional utility sector investment due to the renewable projects in S3 is on average 112% of IFU in BaU in the same period. These mean that there will be increases in the crude oil export revenues (associated with the displaced crude oil) and increases in utility sector investments in S3 compared BaU. We expect that they will create positive changes in the economy and environment in the S3 scenario. The results of the simulations for the selected variables are illustrated in Fig. 8.6.

We would like to point out some observations from the graphs briefly. First, all the indicators demonstrate growth, whereas CO2 emissions declines in the S3 scenario compared to the BaU scenario. Second, macroeconomic indicators increase with different paces. Numerically, on average in 2022–2030, GVANOIL S3 and XGNOIL S3 increase by 3.3% and 3.6%, respectively, while GREVNOIL S3 and DI S3 rise by 1.0% and 0.6%, respectively. GVANOIL S3 increases due to the aggregate demand increases in the short-run and due to the supply-side expansions (mainly capital stock but non-oil employment also increased albeit slight) in the long run both caused by injecting displaced crude oil revenues and renewable investments into the economy. GREVNOIL S3 rises because GVANOIL S3, i.e., non-oil economic activities increase. Increases in DI S3 are mainly caused by increases in LABCOMP S3, which grows by 1.0% on average in 2022–2030 compared to LABCOMP BaU. Third, DEN\_TOT\_KSA S3 increases on average by 2.0% compared to BaU. The increase is caused by the increasing aggregate demand and economic activities. Last, but not least, CO2\_EN\_TOT\_KSA declines in a growing pace in the S3 scenario compared to the BaU scenario. Numerically, the drop is 1.8% in 2022 and reaches to 18.2% in 2030. Apparently, the declining CO2 effect of the renewable deployments, which removes crude oil use, outpaces the increasing effect of additional fossil fuel energy consumption due to the expanded economic activities in the S3 scenario.

Acknowledgments The authors are thankful to the former KGEMM team members, Nayef Al-Musehel, Ziyad Alfawzan, Shahad Al-Arenan, Noha Abdel Razek, Waheed Olagunju, Hanadi Al Sunaid and current members, Abdulelah Darandary, Ryan Al Yamani for their contributions to the project. We also thank the participants of the EcoMod 2019 Economic Modeling and Data Science conference, in particular Jean Louis Brillet and Geoffrey Hewings, for their comments. We are also grateful to Amro Elshurafa, Andrea Carlo Bollino, Anwar Gasim, Axel Pierru, John Qualls, Fatih Karanfil, Lester Hunt, and Walid Matar for their expertise on the relationships in the energy block. We are indebted to Abdulaziz Dahlawi for his great help in data processing. Our thanks also go to Chay Allen for editing discussion paper version of this book and Evelyn Simpson for supporting library resources. The authors accept sole responsibility for any errors and omissions. The views expressed in this book are the authors' and do not necessarily represent the views of their affiliated institutions.

Open Access This 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.

The 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.

## Appendixes

### Appendix A: KGEMM Methodological Framework and Philosophy

Energy, financial, macroeconomic, econometric, forecasting, and policy models need to perform well in a complex environment with numerous interrelated actions and decisions occurring simultaneously by heterogeneous agents. Modern economies are evolving; there are both gradual and sudden structural changes and shifts due to institutional, technological, financial, international competitiveness, political, social and legal changes. The philosophy underlying the econometric methodology for the development and use of KGEMM is derived from the principles espoused by Haavelmo (1943a, b, 1944) on the theory of reduction, followed by the general-tospecific or LSE tradition developed by many including Hendry and Johansen (2015). Empirical (policy and forecasting) models are supposed to capture these factors in an environment where the data are non-stationary; the degree of misspecification is unknown for the data generation process (DGP), but it is no doubt large. The data available may be inaccurate, a proxy for theoretical constructs and an agent's decision-making criteria, produced with a lag, and subject to revision. The methodology for KGEMM begins with a focus on understanding and replicating the Saudi Arabian macroeconomy and energy.

Model building is an attempt to characterize the properties of observed energyeconomic data using simple parametric relationships, which remain reasonably constant over time, account for the findings of previous models, and are interpretable in an energy-economic sense. There are three key aspects in applied empirical model building: data properties, including integration and cointegration; dynamic specification, including the use of equilibrium correction models; and model evaluation and model design.

This Appendix is divided into four sections. The first section lays out the typology of macroeconomic modeling techniques and how their methodologies relate to one another. The second section discusses the econometric philosophy underlying KGEMM's methodology. The third section explains the econometric methodology that the KGEMM team uses to develop the behavioral equations of the model. The fourth section details other issues of the model's development and how it can be used for policy analysis.

### A Typology of the Macroeconomic Modeling Techniques

Pagan (2003a, b) lays out a useful structure for understanding the empirical modeling frontier and the strategic trade-offs faced by researchers. The frontier defines the tension between building models that are consistent or coherent theoretically and empirically. Below are two quotes from his paper and a slight modification of his frontier diagram.

Macroeconomic model building has always faced the challenge of how one trades off theoretical and empirical coherence. If one thinks of a curve which summarizes this tradeoff, where we locate on it ultimately depends on preferences. But how does one get to this point? Broadly two strategies have been employed. One is to start with statistical models that closely fit the data, and then try to impose a theoretical structure upon them – what we will term the 'bottoms-up' approach. The other is to embellish the framework laid out by some miniature theoretical model – what we will call the 'top-down' approach. The first generation of macro models tended to follow the first strategy, although in many ways the vision of the Cowles Commission was the latter. Recently there has been a return to the original vision, particularly in central banks that have attained independence and have an inflation target.

I want to start the lecture with a graph. Over the years I have found it useful to think about the activities engaged in by econometric researchers with this representation. It shows the frontier curve that connects up the degree to which our current modelling methods aim to exhibit coherence with the ideas of economic theories and the degree to which they attempt to cohere with the data. At each end of the frontier the coherence is perfect for one of these characteristics and zero for the other. Crudely speaking, we might say that economics has primacy for those modelling strategies located at the top left hand corner while statistics is dominant at the bottom right hand end. For this reason, I will refer to the models at the top end as 'economic' and those at the bottom as 'statistical.' Those along the frontier are hybrid models. Another way of expressing this is to say that, at the bottom we have models that simply summarize the data, and at the top we have models that aim to interpret the data. Along the curve we have work that attempts to trade off the two objectives. When we are inside the frontier it is possible to improve modelling on either one or both dimensions.

Pagan's original diagram has been augmented to illustrate where the computable general equilibrium (CGE) models and mixed complementarity (MCP) models fit on the frontier. They are located in the upper-left corner because their specifications are based solely on theoretical and technological specifications. The data requirements of these models are typically very large and depend on social accounting matrices and input-output tables obtained from 5 year and/or infrequent censuses. Thus, they are a snapshot for a single year. Arbitrary bridge equations have been used when two or more censuses are linked (Fig. A.1).

Source: Adapted from Pagan (2003a).

Fig. A.1 Adrian Pagan's "empirical" modeling frontier and research strategy trade-offs. (Source: Adapted from Pagan (2003a))

**Degree of statistical coherence**

Pagan notes that "economic" models are popular in academia and are most relevant for "storytelling," while "statistical" models are mostly relevant to approximate data, and hence are widely used in predictions/forecasting. Pagan further states that models involved in policy decision making need to bring together a degree of theoretical and empirical coherence. In other words, the models used for policy purposes should be neither economic nor statistical; they should be hybrid models based on equilibrium correction models (ECMs). Other studies also discuss the relevance of the hybrid models for policy decision making (see, e.g., Hara et al. 2009 inter alia).

### Econometric Philosophy Underlying KGEMM

Haavelmo (1943a, b, 1944) provides the first rigorous treatment of causality. His fundamental contributions to understanding the formulation and identification of causal models linked economic theory and econometrics. Yet their reception and interpretation within the econometrics community is still debated along the tradeoffs of theoretical coherence and empirical coherence suggested by Pagan (2003a). Table A.1 illustrates the different approaches between economists and statisticians. The current dominant orthodoxy can be summarized as economic analysis that formulates the correct (theoretical) model and econometrics, which merely has the task of estimating the parameters of interest from the best available data. This


Table A.1 Approaches and/or methodology for research

approach is wrong on several fronts. There is frequently no one correct theory, alternative theories can be observationally equivalent, theoretical models treat dynamics in an ad hoc manner, and the available data may not be the same as the theoretical constructs.

Haavelmo recommended applying the general principles of analyzing statistical models instead of the economists' methods. The applied econometrician must use a probability approach to their craft. He suggested they should start by assuming a probability structure to the data. This is effectively conjecturing the existence of a DGP and taking the data properties issue seriously. A critical feature of the DGP's components is that it can include competing theories. Thus, in the end the researcher discovers if one theory dominates the other(s) or different aspects seem relevant, leading to theoretical and empirical discoveries.

In reality, the DGP is unknown. The general principles suggest an iterative approach or progressive research strategy.


The approach to deriving a model, or the context of model discovery, is based on the theory of reduction. This explains the origin of empirical models based on reduction steps or operations, which are conducted on the DGP. Ideally, this implies the application of statistical models instead of the researcher imposing theoretical models based on systems of equations.

A second area Haavelmo cautioned empirical modelers about was that measured or available data from official sources are often far from the definitions of the true variables. Moreover, even if they were close, this would not necessarily mean that they correspond to the variables from economic theory. For example, when modeling private consumption what is the relevant period for making decisions regarding expenditures on services, non-durables, and durables? Is that consistent with how the data are collected?

Data from official sources depend on

#### Appendixes 103


In 1989, Haavelmo stated:

The basis of econometrics, the economic theories that we have been led to believe in by our forefathers, were perhaps not good enough. It is quite obvious that if the theories we build to simulate actual economic life are not sufficiently realistic, that is, if the data we get to work on in practice are not produced in a way theory suggests, then it is rather meaningless to confront actual observations with relations that describe something else.

The purpose of empirical investigations is to test a theory. A prerequisite for valid inferences about the theory is that there must be a close correspondence between measured variables and the true variables. In the words of Haavelmo:

It is then natural to adopt the convention that a theory is called true or false according as the hypotheses implied are true or false, when tested against the data chosen as the 'true' variables. Then we may speak interchangeably about testing hypotheses or testing theories.

Data issues will persist but are likely to improve as statistical agencies build up their capacity, processes, and delivery of data. Despite the problems in collecting and measuring macroeconomic data, the information is valuable. It can help in decisionmaking by governments, firms, individuals regarding commerce, and public finance. Theoretical arguments are needed to understand the relationships among variables regardless of the weak to poor correspondences that exist between the theoretical variables and the measured data. The cautions about realism in the data and the value of empirical data are serious but may be framed in terms of the signal-to-noise ratio in the data.

Hendry (2018) emphasizes that the debate can be addressed by nesting the theorydriven and data-driven approaches. This enables the researcher to retain insights from theory while exploring the empirical interactions when evaluating the theoretical and data evidence. Moreover, the combination of a clearly mapped out scientific method with advances in computing power and statistical programs, Autometrics, in particular provide econometricians the ability to distinguish between correlation and causation in developing empirical models. Autometrics is written to follow the theory of reduction through the general-to-specific approach of model testing and evaluation.

Models are designed to satisfy selection criteria through hypothesis tests. These can be from explicit theory and/or "long-run ratios." The reduction from a general model to a specific one relates to no loss of relevant information. Starting from a specific model and then branching out to address econometric issues like serial correlation and dynamics produces path-dependent models. The fundamental concepts from econometrics are natural vehicles for the theory of reduction and correspond to no loss of relevant information.

Consider:


All of these concepts are related to the evaluation of information and the formation of null hypotheses for diagnostic testing, for model evaluation and design criteria for model selection. The reduction process is inherently iterative: Many reduction paths could be considered, which may lead to different terminal specifications. Retain those models which survive. If multiple models survive "testimation," then a new general model is formed. In that case, conduct encompassing tests between these (possibly non-nested) specifications. If no single specific model is chosen, selection can be done by information criterion and sub-sample reliability.

The reduction approach is based on explicit model design criteria. This is based on an approximation of the local data generating process (LDGP). The LDGP is a reduction of the DGP that should include all possible relevant variables. Hypothesis tests examine the information losses from testing reductions. These can include tests for autocorrelation, heteroscedasticity, omitted variables, multicollinearity, and non-constancy. The objective is to find a congruent model.

This is superior to "symptomatology" approach in traditional econometrics where the theoretical model is directly imposed on data. The approach is invalid: there is no unique alternative to any null. Often, following adjustments of the linear model assumptions, the outcome is path dependent.

The reductions can be organized into 11 stages:


#### Congruency

We often read a model is (non) congruent. This is an important concept in our econometric philosophy and methodology.

Empirical models are at best approximations of the true data-generating process. The econometric model should exhibit certain desirable properties that render it a valid representation of the true DGP. Hendry (1995) and Mizon (1995) suggest the following six criteria according to the LSE methodology:


### The Econometric Methodology KGEMM Uses

The theory of reduction is operationalized through the general-to-specific approach advocated by Campos et al. (2005), and Hendry (1993, 1995, 2000). The approach begins with a general hypothesis about the relevant explanatory variables and dynamic process (i.e., the lag structure of the model). The general hypothesis is considered acceptable to all schools of thought. This is referred to as the general unrestricted model or GUM. The autoregressive distributed lag (ADL) model begins with a regression of the variable of interest,yt, on lagged values of itself and current and lagged values of the explanatory variables, zt.

$$\mathbf{y}\_t = a\_0 + \sum\_{i=1}^p a\_i \mathbf{y}\_{t-i} + \sum\_{j=0}^m b\_j \mathbf{z}\_{t-j} + \mathbf{e}\_t$$

The error term is assumed to be white noise. In this case, the model is referred to as an ADL ( p, m) because it contains p lagged dependent variables and m lagged explanatory variables. In the single equation, we are implicitly assuming a conditional model where the z variables are assumed to be weakly exogenous. The intercept can be expanded to include other deterministic variables like seasonal dummy variables, trend, shift dummy variables, and one-off effects. This single equation representation can be generalized to a vector autoregression.

The model is then narrowed by testing for simplifications to/or restrictions on the general model. The final or conditional model attempts to characterize the properties of the sample data in simple parametric relationships, which remain reasonably constant over time. It also accounts for the findings of previous models and is interpretable economically and financially. Rather than using econometrics to illustrate theory, the goal is to "discover" which alternative theoretical views are tenable and test them scientifically.

At this point, we have not made any assumptions on the order of integration for the y and z variables. The later assumptions can be evaluated in specifications of the model in order to avoid the problems of nonsense and spurious regressions.

Before estimating a GUM of an ADL or vector autoregressive regression (VAR) system, the first step involves examining the time series properties of the individual data series. We look at patterns and trends in the data and test for stationarity and the order of integration. Second, we form the GUM of the ADL equation or VAR. This step involves testing for the appropriate lag length of the system, including residual diagnostic tests and tests for model/system stability. Third, we examine the equation or system for potential cointegration relationship(s). Data series which are integrated of the same order may be combined to form economically meaningful series which are integrated of lower order. The cointegrating relations are tested for interpretation as an equilibrium correction mechanism. In the case of a system, we test for weak exogeneity. Based on these results, a conditional ECM of the endogenous variables may be specified and further reduction tests are performed and economic hypotheses tested.

There are two types of testing. The first is model evaluation, which is somewhat "mechanistic." These tests are performed in the context of testing for model congruency with respect to the data and economic theory. Criteria represent the null hypotheses and the associated test statistics are used to test the hypotheses. Model evaluation tests should be interpreted as destructive activity on the model. They are necessary (not sufficient) conditions for inference testing, forecasting, and policy analysis. An example would include the appropriate lag length to use in capturing the dynamics of the relationships between the variables.

The second type of testing is referred to as model design. After a statistical model has been estimated, these tests are source of value added or "art" conducted by the econometrician. Examples include hypotheses for unit elasticities, relative price elasticities, and symmetric responses are part of the toolkit or pallet.

Autometrics Doornik (2009) implements the general-to-specific modeling algorithm following on the program by Hendry (2001). There are five basic steps:


Table A.2 organizes the arguments against data mining by class on the left and refutations on the right.

There have been four generations in the evolution of this argument:


These mechanistic models appear to be immune to the standard criticisms. However, they do require that the initial general model GUM is appropriate. Also, they, PcGets and Autometrics, do not recognize the possible data transformations. These are model design issues where the applied econometrician provides value added before and/or after the reduction program, Autometrics is run.


Table A.2 Data mining: Four pejorative senses and four refutations

The model's econometric specification, especially ECMs, are estimated using the principles outlined above employing the Autometrics program in Gets module of OxMetrics. The software is designed to mimic the principles of general-to-specific in testing. Final equations will be transferred/coded into EViews to use in model building. The program will cut the equation development and testing time to a fraction of what it otherwise would be. The software is designed to follow the scientific method underlying the theory of reduction, taking advantage of advances in computing power and econometric techniques.

### Other Issues of Model Development and Application for Policy Analysis

#### Endogeneity

As it is discussed in the econometric literature, although the OLS estimation of the long-run parameters is super consistent, it may suffer from bias issue and inferences based on them are invalid since the limiting distribution of the estimation is dependent on nuisance parameters (see, e.g., Phillips and Loretan 1991; Pesaran and Shin 1999). The bias issue, especially the simultaneity bias, caused by endogeneity between dependent variable and explanatory variable as they can be jointly determined is widely concerned in economics. Fortunately, the methods mentioned fourth section that we employ to estimate long-run relationships account for this issue and hence, render the endogeneity issue being less important (e.g., see discussion in Pesaran and Shin 1999; Stock 1987; Phillips and Durlauf 1986). In this regard, it appears that one may prefer to build and simulate a macroeconometric model based on long-run equations, i.e., cointegrated relationships, as it effectively rules out the endogeneity issue (for such approach, see Weyerstrass et al. 2018; Khan and ud Din 2011; Weyerstrass and Neck 2007; Musila 2002).

As for the endogeneity issue in the short-run estimations, one can consider instrumental variable (IV) estimations such as Two-stage OLS (TSLS). However, the literature on econometric modeling discusses that the gain from IV estimations is very little compared to those from OLS on many occasions (e.g., see Johansen and Magnussen 1996; Fair 1994, 2004; Lin 1994; Christ 1951). Additionally, as Bradley et al. (1995) discuss among other studies, the gain from IV estimations may be very limited when sample span is small. Moreover, it is a well-known issue that it is quite difficult to find the right instruments. In earlier KGEMM versions, we estimated the short-run equations using different IV methods (such as TSLS, Generalized Method of Moments, Limited Information Maximum Likelihood, and K-class). We have noticed that the difference between the coefficients estimated from IV methods and those from the OLS were very small in many cases as it is discussed in the literature (e.g., see Johansen and Magnussen 1996 for the Saudi case). Therefore, we did not use IV methods in estimating the short-run relationships in this fifth version of KGEMM, but it may be considered in future version.

#### Invariance and the Lucas Critique

The 'Lucas critique' is a criticism of econometric policy evaluation procedures that fail to recognize that optimal decision rules of economic agents vary systematically with changes in policy. In particular, it criticizes using estimated statistical relationships from past data to forecast the effects of adopting a new policy, because the estimated regression coefficients are not invariant and will change along with agents' decision rules in response to a new policy. A classic example of this fallacy was the erroneous inference that a regression of inflation on unemployment (the Phillips curve) represented a structural trade-off for policy to exploit (Ljungqvist 2008).

This brief note describes what the Lucas critique implies and testing for it. The fundamental point and Lucas himself agrees with is that the Lucas critique is a possibility theorem; it is not a truism. The lack of invariance in the parameters of a conditional model can arise from multiple sources.

The basis for the Lucas critique can be illustrated with a simple relationship. In the example below, consider the x variable as a policy variable and y as the variable that policy is trying to influence.

$$\mathbf{y}\_t = \boldsymbol{\beta}\_0 \, \mathbf{x}\_t + \boldsymbol{u}\_t$$

Assume this is estimated by OLS. The properties of the disturbance term depend on the data generating process (DGP). The simple relationship is a special case of a more general model. For example, it could have come from a first order Gaussian VAR for the two time series.

$$
\begin{pmatrix} \mathbf{y}\_t \\ \mathbf{x}\_t \end{pmatrix} = \begin{pmatrix} \boldsymbol{\pi}\_{10} \\ \boldsymbol{\pi}\_{20} \end{pmatrix} + \begin{pmatrix} \boldsymbol{\pi}\_{11} & \boldsymbol{\pi}\_{12} \\ \boldsymbol{\pi}\_{21} & \boldsymbol{\pi}\_{22} \end{pmatrix} \begin{pmatrix} \mathbf{y}\_{t-1} \\ \boldsymbol{\chi}\_{t-1} \end{pmatrix} + \begin{pmatrix} \boldsymbol{\varepsilon}\_{\mathcal{H}} \\ \boldsymbol{\varepsilon}\_{\mathcal{H}} \end{pmatrix},
$$

The joint distribution for error terms is white noise with a mean of zero and possible non-zero variance-covariance matrix. This means there is no autocorrelation.

$$\begin{pmatrix} \boldsymbol{\varepsilon}\_{\rm yt} \\ \boldsymbol{\varepsilon}\_{\rm xt} \end{pmatrix} \sim N\left(\mathbf{0}, \begin{pmatrix} \sigma\_{\rm y}^{2} & \sigma\_{\rm xy} \\ \sigma\_{\rm xy} & \sigma\_{\rm x}^{2} \end{pmatrix} \mid \mathbf{y}\_{t-1}, \boldsymbol{x}\_{t-1}\right),$$

The simple relationship above given the Gaussian distribution and the assumption that the conditional distribution for yt conditional on x<sup>t</sup> is given by

$$\mathbf{y}\_t = \mathcal{J}\_0^\* \operatorname{E}[\mathbf{x}\_t \mid I\_{t-1}] + \varepsilon\_t$$

$$\mathbf{x}\_t = \pi\_{22}\mathbf{x}\_{t-1} + \varepsilon\_{xt}$$

Here the original unrestricted VAR model has been transformed into a conditional model in the first equation and a marginal model in the second equation. This is just an application of Bayes' theorem. The series x is stationary implying that -1 < π<sup>22</sup> < 1. Recall the terms ε<sup>t</sup> and εxt are independent white noise processes. The term It–<sup>1</sup> denotes the information set available to form rational expectations. In this example, the notation can be rewritten as E xt ½ ¼ <sup>j</sup> <sup>I</sup>t-<sup>1</sup> E xt ½ <sup>j</sup> xt-<sup>1</sup> : The conditional expectation for the disturbance given current xt is zero, <sup>E</sup>[εt<sup>|</sup> xt] <sup>¼</sup> 0. Then yt is a normal variable with a mean:

$$E[\mathbf{y}\_t \mid \mathbf{x}\_{t-1}] = \pi\_{22} \ \boldsymbol{\beta}\_0^\* \,\mathbf{x}\_{t-1} \dots$$

Moreover xtis a normal variable with mean.

$$E[\mathbf{x}\_t | \mathbf{x}\_{t-1}] = \pi\_{22} \mathbf{x}\_{t-1}.$$

Then the conditional mean of yt given x<sup>t</sup> is given by

$$E[\mathbf{y}\_t \mid \mathbf{x}\_t \ ] \equiv \mu\_{Y \mid X} = \boldsymbol{\beta}\_0^\* \ \mathbf{x}\_t + E[\boldsymbol{u}\_t \mid \mathbf{x}\_t \ ].$$

The last term is not equal to zero, because xt must be correlated with ut due to

$$u\_t = e\_t - \beta\_0^\* \ e\_{xt}$$

And the two equations from the conditional distribution above. The expectation for ut given xt can be written as

$$E[\mathbf{u}\_t \mid \mathbf{x}\_t] = -\beta\_0^\* \ E[\mathbf{e}\_{xt} \mid \mathbf{x}\_t] \text{ since } E[\mathbf{e}\_t \mid \mathbf{x}\_t] = \mathbf{0} \text{ by assumption from the DGP.}$$

Given the assumption of normality, there is a regression function:

$$E[\varepsilon\_{\rm tt} \mid \mathbf{x}\_{\rm t}] = \delta \,\mathbf{x}\_{\rm t} \,\,\text{where } \delta = \frac{E[\varepsilon\_{\rm tt} \mid \mathbf{x}\_{\rm t}]}{Var[\mathbf{x}\_{\rm t}]} = \frac{\sigma\_{\rm ext}^2}{Var[\mathbf{x}\_{\rm t}]}$$

The numerator in the last term is the variance of εxt . The stationarity condition implies the variance of x is defined.

$$\operatorname{Var}[\mathbf{x}\_t \mid \mathbf{}] = \frac{\sigma\_{\text{ext}}^2}{1 - \pi\_{22}^2}$$

¼ - Therefore the regression coefficient is just δ 1 π<sup>2</sup> <sup>22</sup> .

The conditional expectation for yt when the DGP is characterized by rational expectations becomes

$$
\mu\_{Y|X} := \beta\_0^\* \ge\_t - \beta\_0^\* \left(1 - \pi\_{22}^2\right) \ge\_t = \beta\_0^\* \pi\_{22}^2 \ge\_t
$$

When regressing yt on x<sup>t</sup> by OLS the estimated coefficient β b <sup>0</sup> is always consistent. However, in this case we see that the coefficient in this example is not a single parameter, but β <sup>0</sup> π<sup>2</sup> 22. Thus we obtain

$$\text{plim}\left(\widehat{\beta}\_0\right) = \beta\_0^\* \,\,\pi\_{22}^2 < \beta\_0^\*, \text{because} - 1 < \pi\_{22} < 1.$$

The essence of the Lucas critique is that any change in the formation of expectation, represented here by π22, is predicted to lead to a change in theplim β b 0 .β 0is often referred to as a deep structural parameter. This implies that the estimator forβ <sup>0</sup>will not yield the correct conclusions about how a policy change in xtaffects yt.

This implies that the initial OLS parameter β0 is not invariant to changes in expectations, π22. Therefore, xt is not super-exogenous when the DGP is characterized by rational expectations in the Gaussian VAR above. However, this is not true, β <sup>0</sup> is invariant to changes in xt under the assumption of the conditional distribution for yt conditional on x<sup>t</sup> as written in the two-equation system.

As mentioned above, the important point is that the Lucas critique is a possibility theorem; it is not a truism. The lack of invariance in the parameters of a conditional model can arise from multiple sources. Pitfalls like omitted variables and misspecified dynamics can lead to the lack of invariance even when instrumental variables/generalized method of moments (IV/GMM) estimation techniques are employed. The critical point is the exact set of assumptions. Consider a rational expectations model for planned consumption.

#### Empirical Testing for the Lucas Critique

Engle and Hendry (1993) and Engle et al. (1983) developed tests for invariance and super exogeneity. Their approach checks whether the predicted invariance occurs in the conditional model following a significant structural break in the marginal part of the model.

In the conditional plus marginal model considered above, if there is a structural break in the marginal model for xt, then it is likely that one (or more) parameters in the conditional model for yt will change around the break. But this does not follow logically. A lack of change or invariance in the parameters for the specific break being tested could be as a result of a true structural break. Recursive tests for stability of the conditional and marginal models and parameters should have been conducted previously.

Formal testing can be performed by examining the stability of the marginal model. Structural breaks (impulses, shifts, and/or trends) can be estimated using Autometrics. Impulse and step dummies are introduced into the conditional model to test for super exogeneity. The null hypothesis of invariance is that these do not add explanatory power. An F-test or Wald test is used. If the null hypothesis is rejected, the Lucas critique is "confirmed." But, failure to reject the null does not imply that it is true. Expectations are still important except in this case, where the impact is not that strong. The specification of rational expectations in the model is too focused. It does not reveal or capture how expectations are formed and transmitted through the model.

Favero and David Hendry (1992) and Ericsson and Irons (1994, 1995) provide reviews of the literature on testing for the Lucas critique. Rudebusch (2005), among others, also shows the empirical unimportance of the Lucas critique to monetary policy. Moreover, Ericsson and Irons (1994, 1995) conclude that virtually no evidence exists that empirically substantiates the Lucas critique. They also empirically refute the critique using a super exogeneity test. The mnemonics and descriptions of the variables used in the fifth version of KGEMM are documented below in Appendix B.


### Appendix B: KGEMM Variables

#### Appendixes 113



#### Appendixes 115

(continued)



#### Appendixes 117

(continued)



#### Appendixes 119

(continued)




#### Appendixes 121

(continued)



#### Appendixes 123

(continued)




#### Appendixes 125

(continued)



#### Appendixes 127

(continued)



#### Appendixes 129

(continued)



#### Appendixes 131

(continued)




#### Appendixes 133

(continued)



#### Appendixes 135



#### Appendixes 137

(continued)



(continued)

### Appendix C: Estimated Final ECM Specifications

This appendix records in Sects. A.3.1, A.3.2, A.3.3, A.3.4, and A.3.5 the estimated final ECM specifications associated with the long-run equations reported in Sect. 7. We obtained final ECM specifications from general unrestricted ECM specifications using Autometrics, a machine learning econometric modeling methodology in the Gets module of OxMetrics, as discussed in Sect. 4 and Appendix A.3. The lag orders of two and one are considered in the formulation of general unrestricted ECMs depending on sample spans being available for estimations. Estimated initial and congruent general unrestricted ECM specifications are not reported here to save space, but they are available from the authors on request.

For the readers convenience, we describe one of the ECM equations below and the rest of the equations follow the same logic. As an example, we select the first appeared ECM equation, i.e., Eq. (1).

$$\begin{array}{l} \text{DLOG(IFDIS)} = -0.06 - 0.25^\* \text{ECT\\_IFDIS}(-1) + 2.65^\* \text{ DILOG(GVAIS)} \\ \text{ } -0.01^\* \text{ D(RRLEND1)} - 1.26^\* \text{ DI2014} \end{array}$$

Where, D denotes the first difference. LOG indicates the natural logarithmic transformation. IFDIS, GVADIS, and RRLEND1 are the variables (see Appendix B for the definitions and notations of the variables). DI2014 is a dummy variable taking unity in 2014 and zero otherwise. ECT\_IFDIS is the residuals of the long-run equation of private investment in the retail, wholesale, hotels, and catering sector, i.e., equilibrium correction term. In general, ECT\_X denotes equilibrium correction term for the variable X. (n) attached to a given variable indicates n years of lagged series of a given variable. For example, ECT\_IFDIS(-1) indicates one year lagged ECT\_IFDIS.

This final ECM specification above indicates that the growth rates of private investment in the retail, wholesale, hotels, and catering sector increases by 2.7% if the growth rate of the sector's output increases by 1% while a 1 percentage point rise in the changes of real lending rate causes 1% decrease in the sector's investment, holding other factors constant. Speed of adjustment (SoA), i.e., the coefficient of ECT\_IFDIS lagged by one year shows that in 1 year, 25% of the disequilibrium corrects to the long-run relationship (that private investment in the retail, wholesale, hotels, and catering sector establishes with its output and the real lending rate) given in Eq. (44).

### Final ECM Specifications for the Real Block

#### ECM Equations for Investments by Economic Activity Sector

$$\begin{aligned} \text{DLOG}(\text{IFBIS}) &= -0.060 - 0.25^{\circ} \text{ET} \text{ I} \text{IDS} (-1) \\ &+ 2.6^{\circ} \text{DLOG}(\text{GVAIDS}) - 0.01^{\circ} \text{D(RREIEND)} \\ &- 1.26^{\circ} \text{DDA104} \\ \text{DLOG}(\text{IFCON}) &= -0.06 - 0.71^{\circ} \text{ET} \text{I}\_{\text{IFCON}} (-1) - 0.01^{\circ} \text{D(RELEND)} \\ &- 0.47^{\circ} \text{IDL} (04 - 0.6^{\circ} \text{T}^{2} \text{D004} + 0.83^{\circ} \text{T} \text{2005} \\ &- 0.20^{\circ} \text{T} \text{2008} \\ \text{DLOG}(\text{IFFBI}) &= -0.03 - 0.97^{\circ} \text{ET} \text{I}\_{\text{IFFBI}} (-1) + 2.88^{\circ} \text{DLOG}(\text{GVAFBI}) \\ &+ 0.34^{\circ} \text{DLOG}(\text{IFFBI}(-1)) - 0.86^{\circ} \text{DP} \text{2010} \\ &+ 0.34^{\circ} \text{DLOG}(\text{IFFBI}(\text{XM})) \\ &+ 1.41^{\circ} \text{DLOG}(\text{GVAMNON}) \\ &- 0.01^{\circ} \text{DLOG}(\text{GVAMNON}) \\ &+ 0.39^{\circ} \text{PDLOG}(\text{I}\_{\text{IFF}}) + 1.54^{\circ} \text{DLOG}(\text{RER}) \\ \text{DLOG}(\text{IFFOH}) &= -0.46 + 0.57^{\circ} \text{ET} \text{F}\_{\text{IFF}} \text{IDGS} \\ &+ 0.48^{\circ} \text{T}$$

$$\begin{aligned} \text{DLOG(IFTRACOX)} &= 0.10 - 0.08 \text{'ECT I} \text{IFRACOM}(-1) \\ &+ 1.97 \text{'DLOG(GVAATACOM)} \\ &- 1.83 \text{'DLOG(GVATACOM)} - 1) \\ &- 0.49 \text{'DLOG(GVATACOM)} - 1) \\\\ \text{DLOG(IFU)} &= -0.001 - 0.76 \text{'ECT I} \text{FU}(-1) + 1.35 \text{'DLOG(GVAU)} \\ &+ 5.63 \text{'DLOG(RER}(-1)) - 0.50 \text{'DP995} \\ &- 0.56 \text{'DP2009} - 0.28 \text{'DP2012} \\ \text{DLOG(IFAGR)} &= -0.10 + 8.23 \text{'DLOG(GVAAGR)} \\ &+ 4.31 \text{'DLOG(RERR)} - 0.42 \text{'ECT I} \text{AGR}(-1) \\ &+ 1.83 \text{'DLOG(I)} + 1.78 \text{'DDO11} + 1.63 \text{'DDO12} \end{aligned} (9)$$

#### ECM Equations for Sectoral Gross Value Added by Economic Activity Sector

$$\begin{aligned} \text{DLOG } (\text{GVAAG}) &= -0.00001 - 0.20 \text{"{Cert}.GVAAG} (-1) \\ &+ 0.07 \text{"{DLOG } (\text{DEL.AGR})} - 0.06 \text{"{DDP } 994} \\ &- 0.05 \text{"{DDP } 994} \\ \text{DLOG } (\text{GVACGN}) &= -0.01 - 0.36 \text{"{Cert} } \text{GVACGN} (-1) + 0.04 \text{"{DDP } 998} \\ &+ 0.23 \text{"{DLOG } (\text{TDONC})} + 0.32 \text{"{DLOG } (\text{GVACGN}(-1)) \\ &(11) \\ \text{DLOG } (\text{GVABDIS}) &= -0.02 - 0.66 \text{"{FCT } \text{GVAIDS}(-1) \\ &+ 0.38 \text{"{DLOG } (\text{GVAIDS}(-1)) + 0.31 \text{"{DLOG } (\text{GVAIDS}(-2)) } \\ &+ 0.39 \text{"{DLOG } (\text{TDN } \text{R}) + 0.10 \text{"{DDP } 910} \\ \text{DLOG } (\text{GVAFIBN}) &= -0.01 + 0.30 \text{"{DLOG } (\text{GVAFC}(-1)) \\ &+ 0.86 \text{"{DLOG } (\text{GVAFIB}(-1)) \\ &- 0.35 \text{"{Credit } 974} \text{"{DACK } (\text{GVACG}(-1)) \\ \text{DLOG } (\text{GVACGN}(-1) - 0.04 \text{"{DLOG } 974}) &+ 0.16 \text{"{DLOG } (\text{GVACG}(-1)) \\ &+ 0.68 \text{"{DLOG } (\text{GVACG}(-1)) + 0.16 \text{"{DLOG } (\text{T}$$

$$\begin{aligned} \text{DLOG}(\text{GVAAMNON}) &= 0.04 - 0.40^{\circ} \text{EURG}(\text{GVAAMNON}) - (1) \\ &+ 0.30^{\circ} \text{DLOG}(\text{GVAMON}) + 0.09^{\circ} \text{PLG}(\text{GVACM}\_{\text{L}}) \\ &+ 0.12^{\circ} \text{DLOG}(\text{GVAIM}\_{\text{L}}) + 0.02^{\circ} \text{DLOG}(\text{DLOG}\_{\text{L}}) \\ \text{DLOG}(\text{GVAAME}) &= 0.01 - 0.83^{\circ} \text{EURG}(\text{GVAIM}\_{\text{L}}) - (1) \\ &+ 0.31^{\circ} \text{DLOG}(\text{GVAIM}\_{\text{L}}) - 0.15 \\ &+ 0.31^{\circ} \text{DLOG}(\text{GVAIM}\_{\text{L}}) - 0.45^{\circ} \text{DLOG}(\text{GVAIM}\_{\text{L}}(-1)) \\ &- 0.11^{\circ} \text{DLOG}(\text{DLOG}(\text{L})) \\ &- 0.11^{\circ} \text{DLOG}(\text{DLOG}(\text{L})) \\ \text{DLOG}(\text{GVAOMN}\_{\text{L}}) &= 0.01 - 0.26^{\circ} \text{EURG}(\text{A}) + 0.03^{\circ} \text{SLD}(\text{GVAIM}\_{\text{L}}(-1)) \\ &+ 0.11^{\circ} \text{DLOG}(\text{DLOG}(\text{S})) + 0.03^{\circ} \text{DLOG}(\text{Dike}(\text{L})) \\ &+ 0.25^{\circ} \text{DLOG}(\text{WABN}\_{\text{L}}(-1)) + 0.03^{\circ} \text{DLOG}(\text{L}) \end{aligned} (16)$$
 
$$\begin{aligned} \text{DLOG}(\text{WACM}(\text{FIBM}))$$

ð Þ 21

$$\begin{array}{l} \text{DLOG (DINGA.INID.NEU)} = -0.01 + 0.55^{\circ} \text{ DLOG (DINGA.INID.NEV(-1))} \\ \quad + 0.37^{\circ} \text{ DLOG (GVAFETCH)} \\ \quad - 1.22^{\circ} \text{ ECT } \text{DLOG} \text{.INID.NEV(-1)} \\ \quad + 0.04^{\circ} \text{ S12000} \end{array} (22)$$

$$\begin{array}{l} \text{DLOG (GVATRACOM)} = -0.002 - 0.40^{\circ} \text{ECT } \text{\_GVATRACOM}(-1) \\ \quad + 0.60^{\circ} \text{DLOG (TOTACCOOH)} \\ \quad + 0.44^{\circ} \text{DLOG (DINA-TRA)} \\ \quad + 0.42^{\circ} \text{DLOG (TOTACCOOH(-1))} \\ \quad + 0.23^{\circ} \text{DLOG (GVATRACOM(-1))} \\ \quad - 0.10^{\circ} \text{DDP} (\text{GVAPIACOM}(-1)) \\ \quad - 0.10^{\circ} \text{DDP} (\text{DNG} (\text{DNG} \text{L}) + 0.11^{\circ} \text{DLOG (DCOMP } \text{L}) \\ \quad + 0.23^{\circ} \text{DLOG (DINA} \text{ (-1)} + 0.09^{\circ} \text{DP} 2010 \\ \quad - 0.52^{\circ} \text{ECT } \text{GVAU} (-1) + 0.09^{\circ} \text{DP} 2010 \end{array}$$

#### ECM Equation for Private Consumption

$$\begin{aligned} \text{DLOG(CONS)} &= 0.03 - 0.23^{\text{\*}} \text{ECT\\_CONS} (-1) + 0.16^{\text{\*}} \text{DLOG(DI)} \\ &- 0.01 \, ^{\text{\*}} \text{D(RCB} - @ \text{PCH}(\text{CPI}) \* 100) \\ &- 0.06 \, ^{\text{\*}} \text{DLOG(WEALTH)} - 0.05 \, ^{\text{\*}} \text{DB1112} \end{aligned} \tag{25}$$

### Final ECM Specifications for the Fiscal, Monetary, and External Blocks

#### ECM Equations for Government Expenditure Items and Debt

$$\begin{array}{l} \text{DLOG} \left( \text{GWSA} \,\text{Z} \right) = 0.05 - 0.18 \,\text{'ECT\\_GWSA} \,\text{Z} \,(-1) \\ \quad + 0.07 \,\text{'DLOG} \left( \text{GREV} \right) - 0.13 \,\text{'DI1994} \\ \quad - 0.26 \,\text{'DI1995} - 0.34 \,\text{'DI1996} + 0.14 \,\text{'DP2009} \end{array} \tag{26}$$

$$\begin{aligned} \text{DLOG}\ (\text{GATE}\mathcal{Z}) &= 0.03 - 0.29\text{PET}\ \text{GATE}\ \text{(-1)}\\ &+ 0.12\text{ DLOG}\ (\text{GIVE}(-1)) + 0.25\text{ DI1979} \\ &+ 0.39\text{ PD000} \end{aligned} \tag{27}$$

$$\begin{aligned} \text{DLOG}\ (\text{GMO}\ \mathcal{Z}) &= -0.01 - 0.16^{\circ}\text{ECT}\ \text{GMO}\ \mathcal{Z}(-1) \\ &+ 0.25\text{ DLOG}\ (\text{GMO}\ \mathcal{Z}(-1)) - 0.48^{\circ}\text{DDP}(1977) \\ &+ 0.16^{\circ}\text{DLOG}\ (\text{GBPE}(-1)) - 0.48^{\circ}\text{DDP}(1979) \\ \text{DLOG}\ (\text{GGCPE}) &= 0.02 - 0.25^{\circ}\text{DLOG}\ (\text{GBPE}) + 5.63^{\circ}\text{DDP}\ 981 \\ &- 0.76^{\circ}\text{DDP}\ 56 + 0.92^{\circ}\text{DDP}\ 990 \\ \text{DLOG}\ (\text{GC.Z.OH}) &= 0.10 - 0.41^{\circ}\text{ECC.G.Z.OH}\ (-1) \\ &- 0.14^{\circ}\text{DLOG}\ (\text{GCC.GH}\ (-1)) + 0.62^{\circ}\text{DLOG}\ (\text{GBRW}) \\ &- 0.35^{\circ}\text{DLOG}\ (\text{GREV}(-1)) + 0.05^{\circ}\text{DDP}\ 991 \\ &- 0.73^{\circ}\text{DLOG}\ (\text{GREV}(-1)) + 0.39^{\circ}\text{DDP}\ 999 \\ \text{DLOG}\ (\text{GLZ}) &= 0.04 - 0.17^{\circ}\text{ECTG}\$$

ð Þ 32

#### ECM Equation for M2 Money Balance

$$\begin{array}{l} \text{DLOG (M2/PGDP\*100)} = -4.17 - 0.29^{\circ} \text{DLOG (M2(-1)/PGDP(-1)\*100)} \\ \quad - 0.32^{\circ} \text{DLOG (M2(-2)/PGDP(-2)\*100)} \\ \quad - 0.93^{\circ} \text{ECT } \text{MD} \text{,} \text{(} -1) + 0.63^{\circ} \text{DLOG (GDP)} \\ \quad - 0.20^{\circ} \text{DLOG (WPO AL.R)} \\ \quad - 0.31^{\circ} \text{DLOG (WPO AL.R(-1))} \\ \quad + 0.62^{\circ} \text{DLOG (REER)} - 0.003^{\circ} \text{D(IRD)} \\ \quad - 0.20^{\circ} \text{DLOG (WPO AL.R(-2))} + 0.08^{\circ} \text{DP2004} \\ \quad + 0.06^{\circ} \text{DP2009} \end{array}$$

#### Exports Related ECM Equations

$$\begin{aligned} \text{DLOG}(\text{XGNOIL}) &= -0.62^{\circ}\text{EFT}\_{1}\text{XGONL}(-1) + 0.19^{\circ}\text{DLOG}(\text{XGNOIL}(-1)) \\ &- 1.73^{\circ}\text{DLOG}(\text{REER}) + 0.45^{\circ}\text{DLOG}(\text{REER}(-1)) \\ &- 1.00^{\circ}\text{DLOG}(\text{REER}(-2)) - 0.66^{\circ}\text{DLOG}(\text{GDP}\_{\text{MNA}}\text{\*}\text{RXD}) \\ &- 0.54^{\circ}\text{DLOG}(\text{GDP}\_{\text{MNA}}\text{(-1)} \star \text{RXD}(-1)) \\ &+ 0.56^{\circ}\text{DLOG}(\text{GDP}\_{\text{MNA}}\text{'}(-2) \star \text{RXD}(-2)) \\ &+ 2.88^{\circ}\text{DLOG}(\text{GVANOIL}) - 1.81^{\circ}\text{DLOG}(\text{GVANOIL}(-2)) \\ &- 0.32^{\circ}\text{DP}1992 - 0.15^{\circ}\text{DDP}1994) \\ \text{DLOG}(\text{XOHALEF}) &= - - 0.05 + 1.69^{\circ}\text{DLOG}(\text{WHERE}) + 0.92^{\circ}\text{DLOG}(\text{GVAOILREF}) \\ &- 0.18^{\circ}\text{DLOG}(\text{WPOAL}\text{L.R}(-2)) \\ &- 0.64^{\circ}\text{EECT}\text{XOHILEF}(-1) - 0.23^{\circ}\text{DP}2000 \\ &- 0.40^{\circ}\text{DP}1987 \end{aligned} (2)$$

ð Þ 35

#### Imports Related ECM Equations

$$\begin{array}{l} \text{DLOG (MGCAP)} = -0.04 + 1.31 \text{"DLOG (DOM)} \\ \quad - 0.66 \text{"ECT MGGAP} (-1) + 0.18 \text{"DDP2003}) \\ \quad + 0.07 \text{"S12008} \end{array} \tag{36}$$
 
$$\begin{array}{l} \text{DLOG (MGCONS)} = 0.001 + 0.41 \text{"DLOG (MGCONS} (-1)) + 0.96 \text{"DLOG (DOM)} \\ \quad - 1.15 \text{"ECT MGGCONS} (-1) + 0.09 \text{"DDP1995} \\ \quad + 0.18 \text{"D(DP2003)} - 0.16 \text{"DP2018} \end{array} \tag{37}$$
 
$$\begin{array}{l} \text{DLOG (MGINITER)} = -0.03 + 2.25 \text{"DLOG (GVANOIL + GVAOIL)} \\ \quad - 0.38 \text{"DLOG (GDP307, PGDP)} \end{array} \tag{38}$$

 ð- Þ ð Þ - 0:38 DLOG ðPGDP US=PGDPÞ 0:87 ECT MGINTER 1 38

$$\begin{array}{l} \text{DLOG (MS)} = 0.03 - 0.53^\* \text{ECT } \text{MS} (-1) + 2.67^\* \text{DLOG (DOMD)}\\ \quad + 1.74^\* \text{DLOG (DOMD} (-1)) - 7.64^\* \text{DLOG (RRXD)}\\ \quad - 0.37^\* \text{DP2015} - 0.46^\* \text{DP1998} + 0.38^\* \text{DP2006} \end{array} (39)$$

#### ECM Equation for Outflow Remittances

$$\begin{aligned} \text{DLOG}(\text{REMF}(\text{RXD}/\text{PGD}^\*100) &= -0.0004 - 0.16^\circ \text{ET.REMOF}(-1) \\ &+ 0.31^\circ \text{DLOG}(\text{REMF}(-1) \* \text{RXD}(-1)) \\ &\text{PGDP}(-1) \* 100) \\ &+ 0.74^\circ \text{DLOG}(\text{GDP}) \\ &+ 0.26^\circ \text{DLOG}(\text{ETNS}(-1)) \\ &- 1.36^\circ \text{DLOG}(\text{PGDP}) \\ &- 0.03^\circ \text{DLOG}((\text{EXPL}/\text{PGDP}^\*100) + 1) \\ &+ 0.23^\circ \text{DDP}(\text{DP}1978) \\ &- 0.16^\circ \text{DDP}(\text{DP}2016) - 0.31^\circ \text{DP}1986 \end{aligned}$$

### Final ECM Specifications for Domestic Prices Block


#### ECM Equations for Prices of the Household Consumption Basket Items

$$\text{DLOG}(\text{CPH}) = -0.05 + 0.26 \text{'DLOG}(\text{CPU}(-1)) + 0.56 \text{'CLO}(\text{PGDHREAL}) \tag{41}$$

$$+ 0.10^4 \text{DLOG}(\text{PEH}\_2\text{RES}) - 0.26 \text{'FC} \text{CTU} \text{EUR} \text{LUL} (-1) \tag{42}$$

$$\begin{array}{c} \text{DLOG}(\text{CPH}(\text{CPH}) = 0) \text{ 0 + 0.78^2 \text{LGO}(\text{PGDHGAP}) + 0.61 \text{'DLOG}(\text{PGDHGAP}(-1)) \\ + 0.28^2 \text{'LLOG}(\text{PMIG}) \\ - 0.25^2 \text{'ECT}(\text{CPH}(\text{OD}) \text{LUL} (-1) + 0.001 \\ - 0.25^2 \text{'ECT}(\text{CPH}(\text{CPH}(\text{R}(-2)) + 0.15^2 \text{'LLOG}(\text{PE} + \text{TRACOM}) \\ - 0.34^2 \text{'ECT}(\text{CPH} \text{RAL} (-1) \\ \end{array} \tag{43}$$

$$\begin{array}{c} \text{DLOG}(\text{CPH} \text{H}) = 0.02 + 0.03^2 \text{DLOG}(\text{WID} \text{S}) + 0.09^2 \text{DLOG}(\text{PM} \text{)} \\ - 0.30^2 \text{CET} \text{CPH} \text{H} \text{LUL} (-1) + 0.04^2 \text{S11984} \\ - 0.08^2 \text{S12010} + 0.06^2 \text{S12017} - 0.08^2 \text{DDP} \text{L} \text{T} \end{array} \tag{44}$$

$$\begin{array}{c} \text{DLOG}(\text{CPH} \text{ODM}) = -0.42 + 1.07^2 \text{LLOG$$

ð Þ -

ð Þ ð Þ ð Þ Þ ð Þ ð Þ Þ ð Þ ð Þ DLOG ðCPIHTLÞ¼-0:002 - 0:18ECT CPIHTL ULCð-1Þ þ0:59 DLOG ðCPIHTLð-1ÞÞ þ 0:29DLOG ðPGDPDISÞ þ0:78 DLOG ððVAT RATE þ 100Þ=100Þ þ0:05 DP2011 46 DLOG ðCPICLOTHÞ¼-0:01 þ 0:37DLOG ðCPICLOTHð-1ÞÞ þ0:17DLOG ðPGDPMANNOÞ þ 0:05 DLOG ðWDISÞ -0:57 ECT CPICLOTH ULCð-1Þ þ0:06 DðDP2016Þ 47 DLOG ðCPIMISCÞ¼-0:001 þ 0:31DLOG ðCPIMISCð-1ÞÞ þ 0:27DLOG ðPGDPSERÞ þ0:12DLOG ðPMGÞ - 0:32ECT CPIMISC ULCð-1Þ þ0:04DP2007 48 DLOG ðCPIEDUÞ ¼ 0:01 - 0:67 ECT CPIEDU ULCð-1Þ þ 0:00004 -06GAP GVAGOVð-1Þ þ 0:04DLOG ðPGDPGOVÞ -0:30DLOG ðGVAGOVÞ - 0:28DLOG ðGVAGOVð-1Þ -0:05 DP2007 þ 0:05 SI2010 - 0:05 SI2007 49 DLOG ðCPIARTÞ¼ -1:84 þ 0:35 DLOG ðCPIARTð-1ÞÞ þ 1:02DLOG ðPGDPSERÞ þ1:28 DLOG ðPGDPSERð-1ÞÞ þ 0:10DLOG ðWSERÞ þ0:34 DLOG ðPMÞ þ 0:31 DLOG ðPMð-1ÞÞ þ1:74DLOG ððVAT RATE þ 100Þ=100Þ -0:58 ECT CPIART ULCð-1Þ þ 0:09DP2011 50 DLOG ðCPIHEALÞ¼-0:001 þ 0:51 DLOG ðCPIHEALð-1ÞÞ þ 0:49DLOG ðCPIHEALð-2Þ þ0:14DLOG ðPGDPSERð-2ÞÞ þ 0:02 DLOG ðPMSð-2ÞÞ þ0:60 DLOG ððVAT RATE þ 100Þ=100Þ -0:24ECT CPIHEAL ULCð-1Þ - 0:12DP2017 51 DLOG ðCPITOBCÞ¼-3:73 - 0:33ECT CPITOBC ULCð-1Þ þ0:43DLOG ðCPITOBCð-1ÞÞ þ 0:55 DLOG ðPGDPMANNOÞ þ0:10 DLOGðPELE COMMÞ þ 2:72DLOG ððVAT RATE þ 100Þ=100Þ -0:08DP2015 þ 0:07DP1998 52


#### ECM Equations for Sectoral Producer Prices

$$\begin{aligned} \text{DLOG}(\text{PEDPA}(\text{AG})) &= 0.05 + 0.2^{\text{cpt}} \text{FC}^{\text{PEDPA}}(\text{GDP}(-1)) \\ &+ 0.06^{\text{cpt}} \text{DI}(\text{DDP}(\text{AG}) - 0.02^{\text{cpt}} \text{DI}(\text{DDP}(-1)) \\ &+ 0.06^{\text{cpt}} \text{DI}(\text{DDP}(\text{AG}) - 0.02^{\text{cpt}} \text{DI}(\text{DDP}(-1)) \\ &- 0.3^{\text{cpt}} \text{DI}(\text{DDP}(\text{AG}) \text{DI}(\text{DDP}(\text{OD}) - 1) \\ &- 0.3^{\text{cpt}} \text{DI}(\text{DDP}(\text{AG}) \text{DI}(\text{DDP}(\text{OD}) - 1) \\ &+ 0.04^{\text{cpt}} \text{DI}(\text{DDP}(\text{OD}(-1)) + 0.05^{\text{cpt}} \text{DI}(\text{DDP}(\text{OD}(-1)) - 1) \\ &+ 0.04^{\text{cpt}} \text{DI}(\text{DDP}(\text{OD}(\text{DDP}(\text{DI}(-1)) - 1) \\ &+ 0.04^{\text{cpt}} \text{DI}(\text{DDP}(\text{OD}(\text{DDP}(\text{DI}(-1)) - 1) \\ &+ 0.05^{\text{cpt}} \text{DI}(\text{DDP}(\text{OD}(\text{DDP}(\text{I}(-1)) \\ \text{DDP}(\text{OD}(\text{DDP}(\text{DI}(-1)) - 1) \\ &- 0.3^{\text{cpt}} \text{DI}(\text{DDP}(\text{DDP}(\text{DI}(-1)) - 1) \\ &- 0.3^{\text{cpt}} \text{DI}(\text{DDP}(\text{V}(\text{DDP}(\text{I}(-1))$$

148 Appendixes

$$\begin{aligned} \text{DLOG(PGDPTR)} &= 0.003 + 0.27 \text{'DLOG(PGDOND)} \\ &+ 0.04 \text{'DLOG(ULCRACOM)} \\ &- 0.68 \text{'ECT\\_PGDPRICACOM} (-1) \\ &+ 0.10 \text{'DP2000} + 0.01 \text{'DI2009} - 0.02 \text{'S12008} \\ \end{aligned} (61)$$

$$\begin{aligned} \text{DLOG(PGDPSK)} &= -0.32 \text{'ECT\\_PGDPRER} (-1) + 0.02 \\ &+ 0.28 \text{'DLOG(PDM(-1))} + 0.15 \text{'DLOG(PDM(-2))} \\ &+ 0.10 \text{'DLOG(PLDER\\_COMM)} + 0.03 \text{'DP2018} \\ \text{DLOG(PGDPU)} &= -0.02 + 0.80 \text{'DLOG(PGDPOND)} + 0.12 \text{'DLOG(PEL\\_U)} \\ &+ 0.15 \text{'DLOG(PEL\\_U(-1))} - 0.43 \text{'ECT\\_PGDPRU(-1)} \\ &- 0.06 \text{'DLOG08} + 0.15 \text{'DP2018} \end{aligned} (62)$$

### Final ECM Specifications for Labor and Wages Block

#### ECM Equations for Employment by Economic Activity Sector

$$\begin{aligned} \text{DLOG}(\text{ETAGB}) &= 0.04 - 0.57^{\circ} \text{ETET}\_{\text{ETABGR}}(-1) + 1.98^{\circ} \text{DLOG}(\text{GVAGR}) \\ &- 0.79^{\circ} \text{DLOG}(\text{WARGR}^{\circ}100) \\ &- 0.17^{\circ} \text{DLOG}(\text{ETAGB}(-1)) \end{aligned} \tag{64}$$

$$\begin{aligned} \text{DLOG}(\text{ETCON}) &= -0.09 - 1.11^{\circ} \text{ETET}\_{\text{ETCON}}(-1) \\ &- 0.41^{\circ} \text{DLOG}(\text{WCOMP}(-1)/\text{PDPDCON}^{\circ}100) \\ &- 0.13^{\circ} \text{DLOG}(\text{WCON}(-1)/\text{PDPDCON}(-1) \ast 100) \\ &+ 0.21^{\circ} \text{SID}(-0.12^{\circ} \text{T} \text{EDI8} + 0.11^{\circ} \text{T} \text{2008} \\ \text{DLOG}(\text{ETDIIS}) &= -0.48^{\circ} \text{ECT}\_{\text{ETDIIS}}(-1) - 0.001 + 0.22^{\circ} \text{DLOG}(\text{GVADR}) \\ &- 0.69^{\circ} \text{DLOG}(\text{WDIS}/\text{RGDDIS}^{\circ}100) + 0.12^{\circ} \text{D(DBP2016)} \\ &- 0.07^{\circ} \text{D(DP2003)} \\ \text{DLOG}(\text{ETIBBU}) &= 0.04 - 0.31^{\circ} \text{ECT}\_{\text{ETDI}}(-1) \\ &- 0.38^{\circ} \text{DLOG}(\text{WIBBU}/\text{PGDFIBU}^{\circ}100) \\ &- 0.17^{\circ} \text{DLOG}(\text{WIBBU}/\text{PGDFIBU}^{\circ}100) \\$$

$$\begin{array}{c} + 0.59^\* \text{DLOG}(\text{GVAGOV}) + 0.06^\* \text{DP2008} \\ + 0.08^\* \text{DP2013} - 0.07^\* \text{DP2018} \end{array} \tag{68}$$

$$\begin{aligned} \text{D.OG}(\text{ETAMNO}) &= 0.01 - 0.37 \text{'ETET} \text{EUNANNO}(-1) \\ &+ 0.69 \text{"D2008} \\ &- 0.15 \text{"D2008} \end{aligned} \tag{69}$$

$$\begin{aligned} \text{D.OG}(\text{ETAMNOHT}) &= 0.02 - 0.18 \text{"{DET} $ET$ } \text{EBIT} \text{MNOME}(-1) \\ &+ 0.18 \text{"{DLO}OfETO}(-1) + 0.13 \text{"{D2002}} \\ &- 0.26 \text{"{DBO}log} + 0.21 \text{"{DBI}At} \\ \text{DLOG}(\text{ETOTHS}) &= 0.006 - 1.35 \text{"{EET} $ET$ } \text{EITOS}(-1) + 0.59 \text{"{DLOG}(
\text{GVANOL})} \\ &- 0.29 \text{"{DLO}OfETO}(-1) + 0.59 \text{"{DLOG}(
\text{GVANOL})} \\ &+ 0.29 \text{"{DLO}OfETO}(-1) + 0.12 \text{"{DDI}OfETO} \\ &- 0.09 \text{"{DDI}OfETO} \text{(}(-1) + 0.12 \text{"{DDI}OfETO} \\ &- 0.07 \text{"{DDI}OfETO} \\ \text{DLOG}(\text{ETTHACOM}) &= 0.00 - 0.23 \text{"{EET} $ET$ } \text{EITACOM}(-1) \\ &- 0.17 \text{"{DLO}OfETO} \text{(}(-1) \\ &- 0.12 \text{"{DLO}OfETO} \text{(}(-1) \text{"{DLO}OfETO} \text{(}(-1) \text{@U} \\ &- 0.12 \text{"{DLO}OfETO} \text{(}(-1) \text{@U} \text{"{DDI}OfETO} \text{(}(-10) \text{@$$

#### ECM Equations for Wages by Economic Activity Sector

$$\begin{aligned} \text{DLOG}(\text{WGR}) &= -1.04^{\circ} \text{ECT } \text{WGR}(-1) - 5.29 + 0.31^{\circ} \text{DLOG}(\text{WGR}(-1)) \\ &+ 1.27^{\circ} \text{DLOG}(\text{GVAGR}/\text{ETAGB}) \\ &+ 0.19^{\circ} \text{DLOG}(\text{GVAGR}(-2)/\text{ETAGB}(-2)) \\ &+ 2.14^{\circ} \text{DLOG}(\text{PGDPAGR}) - 1.63^{\circ} \text{DLOG}(\text{PGDPAGR}(-1)) \end{aligned} \tag{75}$$
 
$$\begin{aligned} \text{DLOG}(\text{WCN}) &= -0.36^{\circ} \text{ETCOON}(-1) + 0.001 \\ &+ 0.89^{\circ} \text{DLOG}(\text{GVAGR}/\text{ETCON}) + 0.28^{\circ} \text{DP2010} \\ &- 0.54^{\circ} \text{DP2003} \end{aligned} \tag{76}$$

ÞÞ Þ ÞÞ ð Þ Þ Þ þ ð Þ ð Þ - þ ð Þ ð Þ - ð Þ Þ ð Þ DLOG WFIBU ð Þ¼-0:48 ECT WFIBUð Þþ -1 0:01 - 0:31 DLOG WFIBU ð ð-1 þ1:21 DLOG GVAFIBU ð Þ =ETFIBU þ0:34 DLOG GVAFIBU ð ð Þ -1 =ETFIBUð Þ -1 þ0:72 DLOG PGDPFIBU ð Þ- 0:56 DLOG PGDPFIBU ð ð-1 -0:30 DST0309 77 DLOG WMAN ð Þ¼-0:35 ECT WMANð Þ- -1 0:15 þ1:16 DLOG GVAMAN ð =ETMAN þ1:00 DLOG GVAMAN ð ð Þ -2 =ETMANð Þ -2 0:41 DLOG WMAN 2 ð78Þ DLOG WMIN ð Þ¼-0:65 ECT WMINð Þ- -1 0:01 þ0:32 DLOG GVAMIN ð Þ =ETMIN 0:18 DLOG PGDPMIN 1 0:18 D DP2011 ð79Þ DLOG WTRACOM ð Þ¼-0:87 ECT WTRACOMð Þþ -1 0:01 þ0:42 DLOG GVATRACOM ð =ETTRACOM þ0:86 DLOG PGDPTRACOM ð Þ þ0:27 DLOG WTRACOM ð Þ ð Þ -1 - 0:42 DP2003 -0:26 DP2011 80 DLOG WU ð Þ¼-0:52 ECT WUð Þ- -1 0:02

$$\begin{array}{l} \text{DLOG}(\mathbf{W}\mathbf{U}) = -0.22 \text{ ECI\\_WU}(-1) - 0.02\\ \quad + 1.24 \text{"DLOG}(\text{GVAU}/\text{ETU}) - 0.63 \text{"DP2010} \\ \quad + 0.30 \text{"DP2013} \end{array} \tag{81}$$

### Final ECM Specifications for Energy Block

#### ECM Equations for Demand for Energy Products

Industry

$$\begin{array}{l} \text{DLOG (DCOIL.IND)} = 0.02 + 1.50 \, ^\circ \text{DLOG (GVAIND - GVAU)} \\ \quad - 0.64 \, ^\circ \text{DLOG (PCOIL.IND(-2) / PGDPIND(-2) \* 100)} \\ \quad - 0.32 \, ^\circ \text{ECT } \text{DCOD.IND } \text{N(-1)} \\ \quad + 0.75 \, ^\circ \text{DLOG (PDIS.IND / PGDPIND \*100)} \\ \quad + 0.75 \, ^\circ \text{DPI994} - 1.23 \, ^\circ \text{DBI1516} + 1.48 \, ^\circ \text{DBS990} \\ \quad - 1.54 \, ^\circ \text{DPE2012} - 0.71 \, ^\circ \text{DIB1314} \end{array} \tag{82}$$

$$\begin{aligned} \text{DLOG}(\text{DDISE},\text{IND}) &= 0.04 + 0.10^{\circ}\text{DLOG}(\text{DDISE},\text{IND}(1-\text{DDI})^{-1}) \\ & - 0.11^{\circ}\text{DLOG}(\text{DDISE},\text{IND}(\text{DDISE},\text{NDON}^{-1},\text{100}) \\ & - 1.19^{\circ}\text{ELOG}(\text{DDISE},\text{NDON}^{-1},\text{-1}) - 0.12^{\circ}\text{D(DDIE},\text{DDI}) \\ & - 0.16^{\circ}\text{DDP}(\text{DDISE}) \\ & - 0.12^{\circ}\text{DPD}(\text{DDISE}) - 0.07^{\circ}\text{DPD}21 \\ & - 0.96^{\circ}\text{END}(\text{DDIE},\text{NDON}^{-1},\text{-1}) + 7.5^{\circ}\text{DLD}(\text{PPR}\text{MAND}^{-1},\text{100}) \\ & - 0.96^{\circ}\text{END}(\text{DDIE},\text{NDON}^{-1}) + 0.2^{\circ}\text{DDI}(\text{PPR}\text{MMD}^{-1}) \\ & - 0.29^{\circ}\text{DDI}(\text{SID}(\text{POR}\_{\text{LIN}}) + 0.02^{\circ}\text{DDI}(\text{PDI}(\text{DDI}(\text{SID})) \\ & - 0.42^{\circ}\text{EPT}(\text{DDI}(\text{SID}(\text{SID}(\text{SID})) \\ & - 0.25^{\circ}\text{DPI}197) \end{aligned} (84)$$
 
$$\begin{aligned} \text{DLOG}(\text{DDIG}\_{\text{LIN}}) &= 0.04 + 0.42^{\circ}\text{EPT}\_{\text{LIN}}\text{DDI}(\text{SID}(\text{SID}(\text{SID}(\text{SID}(\text{S$$

Transport

$$\begin{aligned} \text{DLOG}(\text{DDIS\\_TRA}) &= 0.01 - 0.21^\circ \text{DLOG}(\text{PDIS\\_TRA}/\text{PGDPNOIL}^\circ 100) \\ &- 0.82^\circ \text{ECT\\_DDIS\\_TRA\\_N}(-1) + 0.11^\circ \text{D(DP2016)} \end{aligned} \tag{88}$$

$$\begin{aligned} \text{DLOG}(\text{DGAS\\_TRA}) &= -0.48^\circ \text{ECT\\_DGAS\\_TRA\\_N}(-1) \\ &- 0.11^\circ \text{DLOG}(\text{PGAS\\_TRA/PGDPNOIL}^\circ 100) \\ &+ 0.06^\circ \text{DBS990} - 0.05^\circ \text{DBT88} + 0.02^\circ \text{DBT90} + 0.04 \end{aligned} \tag{89}$$

ð94Þ

$$\begin{aligned} \text{DLOG}(\text{DKKR\\_RES}/\text{POP}) &= 0.01 - 0.24^\* \text{DLOG}(\text{PKER\\_RES}/\text{CPI}^\*100) \\ &- 0.49^\* \text{ECT\\_DKER\\_RES} \, \_N(-1) \\ &+ 0.40^\* \text{DLOG}(\text{PELE\\_RES}\\_\text{CONS}/\text{CPI}^\*100) \\ &- 0.29 + 0.36^\* \text{DP}1990 + 0.52 \, ^\bullet \text{DDP}1991) \end{aligned} \tag{90}$$

Residential

$$\begin{aligned} \text{DLOG(DELE\\_RES/POP)} &= 0.03 - 0.31^{\circ} \text{ECT\\_DEL\\_RES\_N(-1)} \\ &\quad - 0.11^{\circ} \text{DLOG(PEL\\_RES\_{NCS}^{\circ})} \\ &\quad + 0.31^{\circ} \text{DLOG(DEL\\_RES(-1)/POP(-1))} \\ &\quad + 0.39^{\circ} \text{DLOG(CDD)} \\ &\quad + 0.08^{\circ} \text{DLOG(CDD)} \\ \end{aligned} (91)$$

$$\begin{aligned} \text{DLOG(DKER\\_RES/POP)} &= 0.01 - 0.24^{\circ} \text{DLOG(PKE\\_RES/CP100)} \\ &\quad - 0.49^{\circ} \text{ECT\\_DER\\_RES\_N(-1)} \\ &\quad + 0.40^{\circ} \text{DLOG(PEL\\_RES\_{NCS}(-1))} \\ &\quad - 0.29^{\circ} \text{DLOG(PEL\\_ES\\_CON/CP100)} \\ &\quad - 0.29^{\circ} \text{DDP1987} + 0.36^{\circ} \text{DDP1990} + 0.52^{\circ} \text{DDP1991}) \\ \text{DLOG(DLPG\\_RES)} &= 0.20 - 0.21^{\circ} \text{ECT\\_DLP\\_N(-1)} + 0.38^{\circ} \text{DLOG(GDP)} \\ &\quad + 0.16^{\circ} \text{DDI} 112 - 0.18^{\circ} \text{DST2014} - 0.03^{\circ} \text{DDP2101} \end{aligned} (9) \end{aligned}$$

Commercial, Government, and Agriculture

$$\begin{aligned} \text{DLOG}(\text{DELE\\_COMM}) &= -1.26^{\circ} \text{ECT\\_DELE\\_COMM\\_NEW}(-1) + 0.07\\ &+ 0.21^{\circ} \text{DLOG}(\text{DELE\\_COMM}(-1)) \\ &+ 0.29^{\circ} \text{DLOG}(\text{GVADIS} + \text{GVATRACOM} \\ &+ \text{GVAFBU} + \text{GVACTHS} + \text{GVACNON}) \\ &- 0.13^{\circ} \text{DLOG}(\text{PELE\\_COMM/\text{CP1}\,100}) \\ &- 0.08^{\circ} \text{DLOG}(\text{IFDIS} + \text{IFTRACOX} + \text{IFFIBU} \\ &+ \text{IFOTHS} + \text{IFCON}) - 0.10^{\circ} \text{D(DBT2017)} \\ &+ 0.13^{\circ} \text{D(DP2012)} - 0.08^{\circ} \text{DB2002} \end{aligned}$$

$$\begin{aligned} \text{DLOG}(\text{DELE\\_GOV}) &= 0.05 - 0.08^\* \text{DLOG}(\text{PELE\\_GOV}/\text{PCDFGOV}^\*100) \\ &- 0.76^\* \text{ECT\\_DLE\\_GOV\\_N} (-1) + 0.11^\* \text{D(DP2017)} \\ &+ 0.19^\* \text{D(DP2018)} \end{aligned}$$

ð Þ 95

$$\begin{aligned} \text{DLOG(DELE\\_AGR)} &= 0.03 + 0.21^\prime \text{DLOG(DELE\\_AGR(-1))} \\ &- 0.05 \, ^\prime \text{DLOG(IFAGB)} + 0.04 ^\prime \text{DLOG(IFAGB(-1))} \\ &+ 0.27 ^\prime \text{DLOG(WAGE/PGDPAGE^\prime 100)} \\ &- 1.02 ^\prime \text{ECT\\_DELE\\_AGR\\_N(-1)} \\ &+ 0.32 ^\prime \text{D(DP2009)} \end{aligned}$$

#### Note


The last three types of dummy variables are created by Autometrics in Ox Metrics.

Fig. D.1 The real block indicators: GDP, GVANOIL, and IFNOILP. (Source: EViews in-sample simulations)

Fig. D.2 The fiscal block indicators: GEXP, GREVOIL, and GREVNOIL. (Source: EViews in-sample simulations)

Fig. D.3 The prices block indicators: CPI inflation, %, CPIFOOD inflation, % and DI\_Z. (Source: EViews in-sample simulations)

Fig. D.4 The labor market block indicators: ET, ETNOIL, and ETMANNO. (Source: EViews in-sample simulations)

Fig. D.5 The monetary block and external block indicators: M0 growth, XGNOIL, M. (Source: EViews in-sample simulations)

Fig. D.6 Energy block, CO2 emissions block, and population and age cohort block indicators: DEN\_TOT\_KSA, CO2\_EN\_TOT\_KSA, POPW. (Source: EViews in-sample simulations)



### Appendix D: KGEMM in-Sample Simulations Results, 1999–2019

### Appendix E. Consumer Price Indexes Weights and CO2 Conversion Factors

¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ ¼ CO2 Conversion Factors and their Sources Used in the CO2 Emissions Block 1 MTOE 10^6 TOE. https://en.wikipedia.org/wiki/Tonne\_of\_oil\_equivalent#:~:text¼6%20External%20 links-Definitions,toe%20%3D%2041.868%20gigajoules%20(GJ) 1 TOE of Crude Oil 7.33 Barrel of Crude Oil. https://en.wikipedia.org/wiki/Tonne\_of\_oil\_equivalent#:~:text¼6%20External%20 links-Definitions,toe%20%3D%2041.868%20gigajoules%20(GJ) 1 Barrel of Crude oil 0.43 Metric Tons of CO2 https://www.epa.gov/energy/greenhouse-gases-equivalencies-calculator-calcula tions-and-references 1 TOE of Diesel 0.99 Ton of Diesel https://en.wikipedia.org/wiki/Tonne\_of\_oil\_equivalent#:~:text¼6%20External%20 links-Definitions,toe%20%3D%2041.868%20gigajoules%20(GJ) 1 Ton of Diesel 7.5 Barrels of Diesel https://qp.com.qa/ar/Pages/ConversionFactor.aspx 1 Barrel 42 Gallon 1 Gallon of Diesel 0.01018 Metric Tons CO2 https://www.epa.gov/energy/greenhouse-gases-equivalencies-calculator-calcula tions-and-references 1 TOE of HFO 1.04 Ton of Diesel https://www150.statcan.gc.ca/n1/pub/57-601-x/00105/4173282-eng.htm 1 Ton of Fuel Oil 6.7 Barrel of Fuel Oil https://qp.com.qa/ar/Pages/ConversionFactor.aspx It is also assumed that 1 TOE HFO 1 Ton HFO. 1 Barrel of Fuel Oil 0.43 metric tons of CO2 https://www.epa.gov/energy/greenhouse-gases-equivalencies-calculator-calcula tions-and-references 1 MTOE of Natural Gas 39.2 Mcf https://www.energy-sea.gov.il/English-Site/Pages/Data%20and%20Maps/calc.aspx 1 Mcf of Natural Gas 0.0548 Metric Tons of CO2 https://www.epa.gov/energy/greenhouse-gases-equivalencies-calculator-calcula tions-and-references


https://www.seai.ie/data-and-insights/seai-statistics/conversion-factors/


https://www.unitjuggler.com/convert-energy-from-toe-to-MWh.html


¼ 1 TOE of Liquefied Petroleum Gas (LPG) 0.887862914 Ton of LPG

https://www.seai.ie/data-and-insights/seai-statistics/conversion-factors/


## References


## Index

#### A

Autometrics, 2, 21, 23, 24, 28, 103, 107, 108, 111, 121, 136, 138, 153

#### E

Equilibrium correction modeling (ECM), 7, 16, 21, 23, 41, 85, 99, 101, 106, 108, 121–125, 138–153

#### G

General-to-specific modeling strategy, 21, 107

#### K

KAPSARC Global Energy Macroeconometric Model (KGEMM), 1–3, 15, 18, 19, 21–82, 85–97, 99–108, 112, 157

#### M

Macroeconometric model, 1–3, 5, 6, 16, 17, 19, 21, 31, 38, 85, 89

#### S

