Policy Research Working Paper 9585 Power Market Sophistication and Sector Outcomes A Focus on Social Performance, Electricity Reliability, and Renewable Energy Penetration Djeneba Doumbia International Finance Corporation Economics and Private Sector Development Vice Presidency March 2021 Policy Research Working Paper 9585 Abstract This paper exploits a novel and comprehensive dataset on markets is associated with higher electricity access, better power market structure over 1989–2020 to analyze the rela- consumer affordability, larger renewable energy penetra- tionship between power market sophistication—defined as tion, and lower system average interruption duration index. the move toward a more competitive market—and final The results also highlight that, for certain steps in power sector outcomes: social performance, electricity reliability, market sophistication, improvements in sector outcomes and renewable energy penetration. Unlike most previous are greater. For instance, moving from vertically integrated studies on the performance of power sector reforms, the utility models to single buyer models is associated with paper relies on the de facto implementation of reforms relatively larger improvements in access to electricity and rather than de jure reform adoption. The results of panel electricity reliability, while moving from wholesale com- regression models suggest that moving from vertically petition to retail competition models is associated with a integrated utility models toward more sophisticated power relatively larger penetration of renewable energy. This paper is a product of the Economics and Private Sector Development Vice Presidency, the International Finance Corporation. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http:// www.worldbank.org/prwp. The author may be contacted at ddoumbia@ifc.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Power Market Sophistication and Sector Outcomes: A Focus on Social Performance, Electricity Reliability, and Renewable Energy Penetration Djeneba Doumbia1 Keywords: Power market sophistication; Power sector reforms; Sector outcomes JEL Codes: D40; Q43; O11; P18; O38 1 I am grateful to Neil Gregory, Camilo Mondragon Velez, Jevgenijs Steinbuks, Justice Tei Mensah, and Nouhoum Traore for their helpful comments. I also thank Elcin Akcura and Emelly Mutambatsere for sharing the global power market structure database. The findings, interpretations and conclusions expressed in this paper do not necessarily reflect the views of the World Bank Group, its Board of Directors, or member states. 1. Introduction Power infrastructure is critical for economic development and growth. Access to reliable electricity services is associated with economic development and growth, through various channels, including the role of electricity in power industrial processes and as a key factor in the production of goods and services in most productive sectors (Burke, Stern, and Bruns 2018; World Bank 2017). For instance, Perez-Sebastian et al. (2020) show that the growth of electricity infrastructure explains 32 percent of the observed increase in GDP per capita in Brazil, including through its effects on the reallocation of inputs toward more productive activities. Studies also show that electrification has a positive effect on industrial output, agricultural and manufacturing employment, and human development (Lee, Miguel, and Wolfram 2020; Rud 2012; Kline and Moretti 2014; Lipscomb, Mobarak, and Barham 2013). At the microeconomic level, the lack of secure and reliable electrical power has been identified as a key constraint on doing business that adversely impacts firm performance in developing countries (Abdisa 2018; Cole et. al 2018). Electricity shortages also adversely impact employment, via their reducing effects on entrepreneurial activity, productivity, output, and export competitiveness (Mensah, 2018). At the household level, electrification positively impacts income and expenditure (Bridge, Adhikari, and Fontenla, 2016; Khandker, Barnes, and Samad 2012), and education (Akpandjar and Kitchens, 2017). The importance of electricity is further highlighted in the Sustainable Development Goal 7 Ensuring access to affordable, reliable, sustainable, and modern energy for all. To improve electricity sector performance and outcomes, many countries have implemented electricity sector reforms to introduce market-type mechanisms. In the 1990s, the pace of liberalization, restructuring and privatization in the power sector has accelerated and spread globally, including a shift toward a more market-based approach (Lee and Usman 2018). While the triggers of the 1990s power sector reforms were different across countries, the prevalent objectives in many developing countries involved addressing poor financial management and technical delivery, increasing access to electricity and investing in sufficient power supply capacity to meet the growing demand for electricity (Besant-Jones 2006; Gratwick and Eberhard 2008). Existing studies suggest that introducing competition in the electricity sector raises labor productivity and capacity utilization, leading to improved performance in electricity generation (for example, see Zhang, Parker, and Kirkpatrick 2008). Other studies show that electricity sector reforms generally provide better outcomes, although the efficiency gains may not always reach the end users (Jamasb, Nepal, and Timilsina 2015). While the literature has analyzed the 2 impact of the de jure adoption of power sector reforms on sector outcomes, limited research has been devoted to the links between the de facto implementation of reforms, that is, the current structure of power markets and outcomes. In fact, the de jure adoption of reforms does not always translate into de facto reform implementation (Urpelainen and Yang 2019). Using a newly assembled and comprehensive dataset on power market structure that depicts the de facto implementation of reforms and that covers the period 1989–2020, this paper complements existing studies by (a) assessing whether the introduction of a more sophisticated power market structure is associated with better final power sector outcomes in social performance, electricity reliability, and renewable energy penetration; and (b) examining which steps in power market sophistication provide better outcomes. The results show that power market sophistication—defined as a move from vertically integrated utility models to more sophisticated power markets—is associated with greater access to electricity, better affordability, reduced system average interruption duration; and improved renewable energy penetration. The findings also suggest that certain steps in power market sophistication provide better sector outcomes than others – the movement from vertically integrated utility models to single buyer models is associated with more improved sector outcomes compared to those that moved from single buyer models to wholesale competition models and from wholesale competition to retail competition models. The remainder of the paper is structured as follows. Section 2 provides background on power sector reforms, including the reasons for reforming the sector, reform implementation, and reform outcomes. Section 3 describes the data and explains the econometric methodology. Section 4 presents empirical results. Section 5 provides concluding remarks. 2. Background: Power sector reforms This section discusses (a) the main reasons for reforming the power sector in both developed and developing countries starting in the 1970s; (b) key elements and reform steps of the 1990s standard reform model as well as the implementation of the standard reform model across countries; and (c) a review of studies on the successes and failures of the standard reform model with a focus on the studies related to access, affordability, reliability, and renewable energy penetration, which are the most relevant to this study. 2.1. Why reform the power sector? Starting in the 1970s, elevated fuel prices, environmental concerns, technological progress, and a desire for more economic efficiency led to a rethink of the vertically integrated monopoly model (Tuttle et al. 2016). The oil crises of the 1970s raised the awareness of governments on 3 the vulnerability of countries to oil price shocks and the associated higher costs of fuel imports, leading to greater attention to the benefits of energy conservation. The late 1970s saw the onset of power sector reforms in many countries. The Government of the United Kingdom initiated a restructuring of the country’s traditional vertically integrated electric power monopoly to introduce competition primarily at the wholesale level where generators compete in the market. In the United States, Congress passed the Public Utilities Regulatory Policy Act in 1978. This introduced competition in generation and opened the system to independent power producers (IPPs). In the developing world, Chile became a pioneer by undertaking comprehensive power sector reforms in 1978 (Bacon 1995). The reasons for restructuring and privatizing the power sector included the poor efficiency levels of state-owned utilities and a desire to mobilize foreign equity capital to help decrease the debt burden (Ljung 2007). These reforms aimed at boosting the emphasis on economic efficiency within the electricity sector by establishing a more independent regulatory authority, the National Energy Commission, which would attract new private investments (Henisz and Zelner 2013). The pioneering reforms, including in Chile and the United Kingdom, inspired many developed and developing countries to follow suit during the 1990s (Bacon and Besant-Jones 2001). In the 1990s, the pace of liberalization, restructuring, and privatization in the power sector accelerated, spread globally, and shifted toward a more market-based approach (Lee and Usman 2018). A mixture of theoretical developments, technological progress in the electricity sector, and political or ideological drivers seemed to underpin the rationale behind the reforms (Sen, Nepal, and Jamasb 2016). The triggers of the 1990s electricity reforms were different across countries. In many advanced countries, reforms in the electricity sector sought to enhance economic efficiency, particularly through reductions in tariffs (Bacon 1995; Victor and Heller 2007). In the OECD countries, the reforms were driven by the desire to restructure the sector into its competitive functions in the context of raising the level of existing commercial standards of performance (Besant-Jones 2006; Sen, Nepal, and Jamasb 2016; Williams and Ghanadan 2006). In many developing countries, the objectives of the reforms were to promote investment in sufficient power supply capacity to meet the growth in demand for electricity, increase access to public electricity, and address poor performance, financial management, and technical delivery (Bacon 2018; Besant-Jones 2006; Gratwick and Eberhard 2008). 2.2. The 1990s standard reform model 2.2.1. Key elements of the standard reform model 4 The market-oriented reforms which were pioneered in the 1980s and 1990s by the United Kingdom, Norway, Chile and the United States involved a set of policy measures. These reforms were labelled differently in the literature (Foster and Rana 2020): a “blueprint for action” (Bacon 1995), a “standard model” (Littlechild 2001), “standard prescription” (Hunt 2002), or “textbook architecture” (Joskow 2008) of sector norms. Some of the key elements of the reform model of the 1990s included (a) corporatization and commercialization, with the objective of strengthening financial discipline; (b) restructuring or unbundling state-owned vertically integrated utilities (VIUs) to allow private sector or foreign participation or ownership in the sector; (c) the establishment of independent regulation to promote efficiency and transparency in sectoral management; (d) the introduction of independent power producers in the sector to promote investment in generation; and (e) privatization of the competitive segments (that is, generation and distribution) to provide incentives for cost efficiency and promote financial discipline (Gratwick and Eberhard 2008; Joskow 2008; Kessides 2012; Sen 2014; Victor and Heller 2007). This set of reforms was strengthened by regulatory interventions, including, for instance, changes in pricing design (Bensch 2019). 2.2.2. Power market structure across countries The implementation of these reforms and the resulting market structure vary based on specific country contexts. For instance, in many developing countries, electricity reform has been “an incomplete, uneven, and irregular process that entails a complex set of interactions between the state and the market” (Kessides 2012, 3). Since the 1990s, albeit at a different pace, many countries have established more competitive power markets. In 1989, about 90 percent of countries and territories followed a state-owned VIU model, compared with only 35 percent in 2020 (figure 1). Many countries have adopted one of the two variants of the single-buyer model. Figure 1 shows that, in 2020, about 21 percent of countries and territories had adopted a model whereby the single buyer owns and controls some generation assets. This compares with fewer than 2 percent of countries in 1989. Currently, about 12 percent of countries have adopted an unbundled single-buyer model versus fewer than 1 percent in 1989. Power markets in only 8 percent (16 percent) of countries were labeled as wholesale competition (retail competition) models in 2020 (figure 1). Yet, reforms have developed at a higher speed in most European and North American countries and parts of Latin America, compared with many African countries. For instance, while two-thirds of North American countries are currently in a retail competition model, VIU models are predominant in East Asia and the Pacific and in Sub-Saharan Africa, with, respectively, 60 percent and 48 percent of countries in these regions engaged in a VIU model that is either state-owned or majority private. Below is an overview of the current status of power market structure across groupings (income, region, fragility 5 status, and political system) based on four groups of power markets that the analysis in the present paper relies on—VIU models, single-buyer models, wholesale competition, and retail competition models. These models are defined in figure 2. • Vertically integrated utility models State-owned VIU models remain predominant across the world. The models in more than half of low-income countries are classified as state-owned VIUs. This compares with 32 percent of lower-middle-income countries, 35 percent of upper-middle-income countries, and 32 percent of high-income countries. Most of the countries with state-owned VIU models are islands or small states. Across regions, VIU models are more predominant in East Asia and the Pacific and in Sub-Saharan Africa. Among countries classified as fragile and conflict-affected states (FCS countries), 70 percent follow state-owned VIU models. Power markets in most countries with autocratic political systems follow state-owned VIUs (table 1). Few countries (about 8 percent) have privatized (to some extent) their VIUs. • Single buyer models Currently, in about 21 percent of countries and territories, a single buyer owns or controls some generation assets. This compares with fewer than 2 percent of countries in 1989. Currently, about 12 percent of countries follow an unbundled single-buyer model versus fewer than 1 percent in 1989. Across income groups, single-buyer models are more predominant in middle- income countries. Many countries in Latin America and the Caribbean, the Middle East and North Africa, and Sub-Saharan Africa currently follow a variant of the single-buyer model. For instance, among Sub-Saharan African countries, 42 percent have power markets in which a single buyer owns or controls some generation assets, and 10 percent follow an unbundled single-buyer model. Regarding fragility status, these models are more prevalent among non- FCS countries (more than 35 percent), compared with FCS countries (about 24 percent). Regarding power structure models across political regimes, more than half of countries with autocratic political systems (11 of 21) have a power market with a single buyer. Among countries classified as anocracies and democracies, 49 percent and 35 percent, respectively, run a variant of the single-buyer model (table 1). • Wholesale competition models Power markets in only 8 percent of countries were labeled as wholesale competition models in 2020 (figure 1). While about 17 percent of upper-middle-income countries have power markets labeled as wholesale competition models, these models are nonexistent in low-income 6 countries, FCS countries, autocracies, the Middle East and North Africa, and Sub-Saharan Africa (table 1). • Retail competition models Since the early 1990s, more countries have adopted retail competition models (fewer than 1 percent of countries in 1989 versus 16 percent in 2020). The majority of high-income countries (38 percent), countries in Eastern Europe and Central Asia (42 percent), North American countries (67 percent), and countries with democratic political systems (35 percent) have power markets labeled as either partial or full retail competition models. In contrast, retail competition is nonexistent in low-income countries, the Middle East and North Africa, South Asia, Sub- Saharan Africa, FCS countries, and countries with autocratic political regimes (table 1). Figure 1. Share of countries by prevailing power market structure, 1989 versus 2020 Power market structure, 1989 versus 2020 100 89.6 1989 Percentage of countries 80 2020 60 40 35.2 20.9 20 11.7 13.5 7.8 7.8 5.7 1.3 0.4 0.0 2.6 0.0 0.9 2.6 0.0 0 State-owned Majority Controlling Unbundled Bilateral Power Partial Full private some trading exchange generation assets VIU Single Buyer Wholesale competition Retail competition Source: Author based on the global power market structure database. Note: The definition of different power market structures is presented in Figure 2. 7 Table 1. Power market structure across country groups Wholesale Retail VIU Single buyer competition competition Share of countries by Controlling power market structure as State- Majority some Unbundle Bilateral Power of 2020 owned private generation d trading exchange Partial Full By income group LIC 48.4 6.5 41.9 3.2 0.0 0.0 0.0 0.0 LMIC 31.9 0.0 34.0 25.5 2.1 4.3 2.1 0.0 UMIC 35.0 5.0 20.0 13.3 8.3 8.3 6.7 3.3 HIC 31.7 10.1 6.3 6.3 0.0 7.6 1.3 36.7 By region EAP 50.0 10.0 17.5 7.5 0.0 2.5 2.5 10.0 ECA 21.7 5.0 3.3 10.0 8.3 10.0 0.0 41.7 LAC 36.2 12.8 23.4 4.3 2.1 10.6 10.6 0.0 MENA 38.1 0.0 23.8 38.1 0.0 0.0 0.0 0.0 NA 0.0 33.3 0.0 0.0 0.0 0.0 0.0 66.7 SA 37.5 0.0 25.0 25.0 0.0 12.5 0.0 0.0 SSA 40.0 8.0 42.0 10.0 0.0 0.0 0.0 0.0 By fragility status FCS 70.3 5.4 10.8 13.5 0.0 0.0 0.0 0.0 Non-FCS 27.4 6.2 23.5 11.7 3.4 7.3 3.4 17.3 By political system* Autocracy 47.6 0.0 28.6 23.8 0.0 0.0 0.0 0.0 Anocracy 42.2 2.2 33.3 15.6 2.2 2.2 0.0 2.2 Democracy 13.1 1.0 21.2 14.1 3.0 12.1 6.1 29.3 Note: The World Bank classification is used for income, region and FCV groupings. For political system, using Polity2 indicator (ranging from -10 to +10 with special values -66, -77, and -88) three groups have been defined as follow: autocracy (-10 to -6); anocracy (-5 to +5 and the three special values); and democracy (+6 to +10). *The latest data available for the political indicator (year 2018) are used. 8 Figure 2. Power market structure taxonomy High Retail Competition – Partial (only large industrial and commercial customers); full (industrial, commercial and residential customers) Allows customers to have access to competing generators and buy electricity from a retailer of their choice Wholesale Competition –bilateral trading; power exchange A set of distribution companies buy electricity from competing generators and distribute it to customers via a bundled transmission service. Single Buyer – controlling some generation assets; fully ownership unbundled One entity buys electricity from independent power producers (IPPs) and owns and controls transmission and distribution. Vertically Integrated Utility – state-owned; majority private A single entity owns and controls generation, transmission, and distribution. Low None Intermediate High Level of competition Source: Author using Nepal and Jamasb (2015). 2.3.The outcomes of the standard reform model Evidence across countries suggests that, if efficiently implemented, the standard reform model can lead to improvements in many dimensions of operating performance and efficiency and generate desirable macroeconomic consequences (Jamasb, Nepal, and Timilsina 2015). For electricity reforms to be successful, the progress of the reforms should be coordinated across aspects of the development process, including macroeconomic, political, sectoral, and financial (Nepal and Jamasb 2015). Over recent years, a body of literature on the performance of electricity sector reforms at both the micro and macro levels has emerged. Some of the key results from international empirical studies (as discussed in Kessides 2012, 3) are as follows: (a) moderate and unstable efficiency gains from privatization reforms; private sector participation provides benefits only if combined with the presence of an independent regulator; (b) strong evidence that competition implies significant improvements in performance; (c) 9 liberalization reforms do not always result in declines in retail electricity prices; these reforms should lead to higher prices in countries where electricity prices are inefficiently low; and (d) liberalization reforms contributed to the reduction of historic pricing distortions in the electricity sector. This section discusses the following empirical studies on the impact of power sector reforms on sector outcomes related to access to, affordability and reliability of electricity, and renewable energy penetration – which are believed to be the most relevant to the present study. Using data from 108 countries over 1982–2016 and relying on an aggregated reform index of eight individual reform variables, Dertinger and Hirth (2020) find a positive effect of a fully- fledged reform on electricity access. Their findings show that a full reform range leads to a 20- percentage point increase in connection rates and a 62 percent increase in per capita residential consumption. However, in regions where access is a lesser challenge, such as Eastern Europe and Central Asia, no evidence of the effects of reforms on access has been found. Foster and Rana (2020) find a significant and positive impact of private sector participation on generation capacity and electricity access, particularly in low-income countries—a percentage point increase in private sector participation is associated with a 0.3 percentage point increase in electricity access. In contrast, their findings hint at a significant negative relationship among regulation, restructuring, and electrification. The effects of electricity reforms on electricity access vary by types of reform. Vagliasindi (2012) finds a significant and positive relationship between privatization, regulation, and residential electricity access in 22 countries, but a negative link between partial unbundling on connection rates. Using panel data on 51 developing countries, Zhang, Parker, and Kirkpatrick (2008) show that competition leads to improved service penetration, capacity expansion, labor efficiency, and favorable prices for industrial users. While, on their own, privatization and regulation do not seem to lead to improved economic performance, the coexistence of these two reforms appears to be correlated with improved electricity availability, enhanced generation capacity, and greater labor productivity. Studies on the effects of electricity reforms on electricity reliability exhibit mixed results. While some studies find that electricity sector reforms increase losses (Erdogdu 2011), others show a reducing effect. For instance, using data on 18 Latin American countries findings from, Balza, Jiménez, and Mercado Díaz (2013) suggest that privatization is associated with improved quality and efficiency in the sector because of the reducing effect on electricity losses and the enhancement effect on generation capacity. Evidence from panel data covering 86 countries shows that reform indicators associated with the entry of independent power 10 producers, unbundling, the establishment of regulatory agencies, and the introduction of a wholesale spot market exhibit a reducing effect on transmission and distribution losses (Nagayama 2010). Using data on 108 developing countries over 1982–2016, Dertinger and Hirth (2020) find no robust evidence for any impact of power sector reforms on transmission and distribution losses. Regarding the effects of electricity reforms on renewable energy penetration, Foster and Rana (2020) find a positive association between private sector participation and regulation, and the expansion of renewable energy. The effect of competition is significant and positive in middle- income countries whereas it is negative in low-income countries. Using data on four Latin American countries, Ruiz-Mendoza and Sheinbaum-Pardo (2010) show no evidence of the reducing effect of restructuring on carbon dioxide emissions. This paper builds on and complements previous research by relying on a newly constructed dataset on the de facto status of power markets covering the period 1989-2020, studying the extent to which power market sophistication is associated with improved final sector outcomes, and examining whether certain steps in power market sophistication provide better sector outcomes. 3. Data and econometric methodology 3.1.Data The paper relies on an unbalanced panel dataset on the structure of power markets and final sector outcomes. This section provides a description of the global power market structure database and the indicators of final sector outcomes, as well as data sources. 3.1.1. Global power market structure database The Global Power Market Structure Database was constructed as part of the flagship report of the International Finance Corporation (IFC), Creating Power Markets. The construction methodology of the database includes the definition of various types of power markets based on an extensive literature review and consultations with power sector specialists; the collection of data and the assessment of power market structures that relies on various data and information sources, including ScienceDirect, World Bank working papers, IFC project documents, the Energy Sector Management Assistance Program; and a validation step using various sources, including existing data, and based on a review by industry experts and power 11 sector specialists. 2 The database covers 230 countries and territories across the world over the period 1989–2020 and includes eight types of power markets as defined in table 1. 3 For the purpose of this study, an indicator that depicts the movements from one power market structure to another (by country and year) has been constructed based on four groups of power market structures—the VIU, single-buyer, wholesale competition, and retail competition models. 3.1.2. Power sector outcomes This section describes the variables used as final sector outcomes in the econometric analysis. The sector outcome indicators include social performance, electricity reliability, and renewable energy penetration. Social performance - Access to electricity is measured as the percentage of population with access to electricity. The data are from the World Bank’s World Development Indicators (WDI) and Sustainable Energy for All (SE4ALL) databases. - Consumer affordability of electricity is an index based on three sub-indicators: cost of subsistence consumption; affordability of the connection fee; and policy to support low-volume consumers. The data are from RISE (Regulatory Indicators for Sustainable Energy) project. Electricity reliability - System Average Interruption Duration Index (SAIDI) is the average total duration of power outages (in hours) experienced by a customer in a year in the largest business city in each country. The data are from the World Bank Doing Business compilation. - System Average Interruption Frequency Index ((SAIFI) measures the average number of service interruptions experienced by a customer in a year in the largest business city in each country. The data have been obtained from the World Bank Doing Business compilation. Renewable energy penetration - Electricity production from renewable sources (excluding hydroelectric) is measured as the percent share of total electricity production. Renewable electricity is the share 2 The detailed methodology for the construction of the Global Power Market Structure Database and the limitations are explained in IFC, “Creating Markets: Power Markets for Development” (forthcoming). 3 The list of countries and territories covered in the database is presented in table A3 in the appendix. 12 of electricity generated by renewable power plants in total electricity generated by all types of plants. The data are from the International Energy Agency statistics. Tables A2 in the appendix shows summary statistics of the outcome variables. The indicators of access to electricity and renewable energy penetration exhibit a larger number of year- country observations compared with electricity reliability (SAIDI and SAIFI). 3.2.Econometric methodology The empirical strategy consists of two steps that correspond to the two main research questions with which this paper grapples, as follow: (a) has the introduction of more sophisticated power markets improved final sector outcomes? and (b) do certain steps or transition paths in power market sophistication provide better sector outcomes than others? The first estimation leverages documented changes in power market structure over time to assess the extent to which the introduction of a more sophisticated power market is associated with sector outcomes, irrespective of the structure of a market following reform implementation. The analysis relies on a newly constructed panel dataset on power market structure covering 230 countries and territories over the period 1989–2020, as described in section 3.1. The model considers changes in outcomes over time across a group of countries that introduced a more sophisticated power market structure over the period of study, moving from a VIU model to a more sophisticated power market relative to countries that did not experience any changes in power market structure. The econometric estimation is described in equation (1), as follows: = α + + x + μ + (1) where denotes a vector of sector outcomes for country at time as described in section 3.1; is a set of controls for a country’s economic and political characteristics and sector- specific characteristics, including GDP per capita; and polity2 is a regime-type indicator that captures a country’s political regime authority, and power system size (installed electricity capacity in gigawatts). To test the robustness of the results, other control variables, including democratic accountability (from the International Country Risk Guide Database) and population density (from WDI) are used. μ are country fixed effects. denotes the error term. Table A1 in the appendix summarizes the description and source of the variables used in the present paper. In case of variation in reform timing, as in this paper, the two-way fixed effects model, which includes both country and time fixed effects, can lead to biased estimates and may not be appropriate (de Chaisemartin and D’Haultfœuille 2020). Therefore, as a robustness 13 test, in addition to country fixed effects, this paper includes dummies for short-term shocks that are common across countries. 4 The variable of interest, , is a dummy variable that switches on (takes the value of 1) in the year of the transition from a VIU model to a more sophisticated power market structure in country and keeps the value 1 in all following years and 0 otherwise. 5 For countries that experienced a reversal in power market structure, corresponding observations were dropped from the analysis. 6 Countries that already had a more sophisticated power market structure than the VIU models at the beginning of the period of study were also dropped from the estimation regressions of equation (1). 7 The indicator was constructed using the four groups of power market structures listed in figure 2—VIU, single buyer, wholesale competition, and retail competition (partial and full). The estimated coefficient captures changes in outcomes before and after the change in power market structure that occurs in the countries that undertook the transition from VIU models to more sophisticated power markets. Given the endogeneity issues that may arise from the fact that power market sophistication might affect sector performance, just as performance can influence the reform decision, the results are interpreted as conditional correlations rather than causal effects. In the second step, the paper examines whether certain steps in power market sophistication provide better outcomes than others. Three types of transition in power market sophistication are defined for this purpose: (a) from VIU to single-buyer models (step 1), (b) from single- buyer to wholesale competition models (step 2), and (c) from wholesale to retail competition models (step 3). Equation (1) is rewritten below; ∈ (1,2,3) represents the three subsamples. _ = α0_ + _ + x _ + μ_ + _ (2) In equation (2), if = 1, the sample includes all countries that moved from a VIU model to a single-buyer model at a certain point over the period of study as well as countries that retained a VIU model. If = 2, the sample includes all countries that moved from a single-buyer model to wholesale competition at a certain point over the period of study as well as countries that maintained in a single-buyer model. If = 3, the sample under study includes all countries that moved from wholesale to retail competition models and countries that retained a wholesale 4 Estimating a model including year fixed effects yields comparable results, except for the coefficient associated with access to electricity, which has a lower magnitude, although it remains significant. 5 To test the robustness of the results, the paper also estimates models using one- to three-year lags. 6 The dropped observations represent less than 1 percent of the total number of observations. 7 This includes six countries, which represent fewer than 3 percent of the total number of countries considered in this analysis. 14 competition model. The regressions rely on sample and treatment periods of interest for each step, fully exploiting the dynamics in the power market structure. 4. Estimation results Scatterplots are used as a first step to visualize the relationship between power market structure and final sector outcomes suggested by the data. Figures 3.A-D provide some evidence on the link between power market sophistication and selected sector outcomes – access to electricity, consumer affordability of electricity, and electricity reliability (SAIDI and SAIFI). Populations in countries with power markets that have advanced toward more competitive or sophisticated models tend to have better access to electricity and enjoy greater affordability. In particular, the rate of access to electricity and the consumer affordability indexes are greater than 50 in all the countries that introduced either retail or wholesale competition models. In addition, the sophistication of power markets is positively and significantly associated with better electricity reliability—SAIDI and SAIFI tend to be lower in countries in which power markets are more competitive. Countries with more sophisticated power markets tend to exhibit fewer system average interruptions and shorter durations in interruptions. A second set of evidence on the extent to which the introduction of more competitive and sophisticated systems is associated with improved sector outcomes relies on visualization using event studies. The date of change in the power market structure of each country is normalized to year t, and the indicators of final outcomes are dated accordingly to gauge sector outcome trends before and after the changes in power market structure. Figures 4.A–4.C illustrate that selected final sector outcomes tend to improve in the years following the move toward a more sophisticated power market. The introduction of a more sophisticated power market structure is followed by greater access to electricity, better consumer affordability, and higher renewable energy penetration. The introduction of private sector participation and competition brings private management and capital into the power sector. It is associated with enhanced operational efficiency and labor productivity which lead to better sector outcomes. 15 Figure 3. Power market structure and selected final sector outcomes Note: The circles depict to which extent countries with more sophisticated power markets tend to be at no less than a certain threshold in electricity access (50 percent), consumer affordability (index=50), and have relatively lower SAIDI and SAIFI. Corr. coeff= correlation coefficient. *** significant at 1 percent. Power market structure is defined as follows: 1 = state-owned VIU, 2=majority private state-owned VIU, 3=single buyer model – controlling some generation assets, 4=single buyer model – unbundled, 5=wholesale competition – bilateral trading, 6=wholesale competition – power exchange, 7=retail competition – partial, and 8=retail competition – full. 16 Figure 4. Introduction of more sophisticated power markets and selected final sector outcomes A. Access to electricity 84 After change 83 Access to electricity, % of 82 population 81 80 79 Before change 78 77 76 t-8 t-7 t-6 t-5 t-4 t-3 t-2 t-1 t t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 B. Consumer affordability of electricity Consumer affordability, index (0- 88 After change 84 80 76 100) 72 68 64 60 Before change 56 t-8 t-7 t-6 t-5 t-4 t-3 t-2 t-1 t t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 C. Electricity production from renewable sources Electricity production from renewable sources, excl. hydroelectric, % of total 3.6 After change 2.8 production Before change 2 1.2 t-8 t-7 t-6 t-5 t-4 t-3 t-2 t-1 t t+1 t+2 t+3 t+4 t+5 t+6 t+7 t+8 Source: Author using power market structure database, WDI database, International Energy Agency statistics, and RISE project. Note: The graphs show the evolution of different outcome variables 8 years before and 8 years after the introduction of a more sophisticated power market by countries with Vertically Integrated Utility models. The year in which the country introduces a more sophisticated power market is normalized to t. Then, averages of the different indicators are calculated across the periods: from t-8 to t+8. The sample is restricted to countries that experienced a movement in their power market structure towards more sophistication. 17 To investigate empirically whether the introduction of more competitive and sophisticated power markets is associated with improved sector outcomes, the econometric estimation based on equation (1)—as explained in section 3.2—is estimated. Table 2 shows the estimation results for the three sets of sector outcomes, controlling for country economic characteristics and sector-specific characteristics (first columns for each outcome) as well as country political characteristics (second columns). Results indicate that an incremental 5.5 percent of the population had access to electricity in countries that moved from VIU models to more sophisticated power markets, including single-buyer models and wholesale and retail competition models (see table 2, column 1). This figure is about 4.8 percentage points when also controlling for country political characteristics (see column 2). In addition, the consumer affordability of electricity index increased by about 3 to 4 percent points for countries that introduced more sophisticated power markets (table 2, columns 3 and 4). The results also show that political stability as measured by the Polity2 score is positively associated with access to electricity and the consumer affordability of electricity. While there has been a significant decline in system average interruption duration in countries after they moved toward more sophisticated markets, the coefficients associated with the frequency of interruption are not statistically significant. This result holds in all three specifications (table 2, columns 5–6 for SAIDI and columns 7–8 for SAIFI). In the preferred specification, the results suggest there is a significant decrease, by 443.5 hours, in the SAIDI in these countries (see table 2, column 6). The introduction of a higher level of market sophistication would allow an enhancement of service penetration, the promotion of an efficient utilization of capacities, the expansion of capacities, and the reduction of energy losses. Regarding renewable energy penetration, the share of electricity production from renewable sources (excluding hydroelectric) increased by about 2 percentage points in countries that introduced more sophisticated power markets (columns 9 and 10). The results indicate that income per capita and system size are positively associated with a higher share of electricity production from renewable sources. The results discussed above are robust to the introduction of dummies for the short-term shocks that are common across countries (table A4). The introduction of additional control variables, including democratic accountability and population density, also yields similar results (see table A5). For instance, the results show that movements from VIU models to more sophisticated power markets are associated with an additional 4.7 percent increase in access to electricity in exposed countries. Population density and democratic accountability are also 18 positively associated with power market sophistication. The coefficient associated with SAIFI—which was not significant in the baseline regressions—is negative and significant (table A5, column 6). The effects of sector reforms on utility management will likely not appear immediately following the de jure adoption of reforms (Dertingler and Hirth 2020). Another robustness check assumes that sector outcomes can be affected by market sophistication with a lag, despite the use of the current states of power markets to capture movements in power market structure. Along these lines, the paper uses one- to three-year lags of the variable as explanatory variables of interest. The results remain in line with the main results (tables A6 and A7 in the appendix). The coefficients associated with electricity access, consumer affordability of electricity, and electricity production from renewable sources become larger as the number of lags grows. Considering low- and middle-income countries, the results show that movements from VIU models to more sophisticated power markets are associated with larger improvements in certain final outcomes. For instance, lower-middle-income countries that introduced a power market with a higher sophistication level had an additional increase of 7 percent in the population segment with access to electricity (see table A8). However, if one controls for political characteristics, the coefficient associated with the consumer affordability of electricity becomes nonsignificant. As in the regressions using the full sample, the coefficients associated with SAIFI are negative, but not significant. Moreover, the coefficients associated with electricity production from renewable sources are positive, yet not significant (table A8, columns 11–13). 19 Table 2. Movements toward more sophisticated power markets and sector outcomes – estimation results System Average System Average Electricity production Access to electricity, % of Consumer affordability Dependent variables Interruption Duration Interruption Frequency from renewable sources, total population index index index excl. hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Introduction of more 5.517*** 4.835*** 3.744** 3.145* -442.9*** -443.5*** -12.48 -12.47 1.730*** 1.617*** sophisticated power markets (θ) (0.371) (0.391) (1.832) (1.863) (58.87) (63.71) (13.63) (14.74) (0.169) (0.174) GDP per capita 0.0318 0.0269 0.163 0.171 -5.501 -6.547 -0.448 -0.417 0.214*** 0.210*** (0.0261) (0.0336) (0.184) (0.182) (7.194) (8.563) (1.666) (1.981) (0.0154) (0.0160) System size 3.829 4.277 1.038 1.572 80.66 90.97 3.151 5.216 3.282** 3.515** (2.662) (2.651) (9.855) (9.690) (415.8) (450.6) (96.34) (104.3) (1.503) (1.504) Polity2 0.781*** 0.713*** 1.628 1.891 -0.0132 (0.0581) (0.206) (10.99) (2.542) (0.0265) Constant 74.00*** 68.41*** 79.06*** 76.21*** 508.7*** 566.6** 40.81 33.54 -2.200*** -1.899*** (0.447) (0.548) (3.395) (3.500) (176.9) (223.2) (41.08) (51.79) (0.261) (0.272) Observations 3,948 3,351 946 920 349 301 347 299 3,144 2,992 Number of countries 180 153 119 117 121 105 120 104 129 124 R-squared 0.068 0.123 0.006 0.021 0.201 0.202 0.004 0.007 0.148 0.138 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. 20 The second estimation examines whether specific steps or transitions in power market sophistication are associated with improved sector outcomes. The results show that certain steps in power market sophistication provide better sector outcomes compared with others. Increases in electricity access and decreases in SAIDI are larger in countries that moved from VIU to single- buyer models, compared with countries that moved from single-buyer models to wholesale competition models and with those that moved from wholesale to retail competition models (see table 3). For countries that moved from single-buyer models to wholesale competition models, only coefficients associated with electricity access are significant. The magnitude of the coefficient associated with consumer affordability of electricity is larger in countries that moved from VIU models to single-buyer models (step 1), compared with countries that moved from single-buyer models to wholesale competition models (step 2) and from wholesale competition models to retail competition models (step 3). However, in all specifications, the coefficients associated with affordability are not significant in steps 2 and 3. When controlling for GDP per capita and system size, the coefficient associated with affordability is significant in step 1 but becomes nonsignificant if one controls also for country political characteristics. Detailed results are presented in tables A9-A11 in the appendix. Regarding renewable energy penetration, countries that moved from wholesale to retail competition models exhibited a larger increase. They experienced a rise of 2 percent in the share of electricity production from renewables, compared with about 0.5 percent in countries that moved from VIU models to single-buyer models. This sheds light on the stronger economic, institutional, and sector-specific preconditions of countries (such as New Zealand, Norway, the United Kingdom, and so on) that made the move toward retail competition models. To assess the robustness of these results, equation (2) is estimated by adding democratic accountability and population density as control variables, in addition to system size and GDP per capita. The results presented in tables A12–A14 are broadly in line with the main results described above. Coefficients associated with access to electricity in steps 2 (1.6 percentage points) and 3 (1.7 percentage points) are similar. As for equation (1), an additional robustness test using one- to three-year lags of the variable has been performed. The results remain broadly in line with the main results. Regarding the transition from VIU to single-buyer models (step 1), the coefficients associated with electricity access, consumer affordability of electricity, and electricity production from renewable sources 21 become larger as the number of lags grows (table A15). This is also the case of access to electricity and consumer affordability in countries that transitioned from single-buyer to wholesale competition models and of electricity production from renewable sources in countries that transitioned from wholesale to retail competition models (tables A16 and A17). Table 3. Steps in power market sophistication and sector outcomes – estimation results Step 1: From Step 2: From Single Buyer Step 3: From Wholesale Vertically Integrated models to Wholesale Competition to Retail Utility to Single Buyer Competition Competition Dependent variables Access to electricity, % of population 7.529*** 3.554*** 1.403** Observations 2,653 1,212 404 Number of countries 152 90 32 Consumer affordability index 3.188 -0.703 0.649 Observations 649 443 175 Number of countries 87 62 26 Electricity production from renewable 0.486*** 0.264 2.080** sources, excl. hydroelectric, % Observations 2,373 1,067 348 Number of countries 124 74 28 System Average Interruption Duration -451.1*** -0.0856 3.194 index Observations 173 143 80 Number of countries 63 52 28 System Average Interruption Frequency -13.05 0.0573 0.821 index Observations 171 141 80 Number of countries 62 51 28 Note: Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita (in thousands of dollars), system size (in gigawatt), Polity2 score, and country fixed effects are included. 5. Concluding remarks This paper first analyzes the extent to which power market sophistication, defined as the move toward a more competitive market, promotes improved final sector outcomes: social performance, electricity reliability, and renewable energy penetration. Second, it investigates whether certain actions taken in power market sophistication are associated with improved final sector outcomes. Using a newly assembled and extensive database on power market structure over the period 1989– 2020, the paper shows that movements from VIU models to more sophisticated power markets are associated with improvements in certain sector outcomes, including access to electricity, consumer affordability of electricity, and the share of electricity production from renewable sources, and 22 decreases in the SAIDI. The findings also suggest that improvements in electricity access and decreases in the system average interruption duration tend to be greater in countries that moved from VIU models to single-buyer models, compared with countries that moved from single-buyer models to wholesale competition models and with those that moved from wholesale competition to retail competition models. Additionally, the improvement of renewable energy penetration is larger in countries that moved from wholesale to retail competition models, highlighting the stronger economic, institutional, and sector-specific preconditions in countries. These results are robust to alternative specifications, including the introduction of additional control variables and lagged indicators of power market sophistication. All in all, the findings indicate that the introduction of private sector participation and competition in the power sector is important in promoting economic development through its role in enhancing electricity access, affordability, and reliability. While the present paper studies the average extent to which the sophistication of power markets improves sector outcomes, the effect may be heterogenous and vary across time. A possible future area of research would involve exploring these heterogenous causal effects using econometric models that are adapted to a staggered adoption design under suitable assumptions and data requirements. 23 APPENDIX Table A1. Definition and source of data Variable Source Description/Definition Final sector outcomes World Development Indicators (WDI), Sustainable Access to electricity is measured as the percentage of population with Access to electricity Energy for All (SE4ALL) databases access to electricity The index (0-100) is constructed based on three dimensions: the Regulatory Indicators for Sustainable Energy (RISE) Consumer affordability of electricity affordability of subsistence consumption; the affordability of the connection project fee; and the existence of a lifeline tariff SAIDI measures the average total duration of outages (in hours) System Average Interruption Duration Index Doing Business, World Bank experienced by a customer in a year in the largest business city of the (SAIDI) country System Average Interruption Frequency Index SAIFI is the average number of service interruptions experienced by a Doing Business, World Bank (SAIFI) customer in a year in the largest business city of the country Electricity production from renewable sources, excluding hydroelectric in Electricity production from renewable sources International Energy Agency Statistics percent of total electricity production Control variables GDP per capita WDI, World Bank GDP per capita, PPP (in constant 2011 international dollars) System size U.S. Energy Information and Administration (eia) Installed electricity capacity in gigawatt The Polity score captures the regime authority spectrum of a country. The Polity 2 The Polity Project, Center for Systemic Peace Polity measure ranges from -10 (strongly autocratic) to +10 (strongly democratic) International Country Risk Guide (ICRG), Political Democratic accountability The index (0-6) captures how responsive government is to its people Risk Services (PRS) group Population density (people per squared km of land area) is midyear Population density WDI, World Bank population divided by land area in square kilometers Table A2. Descriptive Statistics, final sector outcomes Variables Standard Deviation Number of Number of Mean Between Within observations countries Access to electricity (%) 79.9 30.3 6.8 4,888 213 Consumer affordability of electricity 84.1 27.1 7.3 1,056 132 System Average Interruption Duration Index 42.4 285.2 74.3 811 154 System Average Interruption Frequency 18.2 122.3 22.9 809 154 Index Electricity production from renewable 2.7 4.9 3.4 3,705 142 sources excl. hydroelectric (%) Note: The sample of countries used include the sample of countries covered in the global power market structure database. 24 Table A3. List of countries and territories covered in the power market structure database Country Country code Region Income group Afghanistan AFG South Asia LIC Albania ALB Europe & Central Asia UMIC Algeria DZA Middle East & North Africa UMIC American Samoa ASM East Asia & Pacific UMIC Andorra AND Europe & Central Asia HIC Angola AGO Sub-Saharan Africa LMIC Anguilla Latin America & Caribbean Antigua and Barbuda ATG Latin America & Caribbean HIC Argentina ARG Latin America & Caribbean UMIC Armenia ARM Europe & Central Asia UMIC Aruba ABW Latin America & Caribbean HIC Australia AUS East Asia & Pacific HIC Austria AUT Europe & Central Asia HIC Azerbaijan AZE Europe & Central Asia UMIC Bahamas, The BHS Latin America & Caribbean HIC Bahrain BHR Middle East & North Africa HIC Bangladesh BGD South Asia LMIC Barbados BRB Latin America & Caribbean HIC Belarus BLR Europe & Central Asia UMIC Belgium BEL Europe & Central Asia HIC Belize BLZ Latin America & Caribbean UMIC Benin BEN Sub-Saharan Africa LIC Bermuda BMU North America HIC Bhutan BTN South Asia LMIC Bolivia BOL Latin America & Caribbean LMIC Bosnia and Herzegovina BIH Europe & Central Asia UMIC Botswana BWA Sub-Saharan Africa UMIC Brazil BRA Latin America & Caribbean UMIC British Virgin Islands VGB Latin America & Caribbean HIC Brunei Darussalam BRN East Asia & Pacific HIC Bulgaria BGR Europe & Central Asia UMIC Burkina Faso BFA Sub-Saharan Africa LIC Burundi BDI Sub-Saharan Africa LIC Cabo Verde CPV Sub-Saharan Africa LMIC Cambodia KHM East Asia & Pacific LMIC Cameroon CMR Sub-Saharan Africa LMIC Canada CAN North America HIC Cayman Islands CYM Latin America & Caribbean HIC Central African Republic CAF Sub-Saharan Africa LIC Chad TCD Sub-Saharan Africa LIC Channel Islands CHI Europe & Central Asia HIC Chile CHL Latin America & Caribbean HIC China CHN East Asia & Pacific UMIC Colombia COL Latin America & Caribbean UMIC Comoros COM Sub-Saharan Africa LMIC Congo, Dem. Rep. COD Sub-Saharan Africa LIC Congo, Rep. COG Sub-Saharan Africa LMIC Cook Islands East Asia & Pacific Costa Rica CRI Latin America & Caribbean UMIC Cote d'Ivoire CIV Sub-Saharan Africa LMIC Croatia HRV Europe & Central Asia HIC Cuba CUB Latin America & Caribbean UMIC Curacao CUW Latin America & Caribbean HIC Cyprus CYP Europe & Central Asia HIC Czech Republic CZE Europe & Central Asia HIC Denmark DNK Europe & Central Asia HIC Djibouti DJI Middle East & North Africa LMIC 25 Dominica DMA Latin America & Caribbean UMIC Dominican Republic DOM Latin America & Caribbean UMIC Ecuador ECU Latin America & Caribbean UMIC Egypt, Arab Rep. EGY Middle East & North Africa LMIC El Salvador SLV Latin America & Caribbean LMIC Equatorial Guinea GNQ Sub-Saharan Africa UMIC Eritrea ERI Sub-Saharan Africa LIC Estonia EST Europe & Central Asia HIC Eswatini SWZ Sub-Saharan Africa LMIC Ethiopia ETH Sub-Saharan Africa LIC Faroe Islands FRO Europe & Central Asia HIC Fiji FJI East Asia & Pacific UMIC Finland FIN Europe & Central Asia HIC France FRA Europe & Central Asia HIC French Guiana Latin America & Caribbean French Polynesia PYF East Asia & Pacific HIC Gabon GAB Sub-Saharan Africa UMIC Gambia, The GMB Sub-Saharan Africa LIC Georgia GEO Europe & Central Asia UMIC Germany DEU Europe & Central Asia HIC Ghana GHA Sub-Saharan Africa LMIC Gibraltar GIB Europe & Central Asia HIC Greece GRC Europe & Central Asia HIC Greenland GRL Europe & Central Asia HIC Grenada GRD Latin America & Caribbean UMIC Guadeloupe Latin America & Caribbean Guam GUM East Asia & Pacific HIC Guatemala GTM Latin America & Caribbean UMIC Guernsey Europe & Central Asia Guinea GIN Sub-Saharan Africa LIC Guinea-Bissau GNB Sub-Saharan Africa LIC Guyana GUY Latin America & Caribbean UMIC Haiti HTI Latin America & Caribbean LIC Honduras HND Latin America & Caribbean LMIC Hong Kong SAR, China HKG East Asia & Pacific HIC Hungary HUN Europe & Central Asia HIC Iceland ISL Europe & Central Asia HIC India IND South Asia LMIC Indonesia IDN East Asia & Pacific LMIC Iran, Islamic Rep. IRN Middle East & North Africa UMIC Iraq IRQ Middle East & North Africa UMIC Ireland IRL Europe & Central Asia HIC Isle of Man IMN Europe & Central Asia HIC Israel ISR Middle East & North Africa HIC Italy ITA Europe & Central Asia HIC Jamaica JAM Latin America & Caribbean UMIC Japan JPN East Asia & Pacific HIC Jersey, Channel Islands Europe & Central Asia Jordan JOR Middle East & North Africa UMIC Kazakhstan KAZ Europe & Central Asia UMIC Kenya KEN Sub-Saharan Africa LMIC Kiribati KIR East Asia & Pacific LMIC Korea, Dem. People’s Rep. PRK East Asia & Pacific LIC Korea, Rep. KOR East Asia & Pacific HIC Kosovo XKX Europe & Central Asia UMIC Kuwait KWT Middle East & North Africa HIC Kyrgyz Republic KGZ Europe & Central Asia LMIC 26 Lao PDR LAO East Asia & Pacific LMIC Latvia LVA Europe & Central Asia HIC Lebanon LBN Middle East & North Africa UMIC Lesotho LSO Sub-Saharan Africa LMIC Liberia LBR Sub-Saharan Africa LIC Libya LBY Middle East & North Africa UMIC Liechtenstein LIE Europe & Central Asia HIC Lithuania LTU Europe & Central Asia HIC Luxembourg LUX Europe & Central Asia HIC Macao SAR, China MAC East Asia & Pacific HIC Madagascar MDG Sub-Saharan Africa LIC Malawi MWI Sub-Saharan Africa LIC Malaysia MYS East Asia & Pacific UMIC Maldives MDV South Asia UMIC Mali MLI Sub-Saharan Africa LIC Malta MLT Middle East & North Africa HIC Marshall Islands MHL East Asia & Pacific UMIC Martinique Latin America & Caribbean Mauritania MRT Sub-Saharan Africa LMIC Mauritius MUS Sub-Saharan Africa UMIC Mayotte Sub-Saharan Africa Mexico MEX Latin America & Caribbean UMIC Micronesia, Fed. Sts. FSM East Asia & Pacific LMIC Moldova MDA Europe & Central Asia LMIC Monaco MCO Europe & Central Asia HIC Mongolia MNG East Asia & Pacific LMIC Montenegro MNE Europe & Central Asia UMIC Montserrat Latin America & Caribbean Morocco MAR Middle East & North Africa LMIC Mozambique MOZ Sub-Saharan Africa LIC Myanmar MMR East Asia & Pacific LMIC Namibia NAM Sub-Saharan Africa UMIC Nauru NRU East Asia & Pacific UMIC Nepal NPL South Asia LIC Netherlands NLD Europe & Central Asia HIC New Caledonia NCL East Asia & Pacific HIC New Zealand NZL East Asia & Pacific HIC Nicaragua NIC Latin America & Caribbean LMIC Niger NER Sub-Saharan Africa LIC Nigeria NGA Sub-Saharan Africa LMIC Niue Europe & Central Asia North Macedonia MKD Europe & Central Asia UMIC Northern Mariana Islands MNP East Asia & Pacific HIC Norway NOR Europe & Central Asia HIC Oman OMN Middle East & North Africa HIC Pakistan PAK South Asia LMIC Palau PLW East Asia & Pacific HIC Panama PAN Latin America & Caribbean HIC Papua New Guinea PNG East Asia & Pacific LMIC Paraguay PRY Latin America & Caribbean UMIC Peru PER Latin America & Caribbean UMIC Philippines PHL East Asia & Pacific LMIC Poland POL Europe & Central Asia HIC Portugal PRT Europe & Central Asia HIC Puerto Rico PRI Latin America & Caribbean HIC Qatar QAT Middle East & North Africa HIC Reunion Sub-Saharan Africa Romania ROU Europe & Central Asia UMIC 27 Russian Federation RUS Europe & Central Asia UMIC Rwanda RWA Sub-Saharan Africa LIC Samoa WSM East Asia & Pacific UMIC Sao Tome and Principe STP Sub-Saharan Africa LMIC Saudi Arabia SAU Middle East & North Africa HIC Senegal SEN Sub-Saharan Africa LMIC Serbia SRB Europe & Central Asia UMIC Seychelles SYC Sub-Saharan Africa HIC Sierra Leone SLE Sub-Saharan Africa LIC Singapore SGP East Asia & Pacific HIC Sint Maarten (Dutch part) SXM Latin America & Caribbean HIC Slovak Republic SVK Europe & Central Asia HIC Slovenia SVN Europe & Central Asia HIC Solomon Islands SLB East Asia & Pacific LMIC Somalia SOM Sub-Saharan Africa LIC South Africa ZAF Sub-Saharan Africa UMIC South Sudan SSD Sub-Saharan Africa LIC Spain ESP Europe & Central Asia HIC Sri Lanka LKA South Asia UMIC St. Kitts and Nevis KNA Latin America & Caribbean HIC St. Lucia LCA Latin America & Caribbean UMIC St. Martin (French part) MAF Latin America & Caribbean HIC St. Vincent and the Grenadines VCT Latin America & Caribbean UMIC Sudan SDN Sub-Saharan Africa LMIC Suriname SUR Latin America & Caribbean UMIC Sweden SWE Europe & Central Asia HIC Switzerland CHE Europe & Central Asia HIC Syrian Arab Republic SYR Middle East & North Africa LIC Taiwan, China TWN East Asia & Pacific HIC Tajikistan TJK Europe & Central Asia LIC Tanzania TZA Sub-Saharan Africa LIC Thailand THA East Asia & Pacific UMIC Timor-Leste TLS East Asia & Pacific LMIC Togo TGO Sub-Saharan Africa LIC Tonga TON East Asia & Pacific UMIC Trinidad and Tobago TTO Latin America & Caribbean HIC Tunisia TUN Middle East & North Africa LMIC Turkey TUR Europe & Central Asia UMIC Turkmenistan TKM Europe & Central Asia UMIC Turks and Caicos Islands TCA Latin America & Caribbean HIC Tuvalu TUV East Asia & Pacific UMIC Uganda UGA Sub-Saharan Africa LIC Ukraine UKR Europe & Central Asia LMIC United Arab Emirates ARE Middle East & North Africa HIC United Kingdom GBR Europe & Central Asia HIC United States USA North America HIC Uruguay URY Latin America & Caribbean HIC Uzbekistan UZB Europe & Central Asia LMIC Vanuatu VUT East Asia & Pacific LMIC Venezuela, RB VEN Latin America & Caribbean UMIC Vietnam VNM East Asia & Pacific LMIC Virgin Islands (U.S.) VIR Latin America & Caribbean HIC Wallis and Futuna East Asia & Pacific West Bank and Gaza PSE Middle East & North Africa LMIC Yemen, Rep. YEM Middle East & North Africa LIC Zambia ZMB Sub-Saharan Africa LMIC Zimbabwe ZWE Sub-Saharan Africa LMIC Note: Income group and income code are presented as in the World Bank classification. 28 Table A4. Movements toward more sophisticated power markets and sector outcomes – robustness check (1) Electricity production Access to electricity, % of Consumer affordability Dependent variables from renewable sources, total population index excl. hydroelectric, % (1) (2) (3) Introduction of more sophisticated 4.887*** 3.125* 1.666*** power markets (θ) (0.392) (1.862) (0.174) GDP per capita 0.0291 0.174 0.214*** (0.0336) (0.182) (0.0161) System size 4.149 1.663 3.448** (2.651) (9.685) (1.502) Polity2 0.783*** 0.709*** -0.0148 (0.0581) (0.206) (0.0265) Constant 68.52*** 76.31*** -1.856*** (0.551) (3.499) (0.272) Oil price shock dummy yes yes yes Observations 3,351 920 2,992 Number of countries 153 117 124 R-squared 0.124 0.024 0.140 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. Oil price shock dummy was dropped from SAIDI and SAIFI regressions because of missing observations corresponding to the years of oil price shock. 29 Table A5. Movements toward more sophisticated power markets and sector outcomes – robustness check (2) System Average System Average Electricity production from Access to electricity, % of total Consumer affordability Dependent variables Interruption Interruption renewable sources, excl. population index Duration index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) Introduction of more sophisticated power markets (θ) 4.739*** 4.738*** 3.437* 3.406* -502.0*** -13.84** 1.895*** 1.936*** (0.353) (0.354) (1.882) (1.881) (49.98) (5.610) (0.194) (0.194) GDP per capita -0.0233 -0.0234 0.0758 0.0768 -10.69* -0.473 0.257*** 0.262*** (0.0314) (0.0315) (0.194) (0.194) (6.121) (0.689) (0.0186) (0.0187) System size 4.852** 4.853** 1.487 1.607 104.3 2.751 2.683* 2.622* (2.240) (2.241) (9.802) (9.795) (330.3) (37.11) (1.590) (1.588) Democratic accountability 1.091*** 1.090*** 2.130** 2.106** -20.07 -1.897 0.0162 0.0245 (0.169) (0.169) (0.898) (0.898) (51.36) (6.918) (0.0842) (0.0841) Population density 0.00327*** 0.00327*** 0.0134 0.0137 1.439*** 0.0345 -0.00347*** -0.00353*** (0.00109) (0.00109) (0.0107) (0.0107) (0.432) (0.0485) (0.000648) (0.000648) Constant 71.20*** 71.20*** 71.59*** 71.75*** 389.1 40.23 -2.504*** -2.489*** (0.835) (0.835) (5.543) (5.539) (302.4) (38.61) (0.442) (0.441) Oil price shock dummy no yes no yes no no no yes Observations 2,874 2,874 795 795 275 273 2,737 2,737 Number of countries 127 127 100 100 95 94 114 114 R-squared 0.106 0.106 0.018 0.021 0.366 0.035 0.155 0.158 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. 30 Table A6. Movements toward more sophisticated power markets and sector outcomes – one-year lag: robustness check (3) System Average System Average Electricity production from Access to electricity, % of Consumer affordability Dependent variables Interruption Duration Interruption Frequency renewable sources, excl. total population index index index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Introduction of more sophisticated 5.527*** 4.897*** 3.898** 3.356* -226.7*** -226.8*** -6.375 -6.365 1.821*** 1.739*** power markets (θ)t-1 (0.363) (0.382) (1.851) (1.927) (63.93) (69.17) (13.62) (14.71) (0.165) (0.170) GDP per capita 0.0263 0.0169 0.161 0.168 -2.143 -2.480 -0.353 -0.303 0.209*** 0.202*** (0.0261) (0.0337) (0.184) (0.182) (7.813) (9.298) (1.664) (1.977) (0.0154) (0.0160) System size 3.473 3.965 1.171 1.678 7.362 12.45 1.084 3.009 3.020** 3.238** (2.660) (2.648) (9.853) (9.688) (452.5) (490.6) (96.45) (104.4) (1.500) (1.501) Polity2 0.779*** 0.702*** 2.050 1.903 -0.0145 (0.0580) (0.207) (11.97) (2.545) (0.0264) Constant 74.25*** 68.70*** 79.05*** 76.19*** 269.0 290.6 34.04 25.77 -2.079*** -1.771*** (0.447) (0.549) (3.385) (3.492) (189.8) (239.4) (40.53) (51.08) (0.262) (0.273) Observations 3,948 3,351 946 920 349 301 347 299 3,144 2,992 Number of countries 180 153 119 117 121 105 120 104 129 124 R-squared 0.071 0.126 0.007 0.022 0.053 0.053 0.001 0.004 0.152 0.143 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. 31 Table A7. Movements toward more sophisticated power markets and sector outcomes – other lag specifications: robustness check (4) Electricity production from Access to electricity, % of Consumer affordability renewable sources, excl. total population index hydroelectric, % Lag: (1) (2) (3) (4) (5) (6) Introduction of more sophisticated power markets (θ) t-2 4.955*** 5.698*** 1.839*** (0.376) (2.072) (0.167) Introduction of more sophisticated power markets (θ) t-3 5.088*** 6.691*** 1.984*** (0.370) (2.236) (0.164) Observations 3,351 3,351 920 920 2,992 2,992 Number of countries 153 153 117 117 124 124 R-squared 0.128 0.132 0.027 0.029 0.148 0.155 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita, system size, Polity 2 indicator and country fixed effects are included in all specifications. 32 Table A8. Movements toward more sophisticated power markets and sector outcomes, low- and middle-income countries – estimation Results System Average System Average Electricity production from Access to electricity, % of total Dependent variables Consumer affordability index Interruption Duration Interruption Frequency renewable sources, excl. population index index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (13) (14) Introduction of more 7.533*** 6.824*** 6.892*** 4.718* 3.825 3.766 -1,316*** -1,316*** -36.04 -36.09 0.203 0.139 0.140 sophisticated power markets (θ) (0.589) (0.591) (0.593) (2.477) (2.572) (2.569) (95.90) (102.9) (31.16) (33.40) (0.145) (0.150) (0.150) GDP per capita 1.220*** 1.101*** 1.097*** 1.121** 1.128** 1.185** -4.037 -3.817 -1.665 -1.226 0.173*** 0.177*** 0.177*** (0.107) (0.109) (0.109) (0.557) (0.557) (0.558) (16.90) (19.40) (5.491) (6.300) (0.0271) (0.0278) (0.0279) System size -7.805** -6.357** -6.430** -6.080 -5.527 -5.806 18.71 19.23 10.11 9.058 2.304** 2.305** 2.303** (3.263) (3.176) (3.176) (12.32) (12.20) (12.19) (408.9) (440.8) (132.9) (143.2) (1.029) (1.041) (1.042) Polity2 0.731*** 0.731*** 0.749*** 0.747*** 2.021 1.887 0.0491*** 0.0491*** (0.0689) (0.0689) (0.250) (0.250) (10.64) (3.455) (0.0183) (0.0183) Constant 54.14*** 51.28*** 51.45*** 67.49*** 65.02*** 64.86*** 1,043*** 1,122*** 76.61 70.42 0.524*** 0.462** 0.463** (0.654) (0.665) (0.676) (4.142) (4.246) (4.243) (177.7) (220.6) (57.64) (71.57) (0.179) (0.185) (0.186) Oil price shock dummy no no yes no no yes no no no no no no yes Observations 2,567 2,304 2,304 674 648 648 203 179 201 177 2,018 1,970 1,970 Number of countries 124 112 112 85 83 83 72 64 71 63 84 83 83 R-squared 0.158 0.204 0.205 0.014 0.029 0.032 0.596 0.596 0.011 0.014 0.039 0.044 0.044 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. Oil price shock dummy was dropped from SAIDI and SAIFI regressions because of missing observations corresponding to the years of oil price shock. 33 Table A9. Power market sophistication and sector outcomes (step 1) – estimation results System Average System Average Electricity production from Access to electricity, % of total Dependent variables Consumer affordability index Interruption Duration Interruption renewable sources, excl. population index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) From Vertically Integrated Utility 8.636*** 7.529*** 7.594*** 4.128* 3.188 3.129 -447.9*** -451.1*** -12.88 -13.05 0.507*** 0.486*** 0.486*** models to Single Buyer models (0.480) (0.506) (0.506) (2.261) (2.316) (2.315) (76.16) (86.00) (17.69) (19.92) (0.107) (0.109) (0.110) GDP per capita 0.149*** 0.268*** 0.274*** 0.389 0.329 0.352 -13.66 -18.98 -1.101 -1.380 0.0656*** 0.0550*** 0.0551*** (0.0326) (0.0497) (0.0497) (0.348) (0.343) (0.343) (14.67) (19.50) (3.409) (4.517) (0.0105) (0.0110) (0.0111) System size -2.600 -2.938 -3.116 -0.311 0.392 0.340 134.8 169.7 8.108 10.26 2.227*** 2.334*** 2.333*** (2.891) (2.940) (2.938) (12.01) (11.82) (11.82) (551.8) (625.1) (128.3) (144.8) (0.833) (0.826) (0.827) Polity2 0.666*** 0.667*** 0.951*** 0.941*** 5.746 5.663 0.0373** 0.0373** (0.0650) (0.0649) (0.262) (0.262) (23.95) (5.546) (0.0146) (0.0146) Constant 67.75*** 60.70*** 60.85*** 71.86*** 70.82*** 70.76*** 574.5** 695.4** 59.01 56.49 0.0901 0.209 0.210 (0.512) (0.651) (0.654) (4.108) (4.142) (4.141) (264.6) (335.3) (61.66) (77.91) (0.160) (0.159) (0.159) Oil price shock dummy no no yes no no yes no no no no no no yes Observations 3,235 2,653 2,653 675 649 649 215 173 213 171 2,514 2,373 2,373 Number of countries 179 152 152 89 87 87 77 63 76 62 129 124 124 R-squared 0.107 0.155 0.157 0.008 0.031 0.034 0.204 0.207 0.004 0.014 0.037 0.037 0.037 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. Oil price shock dummy was dropped from SAIDI and SAIFI regressions because of missing observations corresponding to the years of oil price shock. 34 Table A10. Power market sophistication and sector outcomes (step 2) – estimation results System Average System Average Electricity production from Access to electricity, % of total Dependent variables Consumer affordability index Interruption Duration Interruption renewable sources, excl. population index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) From Single Buyer models to 3.581*** 3.554*** 3.660*** -1.022 -0.703 -0.820 0.783 -0.0856 0.778 -0.0856 0.282 0.264 0.324 Wholesale Competition models (0.905) (0.909) (0.911) (2.990) (3.055) (3.051) (18.27) (18.96) (10.82) (18.96) (0.304) (0.309) (0.311) GDP per capita 0.372*** 0.344*** 8.297* 0.474 0.496 -0.840 -1.633 -0.990 -0.834 -0.990 0.0765*** 0.0760*** 5.527*** (0.0786) (0.0780) (4.230) (0.405) (0.409) (11.60) (2.698) (2.911) (1.597) (2.911) (0.0242) (0.0243) (1.796) System size 8.926** 8.865** 0.344*** -0.806 -0.833 0.515 -1.778 -3.234 3.862 -3.234 5.771*** 5.760*** 0.0783*** (4.269) (4.213) (0.0780) (11.54) (11.62) (0.409) (75.68) (78.54) (44.82) (78.54) (1.783) (1.792) (0.0243) Polity2 0.582*** 0.578*** 0.204 0.190 2.970 2.970 0.0327 0.0317 (0.0932) (0.0932) (0.329) (0.328) (1.918) (1.918) (0.0320) (0.0320) Constant 73.36*** 70.96*** 71.23*** 79.88*** 78.68*** 78.74*** 53.79 33.41 24.52 33.41 1.179*** 1.068*** 1.144*** (0.994) (1.048) (1.065) (4.900) (5.237) (5.229) (42.43) (48.86) (25.22) (48.86) (0.328) (0.355) (0.358) Oil price shock dummy no no yes no no yes no no no no no no yes Observations 1,274 1,212 1,212 449 443 443 155 143 153 143 1,078 1,067 1,067 Number of countries 94 90 90 63 62 62 56 52 55 52 75 74 74 R-squared 0.048 0.081 0.083 0.004 0.005 0.011 0.004 0.031 0.003 0.031 0.026 0.027 0.030 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. Oil price shock dummy was dropped from SAIDI and SAIFI regressions because of missing observations corresponding to the years of oil price shock. 35 Table A11. Power market sophistication and sector outcomes (step 3) – estimation results System Average System Average Electricity production from Access to electricity, % of total Dependent variables Consumer affordability index Interruption Duration Interruption renewable sources, excl. population index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) From Wholesale Competition models to Retail Competition 1.443** 1.403** 1.443** 0.240 0.649 0.647 3.349 3.194 0.867 0.821 1.593** 2.080** 2.139*** models (0.608) (0.614) (0.614) (1.853) (1.107) (1.110) (2.983) (2.787) (1.287) (1.262) (0.808) (0.814) (0.814) GDP per capita -0.0125 -0.00283 -0.00473 0.111 -0.179 -0.181 0.278 0.451 0.104 0.154 0.195*** 0.0749 0.0845 (0.0513) (0.0545) (0.0545) (0.191) (0.115) (0.116) (0.471) (0.445) (0.203) (0.201) (0.0730) (0.0824) (0.0826) System size 85.78*** 85.37*** 83.15*** -1.566 0.0333 0.0706 -220.0*** -217.7*** -55.64** -54.95** 38.50** 42.55*** 39.61** (9.447) (9.500) (9.620) (18.30) (10.93) (10.97) (58.98) (55.12) (25.44) (24.97) (15.58) (15.44) (15.56) Polity2 -0.244 -0.221 -7.221*** -7.231*** 2.145*** 0.636* 2.326*** 2.265*** (0.459) (0.458) (0.444) (0.447) (0.709) (0.321) (0.772) (0.772) Constant 89.80*** 91.47*** 91.62*** 95.64*** 162.2*** 162.3*** 12.26 -9.246 4.271 -2.058 -0.372 -16.23*** -15.57*** (1.206) (3.449) (3.447) (5.190) (5.134) (5.167) (12.50) (13.84) (5.393) (6.268) (1.710) (5.524) (5.536) Oil price shock dummy no no yes no no yes no no no no no no yes Observations 406 404 404 175 175 175 82 80 82 80 348 348 348 Number of countries 33 32 32 26 26 26 29 28 29 28 28 28 28 R-squared 0.219 0.219 0.224 0.003 0.647 0.647 0.220 0.345 0.087 0.157 0.100 0.125 0.131 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. Oil price shock dummy was dropped from SAIDI and SAIFI regressions because of missing observations corresponding to the years of oil price shock. 36 Table A12. Power market sophistication and sector outcomes (step 1) – robustness check (1) System Average System Average Electricity production from Access to electricity, % of total Consumer affordability Dependent variables Interruption Interruption renewable sources, excl. population index Duration index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) From Vertically Integrated Utility models to Single Buyer models 7.847*** 7.853*** 2.427 2.300 -533.8*** -14.74* 0.577*** 0.578*** (0.472) (0.473) (2.451) (2.450) (70.15) (8.027) (0.129) (0.129) GDP per capita 0.158*** 0.159*** 0.357 0.386 -22.63* -1.005 0.0675*** 0.0676*** (0.0424) (0.0425) (0.368) (0.368) (13.34) (1.529) (0.0123) (0.0123) System size -2.426 -2.442 -1.034 -1.100 189.7 7.578 2.158** 2.156** (2.453) (2.455) (12.25) (12.24) (463.9) (53.09) (0.894) (0.894) Democratic accountability 0.473** 0.474** 3.064*** 3.035*** -26.05 -0.401 0.130*** 0.130*** (0.199) (0.199) (1.154) (1.153) (98.27) (17.80) (0.0502) (0.0502) Population density 0.0105*** 0.0105*** 0.0688** 0.0703*** 2.068*** 0.0493 -0.000253 -0.000254 (0.00241) (0.00242) (0.0270) (0.0270) (0.701) (0.0802) (0.000648) (0.000648) Constant 63.31*** 63.32*** 54.94*** 54.72*** 375.2 48.84 -0.518* -0.516* (1.002) (1.004) (7.523) (7.516) (459.5) (72.04) (0.268) (0.268) Oil price shock dummy no yes no yes no no no yes Observations 2,184 2,184 529 529 149 147 2,122 2,122 Number of countries 126 126 70 70 53 52 114 114 R-squared 0.165 0.165 0.037 0.041 0.390 0.038 0.043 0.043 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. 37 Table A13. Power market sophistication and sector outcomes (step 2) – robustness check (1) System Average System Average Electricity production from Access to electricity, % of total Consumer affordability Dependent variables Interruption Interruption renewable sources, excl. population index Duration index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) From Single Buyer models to Wholesale Competition models 1.550** 1.599** -0.622 -0.758 -1.126 -0.334 0.326 0.376 (0.681) (0.684) (2.995) (2.989) (15.86) (7.285) (0.347) (0.349) GDP per capita 0.338*** 0.338*** 0.434 0.454 -0.765 -0.212 0.0769*** 0.0793*** (0.0555) (0.0555) (0.401) (0.401) (2.487) (1.144) (0.0255) (0.0255) System size 3.893 3.659 -6.084 -6.173 1.950 3.637 5.711*** 5.463*** (2.985) (3.002) (11.42) (11.39) (66.24) (30.42) (1.882) (1.889) Democratic accountability 0.603** 0.617** 1.898 1.837 -6.472 1.919 -0.332*** -0.318*** (0.240) (0.241) (1.252) (1.249) (7.338) (3.695) (0.111) (0.111) Population density 0.120*** 0.119*** 0.303*** 0.308*** -0.760 -0.280 0.00572* 0.00591* (0.00634) (0.00635) (0.0596) (0.0595) (0.652) (0.300) (0.00327) (0.00327) Constant 51.96*** 52.03*** 31.65*** 31.16*** 168.5* 44.01 1.698** 1.683** (1.587) (1.590) (10.58) (10.56) (99.28) (46.82) (0.780) (0.780) Oil price shock dummy no yes no yes no no no yes Observations 1,056 1,056 396 396 124 122 962 962 Number of countries 77 77 56 56 45 44 64 64 R-squared 0.326 0.326 0.090 0.097 0.031 0.016 0.040 0.042 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. 38 Table A14. Power market sophistication and sector outcomes (step 3) – robustness check (1) System Average System Average Electricity production from Access to electricity, % of total Consumer affordability Dependent variables Interruption Interruption renewable sources, excl. population index Duration index Frequency index hydroelectric, % (1) (2) (3) (4) (5) (6) (7) (8) From Wholesale Competition models to Retail Competition 1.732*** 1.721*** 0.465 0.466 3.172 0.824 1.111 1.099 models (0.659) (0.659) (1.593) (1.598) (2.889) (1.056) (0.820) (0.820) GDP per capita -0.0493 -0.0441 -0.0552 -0.0543 0.438 0.0899 0.421*** 0.438*** (0.0795) (0.0799) (0.240) (0.241) (0.572) (0.209) (0.102) (0.104) System size 87.35*** 85.93*** -1.923 -1.940 -216.6*** -52.75** 36.24** 33.86** (9.598) (9.796) (15.72) (15.78) (57.12) (20.88) (14.84) (15.05) Democratic accountability -1.202*** -1.139*** -14.17*** -14.16*** 2.821 1.179 -3.137*** -3.039*** (0.398) (0.407) (1.900) (1.915) (3.476) (1.271) (0.500) (0.510) Population density 0.000775 0.000638 0.00142 0.00142 -0.0169 -0.00263 -0.00634*** -0.00666*** (0.00159) (0.00160) (0.00634) (0.00637) (0.0307) (0.0112) (0.00203) (0.00206) Constant 95.87*** 95.68*** 170.6*** 170.5*** 1.758 -0.0312 13.14*** 12.70*** (2.460) (2.476) (11.17) (11.26) (24.49) (8.952) (3.084) (3.120) Oil price shock dummy no yes no yes no no no yes Observations 392 392 175 175 77 77 339 339 Number of countries 30 30 26 26 27 27 27 27 R-squared 0.239 0.240 0.282 0.282 0.260 0.148 0.225 0.227 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita is in thousands of dollars and system size is in gigawatt. Country fixed effects are included in all specifications. 39 Table A15. Power market sophistication and sector outcomes (step 1) – different lag specifications: robustness check (2) Dependent variables Electricity production from Access to electricity, % of total Consumer affordability index SAIDI SAIFI renewable sources, excl. population hydroelectric, % Lag: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Market sophistication (θ) t-1 7.623*** 3.430 -227.8** -6.514 0.508*** (0.499) (2.418) (93.09) (19.77) (0.108) Market sophistication (θ) t-2 7.709*** 5.697** 0.525*** (0.495) (2.526) (0.107) Market sophistication (θ) t-3 7.886*** 6.329** 0.554*** (0.493) (2.623) (0.107) Observations 2,653 2,653 2,653 649 649 649 173 171 2,373 2,373 2,373 Number of countries 152 152 152 87 87 87 63 62 124 124 124 R-squared 0.159 0.162 0.166 0.032 0.037 0.038 0.054 0.011 0.038 0.039 0.040 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita, system size, Polity 2 indicator and country fixed effects are included in all specifications. SAIDI is Sytem Average Interruption Duration Index and SAIFI stands for System Average Interruption Frequency. Table A16. Power market sophistication and sector outcomes (step 2) – different lag specifications: robustness check (2) Dependent variables Electricity production from Access to electricity, % of total Consumer affordability index SAIDI SAIFI renewable sources, excl. population hydroelectric, % Lag: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Market sophistication (θ) t-1 3.872*** -0.616 12.59 9.276 0.150 (0.907) (3.052) (11.89) (6.857) (0.310) Market sophistication (θ) t-2 4.509*** -0.712 0.508 (0.951) (3.342) (0.321) Market sophistication (θ) t-3 5.282*** -0.727 0.698** (1.001) (3.880) (0.339) Observations 1,274 1,212 1,212 455 443 443 147 145 1,141 1,067 1,067 Number of countries 94 90 90 64 62 62 53 52 80 74 74 R-squared 0.054 0.087 0.091 0.006 0.005 0.005 0.043 0.074 0.027 0.029 0.031 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita, system size, Polity 2 indicator and country fixed effects are included in all specifications. SAIDI is Sytem Average Interruption Duration Index and SAIFI stands for System Average Interruption Frequency. 40 Table A17. Power market sophistication and sector outcomes (step 3) – different lag specifications: robustness check (2) Dependent variables Electricity production from Access to electricity, % of total Consumer affordability index SAIDI SAIFI renewable sources, excl. population hydroelectric, % Lag: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) Market sophistication (θ) t-1 1.119* 0.689 1.750 0.515 2.499*** (0.607) (1.047) (2.743) (1.234) (0.800) Market sophistication (θ) t-2 0.799 0.709 3.014*** (0.612) (0.982) (0.805) Market sophistication (θ) t-3 0.447 0.749 3.428*** (0.617) (0.919) (0.821) Observations 404 404 404 175 175 175 80 80 348 348 348 Number of countries 32 32 32 26 26 26 28 28 28 28 28 R-squared 0.216 0.212 0.209 0.647 0.647 0.648 0.333 0.152 0.134 0.145 0.154 Note: Standard errors in parentheses. Significance level: *** 1 percent, ** 5 percent, * 10 percent. GDP per capita, system size, Polity 2 indicator and country fixed effects are included in all specifications. SAIDI is Sytem Average Interruption Duration Index and SAIFI stands for System Average Interruption Frequency. 41 References Abdisa, L. M. 2018. “Power Outages, its Economic Cost and Firm Performance: Evidence from Ethiopia.” Utility Policies, 53: 111–120. Akpandjar, George, and Carl Kitchens. 2017. “From Darkness to Light: The Effect of Electrification in Ghana, 2000–2010.” Economic Development and Cultural Change 66 (1): 31–54. Bacon, R. W. 1995. “Privatization and Reform in the Global Electricity Supply Industry.” Annual Review of Energy and the Environment. 20 (1): 119–43. Bacon, R. W. 2018. “Taking Stock of the Impact of Power Utility Reform in Developing Countries.” Policy Research Working Paper 8460, World Bank, Washington, DC. Bacon, R. W., and J. Besant-Jones. 2001. “Global Electric Power Reform, Privatization and Liberalisation of the Electric Power Industry in Developing Countries.” Annual Review of Energy and the Environment 26 (November): 331–59. Balza L., R.A. Jimenez, and J. E. Mercado Díaz. 2013. “Privatization, Institutional Reform, and Performance in the Latin American Electricity Sector.” Inter-American Development Bank, Energy Division INE/ENE Technical Note No. IDB-TN-599. Bensch, G. 2019. “The effects of market-based reforms on access to electricity in developing countries: a systematic review.” Journal of Development Effectiveness, 11 (2): 165–188. Besant-Jones, J. 2006. “Reforming Power Markets in Developing Countries: What Have We Learned?” World Bank Discussion Paper 19, World Bank, Washington, DC. Bhattacharyya, S. C. 2011. Energy Economics: Concepts, Issues, Markets, and Governance. London: Springer. Bridge, B. A., Adhikari, D., and M. Fontenla. 2016. “Household-level effects of electricity on income.” Energy Economics, 58 (C): 222–228. Burke, P. J., Stern, D. I., and S. B. Bruns. 2018. “The Impact of Electricity on Economic Development: A Macroeconomic Perspective.” International Review of Environmental and Resource Economics, now publishers, 12 (1): 85–127. Cole. M., Elliott, R. J. R., , Occhiali, G., and E. Strobl. 2018. “Power outages and firm performance in Sub-Saharan Africa.” Journal of Development Economics, Vol. 134: 150 – 159. 42 de Chaisemartin, C., and X. D’Haultfœuille. 2020. “Two-Way Fixed Effects Estimators with Heterogeneous Treatment Effects.” American Economic Review, 110 (9): 2964–2996. Dertinger, A., and L. Hirth. 2020. “Reforming the electric power industry in developing economies evidence on efficiency and electricity access outcomes.” Energy Policy, Vol. 139 (C). Erdogdu, E. 2011. “What Happened to Efficiency in Electricity Industries after Reforms?” Energy Policy, 39 (10): 6551 6560. Foster, V., and A. Rana. 2020. “Rethinking Power Sector Reform in the Developing World” Sustainable Infrastructure. Washington, DC: World Bank. https://openknowledge.worldbank.org/handle/10986/32335. Gratwick, K. N., and A. Eberhard. 2008. “Demise of the standard model for power sector reform and the emergence of hybrid power markets.” Energy Policy, 36 (10): 3948–3960. Hunt, S. 2002. Making Competition Work in Electricity. New York: John Wiley and Sons, Inc. Jamasb, T., Nepal, R., and G.R. Timilsina. 2015. “A Quarter Century Effort Yet to Come of Age: A Survey of Power Sector Reform in Developing Countries.” Policy Research Working paper 7330, June, Development Research Group, The World Bank Group, Washington, D.C. Joskow, P. L. 2008. “Lessons Learned from the Electricity Market Liberalization.” Energy Journal 29 (special issue): 9–42. Kessides, I. 2012. “Electricity Reforms—What Some Countries Did Right and Others Can Do Better.” Viewpoint, Note No. 332, World Bank, Washington, DC. Kessides, I. 2012. “The Impacts of Electricity Sector Reforms in Developing Countries.” The Electricity Journal, 25 (6): 79–88. Khandker, S. R., Barnes D. F., and H. A. Samad. 2012. “The Welfare Impacts of Rural Electrification in Bangladesh.” Energy Journal 33 (1): 187–206. Kline, P. and E. Moretti. 2014. “Local Economic Development, Agglomeration Economies, and the Big Push: 100 Years of Evidence from the Tennessee Valley Authority.” Quarterly Journal of Economics 129 (1): 275–331. Lee, A. D., and Z. Usman. 2018. “Taking Stock of the Political Economy of Power Sector Reforms in Developing Countries - A Literature Review” Policy Research Working paper 8518, June, Development Research Group, The World Bank Group, Washington, D.C. 43 Lee, K., Miguel E., and C. Wolfram. 2020. “Does Household Electrification Supercharge Economic Development?” Journal of Economic Perspectives, 34 (1)—Winter 2020: 122– 144 Littlechild, S. C. 2001. “Electricity: Regulatory Developments from Around the World.” Beesley Lectures on Regulation series XI, Institute of Economic Affairs/London Business School, London, October 9. Ljung, P. 2007. “Energy Sector Reform: Strategies for Growth, Equity and Sustainability”, Sidastudies no. 20, Swedish International Development Cooperation Agency, Stockholm. Lipscomb, M., Mobarak A. M., and T. Barham. 2013. “Development Effects of Electrification: Evidence from the Topographic Placement of Hydropower Plants in Brazil.” American Economic Journal: Applied Economics 5 (2): 200–231. Mensah, J. T. 2018. “Jobs! Electricity Shortages and Unemployment in Africa.” Policy Research working paper; no. WPS 8415 Washington, D.C. : World Bank Group. Nagayama, H. 2010. “Impacts on investments, and transmission/distribution loss through power sector reforms.” Energy Policy, Elsevier, 38 (7): 3453 – 3467. Nepal, R. and T. Jamasb. 2015. “Caught between theory and practice: government, market, and regulatory failure in electricity sector reforms”. Economic analysis and policy, Vol. 46, 16– 24. Perez-Sebastian, F., J. Steinbuks, J. G. Feres, and I. M. Trotter. 2020. “Electricity Access and Structural Transformation: Evidence from Brazil's Electrification.” World Bank Policy Research Working Paper No. 9182. Rud, J. P. 2012. “Electricity Provision and Industrial Development: Evidence from India.” Journal of Development Economics 97 (2): 352–67. Ruiz-Mendoza, B. J., and C. Sheinbaum-Pardo. 2010. “Electricity sector reforms in four Latin- American countries and their impact on carbon dioxide emissions and renewable energy”, Energy Policy, 38 (11): 6755–6766. Sen, A. 2014. “Divergent Paths to a Common Goal? An Overview of Challenges to Electricty Sector Reform in Developing versus Developed Countries.” University of Oxford Institute for Energy Studies Working Paper 10. 44 Sen, A., Nepal, R., and T. Jamasb. 2016. “Reforming Electricity Reforms? Empirical Evidence from Asian Economies.” University of Oxford Institute for Energy Studies Working Paper 18. Tuttle, D. P., Gülen, G., Hebner, R., King, C. W., Spence, D. B., Andrade, J., Wible, J. A., Baldwick, R., and R. Duncan. 2016. “The History and Evolution of the U.S. Electricity Industry.” White Paper UTEI/2016-05-2. Urpelainen, J., and J. Yang. 2019. “Global patterns of power sector reform, 1982–2013.” Energy Strategy Reviews, Vol 23: 152–162. Vagliasindi, M. 2012. “Power Market Structure and Performance” Policy Research Working Paper; No. 6123. World Bank, Washington, DC. https://openknowledge.worldbank.org/handle/10986/11933 Victor, D. G., and T. C. Heller. 2007. “The Political Economy of Power Sector Reform: The Experiences of Five Major Developing Countries.” New York: Cambridge University Press. Williams, James, and Rebecca Ghanadan. 2006. “Electricity Reform in Developing and Transition Countries: A Reappraisal.” Energy, 31 (6–7): 815–44. World Bank. 2017. “State of Electricity Access Report”, World Bank Group, Washington, D.C. https://openknowledge.worldbank.org/bitstream/handle/10986/26646/114841-WP-v2- FINALSEARwebopt.pdf?sequence=6&isAllowed=y Zhang, Y., D. Parker, and C. Kirkpatrick. 2008. “Electricity sector reform in developing countries: an econometric assessment of the effects of privatization, competition and regulation”, Journal of Regulatory Economics, 33 (2): 159–178. 45