- Email: [email protected]
Does Independent Regulation of Public Utilities in Developing Countries Improve Efficiency?
Does Independent Regulation of Public Utilities in Developing Countries Improve Efficiency? An evaluation suggests that introducing independent regulation to the electricity industry in developing countries is effective in stimulating performance improvements, leading to greater generation and better quality of service. The impact on energy efficiency is positive but insignificant. Babacar Sarr is a Program Officer at the International Budget Partnership, where he is responsible for conducting research on public financial management and overseeing the implementation of the Open Budget Survey in Africa. Dr. Sarr received his B.Sc. in Econometrics from Pantheon-Sorbonne University, and his M.Sc., M.Phil., and Ph.D. from CERDI-Auvergne University in France. His research focuses on measuring the impact of public sector reform in developing countries and his work has appeared in academic journals both in the areas of public sector management and public service delivery.
72
Babacar Sarr
I. Introduction Regulatory systems for infrastructure sectors are a relatively new but important phenomenon in many developing countries. It has been estimated that close to 200 new infrastructure regulators have been created around the world in the past 10 years (Brown et al., 2006). These regulatory systems are designed to respond to natural monopolies and market failures associated with network industries such as
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
electricity, gas, water, telecommunications, and transportation. The aim of regulation is to encourage efficient, low-cost, and reliable service provision while ensuring financial viability and new investment. The expectation is that regulation mechanisms can depoliticize tariff setting, reduce costs, improve service quality in a cost-effective way, stimulate the introduction of new products and services, and stimulate efficient investment. Therefore, the establishment of an The Electricity Journal
independent regulation agency (IRA) should lead to improved economic performance. he electricity sector has three components: generation, transmission, and distribution. In most countries this sector is a strategic activity with natural monopoly characteristics resulting from the existence of economies of scale and scope. Regulation is required especially in the areas of electricity supply that remain dominated by one or a very small number of operators, to prevent monopoly abuse. In many countries, instead of direct regulation by a government department, the establishment of independent or quasiindependent regulatory agencies has been favored, drawing on the regulatory models of the United States and the United Kingdom. This form of independent regulation is expected to encourage private capital to invest in capacity in the face of a potential holdup problem under conditions of incomplete contracts (Spiller, 1996). Despite these good intentions, there is little evidence that these regulatory systems have met their expectations. Indeed, the literature on estimating the effect of regulatory governance arrangements on infrastructure outcomes is relatively small to date, particularly for the electricity sector (Jamasb et al., 2004). A particularly important study is that by Kirkpatrick et al. (2008), who explore the impacts of competition, privatization, and
T
July 2015,
Vol. 28, Issue 6
regulation on electricity sector performance1 in 36 developing and transitional countries. In their results, the regulation variable is not a significant, correctly signed explanatory variable in any of the regressions reported. However, in this paper, the regulatory variable is a dummy which is not dated; for instance, a country like Ethiopia establishing an independent regulatory body in 1999 scores 1 for the whole of the
Despite good good intentions, there is little evidence that these regulatory systems have met their expectations. estimation period, as would Costa Rica, which did so in 1928. Therefore, the experience of the regulator is not taken into account in their model. Bortolotti et al. (1998), who use data on the privatization of electricity generation in 38 countries between 1977 and 1997, find that effective regulation is a prerequisite for the success of privatization. Cubbin and Stern (2006) conclude that the quality of regulatory governance is positively and significantly associated with higher per-capita generation capacity levels and that this positive effect increases
over time with the reputation of the regulator. mpirical testing of the performance of regulation seems to concentrate on case studies and the application of panel-data econometrics (Guasch et al., 2007). These methods cannot allow a statistically robust separation of the effects of other changes like privatization from the impact of the establishment of an independent regulatory agency. Therefore, we propose to interpret the introduction of an independent regulator as a natural experiment,2 in order to re-establish the conditions of a randomized experiment and represent the IRA as a treatment. This leads us to perform propensity score matching as an alternative to the widely used regression approach. We seek to overcome the methodological limitations of the usual regression techniques by letting the data select the controls for IRA establishment. Even if the effectiveness or quality of regulation may differ depending on the country, we seek more to assess the impact of country’s propensity to undertake such regulatory reform. This article attempts to provide evidence on the positive or negative impact of independent regulation on electricity supply, and then builds foundations for future reforms in a sector representing one of the biggest challenges for developing countries, particularly in Africa (World Bank, 2010).
E
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
73
The rest of the article is organized as follows. The next section introduces the basic theoretical and conceptual foundations of regulation. Section III sets out our estimation method, based on a combination of propensity score matching and double differencing on preregulation and post-regulation data. Section IV addresses data issues. The results are presented in Section V while Section VI offers some conclusions.
II. Theoretical and Conceptual Foundations The optimism of the late 1980s and early 1990s that electricity and other similar infrastructure industries in developing countries could be commercialized and financed from private investment flows was associated with an optimistic view that viable and effective regulatory arrangements could readily be established in most countries.3 Amid growing dissatisfaction with the performance of publicly owned enterprises, particularly in the developing world, there has been a broad movement toward the regulation option. he traditional rationale for economic regulation of an industry has to do with its natural monopoly characteristics. Electricity industries are characterized by large, unavoidable fixed costs and many of these investments are sunk. The
T
74
electricity generation, transmission, and distribution sectors may have significantly declining average costs overall so that it is less costly for one enterprise to satisfy all demand than it is for more than one. atural monopolies have little chance of being driven out of a market by more-efficient new entrants; consequently, some form of regulation is necessary to protect
N
Distribution companies need to believe that they are able to obtain adequate returns on their investments if the industry is to be maintained as an attractive investment object. the customers’ rights in monopoly industries such as electricity transmission and distribution. Broadly defined, regulation is designed to achieve efficient use of energy, minimize production costs, provide clear investment incentives, and induce efficient organization of the electric services industry. It can play an important role in determining the operational framework for the regulated business, and it also has significant impacts on the overall development of the industry, because firms tend to adapt to their operating environment. In
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
other words, firms aim at optimizing their performance under a given regulatory regime. This sets high requirements to regulation, especially in an industry with long asset lifetimes, because of the far-reaching impacts of today’s decisions. Distribution companies need to believe that they are able to obtain adequate returns on their investments if the industry is to be maintained as an attractive investment object. Another important issue is to make sure that the so called viability constraint of regulation is met, because regardless of stakeholders’ price and quality expectations, the regulated companies will only supply the monopoly services if it is profitable for them to do so. Most developing countries use two common regulatory regimes: rate-of-return (ROR) regulation and price cap regulation. ROR is an approach where regulatory agencies fix the rate of return that a utility can earn on its assets. They set the price the utility can charge so as to allow it to earn a specified rate of return. The regulated price can be adjusted upward if the utility starts making a lower rate of return, and it will be adjusted downward if the utility makes a higher rate. Over the past decade, the price cap approach has become increasingly common internationally because it is thought to give firms stronger incentives to be efficient. Under this approach, the regulated price The Electricity Journal
is adjusted each year by the rate of inflation plus or minus some predetermined amount and without regard to changes in the firm’s profits. Both price caps and ROR regulation have been studied in the theoretical literature but experiences with their operation are not well documented for developing economies. Nevertheless price cap regulation seems to be the preferred model for utility regulation in a growing number of developing countries. In the Latin America and Caribbean region, a study for the World Bank indicated that the price cap method for regulating prices and profits is well established and seemingly preferred to ROR regulation (Guasch, 2001). A related study that analyzed almost 1,000 utility sector concession contracts awarded between 1980 and 2000 revealed that 56 percent of the contracts were regulated under a price cap regime. A further 24 percent of contracts used a hybrid regime, and only 20 percent were based on ROR regulation (Estache et al., 2003). A more recent survey of utility regulatory practice in developing countries sought to identify the difficulties faced when using price cap and ROR regulation (Kirkpatrick et al., 2005). When a price cap form of regulation is used, the difficulty most often cited in answers concerned information asymmetries, serious levels of customer complaints about rising prices, and political pressures.4 July 2015,
Vol. 28, Issue 6
Respondents were also asked to comment on difficulties faced when operating ROR regulation. Again, the most cited difficulty related to information asymmetries, serious levels of customer complaints about rising prices, enterprises overinvesting in capital equipment, and enterprises providing misleading information. To conclude, both price cap and ROR regulation seem to create the same
The degree of independence can be assessed by the amount of a regulator’s budget coming from industries’ fees, government appropriations, donors, or other sources or whether the country’s chief executive can veto regulatory decisions.
difficulties in developing countries; furthermore, a recent paper by Wren-Lewis (2014) concludes that the form of incentive regime is unlikely to be the determining factor in the performance of a given utility firm and suggests that the type of regulation used does not correlate strongly with performance with utilities in Africa. nother important issue to consider when studying regulatory performance is the level of independence of the regulatory agency from political powers. Levy and Spiller (1994)
A
propose to break down regulation into governance and content. Governance includes rules that order the scope, configuration, and process of regulatory decision-making while content (described above) includes the specific laws and rules toward pricing and investment. The degree of independence can be assessed by the amount of a regulator’s budget coming from industries’ fees, government appropriations, donors, or other sources or whether the country’s chief executive can veto regulatory decisions (Wallsten et al., 2004). However, no comprehensive analysis of these aspects of regulatory agencies in developing countries does exist; therefore, the analysis performed in this article will be limited to the de jure and not the de facto independency of regulatory agencies.
III. Methodology A true measure of the impact of an intervention is the difference between the observed outcome for a group of beneficiaries and the (counterfactual) outcome for the same group without the benefit of intervention. Because counterfactuals are never observed, the challenge of the evaluation work is to find the plausible proxies for such unobserved outcomes. In this study, for example, countries that have an IRA are likely to be very different from countries that do
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
75
not on a wide range of characteristics, such as the privatization of their electricity sector, even before the IRA is created. It is then crucial to try to separate out the causal effects of the IRA from the effects of these preexisting differences between the ‘‘treated’’ and ‘‘control’’ groups. To do so, we use matching techniques usually applied in non-experimental contexts, borrowed from the program evaluation literature. To be consistent with this literature in this section we may refer to the establishment of IRA as the treatment, to the IRAers as the treated group, and to the non-IRAers as the control group. atching methods allow one to control for the presence of IRAers in the sample that are not comparable with non-IRAers and vice versa, and to take into account the differences of characteristics between the two groups, which are usual in non-randomized samples like a dataset of countries. The goal is to estimate a suitable counterfactual for each IRAer in order to re-establish the conditions of a randomized experiment when no such data are available. Under these circumstances, the difference between the outcome of the treated and that of a matched counterfactual can be attributed to the treatment effect.5 o assess the impact of a policy one should track changes across different performance dimensions. In practice, several indicators need
to be constructed for each type of intervention and the choice of these indicators is determined by the practicalities of the evaluation and data collection. Impact or outcome indicators have to be measurable with the data at hand, and be linked directly to, and respond relatively quickly to the intervention in question. To measure the impact of creating an
M
T 76
IRA on performance in the electricity sector, we used three indicators: electricity power transmission and distribution losses (EL), electricity generation per capita (GEN), and GDP per unit of energy use (GDPU). GDP per unit of energy use (PPP $ per kg of oil-equivalent6) refers to apparent consumption, which is equal to indigenous production plus imports and stock changes, minus exports and fuels supplied to ships and aircraft engaged in international transportation. This indicator measures energy intensity and therefore a measure of the energy efficiency of a nation’s economy; differences in this ratio over time and across
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
countries reflect structural changes in the economy and changes in the energy efficiency. Energy efficiency offers a powerful and cost-effective tool for achieving a sustainable energy future. Improvements in energy efficiency can reduce the need for investment in energy infrastructure, cut fuel costs, increase competitiveness, and improve consumer welfare. Electricity power transmission and distribution losses (as a percentage of generation) include electricity losses due to operation of the system and the delivery of electricity as well as those caused by unmetered supply. This comprises all losses due to transportation and distribution of electrical energy and heat. Electric power transmission and distribution losses include losses in transmission between sources of supply and points of distribution and in the distribution to consumers, including pilferage. As stipulated in regulation theory, the establishment of an IRA is expected to improve the technical quality of electricity service and lead to a decline in this impact indicator. The third indicator used in this article, electricity generation per capita (gigawatt-hour per capita) captures the extent of electricity available to the economy. This indicator takes into account all the electricity production from different sources.7 The establishment of an IRA would reduce costs, improve service quality, stimulate The Electricity Journal
efficient investment, and then lead to higher generation per capita.
IV. Data The main database used to perform the study is the World Bank World Development Indicators, which presents the most current and accurate global development data from officially recognized sources. For information on the establishment of regulatory policies we rely on Table 1: Sample Countries[2_TD$IF]. Year Independent Treated Countries
Regulator was Established
Albania Armenia
1999 1997
Bulgaria
2000
Cameroon Costa Rica
1998 1996
Dominican Republic Ecuador
1998 1998
El Salvador
1996
Ethiopia Georgia
1996 1997
India Ivory Coast
1998 1998
Jamaica
1997
Kazakhstan Latvia
1999 1996
Lithuania Moldova
1998 1997
Namibia
2000
Nicaragua Panama
1995 1996
Romania South Africa
1999 1996
Thailand
1999
Zambia
1995
July 2015,
Vol. 28, Issue 6
the dataset constructed by Wallsten et al. (2004); the authors set out to compile a comprehensive and consistent dataset through an extensive survey of telecommunications and electricity regulators in developing countries. Our sample includes 24 countries8 that have established an independent regulatory agency of their electricity sector between 1995 and 2000 (Table 1). The choice of the sample countries was based on access to data and especially information on regulation reform. The control group is composed of 34 developing countries that have not created an IRA by 2001. he performance indicators used in the study are, as mentioned above, net electricity generation per capita, gross domestic product per unit of energy used, and electricity power transmission and distribution losses. These indicators capture the extent of electricity available to the economy and quality of the service. A number of variables were included in the model to control and define a particular profile of an economy. These variables, used to generate the propensity score for each country, include GDP per capita, population, gross domestic savings (GDS), the openness of an economy (openness ratio), the gross fixed capital formation of the private sector (GFCF), foreign direct investment (FDI), oil rents to GDP, the population density,
T
the polity score, and dummy variables for privatization and legal origin. We experimented with a wide set of additional variables, but their inclusion did not change our results substantively. We tested for the balancing hypothesis and selected the most parsimonious specification. For each country we calculate the average of the above variables for the five years previous the establishment of an independent regulatory authority. For the control group, we use the five years previous to 19979 given that the treatment did not occur the same year for all treated countries.
V. Results A. Descriptive statistics Our sample is composed of 58 countries and 495 observations. Table 2 displays the mean values of the three impact indicators and the characteristics of IRAers and non-IRAers before and after the introduction on an IRA. For the latter, there is a clear difference between IRAers and non-IRAers and this justifies the use of matching to remove the selection bias due to the observed differences between the treated and the control groups. Regarding the impact indicators, performances are on average somewhat better in IRAer countries compared to non-IRAer countries before the introduction of
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
77
Table 2: Summary Statistics for Variables[3_TD$IF]. Impact Indicators
Treated Group
Control Group
T0 a
T1
T0
T1
Electricity T&D losses
[4_TD$IF]20.61 *** (2.403)
15.91 *** (1.818)
14.70 *** (1.886)
17.88*** (1.490)
GDP/unit of energy use
[5_TD$IF]4.722 *** (0.626)
6.057 *** (0.616)
5.319 *** (0.684)
5.625*** (0.658)
Generation per capita
[6_TD$IF]1.627 *** (0.229)
2.346 *** (0.219)
1.293 *** (0.201)
1.188*** (0.162)
Covariates
Treated Group
Control Group
Mean
S.E.
Mean
S.E.
Log GDP per capita Log population
[7_TD$IF]7.059 *** [9_TD$IF]2.218 ***
0.203 0.329
[8_TD$IF]6.801*** [10_TD$IF]2.230***
0.223 0.216
Privatization
[1_TD$IF]0.714 ***
0.101
[12_TD$IF]0.348***
0.102
Trade Oil rents
[13_TD$IF]78.44 [15_TD$IF]1.474 *
8.411 0.692
[14_TD$IF]72.53 3.125
6.063 1.892
FDI GFCF
16_TD$IF]2[ .020 *** [18_TD$IF]19.18 ***
0.276 1.065
17_TD$IF]1[ .922** [19_TD$IF]21.43 ***
0.716 0.975
Density of population Democracy
[20_TD$IF]89.92 *** [2_TD$IF]5.886 ***
18.98 0.877
[21_TD$IF]52.50 *** 1.470
10.87 1.174
[23_TD$IF]0.286 **
0.101
[24_TD$IF]0.174*
Inflation Urban population
*
[25_TD$IF]2.417 [26_TD$IF]49.07 ***
1.047 3.197
*
1.724 [27_TD$IF]51.30 ***
Fuel consumption
[28_TD$IF]62.40 ***
5.953
[29_TD$IF]54.18 ***
Legal origin
***
***
0.0808 0.748 4.377 6.383
T0 and T1 refer respectively to the five years before and after the IRA adoption[30_TD$IF]. Significant at 1%, significant at 5% and * significant at 10% level. Standard errors are given in parentheses. a
***
independent regulators. Apart from non-IRAers’ electricity losses, they generally moved in the expected direction over time, with a tendency to increase over time in both IRAer and non-IRAer countries. B. Electricity power transmission and distribution losses Losses are estimated from the discrepancy between energy 78
**
produced (as reported by power plants) and energy sold to end customers more simply the difference between what is produced and what is consumed. Given the insufficiency of energy production in developing countries, minimizing power transmission and distribution losses represent a major challenge; that makes it important to ask whether independent regulation can improve the technical quality of the sector.
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
The Kernel propensity score matching double-difference estimation of the impact of IRA on electricity power transmission and distribution losses is shown in Table 3. The first two rows show results of simple difference-in-difference estimates with and without covariates and the last two rows present results for our model specification presented above in Section III. Under our assumptions, these estimations reflect causal effects of IRA establishment. One or two stars indicate whether each change is significantly different from zero at the 10 and 5 percent significance levels, respectively. verall, we see a decline in our outcome indicator; focusing on the Kernel-PS-based estimates, the establishment of IRA resulted in a significant decline in electricity power T&D losses by 7.36 percentage points if we consider the unmatched sample. The matched sample shows an even larger statistically significant difference, the outcome indicator declining by 8.42 percentage points. These results are consistent with the independent regulation’s effects on productivity and efficiency gains.
O
C. Electricity generation per capita Concerning electricity generation, the estimates in Table 3 show an increase in the average generation per capita respectively by 0.766 and The Electricity Journal
Table 3: Average Impact of IRA Adoption[31_TD$IF]. Treatment Effect
(1)
(2)
Diff-in-Diff
(3)
(4)
PS Kernel Matched DD With Covariates
Common Support
Electricity T&D losses
7.919** (3.525)
7.626** (3.476)
7.368** (3.635)
8.425* (4.671)
GDP per unit of energy use
1.402 (1.175)
1.312 (0.945)
1.226 (1.217)
1.029 (1.585)
Generation per capita
0.766**
0.946**
0.862**
0.883*
(0.360)
(0.364)
(0.392)
(0.515)
[32_TD$IF]***
[3_TD$IF]Significant [34_TD$IF]at [35_TD$IF]1%,
[36_TD$IF]**
[37_TD$IF]significant at [38_TD$IF]5% [39_TD$IF]and significant at [40_TD$IF]10% level. [41_TD$IF]Standard [42_TD$IF]errors [43_TD$IF]are [4_TD$IF]given in parentheses. *
0.946 GWh per capita; the Kernel PS based estimates confirm these results, the average impact being 0.883 for the matched sample. This positive impact of regulation can be explained by the fact that it allows more competition and generate gains in productive and allocative efficiency. These results are consistent with those of Cubbin and Stern (2006) who confirm the theoretically positive effect of regulation, which leads to higher generation per capita. D. Energy efficiency Efficiency improvements in electricity sector are fundamental, moreso in developing countries, for two reasons. First, an increase in the GDP per unit of energy used reduces costs and may result in financial cost saving to consumers. Second, reducing energy use could be a key solution in reducing global emissions of greenhouse gases. An independent regulation authority could enhance energy efficiency July 2015,
Vol. 28, Issue 6
and by this means increase competitiveness and improve consumer welfare. As for previous outcome indicators, we first perform simple doubledifference estimates with and without covariates before implementing the Kernel PS method. verall, the establishment of an independent regulation authority is associated with more energy efficiency, the GDP per unit of energy used increased by 1.029 to 1.402 $ per kg of oil equivalent. However, none of the four estimates do provide statistically significant results, meaning that we cannot conclude that IRA establishment has significantly improve energy efficiency in developing countries.
O
VI. Discussion and Conclusion In many developing countries, significant reforms have been undertaken in economic
infrastructure sectors over the past two decades with the objectives of improving efficiency and the quality of service delivery. While many authors have empirically investigated the effects of privatization in infrastructure sectors, less attention has been paid to the potential role that regulatory policies can play in improving infrastructure outputs and outcomes. As in most infrastructure sectors, the electricity sector is characterized by long-term investments, meaning that an effective regulatory system can be crucial to ensure both investor confidence and consumer protection. In fact, one of the primary purposes of a regulation system is to protect consumers from monopoly abuse, while providing investors with protection from arbitrary political action and incentives to promote efficient operation and investment. his article attempts to contribute to this literature by
T
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
79
investigating the impact of the creation of an independent regulatory authority on infrastructure performance. To do so, we investigate the impact of such reforms on power generation, quality, and efficiency of the electricity sector in developing countries. We have combined macroeconomic data on regulation with program evaluation techniques to assess the theoretically foundations of such reform. The interpretation we gave to IRA establishment, that of a natural experiment, allowed us to use powerful evaluation tools normally applied in micro-econometrics, where the odds to identify policy effects are by far higher than in macroeconomics. We remove the selection bias due to observable and time-invariant unobservable characteristics in identifying IRAs’ impacts by combining propensity score matching with the double-difference method. Our findings support the idea that IRA establishment delivers some promised outcomes. It improves the technical quality of electricity service: electricity power T&D losses fall in average by 8.42 percentage points. It also increases the average electricity generated by 0.88 GWh per capita. The impact on energy efficiency measured by GDP per unit of energy used is also positive but insignificant. We would like to measure the treatment impact on more
80
outcome indicators, such as the price of electricity or universal access to electricity, but we face the lack of longitudinal data. These findings do confirm that the establishment of independent regulation agencies in public utilities can significantly improve the productivity and quality of infrastructure, and potentially improve coverage and access to basic infrastructure. This article also demonstrates that robust evaluation of public sector reform is possible even in the absence of a proper baseline survey. Finally, the evidence presented also provides an input to the development and implementation of effective reform in economic infrastructure sectors in developing countries.&
References Bortolotti, B., Fantini, M., Siniscalco, D., 1998. Regulation and Privatization: the Case of Electricity. Department of Economics and Finance Working Paper, University of Turin. Brown, A.C., Stern, J., Tenenbaum, B., 2006. Handbook for Evaluating Infrastructure Regulatory Systems. World Bank, Washington, DC. Cubbin, J., Stern, J., 2006. The impact of regulatory governance and privatization on electricity industry generation capacity in developing economies. World Bank Econ. Rev. 20, 115–141. Estache, A., Rodriguez Pardina, M., Rodriguez, J.-M., Sember, G., 2003. An Introduction to Financial and Economic Modeling for Utility Regulators. Policy Research Working Paper Series 3001. Gassner, K., Pushak, N., 2014. 30 years of British utility regulation: developing country experience and outlook. Util. Policy 31, 44–51.
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
Guasch, L., 2001. Concessions and Regulatory Design: Determinants of Performance, Fifteen Years of Evidence. Mimeo. World Bank and University of California San Diego. Guasch, J.-L., Lafont, J.-J., Straub, S., 2007. Concession of infrastructure in Latin America: government-led renegotiation. J. Appl. Econometr. 22 (7) 1267–1294. Jamasb, T., Mota, R., Newbery, D., Pollitt, M., 2004. Electricity Sector Reform in Developing Countries: A survey of Empirical Evidence on Determinants and Performance. Cambridge Working Papers in Economics, CWPE 0439, Department of Applied Economics. Kirkpatrick, C., Parker, D., Zhang, Y.F., 2005. Competition, regulation and privatization of electricity generation in developing countries: does the sequencing of the reforms matter? Q. Rev. Econ. Financ. 45, 358–379. Kirkpatrick, C., Parker, D., Zhang, Y.F., 2008. Electricity sector reform in developing countries: an econometric assessment of the effects of privatization, competition and regulation. J. Regul. Econ. 33, 159– 178. Levy, B., Spiller, P.T., 1994. The institutional foundations of regulatory commitment: a comparative analysis of telecommunication regulation. J. Law Econ. Org. 10 (2) 201–246. Spiller, P.T., 1996. A positive political theory of regulatory instruments: contracts, administrative law or regulatory specificity? South. Calif. Law Rev. 69, 477–515. Wallsten, S., Clarke, G., Haggarty, L., Kaneshiro, R., Noll, R., Shirley, M., Xu, L.C., 2004. New Tools for Studying Network Industry Reforms in Developing Countries: the Telecommunications and Electricity Regulation Database. World Bank Policy Research Working Paper 3286. World Bank, 2010. In: Foster, V., Briceno-Garmendia, C. (Eds.), Africa’s Infrastructure: A Time for Transformation. World Bank, Washington, DC. Wren-Lewis, L., 2014. Utility regulation in Africa: how relevant is the
The Electricity Journal
British model? Util. Policy, http:// dx.doi.org/10.1016/j.jup. 2014.09.002.
Endnotes: 1. They used the following indicators to measure sector performance: net electricity generation per capita, installed generation per capita, net electricity generation per employee, and generation relative to average capacity. 2. Andres et al. (2008) use a similar approach to assess the impact of private sector participation on infrastructure.
July 2015,
Vol. 28, Issue 6
3. See Levy and Spiller (1994). 4. Gassner and Pushak (2014) also found that political pressures are one of the main barriers to the successful introduction of regulatory agencies in low-income and middle-income countries. 5. Let D be a binary indicator that equals unity if a country has established an IRA and zero otherwise. Also, let Y1t denote the value of certain outcome in period t if the country has an IRA and Y0t if not. Given a set of observable country attributes X, the average effect of the IRA on Yt = E[(Y1t1 Y0t0) jX, D = 1] E[(Y0t1 Y0t0) jX, D = 0].
Following Heckman et al. (1997) and Heckman et al. (1998), we can compute a consistent estimator of the counterfactual by means of a kernelweighted average of outcomes. 6. The kg of oil equivalent is a unit of energy: the amount of energy released by burning one kg of crude oil (as different crude oils have different calorific values, the exact value of the toe is defined by convention). 7. Coal, oil, natural gas, nuclear, and hydroelectric sources. 8. See Table 1 for the list of countries. 9. 1997 is the average establishment date of an IRA in the treated group.
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
81
72
Babacar Sarr
I. Introduction Regulatory systems for infrastructure sectors are a relatively new but important phenomenon in many developing countries. It has been estimated that close to 200 new infrastructure regulators have been created around the world in the past 10 years (Brown et al., 2006). These regulatory systems are designed to respond to natural monopolies and market failures associated with network industries such as
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
electricity, gas, water, telecommunications, and transportation. The aim of regulation is to encourage efficient, low-cost, and reliable service provision while ensuring financial viability and new investment. The expectation is that regulation mechanisms can depoliticize tariff setting, reduce costs, improve service quality in a cost-effective way, stimulate the introduction of new products and services, and stimulate efficient investment. Therefore, the establishment of an The Electricity Journal
independent regulation agency (IRA) should lead to improved economic performance. he electricity sector has three components: generation, transmission, and distribution. In most countries this sector is a strategic activity with natural monopoly characteristics resulting from the existence of economies of scale and scope. Regulation is required especially in the areas of electricity supply that remain dominated by one or a very small number of operators, to prevent monopoly abuse. In many countries, instead of direct regulation by a government department, the establishment of independent or quasiindependent regulatory agencies has been favored, drawing on the regulatory models of the United States and the United Kingdom. This form of independent regulation is expected to encourage private capital to invest in capacity in the face of a potential holdup problem under conditions of incomplete contracts (Spiller, 1996). Despite these good intentions, there is little evidence that these regulatory systems have met their expectations. Indeed, the literature on estimating the effect of regulatory governance arrangements on infrastructure outcomes is relatively small to date, particularly for the electricity sector (Jamasb et al., 2004). A particularly important study is that by Kirkpatrick et al. (2008), who explore the impacts of competition, privatization, and
T
July 2015,
Vol. 28, Issue 6
regulation on electricity sector performance1 in 36 developing and transitional countries. In their results, the regulation variable is not a significant, correctly signed explanatory variable in any of the regressions reported. However, in this paper, the regulatory variable is a dummy which is not dated; for instance, a country like Ethiopia establishing an independent regulatory body in 1999 scores 1 for the whole of the
Despite good good intentions, there is little evidence that these regulatory systems have met their expectations. estimation period, as would Costa Rica, which did so in 1928. Therefore, the experience of the regulator is not taken into account in their model. Bortolotti et al. (1998), who use data on the privatization of electricity generation in 38 countries between 1977 and 1997, find that effective regulation is a prerequisite for the success of privatization. Cubbin and Stern (2006) conclude that the quality of regulatory governance is positively and significantly associated with higher per-capita generation capacity levels and that this positive effect increases
over time with the reputation of the regulator. mpirical testing of the performance of regulation seems to concentrate on case studies and the application of panel-data econometrics (Guasch et al., 2007). These methods cannot allow a statistically robust separation of the effects of other changes like privatization from the impact of the establishment of an independent regulatory agency. Therefore, we propose to interpret the introduction of an independent regulator as a natural experiment,2 in order to re-establish the conditions of a randomized experiment and represent the IRA as a treatment. This leads us to perform propensity score matching as an alternative to the widely used regression approach. We seek to overcome the methodological limitations of the usual regression techniques by letting the data select the controls for IRA establishment. Even if the effectiveness or quality of regulation may differ depending on the country, we seek more to assess the impact of country’s propensity to undertake such regulatory reform. This article attempts to provide evidence on the positive or negative impact of independent regulation on electricity supply, and then builds foundations for future reforms in a sector representing one of the biggest challenges for developing countries, particularly in Africa (World Bank, 2010).
E
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
73
The rest of the article is organized as follows. The next section introduces the basic theoretical and conceptual foundations of regulation. Section III sets out our estimation method, based on a combination of propensity score matching and double differencing on preregulation and post-regulation data. Section IV addresses data issues. The results are presented in Section V while Section VI offers some conclusions.
II. Theoretical and Conceptual Foundations The optimism of the late 1980s and early 1990s that electricity and other similar infrastructure industries in developing countries could be commercialized and financed from private investment flows was associated with an optimistic view that viable and effective regulatory arrangements could readily be established in most countries.3 Amid growing dissatisfaction with the performance of publicly owned enterprises, particularly in the developing world, there has been a broad movement toward the regulation option. he traditional rationale for economic regulation of an industry has to do with its natural monopoly characteristics. Electricity industries are characterized by large, unavoidable fixed costs and many of these investments are sunk. The
T
74
electricity generation, transmission, and distribution sectors may have significantly declining average costs overall so that it is less costly for one enterprise to satisfy all demand than it is for more than one. atural monopolies have little chance of being driven out of a market by more-efficient new entrants; consequently, some form of regulation is necessary to protect
N
Distribution companies need to believe that they are able to obtain adequate returns on their investments if the industry is to be maintained as an attractive investment object. the customers’ rights in monopoly industries such as electricity transmission and distribution. Broadly defined, regulation is designed to achieve efficient use of energy, minimize production costs, provide clear investment incentives, and induce efficient organization of the electric services industry. It can play an important role in determining the operational framework for the regulated business, and it also has significant impacts on the overall development of the industry, because firms tend to adapt to their operating environment. In
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
other words, firms aim at optimizing their performance under a given regulatory regime. This sets high requirements to regulation, especially in an industry with long asset lifetimes, because of the far-reaching impacts of today’s decisions. Distribution companies need to believe that they are able to obtain adequate returns on their investments if the industry is to be maintained as an attractive investment object. Another important issue is to make sure that the so called viability constraint of regulation is met, because regardless of stakeholders’ price and quality expectations, the regulated companies will only supply the monopoly services if it is profitable for them to do so. Most developing countries use two common regulatory regimes: rate-of-return (ROR) regulation and price cap regulation. ROR is an approach where regulatory agencies fix the rate of return that a utility can earn on its assets. They set the price the utility can charge so as to allow it to earn a specified rate of return. The regulated price can be adjusted upward if the utility starts making a lower rate of return, and it will be adjusted downward if the utility makes a higher rate. Over the past decade, the price cap approach has become increasingly common internationally because it is thought to give firms stronger incentives to be efficient. Under this approach, the regulated price The Electricity Journal
is adjusted each year by the rate of inflation plus or minus some predetermined amount and without regard to changes in the firm’s profits. Both price caps and ROR regulation have been studied in the theoretical literature but experiences with their operation are not well documented for developing economies. Nevertheless price cap regulation seems to be the preferred model for utility regulation in a growing number of developing countries. In the Latin America and Caribbean region, a study for the World Bank indicated that the price cap method for regulating prices and profits is well established and seemingly preferred to ROR regulation (Guasch, 2001). A related study that analyzed almost 1,000 utility sector concession contracts awarded between 1980 and 2000 revealed that 56 percent of the contracts were regulated under a price cap regime. A further 24 percent of contracts used a hybrid regime, and only 20 percent were based on ROR regulation (Estache et al., 2003). A more recent survey of utility regulatory practice in developing countries sought to identify the difficulties faced when using price cap and ROR regulation (Kirkpatrick et al., 2005). When a price cap form of regulation is used, the difficulty most often cited in answers concerned information asymmetries, serious levels of customer complaints about rising prices, and political pressures.4 July 2015,
Vol. 28, Issue 6
Respondents were also asked to comment on difficulties faced when operating ROR regulation. Again, the most cited difficulty related to information asymmetries, serious levels of customer complaints about rising prices, enterprises overinvesting in capital equipment, and enterprises providing misleading information. To conclude, both price cap and ROR regulation seem to create the same
The degree of independence can be assessed by the amount of a regulator’s budget coming from industries’ fees, government appropriations, donors, or other sources or whether the country’s chief executive can veto regulatory decisions.
difficulties in developing countries; furthermore, a recent paper by Wren-Lewis (2014) concludes that the form of incentive regime is unlikely to be the determining factor in the performance of a given utility firm and suggests that the type of regulation used does not correlate strongly with performance with utilities in Africa. nother important issue to consider when studying regulatory performance is the level of independence of the regulatory agency from political powers. Levy and Spiller (1994)
A
propose to break down regulation into governance and content. Governance includes rules that order the scope, configuration, and process of regulatory decision-making while content (described above) includes the specific laws and rules toward pricing and investment. The degree of independence can be assessed by the amount of a regulator’s budget coming from industries’ fees, government appropriations, donors, or other sources or whether the country’s chief executive can veto regulatory decisions (Wallsten et al., 2004). However, no comprehensive analysis of these aspects of regulatory agencies in developing countries does exist; therefore, the analysis performed in this article will be limited to the de jure and not the de facto independency of regulatory agencies.
III. Methodology A true measure of the impact of an intervention is the difference between the observed outcome for a group of beneficiaries and the (counterfactual) outcome for the same group without the benefit of intervention. Because counterfactuals are never observed, the challenge of the evaluation work is to find the plausible proxies for such unobserved outcomes. In this study, for example, countries that have an IRA are likely to be very different from countries that do
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
75
not on a wide range of characteristics, such as the privatization of their electricity sector, even before the IRA is created. It is then crucial to try to separate out the causal effects of the IRA from the effects of these preexisting differences between the ‘‘treated’’ and ‘‘control’’ groups. To do so, we use matching techniques usually applied in non-experimental contexts, borrowed from the program evaluation literature. To be consistent with this literature in this section we may refer to the establishment of IRA as the treatment, to the IRAers as the treated group, and to the non-IRAers as the control group. atching methods allow one to control for the presence of IRAers in the sample that are not comparable with non-IRAers and vice versa, and to take into account the differences of characteristics between the two groups, which are usual in non-randomized samples like a dataset of countries. The goal is to estimate a suitable counterfactual for each IRAer in order to re-establish the conditions of a randomized experiment when no such data are available. Under these circumstances, the difference between the outcome of the treated and that of a matched counterfactual can be attributed to the treatment effect.5 o assess the impact of a policy one should track changes across different performance dimensions. In practice, several indicators need
to be constructed for each type of intervention and the choice of these indicators is determined by the practicalities of the evaluation and data collection. Impact or outcome indicators have to be measurable with the data at hand, and be linked directly to, and respond relatively quickly to the intervention in question. To measure the impact of creating an
M
T 76
IRA on performance in the electricity sector, we used three indicators: electricity power transmission and distribution losses (EL), electricity generation per capita (GEN), and GDP per unit of energy use (GDPU). GDP per unit of energy use (PPP $ per kg of oil-equivalent6) refers to apparent consumption, which is equal to indigenous production plus imports and stock changes, minus exports and fuels supplied to ships and aircraft engaged in international transportation. This indicator measures energy intensity and therefore a measure of the energy efficiency of a nation’s economy; differences in this ratio over time and across
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
countries reflect structural changes in the economy and changes in the energy efficiency. Energy efficiency offers a powerful and cost-effective tool for achieving a sustainable energy future. Improvements in energy efficiency can reduce the need for investment in energy infrastructure, cut fuel costs, increase competitiveness, and improve consumer welfare. Electricity power transmission and distribution losses (as a percentage of generation) include electricity losses due to operation of the system and the delivery of electricity as well as those caused by unmetered supply. This comprises all losses due to transportation and distribution of electrical energy and heat. Electric power transmission and distribution losses include losses in transmission between sources of supply and points of distribution and in the distribution to consumers, including pilferage. As stipulated in regulation theory, the establishment of an IRA is expected to improve the technical quality of electricity service and lead to a decline in this impact indicator. The third indicator used in this article, electricity generation per capita (gigawatt-hour per capita) captures the extent of electricity available to the economy. This indicator takes into account all the electricity production from different sources.7 The establishment of an IRA would reduce costs, improve service quality, stimulate The Electricity Journal
efficient investment, and then lead to higher generation per capita.
IV. Data The main database used to perform the study is the World Bank World Development Indicators, which presents the most current and accurate global development data from officially recognized sources. For information on the establishment of regulatory policies we rely on Table 1: Sample Countries[2_TD$IF]. Year Independent Treated Countries
Regulator was Established
Albania Armenia
1999 1997
Bulgaria
2000
Cameroon Costa Rica
1998 1996
Dominican Republic Ecuador
1998 1998
El Salvador
1996
Ethiopia Georgia
1996 1997
India Ivory Coast
1998 1998
Jamaica
1997
Kazakhstan Latvia
1999 1996
Lithuania Moldova
1998 1997
Namibia
2000
Nicaragua Panama
1995 1996
Romania South Africa
1999 1996
Thailand
1999
Zambia
1995
July 2015,
Vol. 28, Issue 6
the dataset constructed by Wallsten et al. (2004); the authors set out to compile a comprehensive and consistent dataset through an extensive survey of telecommunications and electricity regulators in developing countries. Our sample includes 24 countries8 that have established an independent regulatory agency of their electricity sector between 1995 and 2000 (Table 1). The choice of the sample countries was based on access to data and especially information on regulation reform. The control group is composed of 34 developing countries that have not created an IRA by 2001. he performance indicators used in the study are, as mentioned above, net electricity generation per capita, gross domestic product per unit of energy used, and electricity power transmission and distribution losses. These indicators capture the extent of electricity available to the economy and quality of the service. A number of variables were included in the model to control and define a particular profile of an economy. These variables, used to generate the propensity score for each country, include GDP per capita, population, gross domestic savings (GDS), the openness of an economy (openness ratio), the gross fixed capital formation of the private sector (GFCF), foreign direct investment (FDI), oil rents to GDP, the population density,
T
the polity score, and dummy variables for privatization and legal origin. We experimented with a wide set of additional variables, but their inclusion did not change our results substantively. We tested for the balancing hypothesis and selected the most parsimonious specification. For each country we calculate the average of the above variables for the five years previous the establishment of an independent regulatory authority. For the control group, we use the five years previous to 19979 given that the treatment did not occur the same year for all treated countries.
V. Results A. Descriptive statistics Our sample is composed of 58 countries and 495 observations. Table 2 displays the mean values of the three impact indicators and the characteristics of IRAers and non-IRAers before and after the introduction on an IRA. For the latter, there is a clear difference between IRAers and non-IRAers and this justifies the use of matching to remove the selection bias due to the observed differences between the treated and the control groups. Regarding the impact indicators, performances are on average somewhat better in IRAer countries compared to non-IRAer countries before the introduction of
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
77
Table 2: Summary Statistics for Variables[3_TD$IF]. Impact Indicators
Treated Group
Control Group
T0 a
T1
T0
T1
Electricity T&D losses
[4_TD$IF]20.61 *** (2.403)
15.91 *** (1.818)
14.70 *** (1.886)
17.88*** (1.490)
GDP/unit of energy use
[5_TD$IF]4.722 *** (0.626)
6.057 *** (0.616)
5.319 *** (0.684)
5.625*** (0.658)
Generation per capita
[6_TD$IF]1.627 *** (0.229)
2.346 *** (0.219)
1.293 *** (0.201)
1.188*** (0.162)
Covariates
Treated Group
Control Group
Mean
S.E.
Mean
S.E.
Log GDP per capita Log population
[7_TD$IF]7.059 *** [9_TD$IF]2.218 ***
0.203 0.329
[8_TD$IF]6.801*** [10_TD$IF]2.230***
0.223 0.216
Privatization
[1_TD$IF]0.714 ***
0.101
[12_TD$IF]0.348***
0.102
Trade Oil rents
[13_TD$IF]78.44 [15_TD$IF]1.474 *
8.411 0.692
[14_TD$IF]72.53 3.125
6.063 1.892
FDI GFCF
16_TD$IF]2[ .020 *** [18_TD$IF]19.18 ***
0.276 1.065
17_TD$IF]1[ .922** [19_TD$IF]21.43 ***
0.716 0.975
Density of population Democracy
[20_TD$IF]89.92 *** [2_TD$IF]5.886 ***
18.98 0.877
[21_TD$IF]52.50 *** 1.470
10.87 1.174
[23_TD$IF]0.286 **
0.101
[24_TD$IF]0.174*
Inflation Urban population
*
[25_TD$IF]2.417 [26_TD$IF]49.07 ***
1.047 3.197
*
1.724 [27_TD$IF]51.30 ***
Fuel consumption
[28_TD$IF]62.40 ***
5.953
[29_TD$IF]54.18 ***
Legal origin
***
***
0.0808 0.748 4.377 6.383
T0 and T1 refer respectively to the five years before and after the IRA adoption[30_TD$IF]. Significant at 1%, significant at 5% and * significant at 10% level. Standard errors are given in parentheses. a
***
independent regulators. Apart from non-IRAers’ electricity losses, they generally moved in the expected direction over time, with a tendency to increase over time in both IRAer and non-IRAer countries. B. Electricity power transmission and distribution losses Losses are estimated from the discrepancy between energy 78
**
produced (as reported by power plants) and energy sold to end customers more simply the difference between what is produced and what is consumed. Given the insufficiency of energy production in developing countries, minimizing power transmission and distribution losses represent a major challenge; that makes it important to ask whether independent regulation can improve the technical quality of the sector.
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
The Kernel propensity score matching double-difference estimation of the impact of IRA on electricity power transmission and distribution losses is shown in Table 3. The first two rows show results of simple difference-in-difference estimates with and without covariates and the last two rows present results for our model specification presented above in Section III. Under our assumptions, these estimations reflect causal effects of IRA establishment. One or two stars indicate whether each change is significantly different from zero at the 10 and 5 percent significance levels, respectively. verall, we see a decline in our outcome indicator; focusing on the Kernel-PS-based estimates, the establishment of IRA resulted in a significant decline in electricity power T&D losses by 7.36 percentage points if we consider the unmatched sample. The matched sample shows an even larger statistically significant difference, the outcome indicator declining by 8.42 percentage points. These results are consistent with the independent regulation’s effects on productivity and efficiency gains.
O
C. Electricity generation per capita Concerning electricity generation, the estimates in Table 3 show an increase in the average generation per capita respectively by 0.766 and The Electricity Journal
Table 3: Average Impact of IRA Adoption[31_TD$IF]. Treatment Effect
(1)
(2)
Diff-in-Diff
(3)
(4)
PS Kernel Matched DD With Covariates
Common Support
Electricity T&D losses
7.919** (3.525)
7.626** (3.476)
7.368** (3.635)
8.425* (4.671)
GDP per unit of energy use
1.402 (1.175)
1.312 (0.945)
1.226 (1.217)
1.029 (1.585)
Generation per capita
0.766**
0.946**
0.862**
0.883*
(0.360)
(0.364)
(0.392)
(0.515)
[32_TD$IF]***
[3_TD$IF]Significant [34_TD$IF]at [35_TD$IF]1%,
[36_TD$IF]**
[37_TD$IF]significant at [38_TD$IF]5% [39_TD$IF]and significant at [40_TD$IF]10% level. [41_TD$IF]Standard [42_TD$IF]errors [43_TD$IF]are [4_TD$IF]given in parentheses. *
0.946 GWh per capita; the Kernel PS based estimates confirm these results, the average impact being 0.883 for the matched sample. This positive impact of regulation can be explained by the fact that it allows more competition and generate gains in productive and allocative efficiency. These results are consistent with those of Cubbin and Stern (2006) who confirm the theoretically positive effect of regulation, which leads to higher generation per capita. D. Energy efficiency Efficiency improvements in electricity sector are fundamental, moreso in developing countries, for two reasons. First, an increase in the GDP per unit of energy used reduces costs and may result in financial cost saving to consumers. Second, reducing energy use could be a key solution in reducing global emissions of greenhouse gases. An independent regulation authority could enhance energy efficiency July 2015,
Vol. 28, Issue 6
and by this means increase competitiveness and improve consumer welfare. As for previous outcome indicators, we first perform simple doubledifference estimates with and without covariates before implementing the Kernel PS method. verall, the establishment of an independent regulation authority is associated with more energy efficiency, the GDP per unit of energy used increased by 1.029 to 1.402 $ per kg of oil equivalent. However, none of the four estimates do provide statistically significant results, meaning that we cannot conclude that IRA establishment has significantly improve energy efficiency in developing countries.
O
VI. Discussion and Conclusion In many developing countries, significant reforms have been undertaken in economic
infrastructure sectors over the past two decades with the objectives of improving efficiency and the quality of service delivery. While many authors have empirically investigated the effects of privatization in infrastructure sectors, less attention has been paid to the potential role that regulatory policies can play in improving infrastructure outputs and outcomes. As in most infrastructure sectors, the electricity sector is characterized by long-term investments, meaning that an effective regulatory system can be crucial to ensure both investor confidence and consumer protection. In fact, one of the primary purposes of a regulation system is to protect consumers from monopoly abuse, while providing investors with protection from arbitrary political action and incentives to promote efficient operation and investment. his article attempts to contribute to this literature by
T
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
79
investigating the impact of the creation of an independent regulatory authority on infrastructure performance. To do so, we investigate the impact of such reforms on power generation, quality, and efficiency of the electricity sector in developing countries. We have combined macroeconomic data on regulation with program evaluation techniques to assess the theoretically foundations of such reform. The interpretation we gave to IRA establishment, that of a natural experiment, allowed us to use powerful evaluation tools normally applied in micro-econometrics, where the odds to identify policy effects are by far higher than in macroeconomics. We remove the selection bias due to observable and time-invariant unobservable characteristics in identifying IRAs’ impacts by combining propensity score matching with the double-difference method. Our findings support the idea that IRA establishment delivers some promised outcomes. It improves the technical quality of electricity service: electricity power T&D losses fall in average by 8.42 percentage points. It also increases the average electricity generated by 0.88 GWh per capita. The impact on energy efficiency measured by GDP per unit of energy used is also positive but insignificant. We would like to measure the treatment impact on more
80
outcome indicators, such as the price of electricity or universal access to electricity, but we face the lack of longitudinal data. These findings do confirm that the establishment of independent regulation agencies in public utilities can significantly improve the productivity and quality of infrastructure, and potentially improve coverage and access to basic infrastructure. This article also demonstrates that robust evaluation of public sector reform is possible even in the absence of a proper baseline survey. Finally, the evidence presented also provides an input to the development and implementation of effective reform in economic infrastructure sectors in developing countries.&
References Bortolotti, B., Fantini, M., Siniscalco, D., 1998. Regulation and Privatization: the Case of Electricity. Department of Economics and Finance Working Paper, University of Turin. Brown, A.C., Stern, J., Tenenbaum, B., 2006. Handbook for Evaluating Infrastructure Regulatory Systems. World Bank, Washington, DC. Cubbin, J., Stern, J., 2006. The impact of regulatory governance and privatization on electricity industry generation capacity in developing economies. World Bank Econ. Rev. 20, 115–141. Estache, A., Rodriguez Pardina, M., Rodriguez, J.-M., Sember, G., 2003. An Introduction to Financial and Economic Modeling for Utility Regulators. Policy Research Working Paper Series 3001. Gassner, K., Pushak, N., 2014. 30 years of British utility regulation: developing country experience and outlook. Util. Policy 31, 44–51.
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
Guasch, L., 2001. Concessions and Regulatory Design: Determinants of Performance, Fifteen Years of Evidence. Mimeo. World Bank and University of California San Diego. Guasch, J.-L., Lafont, J.-J., Straub, S., 2007. Concession of infrastructure in Latin America: government-led renegotiation. J. Appl. Econometr. 22 (7) 1267–1294. Jamasb, T., Mota, R., Newbery, D., Pollitt, M., 2004. Electricity Sector Reform in Developing Countries: A survey of Empirical Evidence on Determinants and Performance. Cambridge Working Papers in Economics, CWPE 0439, Department of Applied Economics. Kirkpatrick, C., Parker, D., Zhang, Y.F., 2005. Competition, regulation and privatization of electricity generation in developing countries: does the sequencing of the reforms matter? Q. Rev. Econ. Financ. 45, 358–379. Kirkpatrick, C., Parker, D., Zhang, Y.F., 2008. Electricity sector reform in developing countries: an econometric assessment of the effects of privatization, competition and regulation. J. Regul. Econ. 33, 159– 178. Levy, B., Spiller, P.T., 1994. The institutional foundations of regulatory commitment: a comparative analysis of telecommunication regulation. J. Law Econ. Org. 10 (2) 201–246. Spiller, P.T., 1996. A positive political theory of regulatory instruments: contracts, administrative law or regulatory specificity? South. Calif. Law Rev. 69, 477–515. Wallsten, S., Clarke, G., Haggarty, L., Kaneshiro, R., Noll, R., Shirley, M., Xu, L.C., 2004. New Tools for Studying Network Industry Reforms in Developing Countries: the Telecommunications and Electricity Regulation Database. World Bank Policy Research Working Paper 3286. World Bank, 2010. In: Foster, V., Briceno-Garmendia, C. (Eds.), Africa’s Infrastructure: A Time for Transformation. World Bank, Washington, DC. Wren-Lewis, L., 2014. Utility regulation in Africa: how relevant is the
The Electricity Journal
British model? Util. Policy, http:// dx.doi.org/10.1016/j.jup. 2014.09.002.
Endnotes: 1. They used the following indicators to measure sector performance: net electricity generation per capita, installed generation per capita, net electricity generation per employee, and generation relative to average capacity. 2. Andres et al. (2008) use a similar approach to assess the impact of private sector participation on infrastructure.
July 2015,
Vol. 28, Issue 6
3. See Levy and Spiller (1994). 4. Gassner and Pushak (2014) also found that political pressures are one of the main barriers to the successful introduction of regulatory agencies in low-income and middle-income countries. 5. Let D be a binary indicator that equals unity if a country has established an IRA and zero otherwise. Also, let Y1t denote the value of certain outcome in period t if the country has an IRA and Y0t if not. Given a set of observable country attributes X, the average effect of the IRA on Yt = E[(Y1t1 Y0t0) jX, D = 1] E[(Y0t1 Y0t0) jX, D = 0].
Following Heckman et al. (1997) and Heckman et al. (1998), we can compute a consistent estimator of the counterfactual by means of a kernelweighted average of outcomes. 6. The kg of oil equivalent is a unit of energy: the amount of energy released by burning one kg of crude oil (as different crude oils have different calorific values, the exact value of the toe is defined by convention). 7. Coal, oil, natural gas, nuclear, and hydroelectric sources. 8. See Table 1 for the list of countries. 9. 1997 is the average establishment date of an IRA in the treated group.
1040-6190/# 2015 Elsevier Inc. All rights reserved., http://dx.doi.org/10.1016/j.tej.2015.06.004
81