Essays on Electricity Market Reforms: 
A Cross-Country Applied Approach 



Erkan Erdogdu 




Girton College 
University of Cambridge 



This thesis is submitted for the degree of 
Doctor of Philosophy (PhD) 


2013  
LL

Declaration 

,GHFODUHDVWKHDXWKRURIWKLVWKHVLVWKHZRUNDQGUHVHDUFKSUHVHQWHGKHUHLVWRWKHEHVWRI
P\NQRZOHGJHDQGEHOLHIRULJLQDOH[FHSWZKHUHGXHDFNQRZOHGJHPHQW LQDFFRUGDQFHZLWK
WKHVWDQGDUGUHIHUHQFLQJSUDFWLFHVLVPDGHLQWKHWH[WRIWKHWKHVLV7KLVWKHVLVLVWKHUHVXOWRI
P\ RZQ ZRUN DQG LQFOXGHV QRWKLQJ ZKLFK LV WKH RXWFRPH RI ZRUN GRQH LQ FROODERUDWLRQ
H[FHSWZKHUHVSHFLILFDOO\LQGLFDWHGLQWKHWH[W7KLVWKHVLVKDVQRWEHHQVXEPLWWHGLQZKROHRU
LQSDUW IRUFRQVLGHUDWLRQIRUDQ\RWKHUGHJUHHRUTXDOLILFDWLRQDW WKLVXQLYHUVLW\RUDQ\RWKHU
LQVWLWXWLRQ ,Q DFFRUGDQFH ZLWK WKH UHJXODWLRQV RI WKH -XGJH %XVLQHVV 6FKRRO WKH WKHVLV
FRQWDLQVQRPRUHWKDQZRUGVRIWH[W



(UNDQ(UGRJGX
 
iii 
 
Acknowledgements 
 
I would like to take this opportunity to thank various people and institutions without whose 
extremely generous support this thesis would have hardly been prepared. First of all, I would 
like to thank my supervisors, Dr. Michael Pollitt and Prof. David Newbery, for all their 
helpful comments and suggestions. Special thanks are due to Dr. Paul Kattuman, who 
provided me with the econometrics background that I have extensively exploited to prepare 
the thesis. I would like to extend my sincere thanks and appreciation to Mr. Anthony Haynes, 
who helped me to improve the language of the thesis.  
 
I am grateful to my examiners, Prof. Stephen Littlechild and Prof. Catharine Waddams, for 
their time spent on my thesis and constructive comments that made the thesis much better. 
 
My research has benefited greatly throughout the years from comments and suggestions by 
many participants of the conferences or seminars such as 30th USAEE/IAEE North American 
Conference (9-12 October 2011, Capital Hilton Hotel, Washington, DC), various EPRG 
Energy & Environment seminars, the Cambridge Judge Business School doctoral conferences 
and PhD research lunches. Their contribution is sincerely appreciated. 
 
I am also grateful to the Cambridge Overseas Trust for awarding me a full scholarship that 
has financed my doctoral studies. Besides, I really appreciate the contribution of Judge 
Business School (University of Cambridge), Latin-American Energy Organization (OLADE), 
E-Control (Austria), International Energy Agency (OECD/IEA), and World Bank into this 
thesis by providing either essential data or finance for conference attendance. 
 
iv 
 
The first paper of the thesis is published in Energy Policy (ISSN: 0301-4215, Volume 39, 
Issue 3, March 2011, pages 1080-1092, doi:10.1016/j.enpol.2010.11.023). The second and 
third papers are published in EPRG Working Paper Series (No: 1212 & 1216) and 
Cambridge Working Papers in Economics (No: 1227 & 1232). The third paper was also 
granted one of the top 4 best paper awards (out of 160 papers) at 30th USAEE/IAEE North 
American Conference (9-12 October 2011, Capital Hilton Hotel, Washington, DC). In March 
2012, on the occasion of the 10 year anniversary of the Austrian energy regulator (E-Control), 
the thesis as a whole was granted a research award by E-Control in recognition for its 
contribution to research in the area of electricity market liberalization. The second and third 
papers were submitted to Energy Economics (ISSN: 0140-9883) and The Energy Journal 
(ISSN: 0195-6574), respectively; and we have received “revise and resubmit” decisions from 
the Editors of both journals at the time when the thesis is submitted. The second paper has 
also been accepted for presentation at the 2013 Conference of the International Society for 
New Institutional Economics (ISNIE) in Florence/Italy on June 20-22, 2013. I would like to 
express my gratitude to all anonymous referees who took part in the review processes of the 
papers and the thesis in these incidents. 
 
Last but not least, I owe many thanks to my loving wife who has been by my side through all 
my ups and downs, who has supported me and helped me, during the entire course of my 
PhD, which has been a very challenging journey. Without her, things would not be the same. 
 
  
v 
 
Summary 
 
In the last two decades, more than half of the countries in the world have introduced a reform 
process in their power industries and billions of dollars have been spent on liberalizing 
electricity markets around the world. This thesis presents a doctoral research concerned with 
the cross-country empirical analysis of the electricity market reforms. The thesis is in three-
paper format; that is, we present three independent but related stand-alone papers.  
 
The first paper focuses on the impact of power market reforms on electricity price-cost 
margins and industrial/residential price ratios. It investigates this issue by looking at the 
impact of the electricity industry reforms on residential and industrial electricity price-cost 
margins and their effect on industrial/residential price ratios. Using panel data from 63 
developed and developing countries covering the period 1982–2009, empirical models are 
developed and analysed. The results suggest that each individual reform step has different 
impact on price-cost margins and industrial/residential price ratios for each consumer and 
country group. That is to say, our findings imply that similar reform steps may have different 
impacts in different countries, which supports the idea that reform prescription for a specific 
country cannot easily be transferred to another one with similar success. 
 
The second paper explores whether the question of why some countries are able to implement 
more extensive reforms is closely related to the question of why some countries have better 
institutions than others. It analyses this question by using an empirical econometric model 
based on Poisson regression with cross-section data covering 51 states in US, 13 provinces in 
Canada and 51 other countries. The study concludes that both the background of the 
chairperson and the minister/governor and institutional endowments of a country are 
important determinants of how far reforms have gone in a country. 
vi 
 
 
Considering the fact that ideological considerations, political composition of governments and 
educational/professional background of leaders have played and will play a crucial role 
throughout the reform process; the third paper attempts to discover the impact of political 
economic variables on the liberalization process in electricity markets. It develops and 
analyses empirical models using panel data from 55 developed and developing countries 
covering the period 1975–2010. The results suggest that a portion of the differences in the 
reform experiences of reforming countries in the past three decades can be explained by 
differences in the political structure, in the ideology of the government and in the professional 
and educational backgrounds of the political leaders. 
 
  
vii 
 
Contents 
 
Declaration ................................................................................................................................ ii 
Acknowledgements .................................................................................................................. iii 
Summary ................................................................................................................................... v 
Contents ................................................................................................................................... vii 
List of Figures ......................................................................................................................... xii 
List of Tables .......................................................................................................................... xiii 
 
1 Introduction .......................................................................................................................... 1 
1.1 Motivations for the reform .............................................................................................. 2 
1.2 The reform and its measurement .................................................................................... 4 
1.3 The earlier literature ....................................................................................................... 6 
1.4 Connections among the papers ..................................................................................... 13 
1.5 Contribution of the papers into the existing literature .................................................. 15 
 
2 The Reforms So Far ........................................................................................................... 22 
2.1 Countries highly progressed in reforms ........................................................................ 22 
2.1.1 The United Kingdom ........................................................................................... 22 
2.1.2 Chile and Argentina ............................................................................................ 23 
2.1.3 Nordic countries .................................................................................................. 26 
2.2 Countries with limited reforms ..................................................................................... 28 
2.2.1 North America ..................................................................................................... 28 
2.2.2 Australia and New Zealand ................................................................................. 32 
2.2.3 European Union ................................................................................................... 34 
2.2.4 Turkey ................................................................................................................. 40 
viii 
 
2.2.5 Japan, South Korea and Philippines .................................................................... 41 
2.2.6 Brazil, Bolivia and Peru ...................................................................................... 43 
2.3 Countries with suspended or partial reforms ................................................................ 46 
2.3.1 China and India ................................................................................................... 46 
2.3.2 Eurasia ................................................................................................................. 48 
2.3.3 Africa  ................................................................................................................. 50 
 
3 Paper One ........................................................................................................................... 52 
3.1 Introduction ................................................................................................................... 52 
3.2 Literature review ........................................................................................................... 55 
3.3 Hypothesis development ............................................................................................... 59 
3.3.1 Hypothesis on price-cost margins ....................................................................... 59 
3.3.2 Hypothesis on relative price of electricity for industry and households ............. 64 
3.3.3 Control variables ................................................................................................. 66 
3.4 Methodology ................................................................................................................. 67 
3.5 Overview of data ........................................................................................................... 70 
3.6 Empirical analysis ......................................................................................................... 76 
3.7 Discussion of the results ............................................................................................... 80 
3.7.1 Impact of reforms on price-cost margins for industry ......................................... 81 
3.7.2 Impact of reforms on price-cost margins for households .................................... 82 
3.7.3 Impact of reforms on industrial/residential price ratio ........................................ 83 
3.7.4 Results from control variables ............................................................................. 85 
3.8 Conclusion  ................................................................................................................... 88 
 
4 Paper Two ........................................................................................................................... 92 
4.1 Introduction ................................................................................................................... 92 
4.2 Conceptual framework and literature review ............................................................... 96 
ix 
 
4.3 Hypothesis development ............................................................................................. 102 
4.3.1 Hypothesis on the background of key policy makers and implementers .......... 103 
4.3.2 Hypothesis on the impact of democracy on the reforms ................................... 108 
4.3.3 Hypothesis on the impact of corruption on the reforms .................................... 109 
4.3.4 Control variables ............................................................................................... 110 
4.4 Methodology ............................................................................................................... 111 
4.5 Overview of data ......................................................................................................... 115 
4.6 Empirical analysis ....................................................................................................... 126 
4.7 Discussion of the results ............................................................................................. 133 
4.7.1 Impact of the background of key persons on the reforms ................................. 133 
4.7.2 Impact of democracy on the reforms ................................................................. 137 
4.7.3 Impact of corruption on the reforms ................................................................. 138 
4.7.4 Results from control variables ........................................................................... 139 
4.7.5 Results without the states in US and provinces in Canada ............................... 140 
4.7.6 Impact of initial conditions and initial price levels on the reforms ................... 146 
4.8 Conclusion  ................................................................................................................. 151 
 
5 Paper Three ...................................................................................................................... 153 
5.1 Introduction ................................................................................................................. 153 
5.2 Literature review ......................................................................................................... 155 
5.3 Hypothesis development ............................................................................................. 160 
5.3.1 Hypothesis on industry sector ........................................................................... 161 
5.3.2 Hypothesis on foreign financial aid .................................................................. 164 
5.3.3 Hypothesis on government ideology ................................................................. 165 
5.3.4 Hypothesis on the background of chief executive ............................................ 166 
5.3.5 Control variables ............................................................................................... 168 
5.4 Methodology ............................................................................................................... 171 
x 
 
5.5 Overview of data ......................................................................................................... 174 
5.6 Empirical analysis ....................................................................................................... 187 
5.7 Discussion of the results ............................................................................................. 193 
5.7.1 Impact of industry sector on the reforms .......................................................... 193 
5.7.2 Impact of foreign financial aid on the reforms .................................................. 195 
5.7.3 Impact of government ideology on the reforms ................................................ 196 
5.7.4 Impact of the background of chief executive on the reforms ............................ 198 
5.7.5 Results from control variables ........................................................................... 200 
5.7.6 Impact of initial conditions, initial price levels and California crisis on the 
reforms .............................................................................................................. 203 
5.8 Conclusion  ................................................................................................................. 209 
 
6 Conclusion ........................................................................................................................ 212 
6.1 Have research questions been answered? ................................................................... 213 
6.2 Policy repercussions of the results .............................................................................. 215 
6.3 Policy suggestions ...................................................................................................... 219 
6.4 Limitations of the research ......................................................................................... 220 
6.5 What still needs to be done ......................................................................................... 225 
 
Bibliography .......................................................................................................................... 228 
Appendices ............................................................................................................................ 248 
Appendix I-1: Price-fuel cost margins when electricity reforms started ........................... 248 
Appendix I-2: Change in price-fuel cost margins for industry .......................................... 250 
Appendix I-3: Change in price-fuel cost margins for households ..................................... 251 
Appendix I-4: Estimation results for Paper One ................................................................ 252 
Appendix II-1: Summary of previous econometric studies adopting a NIE approach ...... 257 
xi 
 
Appendix II-2: Summary of previous econometric studies based on NIE by their 
focus ......................................................................................................... 277 
Appendix II-3: Classification of variables employed in previous econometric studies 
based on NIE by what they measure ........................................................ 281 
Appendix III-1: Summary of previous applied econometric studies adopting a 
political economy approach ..................................................................... 282 
Appendix III-2: Summary of previous econometric studies based on political 
economy by their focus ............................................................................ 296 
Appendix III-3: Estimation results for Paper Three .......................................................... 300 
Appendix III-4: Correlation among explanatory and control variables ............................. 308 
 
  
xii 
 
List of Figures 
 
Figure 1.1 Reform scores of the countries analysed in the thesis ............................................ 19 
Figure 3.1 Change in price-fuel cost margins for industry ....................................................... 63 
Figure 3.2 Change in price-fuel cost margins for households .................................................. 64 
Figure 3.3 Electricity end user price - fuel cost margins in 1987 and 2007 ............................. 74 
Figure 4.1 The relationship between institutional structure and reform process...................... 94 
Figure 4.2 Poisson distribution for different  values .......................................................... 113 
Figure 4.3 Histogram of reform score variable ...................................................................... 118 
Figure 4.4 Electricity reform in US states as of 2011 ............................................................ 118 
Figure 4.5 Educational backgrounds of chairpersons and ministers/governors ..................... 121 
Figure 4.6 Education level of chairpersons and ministers/governors ..................................... 121 
Figure 4.7 Length of term of chairpersons and ministers/governors ...................................... 122 
Figure 4.8 Civil liberties and political rights scores for the sample countries in 2011 .......... 123 
Figure 4.9 Corruption perceptions index for 2010 ................................................................. 124 
Figure 4.10 Scatter plot of investment freedom index and property rights index .................. 125 
Figure 5.1 Electricity market closeness index in OECD countries (1989, 2007) ................... 181 
Figure 5.2 Electricity market closeness index in countries where 
EBRD operates (1989, 2007) ................................................................................ 182 
Figure 5.3 Histogram of industry value added variable ......................................................... 184 
Figure 5.4 Total development assistance and aid received, 1990-2007 ................................. 185 
Figure 6.1 Reform score frequencies of reported and non-reported countries 
in the second paper ............................................................................................... 222 
 
  
λ
xiii 
 
List of Tables 
 
Table 1.1 Countries analysed in the PhD thesis ....................................................................... 18 
Table 3.4 Descriptive statistics of the variables ....................................................................... 75 
Table 3.5 Summary of estimation results ................................................................................. 78 
Table 4.1 Descriptive statistics of the variables ..................................................................... 116 
Table 4.2 Poisson regression estimation results for the full sample ....................................... 128 
Table 4.3 Poisson regression estimation results for the full sample as 
Incident Rate Ratios (IRR) ..................................................................................... 130 
Table 4.4 Poisson regression estimation results without the states in US and provinces 
in Canada as IRR (Micro Model) ........................................................................... 142 
Table 4.5 Poisson regression estimation results without the states in US and provinces 
in Canada as IRR .................................................................................................... 143 
Table 4.6 Results with and without the states in US and provinces in Canada as IRR .......... 145 
Table 4.7 Poisson regression estimation results for the full sample including initial 
conditions and initial price levels ........................................................................... 148 
Table 5.1 Descriptive statistics of the variables ..................................................................... 175 
Table 5.2 Composition of the electricity market closeness index .......................................... 178 
Table 5.3 Country groups based on electricity market closeness index in 2000 .................... 183 
Table 5.4 Estimation results for the models testing Hypothesis 1 (sub-indicators) ............... 188 
Table 5.5 Estimation results for the models testing Hypothesis 1 (overall indicator) ........... 189 
Table 5.6 Estimation results for the models testing Hypothesis 2 (sub-indicators) ............... 190 
Table 5.7 Estimation results for the models testing Hypothesis 2 (overall indicator) ........... 190 
Table 5.8 Estimation results for the models testing Hypotheses 3 and 4 (sub-indicators) ..... 191 
Table 5.9 Estimation results for the models testing Hypotheses 3 and 4 (overall indicator) . 192 
Table 5.10 Estimation results for the limited model testing Hypotheses 1 and 2 .................. 206 
Table 5.11 Estimation results for the limited model testing Hypotheses 3 and 4 .................. 207 
 
1 
 
 
 
 
 
 
Chapter 1 
 
1 Introduction 
 
This thesis presents the doctoral research concerned with the cross-country analysis of the 
electricity market reforms that have been put into practice in more than half of the countries in 
the world. The research is inspired by my own experiences working for the energy market 
regulator (EMRA1) in my home country before commencing my doctoral studies in 
Cambridge. The thesis is in three-paper format. We present three independent but related 
stand-alone papers which make up Chapters 3, 4 and 5 of the thesis. This introductory chapter 
has five sections. First of all, we discuss motivations for the electricity market reform. 
Second, we briefly explain what reform is and how reform progress is measured in the papers. 
Third; although each paper has its own specific literature review section, here we present a 
discussion on the link between general theme of the thesis and the earlier relevant literature, 
especially theory of regulation and public choice theory. Then, we present the connections 
among the papers. Finally, we mention the expected contribution of the papers into the 
existing literature.  
 
                                                          
1 EMRA: Energy Market Regulatory Authority, Republic of Turkey 
2 
 
1.1 Motivations for the reform 
 
Starting from the early 1980s, a number of political, financial and technical factors converged 
and started to undermine the logic that electricity industry should be handled via a vertically 
integrated (and usually state-owned) monopoly (Gratwick and Eberhard, 2008). Among these 
factors, there were ideological reasons, development of gas-fired combined cycle gas 
turbines2 (CCGTs), improvement in information and communication technologies, questions 
about the efficiency of vertically integrated utilities (whether publicly owned or private and 
regulated by authorities) and poor performance of existing utilities, especially in developing 
countries. 
 
Electricity reform in developed and developing countries has been a fundamentally different 
undertaking in terms of motivations, sector conditions, and institutional context. In developed 
countries, the main targets of the reform has been the improvement in the economic efficiency 
of the sector, encouragement of inter-regional (or cross border) trade, transferring investment 
risks to the private sector and offering customer choice. Other subsidiary motives include the 
demonstration effects of the pioneering reforms of the power sectors in the UK and Norway in 
the early 1990s, the EU directives, the rapid changes in technology especially in the 
generation of electricity that made new industrial structures possible, the desire to overcome 
what might be called sub-optimal regulation, and the policy objective to eliminate tendency to 
over-invest (so called “gold-plating”).  
 
In developing countries, motivation for reform includes the poor performance of state-run 
electricity operators in terms of high costs, inadequate expansion of access to electricity 
                                                          
2 The advent of highly efficient CCGTs made it possible to build small units in relatively short time with little 
risk, which eliminated the significant barriers that had previously existed to entry in power generation. 
3 
 
services and unreliable supply; the inability of the public sector to meet the investment and 
maintenance costs of the electricity industry associated with the increasing demands for 
power resulting from economic development; the need to remove the burden of price 
subsidies (so as to release resources for other areas of public expenditure), low service 
quality, low collection rates, high network losses; the desire to raise immediate revenue for 
the government through the sale of state assets; the policy to attract foreign direct investment 
in the power sector; and encouragement of reform by international financial organizations and 
donor agencies such as the IMF and World Bank (Zhang et al., 2008). Besant-Jones (2006) 
argues that power market reform in developing countries should be assessed against three 
outcomes that reflect the drivers for reform. These outcomes are better service quality for 
electricity consumers to support economic growth and welfare, improvement in government’s 
fiscal position, and more affordable access to electricity for the poor. For him, the most 
important lesson from reforming power markets in developing countries is that “cookbook” 
solutions for reforming their power markets are ruled out by the extensive range of economic 
and institutional endowments of these countries. This lesson emphasizes the importance of 
country and power market initial conditions for reform that determine the initial (and often 
subsequent) scope and composition of the reform. Countries with better endowments should 
be able to achieve more ambitious outcomes from power market reform than countries with 
lesser endowments. Reforms based on substantial market restructuring for large middle-
income countries, for example, may be infeasible for small low-income countries. He lists 
strategic elements for implementing power market reform as: (i) power market reform has 
many dimensions, (ii) power market reform must be adapted to starting conditions, (iii) power 
market reform is a process - not an event, (iv) power market reform is an opportunity to help 
the poor. 
 
4 
 
1.2 The reform and its measurement 
 
Since the 1980s, vast amounts of financial resources and effort have been spent on reforming 
electricity industries in both developed and developing countries. Reforms were pioneered by 
Chile (in 1982), the UK (in 1989), and Norway (in 1991); and have spread all over the world. 
In almost all reforming countries, electricity reform has been a part of wider policies towards 
a liberal market economy. In the process of reform, the former vertically integrated electricity 
utilities were restructured and unbundled, and competition has been introduced into 
generation, wholesale and retail segments of the industry. Transmission and distribution 
businesses have usually remained as regional or national monopolies but they have been put 
under regulation by an independent regulator. Other common elements of the reforms include 
the introduction of wholesale and spot power markets, the establishment of impartial market 
and system operators, the removal of restrictions on third party access to networks and, in 
some cases, privatization. As indicated by World Bank (2004); competition, unbundling, 
private participation, and other reform elements are not ends in themselves, but rather 
intended to contribute to the achievement of broad goals for poverty reduction, economic 
growth and environmental sustainability. Given the differing points at which they find 
themselves, countries adopt power sector reform strategies that reflect the strategic priorities 
for their electricity industry, and the immediate country conditions that influence the 
suitability of particular approaches. The course that power sector reform takes and the speed 
at which reforms can be implemented vary from country to country, but reform is a 
continuous and evolving process. 
 
Many aspects of electricity reforms are not readily quantifiable in physical or monetary units 
(Jamasb et al., 2006). That is to say, objective comparisons across countries concerning the 
success or failure of the reforms are inherently difficult in any study and our analysis is not an 
5 
 
exception. Since it is very difficult, if not impossible, to develop a true measure of success or 
failure of the reform process; in this research, we focus on the reform progress rather than 
reform success or failure. Besides, reform success or failure may be country specific and 
therefore it is not easy to develop a measure of it applicable to all countries. On the other 
hand, reform progress is a variable standardizable through countries and therefore a suitable 
indicator for a cross-country analysis. However, the measurement of reform progress also 
requires a great deal of effort as the main steps of electricity reform are usually established 
progressively and have a qualitative dimension.  
 
Since we do not have a perfect tool or variable to measure reform progress, we prefer to use 
different measures in each paper. In the first paper, we use dummy variables to account for 
various reform steps and their cross-products. In the second paper, we construct an electricity 
market reform score variable to represent overall reform progress. The variable takes the 
values from 0 to 8; depending on how many of the following reform steps have been taken in 
a specific country and specific year: (1) introduction of independent power producers, (2) 
corporatization of state-owned enterprises, (3) law for electricity sector liberalization, (4) 
introduction of unbundling, (5) establishment of electricity market regulator, (6) introduction 
of privatization, (7) establishment of wholesale electricity market, and (8) choice of supplier. 
In the third paper, we do not construct a variable ourselves, but use electricity market reform 
indicators constructed by OECD and EBRD to represent reform progress. OECD provides 
data on the reform progress in its member countries and EBRD provides similar data for 
additional countries where it operates. Each method measures reform progress from a 
different perspective but, apart from this, they are not directly related. Although these three 
different approaches to measurement of reform progress in three different papers seem a 
practical and reasonable representation of reform dimension, we cannot argue that we fully 
reflect all characteristics of the various reform processes in our papers.  
6 
 
 
1.3 The earlier literature 
 
Each paper of the thesis has its own specific literature review section. Therefore, in this 
section, we do not present a review of the literature directly related to the papers. Instead, we 
present a discussion on the earlier relevant literature, especially theory of regulation and 
public choice theory. We specifically focus on theory of regulation and public choice theory 
since “electricity market reforms” may be regarded as a form of regulation (or a change in 
regulation) implemented as a result of a political decision. As discussed below, theory of 
regulation, public choice theory and previous empirical investigations based on these theories 
indicate that economic performance (including performance of prices) is affected by liberal 
reforms. In the first paper, we try to find out whether this relationship holds true when we take 
into account the fuel costs. Besides, within the framework suggested by economic theory of 
regulation, it is expected that the strength of various interest groups determines the outcome 
of electricity market reforms (including relative electricity prices). Based on this expectation, 
the first paper also investigates the impact of the reforms on the relative price of electricity for 
two main interest groups, namely industrial and residential consumers. Moreover, in line with 
the basic arguments of the economic theory of regulation, the third paper explores the impact 
of the industrial consumers (as one of the most organized and powerful interest group in 
electricity sector) and the ideological orientation of politicians on the reform progress. The 
further details of the links between the earlier literature summarized below and the papers 
(especially the first and the third papers) are provided in the hypotheses development sections 
of each paper. 
 
The science of politics and the science of economics have traditionally been separated by the 
types of questions they ask, the assumptions they make about individual motivation, and the 
7 
 
methodologies they employ. Political science has studied the behaviour of homo politicus in 
the public domain; economics has studied homo economicus in the market. Political science 
has often supposed that homo politicus pursues the public interest. Economics has assumed 
that all economic agents pursue their private interests. “Public choice” has been characterised 
as the application of economics to political science. It applies the methodology of economics 
to the study of politics. It is interdisciplinary insofar as it employs the analytic tools of 
economics and chooses as its subject matter the identical fields as political science does 
(Kirchner, 2007). 
 
Public choice theory resists modelling public policymakers as disinterested maximizers of 
society’s welfare (Buchanan and Tullock, 1962). Public choice assumes that all human actors, 
in or out of government, pursue similar objectives (utility maximization) and employ the same 
rational-choice calculus to select the alternative that yields the greatest personal benefit net of 
cost (Buchanan, 1972). This assumption of universal self-interest coupled with the logic of 
collective action (Olson, 1965) implies that the individuals responsible for formulating and 
executing public policies toward business will have powerful incentives, not selflessly to 
promote the public interest, but to enhance their own wellbeing by catering to the demands of 
politically well-organized special-interest groups. 
 
The central economic argument for regulation of an industry is that it is characterized by 
‘natural monopoly’ (Viscusi et al., 2005). The traditional notion of natural monopoly is based 
on the existence of economies of scale throughout the relevant range of production on the 
market. Such scale economies were typically taken to mean that competition might lead to 
greatly inefficient and even wildly fluctuating, unstable prices so that government 
intervention of some sort was necessary. However, economic research has demonstrated that 
even if an industry is characterized by natural monopoly in the sense that there is no room for 
8 
 
competition within a market, under some circumstances competition for the market may 
succeed in allocating resources quite efficiently in the absence of regulation (see Demsetz 
(1968), Stigler (1968), and Posner (1972)). From a political view, perhaps the most significant 
feature of regulation is that it redistributes income, creating winners and losers, thereby 
shaping interest groups and coalitions. Thus, it is not surprising that there is a large literature 
on regulation addressing reasons for regulation far broader than natural monopoly. Joskow 
and Noll (1981) and Noll and Owen (1983) provide excellent discussions on this topic. To 
sum up, economics research on regulation has three main themes. The first and oldest deals 
with market failures and the corrective actions that government can undertake to ameliorate 
them. The second examines the effects of regulatory policies, and asks whether government 
intervention is efficient or more efficient than doing nothing. The third investigates the 
political causes of regulatory policy (Noll, 1989). 
 
Within this context, two general theories of regulation have been developed to explain the 
regulation of markets. The first of these, called the public interest theory, argues that the 
reason for regulation is to avoid market failure when an industry is naturally monopolistic. 
This theory presumes that the goal of regulation is to simulate competition where it is 
allegedly impossible because of economies of scale, and that this intention will, in fact, be 
carried out by the regulatory process. The second approach, promoted by George Stigler and 
Sam Peltzman, is the economic theory of regulation. The essence of this approach is that 
regulators and politicians, like homo economicus, weigh the benefits and costs of various 
courses of action in a political framework where the attainment of a voting majority 
determines success. So, in some sense, the economic theory of regulation is the application of 
public choice approach to domain of regulation. In its broadest interpretation, this approach 
emphasizes a balancing of interest group strengths and weaknesses at the margin, with the 
9 
 
outcome determined by the stake that the various groups have in it and the efficiency by 
which they can influence the regulatory process (Wenders, 1988). 
 
Conventional wisdom, put forward by Pigou’s (1920) classic work on welfare, holds that state 
action is necessary to reduce the impact of externalities, with taxes used to reduce harmful 
effects and subsidies to encourage beneficial ones. Ronald Coase, in a classic article published 
in 1960, challenged the conventional wisdom in economics regarding externalities, taxes, and 
subsidies. Coase argues that the existence of an external effect associated with a given activity 
does not inevitably require government intervention in the form of taxes and subsidies. 
Pareto-optimal resolutions of externality situations can be and often are worked out between 
the affected parties without the help of the government (Coase, 1960). The Coase theorem 
may be summarized as “in the absence of transaction and bargaining costs, affected parties 
to an externality will agree on an allocation of resources that is both Pareto optimal and 
independent of any prior assignment of property rights”. This implies that when transaction 
costs are zero, all collective choices that promise a Pareto improvement are made; no public 
good with benefits greater than costs goes unprovided; no Pareto-relevant external effect is 
left unaltered; no firm that would make a profit fails to get started, no matter how large the 
number of participants needed to bring about the optimal collective choice (Mueller, 2003). 
 
Stigler (1971) modelled regulation largely as a struggle between producers and consumers for 
access to the rents associated with conditions of natural monopoly. He challenged the two 
most popular notions of economic regulation, namely that (i) regulation is for public benefit, 
and (ii) rationality cannot be used to understand politics. Given that the members of the 
regulated industry normally would be better informed about the regulatory process, have 
greater financial stakes in regulatory outcomes, and, owing to their smaller numbers and more 
cohesive objectives, be better organized and, hence, more effective in bringing influence to 
10 
 
bear on the regulatory agency, regulators would tend to favour their interests over those of 
consumers (Shughart, 2004). The essence of the ‘capture’ theory of regulation is that 
‘consumers are the least organized and therefore typically the least effective interest group; 
the long-run consumer interest in particular has no lobby’ (Posner, 1969). 
 
Stigler and Friedland (1962) reported cross-sectional regression results for electricity prices 
across U.S. states for 1922. Their results suggested that state regulation of electricity prices 
had no detectable effect on the level of these prices; prices in regulated states were 
statistically the same as prices in unregulated states, ceteris paribus. Besides, their findings 
suggested that the regulatory bodies could not reduce the amount of price discrimination 
between domestic users and industrial users. Therefore, they concluded that regulation is 
ineffective. According to Stigler and Friedland, the reasons for regulation being ineffective 
are: (1) an individual utility system does not possess any significant amount of long-run 
monopoly power, and (2) regulatory bodies are incapable of forcing the utility to operate at a 
specified combination of output, prices, and costs. These results later provided the basis for 
the Stigler-Peltzman theory of economic regulation, which predicts that regulatory decisions 
will tend to be determined by the relative strengths of various opposing interest groups, such 
as buyers, sellers, competitors, legislators, bureaucrats and so on without necessarily being 
able to predict, a priori, which particular group (if any) will dominate a given decision 
(Upadhyaya et al., 1997). 
 
In an important extension of Stigler’s argument, Peltzman (1976) supplies a more general 
framework for thinking about the problem. In that more general theory, the regulators 
themselves are portrayed as rational, self-interested actors whose objective is to maximize 
their own political support. Where they hold elective office, ‘political support’ can be defined 
in terms of votes, campaign contributions, or both, in which case regulators are assumed to be 
11 
 
motivated by the goal of maximizing their probability of re-election. Where they hold 
appointive office, regulators strive to maximize their probability of reappointment or some 
other index of job security. An even more universal behavioural assumption is utility (wealth) 
maximization, which includes the regulator’s salary and perquisites of public office as well as 
income received from post-government employment.  
 
In Peltzman’s framework, the regulator’s pursuit of self-interest is not unconstrained. The 
regulator selects the price the regulated firm is permitted to charge. This price can be set at the 
competitive level, in which case the regulated firm earns a normal profit and consumers enjoy 
all of the gains associated with regulation. The price can also be set at the monopoly profit-
maximizing level, in which case producers are regulation’s sole beneficiaries. In general, 
however, the politically self-interested regulator must weigh the demands of both groups. 
While an increase in price (and profit) means greater political support from the regulated 
firm(s), it also invites greater opposition from consumers. Lower prices invoke the opposite 
reactions. If the political returns to higher profit or lower price are diminishing at the margin, 
neither group will get all that it wants from regulation. From the regulator’s point of view, the 
optimal price will lie somewhere between the extremes of competition and monopoly. Where 
the balance is struck in any particular case depends on the configurations of the costs and 
benefits of bringing political influence to bear on the regulatory process facing the groups 
having stakes in the outcome (Shughart, 2004).  
 
Becker (1983) presented a theory of competition among pressure groups for political 
influence that built on Peltzman’s analysis. Political equilibrium was shown to depend on the 
efficiency of each group in producing pressure, the effect of additional pressure on the 
influence of the group, the number of persons in different groups, and the deadweight cost of 
taxes and subsidies. 
12 
 
 
De Alessi (1995) also argued that the demand and supply of regulation are driven by 
considerations other than alleviating market limitations. He stated that the evidence suggests 
that regulation typically is established and operated for the benefit of the industries and 
occupations being regulated and of the regulatory agencies. Entrepreneurs have incentive to 
demand state support to restrict entry, enforce collusion, and pursue other wealth-increasing 
activities. Correspondingly, politicians and other government employees have incentive to 
favour organized pressure groups, extort tribute, and expand their activities. Evidence from a 
variety of industries indicates that regulation results in higher prices, lower output and greater 
wealth for the firm’s owners. On the other hand, De Alessi (2001) maintained that although 
those to be regulated may be successful in capturing the regulatory mechanism, it does not 
follow that they reap the full monopoly rents. Among other reasons, transfer of control from 
the market to the political arena admits other rent seekers to the decision-making process. As 
a result, regulated firms typically are subject to a profit constraint, attenuating owners’ 
property rights and increasing managers’ discretionary authority. 
 
As we summarized above, the theory of public choice and economic theory of regulation 
underline that individuals, whether politicians, regulators or voters, will make political 
decisions based on their own self-interest because it is votes that count in the political process, 
which often produces an outcome damaging to general economic welfare. One of the reasons 
for this undesirable outcome is that the voter simply finds it costly to determine the relative 
costs and benefits of alternative courses of political action and to make his/her political power 
felt. Thus, the cost of getting information about alternatives and then acting on this 
information plays a large role in political behaviour. This conclusion explains the desire of 
OECD countries to reduce the role of the state in utility sectors in the 1980s and 1990s. 
 
13 
 
1.4 Connections among the papers 
 
As mentioned before, the thesis consists of three independent but related stand-alone papers. 
The papers have three characteristics in common. First of all, all papers focus on electricity 
reforms and analyse a specific feature of it. Second, they have a cross-country macro 
approach; that is, we do not concentrate on a specific country or region. Instead, we adopt a 
global approach in all papers. The papers analyse the global electricity experiment, which 
many other authors have noted, like Pollitt (1997). Finally, the analysis in the papers is 
predominantly empirical. In other words, we do not just point to spectacular reform failures 
(e.g. the California disaster) or try to get general conclusions from some success stories of a 
few reforming countries (e.g. NordPool). We believe that what is really needed is a complete 
study of the impact of reforms within the context of a well-defined model construction. 
 
Apart from the similarities mentioned above, the papers differ from each other in terms of the 
reform dimension analysed, the relevant related literature, methodology, data and, of course, 
the results. The first paper3 focuses on the impact of power market reforms on electricity 
price-cost margins and industrial/residential price ratios. Adopting a New Institutionalist 
Economics (NIE) perspective, the second paper explores the relationship between reform 
progress and institutional quality. The third paper4 has a political economy focus and attempts 
to discover the impact of political economic variables on the liberalization process in 
electricity markets. Since NIE may be regarded as a subset of political economy literature, the 
second and third papers overlap in some areas in terms of the literature. Besides, some 
variables in the second and third papers seem to be similar but, in fact, we use them in the 
                                                          
3 The first paper is published in Energy Policy (ISSN: 0301-4215, Volume 39, Issue 3, March 2011, pages 1080-
1092, doi:10.1016/j.enpol.2010.11.023). 
4 The third paper was granted one of the top 4 best paper awards (out of 160 papers) at 30th USAEE/IAEE North 
American Conference (9-12 October 2011, Capital Hilton Hotel, Washington, DC). 
14 
 
second and third papers for quite different reasons. For instance, NIE emphasizes the concept 
of “path dependency”, which implies that the head of electricity market regulatory agency and 
the minister responsible for energy related issues when the reforms initiated are important for 
the subsequent progress of the reforms. In line with this concept, like most of other similar 
analyses; we assume that experience in the industry, length of term, education level and 
educational background reflect personalities of policy makers and implementers. Of course, 
there may be some other aspects of the personalities. However, due to lack of data, we cannot 
include them into our analysis. So, in the second paper, we use variables related to the 
characteristics of the chairperson of the regulatory agency and energy minister when reforms 
started. In the third paper, however, we focus on the professional and educational background 
of the chief executive (prime minister or president) as the political economy literature 
suggests that the decision to initiate, develop, suspend or cancel a reform process may be 
affected by the personal backgrounds of the political leaders. In short, although we use data 
on key people in the second and third papers, the persons we concentrate on and the reasons 
for concentrating on them are quite different. Moreover, in terms of data and methodology, 
the second and third papers differ completely. Like the first paper, the third paper analyses a 
panel data set using panel data estimation methods (mainly fixed/random effects models). On 
the other hand, the second paper has a cross-section data; and since the dependent variable in 
the analysis is a count variable, which can take on nonnegative integer values only, we use 
Poisson regression methods in the second paper. Naturally, the total number of maximum 
observations for each variable is quite different for each paper. In the first and third papers, 
we have 1,764 and 1,540 maximum observations for each variable, respectively, while the 
total number of maximum observations for the second paper is 115. 
 
As a final point, in the first and third papers, we divide countries into various groups and 
report the results for each country group separately. In the first paper, we have three country 
15 
 
groups, namely (1) developed countries, (2) developing countries in America, and (3) other 
developing countries. In the third paper, we have two country groups, namely (1) OECD 
countries, and (2) non-OECD countries. In both papers, we do not prefer to carry out a 
regression that includes observations from all countries in our datasets and report an overall 
result from it because pooling variables from samples that appear to be different and have 
different coefficients in the separate regressions does not seem to contribute to the papers 
much. In the second paper, we do not classify countries; therefore, we do not naturally report 
results based on country groups.  
 
1.5 Contribution of the papers into the existing literature 
 
Three decades have elapsed since the introduction of the first reforms and there is now a need 
for a detailed evaluation of the reforms as they appear to be costly and there seems to be a 
growing controversy about their benefits. A number of empirical studies have analysed 
electricity market reforms in a variety of ways but they focused generally on a single country 
or a few countries, as reviewed in Mota (2004) and Pollitt (2009). A number of scholars have 
made a contribution in conducting cross-country analysis of the impact of reforms in the 
electricity industries. However, in these studies, analyses were usually conducted using very 
limited data and the number of countries analysed was small. As summarized by Schiantarelli 
(2005), there are some other empirical contributions that use cross-country data to study the 
effect of product market regulation and reform on a country’s macroeconomic performance. 
They provide cross-country evidence on the effect of product market regulation on mark-ups, 
firm dynamics, investment, employment, innovation, productivity and output growth. 
However, in these studies, electricity industry has not been among the industries examined. 
So, there is a research gap in this important area. Using original panel and cross-section data 
16 
 
sets, this thesis attempts to fill this gap to a certain degree. In addition, to the best of our 
knowledge, the present study is the most extensive one in terms of both scale and scope. 
 
Within this context, the expected contributions of the papers are as follows. The first paper 
focuses on the impact of power market reforms on electricity price-cost margins and 
industrial/residential price ratios. It tries to answer: (i) what is the impact of electricity market 
reforms on electricity price-cost margins? (ii) does liberalization cause a change in 
industrial/residential electricity price ratios? (iii) what are the other factors that influence 
electricity price-cost margins and industrial/residential electricity price ratios and how much 
are they influential relative to reform process? The second paper investigates the relationship 
between reform progress and the quality of institutions. It attempts to find out: (i) do 
differences in institutional structures of countries play an important role in explaining how far 
reforms have gone in these countries? (ii) if they do, how do specific institutional 
endowments of a country affect its reform performance? (iii) does the background of the 
chairperson of the regulatory agency when reforms started or were considered or that of the 
governor or minister responsible for energy policy at that time have an impact on reform 
progress? The third paper constitutes an endeavour to reveal the impact of political economic 
structure of a country on the liberalization process in its electricity market. It aims at 
answering the following questions: (i) what is the impact of industrial electricity consumers 
(as an interest group) on the reform progress in power sector? (ii) does foreign influence 
resulting from the dependence on foreign financial support have an influence on the electricity 
market liberalization process? (iii) do the ideology of ruling party and 
professional/educational background of the chief executive (prime minister or president) 
constitute important determinants of the reform progress? If yes, what is the direction of the 
influences originated from these variables? 
 
17 
 
The analyses in the three papers of the thesis cover 93 countries in total. The list of countries 
and their distribution to the samples of the papers are given in Table 1.1 while Figure 1.1 
presents their reform score based on the scale developed in the second paper. The countries in 
our dataset have introduced at least one reform step (e.g. introduction of IPPs into their 
electricity market), so there is no country in our dataset with a reform score of zero. 
Electricity market reform score assigns a score to each country for each year based on the 
reform status of that country. It is an indicator of reform progress, rather than reform success. 
It does not give an idea about reform success and clearly does not present a full picture of the 
situation in each country; however, it still successfully groups countries in terms of reform 
progress. For instance, Chile and Argentina5 get reform score of 7 as there is no retail 
competition for households and small industrial consumers. In the UK, even households are 
able to choose their suppliers so the UK gets the full score of 8. If we assign a full score to 
Chile and Argentina, they will seem to be the same as the UK in terms of reform progress. 
However, this is not the case. That is, reform score measures the distance between actual 
market structure of a country and an ‘ideal’ one where electricity industry is unbundled, 
privately-owned and fully competitive. The score of each country differs since the distance of 
each country from this ‘ideal’ point is different. In short, reform score is a useful indicator for 
our purposes because we focus on reform progress, rather than reform success, in this thesis 
and it provides us with a tool to measure reform progress in each country and each year. 
Besides, this approach to measuring reform impact is common and used also by international 
agencies like OECD (Conway and Nicolett, 2006). 
 
                                                          
5 For a more detailed discussion on the electricity market reforms in Chile and Argentina, please see Section 
2.1.2. 
18 
 
Table 1.1 Countries analysed in the PhD thesis 
In the samples of three papers 
(26 countries) 
Australia*, Austria, Belgium, Canada*, Croatia, Czech 
Republic, Denmark*, Finland*, France*, Germany*, Greece*, 
Ireland, Italy*, Luxembourg, Mexico, Netherlands, New 
Zealand*, Norway*, Poland*, Portugal, Romania*, Russian 
Federation*, Spain*, Turkey*, United Kingdom*, United States* 
In the samples of 1st and 2nd papers only 
(10 countries) 
Argentina*, Brazil*, China*, Colombia, Cyprus, India*, 
Jamaica, Singapore, South Africa*, Thailand 
In the samples of 2nd and 3rd papers only 
(9 countries) 
Albania, Armenia*, Bosnia and Herzegovina, Bulgaria, 
Estonia, Latvia, Lithuania, Macedonia, Serbia 
In the samples of 1st and 3rd papers only 
(7 countries) 
Hungary, Japan*, Kazakhstan*, Korea*, Slovak Republic, 
Sweden*, Switzerland 
In the sample of 1st paper only 
(20 countries) 
Bolivia*, Chile*, Costa Rica, Cuba, Dominican Republic, 
Ecuador, El Salvador, Guatemala, Haiti, Honduras, Indonesia, 
Israel, Nicaragua, Panama, Paraguay, Peru*, Taiwan (Chinese 
Taipei), Trinidad & Tobago, Uruguay, Venezuela 
In the sample of 2nd paper only 
(8 countries) 
Bangladesh, Cambodia, Kenya, Malaysia, Nigeria*, Pakistan, 
Philippines*, Uganda 
In the sample of 3rd paper only 
(13 countries) 
Azerbaijan*, Belarus, Georgia*, Iceland, Kyrgyz Republic, 
Moldova, Mongolia, Montenegro, Slovenia*, Tajikistan*, 
Turkmenistan, Ukraine*, Uzbekistan 
* The reform experience in these countries is summarized in Chapter 2. 
 
 
19 
 
Figure 1.1 Reform scores of the countries analysed in the thesis 
 
 
27
Argentina
25 Armenia
Australia Belgium
Austria Bolivia
Bulgaria Brazil
Canada Chile
Czech Rep. Colombia
Denmark Croatia
El Salvador Dominican Rep.
Finland Georgia
France Guatemala
Germany India
Greece Ireland
Hungary Japan
Italy Kazakhstan 12
Latvia Lithuania Albania
New Zealand Moldova Ecuador
Norway Netherlands Estonia 9
Panama Nicaragua Indonesia Cuba
Poland Peru South Korea 7 Haiti
Portugal Philippines Luxembourg Bangladesh Israel
Romania Russian Fed. Macedonia China 5 Paraguay
Singapore Slovakia Malaysia Jamaica Bosnia and Herz. 4 4 Tajikistan
Spain Slovenia Mongolia Kenya Cambodia Costa Rica Serbia Taiwan
Turkey Sweden Nigeria South Africa Kyrgystan Cyprus Uruguay Trinidad & Tobago
United Kingdom Ukraine Pakistan Thailand Switzerland Honduras Uzbekistan Belarus
United States Iceland Venezuela Uganda Montenegro Mexico Azerbaijan Turkmenistan
8 7 6 5 4 3 2 1
7
5
3
27
Fr
eq
ue
nc
y
Reform Score
1
25
23
21
19
17
15
13
11
9
20 
 
Another important point regarding the score of each country is that we only account for 
whether a step is taken or not, which may not reflect the full extent of the reform step. Market 
choice in Bulgaria and El Salvador or retail competition in France and UK may be different 
but since they took all 8 reform steps, we assign the same score to them. Although the 
countries with a reform score of, for instance, 8 are not the same, they are still within a range 
and this range is different from the range for countries with a reform score of 7. In other 
words, reform scores represent a specific reform progress range and all countries with the 
same reform score are between the upper and lower limits of this range. That is, in terms of 
the number of reform steps taken, the categories are quite homogeneous. For example, in 
Sweden, there is almost no privatization at all but all other reform steps are taken. So, it gets a 
reform score of 7. On the other hand, in Denmark, there are some privatizations (though not 
on a large scale) and it took all other reform steps too; so, Denmark gets a full reform score of 
8. In the Section “6.4 Limitations of the research”, we clearly acknowledge the problem 
associated with measuring the true scale and scope of electricity market reforms in our papers. 
On the other hand, we carried out an extensive literature review but could not identify any 
other study that is clearly more successful in measuring the scale and scope of reform 
progress than ours. 
 
The final point that we would like to mention about the score of each country is about federal 
countries. Federal countries (Australia, Canada, India etc.) are problematic in terms of data 
collection. In our papers, our level of analysis is the country. That is, the main units we get 
our data are countries, not individuals, cities, states, regions and so on. However, in some 
federal countries, individual states may be free to decide whether to introduce a reform 
program and therefore reform progresses in different parts of a federal country may be quite 
different. In the second paper, we included individual states of US and provinces in Canada as 
separate observations. However, analysing individual states and provinces together with 
21 
 
countries also create some other practical and econometric problems. To avoid these 
problems, in the second paper, we also provide results without states in US and provinces in 
Canada. In the first and third papers, we regard the state with the highest level of reform 
progress as the representative unit for the whole country. That is, we assign a reform score of 
8 to a federal country if any of its states has taken all reform steps even if some states have 
not taken any step at all. We do not prefer to assign the score of the lowest scoring state to the 
whole country as all federal countries would get a reform score of 0 if we did so. We also do 
not prefer to determine a “median state” and assign its score to the whole country because it is 
almost impossible to agree on the criteria to determine such “median states” and any effort to 
specify such criteria will be random and cannot be defended theoretically. Still, we 
acknowledge that our approach is not ideal and the results should be evaluated taking into 
account this fact. 
 
The thesis is structured as follows. Chapter 2 briefly summarizes the reform experience in 37 
countries that have novel characteristics in terms of the reform process. Chapters 3, 4 and 5 
present the first, second and third papers, respectively. Final section concludes with policy 
implications of the findings of the three papers. 
 
  
22 
 
 
 
 
 
 
Chapter 2 
 
2 The Reforms So Far 
 
In most of the European countries, US, Canada, Australia and some countries in Latin 
America, power sector reforms are already highly developed. On the other hand, countries in 
Africa and the Middle East have been late in implementing reforms and reforms have been 
gradually taking effect in Eastern Europe and Asia. This chapter briefly summarizes the 
reform experience in 37 countries analysed in the three papers of the thesis. We start from 
countries that have reformed most and then mention those with less reform progress. 
 
2.1 Countries highly progressed in reforms 
 
2.1.1 The United Kingdom 
 
British electricity market reform has been generally regarded as the example that other 
countries should follow. Consultants, encouraged by the World Bank and other international 
financial institutions, have recommended the adoption of the “British model” in countries 
with as diverse needs as India, Ukraine and Brazil; while the British model was clearly the 
inspiration for the European Commission’s directives. In the UK, vertically integrated state 
owned power utility (CEGB) was restructured in 1990 to separate out transmission (as the 
23 
 
National Grid Company) and three generation companies: National Power, PowerGen, and 
Nuclear Electric. All except Nuclear Electric were privatized, although the modern stations of 
Nuclear Electric were subsequently sold as British Energy in 1996 (Newbery and Pollitt, 
1997). In the very first years of regulation, Littlechild (1992) reviewed the major reforms and 
concluded that the new arrangements had proved entirely workable. 
 
Shortly prior to privatisation, 12 regional electricity distribution companies (RECs) replaced 
the 12 area boards and transmission became the responsibility of the National Grid Company 
(NGC), a company fully owned by the RECs (Jamasb and Pollitt, 2007). Also, a mandatory-
pool system was introduced in 1990. Since then, the British market has gone through two 
more stages of reform: one by New Electricity Trading Arrangements (NETA) introduced in 
2001 and second by the British Electricity Trading and Transmission Arrangements (BETTA) 
in 2003. 
 
The British electricity reform involved all the elements of a full sector reform including 
restructuring, privatisation, regulation, and competition. At present, the UK market is fully 
liberalised. All consumers can choose their own supplier. Since the starting of reforms, prices 
have fallen in real terms by about 25 per cent, system reliability has been maintained at high 
levels (Thomas, 2004). Although a competitive market is achieved through further asset 
divestiture and new entry, vertical integration has not disappeared. 
 
2.1.2 Chile and Argentina 
 
Electricity reforms in Chile and Argentina were the deepest and the most radical (Pollitt, 
2004). Chile was the first country in the world that introduced reforms in its power industry. 
In Chile, the electricity power law was enacted and a wholesale market was created in 1982. 
24 
 
The state-owned electricity enterprise was privatised without its transmission systems being 
unbundled, leading to emergence of a private company with a virtual monopoly on the 
transmission sector. More specifically, new private company group (Endesa group) held a 
share of more than 90% in transmission network, controlled 80% of generating capacity and 
was the distributor for 43% of all customers in Chile (Nagayama, 2007). So, one group 
controlled generation, transmission and distribution, giving it sufficient leverage to exert 
market power. Today, the Chilean power sector is comprised of 31 generating companies, 5 
transmission companies, and 36 distribution utilities, most of which are privately owned. 
Power generation and transmission operations have been liberalised, allowing free entry to 
and withdrawal from these businesses. Since no restrictions are imposed on foreign capital 
companies, numerous businesses have entered the market (Nagayama and Kashiwagi, 2007). 
As expected, the biggest challenges in Chile’s power sector are issues related to reducing 
market power and promoting competition. The problems originated from the fact that a pool 
market model was adopted when there were a few big generation companies with market 
power. The inadequate unbundling of the generation and transmission sectors resulted in the 
transmission company being owned by a specific generation company, which worsened the 
situation. Finally, as result of an acute electricity crisis caused by a collapse of hydro output in 
1998-1999 and ahead of an election, Chilean government intervened in the functioning of the 
market and assumed a greater decision making role in strategic investment and regulation. 
 
Argentina was also one of the first countries in the world to implement an electricity market 
reform. Besides, market reform in Argentina has been regarded as one of the most successful 
ones as it achieved significant reductions in system losses and improvements in quality of 
supply (Haselip and Potter, 2010). The Argentine power sector reform was designed based on 
the lessons learned from privatisation and reforms in Chile and the United Kingdom. 
Especially, full-scale unbundling in Argentina was in response to the problems that had been 
25 
 
experienced in Chile, where insufficient unbundling and limitations on competition had 
damaged reforms. The electricity sector in Argentina was considerably restructured in 1992 as 
part of the reorganisation and privatisation programme. That is, the power sector reform was 
performed as part of the wider structural changes in the overall Argentine economy. It was 
also an attempt to address the impending energy crisis. In reform process, more than 80% of 
the generation, all of the transmission and 60% of the distribution sector were transferred into 
private ownership. Remaining public ownership was limited to the state owned nuclear power 
generating company and two hydro-electric plants (with part foreign ownership) in the 
generation sector and some provincially owned distribution companies (Pollitt, 2008a). The 
vertically integrated state owned company was restructured into 5 generation and 3 
distribution companies. A system operator (CAMMESA) was established with equal equity 
participation by all interested parties in the market except for small customers and generators. 
The generation market was very successful and the most competitive one probably in the 
world in the late 1990s. 
 
One rather novel aspect of reforms in Argentina is the arrangements for transmission 
expansion. With the reforms, transmission expansions in Argentina were no longer the 
responsibility of the transmission owner or regulator, but of the users of the transmission 
system. The public contest method required users to propose, approve and pay for major 
expansions. Approved expansions were then put out to competitive tender (Littlechild and 
Ponzano, 2008; Littlechild and Skerk, 2008a, b, c, d, e). 
 
Until the macroeconomic crisis of 2002, power sector reforms in Argentina proved successful. 
This is illustrated by the decrease in electricity tariffs and the improved investment situation 
for generators in the decade between reforms being implemented and the economic crisis 
(Nagayama and Kashiwagi, 2007). With the devaluation of the peso in 2002, retail prices for 
26 
 
electricity were frozen, which stopped investments and caused generators and distribution 
companies to suffer from losses as they could not pass-through price increases to customers. 
To balance the disequilibrium between demand and supply caused by the tariff freeze, the 
government pursued a policy of price controls, subsidies and demand-side management 
measures. So, the politicisation of tariff setting process resulted in the setting of electricity 
tariffs at a level at which cost recovery was not feasible, which interfered with the functioning 
of the market. 
 
Although after the crisis the achievements of the reforms were severely limited by the 
government’s poor energy policy and intervention into the market for political reasons; today 
the framework of liberalization is sustained and still functional in Argentina. Pollitt (2008a) 
draws two sets of lessons from Argentina’s electricity reforms. First, comprehensive 
electricity reform can work in a developing country. Second, well organised markets and 
effective network regulation are undermined if there is unnecessary political interference in 
the pricing of electricity. 
 
2.1.3 Nordic countries 
 
Norway was the first among the Nordic countries to liberalise its electricity market in 1991, in 
line with British model but without privatisation. Today, the Norwegian electricity industry 
remains almost entirely in public hands. In Sweden, reform process was initiated in 1996. The 
retail market was fully opened and since 1996 the consumers can choose their own supplier. 
The transmission system remained in a non-profit public monopoly regime. The Finnish case 
is unique in the sense that even before the reform the market was very open (Pineau and 
Hämäläinen, 2000). Finland decided to reform the sector in 1995 and since then the market 
was progressively liberalized. In 1997, the consumers got the right to choose their supplier. 
27 
 
The reform also established the separation between the transmission and generation firms 
although the distribution companies are not fully unbundled. The reform experience in 
Denmark was similar to that of the other Nordic countries and begun in 1996 and was 
completed in 2001 when consumers were granted the right to choose the supplier and the 
industry was totally unbundled. In Denmark, electricity sector is characterized by the presence 
of public local companies and, in spite of some privatization, the overall ownership structure 
still remains mostly public. 
 
Rather than implementing the reforms on their own, the Nordic countries chose to reform by 
merging their electricity markets. Nord Pool, the electricity power exchange with equity 
participation from each country’s system operators, was founded by Norway and Sweden in 
1996. Finland joined the NordPool in 1998 and Denmark in 1999. Reforms in the Nordic 
region seem to have been relatively successful, merging the four countries’ (Norway, Sweden, 
Finland and Denmark) systems into one market. This success is apparent in the fact that 
unlike the California electricity market that collapsed following from severe demand and 
supply shocks in 2000-2001, the “lights have stayed on” in the Nordic market in spite of 
similar adverse supply and demand shocks in 2002-3 (Amundsen and Bergman, 2006). 
However, electricity prices in the region have increased as a result of a rise in electricity taxes 
and the introduction of the European system of CO2 emission permits, which limited the 
popularity of the reforms in the region. Since the decisions to raise electricity taxes and to 
introduce emission permits have nothing to do with reform process, Nordic electricity market 
seems to work quite well so far. Amundsen and Bergman (2006) conclude that the main 
factors behind the relatively successful electricity market reform in the Nordic countries 
include (i) a simple but sound market design, to a large extent made possible by the large 
share of hydropower, (ii) successful dilution of market power, attained by the integration of 
the four national markets into a single Nordic market, (iii) strong political support for a 
28 
 
market-based electricity market, (iv) voluntary, informal commitment to public service by the 
power industry. They also argue that the second and third of these factors are “transferable”, 
while the first and fourth to a large extent are country-specific. 
 
Littlechild (2006) questions the ability of regulation to substitute for retail competition in the 
context of Nordic residential electricity markets (Norway, Sweden and Finland) after they 
opened to retail competition in 1998. Although they, he argues, have not been subject to 
regulatory controls on prices or other contract terms; competition is developing well, and 
between 11% and 32% of residential customers have switched to other suppliers, and a further 
19% or more have chosen new terms with their local supplier. He also underlines that the use 
of new products in the Nordic electricity markets is increasing over time, and there is 
considerable product innovation. 
 
2.2 Countries with limited reforms 
 
2.2.1 North America 
 
For most of its history the US electricity sector has been dominated by large, vertically 
integrated, and heavily regulated private utilities. The US has never implemented a mandatory 
comprehensive federal electricity market reform program, leaving the most significant reform 
decisions to the states. As a result, many states in the US have introduced only limited 
liberalization without fundamental electricity sector restructuring (Joskow, 2008). 
 
Beginning in the late 1970s, some steps were taken in the US to reform the traditional 
structure. By the late 1990s, extensive disintegration, considerably looser regulation, and 
more market-oriented operation were characteristics of the new US electricity industry. The 
29 
 
reforms were mainly intended to bring competition to wholesale market. Competition among 
independent generators was supposed to create a framework for wholesale power transactions 
so that retail customers and local distribution utilities could purchase power from a wide 
range of alternative suppliers. The result was supposed to be lower wholesale costs and thus 
lower retail prices. By the year 2000, about half of the states either had restructured their 
electricity sectors or was planning to do so (Kwoka, 2008).  
 
Paul L. Joskow argues that there are fairly dramatic differences in the electricity prices across 
US states and, in high-price states, there are strong incentives for consumers, and certainly for 
new producers, to press for reforms that provide competitive entry and possibility of price 
reductions (White, 1996). So, he maintains that states only with high electricity prices have an 
incentive to introduce reforms in the US. Those with relatively low prices do not have such an 
incentive and therefore do not introduce reforms. That is, for him, the differences in electricity 
prices among US states explain why some US states introduce reforms and why others do not 
(Joskow, 2006). Besides, Joskow (2000) points out that average real electricity prices in the 
US fell rapidly from the early 1900s until the early 1970s; from the mid-1970s until the mid-
1980s, however, real prices for electricity increased sharply in response to increases in basic 
energy prices, high interest rates, tightened environmental standards, and investments in 
capital-intensive nuclear power plants. In the mid-1980s, he continues, average real electricity 
prices began to fall again as input prices declined, and they continued to fall during the 1990s; 
in some areas of the country, however, electricity costs and prices remained well above their 
historical low values. Joskow (2000) argues that these continuing high prices became an 
impetus for restructuring in several states. He maintains that the average price of electricity in 
the US in 1997, before major reforms were implemented, was 6.85 cents a kilowatt hour. The 
average price charged to residential customers was about 8.4 cents/kWh, and the price to 
industrial customers was about 4.5 cents/kWh. Although these prices were at the low end of 
30 
 
the range of prices for developed countries, there were large interstate and interregional 
differences in electricity prices. He claims that the states in the Northeast and California had 
much higher prices than the average for the rest of the country; and therefore, they were the 
first states to restructure their power sectors in 1996 and 1997. White (1996) also provides 
evidence that the states with relatively high electricity prices introduced reforms and 
differential prices explain the pattern of electricity reform in the US. White (1996) 
concentrates on the price gap between the regulated price in a given state compared with the 
prices that arise from allowing competitive new entry in power generation. Then, he examines 
the causes of this price gap and estimates its magnitude for different states. His estimates 
reveal fairly dramatic differences in the price gap across states. He concludes that in high-cost 
states they suggest strong incentives for consumers and new producers to demand competition 
in the power market.  
 
Sioshansi (2008) argues that the pace of growth in retail competition has slowed in recent 
years in the US and the transition to a national competitive electricity market has stalled. He 
cites the reasons for this as (i) the spectacular failure of the California market, (ii) mixed 
results in a number of states that have introduced retail competition, (iii) problems in some 
wholesale markets that have not performed as expected, and (iv) a lack of interest by the US 
Congress to push retail competition at the national level.  
 
Among the reasons above, California crisis needs further focus. The California electricity 
market reform had promised to deliver reliable service at low and stable prices. The California 
electricity deregulation process was put into effect in 1998. In the period prior to reform, there 
was a considerable excess generation capacity and electricity prices were above normal. As 
we mentioned before, there were large differences in prices across the US states and 
California had much higher prices than the average for the rest of the country. Joskow (2000, 
31 
 
Table 4-1, p.136) shows that price of electricity for residential and industrial consumers were 
11.50 and 6.95 cents per kilowatt hour in California, respectively, while the US average was 
8.43 cents per kilowatt hour for residential consumers and 4.53 for industrial ones. So, he 
argues, high electricity rates became a major issue in California in the early 1990s, where 
general concerns about the future of the state’s economy were exacerbated by a severe 
recession. Within this context, California become the first state to respond to pressures from 
industrial customers to reduce the price of electricity and from independent power producers 
to increase opportunities to widen the markets for their power; and introduced a reform 
program into its power market. The program included the introduction of new institutional 
arrangements such as power exchange and independent system operator, restructuring, fixing 
end-user prices at 1996 levels and a ban on new long-term power purchase contracts. Public 
power companies were excluded from the deregulation process but had to continue providing 
cheap electricity. 
 
The summer of 1998 showed tendencies to excessive wholesale prices but apart from this 
there was no particular problem and the market seemed to function fairly well until 2000. 
Prices on the wholesale market started to increase in the early summer of 2000 and continued 
to do so in the following months. The first of several forced blackouts took place in June 
2000. During this period, the three major companies started to lose money on a large scale 
and became unable to pay for their power purchases. Consequently, the power generation 
companies became reluctant to sell power on the power exchange as their contracts were not 
honoured. The cap on end-user prices effectively hindered that rising wholesale prices 
transformed into rising end-user prices that would otherwise result in a reduction of 
consumption. Also, the restrictions on the long-term power contracts implied lacking hedging 
opportunities. Finally, the power exchange broke down and was declared bankrupt in March 
2001 (Amundsen and Bergman, 2006). 
32 
 
 
Woo (2001) identifies the major factors that contributed to California crisis as follows: poor 
market design, market power, demand growth (due to extremely warm weather during the 
summer of 2000) not matched by new capacity, a sizable reduction of hydro power generation 
(due to dry weather conditions), rising marginal cost (due to an increase in the price of natural 
gas by some 70 percent from April to November 2000), and financial insolvency (for further 
details, see Sweeney (2002)). The problems in California and elsewhere brought further 
restructuring to a halt in the US but many states were irreversibly committed to deregulation. 
At present, electricity restructuring is substantially complete in some regions of the US, 
although other regions are much less affected. 
 
In Canada, electricity reform started in the province of Alberta in 1996 where competition 
was introduced into power generation, and a wholesale electric pool was created. Alberta had 
a positive experience with reform leading to substantial new investment and reduction and 
stabilisation of prices. On the other hand, in Ontario, political mismanagement of a power 
crisis led to reform being abandoned and government interference into prices (Sioshansi, 
2008). 
 
2.2.2 Australia and New Zealand 
 
The 1990s witnessed a substantial reform in the Australian electricity sector. Since 1991, the 
industry has been broken up into its constituent parts; a national wholesale market for 
electricity was created, competition was introduced to electricity generation and retail supply 
sectors. The reform process was initiated in the State of Victoria for the first time. Victoria’s 
vertically integrated electricity company was divided into generation, transmission and 
distribution/retail. Later, distribution/retail was further divided into five companies with 
33 
 
separate franchise areas and generation was broken up into seven separate generation 
companies. Until 1997, these companies were privatized. In 1994, a wholesale electricity pool 
was established in Victoria and it was merged with the New South Wales wholesale market in 
1998, creating the national wholesale electricity market. After the creation of national 
wholesale market, ownership and operation of the transmission system was separated and a 
public company was made responsible for the operation of the system. As a whole, the 
introduction of competition and privatization led to substantial improvements in productive 
efficiency. Capital utilization rates greatly increased and staff numbers reduced. The largest 
gainers from the reform process were the large industrial and commercial consumers, who 
were able to take advantage of competition among retailers. Households saw little change in 
the real average price of electricity (Abbott, 2006). Today, the progress of liberalization varies 
from state to state in Australia. New South Wales, Victoria, and Queensland have achieved 
liberalization in the retail sector while West Australia still maintains a vertically integrated 
structure.  
 
From 1992 to 1995, significant reforms took place in New Zealand electricity market. In 
1992, an electricity law was passed and provided liberalization of the market and regulation of 
transmission and distribution segments. In 1998, another law was enacted and required forced 
ownership unbundling of electricity distribution from the rest of the electricity industry. Until 
2001, there was no explicit sector regulator and the regulation was left to general competition 
authority. In 2001, a specific sector-focused regulation was introduced and electricity market 
regulatory commission became operational in 2003. Nillesen and Pollitt (2011) examined the 
impact of the policy of forced ownership unbundling of electricity distribution on electricity 
prices, quality of service and costs. They found that ownership unbundling did not achieve its 
objective of facilitating greater competition in the electricity supply industry but that it led to 
34 
 
lower costs and higher quality of service. They concluded that this experience indicated the 
potential benefits of ownership unbundling but also the danger of unintended consequences. 
 
2.2.3 European Union 
 
Traditionally, electricity utilities were vertically integrated in many European countries, with 
state or municipally owned enterprises playing an important role. The market was highly 
regulated with very limited opportunities for users to switch to alternative suppliers. There 
was no third party access to the transmission grid (Fiorio et al., 2007). After the pioneering 
experiences of some member and neighbour states such as the UK and Norway in the 1980s, 
the European Union (EU) began an effort of gradual electricity liberalization starting with the 
first Directive in 1996. The first directive was a compromise between countries that had 
started liberalization and those that contemplated it as a very remote possibility (Trillas, 
2010). EU directive of 1996 required 15 member countries to open their retail markets at least 
partially by 2000. By 2000, all EU member countries, except Greece, had opened their retail 
markets (Jamasb and Pollitt, 2005; Pollitt, 2009). The other objectives in the directive include 
account separation between potentially competitive and monopolistic segments; freedom of 
choice for large consumers; and increasing autonomy of transmission networks. However, it 
still accepted negotiated third party access to networks. The directive was criticised for 
allowing countries too many ways of avoiding complying with the spirit of the reforms; not 
requiring a wholesale market or a market regulator to be set up. The unbundling requirements 
did not guarantee independence of access to the network and the negotiated third party access 
(TPA) option offered the incumbent companies a way to keep out competitors. Retail 
competition was restricted, with no more than a few thousand consumers able to choose by 
2003 even in the largest countries (Thomas, 2006a). 
 
35 
 
The new Electricity Directive was agreed in 2003 and it placed more stringent requirements 
on member states to disintegrate their electricity industries and introduce competition in 
generation and retail supply. EU directive of 2003 required all member states to open the 
retail market to all customers excluding residential users by July 1, 2004 and to achieve 
complete liberalization by July 1, 2007. The negotiated TPA and single buyer options were 
withdrawn and access to the network has to be via regulated TPA. Member States are also 
required to appoint an independent sector regulator. Other key objectives to be achieved by 1 
July 2007 in each member state include the legal unbundling of transmission and distribution 
businesses from competitive generation and supply, free entry into generation markets and 
regular monitoring of the progress of supply competition. 
 
European Commission adopted a third package of energy market reforms in 2009; however 
the new electricity directive/regulation came into force in 2011. This new package aims at 
extending earlier reform packages in 1996 and 2003. At the centre of the third legislative 
package, there are consumer choice, fairer prices, cleaner energy and security of supply. In 
order to reach those goals, the Commission proposes to separate production and supply from 
transmission networks; to facilitate cross-border trade in energy; to improve the effectiveness 
of national regulators; to promote cross-border collaboration and investment; to increase 
market transparency on network operation and supply and to increase solidarity among the 
EU countries. 
 
Overall, all directives aimed at creating a strongly market-based system and a single European 
electricity market. However, many of the EU member states are reluctant in implementing 
these measures. A particular problem in the EU is the lack of will among member states and 
the EU Commission to reduce the market power of dominant companies. They prefer to 
maintain or allow the emergence of “national champions” in the electricity sector. At present, 
36 
 
in most of the European countries, the incumbents’ shares lie between 85 and 95 per cent and 
the incumbents are not challenged by competition from new entrants. In Italy, Denmark, 
France, Germany, the Netherlands and Belgium, switching rates remain below 10 per cent. 
They are slightly above 10 per cent in Finland and Spain. Only three countries exhibit net 
switching rates exceeding 20 per cent: Great Britain, Sweden and Norway (Defeuilley, 2009). 
Today, a number of electricity market models coexist in Europe and they are different from 
one another in terms of the type of ownership, degree of openness, market concentration, and 
the degree of vertical integration. So, it is very difficult to argue that a unique European 
pattern of reform is emerging.  
 
The French electricity market is extraordinary in terms of its input mix to generate electricity 
and its market structure can be considered as opposite of the British model. In France, in 
2008, 77% of the electricity was generated by nuclear plants, 14% came from hydro and 
renewable sources and just 9% from fossil fuels (US EIA, 2010c). In contrast to the UK, 
France was one of the latecomers in initiating reform and implementing the EU directives. 
Also, the reform in France has not led to a major change in the structure of the sector. Reform 
process in France begun only in 2000 when France approved a law to implement EC Directive 
of 1996. The reform included creation of a sector regulator, a regime of regulated third party 
access, introduction of a wholesale market and a progressive opening of the sector with the 
possibility for the consumer to choose the retailer. However, today, there is still a vertically 
integrated public monopoly or near-monopoly (EdF) in France operating at all stages from 
generation to sales. 
 
Prior to the reform, there were a regime of private regional monopolies in Germany with nine 
vertically integrated regional companies, then merged in four groups which, in 2000, still 
controlled 80% of production, 40% of distribution and all transmission (Florio, 2007). In 
37 
 
1998, Germany adopted EU directives regarding the liberalisation of the sector. Since then, 
Germany has realized overall liberalization but without reorganization of vertically integrated 
power companies. The reform introduced full market opening, an electricity exchange and a 
regulatory body. Today, the German electricity market is still characterised by a high degree 
of vertical and horizontal integration dominated by a few large companies, which prevents 
competition and keeps barriers for new entrants and investments. 
 
In 1999, Italy adopted EU directive of 1996 and liberalized its electricity sector by unbundling 
state owned vertically integrated company (ENEL), creating a state-owned transmission 
system operator, and privatizing some power stations of ENEL (a total capacity of 15,000 
MW) to limit its market share to 50% after 2003. Today, the main problem in Italian power 
market is the dominant position of ENEL in basically all segments of the business. An 
additional obstacle to the development of effective competition is the majority stake of 
government in ENEL, which translates into significant political interference on the definition 
of the objectives as well as the management of the company (Ferrari and Giulietti, 2005). 
Today, ENEL still controls about 40% of the generation and the entire distribution network 
with the exception of few cities where the local municipalities own the distribution 
companies. 
 
The reform of the energy sector begun in Spain with a law enacted in 1994 with the aim of 
liberalizing electricity sector. It mandated the legal unbundling of the transmission network 
and created an independent joint public-private transmission system operator (REE), which 
offered regulated TPA to both the transmission and the distribution networks. A new law was 
adopted in 1997 to accelerate the process of liberalization. Full market opening has been 
implemented in Spain since 2003. In 2006, the Iberian Electricity Market (MIBEL) was 
founded and aimed at creating an integrated electricity wholesale market with Portugal, 
38 
 
notably by creating a single market operator for the wholesale Iberian pool market. Today, 
although there exists a market regulator, what consumers end up paying and firms receiving is 
ultimately determined by regulated tariffs, which are set by the government on an annual 
basis, and in a non-transparent manner. Also, the new system has failed in attracting new 
entry, and in promoting the efficient amount of investment needed to guarantee adequate 
reserve margins. Entry has been dissuaded by the incumbent firms. This has mainly been 
achieved by the strategic announcement of new investment plants that have never been carried 
out (Crampes and Fabra, 2005). The most relevant outcome of the electricity reform in Spain 
so far has been the emergence of some big firms that consolidated their generation assets. The 
market is mostly controlled by the three largest companies. 
 
In Poland, electricity reforms took place within the context of Poland’s post-Cold War 
transition from socialist to market economy. Before the reform, whole energy sector was 
controlled by a single vertically integrated company. Between 1987 and 1990, electricity was 
separated from this structure, but remained a vertically integrated industry. In Poland, 
electricity tariffs were massively subsidized through housing subsidies; tariffs paid by 
residential consumers recovered only 1% of the cost of supply (Williams and Ghanadan, 
2006). In 1990, the electricity sector was unbundled both vertically and horizontally into 
autonomous state-owned enterprises and a transmission company (PSE) was set up. In 1993, 
all distribution utilities and a number of generators were turned into joint stock companies, 
which were to be privatized through stock sales (with a limit of 50% on foreign ownership). 
PSE operated the grid as a single buyer based on power purchase agreements with the 
generators. In 1997, a wholesale market was created and replaced the single-buyer model. A 
spot market was also set up in 2000. Reform process has raised tariffs to near 90% of full cost 
recovery, but at a high cost in public support for reforms. At present, Polish market witnesses 
rebundling and vertical reintegration. Two big state-owned utilities (representing almost half 
39 
 
of Poland’s electricity market) were created by merging a number of generation, distribution 
companies and some coal mines (Williams and Ghanadan, 2006). This move reflects a 
growing view in many small EU member states that large, state-supported utilities will be 
more competitive in a unified EU market than small unbundled companies. The same trend 
has been witnessed in Slovenia as well. As a new member state of the EU, Slovenia has been 
required to adopt EU legislation in full and opened its electricity market fully in 2007 when 
all consumers became eligible. Electricity reforms in Slovenia included market liberalization, 
unbundling of activities, allowing regulated TPA, formation of an organized power market, 
adoption of incentive-based price cap regulation and the establishment of an independent 
regulatory body. Like Poland, Slovenia has merged the majority of the state owned power 
plants into a holding of electricity companies (HSE) in 2001 (Hrovatin et al., 2009). 
 
Electricity market in Greece was also dominated by a vertically integrated, state owned 
company (PPC) until the reforms. Greece embarked on electricity market liberalization in 
2001 both to comply with EU directives and to encourage private investments. PPC was 
converted to a share company but remained under state control. A mandatory pool system was 
set up and full market opening has applied since 2007. At present, PPC still holds a highly 
dominant position in both electricity generation and power supply markets. Customer tariffs 
applied by PPC, which holds 98% of consumers, are regulated by the state and their structure 
still includes large cross-subsidizations among customer categories. It is also claimed that the 
level of regulated electricity prices is below power generation costs in Greece  (Iliadou, 
2009). Besides, compliance with the EU legislation on unbundling has been delayed in Greece 
and is still poorly developed. Legal unbundling was introduced only in relation to 
transmission, while PPC remains the exclusive owner of the transmission and the distribution 
networks. 
 
40 
 
In Romania, the vertically integrated, state owned monopoly was divided into five separate 
state owned enterprises over the 1998-2000 period: one each for nuclear generation, hydro 
generation, thermal generation, transmission, and distribution. Since then the distribution 
function has been further divided into eight regional companies, five of which were 
privatized. The wholesale market has been operating in Romania since 2000 and the market 
was fully liberalized in 2007 and all consumers can choose their supplier since then (Diaconu 
et al., 2009). 
 
2.2.4 Turkey 
 
Being a candidate for EU membership since 1960s, Turkey has also closely followed EU 
directives. Before the reforms, as was the case in many European countries, the Turkish 
electricity industry was dominated by a state-owned vertically integrated company (TEK). In 
1982, public monopoly on generation was abolished and the private sector was allowed to 
build power plants and sell their electricity to TEK. In 1984, TEK was restructured and gained 
the status of state-owned enterprise. In 1993, TEK was incorporated into privatization plan 
and split into two separate state-owned enterprises, one for generation and transmission 
(TEAS) and other for distribution (TEDAS). In 2001, the reform process in electricity market 
was initiated. TEAS was restructured to form three new state-owned public enterprises: a 
transmission company (TEIAS), a generation company (EUAS) and a wholesale company 
(TETAS) (Erdogdu, 2007, 2009, 2010). Turkey’s electricity distribution network was further 
divided into 21 distribution regions. TEDAS, which owns 20 of the 21 regions, was included 
in the privatization programme, and a separate distribution company was established in each 
of these 20 regions. As of the beginning of 2012, multiple rounds of divestments of state 
electricity distribution companies have been completed, and the privatisation authority of 
Turkey is due to release the tender for the privatisation of large coal and hydropower plants. 
41 
 
 
Bagdadioglu et al. (2007) focus on creation of 21 distribution companies in Turkey, 18 of 
them by merger. Their paper investigates the potential production efficiency gains using a 
methodology to assess the potential effect of the mergers and whether the mergers are 
efficiency enhancing. This is performed by comparing the actual efficiency levels of observed 
distribution companies with the merger of proposed aggregated companies. The model is 
calibrated on panel data from 1999 to 2003 which include measures of physical capital and 
labour inputs, as well as customer and energy related outputs. Their results indicate potential 
for considerable efficiency gains from the proposed mergers in Turkey. 
 
2.2.5 Japan, South Korea and Philippines 
 
Electricity reforms in the Japanese electricity industry started in 1995 and for the first time 
IPPs were allowed to enter into the generating market by introducing the competitive bidding 
in the wholesale market. The government also introduced yardstick regulation, under which 
the electricity price of each electricity company is determined partly by comparing its 
performance with that of other companies. Companies with higher costs than others suffer 
losses, while those with smaller costs generate profits. Therefore, this system is expected to 
promote the cost cutting competition (Nakano and Managi, 2008). Partial liberalization in 
retail markets was introduced for large consumers in 2000 when power producers and 
suppliers were allowed to enter the market and use networks. Although the liberalization is 
limited in part by the fact that the retail power market has only about 30% share of total 
electricity demand, the eligible customers now have a choice among the nine major utilities 
and ten new entrants (Asano, 2006). Besides, Ida et al. (2007) found that first-period reforms, 
implemented in 1996-1999, were able to reduce costs by 7.5%; while second-period reforms, 
42 
 
during the period of 2000 to 2002, effectively cut costs by 11.8% in Japanese electricity 
market with respect to the base costs before regulatory reforms. 
 
South Korea began transforming the structure of its electricity industry from public monopoly 
to market competition in 1998. Until then, the electricity industry of the country had been 
dominated by a state-owned vertically integrated company (KEPCO). The restructuring plan 
aimed at introducing market competition and privatization to the power industry, which was 
accompanied by the vertical unbundling and horizontal divestiture of KEPCO. As the first 
step of this plan, in 2001, the power generation function of KEPCO was divided into five 
thermal and hydropower generation companies and one company for nuclear power 
generation. Five companies were planned to be privatized over the next several years. 
However, in 2004, the Korean government suspended its electricity market reform based on 
the recommendation of a joint study team, which concluded in their final report that the 
alleged benefits of reform are theoretical and uncertain, while the real costs and risks are 
substantial. This suspension effectively interrupted the original plan adopted in 1998 by the 
previous administration to divest and privatize KEPCO’s generation assets and introduce 
wholesale and retail competition (Lee and Ahn, 2006). 
 
In Philippines, reform process initiated in 2001 when the government focused on introducing 
structural reform and market mechanism principles into the electric power sector. Reform 
objectives included full privatization of state-owned electricity utility, promotion of private 
participation in power market, establishment of a wholesale spot market and full liberalization 
of the market. However, in practice, many of these steps are behind schedule. The power 
prices were distorted due to the take-or-pay contracts with IPPs. Today, the sector reform is 
still on-going in the Philippines and electricity prices are still among the highest in Asia. A 
wholesale electricity spot market was launched in 2006. Since there is no sufficient number of 
43 
 
market participants to create a competitive environment, wholesale prices have not decreased. 
Toba (2007) reports an empirical investigation into the welfare impacts of introduction of 
private sector participation into the Philippine electricity generation sector and privatization of 
the distribution company (Manila Electric Company, Meralco). The study uses a social cost 
and benefit analysis. According to results, the main benefits came from IPPs, which 
contributed to resolving the power crisis and promoted economic and social development. 
Consumers and investors are net gainers, while the government lost and an air pollution cost 
was incurred. The paper concludes that reform with private sector participation increased 
social welfare. 
 
2.2.6 Brazil, Bolivia and Peru 
 
Reforms in Brazil were cautious and gradual (Gabriele, 2004). Brazil started to reform its 
power sector in 1995 with privatization of its major electricity utility (Eletrobras). IPPs were 
allowed to enter the Brazilian market and generation companies were privatized. Besides, a 
nationwide power grid operator and a wholesale electricity market were established. 
However, the vast complexity of the Brazilian electricity industry, incompletely defined 
regulatory structure, a lack of effective planning and an unstable economy hindered the flow 
of investments necessary to guarantee the system’s expansion. As a result, the Brazil 
experienced a rationing of electrical energy that lasted from June 2001 to February 2002 (De 
Souza and Legey, 2010). The 2001 crisis vividly demonstrated Brazil’s vulnerability to 
drought due to an excessive dependence on hydro-power and its low reserve margins (Lock, 
2005). In response to this crisis, after 2004, Brazil shifted its electricity policy to emphasize 
long-term stability instead of free market. Brazilian government established a new regulatory 
framework for electricity. This new framework has three broad objectives: (1) to create an 
efficient mechanism for the contracting of electricity on behalf of captive consumers; (2) to 
44 
 
ensure security of supply at the lowest possible prices; and (3) to provide universal access to 
electricity to consumers around the country (Dutra and Menezes, 2005). The 2004 revision 
introduced new practices in Brazilian power market. To begin with, two distinct contract 
environments are defined. The first environment is the regulated contracting environment 
(ACR) and second one is free contracting environment (ACL). The former has the purpose of 
protecting captive (small) consumers, while the latter allows for “free” (large) customers to 
choose their electricity suppliers. Within the ACR, a distinction is made between “new” and 
“existing” electricity. The aim is for final consumers to pay a combination of a higher price 
associated with new plants and a lower price associated with existing, partially, or fully 
depreciated plants. In this contracting environment, distributors are required to contract their 
entire forecast demand for captive consumers with generators, importers, and retailers. 
Contracts will be auctioned off over time with different auctions for new and existing 
electricity under a lowest-tariff criterion. Other new practices included the revitalization of 
mid- and long-term planning, the introduction of long-term agreements to guarantee the return 
of investments in new plants, and the uncoupling of distribution services from any other 
activities. 
 
In the period 1995-2000 approximately 60% of the Brazilian electricity distribution market 
has been privatised. Mota (2003) assesses the social welfare impact of the privatization 
process for the distribution and supply markets using a social cost-benefit methodology. The 
results show that net benefits are significant, but producers absorb most of the gains. The 
study also concludes that if the regulation had been tougher since the beginning, consumers 
could have benefited more from the privatization. 
 
Bolivia’s electricity reforms occurred in the context of a debt crisis. Electricity reform was a 
component of wider economic reform. Even before reform, generation and distribution were 
45 
 
already partly unbundled with diverse ownership. However, vertically integrated state utility 
(ENDE) controlled 80% of generation and operated the grid. The power sector in general and 
ENDE in particular provided satisfactory service, operated efficiently with relatively low 
system loses and were profitable at the time of reforms as tariffs were set above cost recovery 
levels. In 1994, the sector was fully unbundled. ENDE was turned into three private 
generation companies and a private transmission company. Privatization of the sector was 
completed by 1998. A wholesale market was created and consisted of regulated contracts 
supplemented by a competitive spot market, with distribution utilities required to buy 80% of 
expected demand on 3-year contracts. These arrangements have so far resulted in significant 
investment in expansion and upgrades. The World Bank closely involved in Bolivia’s 
electricity reform and considered it a success in terms of sector finance and operations, and 
the government’s fiscal goals (Williams and Ghanadan, 2006).  
 
Peru also implemented neo-liberal market reforms in the electricity sector in the early 1990s, 
as part of a broader economic restructuring and in response to a crisis in its electricity system 
from 1986 to 1990 (Pérez-Reyes and Tovar, 2009). In 1990, the electricity rates were 
increased and the state electricity utility (Electro Peru) was restructured. Moreover, several 
state-owned companies were privatized but a significant important group of privatized 
companies were renationalized in 2002. The reforms were an attempt to attract private capital 
to finance the expansion of the power supply mainly in the generation sector. As a result of 
the reforms, electrification levels increased from 45 percent in 1992 to 75 percent by 2002. 
Service quality also improved markedly (Cherni and Preston, 2007). 
 
Anaya (2010) assesses the social welfare impact of the restructuring and privatisation of the 
electricity market in Peru. The results show that privatisation was worthwhile and that the 
46 
 
social welfare of being connected has an important contribution on it. Government and 
producers benefited the most and consumers the least due to price increases. 
 
2.3 Countries with suspended or partial reforms 
 
2.3.1 China and India 
 
China has the second largest electricity industry in the world and is playing an important role 
within the global economy. In the past two decades, it has experienced a series of regulatory 
reforms in its electricity industry. With the development of the economy since the 1980s, the 
demand for electricity grew rapidly and power shortage became more serious than ever. In 
order to attract more investments to develop the electricity industry and relieve the bottleneck 
of power shortage, the investments from local governments, domestic enterprises and foreign 
investors in generation sector have been allowed since 1985. The Ministry of Electric Power 
was abolished in 1998, with its business functions transferred to the newly formed vertically 
integrated public utility (SPC), which was corporatized shortly. In 2002, SPC was divested 
and the generation sector was separated from the transmission and distribution sectors. 
Generation function was allocated to five big generation corporations. A regulatory body was 
set up in 2003 and the introduction of the wholesale electricity market is also in process (Du 
et al., 2009). Currently, the Chinese electricity industry has evolved into a dual system, with 
dominant state planning at the core, and a decentralised generation system at the periphery, 
owned by state organisations at different levels and by private enterprises. While the 
generation sector has some market competition, the transmission and distribution sectors are 
heavily state-controlled. There is still a chronic electricity shortage, with industrial consumers 
are frequently asked to shut down production during peak times and arrange production 
schedules at nights or weekends (Cherni and Kentish, 2007). As also concluded by Yeoh and 
47 
 
Rajaraman (2004), China still has a huge task ahead of it to complete reform process. Because 
it places a higher value on political and economic stability than economic efficiency, and 
because of its unfamiliarity with a market economy, the transition to a competitive market 
could take many more years in China. 
 
In India, electricity theft, corruption, and a highly cross-subsidized pricing structure have 
made it nearly impossible for the utilities to improve power service. By early 2001, State 
Electricity Boards (SEBs) as a whole faced an average 50 per cent level of technical plus non-
technical losses (Ruet, 2005). The quality and reliability of electricity have been so low that 
industrial consumers across India exit the state-run system and rely on their own on-site 
power generation (Joseph, 2010). India initiated power sector reforms in 1991 when the 
country was facing a political and economic crisis and was under pressure to open up the 
economy as part of a reform package agreed with the International Monetary Fund (IMF) and 
World Bank. IPPs were allowed to enter the power generation business and were offered 
attractive incentives. Although the initial interest was overwhelming, the enthusiasm was 
short-lived as only a few projects actually materialised. Second wave of significant attempts 
for reform came in the late 1990s. During this period, the State Electricity Boards (SEBs) 
began to be unbundled and even State of Orissa fully privatized its generation, transmission, 
and distribution assets. Besides, regulatory commissions were set up at the central and state 
levels and single buyer model was introduced. Soon, it was argued that deeper reforms were 
required to manage, regulate and co-ordinate development of the electricity industry in India. 
New legislative framework was adopted in 2003. The new act has de-licensed generation 
(except hydro), provided for the separation of system operation and transmission activities, 
allowed trading at wholesale and retail levels and permitted multiple licensing at transmission 
and distribution levels (Singh, 2006). However, despite the enactment of a comprehensive 
legal framework for governing the electricity industry, limited progress has been made in 
48 
 
terms of achieving widespread sector liberalisation and privatisation in India (Bhattacharyya, 
2007b). At present, the electricity sector continues to perform poorly. There are still peak 
capacity shortages and energy deficits. Some consumers, like those in the agricultural sector, 
receive subsidized electricity and pay little or nothing for the electricity they consume. 
 
2.3.2 Eurasia 
 
In Eurasia region, each country’s power sector consisted of a vertically integrated public 
sector monopoly immediately following the break-up of the Soviet Union. Since then, each 
country has adopted a different strategy with respect to industry structure. These strategies 
have included different types of vertical unbundling, regulation, privatisation and 
restructuring. In Russia, reforms were adopted more as an ideological undertaking than as a 
result of economic necessity. Reforms in general aimed at diminishing the power of party-
state in general and incumbents in particular. Actually, the electricity sector in Russia was 
doing better than many other countries and was as developed as those in the US or UK (Yi-
chong, 2006). Without serious problems, changes were not so radical. The reform’s another 
objective was to attract domestic and foreign private investment to modernise and develop the 
electricity system (Engoian, 2006). The restructuring of Russia’s power generation sector will 
be complete when state monopoly (RAO UES) dissolves. The country’s transmission grid 
will remain under state control. The reform has created a generation sector divided into 
multiple wholesale electricity companies, which participate in a new competitive wholesale 
market. The creation of six wholesale electricity companies was completed in 2006. Today, 
there are seven separate regional power systems in the Russian electricity sector. RAO UES, 
which is 52 percent owned by the Russian government, controls most of the transmission and 
distribution in Russia. It owns 96 percent of the transmission and distribution system, and the 
wholesale electricity market (FOREM).  
49 
 
 
Electricity reform in Ukraine started in 1996. Vertically integrated national companies were 
unbundled and single-buyer model with compulsory pool market was adopted. Privatization 
process of the electricity companies began in 1997 but was cancelled in 1999 due to 
corruption.  
 
Berg et al. (2005) conducts an empirical analysis of 24 Ukraine electricity distribution 
companies from 1998 to 2002 and conclude that privately owned firms respond to incentives 
that add to net cash flows (associated with reducing commercial and non-commercial network 
losses); however, they also respond more aggressively than do state-owned distribution 
utilities to mark-up (cost-plus) regulatory incentives that increase shareholder value but 
decrease cost efficiency. 
 
The reforms in Armenia, Georgia and Azerbaijan were in general in the form of privatization 
and aimed at attracting foreign investment. Armenia began power sector reform in 1997. 
Unbundling and privatization in distribution were carried out, and foreign capital introduction 
was realized. In Georgia, unbundling was introduced in 1997 along with privatization of the 
power generation company and the distribution companies (Nagayama, 2007). In Azerbaijan, 
key restructuring initiative was to separate electricity distribution from generation and 
transmission, and to auction concessions to the private sector for the management of its four 
distribution companies (Mehta et al., 2007). At present, transmission and generation assets are 
held by vertically integrated state-owned enterprise. Further vertical unbundling is unlikely to 
proceed.  
 
Kazakhstan initiated unbundling and privatization in 1996. The government first unbundled 
power generation from transmission, and privatized most of the power generation capacity. 
50 
 
Transmission and distribution were remained under public domain but performed by separate 
government-run companies. In 1998, the power wholesale market was established. Today, 
Kazakhstan has multiple generators that sell bulk power at unregulated prices and wholesale 
prices are negotiated between suppliers and buyers, who may choose which generators to 
contract with (Nagayama, 2007). In Tajikistan, vertically integrated public utility was 
corporatized in 2001 to facilitate further structural changes and improve the commercial 
performance of the sector. But, apart from this, no reform steps has been taken so far. 
Kyrgyzstan also corporatized its vertically integrated electricity utility but it also unbundled it 
into several generation companies, a transmission company, and four distribution companies 
in 2001. Each of these companies is currently publicly owned but has managerial autonomy.  
 
2.3.3 Africa 
 
The reforms in Africa were very limited in terms of scope and scale and almost in all reform 
cases the main motive was to encourage foreign private direct investment in power markets. 
In Africa, only few countries introduced a substantial reform program in their electricity 
industries. In South Africa, under the apartheid government, prior to 1994, government 
policies were geared at serving the needs of the minority white population group. Energy 
policies, including electricity provision, focused on ensuring sufficient supply for the mining, 
chemical and agricultural industries, which formed the backbone of the South African 
economy. When the new, democratic government came into power in 1994, South Africa’s 
energy policy saw a fundamental shift in focus. In 1995, the government established the 
National Electricity Regulator as a successor to the Electricity Control Board that had been 
established in 1987. Electricity generation in South Africa has been dominated by Eskom, the 
state-owned electricity utility. Eskom owns, operates and maintains the national transmission 
grid and is thus a de facto monopolist on both the generation and transmission level. In 2002, 
51 
 
Eskom was converted into a public company pursuant to the Eskom Conversion Act of 2001. 
At present, Eskom is regulated by the National Energy Regulator of South Africa (NERSA) in 
accordance with the Electricity Regulation Act of 2006. However, as suggested by Newbery 
(2009), little progress has been made in South Africa in terms of electricity sector reform. 
Privatisation process was abandoned; regulator was created but the prices are still based on 
historic costs and, most importantly, demand has predictably outstripped the capacity. Today, 
Eskom continues to generate approximately 95% of the electricity in South Africa.  
 
In Nigeria, state owned power utility (NEPA) was commercialized in 1988. In 2005, the 
monopoly of NEPA in electricity industry was broken and wholesale competition model was 
put into practice. NEPA was divided into 18 companies, including 6 generators, 11 
distributors and one transmission company. Currently, the government holds the shares in the 
successor companies but it is planned that these companies would gradually be privatized 
(Ikeme and Ebohon, 2005). 
 
  
52 
 
 
 
 
 
 
Chapter 3 
 
3 Paper One 
 
 
 
The impact of power market reforms on electricity price-cost margins 
and industrial/residential price ratios: a cross country panel data 
analysis* 
 
3.1 Introduction 
 
Relevant economic theories that we summarized in Chapter 1 (theory of regulation and public 
choice theories) and previous empirical research listed in Section 3.2 predict that electricity 
prices are affected by the reforms in power industry. In this paper, we try to find out whether 
this connection holds true when we take into account the fuel costs. In the course of the paper, 
we develop a measure, called “electricity price-cost margin”, to represent the difference 
between end-user electricity prices and fuel import costs. Although this term is not a standard 
                                                          
* An earlier version of this paper is published in Energy Policy (ISSN: 0301-4215, Volume 39, Issue 3, March 
2011, pages 1080-1092, doi:10.1016/j.enpol.2010.11.023). 
53 
 
one, we still prefer to use it as it is the best term that we can find to represent the end-user 
electricity prices net of fuel import costs. To avoid any misunderstanding, we clearly define 
this term in Section 3.5. 
 
The expected direction of price changes in different countries are different, depending on 
whether prices were set at below or above long run marginal costs (LRMC) in the pre-reform 
period. If they are set above LRMC, then we expect a reduction in prices, and therefore in 
price-cost margins. On the other hand, if prices were below LRMC in pre-reform period, then 
prices (and price-cost margins) may increase as a result of reforms. So, we cannot suggest a 
link between reforms and direction of price-cost margins; but we expect that electricity price-
cost margins are affected as the market moves further from monopoly and closer to 
competition. In our analysis, we look at the industrial and residential electricity prices 
separately. Besides, in this paper, we also try to find out whether there is a correlation 
between electricity reform and industrial/residential price ratio. Electricity reforms are 
expected to influence industrial/residential price ratio; however, the expected direction of 
industrial/residential electricity price ratio changes due to reform depends on the starting point 
and cannot be hypothesized theoretically. Therefore, we cannot suggest a link between 
reforms and direction of changes in industrial/residential electricity price ratio; but we expect 
that power market reforms have a statistically significant impact on industrial/residential 
electricity price ratios. 
 
The paper also aims at clarifying whether the effects of power sector reform on electricity 
price-cost margins and industrial/residential electricity price ratios are different between 
industrial and residential consumers and between developed and developing countries. 
Empirical econometric models are estimated and analysed to observe the impact of electricity 
market reform process on price-cost margins and industrial/residential electricity price ratios. 
54 
 
The econometric models are designed using panel data from 63 countries6. The dataset covers 
the period from 1982 to 2009. 
 
We try to answer following research questions: (i) what is the impact of electricity market 
reforms on electricity price-cost margins? (ii) does liberalization cause a change in 
industrial/residential electricity price ratios? (iii) what are the other factors that influence 
electricity price-cost margins and industrial/residential electricity price ratios and how much 
are they influential relative to reform process? 
 
In point of fact, the fluctuations in fossil fuel prices constitute one of the most important 
determinants of final electricity prices and, therefore, price-cost margins. However, to our 
surprise, this variable has been ignored so far in almost all cross country econometric studies 
trying to explain the impact of reforms on electricity prices (see Ernst & Young (2006), Fiorio 
et al. (2007), Nagayama (2007, 2009), Steiner (2001) and Thomas (2006b)). Since fuel costs 
are probably the most important component of end user prices, any study excluding this 
variable destines to fall short. In view of the fact that our study is the first to take into account 
variations in fuel costs in the explanation of impact of reforms, it not only is an important 
contribution to the existing literature but also fills an important gap in this area. 
                                                          
6 Developed countries (32): Australia, Austria, Belgium, Canada, Croatia, Cyprus, Czech Republic, Denmark, 
Finland, France, Germany, Greece, Hungary, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Netherlands, New 
Zealand, Norway, Portugal, Singapore, Slovak Republic, Spain, Sweden, Switzerland, Taiwan (Chinese Taipei), 
Trinidad and Tobago, United Kingdom, United States. 
Developing countries in America (21): Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, 
Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Jamaica, Mexico, Nicaragua, Panama, 
Paraguay, Peru, Uruguay, Venezuela. 
Other developing countries (10): China, India, Indonesia, Kazakhstan, Poland, Romania, Russian Federation, 
South Africa, Thailand, Turkey. 
55 
 
 
The paper proceeds as follows. Next section provides a literature review on the impact of 
electricity sector reform process on electricity prices. Section 3.3 develops research 
hypotheses. Section 3.4 summarizes the methodological framework. Section 3.5 describes 
data. Following two sections present empirical analysis and discuss the results. The last 
section concludes after considering possible policy implications of the results. 
 
3.2 Literature review 
 
In this section, we review empirical literature on the impact of electricity sector reform 
process on electricity prices. There is an extensive volume of literature on electricity market 
reforms but most of it is in the form of opinion and discussion and not based on detailed 
econometric analysis. In line with our objectives, we focus only on those studies which aim at 
revealing the relationship between power market reforms and electricity prices by analysing 
cross-country data or developing a logical framework to evaluate cross-country evidence. 
 
Electricity prices are also expected to be influenced indirectly by any factor that affects the 
reforms, such as institutional structure of a country or political economic environment. In this 
paper, we only concentrate on the factors that directly have an impact on the prices. Some of 
the indirect factors are covered in the second and third papers. For instance, the impact 
foreign influence originating from foreign financial aid and/or assistance by international 
donor agencies (e.g. World Bank) is investigated in the third paper. 
 
Steiner (2001) carried out the first study focusing on the effect of electricity market reform on 
final electricity prices. She studied the effect of regulatory reforms on the retail prices for 
large industrial customers as well as the ratio of industrial price to residential price, using 
56 
 
panel data for 19 OECD countries for the period 1986-1996. In her analysis, she used 
electricity price, ratio of industrial to residential electricity price, capacity utilization rate and 
reserve margin as variables. The study found that electricity market reforms generally induced 
a decline in the industrial price and an increase in the price differential between industrial 
customers and residential customers, indicating that industrial customers benefit more from 
the reform. She also found that unbundling is not associated with lower prices but is 
associated with a lower industrial to residential price ratio and higher capacity utilization rates 
and lower reserve margins. Hattori and Tsutsui (2004) also examined the impact of the 
regulatory reforms on prices in the electricity industry. Like Steiner (2001), they used panel 
data for 19 OECD countries but for the period 1987-1999. They found, first, that expanded 
retail access is likely to lower the industrial price, while at the same time increasing the price 
differential between industrial and household customers. Additionally, they concluded that 
unbundling of generation did not necessarily lower the price and may have possibly resulted 
in higher prices. Like Steiner (2001), their estimation showed that the effect of unbundling on 
the level of industrial price is statistically insignificant. Besides, they found that introduction 
of a wholesale power market did not necessarily lower the price, and may indeed had resulted 
in a higher price. Their estimates showed, without exception, that establishing a wholesale 
power market resulted in statistically significantly higher prices and also increased the ratio of 
industrial price to household price, although not in a statistically significant manner. 
 
Pollitt (2009) mentions two other empirical studies that examine the price impacts of reform 
by Ernst & Young (2006) and Thomas (2006b). Ernst & Young (2006) prepared a report for 
the UK government’s Department of Trade and Industry (DTI). In their study, they used a 
sample of EU-15 countries and produced some policy suggestions for electricity and gas 
industries with a large number of simple regressions. As a result of their study, they 
concluded that liberalization lowers prices; liberalization lowers costs and price-cost margins; 
57 
 
and liberalized markets increase price volatility. Thomas (2006b) examined a number of 
reports including those of European Commission which look at (or comment on) electricity 
prices. Although these studies, he argued, suggest that reforms in the EU have been associated 
with lower prices for consumers, the evidence does not support these assertions. The price 
reductions, he continued, that have occurred in the past decade took place mostly in the period 
1995-2000, before liberalization was effective in most of the European Union and since then, 
prices have risen steeply, in many cases wiping out the gains of the earlier period. For him, 
other factors, not properly accounted for, such as fossil fuel price movements, technological 
innovations and changes to regulatory practices were more likely to have led to the price 
reductions that occurred in the period 1995-2000 than reforms that had not then taken effect. 
He also underlined that the EU reform model’s real test is whether it can deliver timely 
investment to meet the emerging investment gap following the elimination of short run 
inefficiency and initially high reserve margins.  
 
Fiorio et al. (2007) questioned the widespread beliefs that public ownership can be an 
impediment to other reforms and that it leads to production inefficiency. To test for this and 
the reform paradigm in general, they considered electricity prices and survey data on 
consumer satisfaction in the EU-15. Their empirical findings rejected the prediction that 
privatization leads to lower prices, or to increased consumer satisfaction. They also found that 
country specific features tend to have a high explanatory power, and the progress toward the 
reform paradigm is not systematically associated with lower prices and higher consumer 
satisfaction. 
 
Other two studies on econometric modelling of electricity market reforms come from two 
papers by Nagayama (2007, 2009). Nagayama (2007) used panel data for 83 countries 
covering the period 1985-2002 to examine how each policy instrument of the reform 
58 
 
measures influenced electricity prices for countries in Latin America, the former Soviet 
Union, and Eastern Europe. The study found that variables such as entry of independent 
power producers (IPP), unbundling of generation and transmission, establishment of a 
regulatory agency, and the introduction of a wholesale spot market have had a variety of 
impacts on electricity prices, some of which were not always consistent with expected results. 
The research findings suggested that neither unbundling nor introduction of a wholesale pool 
market on their own necessarily reduces the electricity prices. In fact, contrary to 
expectations, there was a tendency for the prices to rise. He argued, however, coexistent with 
an independent regulator, unbundling may work to reduce electricity prices. He found that 
privatization, the introduction of foreign IPP and retail competition lower electricity prices in 
some regions, but not in all regions. In his second paper, Nagayama (2009) aimed at 
clarifying whether the effects of power sector reforms should be different either across 
regions, or between developing and developed countries. He analysed an empirical model to 
observe the impact of power prices on the selection of a liberalization model in the power 
sector. This was achieved by the use of ordered response, fixed effect and random effect 
models. An instrument variable technique was also used to estimate the impact of the 
liberalization model on the power price. These econometric models were designed using panel 
data from 78 countries in four regions (developed countries, Asian developing countries, the 
former Soviet Union and Eastern Europe, and Latin America) for the period from 1985 to 
2003. The research findings suggested that higher electricity prices are one of the driving 
forces for governments to adopt liberalization models. However, the development of 
liberalization models in the power sector does not necessarily reduce electricity prices. In fact, 
contrary to expectations, the study found that there was a tendency for the prices to rise in 
every market model. 
 
59 
 
3.3 Hypothesis development 
 
This section of the paper develops two hypotheses as to the impact of power market reforms 
on electricity price-cost margins and industrial/residential electricity price ratios, which are 
then tested empirically in the next sections of the paper. 
 
3.3.1 Hypothesis on price-cost margins 
 
Electricity industry reforms are multi-dimensional activities with interrelating dimensions and 
a variety of impacts, and the process generally involves a set of specific steps or measures 
based on the reform model. The predictive power of theory is limited with regard to the 
outcome of the reform of natural monopolies (such as the electricity sector) as this is 
dependent on how the sector is structured and regulated (Jamasb et al., 2005). The reforms are 
expected to result in short and medium term cost savings (through privatization) and cost 
reductions (through competition), and to lower real prices in those countries where electricity 
prices have been set at compensatory levels in the pre-reform period (Joskow, 1998). 
Although the size of cost savings and reductions may differ from one country to another, the 
primary sources of cost savings are identified and extensively discussed in the literature. In 
this context, two important streams of thought that are relevant to forming a hypothesis of the 
impact of reforms on economic performance (including performance of prices) are agency 
and public choice theories (Boycko and Vishny, 1996; Niskanen, 1971; Zeckhauser and Horn, 
1989). According to them, the reforms (and especially privatization) is expected to enhance 
economic efficiency by (i) changing the allocation of property rights, which leads to a 
different structure of incentives for management and hence to changes in managerial 
behaviour; (ii) removing the ‘soft budget’ constraint of taxpayer support and exposing 
enterprises to the disciplines of the private capital market (De Alessi, 1980); (iii) introducing 
60 
 
more precise and measurable objectives, thus reducing transaction costs, especially associated 
with principals monitoring management (agent) behaviour; and (iv) removing political 
interference in the management of enterprises and capture by special interest groups (Zhang et 
al., 2008). For electricity sector, other sources of potentially significant short term and 
medium term cost savings are associated with changes in policies that favour high-cost 
domestic fuels when lower-cost alternatives are available; the performance improvements 
including increases in generating unit availability, as well as savings in both physical and 
financial losses (theft of service) on the distribution system. However, we should also be 
aware that in some cases electricity reform may raise prices because they were set ‘too low’ 
(i.e. below LRMC) in the pre-reform period. 
 
The literature we summarized in the previous section focuses on the relationship between 
reforms and electricity prices and provides us with various insights on this relationship. Since 
this paper may be regarded as an extension to the literature investigating reform-price 
relationship, we need a hypothesis focusing on this relationship with the extension that we 
concentrate on price-cost margins, rather than prices only. In short, relevant economic 
theories (theory of regulation, agency and public choice theories) and previous empirical 
investigations indicate that electricity prices are affected by the reforms in power industry. In 
this paper, we try to find out whether this relationship holds true when we take into account 
the fuel costs. The expected direction of price changes in different countries is different, 
depending on the starting point at the beginning of the reforms. That is, it is mainly 
determined by whether prices were set at below or above the long run marginal costs (LRMC) 
in the pre-reform period. If they are set above LRMC, then we expect a reduction in prices, 
and therefore in price-cost margins. However, if prices are below LRMC in pre-reform 
period, then prices (and price-cost margins) may increase as a result of the reforms. Therefore, 
61 
 
we cannot hypothesize a particular connection between reforms and the direction of price-cost 
margins. From the above arguments the following hypothesis is derived: 
  
Hypothesis 1: Electricity price-cost margins are affected as countries introduce 
reform steps (that is, as the market moves further from monopoly and closer to 
competition). 
 
This hypothesis is both an immediate application of the relevant economic theory (theory of 
regulation, agency and public choice theories) and a direct parallel with the empirical findings 
reported in the literature we summarized in the previous section. While investigating the 
relationship between reforms and price-cost margins, the latter (dependent variable) is 
represented by the variables “price-cost margin for industry” and “price-cost margin for 
households”. On the other hand, six dummy variables (independent power producers, 
wholesale electricity market, choice of supplier, unbundling, privatization, electricity market 
regulator) and their cross-products (privatization and regulator, privatization and 
unbundling, unbundling and regulator) constitute causal variables that represent electricity 
market reforms. The details of these variables are provided in the following sections.  
 
Before proceeding to the development of the second hypothesis, let me comment on the 
potential impact of causal variables on the prices, and therefore price-cost margins, if prices 
are set above LRMC in the pre-reform period. Independent power producers (IPPs) promote 
competition, which is expected to lower electricity prices. However, introduction of IPPs may 
also result in price increases if ‘high’ electricity prices are guaranteed to IPPs with power 
purchase agreements in order to promote investment (especially in developing countries). 
Wholesale electricity markets are expected to lead to lower prices through the lower 
wholesale prices induced by competition. However, as found by Nagayama (2007), if a 
62 
 
wholesale power market is vulnerable to the market power of large generators, prices may go 
up. Similarly, choice of supplier usually means competition among potential electricity 
service providers and therefore puts a downward pressure on prices; but when number of 
suppliers is limited then prices may go upward due to oligopolistic nature of the market. Also, 
unbundling is required to separate competitive and uncompetitive segments of the electricity 
industry and subject competitive segments to competition, which reduces prices. However, if 
unbundling is not effectively carried out, as in the situation where a specific firm exercises 
market power, prices may rise. Moreover, privatization usually results in capital investment 
and advanced technology to increase management efficiency, which is expected to lower 
electricity prices. However, if the investing company requires guarantees on the expected 
return on its investment and this return is assured, then electricity prices may rise. Finally, the 
electricity price is expected to be lowered with the establishment of an independent regulatory 
agency, which makes transparent power market rules to promote competition (Nagayama, 
2007). Nevertheless, a captured (by the industry) or incompetent regulator may easily result in 
an increase in electricity prices. Also, if the regulatory institution is not independent, politics 
may influence decisions on electricity prices and prices may go up. 
 
Actually, we would also like to test the impact of initial electricity price level on price-cost 
margin by including a dummy variable representing whether initial level of price (that is, 
electricity prices when reforms started) was above or below the LRMC. We have data on 63 
countries in our dataset but, due to missing observations, we are able to calculate price-cost 
margins for industry and households when reforms started for 44 countries. As shown in 
Appendix I-1, both industrial and residential prices seem to be above the fuel import costs in 
all of these countries when the reforms started. Since we do not have data on LRMC in 
countries, we find it difficult to assign a country to one of two categories of the dummy 
variable. So, due to the difficulty of establishing whether the initial prices are above or below 
63 
 
LRMC, we do not include the dummy variable representing whether initial price level was 
above or below LRMC when electricity reforms started. Although initial prices are above the 
fuel import costs for both households and industrial consumers in 44 countries in our dataset, 
it does not necessarily mean that they are also above LRMC or that this is the case in all other 
countries. If necessary data become available in the future, the analysis of initial price levels 
may produce interesting results. However, due to limited nature of our dataset, we cannot 
carry out such an analysis in this study. 
 
To investigate possible impact of the initial prices on subsequent price-cost margins further, 
we focus on the relationship between initial prices and the change in price-cost margins from 
the year when reforms started to 2007. Using data from our dataset, Appendix I-2 and 
Appendix I-3 present electricity prices for industry and households, respectively, when 
electricity reform started and the change in price-fuel cost margin between the year the 
 
Figure 3.1 Change in price-fuel cost margins for industry 
 
-150
-100
-50
0
50
100
150
200
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Electricity prices for industry  when electricity reform started (US$/MWh)
Change in price-fuel cost margin for industry since reform started, as of 2007 (US$/MWh)
64 
 
Figure 3.2 Change in price-fuel cost margins for households 
 
 
reform started and 2007. In Appendix I-2 and Appendix I-3, the countries are ranked based on 
electricity prices when electricity reform started in ascending order. Figure 3.1 and Figure 3.2 
provide the visual plot of the data in Appendix I-2 and Appendix I-3, respectively. As a result 
of a visual inspection of Figure 3.1 and Figure 3.2, we could not detect a uniform relationship 
between initial prices and the change in price-cost margins. To sum up, initial electricity price 
levels do not seem to be an effective explanatory determinant of price-cost margins. 
 
3.3.2 Hypothesis on relative price of electricity for industry and households 
 
Cross-subsidization refers to the financing of the cost of electricity service provision to a 
particular consumer group(s) by another consumer group(s). Through cross-subsidization, 
instead of following the rule of attributing costs properly to each consumer group to make 
each of them pay for the electricity they consumed, an electricity utility allows a consumer 
-150
-100
-50
0
50
100
150
200
250
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Electricity prices for households when electricity reform started (US$/MWh)
Change in price-fuel cost margin for households since reform started, as of 2007 (US$/MWh)
65 
 
group to finance another one. Traditionally, electricity tariffs have been used to engage in 
cross-subsidization, benefitting some interest groups and burdening others. In some countries, 
for social and economic policy reasons, electricity prices are uniform across the country 
despite the fact that there are regional variations in supply costs. Some other countries have 
provided direct and indirect subsidies to electricity supplied to farms, residences and small 
businesses in rural areas. In some countries, electricity tariffs either provide lower prices for 
low-income consumers or specify price structures so that consumers with low-consumption 
levels pay prices below the cost of providing them with service under the assumption that 
there is a high correlation between income and electricity consumption (Joskow, 1998). 
 
As we mentioned in Section 1.3, the economic theory of regulation may be regarded as the 
application of public choice approach to domain of regulation. The essence of this approach is 
that regulators and politicians, like economic man, weigh the benefits and costs of various 
courses of action in a political framework where the attainment of a voting majority 
determines success. In its broadest interpretation, this approach emphasizes a balancing of 
interest group strengths and weaknesses at the margin, with the outcome determined by the 
stake that the various groups have in it and the efficiency by which they can influence the 
regulatory process.  Within the framework suggested by economic theory of regulation, the 
strength of various interest groups determines the outcome of electricity market reforms 
(including the change in relative electricity prices). In pre-reform period, residential 
customers are subsidised by industrial users in some countries while the reverse holds true in 
some other countries. The expected direction of industrial/residential electricity price ratio 
changes due to reform depends on the starting point, county-specific conditions and the 
relative strength of interest groups; and therefore cannot be hypothesized theoretically. The 
following hypothesis is therefore relevant on the effect of reform on industrial/residential 
electricity price ratio. 
66 
 
 
Hypothesis 2: Power market reforms have a statistically significant impact on 
industrial/residential electricity price ratio. 
 
As mentioned above, electricity reforms are expected to influence industrial/residential price 
ratios. In this context, the change in industrial/residential price ratio constitutes a possible 
outcome and the electricity reform is the possible cause of this outcome. By developing this 
second hypothesis, we try to find out whether there is a correlation between electricity reform 
and industrial/residential price ratio. We may think about a causal link in the event that the 
coefficients of the variables representing reform steps turn out to be statistically significant. 
 
While considering the relationship between reforms and industrial/residential price ratio, the 
latter (dependent variable) is represented by the variable “industrial/residential price ratio”. 
As in the previous case, six dummy variables and their cross-products constitute causal 
variables that represent electricity market reforms. The details of these variables are provided 
in the following sections. 
 
3.3.3 Control variables 
 
Control variables are independent variables not directly related to electricity market reforms 
but still may explain a portion of the variations in price-cost margins and industrial/residential 
price ratios. In addition to electricity reforms, we expect that prices are affected by 
consumption levels, income level and transmission and distribution (T&D) losses. It is a very 
basic economic knowledge that prices are influenced by demand (consumption) and income. 
In some countries with high T&D loss levels, it may be necessary to raise electricity prices to 
secure some receipts. As a result of such considerations, we include following variables into 
67 
 
our analysis as control variables: electricity consumption by industry, electricity consumption 
by households, electricity losses in total supply, and GDP per capita. The details of these 
variables are presented in the following sections, too. 
 
3.4 Methodology 
 
It is almost impossible to observe the real impact of power market reforms on prices without 
separating the effects of market reform from variations in fuel costs and other country specific 
features. Therefore, instead of using prices directly in our analysis, we calculate electricity 
price-cost margins for each country and for each year and use this variable in our models as 
dependent variable. However, it is important to remember that what we refer as “price-cost 
margin” in this study is actually “electricity end use price - fuel cost margin”; it is not a 
measure of “economic profit”7 and, therefore, not expected to be zero. Electricity price-cost 
margin in this study includes items such as capital costs, transmission and distribution costs, 
accounting profit of the electricity utilities and so on. Since fuel costs are usually external to 
electricity industry (that is, fuel prices are determined by international markets), they should 
be separated from final electricity prices to observe the real impact of reforms on electricity 
price trends. Therefore, we deduct fuel prices from final electricity prices. We take into 
account only coal and natural gas import costs in the calculation of fuel costs because the cost 
of all remaining inputs (like those for nuclear and renewable power plants) are so low that 
they can be ignored. Additionally, apart from electricity prices, we also look at the impact of 
power market reforms on industrial/residential electricity price ratios. 
 
                                                          
7 Economic profit refers to the difference between the revenue received from the sale of an output and the 
opportunity cost of the inputs used. In the calculation of economic profit, opportunity costs are deducted from 
revenues earned. Therefore, at an optimum level, economic profits are expected to be zero. 
68 
 
We specify price-cost margins and industrial/residential electricity price ratios as a function of 
(i) electricity market reform indicators (dummy variables for individual reform steps and their 
cross-products), (ii) a set of controls (electricity consumption, transmission and distribution 
losses and income level), (iii) country-specific effects (these are assumed to be exogenous and 
to exist independently of reform process, but may explain a portion of the variation in 
electricity prices) and (iv) other unobserved variables that influence electricity price-cost 
margins and industrial/residential price ratios. These variables are then used in panel 
regressions to assess their impact on price-cost margins and industrial/residential price ratios. 
In panel regressions, the exploitation of both cross-country and time-series dimensions of the 
data allows for control of country-specific effects. Apart from reform process, price-cost 
margin and industrial/residential price ratio in a specific country and year are expected to be 
influenced by some other variables like electricity consumption, income level, transmission & 
distribution losses and so on. In our models, we include these variables in order to isolate the 
effect of the reforms on price-cost margins and industrial/residential price ratios. Besides, 
prices for industrial consumers are usually supposed to be more cost-reflective than prices for 
households. Hence, in our analysis, we make a distinction between industrial and residential 
electricity prices. 
 
Panel data analysis has three more-or-less independent approaches: (i) independently pooled 
panels, (ii) random effects models, (iii) fixed effects models. The selection between these 
methods depends upon the objective of our analysis and relevant tests developed to determine 
which one is the most appropriate. We start with the standard panel data regression equation 
below to analyse the impact of electricity industry reforms on industrial and residential 
electricity price-cost margins and industrial/residential price ratios. 
 
69 
 
 1
2 1
k s
it j jit p pi it
j p
Y X Z tβ β γ δ ε
= =
= + + + +∑ ∑  (3.1) 
  
In the model, i and t represent unit of observation and time period, respectively. j and p are 
indices used to differentiate between observed and unobserved variables. Xji and Zpi represent 
observed and unobserved variables, respectively. Xji includes both reform indicators and 
control variables. Yit is dependent variable (that is, electricity price-cost margins and 
industrial/residential price ratios). itε is the disturbance term and t is time trend term. Because 
the Zpi variables are unobserved, there is no means of obtaining information about the 
p piZγ∑ component of the model. For convenience, we define a term iα , known as the 
unobserved effect, representing the joint impact of the Zpi variables on Yit. So, our model may 
be rewritten as follows: 
 
 1
2
i it
k
it j jit
j
tY X α δ εβ β
=
+ += + +∑  (3.2) 
 
Now, the characterization of the iα  component is crucially important in the analysis. If 
control variables are so comprehensive that they capture all relevant characteristics of the 
individual, there will be no relevant unobserved characteristics. In that case, the iα  term may 
be dropped and pooled data regression (OLS) may be used to fit the model, treating all the 
observations for all time periods as a single sample. However, since we are not sure whether 
control variables in our models capture all relevant characteristics of the countries, we cannot 
directly carry out a pooled data regression of Y on X. If we were to do so, it would generate 
an omitted variable bias. Therefore we prefer to use either a Fixed Effects (FE) or Random 
Effects (RE) regression. In FE model, the country-specific effects ( iα ) are assumed to be the 
70 
 
fixed parameters to be estimated. In RE model, the country-specific effects ( iα ) are treated as 
stochastic. The fixed effect model produces consistent estimates, while the estimates obtained 
from the random effect model will be more efficient. There are more than 90 countries in the 
world where a reform process has been initiated but data is available only for 63 countries. 
That is, our sample is limited by data availability. Besides, electricity prices may or may not 
be country specific as in some regions there are regional electricity markets where prices are 
determined across countries. Therefore, we cannot be sure whether the observations in our 
model may be described as being a random sample from a given population; and cannot 
directly decide which regression specification (FE or RE) to use. It will be decided in the 
course of the analysis based on Hausman test. 
 
3.5 Overview of data 
 
Our data set is based on a panel of 63 countries for a period beginning in 1982 and extending 
through 2009. Year 1982 is selected as the starting date for the study because at that time 
electricity market reform was initiated for the first time in Chile. The final date, 2009, 
represents the last year for which data are available at the time the research is conducted. The 
data on electricity prices are not available for many countries for the period before 1982. That 
is, data availability also dictates the starting year. Besides, Chile initiated a large scale 
privatization program only in its electricity industry for the first time in 1982; but the reforms 
started in real sense in the UK in 1989. If we assume that the UK implemented the first reform 
program in full sense, our dataset has 7 years as base case. The countries in our sample are 
determined by data availability, especially by data on electricity prices for residential and 
industrial consumers and fuel costs in electricity generation. Since our panel dataset includes 
data on 63 countries for 28 years, the total number of maximum observations is 1,764 
(63x28). Because of the missing observations, our panel is unbalanced. 
71 
 
 
The variables used in the study are dummy variables representing individual reform steps and 
their cross-products, price-cost margin for industry and households, industrial/residential 
price ratio, electricity consumption by industry and households, electricity losses and income 
level (GDP per capita). We also divided all countries in our dataset into three groups 
(developed countries, developing countries in America and other developing countries) based 
on classification made by World Bank (2010a) and included a dummy variable for each group 
of country into our dataset. 
 
The dummy variables representing the existence of individual reform steps are as follows: (1) 
independent power producers (IPPs), (2) wholesale electricity market, (3) choice of supplier, 
(4) unbundling, (5) privatization, (6) electricity market regulator. In addition to these 
variables, we also include the cross-products of the last three variables into our analysis as we 
expect that combination of them may have different impact than when they exist alone. So, 
our additional three variables are (7) privatization and regulator, (8) privatization and 
unbundling, (9) unbundling and regulator. The dummy variables for reform steps are created 
using data collected and cross-checked from various international and national energy 
regulators’ web sites8. 
 
Data on electricity prices are obtained from International Energy Agency (IEA, 2010c) and 
Latin-American Energy Organization (OLADE, 2010). The unit of observation is current 
US$/kWh. Electricity price data are available separately for residential and industrial users 
and cover 63 countries. 
 
                                                          
8 The full list of sources from which data are obtained can be found at IERN web site (http://www.iern.net). 
72 
 
Fuel cost data are taken from IEA and consist of two sets of data on natural gas import costs 
(USD/MBtu) and coal import costs (USD/tonne) (IEA, 2010a, b). For US, Japan and South 
Korea, we use LNG import costs as natural gas import cost data while pipeline import costs 
are used for the rest. Also, we utilized average EU natural gas pipeline import prices as a 
proxy for natural gas import costs in the countries for which the natural gas import cost data 
are not available. Coking coal is required for production of coke used in steel industries and 
steam coal is used in thermal power plants for steam production. Since we are concentrating 
on electricity generation costs in our study, we used steam coal import costs in our analysis. 
Coal data is missing for some countries in our sample. We used average EU steam coal import 
costs as a proxy for coal import cost for Norway, Switzerland and EU member or candidate 
countries for which data are missing. For other countries with missing observations, we used 
OECD averages. As we take into account the fact that energy markets (including natural gas 
and coal markets) have been internationalized in the last two decades, utilization of average 
EU or OECD import prices as a proxy for import costs in other countries seems to be 
justified. 
 
Having collected data on end-user electricity prices and fuel import costs, we calculated price-
cost margins as follows. First of all, we converted electricity prices into US$/MWh by 
multiplying prices in US$/kWh by 1,000. Then, we converted the data on fuel import costs 
into a common unit, USD/MBtu. In the conversion process, we used the equation 1 MBtu ≈ 
0.036 tonne of coal equivalent. After conversion, we weighted these two variables by both the 
output of electricity from natural gas and coal within each country and year and heat rate9 of 
                                                          
9 The term “heat rate” refers to a power plant’s efficiency in converting fuel to electricity. Heat rate is expressed 
as the number of British thermal units (Btu) required generating a kilowatt hour (kWh) of electricity. Lower heat 
rates are associated with more efficient power generating plants. In the literature, spark spread refers to the 
theoretical gross income of a gas-fired power plant from selling a unit of electricity, having bought the fuel 
required to produce this unit of electricity. All other costs (operation and maintenance, capital and other financial 
73 
 
these two fuels. Data on electricity production from natural gas and coal are obtained from 
IEA (2010d). For instance, if we assume that data for a specific country and a specific year 
are as follows, price-cost margin for industry in this country and year is calculated as 82.2 
US$/MWh, as shown below. 
− Electricity price for industry: 145 US$/MWh 
− Natural gas import cost: 9 USD/MBtu 
− Coal import cost: 5 USD/MBtu  
− Electricity generation from natural gas: 175 TWh 
− Electricity generation from coal: 125 TWh  
− Heat rate for gas-fired plants: 8,000 Btu/kWh (= 8000/1000 Btu/MWh)  
− Heat rate for coal-fired plants: 10,000 Btu/kWh (= 10000/1000 Btu/MWh) 
 
 
9*(8000 1000)*175 5*(10000 /1000)*125
145 82.2
(175 125)
+
− ≈
+
 (3.3) 
 
In 2007, on average, 42.3% of total electricity generation came from natural gas and coal in 
our sample countries (IEA, 2010d) and in 20 of them, gas and coal were responsible for more 
than 65% of total generation. Nuclear, hydro and other renewable sources accounted for most  
 
                                                                                                                                                                                     
costs) must be covered from the spark spread. The term dark spread refers to the similarly defined difference 
between cash streams (spread) for coal-fired power plants. In short; spark/dark spread is the difference between 
the wholesale price of electricity and the cost of the fuel used to generate it taking into account the heat rate of 
each fuel. In our study, however, we calculate price-cost margin as the difference between end-user (not 
wholesale) electricity prices and fuel costs. Actually, price-cost margin varies between plants using different 
fuels and may vary even between plants using the same fuels. However, for simplicity, we assume a heat rate of 
10,000 Btu/kWh for coal-fired plants and 8,000 Btu/kWh for gas-fired ones (For more details see US EIA, 
2010a. Average Operating Heat Rate for Selected Energy Sources. U.S. Energy Information Administration.). 


)LJXUH(OHFWULFLW\HQGXVHUSULFHIXHOFRVWPDUJLQVLQDQG


-50
0
50
100
150
200
250
300
Ar
ge
nt
in
a
Au
st
ria
Br
az
il
Ca
na
da
Ch
ile
Co
lo
m
bi
a
Cz
ec
h 
Re
pu
bl
ic
De
nm
ar
k
Do
m
in
ic
an
 R
ep
.
Fr
an
ce
G
er
m
an
y
Hu
ng
ar
y
Ire
la
nd
Ita
ly
Ja
pa
n
Ko
re
a
M
ex
ic
o
N
et
he
rla
nd
s
N
ew
 Z
ea
la
nd
N
or
w
ay
Pe
ru
Po
la
nd
Po
rt
ug
al
Sl
ov
ak
 R
ep
ub
lic
Sp
ai
n
Sw
ed
en
Sw
itz
er
la
nd
Ta
iw
an
Th
ai
la
nd
Tu
rk
ey
U
ni
te
d 
Ki
ng
do
m
U
ni
te
d 
St
at
es
Ve
ne
zu
el
a
U
S$
/M
W
h 
Price-fuel cost margin for industry in 1987 Price-fuel cost margin for households in 1987
Price-fuel cost margin for industry in 2007 Price-fuel cost margin for households in 2007
75 
 
Table 3.1 Descriptive statistics of the variables 
Variables (Units) 
# of 
obs. 
# of 
countries 
Mean Std. Dev. Min. Max. 
Dependent Variables       
Price-cost margin for industry (US$/MWh) 1,127 54 40.90 28.12 -32.03 212.55 
Price-cost margin for households (US$/MWh) 1,179 54 74.18 50.77 -33.13 344.40 
Industrial/residential electricity price ratio 1,428 61 0.81 0.43 0.20 3.86 
Explanatory Variables       
Independent power producers (IPPs) 1,764 63 0.49 0.50 0 1 
Wholesale electricity market 1,764 63 0.27 0.44 0 1 
Choice of supplier 1,764 63 0.21 0.41 0 1 
Unbundling 1,764 63 0.37 0.48 0 1 
Privatization 1,764 63 0.35 0.48 0 1 
Electricity market regulator 1,764 63 0.40 0.49 0 1 
Privatization and regulator 1,764 63 0.29 0.45 0 1 
Privatization and unbundling 1,764 63 0.29 0.45 0 1 
Unbundling and regulator 1,764 63 0.33 0.47 0 1 
Control Variables       
Electricity consumption by industry sector (GWh) 1,614 63 68,257 159,064 41 1,867,656 
Electricity consumption by households (GWh) 1,614 63 43,490 137,925 0 1,392,241 
Proportion of loses in total supply (%) 1,614 63 11.05 7.52 0 55.87 
GDP per capita (current thousand US$) 1,650 63 11.81 13.61 0.20 109.90 
 
of the remaining generation. Since the fuel costs in nuclear power plants and renewable 
electricity generating facilities constitute a very limited portion of the total cost, we focus only 
on the fuel cost in natural gas or coal-fired power plants where fuel costs have the largest 
share in total cost. Figure 3.3 shows the changes in price-cost margins for industry and 
households during the last two decades in countries for which data are available. 
 
76 
 
In our study, we created industrial/residential electricity price ratio variable to measure the 
relative price of electricity for industrial and residential consumers. This variable is obtained 
by dividing industrial electricity prices by residential prices. 
 
Data on electricity consumption and transmission & distribution losses come from IEA 
(2010e). Data on GDP per capita are obtained from World Bank (2010b). Table 3.4 shows 
descriptive statistics of the variables in our analysis. 
 
3.6 Empirical analysis 
 
Throughout our analysis, we estimate three groups of models to explain electricity price-cost 
margins for industry & households and industrial/residential price ratios. Each group includes 
three models for specific country groups10. In total, we estimate 9 models.  
 
Since using logarithms of variables enables us to interpret coefficients easily and is an 
effective way of shrinking the distance between values, we transform price-cost margin, 
electricity consumption and income level variables into logarithmic form and use these new 
transformed variables in our models. Although using logs and fixed or random effects models 
may have important repercussions on the quantitative interpretation of the results, they do not 
have an impact on the interpretation of the results as we concentrate on the sign of the 
coefficients only. That is, we do not concentrate on the quantitative impact of an explanatory 
variable on the dependent variable; instead, we focus on the sign of the relationship (positive 
or negative) between the dependent and explanatory variables. 
                                                          
10 FE estimation results do not let us detect the differences between country groups as variables that do not vary 
over time (like dummies for separating country groups) are dropped in FE estimation. In order to observe 
possible differences between country groups, we estimate separate models for each country group. 
77 
 
 
We perform the empirical analysis by estimating the specification given in Equation 3.2 for 
each model11. However, as mentioned before, we cannot directly decide which regression 
specification (FE or RE) to use. Therefore, we apply Hausman test for fixed versus random 
effects in each model. To perform this test, we first estimate the fixed effects model (which is 
consistent) and store the estimates, then estimate the random-effects model (which is 
efficient) and run the test. Since we prefer 5% significance level, any p-value less than 0.05 
implies that we should reject the null hypothesis of there being no systematic difference in the 
coefficients. In short, Hausman test with a p-value up to 0.05 indicates significant differences 
in the coefficients. Therefore, in such a case, we choose fixed effects model. However, if p-
value from Hausman test is above 0.05, we cannot reject the null hypothesis of there being no 
systematic difference in the coefficients at 5% level. In such cases, Hausman test does not 
indicate significant differences in the coefficients. Therefore, we provisionally choose random 
effects. After that, we apply Breusch and Pagan Lagrangian Multiplier (BPLM) test for 
random effects in order to decide on using either pooled OLS or random effects in our 
analysis. This test is developed to detect the presence of random effects. In this test, the null 
hypothesis is that variances of groups are zero; that is, there is no unobserved heterogeneity, 
all groups are similar. If the null is not rejected, the pooled regression model is appropriate. 
That is, if the p-value of BPLM test is below 0.05, we reject the null, meaning that random 
effects specification is the preferred one. If it is above 0.05, we prefer pooled OLS 
specification to carry out our regression. Table 3.5 shows a summary of estimation results that 
present statistically significant coefficients and their standard errors. Full details of estimation 
results are provided in Appendix I-4 including estimation output, number of observations and 
countries included in each model estimation, results of Hausman and BPLM tests and 
preferred specifications based on these tests. 
                                                          
11 Throughout the paper, model estimations are carried out and cross-checked by Stata 11.1 and Eviews 7.1. 
78 
 
Table 3.2 Summary of estimation results 
Dependent Variables 
Log of price-cost margin  
for industry 
Log of price-cost margin  
for households 
Industrial/residential 
price ratio 
Explanatory Variables 
D
ev
el
op
ed
 C
ou
nt
ri
es
 
D
ev
el
op
in
g 
C
ou
nt
ri
es
 
in
 A
m
er
ic
a 
O
th
er
 D
ev
el
op
in
g 
C
ou
nt
ri
es
 
D
ev
el
op
ed
 C
ou
nt
ri
es
 
D
ev
el
op
in
g 
C
ou
nt
ri
es
 
in
 A
m
er
ic
a 
O
th
er
 D
ev
el
op
in
g 
C
ou
nt
ri
es
 
D
ev
el
op
ed
 C
ou
nt
ri
es
 
D
ev
el
op
in
g 
C
ou
nt
ri
es
 
in
 A
m
er
ic
a 
O
th
er
 D
ev
el
op
in
g 
C
ou
nt
ri
es
 
Existence of IPPs -0.116
*** 
(0.045) 
-0.641*** 
(0.180)     
 
0.126** 
(0.061) 
 
Wholesale Electricity Market 
 
0.746** 
(0.293) 
-0.361* 
(0.200)   
-0.474*** 
(0.152) 
   
Choice of Supplier 
     
0.951*** 
(0.174) 
0.099***  
(0.031) 
0.245** 
(0.143) 
-0.367*** 
(0.125) 
Unbundling 
  
0.533*** 
(0.190)  
0.695* 
(0.370)  
-0.1***  
(0.036) 
 
-0.217** 
(0.116) 
Privatization 0.188
** 
(0.074)   
0.172*** 
(0.057) 
1.030* 
(0.574)  
-0.108***  
(0.035) 
 
-0.314** 
(0.146) 
Existence of Market Regulator -0.238
** 
(0.100)    
-0.579** 
(0.235)  
-0.099**  
(0.047) 
  
79 
 
Privatization and Regulator 0.193
* 
(0.103)      
   
Privatization and Unbundling -0.280
*** 
(0.105)    
-1.512* 
(0.793)  
   
Unbundling and Regulator 
  
-0.913*** 
(0.284)    
   
Log of electricity consumption by industry -0.947*** 
(0.076)  
-1.295*** 
(0.179) 
NV NV NV 
0.201***  
(0.047) 
0.137** 
(0.067) 
 
Log of electricity consumption by households NV NV NV -1.217*** 
(0.069) 
-0.649*** 
(0.176) 
-0.806*** 
(0.144) 
 
0.143** 
(0.075) 
 
Electricity losses in total supply (%, 0-100) -0.094*** 
(0.013)  
0.049** 
(0.022) 
-0.039*** 
(0.009) 
0.030** 
(0.012) 
0.126*** 
(0.018) 
  
-0.05*** 
(0.014) 
Log of GDP per capita 0.617*** 
(0.046) 
0.865*** 
(0.177) 
1.263*** 
(0.164) 
0.925*** 
(0.041) 
1.361*** 
(0.145) 
0.926*** 
(0.146) 
-0.107***  
(0.026) 
-0.277*** 
(0.066) 
-0.394*** 
(0.088) 
Constant 12.592
*** 
(0.767) 
5.470*** 
(2.052) 
15.555*** 
(1.861) 
14.088*** 
(0.625) 
7.608*** 
(1.469) 
9.035*** 
(1.279) 
-1.525***  
(0.402) 
-0.911** 
(0.534) 
-1.907** 
(1.026) 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Standard errors are shown in parentheses () under coefficients. 
NV: Not a variable in this model. 
 
80 
 
It is not easy to draw conclusions about the impact of extensive electricity market reforms in 
various countries from empirical work that focuses on a single market or from other country-
specific anecdotal discussion of reform processes because neither type of study distinguishes 
the effects of reform from country-specific features. Therefore, our empirical approach was to 
take advantage of the diversity in electricity reform patterns in various countries and to 
control for a number of potential explanatory variables to predict three indicators: electricity 
price-cost margin for households, electricity price-cost margin for industry and 
industrial/residential price ratio. Panel analysis of price-cost margin and industrial/residential 
price ratio trends (using reform indicators, country macroeconomic and other structural 
features) offers objective evidence on the observed impact of reforms at a macro level. 
 
3.7 Discussion of the results 
 
The empirical results presented above are consistent with our hypotheses. So, we fail to reject 
our hypotheses. Reviewing our findings in more detail and in relation to the research 
hypotheses, we find that individual reform steps have statistically significant impact not only 
on price-cost margins for both industrial and residential consumers (Hypothesis 1) but also on 
industrial/residential electricity price ratio (Hypothesis 2). When we look at the results, we see 
that the signs of the coefficients for variables representing various reform steps differ in each 
model, meaning that we cannot observe a uniform pattern concerning the impact of individual 
reform steps on price-cost margins and industrial/residential price ratios, that is, different 
reform steps seem to have different impacts on price-cost margins and industrial/residential 
price ratios. The interpretation of the results in detail and their policy implications are 
provided in the following sub-sections. 
 
81 
 
3.7.1 Impact of reforms on price-cost margins for industry 
 
In the first group of models, our empirical findings suggest, first, that existence of 
independent power producers (IPPs) and electricity price-cost margins are negatively 
correlated for industrial users, meaning that participation of IPPs into generation market 
decreases price-cost margins, especially in developed countries and developing countries in 
America. Second, existence of wholesale electricity markets seems to decrease price-cost 
margins in developing countries in America while it has an increasing effect in other 
developing countries. Third, we could not detect statistically significant results for choice of 
supplier and unbundling. Only exception is that unbundling alone is found to increase 
industrial price-cost margins in other developing countries. However, with regulator, it has a 
decreasing effect. Similarly, unbundling alone does not have a statistically significant impact 
on industrial price-cost margins in developed countries; however, with privatization, it has a 
decreasing impact. Fourth, our analysis reveals that privatization and market regulator have 
statistically significant impacts only in developed countries. On their own, privatization 
increases industrial price-cost margins while existence of an electricity market regulator 
decreases them. If they exist together, they raise industrial price-cost margins in developed 
countries. 
 
Based on results from the first group of models, we clearly fail to reject Hypothesis 1 for 
industrial electricity price-cost margins. That is, industrial electricity price-cost margins are 
clearly affected as countries introduce reform steps. Since Hypothesis 1 is both an immediate 
application of the relevant economic theory (theory of regulation, agency and public choice 
theories) and a direct parallel with the empirical findings reported in the literature we 
summarized in Section 3.2, we may conclude that theory of regulation, agency and public 
choice theories are applicable to the study of the impact of reforms on price-cost margins and 
82 
 
that our findings confirm and strengthen the empirical findings reported in the previous 
literature that the reforms are correlated with price movements in electricity markets. 
 
The policy implications of these results may be summarized as follows. If a developed 
country wishes to reduce price-cost margins for industrial consumers, it should introduce IPPs 
and set up a market regulator in its power market and implement privatization and unbundling 
together. In developed countries, privatization without unbundling results in an increase in 
price-cost margins for industry. On the other hand, a developing country in America with the 
same objective should concentrate on introduction of IPPs, as this is the only reform step that 
reduces price-cost margins for industry; and definitely avoid setting up a wholesale market, 
which rises price-cost margins for industry. However, in other developing countries, existence 
of a wholesale electricity market seems to reduce price-cost margins for industry. Besides, in 
other developing countries, unbundling with regulator also reduces the margins while 
unbundling alone increases them. 
 
3.7.2 Impact of reforms on price-cost margins for households 
 
Unlike the first group of models, we could not detect any significant impact of IPPs on price-
cost margins for households in the second group of models. This result suggests that IPPs 
affect industrial prices only. When we take into account the fact that IPPs usually sell the 
electricity they produced to large industrial consumers, this result seems reasonable. Second, 
the establishment of wholesale electricity markets has a decreasing effect on residential price-
cost margins in other developing countries. Third, choice of supplier seems to increase 
residential price-cost margins in other developing countries. Fourth, on their own, unbundling 
and privatization raise residential price-cost margins. However, with privatization, unbundling 
has a decreasing effect on residential price-cost margins in developing countries in America. 
83 
 
Fifth, existence of a market regulator reduces residential price-cost margins in developing 
countries in America. 
 
Similar to the results from the first group of models, those from the second group of models 
fail to reject Hypothesis 1 for residential electricity price-cost margins. That is, residential 
electricity price-cost margins are also affected as countries introduce reform steps. Again, the 
results from the second group of models imply that theory of regulation, agency and public 
choice theories are relevant to the study of the impact of reforms on price-cost margins and 
that our findings verify and reinforce the empirical findings reported in the previous literature 
that the reforms are associated with price movements in power markets. 
 
The main policy implication from the results of the second group of models is that neither 
unbundling nor privatization alone reduces price-cost margins in developing countries in 
America. However, if they exist together, they cause a decline in price-cost margins for 
residential consumers. Existence of a market regulator further contributes to this trend. In 
other developing countries, wholesale electricity market again seems to be the key step to 
reduce price-cost margins for households. 
 
3.7.3 Impact of reforms on industrial/residential price ratio 
 
The signs of the coefficients for variables representing various reform steps and their cross-
products are important as they show how reforms have influenced the relative price of 
electricity for industrial and residential consumers. Any statistically significant positive 
coefficient for a specific reform step variable implies that this reform step resulted in an 
increase in the relative price of electricity for industrial consumers and a decline in the 
relative price of electricity for residential consumers. Likewise, a statistically significant 
84 
 
negative coefficient for a specific reform step variable means that this reform step resulted in 
a change in the relative price of electricity for the benefit of industrial consumers and at the 
expense of households. 
 
Our results do not suggest a statistically significant impact of the existence of wholesale 
electricity market on industrial/residential price ratio. On the other hand, the results show that 
the existence of IPPs changes industrial/residential price ratio for the benefit of residential 
consumers in developing countries in America. In developed countries and developing 
countries in America, choice of supplier alters price ratio for the benefit of residential 
consumers while it reduces relative price of electricity for industrial consumers in other 
developing countries. Besides, unbundling or privatization changes industrial/residential price 
ratio for industrial consumers and against residential ones in developed countries and other 
developing countries. Market regulator also results in a change in price ratio for the benefit of 
industrial consumers in developed countries. 
 
Based on results from the third group of models, we clearly fail to reject Hypothesis 2. 
Explicitly, power market reforms seem to have a statistically significant impact on 
industrial/residential electricity price ratios. So, according to our results, we confirm the 
correlation between electricity reform and industrial/residential price ratio. When we evaluate 
these results within the framework suggested by economic theory of regulation, it seems that 
the relative strength of industrial and residential consumers may be one of the determinants of 
the outcome of electricity market reforms. However, our results imply that different reform 
steps influence industrial/residential electricity price ratio differently. For instance, in 
developed countries, the existence of an electricity market regulator reduces the relative price 
of electricity for industrial consumers, which may mean that industrial consumers in 
developed countries have more influence on the regulator than residential ones and therefore 
85 
 
may manipulate it to put into practice policies that reduce the relative price of electricity for 
industrial consumers. On the other hand, our results show that choice of supplier reduces the 
relative price of electricity for residential consumers in developed countries, meaning that 
residential consumers may be more powerful in retail markets than the industrial consumers in 
developed countries. 
 
The results above may have important policy implications. They present whether and in what 
direction a specific reform step affects industrial/residential price ratio. For instance, if the 
government of a developed country wishes to alter industrial/residential price ratio for the 
benefit of industrial consumers, it should introduce unbundling, privatization and market 
regulator into its power market and avoid choice of supplier. 
 
3.7.4 Results from control variables 
 
We observe a negative relationship between industrial electricity consumption and industrial 
price-cost margin. This result implies that as industrial electricity consumption rises industrial 
price-cost margins decline. Similarly, we see a negative relationship between residential 
electricity consumption and residential price-cost margins, which suggests that residential 
price-cost margins decline as electricity consumption by households increases. From a public 
policy perspective, the last two results imply that regulators should be vigilant and take 
corrective measures, if necessary, when electricity consumption declines as in such a case 
both industrial and residential price-cost margins tend to increase, signalling possible excess 
profits made by the regulated electricity utilities. 
 
The results also show that any increase in electricity consumption changes 
industrial/residential price ratio for the benefit of residential consumers and at the expense of 
86 
 
industrial consumers in developed countries and developing countries in America. Electricity 
consumption does not have a statistically significant impact on price ratios in other 
developing countries. 
 
We see a positive relationship between income level and industrial and residential price-cost 
margins in all countries, meaning that an increase in income level causes price-cost margins to 
increase. So, the regulators should be watchful when income level increases in a country as in 
such a case both industrial and residential price-cost margins tend to increase, indicating 
possible excess profits made by the regulated electricity utilities. Besides, income level is 
negatively correlated with industrial/residential price ratio in all countries. This result signals 
that an increase in income level changes relative electricity price for the benefit of industrial 
consumers and at the expense of households. 
 
Proportion of electricity losses in total power supply has different impacts on industrial price-
cost margins in developed and developing countries. In developed countries, it decreases 
industrial price-cost margins while the opposite holds true in developing countries. As for the 
impact of electricity losses on residential price-cost margins, it seems to have an increasing 
impact for developing countries and decreasing one in developed ones. The last two points 
show that it is more costly to decrease losses in developing countries. Any effort by an 
electricity utility in a developing country to reduce losses may mean a decline in profit (i.e. 
price-cost margins). In order to encourage electricity utilities to reduce losses, regulators in 
developing countries should take additional measures as electricity utilities are very unlikely 
to take measures to reduce losses if there is no additional motivation provided by the 
regulators. 
87 
 
 
Our findings also reveal that the proportion of electricity losses in total supply is negatively 
correlated with industrial/residential price ratios in other developing countries. So, any 
increase in electricity losses in total supply changes price ratios for the benefit of industrial 
consumers in other developing countries. 
 
It should also be mentioned that dummy variables representing various reform steps have 
relatively weaker impact on price-cost margins and industrial/residential price ratios in almost 
all models. Based on our results, we may argue that electricity consumption, income level and 
network losses are more influential in explaining price-cost margins and industrial/residential 
price ratios than reform process. Finally, we see that country specific features tend to have a 
high power in explaining price-cost margins and industrial/residential price ratios. 
 
We considered adding another variable into our analysis, called “number of years passed 
since reforms started”, to measure the impact of the time since reforms were implemented. 
However, when we do so we face with two important problems. First of all, since our dataset 
is panel and covers the period 1982-2009, the values of this variable for pre-reform periods is 
not clear. For instance, the reforms started in the UK in 1989; therefore such a new variable 
takes the values 1 for 1990, 2 for 1991 and so on for the UK. However, for 1982, it should 
take either -7 (meaning 7 years left until reforms started) or 0 (representing reforms have not 
started yet). Second problem relates to the problem of multicollinearity. Since the new 
variable increases each year, it is highly correlated with other variables that usually increase 
year by year such as electricity consumption and income level; and therefore its inclusion 
results in multicollinearity problem in our models. When we estimate our regressions twice by 
adding two versions of this new variable, one for each method to deal with pre-reform years; 
in both cases, we noticed not only that the coefficient of this new variable is not significant 
88 
 
even at 10% significance level but also that inclusion of this variable makes some of the 
previously significant variables insignificant. Multicollinearity may explain why the inclusion 
of new variable makes some of the previously significant variables insignificant. Therefore, 
based on the results from re-estimated regressions, we decided not to include this new 
variable into our analysis. 
 
To summarize, based on our results, we could not argue that the reform process as a whole 
decreases or increases price-cost margins and changes industrial/residential price ratios 
towards a specific direction. Individual reform steps have diverse impacts on the price-cost 
margins and industrial/residential price ratios in different countries. However, our results 
clearly show that reform steps have a statistically significant impact on both price-cost 
margins and industrial/residential price ratios. 
 
3.8 Conclusion 
 
The true value of electricity reform is a matter of empirical testing rather than theoretical 
debate. Opponents of the reform may point to spectacular reform failures (e.g. California 
disaster), or its advocates may try to get general conclusions from some success stories of a 
few reforming countries (e.g. NordPool). However, what is really needed is a complete study 
of the impact of reforms within the context of a well-defined model construction. Besides, 
today, there are data on electricity market reforms going back about three decades and 
available data start to let us meaningfully establish which market model and industry structure 
optimize social welfare. This study tried to fill the gap by offering a macro level econometric 
analysis on the possible effects of reform process on electricity price-cost margins and 
industrial/residential price ratios. 
 
89 
 
Relevant economic theories and previous empirical research indicate that electricity prices are 
affected by the reforms. In this paper, we tried to find out whether this connection holds true 
when we take into account the fuel costs. Moreover, we also investigated whether there is a 
correlation between electricity reform and industrial/residential price ratio. Throughout the 
study, we focused on these issues by using empirical econometric models to observe the 
impact of electricity market reforms on price-cost margins and industrial/residential price 
ratios. Panel data from 63 countries covering the period from 1982 to 2009 were employed. 
As a result of the study, we could not detect a uniform pattern for the impact of reform 
process as a whole on price-cost margins and industrial/residential price ratios. Our results 
suggest that each individual reform step has different impact on price-cost margins and 
industrial/residential price ratios for each consumer and country group. In a word, our 
findings imply that similar reform steps may have different impacts in different countries, 
which supports the idea that reform prescription for a specific country cannot easily be 
transferred to another one with similar success. Therefore, while deciding whether to initiate a 
reform process or in the process of making decisions on the direction of an already initiated 
reform process, policy makers should take into account the fact that each reform step has a 
specific impact in each country based on each country’s specific circumstances. More than 
that, countries that try to adopt the power market structure of another country (for example, 
the adoption by developing countries of electricity industry model similar to the UK) will 
have very different performance characteristics than the original country because their 
country-specific conditions will be different. The main implication of our results is that 
transferring the formal and economic structure of a successful power market in a developed 
country to developing countries is not a sufficient condition for good economic performance 
of the electricity industries in developing countries. 
 
90 
 
Our conclusions do not necessarily involve a judgement on the overall success or failure of 
the reform process. The change in electricity price-cost margins and industrial/residential 
price ratios is just one of the impacts of the reform and the process should be judged based on 
its overall impact (not only its impact on price-cost margins and industrial/residential price 
ratios). What’s more, it may well be argued that the reform process has just started or is still 
under progress in many countries and today it is too early to measure its impact on price-cost 
margins and industrial/residential price ratios.  
 
Electricity reform is also usually a part of wider economic liberalization program in general 
and energy industry reform in particular. Power market reform affects and is affected by 
reforms in other energy and non-energy sectors. For instance, it is obvious that the progress in 
telecommunication reform has facilitated electricity reform, which in turn has contributed to 
the progress in gas market reform. That is, power sector reform is highly inter-dependent with 
wider reforms in other sectors of the economy and failure to harmonize inter-sector reforms 
may lead to power sector reform measures being ineffective. The success of power sector 
reforms especially in developing countries largely depend on the extent to which inter-sector 
reforms in the economy are synchronized. We did not take into account possible spill-over 
effects from or to other energy and non-energy sectors in this research mainly due to lack of 
essential data. It may however be argued that our results are not comprehensive because of the 
lack of investigation of the impact of inter-reform relationship on electricity price-cost 
margins and industrial/residential price ratios.  
 
The arguments above and similar ones cannot be rejected straight away. Moreover, it is 
obvious that present econometric evidence on the impact of the reform process is quite 
limited. So, there is a definite need for the continued analyses of the effect of reforms in the 
electricity industry. Much work needs to be done and there are ample opportunities for 
91 
 
research in this area. In many countries, power market reform is still an on-going process, a 
fact that also underlines the need for continued and up-to-date study. We believe that panel 
datasets rather than simple cross-section models should be used in future studies, preferably 
including pre- and post-reform data. Furthermore, so far, most of the studies have focused on 
a single reform element or outcome (e.g. reform steps, prices, performance, costs and so on) 
but there is a need for cross-country econometric studies measuring overall impact of the 
reform process. 
 
We admit that power market reform is complex and the evidence is difficult to evaluate. We 
also recognize that it is too early to reach any concrete judgment for future policy suggestions 
based on the results from this paper and other comparable studies. An exact reckoning of the 
long-term effects of reforms on price-cost margins and industrial/residential price ratios will 
require much additional study over longer periods of time. 
 
 
92 
 
 
 
 
 
 
Chapter 4 
 
4 Paper Two 
 
 
 
A cross-country analysis of electricity market reforms: potential 
contribution of New Institutional Economics 
 
4.1 Introduction 
 
One of the main objectives of any economic reform is to bring changes in the institutional 
arrangement so that economic activities can be performed more efficiently. Besides, 
reforming any sector in an economy requires changing the institutional environment, 
changing the organisational structure and modifying the governance mechanism. Since the 
late 1980s, power market reform has become the standard prescription of the multilateral 
donor agencies like the IMF and the World Bank and the reform program has been 
implemented vigorously for about three decades now. Although the content of each reform 
program has differed from one country to another, the policy of functional disintegration, the 
establishment of regulatory authorities, the formation of wholesale and retail power markets 
and the privatization of the electricity industry have been generally regarded as the natural 
93 
 
components of a reform program without paying much attention to the institutional 
environment of the country. 
 
Figure 4.1 presents the relationship between institutional structure and reform process. As can 
be seen in Figure 4.1, whole reform process takes place and is directly affected by the macro 
level institutional structure of the country in which the reforms are put into practice. The 
examples of macro level institutional structure of a country include its legal system, measures 
that guarantee security of property rights in this country, the degree of political and civil 
rights provided by the political regime, investment environment in the country and so on. 
Through reform measures, the pre-reform structure of a power market is transformed into 
post-reform structure. In general, pre-reform structure corresponds to public monopolies or 
regulated private monopolies and post-reform structure refers to a competitive electricity 
market where competition at retail or, at least, wholesale level is possible. Throughout the 
paper, the concept of “reform” refers to a specific process started in Chile in 1982 for the first 
time; so, we consider “regulated private monopoly” in pre-1982 period as a form of 
regulation, not reform. Usually, post-reform structure has some undesirable features that 
trigger further reforms in the power market. So, post-reform structure of the previous wave of 
reforms constitutes the pre-reform structure of the latter wave of reforms and the process goes 
on as such. These cycles of reforms produce economic, social, political and environmental 
impacts, which may have an impact on the decisions concerning the direction of reforms. 
 
Today, most countries have initiated some reform of their power sector despite the fact that 
not much progress has been made in many parts of the world, especially in developing 
countries. As reform pauses or progresses slowly, developing countries in particular face 
problems such as lack of adequate funding for new capacity addition, neglect of utility 
operation and management, and increase in government involvement in the management and 
94 
 
decision-making of the industry, contrary to the expected objectives of the reform 
(Bhattacharyya, 2007a). 
 
Figure 4.1 The relationship between institutional structure and reform process 
 
The purpose of this paper is to find out whether the question of why some countries are able 
to implement more extensive reforms is closely related to the question of why some countries 
have better institutions than others. This paper develops the empirical case that differences in 
institutions are the fundamental cause of differences in the extent of the reforms implemented 
in each country. In this paper, we focus only on the impact of institutional factors on the 
reform progress while the third paper investigates the impact of political economic 
environment on the reform progress. For instance, the impact foreign influence originating 
95 
 
from foreign financial aid and/or assistance by international donor agencies (e.g. World Bank) 
is studied in the third paper. 
 
The interest and motivation for this topic arises from the relatively recent agreement that has 
emerged among scholars in regarding institutions as a key factor shaping the outcome of an 
economic transformation. This objective is pursued by discussing implications of the 
conceptual framework proposed by the New Institutional Economics for power market 
reform. Besides, evidence resulting from an econometric empirical analysis that investigates 
the relationship between institutions and reforms is presented as well. 
 
The reform experience so far (especially in developing countries) suggests two consistent 
findings. First, institutional endowments of a country (such as judicial independence, integrity 
of the legal system, protection of property rights, legal enforcement of contracts and degree of 
polity) largely determine the extent of the reforms (Acemoglu et al., 2008; Acemoglu et al., 
2001, 2005). Second, despite the different approaches in the design of regulatory institutions, 
a separate agency from the government with reasonable levels of autonomy and technical 
expertise has emerged as the preferred model for a regulatory institution. Due to path 
dependency, the chairperson of electricity market regulatory agency when reforms started or 
were considered and the governor/minister responsible for energy policy at that time may play 
a critical role in the process. Therefore, in this study, we focus on general institutional 
endowments and backgrounds of the chairperson and the minister/governor as key factors 
explaining differences in the extent of the reforms implemented in various countries. 
 
We try to answer the following research questions: (i) do differences in institutional structures 
of countries play an important role in explaining how far reforms have gone in these 
countries? (ii) if they do, how do specific institutional endowments of a country affect its 
96 
 
reform performance? (iii) does the background of the chairperson of the regulatory agency 
when reforms started or were considered or that of the governor or minister responsible for 
energy policy at that time have an impact on reform progress? 
 
At this point, it is important to note that we focus on “reform progress” rather than “reform 
success” in our study. In our analysis, the term “progress of reform” does not indicate an 
assessment of success; and the phrase “progress in reform process” constitutes an evaluation 
of the extent of the reform at one point in time, rather than the change in the extent of the 
reform over a period of time. 
 
The paper proceeds as follows. The next section provides a conceptual framework and 
literature review on the New Institutional Economics. Section 4.3 develops research 
hypotheses. Section 4.4 summarizes the methodological framework. Section 4.5 and Section 
4.6 describes data and presents empirical analysis, respectively. Following section discusses 
the results. The last section concludes. 
 
4.2 Conceptual framework and literature review 
 
In recent years the role of institutions in promoting and sustaining economic change has been 
an issue of interest for both theoretical and empirical analyses. The main question is “what 
determines the divergent patterns of evolution of countries or economies over time?” Africa’s 
disappointing economic performance, the East Asian financial crisis, and the weak record of 
the former Soviet Union have also contributed to an increasing focus on the role of 
institutions in determining a country’s economic growth and performance (Aron, 2000). 
Within this context, New Institutional Economics (NIE) has emerged as the body of economic 
thought that considers institutions to be relevant to economic theory, and criticizes the neo-
97 
 
classical mainstream for having pushed them out of the discipline; it deals especially with the 
nature, origin and evolution of institutions, and their effects on economic performance 
(Chavance, 2009). 
 
The increasing focus on NIE is also evident in World Bank publications. World Bank (1997, 
2002) recommends that states develop strong regulatory mechanisms to encourage legal 
accountability, minimize corruption, and foster competition via privatization. The World 
Bank regards privatization as a solution to rent-seeking behaviour of corrupt officials. In 
response to the bureaucracy’s drain on public resources, competition, it is argued, will raise 
the transaction cost of seeking protection and subsidy from the state, and henceforth promote 
efficiency between firms. 
 
Presenting an extensive literature review on NIE is both outside the scope of this paper and 
not possible given limitations on the length of the study. Therefore, in this section, we 
summarize the main characteristics of NIE, mention its difference from “old” institutional 
economics, review central themes in NIE (such as property rights, transaction costs, path 
dependency and the difference between institutions and organizations) and cite main 
criticisms against it. Appendix II-1 presents some examples of empirical work based on NIE. 
Although there is some academic work that investigates the impact of institutions on 
electricity market reform practices (e.g. Haney and Pollitt, 2011); to the best of our 
knowledge, this study constitutes the first empirical work that focuses on the possible 
implications of NIE for electricity market reform. 
 
Oliver Williamson is the inventor (in 1975) of the term ‘new institutional economics’, which 
from the 1990s on came to refer to idea that ‘institutions matter’ and that these can be 
analysed (Chavance, 2009). New institutional economics abandons the standard neoclassical 
98 
 
assumptions that individuals have perfect information and unbounded rationality and that 
transactions are costless and instantaneous (Ménard and Shirley, 2008). The NIE starts from 
the reality that information is rarely complete, and transactions thus have costs associated 
with them, such as costs of finding out what the relevant prices are, of negotiating and of 
concluding contracts, and then of monitoring and enforcing them. Institutions are broadly 
defined as means of reducing these information and transaction costs (Harriss et al., 1997). 
So, the NIE can be seen as a development of neo-classical economics to include the role of 
transaction costs in exchange and so to take account of institutions as critical constraints on 
economic performance. For new institutionalists, the performance of a market economy (or 
scale and scope of an economic reform process) depends upon the formal and informal 
institutions and modes of organization that facilitate private transactions and cooperative 
behaviour. 
 
Douglass North is a particularly significant exponent of the NIE. The main message stemming 
from North’s analysis is that institutions affect economic performance by influencing the level 
of transaction costs and, hence, the feasibility and profitability of engaging in economic 
activity. In other words, institutions determine the opportunity set and provide a stable 
structure to human interaction by reducing uncertainty (North, 1990). For him, institutions are 
not necessarily or even usually created to be socially efficient; rather they are created to serve 
the interests of those with greater political and economic power (Acemoglu et al., 2005). 
 
The NIE is ‘new’ because there is an older school of institutionalism in economics. The old 
institutionalists (e.g. Thorstein Veblen, John R. Commons, Wesley Mitchell, Clarence Ayres) 
did not attribute so significant and beneficial a role to rational choice, scarcity and 
competition as orthodox economic theory did. The assumption of standard economic theory 
that transaction costs were zero was a great analytical convenience and, for a long time, went 
99 
 
largely unquestioned. Upon pushing the logic of zero transaction costs to completion, 
however, serious gaps, errors, and anomalies were exposed by Ronald Coase, who was the 
first to perceive and demonstrate the conceptual problems that resided therein (Ghertman and 
Ménard, 2009). Therefore, according to many scholars, the origins of NIE can be traced back 
to Coase (1937), whose most important message was that when it is costly to transact, 
institutions matter (Gagliardi, 2008). While the new institutionalists reject the neoclassical 
assumption of perfect information and instrumental rationality, they put more emphasis on 
orthodox assumptions of scarcity and competition than the old institutionalists did. 
 
Having briefly mentioned the basic idea behind NIE let me focus on central themes in it. NIE 
assumes the existence of a fundamental relationship between property rights and transaction 
costs, on the one hand, and property rights and institutions, on the other. The establishment 
and maintenance of property rights entail transaction costs and property rights are institutions 
by themselves (Musole, 2009). So, property rights, which define the use, income rights and 
transferability of assets, constitute the core of the economic institutions; and the concept of 
transaction costs is central to the discussion of property rights. When rights are not clearly 
defined, transaction costs increase and market failures occur. For new institutionalists, the aim 
of institutions is to reduce transaction costs so as to allow agents to seize on economic 
opportunities, and an efficient institution is simply an arrangement that minimizes such costs, 
or one which maximizes the joint wealth of all the parties concerned net of transaction costs 
(Brousseau and Glachant, 2008). 
 
Actually, there is no consensus on how transaction costs should be defined. Instead, several 
definitions exist in the literature. For example, transaction costs have variously been defined 
as the cost of using the price mechanism; the costs of exchanging ownership titles; the costs 
of running the economic system; the costs associated with the transfer, capture and protection 
100 
 
of rights; the costs of measuring valuable attributes of that which is being exchanged, as well 
as the costs of monitoring and enforcing agreements; the ex-ante costs of drafting, negotiating 
and safeguarding an agreement and the ex-post costs of haggling, contract governance, and 
bonding costs to secure commitment, the resources used to establish and maintain property 
rights; or simply the ‘economic equivalent of friction in physical systems’ (Musole, 2009).  
 
North (1990) considers transaction costs as partly market costs and partly the costs of time 
that each party must devote to gathering information, to searching, and so on. In addition to 
this categorisation, North mentions a type of transaction cost that does not go through the 
marketplace, called “non-market transaction costs”. This type of transaction cost includes not 
only the high costs of searching where information is not efficiently distributed, and the 
substantial costs of undertaking economic activity in compliance with rules and regulations; 
but also the costs of queuing, bribing officials, cutting through red tape, time involved in 
obtaining permits to do business, and so forth. Besides, it is argued that transaction costs not 
only exist but also they are huge. It is thought that transaction costs may represent about 50-
60% of net national product of modern market economies. In less developed economies, 
transaction costs are thought to make up an even higher fraction of the overall GDP, and 
sometimes no exchange takes place due to these high costs (Musole, 2009).  
 
Another central theme in NIE is path dependency. For institutionalists, the process of 
institutional change is incremental and largely path dependent. Path dependence implies that 
if the process that leads to the emergence of a particular set of institutions is relevant and 
constrains future choice, then not only history matters but, more important, poor performance 
and long-run divergent patterns of development are determined by the same source. Path 
dependence may explain why some countries succeed and others do not (Gagliardi, 2008). So, 
path dependency is a process that can lead either to good or bad outcomes. Therefore, getting 
101 
 
the institutions right is critical because getting them wrong can lead to path-dependency, 
whereby inefficient economic systems persist. Relatively inefficient paths can, for example, 
be persistently followed over fairly long historical periods. In fact, according to North, this is 
the case most frequently found in history (North, 1990). The NIE approach suggests that the 
differences in economic performance are related to institutional endowments. The economic 
performance of a country depends on whether an independent judiciary, clearly defined 
property rights, control structures for enforcing property rights and enforceable contracting 
arrangements exist or not. Given the institutional environment, the opportunities provided by 
the institutional environment will be reflected in the nature and performance of organisations 
that develop. Hence transition from one state to another is constrained by the institutional 
arrangements (Bhattacharyya, 2007a). In short, NIE maintains that once an economy is on an 
“inefficient” path that produces stagnation it can persist (and historically has persisted) 
because of the nature of path dependence. 
 
Unlike ‘old’ institutional economics, North (1990) states that it is essential to distinguish 
institutions from organizations. For him, if institutions are the rules of the game, organizations 
and their entrepreneurs are the players. Organizations are groups of individuals bound by 
some common purpose to achieve objectives. They include political bodies (e.g. political 
parties), economic bodies (e.g. firms, trade unions, regulatory agencies), social bodies (e.g. 
clubs, associations) and educational bodies (e.g. schools, universities). Moreover, North 
(1990) argues that institutions and organizations are interdependent entities. The institutional 
framework determines the opportunities in a society and the emergence and evolution of 
certain organizations. In turn, organizations influence how institutions evolve, thus being 
agents of institutional change. 
 
102 
 
As mentioned before, there is no previous empirical study that applies NIE approach to the 
analysis of power market reforms, so we cannot provide a review of previous studies similar 
to this one. Appendix II-1 presents details of some examples of applied work based on the 
NIE approach, including hypotheses tested, dependent variables, explanatory variables, 
results, data and methodology. Appendix II-2 classifies these studies by their focus. Besides, 
in applied NIE studies, various variables are used to measure different aspects of institutional 
structure. Basically, there are three aspects of institutions: the presence of institutions, the 
organization (or form) of institutions or the outcome of institutions. The presence of 
institutions concerns whether specific institutions exist, without paying attention to their 
organization or outcome. The organization (or form) of institutions is to do with the way 
institutions are actually operated. The outcome of institutions is to do with an overall 
assessment of the impact of the institutions on the performance of the countries or industries 
(Green et al., 2009). Variables in previous studies either measure one of these aspects or are 
used as control variables. Appendix II-3 presents variables employed in previous econometric 
studies by what they measure. 
 
The examples of the previous studies listed in Appendix II-1 show that NIE approach has the 
potential for application in very diverse areas. In the following sections, we will present the 
first empirical study that analyses electricity market reforms with tools provided by NIE. 
 
4.3 Hypothesis development 
 
The literature that we summarized above shows that institutions matter for any economic 
reform and electricity market reform is not an exception. In essence, electricity market reform 
is an institutional reform that necessitates de facto or de jure regime change, creation of new 
institutional structures and rearrangement or removal of existing ones. Institutions may 
103 
 
determine the divergent patterns of evolution of reform processes in various countries over 
time. In the literature, the relationship between institutions and economic transformations has 
been investigated by many scholars. For instance, as we mentioned in the previous section, 
the link between political institutions and economic change is explored by Aghion et al. 
(2008), Alesina et al. (1996), Alesina and Rodrik (1994), Besley and Kudamatsu (2008), 
Caselli et al. (1996), Clague et al. (1996), Drury et al. (2006), Helliwell (1994), Isham et al. 
(1997), Mauro (1995), Persson and Tabellini (2008) and Scully (1988); while the impact of 
economic institutions on economic change is investigated by Assane and Grammy (2003), 
Barro (1991, 1996, 2000), Keefer and Knack (1997), Persson and Tabellini (1994), Spindler 
(1991) and Vanssay and Spindler (1994). In short, the NIE approach suggests that the 
differences in performances of different reform processes are related to institutional 
endowments. This section of the paper develops three hypotheses derived from the NIE 
approach, which are then tested empirically in the next sections of the paper. 
 
4.3.1 Hypothesis on the background of key policy makers and implementers 
 
One of the main repercussions of NIE relevant to electricity reform is that the process of 
electricity market reform is largely path dependent, which may explain why some countries 
succeed and others do not in reforming their power sectors. So, getting the institutions right is 
critical to reform success as getting them wrong can lead to path-dependency, whereby 
inefficient electricity markets may persist. So, to prevent inefficient institutional structures in 
the subsequent reform phases, the utmost attention should be paid to arrangements at the very 
beginning of the reform programs. Right people should set up right structures. In this context, 
the chairperson of the electricity market regulator and the minister responsible for energy 
policy when reforms started may have an important impact on subsequent reform progress. 
 
104 
 
The ministers responsible for energy-related issues in countries or governors in US states or 
Canadian provinces set general policies for electricity industry and the regulatory agencies put 
these policies into practice. Both policy setting and policy implementation are crucial factors 
that explain the reform progress in any country. Besides, path dependency implies that the 
chairperson of electricity market regulatory agency when reforms started or were considered 
and the governor or minister responsible for energy policy at that time play a critical role in 
the progress of subsequent reform process. For instance, in Argentina, Carlos Bastos, 
Secretary of Energy between 1991-96, led the privatisation of the electricity sector within the 
general policy framework of the Minister of Economy. Bastos was formerly an electrical 
engineer, researcher and a consultant on electricity issues for the Inter-American 
Development Bank and the Harvard Institute for International Economic Development. He 
brought the conceptual vision and insistence on a reformed, privately owned and competitive 
sector. He gave general direction and control to the privatisation of the energy sector, and 
took on the political battles, including with parties from the existing industry. The reform was 
along similar lines to the UK, and even went further with respect to restructuring (Littlechild 
and Skerk, 2004). 
 
Similarly, UK has been successful in market reform because it managed to find a set of quite 
able, fair-minded regulators. Prof. Stephen C. Littlechild was Director General of Electricity 
Supply (DGES), in charge of the Office of Electricity Regulation (OFFER), from its 
foundation in September 1989 to 1998. Littlechild, one of the architects of the successful UK 
electricity reform, has been a true believer in competition in electricity markets. Before the 
appointment, he was Professor of Commerce and Head of Department of Industrial 
Economics and Business Studies at the University of Birmingham from 1975 to 1989, and a 
member of the Monopolies and Mergers Commission from 1983 to 1989. In response to the 
apparent problems of the cost-recovery methods, in 1983, Professor Littlechild proposed a 
105 
 
“high-powered” incentive scheme, popularly known as RPI-X or price cap, in which the 
regulator caps the allowable price or revenue for each firm for a pre-determined period. Thus 
far, in terms of economic efficiency, RPI-X has been a clear success. In the United Kingdom, 
the RPI-X regulatory approach has induced cost reductions well beyond expectations. 
Electricity companies have been able to greatly reduce operating costs in large part through 
substantial work force reductions. To sum up, the educational and professional backgrounds 
of energy minister and regulator played an important role in the reform progress in Argentina 
and the UK, respectively. 
 
As summarized in Dreher et al. (2009), there is also a growing literature connecting personal 
traits of policy makers and implementers with policy outcomes. Jones and Olken (2005) find 
that unpredictable changes in a country’s leadership due to the incumbent’s death can trigger 
changes in gross domestic product growth. Göhlmann and Vaubel (2007) provide recent 
empirical evidence on the impact of education and profession on policy. Their results show 
that education and profession of the central bank’s governing council members matter for the 
effectiveness in controlling inflation. Regarding education of policy makers, Duflo (2005) 
shows that reservation of political power for historically disadvantaged groups and women in 
India does not come at the expense of the quality of decision making, even though reservation 
brings to power a group of relatively inexperienced and less educated politicians. Besley et al. 
(2005), to the contrary, use household survey data from India and find that differences in the 
performance of Indian village politicians are systematically linked to politicians’ education. 
This evidence leads a corresponding World Bank (2005) report to conclude “that more 
educated politicians are better” adding to “a growing appreciation among economists that 
education [of politicians] may be important because of its role in inculcating civic values.” 
Besides, previous research suggests that policy outcomes may depend on the time a politician 
106 
 
spends in office. For example, Roubini and Sachs (1989) suggest that there is a tendency 
towards larger deficits in countries characterized by a short average tenure of government. 
 
To sum up, policy makers decide on policies and reform strategies and policy implementers 
put them into practice. When it comes to designing and implementing a reform policy, 
politicians’ education and professional experience is likely to be important. As Kotsogiannis 
and Schwager (2006) maintain, “the implementation of new and unknown policies is more 
demanding than running ‘business as usual’ since it requires imaginative leadership on the 
part of a governor, rather than operational routine.” Politicians with a certain educational or 
professional background may be more likely than others to demonstrate such kind of 
leadership. These politicians thus may have an advantage in fostering reforms. According to 
Rajan (2004), for example, the “gains from reform are never as clear to the wider public as 
they are to economists.” Hence, reforms might be delayed due to a lack of understanding and 
education. Some background education in economics could be advantageous for politicians in 
implementing reforms as they are more likely to distinguish good from bad advice and might 
be more able to resist the pressure of lobbying groups preferring the status quo. This 
knowledge might bring them into a better bargaining position as well. They might also have 
an advantage in communicating the consequences of reforms to the public and the parliament, 
thus decreasing uncertainty and overcoming the deadlock (Dreher et al., 2009). These 
considerations lead us to posit the following hypothesis. 
 
Hypothesis 1: Due to path-dependency, the background of the chairperson of the 
regulator and that of the governor or minister responsible for energy policy when 
reforms started have an impact on overall reform progress. 
 
107 
 
The New Institutional Economics and the literature summarized above suggest that personal 
qualifications may matter for the reform progress and some applied studies mentioned above 
confirm this. In our analysis, we check whether this is also the case for the electricity industry. 
We do not support or oppose the idea that personal qualifications are important determinants 
of any reform progress. Hence, Hypothesis 1 is both an immediate application of the NIE and 
a direct parallel with the empirical findings reported in the literature we cited above. While 
examining the relationship between the backgrounds of policy makers and implementers and 
electricity market reforms, the latter (dependent variable) is represented by the electricity 
market reform score variable. On the other hand, variables on experience, length of term, 
education level and educational background of chairpersons and ministers constitute causal 
variables that represent backgrounds of policy makers and implementers. The details of these 
variables are provided in the following sections. Like most of other similar analyses; we 
assume that experience in the industry, length of term, education level, educational 
background reflect personalities of policy makers and implementers. Of course, there may be 
some other relevant aspects of the personalities that cannot be measured. However, since we 
cannot measure such aspects, we cannot include them into our analysis. 
 
We expect a statistically significant relationship between the backgrounds of policy makers 
and implementers and electricity market reform progress; however, direction of this 
relationship cannot be predicted theoretically. So, we do not develop hypotheses on the 
direction of this relationship. 
 
Appointment of regulators, their term time, salaries and additional benefits, rules concerning 
their removal and other characteristics of the office of regulators may be important 
determinants of the reform progress. However, these are usually not determined freely by 
countries. In the case of developing countries, establishment of regulatory institutions are 
108 
 
usually requested by international donor agencies (like World Bank and IMF); and in the case 
of EU member states, the EU requires them to set up regulatory agencies. Both international 
donor agencies and the EU also specify the characteristics of the office of regulators (their 
appointment, removal, term time, salaries, other benefits and so on). So, since characteristics 
of the office of regulators do not reflect genuinely the political attitudes in specific countries, 
we prefer not to include them into our analysis. 
 
4.3.2 Hypothesis on the impact of democracy on the reforms 
 
According to the NIE theory, institutions are means of reducing information and transaction 
costs related to reform design and implementation; and they may easily turn into critical 
constraints on reform performance if not taken into account properly. So, the performance of 
a reform program largely depends upon the formal and informal institutions, which affect the 
reforms by influencing the level of transaction costs and, hence, the feasibility of engaging in 
a reform initiative. Besides, the NIE approach maintains a fundamental relationship between 
property rights, transaction costs and institutions. Institutional environments that fail to offer 
credible commitments against arbitrary changes in the rules of the game, including 
expropriation, raise transaction costs throughout the economy. Also, when property rights are 
not clearly defined in the course of an electricity market reform, transaction costs increase and 
reforms may fail. In this context, democracy is one of the most important institutions in a 
country that determines the credibility of commitments concerning property rights and 
transaction costs; and therefore, is expected to influence power market reforms. Based on the 
considerations above, we develop the hypothesis below. 
 
Hypothesis 2: The level of democracy in a country has an impact on reform 
progress. 
109 
 
 
This hypothesis is an immediate application of the New Institutional Economics. According 
to NIE approach, the progress in electricity market reforms constitutes a possible outcome and 
the level of democracy in a country may be one of the determinants of this outcome. By 
developing this hypothesis, we check whether this correlation actually exists. While 
examining the relationship between the level of democracy and electricity market reforms, 
reform progress is represented by the electricity market reform score variable while the level 
of democracy is signified by two variables, namely political rights score and civil liberties 
score. The details of these variables are provided in the following sections.  
 
Democracy is a form of government in which all eligible citizens have an equal say in the 
decisions that affect their lives. Democracy allows eligible citizens to participate equally -
either directly or through elected representatives- in the proposal, development, and creation 
of laws. So, democracy enables both pro- and anti-reform groups to voice their concerns 
about power market reforms. Therefore, the direction of the impact of democracy on 
electricity market reforms depends on the composition and relative strength of interest groups 
in each country, and therefore, cannot be established theoretically. 
 
4.3.3 Hypothesis on the impact of corruption on the reforms 
 
NIE suggests that the standard neoclassical assumptions that we have perfect information and 
unbounded rationality and that transactions are costless and instantaneous should be 
abandoned while analysing reforms in electricity markets. The NIE implies that information 
during whole reform process is rarely complete, and transactions related to reform process 
have costs associated with them, such as costs of finding out what and how to reform, of 
negotiating the reform direction with interested parties, of passing necessary legislation, and 
110 
 
then of monitoring and enforcing it. NIE pays attention also to non-market transaction costs 
faced by the firms. In addition to the costs of searching where information is not efficiently 
distributed and undertaking economic activity in compliance with rules and regulations; this 
type of transaction cost includes the cost associated with corruption in a country (e.g. bribing 
officials to cut through red tape). Besides, the literature on the factors that influence the 
decision to reform an infrastructure industry by introducing competition has emphasized the 
role of corruption of the political system. For instance, Emerson (2006) develops a model 
where rent seeking firms agree to pay bribes to policy makers to limit market entry. Since 
corruption increases transaction costs in a country, we expect a negative relationship between 
the level of corruption and reform progress. Therefore, we suggest the following hypothesis. 
 
Hypothesis 3: Countries with relatively high corruption tend to introduce less 
electricity market reforms. 
 
This hypothesis is also an immediate application of the New Institutional Economics. While 
examining the relationship between corruption and electricity market reforms, reform 
progress is represented by the electricity market reform score variable while corruption 
perceptions index represents the level of corruption in a country. The details of these variables 
are also provided in the next sections. 
 
4.3.4 Control variables 
 
Control variables are independent variables not directly related to the background of key 
policy makers and implementers, level of democracy or corruption; but still may explain a 
portion of the variations in reform progress. Along with the background of key policy makers 
and implementers, level of democracy and corruption; we expect that reforms may be affected 
111 
 
by the size of population, income level, investment environment, the level of authoritarianism 
and security of property rights in a country. We also expect that OECD membership may have 
an impact on the reform progress, and therefore include a dummy variable representing 
OECD membership into our analysis. The details of these variables are presented in the 
following sections, too. 
 
4.4 Methodology 
 
In our study, the scope of the reforms in each country is represented by the electricity market 
reform score variable. Therefore, in our analysis, we describe the electricity market reform 
score as a function of 
(a) the background of the chairperson of electricity market regulatory agency when 
reforms started or were considered (his/her experience in electricity industry, his/her 
length of term, his/her education level, his/her educational background in business or 
economics, in engineering or in law); 
(b) the background of the governor or minister who was responsible for energy policy at 
that time (his/her experience in electricity industry, his/her length of term after reforms 
started or were considered, his/her education level, his/her educational background in 
business or economics, in engineering or in law); 
(c) corruption perceptions index, civil liberties and political rights scores; 
(d) control variables (i.e. population, GDP per capita, dummy variable for being an OECD 
country, investment freedom index, polity score and property rights index). 
 
In our analysis, our dependent variable is limited, that is, it is a count variable, which can take 
on nonnegative integer values, . We cannot take the logarithm of a {0,1,2,3,4,5,6,7,8}Yi ∈
112 
 
count variable because it takes on the value zero. An appropriate approach is to model the 
expected value as an exponential function 
  (4.1) 
Since is always positive, the predicted values for y will also be positive. Although this 
is more complicated than a linear model, we know how to interpret the coefficients. Taking 
the log of Equation 4.1 shows that 
  (4.2) 
so that the log of the expected value is linear. Using the approximation properties of the log 
function, we write 
  (4.3) 
In other words,  is roughly the percentage change in E(y|x), given a one-unit increase in 
xi. 
 
Because Equation 4.1 is nonlinear in its parameters, we cannot use linear regression methods. 
We could use nonlinear least squares, which, just as with OLS, minimizes the sum of squared 
residuals. It turns out, however, that all standard count data distributions exhibit 
heteroskedasticity and nonlinear least squares does not exploit this. Instead, we will rely on 
maximum likelihood and the important related method of quasi-maximum likelihood 
estimation. Besides, as we know, normality is the standard distributional assumption for linear 
regression. The normality assumption is reasonable for continuous dependent variables that 
can take on a large range of values. A count variable cannot have a normal distribution 
(because the normal distribution is for continuous variables that can take on all values), and if 
it takes on very few values, the distribution can be very different from normal. Instead, the 
nominal distribution for count data is the Poisson distribution. A random variable Y, which 
only takes on nonnegative integer values, follows the Poisson distribution if, for k = 0, 1, 2, ... 
1 2 0 1 1( | , ,..., ) exp( ... )k k ky x x x x xβ β β= + + +E
exp( )⋅
1 2 0 1 1log[ ( | , ,..., )] ...k k ky x x x x xβ β β= + + +E
% ( | ) (100 )i iy x xβ∆ ≈ ∆E
100 iβ
113 
 
  (4.4) 
where . The mean and variance of Poisson random variable is  and  
  (4.5) 
Figure 4.2 shows the Poisson distribution for different  values. Because we are interested in 
the effect of explanatory variables on y, we must look at the Poisson distribution conditional 
on x. The Poisson distribution is entirely determined by its mean, so we only need to specify 
E(y|x). Then, the probability that y equals the value k, conditional on x, is (for k = 0, 1, 2, ...) 
  (4.6) 
  (4.7) 
  (4.8) 
Here, the interpretation of  is that when there is a one unit increase in Xi, the percentage 
change of is 100 x . This distribution, which is the basis for the Poisson regression 
model, allows us to find conditional probabilities for any values of the explanatory variables. 
 
Figure 4.2 Poisson distribution for different  values 
 
 
exp( )
Pr( )
!
k
Y k
k
λ λ−
= =
0λ > λ
( ) var( )Y Y λ= =E
λ
exp( )
Pr( | )
!
k
i i
i iY k X k
λ λ−
= =
0 1( | ) exp( )i i i iY X Xλ β β= = +E
0 1ln( )i iXλ β β= +
1β
( | )Y XE 1β
λ
114 
 
In principle, the Poisson model is simply a nonlinear regression. It is much easier to estimate 
the parameter with a maximum likelihood method. The log-likelihood function is  
 
  (4.9) 
  (4.10) 
  (4.11) 
where we can drop the term  because it does not depend on . So, we get 
  (4.12) 
While employing Poisson regression, we should keep in mind two important points. First, we 
cannot directly compare the magnitudes of the Poisson estimates of an exponential function 
with the OLS estimates of a linear function. Second, although Poisson analysis is a natural 
first step for count data, it may be restrictive. All of the probabilities and higher moments of 
the Poisson distribution are determined entirely by the mean. In particular, the variance is 
expected to be equal to the mean. This is restrictive but, fortunately, the Poisson distribution 
has a very nice robustness property: whether or not the Poisson distribution holds, we still get 
consistent, asymptotically normal estimators of the . 
 
Since interpretation of coefficients from a Poisson regression is not straightforward, the 
incidence rate ratios (IRR) are obtained by exponentiating the Poisson regression coefficients. 
When we use IRR option, estimated coefficients are transformed to incidence-rate ratios, that 
is  rather than . Standard errors and confidence intervals are similarly transformed. 
This option affects how results are displayed, not how they are estimated. As we discussed 
before, Poisson regression coefficients are interpreted as the difference between the log of 
0 1( , ;{ | } )1
NInL Y Xi i iβ β =
1
Pr( | )
N
i i i
i
Ln Y y X
=
= =∑
0 1 0 1
1
[ exp( ) ( ) ( !)]
N
i i i i
i
X Y X Ln Yβ β β β
=
= − + + + −∑
( !)iLn Y− β
0 1 0 1
1
[ exp( ) ( )]
N
i i i
i
X Y Xβ β β β
=
= − + + +∑
iβ
ieβ iβ
115 
 
expected counts. We also know that the difference of two logs is equal to the log of their 
quotient, log(a) – log(b) = log(a/b), and therefore, we could have also interpreted the 
parameter estimate as the log of the ratio of expected counts: this explains the “ratio” in 
incidence rate ratios. In addition, what we referred to as a count can also be called a rate. By 
definition a rate is the number of events per time (or space), which our response variable 
qualifies as. Hence, we could also interpret the Poisson regression coefficients as the log of 
the rate ratio: this explains the “rate” in incidence rate ratio. Finally, the rate at which events 
occur is called the incidence rate; thus we arrive at being able to interpret the coefficients in 
terms of incidence rate ratios. 
 
Because of the restrictions on the length of the paper and because it is not one of the aims of 
this paper, further details of Poisson regression is not presented here but available from 
Winkelmann (2008), Cameron and Trivedi (1998) and Wooldridge (2009). 
 
4.5 Overview of data 
 
Our data set is cross-section and covers 51 states in US, 13 provinces in Canada and 51 other 
countries12. In total, we have 115 potential observations for each variable. The sample 
countries and states in our analysis are determined by data availability. We have some missing 
observations in our dataset. We acknowledge that the states in US and provinces in Canada 
are not comparable to other countries. We are aware of the fact that analysing individual 
states and provinces together with countries create some practical and econometric problems. 
                                                          
12 Albania, Argentina, Armenia, Australia, Austria, Bangladesh, Belgium, Bosnia and Herzegovina, Brazil, 
Bulgaria, Cambodia, China, Colombia, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, 
Germany, Greece, India, Ireland, Italy, Jamaica, Kenya, Latvia, Lithuania, Luxembourg, Macedonia, Malaysia, 
Mexico, Netherlands, New Zealand, Nigeria, Norway, Pakistan, Philippines, Poland, Portugal, Romania, Russia, 
Serbia, Singapore, South Africa, Spain, Thailand, Turkey, Uganda, United Kingdom. 
116 
 
To avoid these problems, we provide results with and without states in US and provinces in 
Canada separately. 
 
The variables used in the study are electricity market reform score in 2011; experience of the 
chairperson of electricity market regulatory agency in electricity industry when reforms 
started or were considered, his/her length of term after that time, his/her education level, 
his/her educational background in business or economics, in engineering or in law; electricity 
industry experience of the governor or minister who is responsible for energy policy when 
reforms started or were considered, his/her length of term after that time, his/her education 
level, his/her educational background in business or economics, in engineering or in law; civil 
liberties score in 2011, political rights score in 2011; corruption perceptions index in 2010; 
population in 2010, GDP per capita in 2010, dummy for being an OECD country, investment 
freedom index in 2011, polity score in 2010, and property rights index in 2011. Table 4.1 
shows descriptive statistics of the variables. 
 
Table 4.1 Descriptive statistics of the variables 
Variables # of obs. Mean Std. Dev. Min. Max. 
Electricity market reform score in 2011 115 5.48 2.24 1 8 
Chairperson*      
His/her experience in electricity industry at appointment (years) 95 6.59 8.58 0 36 
Length of term (years) 100 4.97 3.05 0 14 
Education level (1: BSc, 2: MSc, 3: PhD) 94 1.89 0.99 0 3 
Educational background in      
 - Business or economics 94 0.40 0.49 0 1 
 - Engineering 94 0.20 0.40 0 1 
 - Law 94 0.44 0.50 0 1 
 - Other 94 0.15 0.36 0 1 
Energy minister/governor**      
His/her experience in electricity industry (years) 101 3.40 4.92 0 36 
Length of term (years) 106 3.48 2.06 0 10 
Education level (1: BSc, 2: MSc, 3: PhD) 103 1.64 0.95 0 3 
117 
 
Variables # of obs. Mean Std. Dev. Min. Max. 
Educational background in      
 - Business or economics 103 0.26 0.44 0 1 
 - Engineering 103 0.17 0.37 0 1 
 - Law 103 0.31 0.47 0 1 
 - Other 103 0.44 0.50 0 1 
Other explanatory variables      
Civil liberties score in 2011 115 1.60 1.15 1 6 
Political rights score in 2011 115 1.62 1.32 1 7 
Corruption perceptions index in 2010 115 6.37 2.06 2 9 
Control variables      
Population in 2010 (million people) 115 41.80 166.67 0.03 1,338.30 
Log of population in 2010 115 1.83 1.86 -3.40 7.20 
GDP per capita in 2010 (thousand $) 115 36.43 22.49 1.26 172.25 
Log of GDP per capita in 2010 115 3.32 0.91 0.23 5.15 
OECD country dummy 115 0.75 0.44 0 1 
Investment freedom index in 2011 115 70.83 12.94 25 95 
Polity score in 2010 114 8.79 3.13 -10 10 
Property rights index in 2011 115 72.61 22.33 20 95 
* The Chairperson refers to the chairperson of electricity market regulatory agency when reforms started or were considered. 
** Energy minister/governor refers to the governor or minister who was responsible for energy policy when reforms started or 
were considered. 
 
Electricity market reform score in 2011 variable takes the values from 0 to 8; depending on 
how many of the following reform steps have been taken in each country as of 2011: (1) 
introduction of independent power producers, (2) corporatization of state-owned enterprises, 
(3) law for electricity sector liberalization, (4) introduction of unbundling, (5) establishment 
of electricity market regulator, (6) introduction of privatization, (7) establishment of 
wholesale electricity market, and (8) choice of supplier. To build this variable, we created 8 
dummy variables for each of the reform steps mentioned above and calculated the total 
number of reform steps taken in each country. Dummy variables for reform steps are created 
based on the data collected and cross-checked from various international and national energy 
regulators’ web sites13. Figure 4.3 provides the histogram of the reform score variable 
                                                          
13 The full list of sources from which data are obtained can be found at IERN web site (http://www.iern.net). 


VKRZLQJ WKH IUHTXHQF\ RI REVHUYDWLRQVZKLOH )LJXUH  VKRZV FXUUHQW VWDWXV RI HOHFWULFLW\
UHIRUPLQ86VWDWHV86(,$E:KHQZHHYDOXDWH)LJXUHZHVHHWKDWDOOFRXQWULHV
LQRXUGDWDVHWKDYHWDNHQDWOHDVWRQHUHIRUPVWHSDQGPRUHWKDQKDOIRIWKHPKDYHWDNHQRU
PRUHUHIRUPVWHSV

)LJXUH+LVWRJUDPRIUHIRUPVFRUHYDULDEOH


)LJXUH(OHFWULFLW\UHIRUPLQ86VWDWHVDVRI


119 
 
Figure 4.4 seems to confirm Paul L. Joskow’s argument that the differences in electricity 
prices among US states explain why some US states introduce reforms and why others do not. 
As we mentioned in Section 2.2.1, Joskow argues that there are fairly dramatic differences in 
the electricity prices across US states and, in high-price states, there are strong incentives for 
consumers, and certainly for new producers, to press for reforms that provide competitive 
entry and possibility of price reductions (White, 1996). So, he maintains that states only with 
high electricity prices have an incentive to introduce reforms in the US and those with 
relatively low prices do not have such an incentive and therefore do not introduce reforms. 
Joskow (2000) shows that the states in the Northeast and California had much higher prices 
than the average for the rest of the country; and therefore, they were the first states to 
restructure their power sectors in 1996 and 1997. White (1996) also provides evidence that the 
states with relatively high electricity prices introduced reforms and differential prices explain 
the pattern of electricity reform in the US. 
 
Some countries considered introducing electricity market reforms but decided not to 
implement such reforms in their power markets. Some other countries also considered 
implementing reforms and decided to do so. For reforming countries, we collected data on the 
background of the chairperson of electricity market regulatory agency when reforms started, 
and the governor or minister who was responsible for energy policy at that time. For non-
reforming countries, we collected data on the background of the chairperson of electricity 
market regulatory agency when reforms were considered, and the governor or minister who 
was responsible for energy policy at that time. Data collection for these variables lasted 10 
months from February to November 2011. The data on chairperson include his/her experience 
in electricity industry, his/her length of term, his/her education level (BSc, MSc or PhD) and 
his/her educational background (business or economics, engineering, law, other/unknown). 
We also gathered data about the governor or minister who was responsible for energy policy 
120 
 
when reforms started or were considered. Similarly, these data include his/her experience in 
electricity industry, his/her length of term, his/her education level (BSc, MSc or PhD) and 
his/her educational background (business or economics, engineering, law, other/unknown). 
Here, two points are important. First, in our study, the length of term refers to time period that 
the regulator or minister/governor remained in office after the reforms started or were 
considered. Second, we are only interested in the regulators or ministers/governors when the 
reforms started or were considered; and collected data on their background. We are not 
interested in the regulators or ministers/governors in office when the data were collected, and 
our dataset do not include observations on them.  
 
The data on chairpersons and the ministers/governors are obtained from various reports and 
documents published by regulatory agencies and ministries of the countries. While deciding 
on which educational backgrounds to include into our analysis, we select the three most 
common backgrounds, namely business or economics, engineering and law. We also create an 
“other/unknown” category to represent other educational backgrounds. For instance, when we 
look at the educational backgrounds of chairpersons, we see that 36.6% of them have a 
background in law, 33.9% in business or economics, 17% in engineering and 12.5% in 
other/unknown educational backgrounds. In a very limited number of countries and for the 
last few years, there is a chief executive in regulatory institutions together with the chair of 
the agency. Since we concentrate on the chairperson of the regulatory agency when reforms 
started (which is usually 1990s), there was almost no chief executive at that time in regulatory 
institutions. In very few cases, there were chief executives when reforms started, and in these 
cases, we ignore them and concentrate on the chairs. Figure 4.5 shows the number of 
chairpersons and ministers/governors in the sample countries by their educational background 
while Figure 4.6 presents this by education level. 
 
121 
 
Figure 4.5 Educational backgrounds of chairpersons and ministers/governors 
 
 
Figure 4.6 Education level of chairpersons and ministers/governors 
 
 
Figure 4.7 provides the number of chairpersons and ministers/governors in the sample 
countries by their length of term. While evaluating Figure 4.7, it is important to keep in mind 
that length of term refers to length of term after the reforms started or were considered. 
 
122 
 
Figure 4.7 Length of term of chairpersons and ministers/governors 
 
Civil liberties and political rights scores for 2011 are taken from Freedom House (2011). The 
Freedom in the World survey conducted by Freedom House provides an annual evaluation of 
the state of global freedom as experienced by individuals. The survey measures freedom -the 
opportunity to act spontaneously in a variety of fields outside the control of the government 
and other centres of potential domination- according to two broad categories: political rights 
and civil liberties. Political rights enable people to participate freely in the political process, 
including the right to vote freely for distinct alternatives in legitimate elections, compete for 
public office, join political parties and organizations, and elect representatives who have a 
decisive impact on public policies and are accountable to the electorate. Civil liberties allow 
for the freedoms of expression and belief, associational and organizational rights, rule of law, 
and personal autonomy without interference from the state. Political rights and civil liberties 
scores range from 1 to 7, with 1 representing the highest and 7 the lowest level of freedom. 
Since civil liberties and political rights are two important but different components of 
democracy, Freedom House does not provide a composite variable. A composite variable may 
be created by taking the average of these two scores, but then it cannot be argued that new 
variable represents democracy and therefore we do not prefer to use it in our analysis. Figure 
123 
 
4.8 shows the number of countries and states in the sample by their civil liberties and political 
rights scores. 
 
Figure 4.8 Civil liberties and political rights scores for the sample countries in 2011 
 
 
Corruption perceptions index for 2010 is taken from Transparency International (2011). It 
ranks countries according to their perceived levels of public sector corruption on a scale of 0 
(highly corrupt) to 10 (very clean). Perceptions are used because corruption is to a great 
extent a hidden activity that is difficult to measure. The index is an aggregate indicator that 
combines different sources of information about corruption, making it possible to compare 
countries. The index draws on assessments and opinion surveys carried out by independent 
and reputable institutions. These surveys and assessments include questions related to the 
bribery of public officials, kickbacks in public procurement, embezzlement of public funds, 
and the effectiveness of public sector anti-corruption efforts. Figure 4.9 provides a visual plot 
of corruption perceptions index for 2010. 
 
124 
 
Standardization of the scales for institutional variables does not change the results, but 
changes the coefficients. Since we transform coefficients into Incident Rate Ratios (IRRs) in 
our analysis before commenting on them, standardization does not contribute to our analysis 
and the results. So, we do not prefer to standardize the scales for institutional variables in this 
study. 
Figure 4.9 Corruption perceptions index for 2010 
 
Data on populations and GDP per capita of the countries and the states in 2010 are taken from 
World Bank (2010b), Statistics Canada and US Census Bureau. Since using the logarithm of a 
variable enables us to interpret coefficients easily and is an effective way of shrinking the 
distance between values, we transform population and GDP per capita variables into 
logarithmic form and use these transformed variables in our model. We also include a dummy 
variable into our dataset to represent OECD member countries. 
 
The data on polity score for each country in 2010 are obtained from Center for Systemic 
Peace (CSP, 2010). The polity score ranges from +10 (strongly democratic) to -10 (strongly 
autocratic). Investment freedom index and property rights index scores for 2011 are provided 
by Heritage Foundation (2011) and both indexes range from 0 to 100. In an economically free 
country, there would be no constraints on the flow of investment capital. Individuals and 
125 
 
firms would be allowed to move their resources into and out of specific activities both 
internally and across the country’s borders without restriction. Such an ideal country receives 
a score of 100 in Heritage Foundation’s Investment Freedom Index. In practice, however, 
most countries have a variety of restrictions on investment. Some have different rules for 
foreign and domestic investment; some restrict access to foreign exchange; some impose 
restrictions on payments, transfers, and capital transactions; in some, certain industries are 
closed to foreign investment. Moreover, labour regulations, corruption, red tape, weak 
infrastructure, and political and security conditions can also affect the freedom that investors 
have in a market. The index evaluates a variety of restrictions typically imposed on 
investment. Points are deducted from the ideal score of 100 for the restrictions found in a 
country’s investment regime. Moreover, the property rights index assesses the ability of 
individuals to accumulate private property, secured by clear laws that are fully enforced by 
the government. It measures the degree to which a country’s laws protect private property 
rights and the degree to which its government enforces those laws. It also assesses the 
 
Figure 4.10 Scatter plot of investment freedom index and property rights index 
 
 
126 
 
likelihood that private property will be expropriated and analyses the independence of the 
judiciary, the existence of corruption within the judiciary, and the ability of individuals and 
businesses to enforce contracts. The more certain the legal protection of property, the higher a 
country’s score; similarly, the greater the chances of government expropriation of property, 
the lower a country’s score. Figure 4.10 presents a scatter plot of investment freedom index 
and property rights index. 
 
4.6 Empirical analysis 
 
Throughout our analysis, we explain electricity market reform score as a function of (i) the 
background of the chairperson of electricity market regulatory agency when reforms started or 
were considered, (ii) the background of the governor or minister who was responsible for 
energy policy at that time, (iii) corruption perceptions index, civil liberties and political rights 
scores, and (iv) control variables. 
 
The assumption of the Poisson model is that the conditional mean is equal to the conditional 
variance. Poisson regression will have difficulty with over dispersed data, i.e. variance much 
larger than the mean. Therefore, before starting our analysis, we need to look at the mean and 
variance of our dependent variable, that is, electricity market reform score. In our case, the 
mean of reform score variable is 5.48 and the variance is 5.01. Even though these numbers are 
for the unconditional mean and variance it can be informative because it gives us some 
indication of whether a Poisson regression should be used. In our analysis, reform score 
variable appears not to be overdispersed, as the mean is larger than the variance, and the 
predictor variables should help, so it may be reasonable to fit a Poisson regression model. 
Moreover, to make sure that Poisson regression is an appropriate tool to analyse our dataset, 
we report the results of the two Poisson goodness-of-fit tests (Deviance and Pearson 
127 
 
goodness-of-fit tests) in the regression output table. The large value for chi-square in these 
tests may be an indicator that the Poisson distribution is not a good choice. A significant 
(p<0.05) test statistic from the tests indicates that the Poisson model is inappropriate. In our 
model, values for chi-square in these tests are quite small and the test statistics are 
insignificant even at 80% level. So, it is obvious that Poisson regression is an appropriate 
method for our analysis. 
 
We start the empirical analysis by estimating a Poisson regression for our model14. Cameron 
and Trivedi (2009) recommend the use of robust standard errors when estimating a Poisson 
model, so we use robust standard errors for the parameter estimates. Table 4.2 presents 
Poisson estimation results. In the output table, we also report “Log pseudolikelihood”, which 
is the log likelihood of the fitted model. It is used in the calculation of the Likelihood Ratio 
(LR) chi-square test of whether all predictor variables’ regression coefficients are 
simultaneously zero. Moreover, we provide the number of observations. This is the number of 
observations used in the Poisson regression. It may be less than the number of cases in the 
dataset if there are missing values for some variables in the model. By default, Stata and 
Eviews do a listwise deletion of incomplete cases. Besides, we also report Wald chi2 value, 
which is the LR test statistic for the omnibus test that at least one predictor variable regression 
coefficient is not equal to zero in the model. The degrees of freedom (the number in 
parenthesis) of the LR test statistic are defined by the number of predictor variables. Finally, 
“Prob>chi2” value indicates the probability of getting a LR test statistic as extreme as, or 
more so, than the one observed under the null hypothesis that all of the regression coefficients 
are simultaneously equal to zero. In other words, this is the probability of obtaining this chi- 
square test statistic if there is in fact no effect of the predictor variables. This p-value is 
 
                                                          
14 Throughout the paper, model estimations are carried out and cross-checked by Stata 12.0 and Eviews 7.1. 
128 
 
Table 4.2 Poisson regression estimation results for the full sample 
Variables Variable Type Coefficient Robust Std. Err. z P>|z| [95% Conf. Interval] 
Electricity market reform score Dependent 
      Chairperson of the regulator when reforms started/considered 
       His/her experience in electricity industry at appointment Explanatory 0.001 0.0039 0.35 0.730 -0.0063 0.0090 
Length of term Explanatory 0.019 0.0129 1.44 0.149 -0.0066 0.0438 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 0.073* 0.0442 1.65 0.098 -0.0136 0.1597 
Educational background in 
        - Business or economics Explanatory -0.186* 0.1106 -1.69 0.092 -0.4031 0.0303 
 - Engineering Explanatory 0.083 0.1085 0.76 0.447 -0.1301 0.2953 
 - Law Explanatory 0.002 0.1117 0.02 0.984 -0.2167 0.2213 
 - Other Explanatory 0.019 0.1389 0.14 0.891 -0.2532 0.2912 
Energy minister/governor when reforms started/considered 
       His/her experience in electricity industry Explanatory -0.007 0.0092 -0.80 0.424 -0.0253 0.0106 
Length of term Explanatory -0.001 0.0217 -0.05 0.963 -0.0436 0.0416 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 0.096* 0.0572 1.67 0.095 -0.0165 0.2077 
Educational background in 
        - Business or economics Explanatory -0.194* 0.1117 -1.74 0.083 -0.4129 0.0250 
 - Engineering Explanatory -0.339** 0.1624 -2.09 0.037 -0.6574 -0.0209 
 - Law Explanatory -0.154 0.1394 -1.10 0.270 -0.4273 0.1193 
 - Other Explanatory -0.275** 0.1160 -2.37 0.018 -0.5021 -0.0474 
129 
 
Variables Variable Type Coefficient Robust Std. Err. z P>|z| [95% Conf. Interval] 
Other explanatory variables 
       Civil liberties score in 2011 Explanatory 0.281*** 0.1071 2.62 0.009 0.0711 0.4910 
Political rights score in 2011 Explanatory -0.148 0.0926 -1.59 0.111 -0.3290 0.0339 
Corruption perceptions index in 2010 Explanatory 0.203*** 0.0772 2.63 0.008 0.0520 0.3546 
Control variables 
       Log of population in 2010 Control 0.177*** 0.0330 5.37 0.000 0.1126 0.2421 
Log of GDP per capita in 2010 Control 0.315*** 0.0943 3.34 0.001 0.1300 0.4997 
Dummy (1: OECD country, 0: non-OECD country) Control -0.306* 0.1567 -1.95 0.051 -0.6134 0.0010 
Investment freedom index in 2011 Control 0.012*** 0.0043 2.87 0.004 0.0039 0.0207 
Polity score in 2010 Control 0.045 0.0328 1.38 0.167 -0.0189 0.1097 
Property rights index in 2011 Control -0.021*** 0.0062 -3.43 0.001 -0.0332 -0.0090 
Constant Constant -0.772 0.8297 -0.93 0.352 -2.3986 0.8539 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Note: Log pseudolikelihood: -173.87, Number of obs: 86 
         Wald chi2(23): 107.98, Prob > chi2: 0.000 
         Deviance goodness-of-fit: 48.87, Prob > chi2(62): 0.8876 
         Pearson goodness-of-fit: 48.28, Prob > chi2(62): 0.8989 
 
130 
 
Table 4.3 Poisson regression estimation results for the full sample as Incident Rate Ratios (IRR) 
Variables Variable Type IRR Robust Std. Err. z P>|z| [95% Conf. Interval] 
Electricity market reform score Dependent 
      Chairperson of the regulator when reforms started/considered 
       His/her experience in electricity industry at appointment Explanatory 1.001 0.0039 0.35 0.730 0.9937 1.0091 
Length of term Explanatory 1.019 0.0131 1.44 0.149 0.9934 1.0447 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 1.076* 0.0475 1.65 0.098 0.9865 1.1731 
Educational background in 
 
      
 - Business or economics Explanatory 0.830* 0.0918 -1.69 0.092 0.6682 1.0307 
 - Engineering Explanatory 1.086 0.1179 0.76 0.447 0.8780 1.3435 
 - Law Explanatory 1.002 0.1120 0.02 0.984 0.8051 1.2477 
 - Other Explanatory 1.019 0.1415 0.14 0.891 0.7763 1.3380 
Energy minister/governor when reforms started/considered 
 
      
His/her experience in electricity industry Explanatory 0.993 0.0091 -0.80 0.424 0.9750 1.0107 
Length of term Explanatory 0.999 0.0217 -0.05 0.963 0.9573 1.0425 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 1.100* 0.0629 1.67 0.095 0.9836 1.2308 
Educational background in 
 
      
 - Business or economics Explanatory 0.824* 0.0920 -1.74 0.083 0.6617 1.0253 
 - Engineering Explanatory 0.712** 0.1157 -2.09 0.037 0.5182 0.9793 
 - Law Explanatory 0.857 0.1195 -1.10 0.270 0.6523 1.1267 
 - Other Explanatory 0.760** 0.0881 -2.37 0.018 0.6052 0.9537 
131 
 
Variables Variable Type IRR Robust Std. Err. z P>|z| [95% Conf. Interval] 
Other explanatory variables 
 
      
Civil liberties score in 2011 Explanatory 1.325*** 0.1419 2.62 0.009 1.0737 1.6339 
Political rights score in 2011 Explanatory 0.863 0.0799 -1.59 0.111 0.7196 1.0345 
Corruption perceptions index in 2010 Explanatory 1.225*** 0.0946 2.63 0.008 1.0534 1.4256 
Control variables 
 
      
Log of population in 2010 Control 1.194*** 0.0394 5.37 0.000 1.1192 1.2739 
Log of GDP per capita in 2010 Control 1.370*** 0.1292 3.34 0.001 1.1388 1.6482 
Dummy (1: OECD country, 0: non-OECD country) Control 0.736* 0.1154 -1.95 0.051 0.5415 1.0010 
Investment freedom index in 2011 Control 1.012*** 0.0043 2.87 0.004 1.0039 1.0210 
Polity score in 2010 Control 1.046 0.0343 1.38 0.167 0.9812 1.1159 
Property rights index in 2011 Control 0.979*** 0.0060 -3.43 0.001 0.9674 0.9910 
Constant Constant 0.462 0.3833 -0.93 0.352 0.0908 2.3489 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Note: Log pseudolikelihood: -173.87, Number of obs: 86 
         Wald chi2(23): 107.98, Prob > chi2: 0.000 
         Deviance goodness-of-fit: 48.87, Prob > chi2(62): 0.8876 
         Pearson goodness-of-fit: 48.28, Prob > chi2(62): 0.8989 
 
 
132 
 
compared to a specified alpha level, our willingness to accept a Type I error, which is 
typically set at 0.05 or 0.01. The small p-value from the LR test, p < 0.0001, would lead us to 
conclude that at least one of the regression coefficients in the model is not equal to zero. 
 
While analysing the estimated Poisson regression coefficients, we should keep in mind that 
the dependent variable is a count variable, and Poisson regression models the log of the 
expected count as a function of the predictor variables. We can interpret the Poisson 
regression coefficient as follows: for a one unit change in the predictor variable, the difference 
in the logs of expected counts is expected to change by the respective regression coefficient, 
given the other predictor variables in the model are held constant. For instance, the coefficient 
of the variable “Chairperson’s education level” can be interpreted as follows: If Chairperson’s 
education increases by one level (e.g. from MSc to PhD), the difference in the logs of 
expected counts would be expected to increase by 0.073 unit, while holding the other 
variables in the model constant. Since interpretation of coefficients from a Poisson regression 
is not straightforward, the incidence rate ratios (IRR) are obtained by exponentiating the 
Poisson regression coefficients. Table 4.3 shows Poisson estimation results as incident rate 
ratios. The output tables also present the standard errors of the individual regression 
coefficients. They are used both in the calculation of the z test statistic and the confidence 
interval of the regression coefficient. P-value gives the probability that a particular z test 
statistic is as extreme as, or more so, than what has been observed under the null hypothesis 
that an individual predictor’s regression coefficient is zero given that the rest of the predictors 
are in the model. 
 
133 
 
4.7 Discussion of the results 
 
The empirical results presented above are in line with our hypotheses. So, we fail to reject our 
hypotheses. Reviewing our findings in relation to the research hypotheses, we find that the 
background of the chairperson of the regulator and that of the governor or minister 
responsible for energy policy when reforms started have a statistically significant impact on 
overall reform progress (Hypothesis 1). Besides, we see that the democracy has also an 
impact on reform progress (Hypothesis 2). Finally, our results show that countries with 
relatively high corruption tend to introduce less electricity market reforms (Hypothesis 3). 
The interpretation of the results in detail and their policy implications are provided in the sub-
sections below. 
 
4.7.1 Impact of the background of key persons on the reforms 
 
Our empirical findings suggest that the educational background and education level of the 
chairperson of the electricity market regulatory agency are two determinants of the scope of 
power industry reform in a country. We could not detect any statistically significant 
relationship between experience or length of term of the chairperson and scope of reforms. 
We find that if the chairperson’s education were to increase by one level (e.g. from MSc to 
PhD), its rate ratio for reform score would be expected to increase by a factor 1.076, while 
holding all other variables in the model constant (see Table 4.3). Besides, we detect a negative 
relationship between educational background of the chairperson in business or economics and 
scope of reforms. Our results imply that if the chairperson holds a degree in business or 
economics, the reform score is expected to decrease by a factor 0.830, while holding all other 
variables in the model constant (see Table 4.3). One difficulty with our analysis here is that 
the perception among the majority of the economics profession of what sound economic 
policy actually is did change substantially since the 1980s. The impact of education on market 
134 
 
reforms might thus change over time. That is, most of those with a degree in business or 
economics may regard the vertical integration as the best model for electricity markets if they 
were educated in pre-1980s period as this was the dominant view at that time. 
 
Let me illustrate these results using data from our dataset. In 2004, South Africa started a 
reform process in its electricity market and set up its regulatory agency (National Energy 
Regulator, NERSA) and its first chair held an MSc degree. One year later, Nigeria also started 
a reform process and established its regulatory agency (Nigerian Electricity Regulatory 
Commission, NERC) and its first chair had a PhD degree. For 2011, the electricity market 
reform scores of South Africa and Nigeria are 5 and 6, respectively. Our results suggest that if 
the chairperson’s education in a country were to increase by one level, its rate ratio for reform 
score would be expected to increase by a factor 1.076. Therefore, holding all other variables 
constant and assuming that two countries are the same apart from the education levels of 
chairpersons of their regulatory agencies when reforms started, our results require that South 
Africa’s reform score would be 5.38 (5*1.076) if the education of the first chairperson of 
South Africa’s regulatory agency were to increase by 1 level (from MSc to PhD). So, our 
findings imply that 0.38 point of 1 point difference between the reform scores of two 
countries may be explained by the difference between education levels of two chairpersons. 
 
The educational background and education level of the governor or minister responsible for 
energy policy when the reforms started or were considered seem to be other two determinants 
of the scope of power industry reform in a country. We could not detect any statistically 
significant relationship between length of term or experience of the minister/governor and 
scope of reforms (see Table 4.3). Our findings show that if the minister/governor’s education 
were to increase by one level, its rate ratio for reform score would be expected to increase by 
a factor 1.1, while holding all other variables in the model constant (see Table 4.3). This result 
135 
 
implies that a minister/governor’s education level positively contributes to the reform process. 
The results also show that if the minister/governor holds a degree in business/economics or 
engineering, reform score is expected to be 0.824 and 0.712 times less, respectively (see 
Table 4.3). 
 
To illustrate these results, we may use data from our dataset. In 1996, Pennsylvania State of 
US considered whether to initiate a reform process. The Governor of Pennsylvania at that 
time held a non-engineering (law) university degree. Four years later, Kentucky State also 
considered the reforms in its electricity market and, at that time, the Governor of Kentucky 
had a degree in engineering. In 2011, the electricity market reform scores of Pennsylvania and 
Kentucky were 8 and 3, respectively. Our results suggest that if the governor holds a degree in 
engineering, reform score is expected to be 0.712 times less. Therefore, holding all other 
variables constant and assuming that two states are the same apart from educational 
background of the governors when the reforms were considered, our results require that 
reform score of Pennsylvania would be 5.7 (8*0.712) if the Governor of Pennsylvania were to 
have an educational background in engineering. So, our findings imply that 2.3 (8-5.7) points 
of 5 points difference between the reform scores of two states might be explained by the 
difference between educational backgrounds of two governors. However, it is important to 
keep in mind that this result is valid holding all other variables constant and assuming that 
two states are the same apart from educational background of the governors when the 
reforms were considered. If these assumptions are removed, then there may be some other 
explanations for the difference between the reform scores of two states. For instance, as we 
mentioned before, Paul L. Joskow argues that the differences in electricity prices among US 
states explain why some US states introduce reforms and why others do not (White, 1996). 
Joskow (2000, Table 4-1, page 135) shows that in Pennsylvania the electricity prices for 
residential and industrial consumers were 9.90 and 5.89 cents per kilowatt hour in 1997, 
136 
 
respectively. In the same year, prices for residential and industrial consumers were 5.58 and 
2.80, respectively, in Kentucky. So, it may easily be concluded that the difference between the 
reform scores of two states is explained by the difference between different electricity prices 
in these two states. From Joskow’s point of view, Pennsylvania had relatively higher prices 
than Kentucky did and therefore introduced more reform steps. In short, it is important to 
remember that our results are valid under a set of assumptions and there may be alternative 
explanations that are valid under another set of assumptions. 
 
Based on the results above, we clearly fail to reject Hypothesis 1. That is, due to path-
dependency, the background of the chairperson of the regulator and that of the governor or 
minister responsible for energy policy when reforms started seem to have a statistically 
significant impact on overall reform progress. Since Hypothesis 1 is both an immediate 
application of the New Institutional Economics and a direct parallel with the empirical 
findings reported in the literature we summarized in Section 4.3.1, our results not only imply 
that NIE may be a useful approach to explain differences among various countries in terms of 
reform progress but also contribute to the literature connecting personal traits of policy 
makers and implementers with policy outcomes. Our findings confirm one of the main 
assumptions of NIE that the process of any economic reform (e.g. electricity market reform) 
is largely path dependent; so, getting the institutions right is critical to reform success as 
getting them wrong can lead to path-dependency, whereby inefficient electricity markets may 
persist. Therefore, to prevent inefficient institutional structures in the subsequent reform 
phases, the utmost attention should be paid to arrangements at the very beginning of the 
reform programs. 
 
Overall, our results show that the background of policy makers and policy implementers are 
associated with the progress in liberalizing reforms in electricity markets. We also 
137 
 
acknowledge the possibility of reverse causation; that is, a country may be more likely to 
choose a regulator or minister/governor with a specific background if it is contemplating 
reform. 
 
The most important policy repercussion of these results is that, to secure a rapid reform 
progress, right people should set up reform policies and structures; and right people should 
put them into practice. Therefore, while appointing regulators and policy makers, their 
backgrounds should be taken into account. 
 
4.7.2 Impact of democracy on the reforms 
 
Our results imply that any improvement in civil liberties score of a country results in a decline 
in the reform score of that country. Civil liberties score ranges from 1 to 7, 1 representing the 
highest and 7 the lowest level of freedom. If civil liberties score of a country were to increase 
by one unit (that is when civil liberties become more limited), its rate ratio for reform score 
would be expected to increase by a factor 1.325 (see Table 4.3). We could not detect a 
statistically significant relationship between political rights score and reform progress.  
 
Based on these results, we clearly fail to reject Hypothesis 2, meaning that the level of 
democracy in a country seems to have a statistically significant impact on reform progress. 
Since Hypothesis 2 is also an immediate application of the New Institutional Economics, the 
results above reconfirm that NIE may be a useful approach to explain the differences among 
various countries in terms of reform progress. The findings in this sub-section verify another 
main assumption of NIE that institutional structure of a country (like democracy) is one of the 
determinants of the extent of an economic reform program (such as electricity market reform) 
in that country. 
138 
 
 
This result may imply that some dimension(s) of democratic systems may obstruct or 
interrupt reforms, which is in line with the empirical evidence provided by World Bank 
(1995). They may be in the form of prolonged legislation periods due to opposition in the 
parliament, frequent changes in government (and therefore, reform policy), and the negative 
impact of anti-reform pressure groups or populist policies that damage reform progress 
especially in election times. Therefore, policy makers and implementers in countries with 
strong democratic traditions should take into account these factors, and design and carry out 
reforms accordingly. 
 
4.7.3 Impact of corruption on the reforms 
 
Our results indicate a positive relationship between reform progress and corruption 
perceptions index (which increases as corruption declines in a county). If corruption 
perceptions index of a country were to increase by one unit, its rate ratio for reform score 
would be expected to increase by a factor 1.225, while holding all other variables in the model 
constant (see Table 4.3).  
 
This result clearly suggests that we fail to reject Hypothesis 3. That is, our results indicate that 
countries with relatively high corruption tend to introduce less electricity market reform, 
which reinforces another key assumption of NIE that non-market transaction costs (like 
corruption) is one of the determinants of the reform progress. This result also constitutes 
another confirmation that NIE may be a valuable approach to explain the differences among 
various countries in terms of reform progress.  
 
139 
 
Our findings clearly show that reforms go much further in corruption-free countries. So, 
countries should do their best to reduce and, if possible, eliminate all structures, rules and 
procedures that may cause or tolerate corruption. 
 
4.7.4 Results from control variables 
 
Population and per capita income of a country seem to be other important factors in the 
reform progress. According to our results, population and GDP per capita of a country are 
positively correlated with its reform score. Moreover, being an OECD country has a 
statistically significant negative impact on reform progress. If a country is a member of 
OECD, then the reform score in this country is expected to be 0.736 times less (see Table 
4.3). Although we cannot conclude this from the econometric analysis in this paper, this result 
may be regarded as an indication that in countries with well-established institutions the 
backgrounds of the chairpersons and the ministers/governors are much less important than in 
those with weaker institutions in terms of reform progress. We also see that reform progress is 
highly correlated with investment freedom index. Our findings suggest a positive relationship 
between investment freedom index and reform progress. If investment freedom index of a 
country were to increase by one unit, its rate ratio for reform score would be expected to 
increase by a factor 1.012, while holding all other variables in the model constant (see Table 
4.3). We could not detect a statistically significant relationship between reform score and 
polity score. Additionally, we detect a negative relationship between property rights index and 
reform score. Although this negative relationship is statistically significant, its impact is 
extremely limited. If property rights index of a country were to increase by one unit, its rate 
ratio for reform score would be expected to decrease by a factor 0,979, while holding all other 
variables in the model constant (see Table 4.3).  
 
140 
 
The results above imply that wealthier and highly-populated countries progress faster in the 
reform process. So, middle and low income countries with limited population should be 
cautious in designing power market reforms and should take into account income and size 
effects during reform implementation. Besides, an appropriate investment environment seems 
to be one of the important determinants of the reform progress. So, countries wishing a rapid 
reform progress may consider putting into practice measures to set up and maintain a proper 
investment climate.  
 
4.7.5 Results without the states in US and provinces in Canada 
 
In our analysis so far, we used observations from countries together with those from the states 
in US and provinces in Canada. The states in US and provinces in Canada are similar in terms 
of their geographical location (i.e. North America) and income level (i.e. high income group), 
which implies that they may have common tendencies towards electricity market reform. 
Taking into account also the fact that the observations from the states in US and provinces in 
Canada constitute more than half of the observations in our dataset (64 of 115 observations), 
our results may be dominated by common characteristics of the states in US and provinces in 
Canada that may or may not be relevant to reform process. Therefore, re-estimating our 
models without the states in US and provinces in Canada may produce useful insights into our 
analysis. In this second phase of estimation, we look at the impact of the backgrounds of 
policy makers/implementers and democracy & corruption on reforms separately. Table 4.4 
presents Poisson regression estimation results without the states in US and provinces in 
Canada as Incident Rate Ratios (IRR) for the impact of the backgrounds of chairpersons and 
ministers on the reforms while Table 4.5 does the same for the impact of democracy and 
corruption. 
 
141 
 
The empirical findings from the second phase of estimation (that is, the results without the 
states in US and provinces in Canada) suggest that the length of term and educational 
background of the chairperson of the electricity market regulatory agency are two 
determinants of the scope of power industry reform in a country. We could not detect any 
statistically significant relationship between experience in electricity industry or education 
level of the chairperson and scope of reforms in a country. We find that if the chairperson’s 
length of term in a country were to increase by one year, its rate ratio for reform score would 
be expected to increase by a factor 1.042, while holding all other variables in the model 
constant. Moreover, we detect a positive relationship between educational background of the 
chairperson in engineering and law and scope of reforms in a country. Our results imply that 
if the chairperson holds a degree in engineering or law, the reform score is expected to be 
1.459 or 1.477 times greater, respectively; while holding all other variables in the model 
constant. 
 
Experience of the minister responsible for energy policy and his/her educational background 
seem to be other important determinants of the scope of power industry reform in a country. 
We could not detect any statistically significant relationship between length of term or 
education level of the minister and scope of reforms. Our findings show that if the minister’s 
experience in electricity industry were to increase by one year, its rate ratio for reform score 
would be expected to decrease by a factor 0.978, while holding all other variables in the 
model constant. This result implies that minister’s experience in electricity industry adversely 
affects the reform process. The results also show that if the minister holds a degree in 
economics or business, reform score in this country is expected to be 1.601 times greater. 
However, if s/he holds a degree in law, then the reform score in this country is expected to be 
0.737 times smaller. 
 
142 
 
Table 4.4 Poisson regression estimation results without the states in 
US and provinces in Canada as IRR (Micro Model) 
Variables 
Variable 
Type 
IRR 
Robust Std. 
Err. 
p-value 
(P>|z|) 
Electricity market reform score Dependent 
   
Chairperson of the regulator when reforms started/considered 
His/her experience in electricity industry at appointment Explanatory 1.007 0.007 0.351 
Length of term Explanatory 1.042** 0.021 0.044 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 1.120 0.133 0.338 
Educational background in 
 
   
- Business or economics Explanatory 0.946 0.165 0.752 
- Engineering Explanatory 1.459* 0.313 0.078 
- Law Explanatory 1.477** 0.228 0.011 
Energy minister when reforms started/considered 
His/her experience in electricity industry Explanatory 0.978** 0.011 0.038 
Length of term Explanatory 0.998 0.045 0.963 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 1.012 0.095 0.900 
Educational background in 
 
   
- Business or economics Explanatory 1.601*** 0.252 0.003 
- Engineering Explanatory 1.403 0.333 0.154 
- Law Explanatory 0.737* 0.127 0.077 
Control variables 
Population in 2010 (million people) Control 1.001*** 0.000 0.000 
Dummy (1: OECD country, 0: non-OECD country) Control 1.471** 0.269 0.035 
Polity score in 2010 [-10, +10] Control 1.090*** 0.034 0.006 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Note: Log pseudolikelihood: -78.25, Number of obs: 35 
         Wald chi2(15): 3384.83, Prob > chi2: 0.000 
         Deviance goodness-of-fit: 26.42, Prob > chi2(19): 0.1188 
         Pearson goodness-of-fit: 27.68, Prob > chi2(19): 0.0898 
 
143 
 
Table 4.5 Poisson regression estimation results without the states in 
US and provinces in Canada as IRR 
Variables 
Variable 
Type 
IRR 
Robust Std. 
Err. 
p-value 
(P>|z|) 
Electricity market reform score Dependent 
   
Explanatory variables 
Civil liberties score in 2011 Explanatory 1.339*** 0.097 0.000 
Political rights score in 2011 Explanatory 0.870 0.076 0.110 
Corruption perceptions index in 2010 Explanatory 1.054* 0.032 0.081 
Control variables 
Log of population in 2010 Control 1.094*** 0.291 0.001 
Log of GDP per capita in 2010 Control 1.413*** 0.099 0.000 
Dummy (1: OECD country, 0: non-OECD country) Control 0.989 0.168 0.949 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Note: Log pseudolikelihood: -110.86, Number of obs: 51 
         Wald chi2(5): 3979.59, Prob > chi2: 0.000 
         Deviance goodness-of-fit: 32.16, Prob > chi2(43): 0.9072 
         Pearson goodness-of-fit: 33.90, Prob > chi2(43): 0.8642 
 
Population of a country seems to be an important factor in the reform progress but its impact 
is quite limited. According to our results, if the population of a country were to increase by 
one million, its rate ratio for reform score would be expected to increase by a factor 1.001, 
while holding all other variables in the model constant. Being an OECD country has also a 
significant positive impact on reform progress. If a country is a member of OECD, then the 
reform score in this country is expected to be 1.471 times greater. 
 
The results from Table 4.5 confirm that variables representing civil liberties and corruption 
perceptions have a strong impact on the reform progress. As in the case of the first phase of 
estimations, the results from the second phase verify that any improvement in civil liberties 
score of a country results in a decline in the reform score of that country. As mentioned 
before, civil liberties score ranges from 1 to 7, 1 representing the highest and 7 the lowest 
144 
 
level of freedom. If civil liberties score of a country were to increase by one unit (that is when 
civil liberties become more limited), its rate ratio for reform score would be expected to 
increase by a factor 1.339 (see Table 4.5). 
 
As we mentioned before, the data on macro variables (i.e. civil liberties score, political rights 
score and corruption perceptions index) are provided on country level and not available for 
specific states in US and provinces in Canada. Therefore, we cannot estimate our regressions 
for states in US and provinces in Canada alone. Nonetheless, it is possible to estimate our 
regressions to detect the impact of the backgrounds of regulators and governors on the reform 
progress in US states and Canadian provinces only. However, when we estimate our 
regressions with data from US states and Canadian provinces only, we see that the 
coefficients of all variables are statistically not significant even at 10% level with the 
exception that a governor with an engineering degree seems to decrease the reform score by a 
factor 0.647. It seems that the impact of the backgrounds of regulators and governors on the 
reform progress is quite limited in the states in US and the provinces in Canada. Since almost 
all coefficients are not significant, we do not provide the further details of the estimation table 
here. 
 
Table 4.6 compares the results from the first and second phase of estimations. It presents 
statistically significant coefficients only. As can be seen in Table 4.6, the results from two 
groups of estimations are similar in general with some slight differences in details. When we 
take into account the fact that number of observations are quite different for the first (115 
observations) and second (51 observations) group of estimations, the similar results confirm 
the robustness of our results. In both groups of estimations, we see that backgrounds of 
chairperson and minister/governor and variables representing democracy and corruption 
levels have a statistically significant impact on reform progress. The only noteworthy 
145 
 
difference between two groups of estimations is the impact of OECD membership on the 
reforms. The results from the observations with the states in US and provinces in Canada 
indicate that OECD membership negatively affects the reform progress while those from the 
observations without the states in US and provinces in Canada imply that OECD membership 
has a positive impact on reform progress. When we evaluate these two results together, we 
may conclude that OECD membership contributes to reform progress in reforming countries 
other than US and Canada. When we include observations from US and Canada into our 
analysis, some common features of the states in US and provinces in Canada (that may or 
may not be relevant to reform process) seem to distort the impact of OECD membership on 
the reform progress. Removing observations from the states in US and provinces in Canada 
makes the true impact of OECD membership clear. 
 
Table 4.6 Results with and without the states in US and provinces in Canada as IRR 
Variables 
Coefficient 
(with US and Canada) 
Coefficient 
(without US and Canada) 
Electricity market reform score 
  Chairperson of the regulator when reforms started/considered 
His/her experience in electricity industry at appointment   
Length of term  1.042** 
Education level (1: BSc, 2: MSc, 3: PhD) 1.076*  
Educational background in   
 - Business or economics 0.830*  
 - Engineering  1.459* 
 - Law  1.477** 
 - Other   
Energy minister/governor when reforms started/considered 
His/her experience in electricity industry  0.978** 
Length of term   
Education level (1: BSc, 2: MSc, 3: PhD) 1.100*  
Educational background in   
 - Business or economics 0.824* 1.601*** 
 - Engineering 0.712**  
 - Law  0.737* 
 - Other 0.760**  
146 
 
Variables 
Coefficient 
(with US and Canada) 
Coefficient 
(without US and Canada) 
Institutional variables 
Civil liberties score in 2011 1.325*** 1.339*** 
Political rights score in 2011   
Corruption perceptions index in 2010 1.225*** 1.054* 
Control variables 
Log of population in 2010 1.194*** 1.094*** 
Log of GDP per capita in 2010 1.370*** 1.413*** 
Dummy (1: OECD country, 0: non-OECD country) 0.736* 1.471** 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
 
4.7.6 Impact of initial conditions and initial price levels on the reforms 
 
As we mentioned in Section 2.2.1, Paul L. Joskow argues that states only with high electricity 
prices have an incentive to introduce reforms in the US; and those with relatively low prices 
do not have such an incentive and therefore do not introduce reforms. In short, the differences 
in electricity prices among US states may explain why some US states introduce reforms and 
why others do not (White, 1996). So, for him, price is an effective explanatory determinant of 
electricity reform in the US. In this sub-section, we would like to investigate whether price is 
an equally effective explanatory determinant of electricity reform internationally as it is in the 
US. 
 
Starting point of the countries in terms of reform steps are not the same. Some countries have 
a tradition of private participation in electricity industry (IPPs, private distribution and 
transmission utilities and so on) and long-established regulatory commissions, so they have 
higher reform scores at the beginning of the reform process. On the other hand, most of the 
countries do not have such traditions and structures in their power industry, so they have very 
low reform scores (o or 1) at the beginning of the reforms. We would also like to investigate 
147 
 
whether the differences in initial conditions in terms of reform steps at the beginning of the 
reform process have an impact on the subsequent reform progress. 
 
In order to carry out two investigations we summarized above, we include three new variables 
into our analysis, namely “electricity prices for industry when electricity reform started”, 
“electricity prices for households when electricity reform started” and “reform score when the 
reforms in electricity market started”. The first two variables represent the impact of initial 
price level on the reform progress while the last one accounts for the impact of starting point 
of the countries in terms of reform steps. The data on initial price levels are taken from the 
first paper, and the data on reform scores when the reforms in electricity market started are 
collected by the author. 
 
As indicated in Table 4.2 and Table 4.3, the main analysis in this paper includes data from 86 
countries, states or provinces. However, data on initial price levels are extremely limited and 
available only for 18 countries. In order not to significantly reduce the number of 
observations in our analysis, we prefer not to include these three new variables directly into 
our main analysis. Besides, we have 23 explanatory or control variables in our main model, 
which is not a problem when we have 86 observations. However, when the number of 
observations is reduced to 18 and there new explanatory variables are added to the analysis; 
Stata drops some variables (including new ones) to avoid various problems associated with 
over specification problem. So, in practice, it is also not possible to include these three new 
variables directly into our main analysis.  
 
To overcome two problems mentioned above, we specify a new Poisson regression including 
three new variables but excluding some of the explanatory or control variables we used in our 
main model. Table 4.7 presents estimation results of our new Poisson regression. Since we  
148 
 
Table 4.7 Poisson regression estimation results for the full sample including initial conditions and initial price levels 
Variables Variable Type IRR Robust Std. Err. z P>|z| [95% Conf. Interval] 
Electricity market reform score Dependent 
      
Chairperson of the regulator when reforms started/considered 
       
His/her experience in electricity industry at appointment Explanatory 1.001 0.004 0.28 0.782 0.994 1.008 
Length of term Explanatory 1.034*** 0.010 3.47 0.001 1.015 1.054 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 0.888** 0.041 -2.59 0.010 0.811 0.971 
Energy minister/governor when reforms started/considered 
 
      
His/her experience in electricity industry Explanatory 1.005 0.012 0.44 0.662 0.982 1.029 
Length of term Explanatory 0.969* 0.016 -1.88 0.060 0.938 1.001 
Education level (1: BSc, 2: MSc, 3: PhD) Explanatory 0.972 0.055 -0.50 0.614 0.870 1.086 
Other explanatory variables 
 
      
Reform score when the reforms in electricity market started Explanatory 0.988 0.027 -0.44 0.663 0.937 1.042 
Electricity prices for industry when electricity reform started Explanatory 1.013*** 0.004 3.24 0.001 1.005 1.021 
Electricity prices for households when electricity reform started Explanatory 0.999 0.002 -0.44 0.662 0.996 1.002 
Civil liberties score in 2011 Explanatory 0.712*** 0.061 -3.98 0.000 0.603 0.842 
Political rights score in 2011 Explanatory 0.915 0.087 -0.93 0.352 0.759 1.103 
Corruption perceptions index in 2010 Explanatory 0.900*** 0.024 -3.98 0.000 0.855 0.948 
149 
 
Variables Variable Type IRR Robust Std. Err. z P>|z| [95% Conf. Interval] 
Control variables 
 
      
Log of population in 2010 Control 0.864*** 0.019 -6.68 0.000 0.827 0.902 
Log of GDP per capita in 2010 Control 0.770* 0.104 -1.93 0.054 0.591 1.005 
Constant Constant 66.702*** 47.131 5.94 0.000 16.699 266.432 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Note: Log pseudolikelihood: -34.85, Number of obs: 18 
         Wald chi2(14): 76965.55, Prob > chi2: 0.000 
         Deviance goodness-of-fit: 0.74, Prob > chi2(3): 0.8648 
         Pearson goodness-of-fit: 0.74, Prob > chi2(3): 0.8649 
 
 
150 
 
have quite different number of observations and use different variables in this limited model 
and the main model, it is natural that the results may also be different. The limited model in 
this section is developed to give an idea about the impact of initial price levels and initial 
conditions on the reform progress using the limited dataset we have.  
 
As can be seen in Table 4.7, variables representing the backgrounds of the chairperson and the 
minister/governor, democracy and corruption are still significant, meaning that they still seem 
to be important determinants of how far reforms have gone in a country. Out of three new 
variables, only one of them is significant. Our results indicate that initial industrial electricity 
prices are positively correlated with reform progress, which verifies Joskow’s argument. So, 
we find that countries with high initial electricity prices for industry progress rapidly in the 
reform process. On the other hand, we could not detect a relationship between initial 
residential prices or initial conditions and the reform progress. 
 
As a result of the analysis in this sub-section, we find out that electricity prices (especially for 
industrial consumers) are an effective explanatory determinant of electricity reform 
internationally as it is in the US. Besides, even after allowing for differences in prices, the 
factors suggested by the New Institutional Economics are still significant determinants of the 
extent of reform. 
 
As we mentioned before, the concept of interest group competition has its roots in the 
economic theory of regulation (Peltzman, 1976; Stigler, 1971), which predicts that different 
interest groups try to capture policy-makers during the reform process and the direction of the 
reform process is shaped by the relatively strong interest groups. Our findings imply that 
reform progress is affected by high initial electricity prices for industrial consumers but not by 
those for households. When we take into account the fact that industrial consumers are much 
151 
 
more organized (and therefore powerful) than the residential ones, the results from our 
analysis seem to be in conformity with the predictions of economic theory of regulation. 
 
Overall, our results clearly show that both the backgrounds of the chairperson and the 
minister/governor, democracy and corruption are important determinants of how far reforms 
have gone in a country. The negative relationship between reform progress and democracy 
indicate that democracy may delay or hinder the reforms by magnifying the voices of anti-
reform interest groups, and therefore, reforms may be limited in countries with strong civil 
society institutions such as trade unions or other organized structures in the society that may 
consider reforms as ‘harmful’ to their self-interest. It should also be noted that any reduction 
in corruption level or improvement in the investment environment positively contributes to 
the scope of reforms in a country. 
 
4.8 Conclusion 
 
This study offered an analysis that focuses on the possible contribution of New Institutional 
Economics to the study of electricity market reforms. In the course of the study, we checked 
the validity of three important arguments of NIE for the power market liberalization process. 
The first argument is the “path-dependency”. To test its impact on the reform progress, we 
tried to explain whether the background of the chairperson of the regulatory agency when 
reforms started or were considered or that of the governor/minister responsible for energy 
policy at that time has an impact on the subsequent reform progress. The second argument is 
the impact of “democracy” as an institution on the reform progress. We looked at the effect of 
two important indicators of democracy (i.e., civil liberties and political rights) on the reform 
progress. The final argument of NIE we tested in this study is about transaction costs. We 
152 
 
concentrated on the level of corruption in a country as one of the key factors that determine 
transaction costs and tried to explore its impact on the reforms.  
 
To observe the impact of these variables, we constructed empirical econometric models using 
cross-section data from 51 states in US, 13 provinces in Canada and 51 other countries. As a 
result of the study, we found that the backgrounds of the chairperson and the 
minister/governor, and the level of democracy and corruption in a country are significantly 
correlated with how far reforms have gone in that country. Although we tested the 
applicability of only three arguments of NIE to the study of reform process in electricity 
markets, the results from our study confirm that NIE as a school of thought has the potential 
to provide a useful framework that can possibly be used to investigate power market reforms. 
Besides, we have many variables in our analysis but our cross-section dataset is limited due to 
lack of data. Although current number of observations in our study let us investigate the 
impact of the backgrounds of key persons on the reform progress, we cannot carry out a 
regression and get meaningful results when we divide our already limited dataset into two set 
of observations as one on “developed countries” and another on “developing countries”. 
Actually, a study looking at the impact of backgrounds on reform progress separately in 
developed and developing countries may produce useful results; but due to limited nature of 
our dataset, we cannot do this in this study. Finally, as a result of the analysis of a limited 
model, we also discovered that initial electricity prices (especially for industrial consumers) 
are an effective explanatory determinant of electricity reform. 
 
 
153 
 
 
 
 
 
 
Chapter 5 
 
5 Paper Three 
 
 
 
The political economy of electricity market liberalization: a cross-
country approach* 
 
5.1 Introduction 
 
Since the 1980s, the structure of electricity industry has shifted from a vertically integrated 
(and usually state-owned) monopoly towards unbundled (and usually privately owned) 
regulated utilities. This shift has also been strongly encouraged by the World Bank, IMF and 
other international financial institutions (Williams and Ghanadan, 2006). The power sector 
reform began in Chile in 1982 for the first time and then spread through various countries in 
the world especially after the 1990s. Therefore, the last three decades have witnessed 
widespread power market reforms in both developed and developing countries that cost 
                                                          
* This paper was presented at 30th USAEE/IAEE North American Conference (9-12 October 2011, Capital 
Hilton Hotel, Washington, DC) and granted one of the top 4 best paper awards (out of 160 papers) at this 
conference. 
154 
 
billions of dollars. Today, reforms are on-going in many countries and the reform process in 
power sector is regarded as not only possible and necessary, but also inevitable. 
 
In all reforming countries (whether developed or developing), reforms take place in a political 
economic environment and are directly affected by the developments taking place in it. In 
most cases, political structure of a country largely determines the extent of the reforms in that 
country. In the United Kingdom, for example, privatization of state owned electricity utility 
reflected the ideology of the Thatcher government and its interest in reducing the costs of 
domestic coal subsidies, among other things. Similar ideological and political explanations 
can be found from Norway to New Zealand (Hogan, 2002). There is no doubt that without 
political support the reforms cannot go further in any country. This paper attempts to discover 
the impact of political economic variables on the liberalization process in electricity markets. 
 
We try to answer the following research questions: (i) what is the impact of industrial 
electricity consumers (as an interest group) on the reform progress in power sector? (ii) does 
foreign influence resulting from the dependence on foreign financial support have an 
influence on the electricity market liberalization process? (iii) do the ideology of ruling party 
and professional/educational background of the chief executive (prime minister or president) 
constitute important determinants of the reform progress? If yes, what is the direction of the 
influences originated from these variables? 
 
The paper proceeds as follows. Next section presents a literature review regarding the applied 
empirical studies focusing on the political economy of liberalization processes. Section 5.3 
develops the hypotheses tested in the study. Section 5.4 summarizes the methodological 
framework. Section 5.5 describes data. Following two sections present empirical analysis and 
discuss the results. The last section concludes. 
155 
 
 
5.2 Literature review 
 
Presenting an extensive literature review on the political economy of economic reform is both 
outside the scope of this paper and not possible given the limitations on the length of the 
study. Although there is some preliminary academic work that investigates the impact of 
political economic variables on electricity market reform outcome (Pollitt, 2012); to the best 
of our knowledge, this study constitutes one of the first empirical applied investigations that 
focus on the possible implications of political economic environment for electricity market 
reform process. So, there is a gap in the empirical literature with regard to the analysis of the 
possible repercussions of the political economic variables for the power market reforms. This 
is quite surprising given the economic importance of the sector for both individual countries 
and the world economy in general, as well as the significant number of reform programs that 
have already initiated in many power sectors. 
 
In this section, we will mention only applied studies on the relationship between economic 
reform processes and political economic variables. The studies presenting an anecdotal 
discussion of the political economy of the various reform programs without any applied 
analysis are outside the scope of this section. Within this framework, we will concentrate on 
three groups of studies: (1) those providing applied evidence from power industry; (2) those 
on the political economy of reform process in telecommunications industry; (3) studies 
presenting the results of applied work from non-infrastructure industries. Appendix III-1 
presents details of the econometric studies mentioned here including hypotheses tested, 
dependent variables, explanatory variables, results, data and methodology. Appendix III-2 
classifies previous econometric studies by their focus. 
 
156 
 
The first group of studies (those focusing on the political economy of electricity market 
reforms) include only two papers by Chang and Berdiev (2011) and Cubbin and Stern (2006). 
Chang and Berdiev (2011) examine the effect of government ideology, political factors and 
globalization on energy regulation in electricity and gas industries using the bias-corrected 
least square dummy variable model in a panel of 23 OECD countries over the period 1975-
2007. They find that left-wing governments promote regulation in gas and electricity sectors; 
and less politically fragmented institutions contribute to deregulation of gas and electricity 
industries. Their results also suggest that long tenures of incumbent government have limited 
impact on regulation in electricity sector, while it is associated with an increase in regulation 
of gas sector. Further, they conclude that higher political constraints and more globalization 
lead to deregulation in electricity and gas sectors; and economic and social integration are the 
forces that promote deregulation in the gas industry, whereas political integration advances 
deregulation in the electricity industry. Cubbin and Stern (2006) assess whether a regulatory 
law and higher quality regulatory governance are associated with superior outcomes in the 
electricity industry. Their analysis, for 28 developing economies over 1980–2001, draws on 
theoretical and empirical work on the impact of telecommunications regulators in developing 
economies. Their study show that, controlling for privatization and competition and allowing 
for country-specific fixed effects, both regulatory law and higher quality regulatory 
governance are positively and significantly associated with higher per capita generation 
capacity. 
 
The studies providing applied evidence from telecommunications industry are Duso and 
Seldeslachts (2010), Gasmi et al. (2009), Gasmi and Recuero Virto (2010) and Li and Xu 
(2002). Duso and Seldeslachts (2010) empirically investigate the cross-sectional and temporal 
variation in entry liberalization in the mobile telecom industries of OECD countries during 
the 1990s. Their findings indicate that majoritarian electoral systems are important drivers for 
157 
 
change, while independent industry regulators slow down such reforms. They conclude that 
powerful industry incumbents hold up the liberalization process and governing bodies that 
favour a small welfare state accelerate it. Taking the view that political accountability is a key 
factor linking political and regulatory structures and processes, Gasmi et al. (2009) 
empirically investigate its impact on the performance of regulation in telecommunications 
using a time-series cross-sectional data set for 29 developing and 23 developed countries 
during 1985–99. They provide empirical evidence on the impact of the quality of political 
institutions and their modes of functioning on regulatory performance. Their analysis finds 
that the impact of political accountability on the performance of regulation is stronger in 
developing countries.  
 
The paper by Gasmi and Recuero Virto (2010) has two related objectives. First, it seeks to 
identify the key determinants of policies that have been at the heart of the reforms of the 
telecommunications industry in developing countries, namely, liberalization, privatization, 
and the (re)structuring of regulation. Second, it attempts to estimate the extent to which these 
policies have translated into actual deployment of telecommunications infrastructure. They 
conduct this simultaneous investigation by means of an econometric analysis of a 1985-1999 
time-series cross-sectional database on 86 developing countries. Their study finds that 
sectoral as well as institutional and financial factors are important determinants of the actual 
reforms implemented. They uncover that countries facing increasing institutional risk and 
financial constraints are more likely to introduce competition in the digital cellular segment 
and to privatize the fixed-line incumbent, these policies being economically attractive to both 
investors and governments. Finally, Li and Xu (2002) examine the political economy of 
privatization and liberalization in the telecommunications sector in recent decades. They find 
that countries with stronger pro-reform interest groups, namely the financial services sector 
and the urban consumers, are more likely to reform in more democratic countries. However, 
158 
 
their result suggest that less democratic countries are more likely to maintain the public sector 
monopoly when the government benefits more from such a governance mode, e.g., when the 
fiscal deficit is higher. 
 
The final group of studies presents the results of applied investigations from non-
infrastructure industries. The examples from this group include Alesina et al. (2006), Boschini 
(2006), Dreher et al. (2009), Duval (2008), Fredriksson and Wollscheid (2008), Goldberg and 
Pavcnik (2005), Huang (2009), Ickes and Ofer (2006), Kim and Pirttilä (2006), Olper (2007), 
Volscho (2007) and Wagner et al. (2009). We will briefly mention them while their details are 
presented in Appendix III-1. 
 
Alesina et al. (2006) question why countries delay stabilizations of large and increasing 
budget deficits and inflation and what explains the timing of reforms. They find that 
stabilizations are more likely to occur during crisis, at the beginning of term of office of a new 
government, in countries with “strong” governments (i.e. presidential systems and unified 
governments with a large majority of the party in office), and when the executive faces less 
constraints. Boschini (2006) analyses how incentives under different sets of political 
institutions map into policies that promote industrialisation. The results show that a flat 
wealth distribution and skilled political elite enhance development the most in elitist regimes, 
while democracies perform as well as elitist regimes in terms of industrialisation. Dreher et al. 
(2009) analyse whether the educational and professional background of a head of government 
matters for the implementation of market-liberalizing reforms. Their results show that reforms 
are more likely during the tenure of former entrepreneurs. Duval (2008) provides an empirical 
attempt to determine whether macroeconomic policies determined as a result of political 
processes influence reform patterns in labour and product markets. 
 
159 
 
Fredriksson and Wollscheid (2008) seek to explain the implications of corruption and political 
instability for firm investment in abatement technology. Their results suggest that political 
instability raises abatement technology investment. Goldberg and Pavcnik (2005) exploit 
drastic trade liberalizations in Colombia in the 1980s and 1990s to investigate the relationship 
between protection and industry wage premiums. Huang (2009) focuses on the forces that 
induce governments to undertake financial sector reform. Ickes and Ofer (2006) examine 
changes in the industrial structure of employment across Russian regions and assess the 
importance of legacy factors, political factors, and success factors in explaining this process. 
They find that initial conditions such as natural resource potential, climate, and industrial 
specialization explain more of the variation in industrial restructuring than political variables. 
Using data from transition economies, Kim and Pirttilä (2006) examine linkages between 
political constraints and economic reforms. Their results suggest that progress in reform is 
positively associated with public support for reforms, which is affected by income inequality 
and expected individual performance during future reforms. They also find evidence to 
support reform sequencing starting with a reform that is both popular and stimulatory to other 
reforms. 
 
Olper (2007) presents an empirical investigation of how agricultural land ownership 
inequality and government ideology (right-wing vs. left-wing) affect agricultural protection. 
Their data show, overall, that protection is decreasing in land inequality and with left-wing 
government orientation, but not in a linear fashion: left-wing governments tend to support 
agriculture in more unequal societies. Using data on 160 US metropolitan statistical areas 
from the 2000 census, Volscho (2007) examines how quintile shares of size-adjusted family 
income are impacted by union density and federal, state, and local government employment. 
Finally, Wagner et al. (2009) analyse how institutional factors affect satisfaction with 
democracy. They find that high-quality institutions like the rule of law, well-functioning 
160 
 
regulation, low corruption, and other institutions that improve resource allocation have a 
positive effect on average satisfaction with democracy. 
 
5.3 Hypothesis development 
 
Acemoglu et al. (2005) maintain that “[e]conomic institutions … are determined as collective 
choices of the society, in large part for their economic consequences. However, there is no 
guarantee that all individuals and groups will prefer the same set of economic institutions 
because different economic institutions lead to different distributions of resources. 
Consequently, there will typically be a conflict of interest among various groups and 
individuals over the choice of economic institutions.” So, among other factors, whether a 
country liberalizes its power sector depends on its political structure and the configuration of 
interest groups. While the liberalization in an electricity market may result in huge efficiency 
gains, the distribution of benefits and costs from the reform is usually determined politically. 
In this study, we use public choice and private interest theories that incorporate the political 
economy as one of the theoretical frameworks to explain the reform progress.  
 
As we mentioned in Section 1.3, the public choice and private interest theories view policy 
outcomes as equilibrium results from competition in the policy market (Becker, 1983; 
Peltzman, 1976; Stigler, 1986). Politicians supply and constituents demand policies. 
Politicians seek elected public offices and their votes increase with campaign contributions 
from interest groups. However, public officials are also agents of their constituents. With 
imperfect information and imperfect public oversight, their private interest, i.e., preference 
differing from those of their constituents, has the potential to shape policy outcomes. Private 
interest theory implies that efficient policies are the exception rather than the rule. Indeed, the 
161 
 
politicians may pursue private interests or may be captured by special interest groups at the 
expense of the general public (Li and Xu, 2002).  
 
In this section, we apply this theoretical framework to develop some of our hypotheses 
regarding the determinants of power market liberalization. Our study is based upon the idea 
that, while political ideology (Olson, 1965; Romer and Rosenthal, 1987) and the relative 
power of interest groups (Peltzman, 1976; Stigler, 1971) shape the direction of power market 
reforms; foreign influence and the background of the chief executive are also among the 
determinants of the reform progress. Hence, we argue that power market reform policies are 
set as a result of, inter alia, pressure by interest groups and politicians pursuing their private 
interests and ideologies, subject to foreign influence. We bring together different data sets 
which enables us to take into account all of the above dimensions. The explanation of each 
hypothesis in detail is provided below. 
 
5.3.1 Hypothesis on industry sector  
 
The concept of interest group competition has its roots in the economic theory of regulation 
(Peltzman, 1976; Stigler, 1971). This literature emphasizes the role of non-market 
competition among interest groups that face distributive conflicts with one another in 
determining policy outcomes. Within this context, political actors are viewed as maximizing 
political gain by balancing electoral support in the form of votes, campaign contributions and 
the like from competing interest groups to whom they “sell” policy. The economic theory of 
regulation also emphasizes the relative strength of competing interest groups in determining 
policy outcomes. That is, the members of more concentrated groups receive higher per capita 
net benefits from favourable policies, and also suffer from a lower incidence of free-riding 
than do members of less concentrated groups. As a result, more concentrated groups are a 
162 
 
more powerful source of political influence than are less concentrated groups, ceteris paribus. 
In this context, Peltzman et al. (1989, page 13) list some of the important characteristics of 
regulation that emerge from the literature on the economic theory of regulation, and the first is 
that compact, well-organized groups will tend to benefit more from regulation than broad, 
diffuse groups, which probably creates a bias in favour of a coalition of producer groups, 
because they are usually well organized relative to all consumers. Peltzman et al. (1989) also 
argues that regulatory policy will seek to preserve a politically optimal distribution of rents 
across this coalition. 
 
The economic theory of regulation predicts that different interest groups try to capture policy-
makers during power market reform process, which is confirmed in empirical studies such as 
Kroszner and Strahan (1999) and Duso (2005). If all parties are equally represented, 
competition is tougher and the policy outcome should be more efficient (Becker, 1983). 
Generally though, the industrial consumers have higher stakes and are better organized in 
shaping power market reform policies. As we know, industrial electricity consumers have a 
greater concentration relative to residential and smaller commercial customers. Also, 
industrial consumers tend to have a greater stake in electricity pricing policies because they 
typically consume more electricity than do residential and small commercial users, both in 
absolute terms and as a fraction of income. Moreover, industrial consumers are typically well-
organized through their affiliations with industry associations and trade groups. These factors 
suggest that the level of interest group pressure for politicians to shape the direction of 
electricity market reforms is greater when industrial consumers constitute a larger fraction of 
the customers. In short, due to reasons summarized above, we look at the impact of industrial 
consumers as the most organized consumer group with substantial electricity consumption on 
the reform progress. Although agricultural consumers are subsidized in many countries; this is 
because of more practical reasons than political ones. Besides, given that residential and 
163 
 
agricultural consumers face a typical free rider problem (Olson, 1965), they are often not a 
relatively powerful interest group. 
 
Depending on the perceptions of the industrial consumers regarding the benefits/losses from 
the liberalization process, countries with a larger industrial sector would be more likely to 
push for or oppose liberalization of the electricity sector. For instance, in some countries, the 
most important beneficiaries (and therefore potential supporters) of the reform programs may 
be large industrial consumers because increased efficiency and careful regulation in the sector 
may transfer huge benefits to them in the form of reduced electricity prices and better service. 
Therefore, it is reasonable to expect that industry sector in these countries supports the reform 
initiatives in the power industry; and as its size gets bigger and bigger so does its influence. 
On the other hand, in some other countries, industrial electricity prices are highly subsidized 
by the government so industrial consumers may regard liberalization policies as a threat to 
their self-interest since liberalization usually means removal of such subsidies. Taking into 
account these concerns, we formulate our first hypothesis as follows. 
 
Hypothesis 1: Industry sector has a significant impact on the extent of electricity 
market reform. As its size gets larger, so does its influence. 
 
This hypothesis is an immediate application of the economic theory of regulation. While 
investigating the impact of industrial consumers on reform progress, the latter (dependent 
variable) is represented by an “overall electricity market closeness index” and three other sub-
indicators, namely “entry barriers in electricity market”, “public ownership in electricity 
market” and “vertical integration in electricity market”. On the other hand, “industry value 
added (% of GDP)” constitutes the causal variable that represents the size of industry sector. 
The details of these variables are provided in the following sections. 
164 
 
 
5.3.2 Hypothesis on foreign financial aid 
 
In 1992, the World Bank officially changed its lending policy for electricity development 
from traditional project lending to policy lending (the Washington consensus). That is, any 
country borrowing from the Bank on power projects would have to agree to move away from 
a “single national electricity utility as a public monopoly” and adopt ownership, structural and 
regulatory reforms (Yi-chong, 2006). Other international financial institutions, such as the 
Asian Development Bank, European Bank for Reconstruction and Development, and the 
Inter-American Development Bank have followed suit (Williams and Ghanadan, 2006). 
Today, the liberalization of the infrastructure (including electricity) industries is one of the 
preconditions of any financial support program. So, we expect that countries receiving foreign 
financial support are likely to initiate a reform program in their power markets. Therefore, our 
next hypothesis is formulated as below. 
 
Hypothesis 2: Foreign financial aid and/or assistance increase the extent of 
electricity market reform. 
 
As mentioned above, electricity reform progress is expected to be influenced by international 
donor agencies through financial aid or assistance. In this context, the change in reform 
progress constitutes a possible outcome and foreign financial aid and/or assistance is one of 
the possible causes of this outcome. By developing this second hypothesis, we try to find out 
whether there is a correlation between electricity reform progress and foreign financial aid 
and/or assistance. We may think about a causal link in the event that the coefficient of the 
variable representing foreign financial aid and/or assistance turns out to be statistically 
significant. As in the case of the first hypothesis, the reform progress is represented by an 
165 
 
overall indicator and three sub-indicators while foreign financial aid and/or assistance is 
signified by the variable “net official development assistance and official aid received”, the 
details of which is presented in the following sections. 
 
5.3.3 Hypothesis on government ideology 
 
If we assume that politicians are perfect agents of their constituents and act based entirely on 
constituent interests, ideology should not affect the policy outcomes. However, in practice, 
politicians’ interests are not perfectly aligned to that of their constituents and their ideologies 
may affect policy outcomes. Parties with different ideologies may prefer different policies. 
Actually, empirical literature shows that party appears to matter (Alt and Lowry, 1994; Clarke 
and Cull, 2002; Jones et al., 2000). Alesina (1987), Hibbs (1977) and Pearce (2006) explain 
that political parties promote policies in line with their ideological preferences. Political 
ideological differences may therefore explain the reform progress in electricity industry.  
 
Although the literature has documented that political parties develop and follow policies in 
line with their ideology (Hibbs, 1977), there is a controversy on the possible direction of the 
ruling party ideology on the reform programs. On the one hand, it is usually assumed that 
right-wing governments favour privatization and liberalization process, and left-wing 
governments prefer government involvement (Bjørnskov, 2005; Duval, 2008; Pitlik, 2007; 
Potrafke, 2010). In line with this assumption, some empirical studies have found that market-
oriented and right-wing governments pursue privatization and liberalization (Bjørnskov and 
Potrafke, 2011; Bortolotti and Pinotti, 2008; Duso, 2002; Pitlik, 2007; Potrafke, 2010). On the 
other hand, some other studies have concluded that left-wing governments are associated with 
more liberalization and regulation in infrastructure industries (Chang and Berdiev, 2011). 
Dreher et al. (2009), for instance, show that reforms are more likely during the tenure of 
166 
 
former entrepreneurs, and entrepreneurs belonging to a left-wing party are more successful in 
inducing reforms than a member of a right-wing party with the same previous profession. 
 
In short, the ideological orientation of the political parties may be critically important in 
explaining the power market reform process in a country (Bortolotti and Pinotti, 2008; Duso, 
2002); however, it is not possible to determine the direction of this impact analytically.  These 
considerations lead us to the following hypotheses. 
 
Hypothesis 3: Government ideology is one of the determinants of the extent of 
electricity market reform. 
 
This hypothesis is a direct parallel with the empirical findings reported in the literature we 
summarized above. By developing this hypothesis, we try to find out whether the suggested 
correlation between a reform process and government ideology is valid for electricity markets, 
too. We may think that the findings of the previous literature hold true also for electricity 
markets if the coefficients of the variables representing government ideology turn out to be 
statistically significant. In this framework, as we mentioned before, reform progress is 
represented by an overall indicator and three sub-indicators while government ideology is 
indicated by three dummy variables that represent party orientation with respect to economic 
policy, namely “right”, “left” or “center”. The details of these variables are provided in the 
following sections. 
 
5.3.4 Hypothesis on the background of chief executive 
 
As we discussed in Section 4.3.1, there is a growing literature within political economy 
framework connecting personal traits of policy makers and implementers with policy 
167 
 
outcomes15. In the second paper, we have employed variables related to the characteristics of 
the chairperson of the regulatory agency and energy minister when reforms started. In this 
paper, on the other hand, we focus on the professional and educational background of the 
chief executive (prime minister or president) as the literature summarized in Section 4.3.1 
suggests that the decision to initiate, develop, suspend or cancel a reform process may also be 
affected by the personal backgrounds of the chief political leaders. So, although we use data 
on the backgrounds of key people in the second paper and this paper, the persons we 
concentrate on are different. Moreover, in terms of data, the second and third papers are also 
different. This paper analyses a panel data set with 1,540 maximum possible observations 
while the second paper has a cross-section dataset with 115 observations. Besides, due to lack 
of data, we could not focus on the professional backgrounds in the second paper; however, in 
this paper we investigate the impact of professional backgrounds of political leaders on the 
reform progress as well. In summary, the prior knowledge, education and experience of the 
chief executive regarding the power market liberalization process may encourage or 
discourage the reform measures. Hence, our final hypothesis is as follows. 
 
Hypothesis 4: Educational and professional background of the chief executive 
matters for the extent of electricity market reform. 
 
This hypothesis is a direct parallel with the empirical findings reported in the literature we 
summarized in Section 4.3.1. Throughout our analysis, as in the case of previous hypotheses, 
reform progress is represented by an overall indicator and three sub-indicators on reform 
progress. On the other hand, the backgrounds of the chief executives are represented by ten 
dummy variables, six of which are for professional background of the chief executive and 
                                                          
15 Since we have already provided the details of this literature in Section 4.3.1, we do not duplicate it here. 
168 
 
four for educational background. The details of these variables are available in the following 
sections. 
 
5.3.5 Control variables 
 
Control variables are independent variables not directly related to the size of industry sector, 
foreign financial aid, government ideology and the background of the chief executive, but 
may still explain a portion of the variations in reform indicators. However, we acknowledge 
that there are no clear-cut criteria to decide which variables to include as part of hypotheses 
and which ones as control variables. In some cases, the impacts of some variables are so well 
known that formulating them as hypotheses would be pointless so we relegate them to control 
variables. In some other cases, the impact of some variables is not known but these variables 
might be important; so we include them as control variables. Finally, we try to avoid testing 
too many hypotheses in a single paper and therefore limit the maximum number of 
hypotheses to four. Hence, if there are more than four variables as potential hypotheses, we 
include the first four most important ones as hypotheses and categorize the remaining as 
control variables. In addition to explanatory variables discussed above, whether and how 
much a country reforms its electricity industry depends also on other factors such as the state 
of economic development, history, political structure and so on. To improve the accuracy of 
our estimates, we control for these factors in our empirical analysis using a rich set of control 
variables, namely the size of rural population, the fairness of income distribution, the level of 
authoritarianism, party structure, political stability, electoral system, EU and OECD 
membership status, existence of electricity market reform idea, population density, electricity 
consumption, income level, average education level in the country and dependence on foreign 
trade. We briefly summarize the link between the control variables in our analysis and power 
169 
 
market reform progress below while the details of these variables are given in the following 
sections. 
 
Compared with rural consumers, urban consumers are more likely to benefit from reforms 
that reduce cross-subsidization and increase electricity service offerings in densely populated 
areas. One indicator of the relative gains of urban consumers is the tariff rebalancing 
associated with liberalization. In addition to gains from tariff rebalancing, liberalization 
usually results in an improvement in electricity service (e.g. fewer interruptions), especially in 
urban areas. Since urban consumers tend to have larger electricity consumption volumes than 
rural residents, partly due to income effect, they should benefit more from liberalization than 
rural consumers; and therefore support liberalization in electricity markets. However, in some 
countries, urban consumers may object to reform process if the price of electricity for urban 
consumers is subsidized in pre-reform period and the reforms mean removal of these 
subsidies. The share of rural population (as % of total population) may be used as a proxy for 
the relative effectiveness of the urban consumers in influencing policies. 
 
We also expect that consumers are better organized in less authoritarian and more egalitarian 
societies; and therefore have more tools to affect reform policies in these societies. Hence, we 
also use the fairness of income distribution and the level of authoritarianism as control 
variables in our analysis. Gini coefficient represents the fairness of income distribution while 
polity score indicates the level of authoritarianism in a country. However, the direction of the 
relationship between these variables and reform progress cannot be established analytically.  
 
Political stability in a country constitutes another factor with the potential to affect power 
market reforms. However, again, its impact on the reform progress cannot be predetermined 
theoretically. On the one hand, subjecting a reform program to the scrutiny of both the ruling 
170 
 
and the opposition parties may increase the credibility of the power market reform program 
for private investors (Levy and Spiller, 1996), thus increasing the possibility of success in 
implementing the program. Moreover, the division of power may help counterbalance ruling 
politicians’ discretion to pursue self-interest, creating a more competitive environment in 
policy-making (Li and Xu, 2002). On the other hand, implementation of reforms may require 
a stable political environment in the form of single-party governments (rather than coalition 
governments), presidential regimes (rather than parliamentary ones) and chief executives with 
longer years in office. Therefore, we include three additional control variables that represent 
political stability into our analysis, namely party structure (single-party or coalition), the 
years the chief executive has been in office, and electoral system (presidential or 
parliamentary). 
 
It is quite possible that countries prefer to liberalize their power markets after reforms are 
implemented successfully in other countries. The dummy variable the existence of electricity 
market reform idea represents the impact of previous international experience on subsequent 
reforms in other countries. It takes the value 1 after 1989 when the electricity market reform 
was implemented, for the first time, in a full scale in a developed country (i.e. the UK); the 
years before 1989 take the value 0. 
 
Indictors of economic development and population structure are also expected to affect 
reforms as they may have implications for the demand for power market reforms. So, we also 
include control variables representing them, namely dummy variables for EU and OECD 
membership status, population density, electricity consumption, GDP per capita, average 
number of years of education and imports of goods and services. 
 
171 
 
5.4 Methodology 
 
As underlined by Jamasb et al. (2006), there is a lack of generally accepted and measured 
indicators for monitoring the progress, impacts, and performance of electricity sector reforms. 
Since the aim of this paper is to propose a framework for analysing the power market reforms 
from a political economy perspective, we face with the same problem. That is, we need to, 
first, evaluate possible impact of political economic environment of a country on electricity 
market reform process in this country; second, decide which indicators to use in our study 
and; finally, specify methods to measure them. Let me focus on these tasks one by one. 
 
Many scholars have studied many aspects of power market reforms. For instance, Stigler and 
Friedland (1962) analysed the effects of regulation of electricity rates and concluded that 
regulation had not resulted in lower electricity rates. Some other scholars discussed the impact 
of political economic environment on the reforms. However, to the best of our knowledge, no 
econometric study has been done so far on the relationship between political economic 
environment and power market reform progress. Therefore, we cannot find empirical 
evidence in the applied literature concerning the direction of this relationship. To carry out 
our analysis, we need to decide which indicators to be used in the study. Since we are 
interested in the impact of political economic variables on power market reform process, we 
need variables representing political economic environment of a country and those 
representing the scale and intensity of the reform process. In addition to these variables, we 
also utilize a set of control variables which are assumed to be endogenous to reform process 
and explain a portion of the variations in reform progress. Another challenge we face in this 
study relates to the measurement of the variables. For an indicator to be useful it needs to be 
based on a clear definition and to be measurable. This is equally important whether it is 
expressed in physical, monetary or qualitative terms. In fact, most of the economic and 
172 
 
industry indicators in our study are measured in some form of monetary or physical unit; and 
therefore, easy to include into the study. However, the extent and scope of electricity reforms 
are not quantifiable in physical or monetary units. The main electricity reform measures, such 
as privatization, unbundling of functions, wholesale markets and independent regulation, are 
generally established gradually and have a qualitative dimension. Accounting for these 
measures with the use of dummy variables, as sometimes done, does not reflect extent or 
intensity. To overcome this problem, we used electricity market reform indicators constructed 
by international organizations (namely, OECD and EBRD). 
 
We specify our dependent variables (that is, reform indicators) as a function of (i) political 
economic variables (comparable cross-country indicators), (ii) a set of controls, (iii) country-
specific effects (these are assumed to be exogenous and to exist independently of reform 
process, but may explain a portion of the variation in reform progress) and (iv) other 
unobserved variables that influence the reform process. These variables are then used in panel 
regressions to assess their impact on variables we are interested in. In panel regressions, the 
exploitation of both cross-country and time-series dimensions of the data allows for control of 
country-specific effects. 
 
We start with the standard panel data regression equation below. 
 
  (5.1) 
  
In the model, i and t represent unit of observation and time period, respectively. j and p are 
indices used to differentiate between observed and unobserved variables. Xji and Zpi represent 
observed and unobserved variables, respectively. Xji includes both political economic 
1
2 1
k s
it j jit p pi it
j p
Y X Z tβ β γ δ ε
= =
= + + + +∑ ∑
173 
 
variables and control variables. Yit is dependent variable (that is, electricity market reform 
indicators). is the disturbance term and t is time trend term. Because the Zpi variables are 
unobserved, there is no means of obtaining information about the component of the 
model. For convenience, we define a term , known as the unobserved effect, representing 
the joint impact of the Zpi variables on Yit. So, our model may be rewritten as follows: 
 
  (5.2) 
 
Now, the characterization of the  component is crucially important in the analysis. If 
control variables are so comprehensive that they capture all relevant characteristics of the 
individual, there will be no relevant unobserved characteristics. In that case, the  term may 
be dropped and pooled data regression (OLS) may be used to fit the model, treating all the 
observations for all time periods as a single sample. However, since we are not sure whether 
control variables in our models capture all relevant characteristics of the countries, we cannot 
directly carry out a pooled data regression of Y on X. If we were to do so, it would generate 
an omitted variable bias. Therefore we prefer to use either a Fixed Effects (FE) or Random 
Effects (RE) regression. In FE model, the country-specific effects ( ) are assumed to be the 
fixed parameters to be estimated. In RE model, the country-specific effects ( ) are treated as 
stochastic. The fixed effect model produces consistent estimates, while the estimates obtained 
from the random effect model will be more efficient. There are more than 90 countries in the 
world where a reform process has been initiated so far but data are available only for 55 
countries. That is, our sample is limited by data availability. Therefore, we cannot be sure 
whether the observations in our model may be described as being a random sample from a 
given population; and cannot directly decide which regression specification (FE, RE or OLS) 
itε
p piZγ∑
iα
1
2
i it
k
it j jit
j
tY X α δ εβ β
=
+ += + +∑
iα
iα
iα
iα
174 
 
to use. It will be decided in the course of the analysis based on Hausman test and Breusch and 
Pagan Lagrangian Multiplier (BPLM) test. 
 
5.5 Overview of data 
 
Our data set is based on a panel of 55 countries for a period beginning in 1975 and extending 
through 2010. List of countries in our data set is available in Figure 5.1 and Figure 5.2. Years 
1975 and 2010 represent, respectively, the earliest and the last year for which data are 
available at the time the research is conducted. The countries in our sample are determined by 
data availability, especially by data on electricity market reform indicators. In this study, we 
assume that countries are comparable to each other; however, we also acknowledge the 
possible lack of comparability among countries. The total number of maximum observations 
for each variable is 1,540. Because of missing observations, our panel is unbalanced. 
 
In total, we use 37 variables in our study. Out of 37 variables, 4 of them are used as dependent 
variables, 15 as explanatory variables and remaining 18 as control variables. Since reform 
progress may be affected by many factors, it seems worth using 33 variables as explanatory or 
control variables. The variables used in the study are entry barriers, public ownership and 
vertical integration in electricity market; overall electricity market closeness index; industry 
value added (% of GDP); net official development assistance and official aid received (current 
billion US$); party orientation with respect to economic policy (right, left or centre); 
professional background of the chief executive (entrepreneur, scientist (economist), military, 
politician, scientist (other) or unknown/other); educational background of the chief executive 
(economics, natural science, other university or unknown/other); party structure (single-party 
or coalition); the years the chief executive has been in office; electoral system (parliamentary 
or presidential regime); rural population (% of total population); gini coefficient; polity score 
175 
 
(-10,+10); dummy variables representing EU members, OECD members or the existence of 
electricity market reform idea; population density (people per square km of land area); 
electricity consumption (MWh per capita); GDP per capita (PPP, current thousand 
international $); average number of years of adult (15+) education; imports of goods and 
services as % of GDP. Table 5.1 shows descriptive statistics of the variables in our analysis. 
 
Table 5.1 Descriptive statistics of the variables 
Variables (units) Mean Std. Dev. Min Max 
# of 
Obser. 
# of 
Ctrys 
Dependent Variables       
Entry barriers in electricity market (0-6) 4.59 2.26 0 6 990 30 
Public ownership in electricity market (0-6) 4.56 1.80 0 6 990 30 
Vertical integration in electricity market (0-6) 4.65 2.03 0 6 990 30 
Overall electricity market closeness index (0-6) 4.46 1.61 0 6 1,540 55 
Explanatory Variables       
Industry value added (% of GDP) 32.39 7.43 10.29 69.92 1,415 55 
Net official development assistance and official 
aid received (current billion US$) 
0.11 0.28 -0.46 3.79 1,408 55 
Party orientation with respect to economic policy       
     - Right 0.40 0.49 0 1 1,218 51 
     - Left 0.44 0.50 0 1 1,218 51 
     - Center 0.15 0.36 0 1 1,218 51 
Professional background of chief executive       
     - Entrepreneur 0.06 0.24 0 1 1,429 54 
     - Scientist (Economist) 0.04 0.21 0 1 1,429 54 
     - Military 0.07 0.25 0 1 1,429 54 
     - Politician 0.63 0.48 0 1 1,429 54 
     - Scientist (Other) 0.27 0.45 0 1 1,429 54 
     - Unknown/other 0.37 0.48 0 1 1,429 54 
Educational background of chief executive       
     - Economics 0.25 0.43 0 1 1,429 54 
     - Natural science 0.18 0.38 0 1 1,429 54 
     - Other university 0.47 0.50 0 1 1,429 54 
     - Unknown/other 0.14 0.35 0 1 1,429 54 
176 
 
Variables (units) Mean Std. Dev. Min Max 
# of 
Obser. 
# of 
Ctrys 
Control Variables       
Party Structure (1: single-party, 0: coalition) 0.46 0.50 0 1 1,493 53 
The years the chief executive has been in office 4.35 3.84 1 35 1,437 54 
Electoral system (parliamentary regimes) 0.68 0.47 0 1 1,475 55 
Rural population (% of total population) 33.95 14.47 2.66 73.60 1,514 55 
Gini coefficient (0-100) 30.43 6.75 16.63 57.40 760 54 
Polity score (-10,+10) 6.31 6.13 -10 10 1,357 53 
EU member (0-1) 0.30 0.46 0 1 1,540 55 
OECD member (0-1) 0.56 0.50 0 1 1,540 55 
Existence of electricity market reform idea (0-1) 0.73 0.45 0 1 1,540 55 
Population density 
(people per sq. km of land area) 
101.26 104.35 1.40 499.96 1,428 55 
Log of population density 4.00 1.34 0.33 6.21 1,428 55 
Electricity consumption (MWh per capita) 5.90 4.99 0.34 36.85 1,450 54 
Log of electricity consumption 1.47 0.80 -1.07 3.61 1,450 54 
GDP per capita (PPP, current thousand int. $) 14.34 10.83 0.73 84.41 1,307 55 
Log of GDP per capita 2.32 0.92 -0.32 4.44 1,307 55 
Average number of years of education received 
by people ages 15 and older 
9.27 1.68 2.92 12.75 1,364 47 
Imports of goods and services (% of GDP) 41.33 21.00 5.88 143.72 1,427 55 
Log of imports of goods and services 3.59 0.53 1.77 4.97 1,427 55 
 
Appendix III-4 presents the correlation among explanatory and control variables. As can be 
seen in Appendix III-4, independent variables are not highly correlated with each other. Out 
of 196 correlations (between 15 explanatory and 14 control variables) shown in Appendix III-
4, 159 correlations are between -0.3 and +0.3. When we look at the limited number of more 
correlated variables, we could not see a counter intuitive or contradictory relationship. Almost 
all high correlations are understandable. For instance, the correlation between OECD 
membership and GDP per capita is +0.6403, which is quite natural if we take into account the 
fact that OECD countries are usually the wealthy ones. So, from an econometric point of 
view, correlation among independent variables does not constitute a problem in our analysis. 
177 
 
 
Data on overall electricity market closeness index are obtained from Conway and Nicolett 
(2006) and EBRD16 (2011). Conway and Nicolett (2006) provide data for 30 OECD 
countries. They also provide data on sub indicators of reform process; namely entry barriers, 
public ownership and vertical integration. Indicators for entry barriers focus on terms and 
conditions for third party access (TPA) and the extent of choice of supplier for consumers. In 
the electricity sector this is supplemented by information on the existence of a liberalised 
wholesale market for power, which is an important issue in most OECD countries. Indicators 
for public ownership record the prevailing ownership structure in the various segments of the 
electricity sector, ranging from fully private to fully public. The scoring allows for mixed 
ownership arrangements in which the natural monopoly segments remain under public hand. 
Indicators for vertical integration focus on whether competitive activities such as generation 
and supply of electricity are separated from natural monopoly activities such as the national 
grid and/or local distribution. The degree of separation ranges from full integration to mere 
legal/accounting separation to separation into different companies owned by different 
shareholders. The assumption here, reflecting industrial organisation theory, is that the scope 
for anti-competitive behaviour is largest when an electricity company simultaneously controls 
the network and operates in upstream or downstream competitive markets. Table 5.2 shows 
the composition of the electricity market closeness index (Conway and Nicolett, 2006). As 
can be seen in Table 5.2, the index ranges from 0 to 6 where 0 represents the fully open 
market in which entry barriers, public ownership and vertical integration are minimized and a 
score of 6 is given to a closed market. 
 
                                                          
16 European Bank for Reconstruction and Development. 
178 
 
Table 5.2 Composition of the electricity market closeness index 
 
Weights 
by theme 
(bj) 
Question 
Weights 
(ck) 
Coding of Data 
Entry regulation: 1/3     
o How are the terms and conditions of 
third party access (TPA) to the electricity 
transmission grid determined? 
 1/3 
Regulated TPA 
0 
Negotiated TPA 
3 
No TPA 
6 
o Is there a liberalised wholesale market 
for electricity (a wholesale pool)? 
 1/3 
Yes 
0 
No 
6 
o What is the minimum consumption 
threshold that consumers must exceed in 
order to be able to choose their electricity 
supplier? 
 1/3 
No threshold 
0 
<250 gigawatts 
1 
Between 250 
and 500 
gigawatts 
2 
Between 500 
and 1000 
gigawatts 
3 
More than 
1000 gigawatts 
4 
No consumer 
choice 
6 
Public ownership: 1/3        
o What is the ownership structure of the 
largest companies in the generation, 
transmission, distribution, and supply 
segments of the electricity industry? 
 1 
Private 
0 
Mostly Private 
1.5 
Mixed 
3 
Mostly Public 
4.5 
Public 
6 
Vertical Integration: 1/3        
o What is the degree of vertical separation 
between the transmission and generation 
segments of the electricity industry? 
 1/2 
Separate Companies 
0 
Accounting separation 
3 
Integrated 
6 
o What is the overall degree of vertical 
integration in the electricity industry? 
 1/2 
Unbundled 
0 
Mixed 
3 
Integrated 
6 
Country scores (0-6) Σjbj Σkck answerjk 
 
179 
 
EBRD (2011) provides a similar indicator for additional 25 developing countries where it 
operates. The data from EBRD are available on a 1-4 scale. EBRD ratings are based on the 
following scale: 
 
Score 1:  Power sector operates as government department with few commercial 
freedoms or pressures. Average prices well below costs, with extensive 
cross-subsidies. Monolithic structure, with no separation of different 
parts of the business. 
 
Score 2: Power company distanced from government, but there is still political 
interference. Some attempt to harden budget constraints, but effective 
tariffs are low. Weak management incentives for efficient performance. 
Little institutional reform and minimal, if any, private sector 
involvement. 
 
Score 3: Law passed providing for full-scale restructuring of industry, including 
vertical unbundling through account separation and set-up of regulator. 
Some tariff reform and improvements in revenue collection. Some 
private sector involvement. 
 
Score 4: Separation of generation, transmission and distribution. Independent 
regulator set up. Rules for cost-reflective tariff-setting formulated and 
implemented. Substantial private sector involvement in distribution 
and/or generation. Some degree of liberalisation. 
 
180 
 
Score 4+: Tariffs cost-reflective and provide adequate incentives for efficiency 
improvements. Large-scale private sector involvement in the unbundled 
and well-regulated sector. Fully liberalised sector with well-functioning 
arrangements for network access and full competition in generation. 
 
To establish uniformity between two data sets, the data from EBRD (2011) are converted into 
6-0 scale. Figure 5.1 and Figure 5.2 provide the change in electricity market closeness index 
from 1989 to 2007 for the countries in our dataset. 
 
Electricity market closeness index assigns a score to each country for each year based on the 
reform status of that country in that year. It is an indicator of reform progress, rather than 
reform success. Table 5.3 classifies 55 countries in our dataset into four groups based on their 
market closeness index in 2000. This classification does not give an idea about reform success 
and clearly does not present a full picture of the situation in each country; however, it still 
successfully groups countries in terms of reform progress. For instance, one of the criteria 
used to determine market closeness score is private participation into electricity industry. In 
2000, Norway had probably much more competitive (and therefore successful) power market 
structure than Germany; but it is placed into the second group while Germany is placed into 
the first one in Table 5.3. The main reason for this is the fact that Norway prefers a market 
model based on “competition without privatization” while Germany has a long tradition of 
private participation in its electricity industry, which enables Germany to get a higher score 
than Norway. That is, electricity market closeness index measures the distance between actual 
market structure of a country and an ‘ideal’ one where electricity industry is unbundled, 
privately-owned and fully competitive. The score of each country differs since the distance of 
each country from this ‘ideal’ point is different. However, the evidence so far does not fully 
confirm that this ‘ideal’ structure is also the most successful one. As we mentioned before, in 


1RUZD\WKHUHKDVQRWEHHQDODUJHVFDOHSULYDWL]DWLRQSURJUDPEXWVWLOOLWVPDUNHWLVSUREDEO\
PXFKPRUH VXFFHVVIXO WKDQ WKRVH RIPDQ\ FRXQWULHV WKDW LPSOHPHQWHG VXFK D SULYDWL]DWLRQ
SURJUDP LQ WKHLU SRZHU LQGXVWULHV ,Q VKRUW HOHFWULFLW\ PDUNHW FORVHQHVV LQGH[ LV D XVHIXO

)LJXUH(OHFWULFLW\PDUNHWFORVHQHVVLQGH[LQ2(&'FRXQWULHV


0 1 2 3 4 5 6
Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
Electricity market closeness index in 2007
Electricity market closeness index in 1989


)LJXUH(OHFWULFLW\PDUNHWFORVHQHVVLQGH[LQFRXQWULHVZKHUH(%5'RSHUDWHV


LQGLFDWRUIRURXUSXUSRVHVEHFDXVHZHIRFXVRQUHIRUPSURJUHVVUDWKHUWKDQUHIRUPVXFFHVVLQ
WKLVWKHVLVDQGLWSURYLGHVXVZLWKDWRROWRPHDVXUHUHIRUPSURJUHVVLQHDFKFRXQWU\DQGHDFK
\HDU$SDUWIURPWKLVLWVKRXOGQRWDQGFDQQRWEHXVHGWRPHDVXUHUHIRUPVXFFHVVRUDVLPLODU
SKHQRPHQRQLQRWKHUVWXGLHV

0 1 2 3 4 5 6
Albania
Armenia
Azerbaijan
Belarus
Bosnia and Herz.
Bulgaria
Croatia
Estonia
FYR Macedonia
Georgia
Kazakhstan
Kyrgyz Republic
Latvia
Lithuania
Moldova
Mongolia
Montenegro
Romania
Russian Federation
Serbia
Slovenia
Tajikistan
Turkmenistan
Ukraine
Uzbekistan
Electricity market closeness index in 2007
Electricity market closeness index in 1989
183 
 
Table 5.3 Country groups based on electricity market closeness index in 2000 
Score 
0-1.5 
(Group 1) 
Score 
1.5-3.0 
(Group 2) 
Score 
3.0-4.5 
(Group 3) 
Score 
4.5-6.0 
(Group 4) 
United Kingdom New Zealand Czech Republic Canada 
Spain Norway Ukraine Luxembourg 
Germany Australia Denmark Korea 
Finland Armenia Italy Slovak Republic 
Estonia Bulgaria Albania Greece 
 
Georgia Bosnia And Herzegovina Switzerland 
 
Moldova Croatia Iceland 
 
Portugal FYR Macedonia Mexico 
 
United States Kyrgyz Republic Turkey 
 
Sweden Austria Belarus 
 
Netherlands France Mongolia 
 
Kazakhstan Ireland Montenegro 
 
Latvia Poland Tajikistan 
 
Lithuania Azerbaijan Turkmenistan 
 
Romania Russian Federation Uzbekistan 
 
Slovenia Serbia 
 
 
Belgium Hungary 
 
 
Japan 
   
The data regarding industry value added as % of GDP, net official development assistance 
and official aid received in current billion US$ and rural population as % of total population 
are taken from World Bank (2011). Figure 5.3 shows histogram of industry value added 
variable. Figure 5.4 presents total development assistance and aid received between 1990 and 
2007. Countries that did not receive any aid or assistance during this period are excluded from 
Figure 5.4. We prefer to use absolute values of aid/assistance in our analysis instead of the 
size of aid as a percentage of GDP because when we express the size of aid as a percentage of 
GDP we get very small (very close to zero) values for many countries, which may create 
serious econometric problems. Also, for aid-receiving countries, what is really important is 
usually not the size of the aid or assistance but the credibility provided by donor agencies 
 


)LJXUH+LVWRJUDPRILQGXVWU\YDOXHDGGHGYDULDEOH

,0):RUOG%DQNHWFE\TXDOLI\LQJWKHPIRUWKHDLGDVVLVWDQFHZKLFKXVXDOO\HQDEOHVDLG
UHFHLYLQJ FRXQWULHV WR JHW IXUWKHU ORZ FRVW FUHGLW IURP LQWHUQDWLRQDO SULYDWH ILQDQFLDO
LQVWLWXWLRQV%HVLGHVWKHDEVROXWHYDOXHVRIDLGDVVLVWDQFHPD\PHDVXUHILQDQFLDOVLJQLILFDQFH
DQGZLOOLQJQHVVWREHRSHQWRUHIRUPUHFRPPHQGDWLRQVRIGRQRUFRXQWULHV

'DWDRQSDUW\RULHQWDWLRQZLWKUHVSHFWWRHFRQRPLFSROLF\SDUW\VWUXFWXUHWKH\HDUVWKHFKLHI
H[HFXWLYHKDVEHHQLQRIILFHDQGHOHFWRUDOV\VWHPRULJLQDWHIURP.HHIHU,QWKLVVWXG\
FKLHIH[HFXWLYHUHIHUVWRWKHKHDGRIWKHH[HFXWLYHEUDQFK,QVRPHFRXQWULHVFKLHIH[HFXWLYH
LV ³SULPHPLQLVWHU´ZKLOH LW LV FDOOHG ³SUHVLGHQW´ LQ VRPHRWKHU FRXQWULHV 3URIHVVLRQDO DQG
HGXFDWLRQDO EDFNJURXQG RI WKH FKLHI H[HFXWLYH GDWD DUH SDUWO\ FROOHFWHG E\ WKH DXWKRU DQG
SDUWO\SURYLGHGE\'UHKHUHWDO:KLOHGHFLGLQJRQZKLFKHGXFDWLRQDODQGSURIHVVLRQDO
EDFNJURXQGV WR LQFOXGH LQWR RXU DQDO\VLV ZH VHOHFWHG WRS ILYH PRVW FRPPRQ SURIHVVLRQDO
EDFNJURXQGVHQWUHSUHQHXUVFLHQWLVWHFRQRPLVWPLOLWDU\SROLWLFLDQDQGVFLHQWLVWRWKHUDQG
WRS WKUHH HGXFDWLRQDO EDFNJURXQGV HFRQRPLFV QDWXUDO VFLHQFH DQG RWKHU XQLYHUVLW\0DQ\
FKLHIH[HFXWLYHVSULPHPLQLVWHUVRUSUHVLGHQWVHVSHFLDOO\LQGHYHORSLQJFRXQWULHVFRPHIURP
DPLOLWDU\EDFNJURXQGVRZHLQFOXGHDGXPP\YDULDEOHWRUHSUHVHQWWKLVEDFNJURXQGDVZHOO
%HVLGHV WKH FKLHI H[HFXWLYHVZLWK DPLOLWDU\ EDFNJURXQGPD\ZRUU\ UHODWLYHO\PRUH DERXW

0
5
10
15
20
25
N
um
be
r o
f c
ou
nt
rie
s 
% of GDP 
Industry value added in 1992
Industry value added in 2007


)LJXUH7RWDOGHYHORSPHQWDVVLVWDQFHDQGDLGUHFHLYHG

HQHUJ\ VHFXULW\ DQG KDYH D VSHFLILF XVXDOO\ QHJDWLYH DWWLWXGH WRZDUGV WKH UHIRUPV ,Q RXU
GDWDVHWSROLWLFDO OHDGHUVPD\KDYH WZRRUPRUHSURIHVVLRQDORUHGXFDWLRQDOTXDOLILFDWLRQVDW
WKH VDPH WLPH +RZHYHU WKLV GRHV QRW DPRXQW WR GRXEOHFRXQWLQJ DV ZH EHOLHYH WKDW WKH
LPSDFWRID FKLHIH[HFXWLYHRQWKHUHIRUPSURJUHVVLVGHWHUPLQHGE\KLVKHUHQWLUHSURIHVVLRQDO
RUHGXFDWLRQDOEDFNJURXQGQRWE\RQHRIWKHPRQO\)RULQVWDQFHLIZHDVVXPHWKDWDSULPH
PLQLVWHUKDVDQXQGHUJUDGXDWHGHJUHHLQHFRQRPLFVDQGDJUDGXDWHGHJUHHLQODZLW LVTXLWH
QDWXUDO WKDW KLVKHUGHFLVLRQVRQSRZHUPDUNHW UHIRUPDUH DIIHFWHGE\KLVKHUNQRZOHGJH LQ
0 5 10 15 20 25
Albania
Armenia
Azerbaijan
Belarus
Bosnia and Herz.
Bulgaria
Croatia
Czech Republic
Estonia
FYR Macedonia
Georgia
Hungary
Kazakhstan
Kyrgyz Republic
Latvia
Lithuania
Mexico
Moldova
Mongolia
Montenegro
Poland
Romania
Russian Federation
Serbia
Slovak Republic
Slovenia
Tajikistan
Turkey
Turkmenistan
Ukraine
Uzbekistan
current billion US$ 
186 
 
both economics and law. Therefore, we had to specify both economics and law as the 
educational background of this prime minister, which clearly may not regarded as double-
counting. We also created “unknown/other” category to represent other educational and 
professional backgrounds. 
 
Gini coefficient17 and polity score data come from UNU-WIDER (2011) and Center for 
Systemic Peace (CSP, 2010), respectively. The EU has the sustained commitment of the 
European Commission within a wider single market agenda (which encompasses all industrial 
sectors) and a political system where energy reforms can become a part of the international 
bargaining process, which maintains the pressure for further electricity market reforms 
(Pollitt, 2008b). EU directives are expected to have an influence on the reform progress 
especially in EU member countries. They are usually decided as a result of a political process 
among member countries; that is, they are not pure technical documents. So, we put a dummy 
variable for the EU countries. Dummy variables representing being an EU member, an OECD 
member and the existence of electricity market reform idea are constructed by the author. In 
our dataset, OECD membership is a variable in a set of OECD countries because our panel 
dataset covers 1975-2010 period and some of current OECD members became a member after 
1975. For instance, Mexico joined OECD on 18 May 1994. So, for Mexico, “OECD 
membership” variable takes the value of “0” for 1975-1993 period and “1” for 1994-2010 
period. The dummy variable for the existence of electricity market reform idea takes the value 
1 after 1989 when the electricity market reform was implemented, for the first time, in a full 
scale in a developed country (i.e. the UK); the years before 1989 take the value 0. 
 
                                                          
17 The Gini coefficient is a measure of the inequality of a distribution, a value of 0 expressing total equality and a 
value of 1 maximal inequality. 
187 
 
World Bank (2011) provides data on population density (people per sq. km of land area), 
electricity consumption (MWh per capita), GDP per capita (PPP, current thousand int. $) and 
imports of goods and services as % of GDP. Average number of years of adult (15+) 
education is taken from Barro and Lee (2010). The data from Barro and Lee (2010) are 
available with 5-year intervals; to ensure conformity with other data, we converted them into 
yearly data by linear interpolation. 
 
5.6 Empirical analysis 
 
Our analysis is composed of estimation of three main groups of models to test our hypotheses. 
Each main group includes two sub-groups of models: one for sub-indicators (entry barriers, 
public ownership and vertical integration in OECD countries) and another for overall 
indicator (OECD countries and non-OECD countries). In total, we estimate 15 models. Since 
using logarithms of variables enables us to interpret coefficients easily and is an effective way 
of shrinking the distance between values, we transform population density, electricity 
consumption per capita, GDP per capita and imports of goods and services as % of GDP 
variables into logarithmic form and use these transformed variables in our models. 
 
We start our analysis by applying Hausman test for fixed versus random effects in each 
model18. As usual, we prefer 5% significance level so any p-value less than 0.05 from 
Hausman test implies that we should reject the null hypothesis of there being no systematic 
difference in the coefficients. In other words, Hausman test with a p-value up to 0.05 indicates 
significant differences in the coefficients. Therefore, in our analysis, if we get a p-value less 
than 0.05, we choose fixed effects model. However, if p-value from Hausman test is above 
0.05, we cannot reject the null hypothesis of there being no systematic difference in the 
                                                          
18 Throughout the paper, model estimations are carried out and cross-checked by Stata 11.2 and Eviews 7.1. 
188 
 
coefficients at 5% level. In such a case, we apply Breusch and Pagan Lagrangian Multiplier 
(BPLM) test for random effects in order to decide on using either pooled OLS or random 
effects in our analysis. This test is developed to detect the presence of random effects. In this 
test, the null hypothesis is that variances of groups are zero; that is, there is no unobserved 
heterogeneity, all groups are similar. If the null is not rejected, the pooled regression model is 
appropriate. That is, if the p-value of BPLM test is below 0.05, we reject the null, meaning 
that random effects specification is the preferred one. If it is above 0.05, we prefer pooled 
OLS specification to carry out our regression. Tables below show a summary of estimation 
results that present statistically significant coefficients and their standard errors. Full details of  
 
Table 5.4 Estimation results for the models testing Hypothesis 1 (sub-indicators) 
Dependent Variables 
Entry barriers 
(0-6) 
Public Ownership 
(0-6) 
Vertical integration 
(0-6) 
(OECD countries) (OECD countries) (OECD countries) 
Explanatory variable    
Industry value added (% of GDP) 0.129*** (0.027) 0.076*** (0.015) 0.128*** (0.025) 
Control variables    
Rural population (% of total population) NS NS NS 
Gini coefficient (0-100) NS NS NS 
Polity score (-10,+10) 0.178*** (0.046) NS 0.109*** (0.042) 
EU member (0-1) -1.61*** (0.319) NS -1.407*** (0.294) 
OECD member (0-1) 1.717*** (0.419) NS 0.907** (0.386) 
Existence of electricity market reform idea (0-1) 1.078*** (0.274) NS 0.521** (0.252) 
Log of population density NS 9.221*** (1.357) NS 
Log of electricity consumption per capita (MWh) 2.566*** (0.796) -1.188*** (0.423) NS 
Log of GDP per capita (PPP, cur. thousand int. $) -5.201*** (0.536) -1.157*** (0.285) -3.679*** (0.494) 
Average number of years of adult (25+) education NS NS NS 
Log of imports of goods and services (% of GDP) -2.415*** (0.582) -0.936*** (0.31) -1.444*** (0.536) 
Constant NS -25.833*** (5.876) 17.055* (10.177) 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
189 
 
estimation results are provided in Appendix III-3 including the full estimation output, the 
number of observations and the countries included in each model, results of Hausman and 
BPLM tests and preferred specifications based on these tests. 
 
Table 5.5 Estimation results for the models testing Hypothesis 1 (overall indicator) 
Dependent Variables 
Overall indicator 
(0-6) 
Overall indicator 
(0-6) 
(OECD countries) (Non-OECD countries) 
Explanatory variable   
Industry value added (% of GDP) 0.111*** (0.018) 0.057*** (0.012) 
Control variables   
Rural population (% of total population) NS -0.157* (0.082) 
Gini coefficient (0-100) NS NS 
Polity score (-10,+10) 0.091*** (0.03) NS 
EU member (0-1) -0.927*** (0.21) 0.463* (0.25) 
OECD member (0-1) 0.889*** (0.275) (omitted) 
Existence of electricity market reform idea (0-1) 0.569*** (0.18) (omitted) 
Log of population density 3.153* (1.675) NS 
Log of electricity consumption per capita (MWh) NS 4.001*** (0.568) 
Log of GDP per capita (PPP, cur. thousand int. $) -3.345*** (0.352) -3.009*** (0.351) 
Average number of years of adult (25+) education NS NS 
Log of imports of goods and services (% of GDP) -1.598*** (0.382) NS 
Constant NS NS 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
 
  
190 
 
Table 5.6 Estimation results for the models testing Hypothesis 2 (sub-indicators) 
Dependent Variables 
Entry barriers 
(0-6) 
Public Ownership 
(0-6) 
Vertical integration 
(0-6) 
(OECD countries) (OECD countries) (OECD countries) 
Explanatory variable    
Net official assistance and aid received -0.628** (0.311) NS NS 
Control variables    
EU member (0-1) -1.06*** (0.234) 0.227* (0.119) -1.171*** (0.214) 
OECD member (0-1) 2.136*** (0.287) -0.371** (0.147) 1.125*** (0.262) 
Existence of electricity market reform idea (0-1) 1.125*** (0.178) NS 0.626*** (0.162) 
Log of population density NS 7.314*** (0.73) 2.843** (1.307) 
Log of electricity consumption per capita (MWh) 2.984*** (0.408) -0.354* (0.208) 1.297*** (0.373) 
Log of GDP per capita (PPP, cur. thousand int. $) -5.987*** (0.347) -1.273*** (0.177) -4.536*** (0.316) 
Average number of years of adult (25+) education -0.226** (0.103) -0.226*** (0.052) NS 
Log of imports of goods and services (% of GDP) -2.491*** (0.399) -0.621*** (0.203) -2.002*** (0.364) 
Constant 23.5*** (5.198) -16.994*** (2.651) 7.825* (4.747) 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level 
 
Table 5.7 Estimation results for the models testing Hypothesis 2 (overall indicator) 
Dependent Variables 
Overall indicator 
(0-6) 
Overall indicator 
(0-6) 
(OECD countries) (Non-OECD countries) 
Explanatory variable   
Net official assistance and aid received -0.334* (0.202) -0.557** (0.252) 
Control variables   
EU member (0-1) -0.668*** (0.152) NS 
OECD member (0-1) 0.964*** (0.187) (omitted) 
Existence of electricity market reform idea (0-1) 0.576*** (0.116) (omitted) 
Log of population density 3.429*** (0.933) NS 
Log of electricity consumption per capita (MWh) 1.309*** (0.266) 2.208*** (0.309) 
Log of GDP per capita (PPP, cur. thousand int. $) -3.932*** (0.226) -1.898*** (0.219) 
Average number of years of adult (25+) education -0.115* (0.067) -1.353*** (0.209) 
Log of imports of goods and services (% of GDP) -1.705*** (0.26) NS 
Constant NS 11.222* (6) 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
 
191 
 
Table 5.8 Estimation results for the models testing Hypotheses 3 and 4 (sub-indicators) 
Dependent Variables 
Entry barriers 
(0-6) 
Public Ownership 
(0-6) 
Vertical integration 
(0-6) 
(OECD countries) (OECD countries) (OECD countries) 
Explanatory variables    
Economic policy orientation of ruling party    
     - Right NS NS -0.526** (0.211) 
     - Left -0.423* (0.246) NS -0.38* (0.214) 
     - Center (omitted) (omitted) (omitted) 
Professional background of chief executive    
     - Entrepreneur NS -0.457*** (0.161) -0.591** (0.264) 
     - Economics 1.333*** (0.389) NS 1.982*** (0.335) 
     - Military NS NS NS 
     - Politician 0.482** (0.213) -0.201* (0.117) 0.443** (0.191) 
     - Scientist, Other 0.446* (0.243) -0.484*** (0.132) NS 
     - Unknown/other 0.516** (0.227) -0.302** (0.124) 0.725*** (0.202) 
Educational background of chief executive    
     - Economics NS NS 0.814* (0.468) 
     - Natural science NS 1.123*** (0.33) 1.75*** (0.541) 
     - Other university NS NS NS 
     - Unknown/other NS NS NS 
Control variables    
Single-party government (0-1) NS -0.144* (0.085) -0.229* (0.139) 
The years the chief executive has been in office 0.042** (0.018) 0.033*** (0.009) 0.044*** (0.015) 
Parliamentary regimes (0-1) NS -0.407* (0.231) NS 
EU member (0-1) -0.829*** (0.233) NS -1.282*** (0.206) 
OECD member (0-1) 1.697*** (0.327) -0.474*** (0.181) 0.966*** (0.293) 
Existence of electricity market reform idea (0-1) 0.749*** (0.181) NS 0.384** (0.158) 
Log of population density 0.606*** (0.14) NS 0.47** (0.218) 
Log of electricity consumption per capita (MWh) 2.886*** (0.318) NS 1.778*** (0.349) 
Log of GDP per capita (PPP, cur. thousand int. $) -5.73*** (0.321) -0.638*** (0.187) -4.266*** (0.299) 
Average number of years of adult (25+) education -0.24*** (0.085) -0.333*** (0.055) NS 
Log of imports of goods and services (% of GDP) -1.202*** (0.272) NS -1.484*** (0.296) 
Constant 16.661*** (1.214) 9.582*** (0.972) 14.172*** (1.352) 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
192 
 
Table 5.9 Estimation results for the models testing Hypotheses 3 and 4 (overall indicator) 
Dependent Variables 
Overall indicator 
(0-6) 
Overall indicator 
(0-6) 
(OECD countries) (Non-OECD countries) 
Explanatory variables   
Economic policy orientation of ruling party   
     - Right NS  (omitted) 
     - Left -0.268* (0.159) NS  
     - Center (omitted) NS  
Professional background of chief executive   
     - Entrepreneur -0.431** (0.196) NS  
     - Economics 1.195*** (0.248) NS  
     - Military NS  NS  
     - Politician 0.262* (0.141) NS  
     - Scientist, Other NS  NS  
     - Unknown/other 0.342** (0.149) NS  
Educational background of chief executive   
     - Economics NS  NS  
     - Natural science 0.948** (0.402) NS  
     - Other university NS  NS  
     - Unknown/other NS  NS  
Control variables   
Single-party government (0-1) NS  NS  
The years the chief executive has been in office 0.038*** (0.011) NS  
Parliamentary regimes (0-1) NS  NS  
EU member (0-1) -0.752*** (0.152) NS  
OECD member (0-1) 0.791*** (0.215) (omitted) 
Existence of electricity market reform idea (0-1) 0.465*** (0.117) (omitted) 
Log of population density 0.272* (0.141) NS  
Log of electricity consumption per capita (MWh) 1.696*** (0.249) 2.266*** (0.43) 
Log of GDP per capita (PPP, cur. thousand int. $) -3.628*** (0.218) -1.245*** (0.37) 
Average number of years of adult (25+) education -0.169*** (0.063) -1.613*** (0.292) 
Log of imports of goods and services (% of GDP) -0.954*** (0.212) NS  
Constant 13.918*** (0.94) NS  
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level.  
193 
 
5.7 Discussion of the results 
 
Our empirical analysis yields findings that are broadly consistent with our hypotheses. 
Accordingly, we fail to reject our hypotheses. We find that industry sector has a significant 
impact on the pace of power market liberalization process; and as its size gets larger, so does 
its influence (Hypothesis 1). Our results also reveal that foreign financial aid and/or assistance 
make liberalization in electricity industry more likely (Hypothesis 2). Moreover, our findings 
imply that government ideology is one of the determinants of the progress in electricity 
market reform process (Hypothesis 3). Finally, based on the results from our analysis, we may 
conclude that educational and professional background of the chief executive matters for 
electricity market liberalization (Hypothesis 4). Below we interpret the results in detail and 
comment on their policy implications. 
 
5.7.1 Impact of industry sector on the reforms 
 
When we look at the results presented in Table 5.4 and Table 5.5, we notice a significant 
negative relationship between electricity market liberalization and the size of industry sector, 
meaning that countries with larger industry sectors tend to liberalize less. As industry value 
added (as % of GDP) increases in a country, power market structure of that country becomes 
less liberal. For example, if industry value added of an OECD country increases from 40% to 
50% of GDP; entry barriers, public ownership and vertical integration scores (on a 0-6 scale) 
of that country increase by 1.29, 0.76 and 1.28 points, respectively. So, our study confirms the 
previous works on market liberalization in showing that interest groups have a significant 
influence on the reform progress and may slow it down. 
 
194 
 
To illustrate the impact of the industry sector on the reform progress, we can use data from 
our dataset. For instance, Turkey and Portugal are OECD countries and their overall 
electricity market closeness indexes (on a 0-6 scale) for 2007 are 2.1 and 1.0; and industry 
value added (as % of GDP) in these countries was 28.3 and 24.9 in the same year, 
respectively. Our results suggest that if industry value added (as % of GDP) increases by one 
unit, electricity market closeness index is expected to rise by 0.111 point (see Table 5.5). 
Therefore, holding all other variables constant and assuming that two countries are the same 
apart from their industry value added figures and electricity market closeness indexes, our 
results suggest that Portugal’s electricity market closeness index might be 1.38 
[1+0.111*(28.3-24.9)] if Portugal’s industry value added figure were to be equal to that of 
Turkey (i.e. 28.3). So, our findings imply that 0.38 of 1.1 points difference between the 
electricity market closeness indexes of two countries may be explained by the difference 
between their industry value added figures. 
 
Based on the results above, we clearly fail to reject Hypothesis 1. Our findings show that 
industry sector may have a significant impact on the pace of power market liberalization 
process; and, as its size gets larger, so does its influence. Since Hypothesis 1 is an immediate 
application of the economic theory of regulation, our results also imply that economic theory 
of regulation is a practical and valuable tool that can be used in the analysis of electricity 
market reform processes. According to our results, we can conclude that the relative strength 
of interest groups is important in determining the outcome of electricity market reform 
programs. 
 
Although the preferences of industrial consumers cannot directly be deducted from this 
research, our findings may be an indication that industrial consumers prefer guaranteed 
subsidized prices in a closed market to the possibility of future reduced prices in a liberal 
195 
 
market. Hence, taking into account this result, reform designers and implementers should do 
their best to reduce the opposition from industrial consumers by informing them of the 
benefits of the reform programs for them. 
 
5.7.2 Impact of foreign financial aid on the reforms 
 
The results from Table 5.6 and Table 5.7 clearly show that countries that receive foreign 
financial aid or assistance are more likely to liberalize their electricity markets and especially 
tend to reduce entry barriers to their power sector. So, our analysis reveals that countries that 
receive foreign financial assistance or aid tend to liberalize their electricity market more than 
a country that does not receive any assistance or aid. This finding holds true for both OECD 
and non-OECD countries. However, the tendency of liberalization in OECD countries is 
towards reducing entry barriers to their electricity markets. The results suggest that if an 
OECD country receives foreign financial assistance or aid, its entry barriers score (on a 0-6 
scale) reduces by 0.6 point. We could not detect any statistically significant impact of 
assistance or aid on public ownership or vertical integration.  
 
We can use data from our dataset once more to illustrate this result. Our results show that if 
foreign financial aid and/or assistance in a non-OECD country increase by $1 billion, overall 
electricity market closeness index (on a 0-6 scale) of that country decreases by 0.557 point 
(see Table 5.7). Azerbaijan and Armenia are non-OECD countries and, for 2009, their overall 
electricity market closeness indexes (on a 0-6 scale) are 3.8 and 2.2; and foreign financial aid 
and/or assistance to these countries in that year was $0.23 and $0.53 billion, respectively. 
Hence, holding all other variables constant and assuming that two countries are the same apart 
from the amount of foreign financial aid and/or assistance and electricity market closeness 
indexes, our results suggest that Azerbaijan’s electricity market closeness index might be 3.64 
196 
 
[3.8-0.557*(0.53-0.23)] if Azerbaijan’s foreign financial aid and/or assistance were to be 
equal to that of Armenia (i.e. 0.53). So, our findings imply that 0.16 of 1.6 points difference 
between the electricity market closeness indexes of two countries may be explained by the 
difference between the amount of foreign financial aid and/or assistance received by each 
country. 
 
Our results show that we fail to reject Hypothesis 2. That is, foreign financial aid and/or 
assistance seem to make liberalization in electricity industry more likely. Our findings 
indicate that the proactive involvement of the World Bank and other donor agencies helps 
diffusing the pro-market ideology of the Washington Consensus. 
 
Taking into account this conclusion, reform implementers should be very cautious while 
putting market reforms into practice and do their best to avoid creating a public perception 
that the reforms are imposed by foreign third parties, which may create very important 
problems related to internalization of the reforms by general public. 
 
Since one of the objectives in this paper is to capture the impact of foreign assistance and aid 
on the reform progress; we do not prefer to evaluate aid-receiving and non-receiving countries 
separately. Evaluating aid receiving and non-receiving countries together enables us to 
capture this impact from a richer dataset. However, we acknowledge that analysing two 
groups separately in future studies may provide interesting results. 
 
5.7.3 Impact of government ideology on the reforms 
 
As for economic policy orientation of ruling party, our results imply that right wing 
governments do not have a statistically significant overall effect on the reform process. 
197 
 
However, we see that they reduce vertical integration in OECD countries. On the other hand, 
left wing governments seem to speed up the reform process in OECD countries. Economic 
policy orientation of the ruling party does not affect the reform process in non-OECD 
countries. Left wing governments in OECD countries reduce entry barriers and vertical 
integration scores (on 0-6 scales) by 0.42 and 0.38 points, respectively. So, our results provide 
evidence that government ideology considerably influences electricity market reform 
progress, especially in OECD countries.  
 
The results above suggest that we fail to reject Hypothesis 3, meaning that government 
ideology may be one of the determinants of the progress in electricity market reform process 
and the ideological orientation of the political parties may be important in explaining the 
power market reform process. Our results confirm that the findings of the previous literature 
we mentioned in Section 5.3.3 hold also true for electricity markets; that is, we detect the 
suggested correlation between a reform process and government ideology for electricity 
markets too. 
 
As a result of our analysis, we did not identify any economic policy orientation that slows 
down the reform progress. That is, our findings imply that during the terms of political parties 
with some specific ideologies reforms may go further but none of them remarkably slows the 
reforms down. So, the most important policy implication of this result is that reform designers 
may wait for the suitable political environment to introduce a specific reform step. For 
instance, our results show that the best time to reduce entry barriers to the electricity market 
of an OECD country is during the term of a party with left ideological orientation. 
 
198 
 
5.7.4 Impact of the background of chief executive on the reforms 
 
Throughout the paper, we have also analysed whether the educational and professional 
background of the chief executive (prime minister or president) matters for the 
implementation of reforms, and found that it has a very significant impact on reform process 
in OECD countries. The background of the chief executive does not seem to have a 
statistically significant impact on reform progress in non-OECD countries.  
 
In OECD countries, leaders’ background in economics or natural sciences influences the 
reform process. We could not detect such an effect for other university degrees. The same 
influence holds true for leaders with a professional background as businessman, scientist 
(economist and others), or politician. Our results do not indicate significant results for 
military officers. 
 
We observe a negative relationship between an educational background in economics or 
natural sciences and the vertical integration score in OECD countries. This relationship is 
much stronger with an educational background in natural sciences. Our findings suggest that 
if the chief executive of a country has an educational background in economics or natural 
sciences, vertical integration score (on a 0-6 scale) of that country increases by 0.81 and 1.75 
points, respectively. As for entry barriers and public ownership, we could not detect a 
meaningful relationship for an educational background in economics but leaders with a 
background in natural sciences seem to increase public ownership by 1.1 points. 
 
As for professional backgrounds, our study finds that businessmen speed up the regulatory 
reform in OECD countries while scientists (economists) and politicians slow the liberalization 
process down. If the chief executive of a country has a professional background as 
199 
 
entrepreneur, then public ownership and vertical integration scores (on a 0-6 scale) of that 
country reduce by 0.45 and 0.59 points, respectively. On the other hand, if s/he has a 
professional background as scientist (economist), entry barriers and vertical integration scores 
increase by 1.33 and 1.98 points, correspondingly. 
 
In OECD countries, the chief executives with a professional background as politicians 
decrease public ownership but increase entry barriers and vertical integration. On the other 
hand, those with a background as scientists (other than economists) have a tendency to 
increase entry barriers but to reduce public ownership. 
 
The following example illustrates the impact of the background of the chief executive on 
reform progress. Our findings suggest that if the chief executive has a professional 
background as entrepreneur in an OECD country, electricity market closeness index decreases 
by 0.431 point in that country; and if s/he has a background as economist (scientist), the index 
increases by 1.195 points (see Table 5.9). Hungary and Canada are OECD countries and, in 
2007, the professional backgrounds of the chief executives were entrepreneur and economist 
in these countries, respectively. Electricity market closeness indexes were 1 for Hungary and 
3.3 for Canada for the same year. Holding all other variables constant and assuming that two 
countries are the same apart from the professional backgrounds of the chief executives and 
electricity market closeness indexes, our results suggest that Canada’s electricity market 
closeness index might be 1.67 [3.3-1.195-0.431] if the professional background of Canada’s 
chief executive were to be the same as that of Hungary (i.e. entrepreneur). So, our findings 
imply that 1.63 of 2.3 points difference between the electricity market closeness indexes of 
two countries may be explained by the difference between the professional backgrounds of 
the chief executives in two countries. 
 
200 
 
The results above show that we fail to reject Hypothesis 4. It seems that educational and 
professional background of the chief executive matters for electricity market liberalization; 
and the prior knowledge, education and experience of the chief executive regarding the power 
market liberalization process may encourage or discourage the reform measures. Since 
Hypothesis 4 is a direct parallel with the empirical findings reported in the literature 
connecting personal traits of policy makers and implementers with policy outcomes 
(summarized in Section 4.3.1), our results contribute to this literature as well.  
 
Taking into account these results, it is clear that the professional and educational backgrounds 
of the chief executives constitute another determinant of the reform progress, especially in 
OECD countries; and should be taken into account during reform design and implementation. 
Since it is not likely that a particular chief executive (prime minister or president) is elected 
by the public or the legislative body based mainly on this country’s policies regarding the 
extent of electricity market reform, reverse causation does not constitute a potential limitation 
in this study. However, we acknowledge that some variables and results regarding the 
background of the chief executive may apply to only one or two countries, whereas others 
may apply to many countries. 
 
5.7.5 Results from control variables 
 
The results from the control variables may also explain some variations in reform progress 
across countries. First of all, in Table 5.8 and Table 5.9, we see that government structure 
(coalition or single-party) has an impact on the reform process in OECD countries but does 
not seem to affect liberalization process in non-OECD countries. In OECD countries, single-
party governments accelerate the reform process by reducing public ownership and vertical 
integration. Moreover, we detect a negative relationship between the years the chief executive 
201 
 
has been in office and the reform progress in OECD countries. Again, the same relationship is 
not observed in non-OECD countries. Therefore, it may be the case that political stability is 
not an indispensable feature of a reform-enhancing political economic environment in OECD 
countries, and reforms may be put into practice successfully even in a politically unstable 
environment. Besides, electoral system (majoritarian or presidential) does not seem to 
influence liberalization process much while entry barriers seem to be lower in countries with 
parliamentary systems. Public ownership score (on a 0-6 scale) of a country with a 
parliamentary system tends to be 0.4 point less than that of one with a presidential system. 
 
The results from Table 5.4 and Table 5.5 indicate that urbanization and income equality have 
almost no significant impact on regulatory reform in electricity markets. The only statistically 
significant impact is that an increase in rural population in non-OECD countries (as % of total 
population) seems to speed up liberalization process in electricity industry; however this 
impact is quite limited. For instance, if rural population in a non-OECD country increases 
from 20% to 30% of total population, overall indicator (on a 0-6 scale) of that country 
decreases by 1.57 points.  
 
Although there seems to be no relation between public ownership and polity score, overall we 
detect a negative correlation between polity score and power sector liberalization in OECD 
countries, meaning that politically more liberal OECD countries prefer to liberalize their 
electricity markets less. Polity score does not seem to be an important factor explaining the 
reform process in non-OECD countries. For example, if polity score (on a -10 to +10 scale) of 
an OECD country increases from 3 to 8, entry barriers and vertical integration scores (on a 0-
6 scale) of that country increases by 0.89 and 0.55 points, respectively. 
 
202 
 
Out of 15 models we estimate, 9 models suggest that being an EU member country 
considerably contributes to efforts for electricity market liberalization. In most cases, this 
effect is large and statistically significant even at 1% level. The reverse holds true for being an 
OECD country. The results from 9 models imply that being an OECD country slows down 
electricity market liberalization process. The relative magnitude of these effects changes from 
one model to another. Therefore, being a member of both EU and OECD does not have a 
uniform effect on the reform process. This result indicates the importance of institutional 
policy for reform progress. EU not only encourages its member countries to introduce reforms 
but also requires them to carry out various reform programs in their electricity industries 
within an institutionalized framework. 
 
The existence of electricity market reform idea seems to limit the reform progress, which is 
consistent with the view that the early reformers had an advantage than the late comers in 
terms of reform implementation. This result may be explained by reform failures in some 
countries (e.g. California disaster). Therefore, policy makers and implementers in reforming 
countries should be able to explain what went wrong in some countries and the measures they 
took to avoid these problems in their markets. 
 
Population density and electricity consumption per capita seem to have a negative correlation 
with liberalization process in power industry, meaning that densely populated countries with 
higher per capita electricity consumption tend to liberalize their electricity markets less. So, 
this result implies that reforms may go further in countries with a higher rural population and 
higher income inequality as well. 
 
On the other hand, per capita income, education level and imports of goods and services (% of 
GDP) tend to have a positive correlation with liberalization process. Countries with higher per 
203 
 
capita income and education level that import a higher portion of goods and services from 
abroad seem to introduce more reform elements in their electricity markets. This result clearly 
shows that countries in the higher income group are more likely to progress more in the 
reform process. Therefore, countries in the low or middle income groups should be careful 
while importing the market structure of a developed country into their electricity industry as 
market structure of a developed country may not be suitable for a developing country. Finally, 
we see that country specific features tend to have a high power in explaining regulatory 
reform in electricity industries. 
 
5.7.6 Impact of initial conditions, initial price levels and California crisis on the reforms 
 
In Paper 2 (Section 4.7.6), we tested Paul L. Joskow’s argument that states only with high 
electricity prices have an incentive to introduce reforms in the US and those with relatively 
low prices do not have such an incentive and therefore do not introduce reforms (White, 
1996); and found that electricity prices (especially for industrial consumers) are an effective 
explanatory determinant of electricity reform internationally as it is in the US. Besides, in the 
same section, we also investigated whether the differences in initial conditions in terms of 
reform progress at the beginning of the reform process have an impact on the subsequent 
reform progress; but could not detect a relationship between initial conditions and the reform 
progress. In this section, using a different dataset, we would like to investigate again the 
impact of initial conditions and initial price levels on the reform progress. 
 
As we indicated in Section 2.2.1, the reform process in California resulted in several 
blackouts in 2000, which may have a profound impact on the decisions whether to initiate an 
electricity market reform process in other countries. Cross-section dataset in the second paper 
did not allow us to investigate the impact of so-called “California crisis” on the reform 
204 
 
progress in other countries; however, panel dataset in this paper lets us do so. So, in addition 
to initial conditions and initial price levels; in this paper, we also try to discover the impact of 
California crisis on the reform progress. 
 
To perform three investigations we mentioned above, we include four new variables into our 
analysis, namely “electricity market closeness index when reforms started”, “electricity prices 
for industry when reforms started”, “electricity prices for households when reforms started” 
and a dummy variable representing “California crisis” effect. The first variable represents the 
impact of starting point of the countries in terms of market closeness index on the reform 
progress while the second and third variables account for the impact of initial price levels. 
The last variable signifies the impact of “California crisis” on the reform process in other 
countries. 
 
The main analysis presented in this paper includes data from 55 countries. However, data on 
initial price levels are quite limited and available only for 28 of them. In order not to lose a 
significant number of observations, we prefer not to include these new variables directly into 
our main analysis. Instead, we prefer to develop two new but limited models to test the impact 
of initial conditions, initial price levels and California crisis on the reform processes in 
various countries. Besides, the data on our new variables are not compatible with fixed-effects 
estimation techniques that we have extensively used in the analysis of the main model. The 
variables representing “electricity market closeness index when reforms started”, “electricity 
prices for industry when reforms started” and “electricity prices for households when reforms 
started” have a separate value for each country but this value is naturally the same across all 
years. For instance, Finland and Hungary had an electricity market closeness index of 3.4 and 
6.0 when they initiated a reform program in their power markets, respectively. We have 33 
observations for each variable and for each country; and each observation represents a 
205 
 
specific year during 1975-2007 period. So, 33 observations representing “electricity market 
closeness index when reforms started” for France are the same and equal to 3.4. Similarly, 
those for Hungary are the same and equal to 6.0. In other words, the observations on these 
three variables vary across countries but not years. When we include these new variables into 
our analysis and try to perform a fixed-effects estimation, Stata drops all of them to avoid 
collinearity. So, in practice, it is not possible to include these new variables directly into our 
main analysis, which is mostly based on fixed-effects estimation methods. 
 
To overcome the problems mentioned above, we specify two new models including all 
explanatory and control variables we used in our main model and four new variables we 
mentioned above; and estimate them using random-effects estimation methods. Table 5.10 
and Table 5.11 present estimation results from two new regressions. Compared to the main 
one, these new models are limited in the sense that we have much less data to estimate 
equations based on them. Table 5.10 shows estimation results for the limited model testing 
Hypotheses 1 and 2 while Table 5.11 presents estimation results for the limited model testing 
Hypotheses 3 and 4. Since we have quite different number of observations in these models 
and the main model and use additional variables in these models, it is natural that the results 
may be different. The limited models in this section are formulated to give an idea about the 
impact of initial conditions, initial price levels and California crisis on the reform progress 
using the limited dataset we have.  
 
As can be seen in Table 5.10 and Table 5.11, the empirical analysis in this section yields 
results quite similar to those from our main models. Coefficients of the variables representing 
industry value added, net official assistance and aid received, economic policy orientation of 
ruling party and professional background of the chief executive are still significant and have  
 
206 
 
Table 5.10 Estimation results for the limited model testing Hypotheses 1 and 2 
 
Electricity market closeness index (0-6) 
(Dependent variable) 
Explanatory variables  
Industry value added (% of GDP) 0.09*** (0.015) 
Net official assistance and aid received -0.397* (0.233) 
Electricity market closeness index when reforms started (0-6) 0.478*** (0.117) 
Electricity prices for industry when reforms started (US$/MWh) -0.047*** (0.011) 
Electricity prices for households when reforms started (US$/MWh)  0.021*** (0.006) 
California crisis (dummy, =1 after 2000) -1.294*** (0.146) 
Control variables  
Rural population (% of total population) NS 
Gini coefficient (0-100) NS 
Polity score (-10,+10) 0.081*** (0.027) 
EU member (0-1) -1.167*** (0.184) 
OECD member (0-1) NS 
Existence of electricity market reform idea (0-1) NS 
Log of population density NS 
Log of electricity consumption per capita (MWh) NS 
Log of GDP per capita (PPP, cur. thousand int. $) -1.34*** (0.346) 
Average number of years of adult (25+) education -0.148* (0.083) 
Log of imports of goods and services (% of GDP) NS 
Constant 4.594*** (1.681) 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Number of countries: 28, number of observations: 456. 
 
207 
 
Table 5.11 Estimation results for the limited model testing Hypotheses 3 and 4 
 
Electricity market closeness index (0-6) 
(Dependent variable) 
Explanatory variables  
Economic policy orientation of ruling party  
     - Right -0.753*** (0.144) 
     - Left -0.719*** (0.146) 
     - Center (omitted) 
Professional background of chief executive  
     - Entrepreneur NS 
     - Economics 0.826*** (0.242) 
     - Military NS 
     - Politician 0.245** (0.114) 
     - Scientist, Other -0.414*** (0.128) 
     - Unknown/other NS 
Educational background of chief executive  
     - Economics NS 
     - Natural science NS 
     - Other university NS 
     - Unknown/other NS 
Electricity market closeness index when reforms started (0-6) 0.338*** (0.04) 
Electricity prices for industry when reforms started (US$/MWh) -0.019*** (0.003) 
Electricity prices for households when reforms started (US$/MWh)  0.009*** (0.002) 
California crisis (dummy, =1 after 2000) -1.749*** (0.128) 
Control variables  
Single-party government (0-1) -0.288*** (0.091) 
The years the chief executive has been in office 0.03** (0.012) 
Parliamentary regimes (0-1) -0.279* (0.162) 
EU member (0-1) -0.83*** (0.131) 
OECD member (0-1) 0.449** (0.193) 
Existence of electricity market reform idea (0-1) -0.254** (0.125) 
Log of population density NS 
Log of electricity consumption per capita (MWh) NS 
Log of GDP per capita (PPP, cur. thousand int. $) -1.344*** (0.25) 
Average number of years of adult (25+) education NS 
Log of imports of goods and services (% of GDP) NS 
Constant 7.446*** (0.764) 
Standard errors are shown in parentheses () with coefficients. 
“NS”: The coefficient is not significant even at 10% level. 
Coefficient that is significant at ***1% level, **5% level, *10% level. 
Number of countries: 27, number of observations: 674. 
 
208 
 
signs mostly similar to those of the main model. The findings from limited models still 
confirm (i) industry sector has a significant impact on the pace of power market liberalization 
process; and as its size gets larger, so does its influence (Hypothesis 1), (ii) foreign financial 
aid and/or assistance make liberalization in electricity industry more likely (Hypothesis 2), 
(iii) government ideology is one of the determinants of the progress in electricity market 
reform process (Hypothesis 3) and (iv) the background of the chief executive matters for 
electricity market liberalization (Hypothesis 4). The similarity of the results from limited 
model and the main model confirms the robustness of our results.  
 
In addition to verifying the robustness of our previous findings, the results from this section 
show that initial conditions, initial price levels and California crisis have a statistically 
significant impact on the reform progress. That is, we see that four new variables are 
significant. Our results indicate that market closeness index when reforms started and initial 
residential electricity prices are positively correlated with market closeness index (and 
therefore, negatively correlated with reform progress), meaning that reform progress may be 
limited in countries where the market has a closed structure at the beginning of the reforms 
and where residential electricity prices were relatively higher when reforms started. On the 
other hand, we detect a negative relationship between market closeness index and initial 
industrial electricity prices, implying that the reforms advance more rapidly in countries 
where industrial electricity prices were relatively higher at the beginning of the reform, 
confirming Joskow’s argument. Finally, we also observe a negative relationship between 
market closeness index and California crisis. That is, our results show that California crisis 
has not had a negative effect on the reform progress in other countries. Positive reform 
experiences in successfully reforming countries (e.g. UK and Norway) have probably 
dominated the negative effect of the California experience. 
 
209 
 
As a result of the analysis in this sub-section, we find out that initial market conditions, initial 
price levels and California crisis are effective explanatory determinants of electricity reform. 
Even after allowing for differences in these variables, the political economic factors are still 
significant determinants of the extent of reforms. Our findings imply that reform progress is 
positively affected by high initial electricity prices for industrial consumers, but negatively 
influenced by those for households. This outcome is partially in conformity with Joskow’s 
argument. That is, high initial electricity prices for industry seem to speed up reform process 
as suggested by Joskow; but high initial prices for households are likely to slow down the 
reform process. Besides, similar to the results from the second paper, when we take into 
account the fact that industrial consumers are much more organized (and therefore influential) 
than the residential ones, the results from our analysis seem to be in conformity with the 
predictions of economic theory of regulation. 
 
5.8 Conclusion 
 
This paper examined the political economy of liberalization in the electricity industries in the 
last decades. We empirically analysed the political economy of reform in the electricity 
industries of 55 countries during the period 1975–2010 with the aim of shedding light on the 
differing paces of reform in different countries. The use of a unique data set obtained by 
merging different data sources on political, government and reform structures as well as 
private interests and government ideologies allowed us to explore time-series and cross-
sectional variation in the political process of economic liberalization. Our findings are 
consistent with the rationale that the structure of political economic system has a strong effect 
on reform outcomes, and that the relative strength of economic and political variables matters 
for the implementation of the reforms.  
 
210 
 
Consistent with public choice theory and economic theory of regulation, our results suggest 
that a portion of the differences in the reform experiences of reforming countries in the past 
three decades can be explained by differences in the relative strength of interest groups. In the 
course of the study, we discover that industry sector has a significant impact on the pace of 
power market liberalization process; and as its size gets larger, so does its influence. Besides, 
our results imply that countries receiving foreign financial support are more likely to 
liberalize their electricity markets, which underlines the point that reforms may not be always 
voluntary. In addition, our findings suggest that government ideology is one of the 
determinants of the progress in electricity market reform process. We identify a decrease in 
vertical integration in electricity industry during the terms of parties with “right” or “left” 
ideologies in OECD countries. The ruling parties with “left” ideology seem also to reduce 
entry barriers in OECD countries. 
 
The study also questions whether politicians’ education and profession matter for the 
electricity market reforms. Overall, our results show that they do. According to our results, 
reforms are more likely to occur if the head of government has been an entrepreneur before 
entering into politics. Personal capabilities required to manage a company thus seem to be 
advantageous in promoting economic reform. Moreover, during the tenure of former 
professional economists, reforms are less likely. This result may also be plausible if we take 
into account the fact that many economists educated before the 1990s are taught that 
electricity industry is a natural monopoly and, therefore, an unbundled power sector may 
result in inefficiency in the provision of electricity service. We also provide evidence that the 
reforms seem to progress slower in OECD countries if the chief executive has an educational 
background in economics or natural science. Especially, those with a background in 
economics increase vertical integration while those with a background in natural science 
increase both vertical integration and public ownership in the sector. In summary, our analysis 
211 
 
confirms that the personal background of political leaders may be important. Clearly, other 
characteristics of politicians also matter for successful policy, and profession and education 
alone do not guarantee success. Besides, the focus of our analysis is restricted to economic 
reforms. Arguably, other policy dimensions are as equally important as the economic policy. 
Whether and to what extent those types of education and profession identified here as being 
supportive for market-oriented liberal reforms are also successful in other areas remain for 
future research. 
 
As a final point, by means of the estimation of a limited model, we also found that initial 
market conditions, initial price levels and California crisis are effective explanatory 
determinants of electricity market reform progress. 
 
The most important single policy implication that can be derived from these findings for the 
electricity industry and, to some extent, for other infrastructure industry reform is that future 
liberalization programs should give due attention to the political economic environment of the 
countries. 
212 
 
 
 
 
 
 
Chapter 6 
 
6 Conclusion 
 
Electricity is an indispensable good for households and a key input for industry in almost 
every economy. The three decades of electricity liberalization so far have taken place in line 
with a general trend towards liberalization of the economy in general and the energy industry 
in particular. In this process, the extent of reforms has been largely determined by country 
specific local conditions, quality of institutions required for the reform and political 
preferences related to the reform agenda. Today, the direct benefits of the reforms to 
households are still not directly visible in many reforming countries, which underlines the 
need for further analyses of the reforms. This thesis contributes to efforts to analyse electricity 
market reforms with an applied macro level cross-country approach.  
 
This concluding chapter has five sections. First of all, we discuss whether we have answered 
the research questions asked in the introductory chapter. Then, we mention main policy 
repercussions of the results from the three papers. The third section of this chapter aims at 
providing key policy suggestions either to a country that seriously considers implementing or 
redesigning a reform program in its electricity market or to a consultant that advises such a 
country. A discussion of the limitations of the research constitutes the fourth section. The 
final section in this chapter discusses what we have done and what still needs to be done. 
213 
 
 
6.1 Have research questions been answered? 
 
Let me start by discussing whether we have answered all research questions that we asked in 
Chapter 1. The first two questions were “what is the impact of electricity market reforms on 
electricity price-cost margins?” and “does liberalization cause a change in 
industrial/residential electricity price ratios?” The analysis in the first paper shows that there 
isn’t a uniform pattern for the impact of reform process as a whole on price-cost margins and 
industrial/residential price ratios. Each individual reform step has different impact on price-
cost margins and industrial/residential price ratios for each consumer and country group. That 
is, similar reform steps may have different impacts in different countries. The third question 
was “what are the other factors that influence electricity price-cost margins and 
industrial/residential electricity price ratios and how much are they influential relative to 
reform process?” In the course of the analysis in the first paper, we find that that power 
consumption, income level, electricity losses and country specific features constitute other 
important determinants of electricity price-cost margins and industrial/residential price ratios; 
and compared with reform process, they are more significant. 
 
The fourth and fifth research questions were “do differences in institutional structures of 
countries play an important role in explaining how far reforms have gone in these countries?” 
and “if they do, how do specific institutional endowments of a country affect its reform 
performance?” The results from the second paper show that institutional endowments of 
countries are important determinants of how far reforms have gone in different countries. Our 
findings suggest that the level of democracy and corruption in a country is significant in terms 
of reform progress. We see a positive relationship between reform progress and corruption 
perceptions index (which increases as corruption declines in a county), underlining the fact 
214 
 
that reforms go further in corruption-free countries. We also identify a negative relationship 
between reform progress and civil liberties, which may prove that reforms may be limited in 
democratic countries with strong civil society institutions such as trade unions or other 
organized structures in the society that may consider reforms as ‘harmful’ to their self-
interest. 
 
The sixth research question was “does the background of the chairperson of the regulatory 
agency when reforms started or were considered or that of the governor or minister 
responsible for energy policy at that time have an impact on reform progress?” Our empirical 
findings suggest that the educational background and education level of the chairperson and 
governor/minister are important for reform progress. We could not detect any statistically 
significant relationship between experience or length of term of the chairperson or 
governor/minister and the scope of reforms. 
 
The seventh research question was “what is the impact of industrial electricity consumers (as 
an interest group) on the reform progress in power sector?” The findings of the third paper 
imply that the structure of interest groups in a country has an impact on the reform progress. 
The findings suggest that there is an inverse relationship between the size of the industry 
sector and electricity market liberalization process. That is, as industry value added (as % of 
GDP) increases in a country, power market structure of that country becomes less liberal. The 
eighth question was put forward as “does foreign influence resulting from the dependence on 
foreign financial support have an influence on the electricity market liberalization process?” 
Our analysis reveals that a country that receives foreign financial assistance or aid tends to 
liberalize its electricity market more than a country that does not receive any assistance or aid. 
 
215 
 
The final research question was “do the ideology of ruling party and professional/educational 
background of the chief executive (prime minister or president) constitute important 
determinants of the reform progress? If yes, what is the direction of the influences originated 
from these variables?” In the course of our analysis, we found that right wing governments do 
not have a statistically significant overall effect on the reform process but they reduce vertical 
integration in OECD countries. On the other hand, left wing governments seem to speed up 
the reform process in OECD countries. Economic policy orientation of the ruling party does 
not affect the reform process in non-OECD countries. Besides, we also found that the 
educational and professional background of the chief executive (prime minister or president) 
has a very significant impact on reform process in OECD countries. The background of the 
chief executive does not seem to have a statistically significant impact on reform progress in 
non-OECD countries. 
 
6.2 Policy repercussions of the results 
 
In this section, we mention main policy repercussions of the results only while detailed results 
are available from the specific papers. Although each paper has its own specific results, the 
results from the thesis as a whole may be grouped into eight main policy implications. 
However, the validity of our results are limited by some factors specified in Section “6.4 
Limitations of the research”. Besides, our results are obtained by analysing a specific dataset 
covering a limited number of countries and a limited period of time. In the future, other 
scholars using different datasets may get different conclusions than ours. The policy 
recommendations below should be evaluated taking into account this fact. 
 
After briefly mentioning the reform progress in some of the countries we focus on in this 
thesis in Chapter 2, we explored the impact of power market reforms on electricity price-cost 
216 
 
margins and industrial/residential price ratios in the first paper. To observe the impact of 
electricity market reforms on electricity price-cost margins and industrial/residential price 
ratios, we used empirical econometric models with panel data from 63 countries covering the 
period from 1982 to 2009. The results from the first paper indicate that there is not a uniform 
pattern for the impact of reform process as a whole on price-cost margins and 
industrial/residential price ratios; and, each individual reform step has different impact on 
price-cost margins and industrial/residential price ratios for each consumer and country group. 
That is to say, the results suggest that similar reform steps may have different impacts in 
different countries. So, the first policy implication of the results from this thesis is that reform 
prescription for a specific country cannot easily be transferred to another one with similar 
success. Thus, while deciding whether to initiate a reform program or in the process of 
making decisions on the direction of an already initiated reform program, policy makers need 
to take into account the fact that each reform step has a specific impact in each country based 
on each country’s specific circumstances. Also, countries that try to adopt the power market 
structure of another country (for example, the adoption by developing countries of an 
electricity industry model similar to the UK) may have very different performance 
characteristics than the original country because their country-specific conditions will be 
different. In short, the first policy implication states that transferring the formal and economic 
structure of a successful power market in a developed country to a developing country is not a 
sufficient condition for good economic performance of the electricity industries in developing 
countries.  
 
The results from the first paper also underline the fact that power consumption, income level, 
electricity losses and other country specific features constitute important determinants of 
electricity price-cost margins and industrial/residential price ratios. So, the second policy 
implication of the results from this thesis is that electricity price-cost margins and 
217 
 
industrial/residential price ratios are partly a function of country specific features such as 
power consumption, income level and so on; therefore, policy makers need to pay due 
attention to specific characteristics of the country while formulating policies on electricity 
prices and industrial/residential price ratios. 
 
The second paper tried to explain whether differences in institutional structures of countries 
play an important role in explaining how far reforms have gone in these countries; how 
specific institutional endowments of a country affect its reform performance and, finally, 
whether the background of the chairperson of the regulatory agency when reforms started or 
were considered or that of the governor/minister responsible for energy policy at that time has 
an impact on reform progress. It focused on these issues by using an empirical econometric 
model with cross-section data from 51 states in US, 13 provinces in Canada and 51 other 
countries to observe the impact of institutional variables on the reform progress. Based on the 
results from the second paper, we have three more policy implications. The third policy 
implication of the results from this thesis is that the background of the chairperson and the 
minister/governor and institutional endowments of a country are important indicators of 
how far reforms will go in a country. Therefore, while designing electricity market reform 
programs, institutional structure of a country and qualifications of the policy 
makers/implementers need to be taken into account. 
 
The fourth policy implication, based on the results from the second paper, is that reforms go 
further in corruption-free countries. So, countries that want to implement more extensive 
electricity markets reforms need to do their best to reduce and, if possible, eliminate all 
structures, rules and procedures that may cause or tolerate corruption. The fifth policy 
implication suggests that there is a negative relationship between reform progress and civil 
liberties. This result points to the possibility that democracy may delay or hinder the reforms 
218 
 
by magnifying the voices of anti-reform interest groups. Hence, to lessen the negative impact 
of democracy on the reforms, a public awareness campaign may be launched concerning the 
objectives and benefits of the reforms. 
 
The third paper examined the political economy of liberalization in the electricity sector using 
data from 55 countries for the period 1975–2010 with the aim of shedding light on the 
differing pace of reforms in different countries. The main conclusion of the third paper is that 
the structure of political economic system is strongly correlated with reform outcomes; so, the 
relative strength of economic and political variables seems to matter for the implementation of 
the reforms. Based on this conclusion, the sixth policy implication of the results from this 
thesis is that a portion of the differences in reform experiences of countries can be 
explained by differences in the political structure, in the ideology of the government and in 
the professional and educational backgrounds of the political leaders. 
 
The seventh policy implication, provided by the third paper, states that countries with larger 
industry sectors tend to liberalize less. Although the preferences of industrial consumers 
cannot directly be deducted from our research, this conclusion may indicate a 
lack/shortcoming of trust by industrial consumers concerning the positive implications of the 
reforms for them. The fact that large industrial consumers tend to be subsidized in the pre-
reform period in many countries may also explain the negative attitude of industry sector 
towards the reforms. So, policy makers need to take additional measures to improve the 
confidence in benefits of the reforms among industrial consumers. The eighth policy 
implication stipulates that countries receiving foreign financial support are more likely to 
liberalize their electricity markets. Countries receiving foreign financial aid or assistance 
need to be careful in designing and implementing market reforms and do their best to avoid 
creating a public perception that the reforms are imposed by foreign third parties on the 
219 
 
country, which may create important internalization problems in the course of reform 
implementation. 
 
6.3 Policy suggestions 
 
As stated before, another aim of this chapter is to provide policy suggestions either to a 
country that seriously considers implementing or redesigning a reform program in its 
electricity market or to a consultant that advises such a country. The results from the thesis 
may be used as guidelines by policy makers in the process of formulating, implementing, 
evaluating and redesigning electricity market liberalization programs. The six main policy 
advises from the three papers are as follows: (1) do not copy or totally transfer the formal and 
economic structure of a successful power market of another country into your country without 
paying due attention to local conditions; (2) institutions matter, so take into account all 
institution-related variables while deciding on any aspect of the reforms; (3) people also 
matter; so the right people in the right positions are required for a successful reform progress; 
(4) political economic conditions and structure of interest groups have a substantial impact on 
the reform outcome, so try to implement reforms under appropriate political conditions with 
support from key interest groups; (5) democracy may delay or hinder the reforms by 
magnifying the voices of anti-reform interest groups, so launch a public awareness campaign 
about the objectives and benefits of the reforms; (6) countries receiving foreign financial 
support are more likely to liberalize their electricity markets, so be careful in designing and 
implementing market reforms and do your best to avoid creating a public perception that the 
reforms are imposed by foreign third parties. 
 
Policy suggestions above are derived from testing of hypotheses that we formulated. 
However, we used many control variables in our analyses but didn’t formulate them as 
220 
 
hypotheses. We acknowledge that the results from control variables are as plausible as those 
from testing of hypotheses, and most of the results from the control variables can be 
rationalized in terms of political economy or NIE literature. 
 
6.4 Limitations of the research 
 
The doctoral research presented in this thesis may have a number of limitations that we 
acknowledge. In fact, we have no reason to believe that any of these limitations should 
undermine our analyses, but cannot of course rule them out. The issue of endogeneity, the 
limited nature of our datasets and the qualitative nature of electricity market reform indicator 
data constitute three potential limitations of the analyses presented in this thesis. 
 
To begin with, like all other econometric studies on reform, the issue of endogeneity may be 
raised in our papers too. In the context of electricity reform, it is likely that just as non-reform 
variables (e.g. institutional variables, political economic structure and so on) have an impact 
on reform variables; reform process can also affect non-reform variables in a country. 
Besides, some variables in our models may be endogenously determined. In other words, 
some explanatory variables in our models may influence each other as well as the electricity 
market reform progress. Actually, instrumental variable (IV) methods allow consistent 
estimation when the explanatory variables (covariates) are correlated with the error terms of a 
regression relationship. Such correlation may occur when the dependent variable causes at 
least one of the covariates (“reverse” causation), when there are relevant explanatory variables 
which are omitted from the model, or when the covariates are subject to measurement error. 
There are two main requirements for using an IV: (1) the instrument must be correlated with 
the endogenous explanatory variables, conditional on the other covariates, (2) the instrument 
cannot be correlated with the error term in the explanatory equation, that is, the instrument 
221 
 
cannot suffer from the same problem as the original predicting variable. Credibility of the 
estimates from a regression using IV methods hinges on the selection of suitable instruments 
that do not violate these two conditions. If these conditions are violated then IV method may 
create more problems than it solves. Since we could not find suitable instrumental variables in 
our analyses, we prefer not to use IV methods in our papers. The analyses in the first and third 
papers deal to some extent with this potential problem by including country and year fixed 
effects. The country fixed effects control for country-specific propensities to reform and year 
fixed effects control for any general trend in the reform of electricity sector.  
 
The second shortcoming may originate from the limited nature of our data sets. In the first and 
third papers, our samples are composed of 63 and 55 countries, respectively, for which we 
could obtain data on all variables in our models. There will be sample selection bias if the 
countries making this data available have differing results for the dependent variables than 
those which do not make data available. Moreover, different countries may have different 
classifications and reporting conventions, so observations in a given data series may not have 
the same meaning across all countries. Taken together, any measurement error and omission 
of explanatory variables may bias estimates of all coefficients in the models. However, in the 
first and third papers, omitted variables may be captured at least in part by the country-
specific effects, mitigating the potential for bias. In the second paper, the sample is composed 
of 51 states in US, 13 provinces in Canada and 51 other countries. There are two main reasons 
for the limited nature of the dataset in the second paper. First, since our analysis in the second 
paper requires data on the chairperson of the regulatory agency when reforms started or were 
considered, we automatically exclude all countries without a regulatory agency for electricity 
industry. Here, it is important to note that we include all countries where a kind of regulation 
exists via an independent commission or a government department into our analyses; that is, 
we do not exclude countries where regulation is made via a government department. We only 
222 
 
exclude those countries where there is not any kind of regulation. We could detect the 
existence of such regulatory agencies in 135 countries. Second, out of these 135 countries, we 
could include only those for which we could obtain data on all variables in our model. If 
excluded countries have some characteristics that cause them to be less likely to be included 
than others, that is, if the sampling of the population is non-random; then there may be a 
sampling bias. Moreover, we considered including dummy variables for ex-Communist 
countries and former colonies. However, when we repeat our analyses including dummy 
variables for ex-Communist countries and former colonies, the results showed that the 
coefficients of these dummies are not significant even at 10% level. Besides, there is not a 
strong argument in the literature on the relationship between reform progress and being an ex-
Communist country or a former colony. Therefore, we prefer not to include additional 
dummies into our analyses. 
 
Figure 6.1 Reform score frequencies of reported and non-
reported countries in the second paper 
 
 
Figure 6.1 provides electricity market reform score frequencies of reported and non-reported 
countries in the second paper. As can be seen in Figure 6.1, more than half of the non-
223 
 
reported countries have a reform score of either 0 or 1. A de facto reform program refers to 
one that is put in practice but not necessarily ordained by law. That is, there may or may not 
be a specific law for the reform but a de facto reform program is implemented in practice and 
transforms the power market structure. On the other hand, a de jure reform program refers to 
one designed on paper and “in law” but usually not put into practice due to various problems 
related to reform implementation (Acemoglu et al., 2005). In our case, most of the non-
reported countries have highly inefficient and corrupted institutional structures that are 
capable of organizing neither a de facto reform program nor a de jure one. That is, they do not 
have necessary means and capabilities both to set up a proper legal basis for the reform and to 
put the reforms in practice. However, we acknowledge that our dataset is limited by data 
availability and some of the missing countries have high reform scores. We did our best to 
collect as much data as possible and include as many countries as possible; however, this is 
the best that we can do with available resources. For instance, we would like to include much 
more data into our dataset and investigate the determinants of reform progress in African 
countries; however, countries in Africa are the most problematic ones in terms of data 
collection. In most cases, ministries and regulators do not have proper web sites. When they 
do, their web sites are not available in English. We sent many e-mails to get data but very few 
are replied. In short, current data available on African countries do not let us carry out any 
kind of meaningful regression on reform progress in African countries. 
 
Due to lack of data, we could not properly account for the impact of some other variables on 
dependent variables in our analyses. For instance, in the first paper, we could not properly 
account for the impact of some variables on electricity price-cost margins like technological 
innovations, changes to regulatory practices, market power and so on. Besides, problems 
associated with price conversions using exchange rates tend to reduce the usefulness of cross-
country data in the first paper. In the first paper, we use electricity prices in national 
224 
 
currencies converted by IEA and OLADE into US$/kWh using the exchange rates to the U.S. 
dollar. As we know, if two countries have differing rates of inflation, then the relative prices 
of goods in the two countries, such as electricity, will change. The relative price of goods is 
linked to the exchange rate through the theory of Purchasing Power Parity (PPP), which states 
that the exchange rate between one currency and another is in equilibrium when their 
domestic purchasing powers at that rate of exchange are equivalent. Purchasing power parities 
take into account different rates of inflation among different economies and equalise the 
purchasing power of different currencies. In other words, they eliminate the differences in 
price levels between countries in the process of conversion. However, due to problematic 
nature of calculation process of PPPs, we do not use them in the first paper. Although our 
approach ignores the inflation in the US, it does so consistently and uniformly across 
countries. Therefore, it does not pose an important limitation to our analysis. 
 
Another limitation in our papers is the problem associated with measuring the true scale and 
scope of electricity market reforms. To lessen the impact of problems related to qualitative 
nature of electricity market reform indicator, we adopt a different approach in each paper. In 
the first paper, we use separate dummy variables for each reform step; in the second paper, we 
construct an aggregate reform score variable to reflect the extent of the reform process and, in 
the third paper, we do not construct a reform indicator ourselves and use various electricity 
market reform indicators constructed by OECD and EBRD. Nonetheless, we cannot claim 
that any of our reform indicators fully reflects all characteristics and intensity of the reforms 
in various countries. 
 
225 
 
6.5 What still needs to be done 
 
Our final purpose in this chapter is to discuss what we have done and what still needs to be 
done. In the first paper, we used panel data to explain the impact of reforms on electricity 
price-cost margins and industrial/residential price ratios. In the second paper, we employed 
cross-section data to observe the relationship between quality of institutions and the reform 
progress. In the third paper, we aimed at investigating the impact of political economic 
variables on the reform outcome. In short, this thesis tried to fill the gap in the applied 
literature concerning the analysis of electricity market liberalization programs. Our results do 
not allow a judgement on success or failure of the reform processes. Even with the results 
from this thesis, the present econometric evidence on the impact of the reforms is still limited. 
 
In many countries, power market reform is still an on-going process; so, there is a definite 
need and ample opportunities for continued analyses of the reforms in the electricity 
industries. The hope is that future research will continue developing econometric models to 
analyse electricity market reforms. We suggest the following for future research. First of all, 
due to lack of essential data, we focus on the reform progress rather than reform success or 
failure. However, there is a definite need for identifying the determinants of reform success or 
failure. So, future research on electricity market reforms should focus on identifying what 
successful reform is and developing new variables that measure the relative success of 
reforms. Second, in the first paper, we investigate the impact of reforms only on electricity 
price-cost margins and industrial/residential price ratios. These are just two dimensions of 
reform outcome and the reform process should be judged based on its overall impact (not only 
its impact on price-cost margins and industrial/residential price ratios). Therefore, there is 
clearly a need for further analysis regarding the impact of reforms on other dimensions such 
as efficiency in electricity industry, security of supply, private investments into power sector 
226 
 
(especially in developing countries), sustainable development and so on. Third, although 
some academic work on the social cost-benefit analysis of the reforms exist, they mostly use 
data from one country or few countries and only deal with a single reform step (usually 
privatization). However, what is needed is a comprehensive social-cost benefit analysis that 
takes into account as many countries as possible and all reform elements. Without such a 
study, no one can truly talk about reform success or failure. 
 
The fourth possible extension in future research may be the identification of suitable 
instrumental variables (IV) to overcome the possible endogeneity problem. In spite of our 
efforts, we could not find any suitable instrumental variable in our analyses and, therefore, did 
not use IV methods in our papers. As we know credibility of the estimates from a regression 
using IV methods hinges on the selection of suitable instruments. Utilization of IV methods 
with inappropriate instruments creates more problems than it solves. If suitable instrumental 
variables exist for the analysis of electricity market reforms and they are used in the future 
research, our understanding of the reforms may improve. The fifth task for future research 
should be the extension of the data set in terms of number of countries, time period, frequency 
of data and number (and quality) of variables. In the first, second and third papers, we 
employed data from 63, 53 and 55 countries, respectively. There may be sample selection bias 
if the countries making this data available have differing results for the dependent variables 
than those which do not make data available. Besides, due to lack of data, we could not 
properly account for the impact of some other variables on dependent variables in our 
analyses. For instance, we could not properly account for the impact of technological 
innovations, changes to regulatory practices, market power and so on in our analyses. Future 
research should focus on developing techniques to overcome these data-related problems. 
 
227 
 
The sixth extension may be realized by taking into account the fact that electricity reform is a 
part of wider economic reform (or liberalization) in general and energy industry reform in 
particular. In the period 1990-2011, total private investments in the infrastructure industries 
(energy, telecom, transport, water and sewerage) were about $1,695 billion. Out of this figure, 
$573 billion (33.8%) went to energy industry in general and $508 billion (30%) to electricity 
industry in particular. Power market reform affects and is affected by reforms in other energy 
and non-energy sectors. For instance, it is obvious that the progress in telecommunication 
reform has facilitated electricity reform, which in turn has contributed to the progress in gas 
market reform. Nepal and Jamasb (2012) investigate the link between power sector reforms 
and wider institutional reforms in the economy across different groups of transition countries. 
Their results indicate that power sector reform is highly inter-dependent with wider reforms in 
other sectors of the economy and failure to harmonize inter-sector reforms leads to power 
sector reform measures being ineffective. They argue that the success of power sector reforms 
in developing countries largely depend on the extent to which they synchronize inter-sector 
reforms in the economy. In this research, we did not take into account possible spill-over 
effects from or to other energy and non-energy sectors but inter-reform relationship is clearly 
an important research area that is open to exploration. 
 
Finally, in the second and third papers, we studied certain aspects of institutional quality and 
political economic variables. Of course, there remain many other characteristics of institutions 
and political economic structure that we did not investigate. They may constitute possible 
topics for future research if data on them become available. 
 
  
228 
 
 
 
 
 
 
Bibliography 
 
Abbott, M., 2006. The Performance of an Electricity Utility: The Case of the State Electricity 
Commission of Victoria, 1925–93. Australian Economic History Review 46, 23-44. 
Acemoglu, D., Bautista, M.A., Querubin, P., Robinson, J.A., 2008. Economic and Political 
Inequality in Development: The Case of Cundinamarca, Colombia, in: Helpman, E. 
(Ed.), Institutions and economic performance. Harvard University Press, Cambridge, 
Mass. 
Acemoglu, D., Johnson, S., Robinson, J.A., 2001. The Colonial Origins of Comparative 
Development: An Empirical Investigation. American Economic Review 91, 1369-
1401. 
Acemoglu, D., Johnson, S., Robinson, J.A., 2005. Chapter 6 Institutions as a Fundamental 
Cause of Long-Run Growth, in: Philippe, A., Steven, N.D. (Eds.), Handbook of 
Economic Growth. Elsevier, pp. 385-472. 
Aghion, P., Alesina, A., Trebbi, F., 2008. Democrary, Technology, and Growth, in: Helpman, 
E. (Ed.), Institutions and economic performance. Harvard University Press, 
Cambridge, Mass. 
Alesina, A., 1987. Macroeconomic policy in a two-party system as a repeated game. 
Quarterly Journal of Economics 102, 651-678. 
229 
 
Alesina, A., Ardagna, S., Trebbi, F., 2006. Who Adjusts and When? On the Political 
Economy of Reforms. National Bureau of Economic Research Working Paper Series 
No. 12049. 
Alesina, A., Özler, S., Roubini, N., Swagel, P., 1996. Political instability and economic 
growth. Journal of Economic Growth 1, 189-211. 
Alesina, A., Rodrik, D., 1994. Distributive Politics and Economic Growth. The Quarterly 
Journal of Economics 109, 465-490. 
Alt, J.E., Lowry, R.C., 1994. Divided Government, Fiscal Institutions, and Budget Deficits: 
Evidence from the States. The American Political Science Review 88, 811-828. 
Amundsen, E.S., Bergman, L., 2006. Why has the Nordic electricity market worked so well? 
Utilities Policy 14, 148-157. 
Anaya, K.L., 2010. The Restructuring and Privatisation of the Peruvian Electricity 
Distribution Market, EPRG Working Paper, No.1009. University of Cambridge, 
Electricity Policy Research Group, Cambridge, UK. 
Aron, J., 2000. Growth and Institutions: A Review of the Evidence. The World Bank 
Research Observer 15, 99-135. 
Asano, H., 2006. Regulatory reform of the electricity industry in Japan: What is the next step 
of deregulation? Energy Policy 34, 2491-2497. 
Assane, D., Grammy, A., 2003. Institutional framework and economic development: 
international evidence. Applied Economics 35, 1811-1817. 
Bagdadioglu, N., Waddams, C., Weyman-Jones, T., 2007. Measuring Potential Gains from 
Mergers among Electricity Distribution Companies in Turkey using a Non-
Parametric Model. Energy Journal 28, 83-110. 
Barro, R., Lee, J.-W., 2010. A New Data Set of Educational Attainment in the World, 1950-
2010, NBER Working Paper No. 15902. 
230 
 
Barro, R.J., 1991. Economic Growth in a Cross Section of Countries. The Quarterly Journal 
of Economics 106, 407-443. 
Barro, R.J., 1996. Democracy and growth. Journal of Economic Growth 1, 1-27. 
Barro, R.J., 2000. Inequality and Growth in a Panel of Countries. Journal of Economic 
Growth 5, 5-32. 
Becker, G.S., 1983. A Theory of Competition Among Pressure Groups for Political Influence. 
The Quarterly Journal of Economics 98, 371-400. 
Berg, S., Lin, C., Tsaplin, V., 2005. Regulation of State-Owned and Privatized Utilities: 
Ukraine Electricity Distribution Company Performance*. Journal of Regulatory 
Economics 28, 259-287. 
Besant-Jones, J.E., 2006. Reforming Power Markets in Developing Countries: What Have We 
Learned?, Energy and Mining Sector Board Discussion Paper, Paper No.19. The 
World Bank, Washington, DC. 
Besley, T., Kudamatsu, M., 2008. Making Autocracy Work, in: Helpman, E. (Ed.), 
Institutions and economic performance. Harvard University Press, Cambridge, Mass. 
Besley, T., Pande, R., Rao, V., 2005. Political selection and the quality of government: 
Evidence from South India. CEPR Discussion Paper 5201. 
Bhattacharyya, S.C., 2007a. Power sector reform in South Asia: Why slow and limited so far? 
Energy Policy 35, 317-332. 
Bhattacharyya, S.C., 2007b. Sustainability of power sector reform in India: what does recent 
experience suggest? Journal of Cleaner Production 15, 235-246. 
Bjørnskov, C., 2005. Political ideology and economic freedom, Working Paper 05–8. 
Department of Economics, Aarhus School of Business. 
Bjørnskov, C., Potrafke, N., 2011. Politics and privatization in Central and Eastern Europe: A 
panel data analysis. The Economics of Transition 19, 201-230. 
Bortolotti, B., Pinotti, P., 2008. Delayed privatization. Public Choice 136, 331-351. 
231 
 
Boschini, A.D., 2006. The political economy of industrialisation. European Journal of 
Political Economy 22, 887-907. 
Boycko, A.S., Vishny, R., 1996. A theory of privatization. Economic Journal 106, 309-319. 
Brousseau, E., Glachant, J.-M., 2008. New institutional economics: a guidebook. Cambridge 
University Press, Cambridge, UK; New York. 
Buchanan, J.M., 1972. Toward analysis of closed behavioral systems, in: Buchanan, J.M., 
Tollison, R.D. (Eds.), Theory of Public Choice. University of Michigan Press, Ann 
Arbor. 
Buchanan, J.M., Tullock, G., 1962. The Calculus of Consent: Logical Foundations of 
Constitutional Democracy. University of Michigan Press, Ann Arbor. 
Cameron, A.C., Trivedi, P.K., 1998. Regression analysis of count data. Cambridge University 
Press, Cambridge, UK; New York, USA. 
Cameron, A.C., Trivedi, P.K., 2009. Microeconometrics using Stata. Stata Press, College 
Station, Tex. 
Caselli, F., Esquivel, G., Lefort, F., 1996. Reopening the convergence debate: A new look at 
cross-country growth empirics. Journal of Economic Growth 1, 363-389. 
Chang, C.P., Berdiev, A.N., 2011. The political economy of energy regulation in OECD 
countries. Energy Economics 33, 816–825. 
Chavance, B., 2009. Institutional economics. Routledge, London; New York. 
Cherni, J.A., Kentish, J., 2007. Renewable energy policy and electricity market reforms in 
China. Energy Policy 35, 3616-3629. 
Cherni, J.A., Preston, F., 2007. Rural electrification under liberal reforms: the case of Peru. 
Journal of Cleaner Production 15, 143-152. 
Clague, C., Keefer, P., Knack, S., Olson, M., 1996. Property and contract rights in autocracies 
and democracies. Journal of Economic Growth 1, 243-276. 
232 
 
Clarke, G., Cull, R., 2002. Political and Economic Determinants of the Likelihood of 
Privatizing Argentine Public Banks. Journal of Law and Economics 45, 1515-1535. 
Coase, R.H., 1937. The Nature of the Firm. Economica 4, 386-405. 
Coase, R.H., 1960. The Problem of Social Cost. Journal of Law and Economics 3, 1-44. 
Conway, P., Nicolett, G., 2006. Product Market Regulation in non-manufacturing sectors in 
OECD countries: measurement and highlights, OECD Economics Department 
Working Paper No.530. 
Crampes, C., Fabra, N., 2005. The Spanish Electricity Industry: Plus ça change. Energy 
Journal 26, 127-153. 
CSP, 2010. Center for Systemic Peace, Polity IV Project: Political Regime Characteristics and 
Transitions, 1800-2009. 
Cubbin, J., Stern, J., 2006. The impact of regulatory governance and privatization on 
electricity industry generation capacity in developing economies. World Bank 
Economic Review 20, 115-141. 
De Alessi, L., 1980. The economics of property rights: A review of the evidence. Research in 
Law and Economics 2, 1-47. 
De Alessi, L., 1995. The public choice model of antitrust enforcement, in: McChesney, F.S., 
Shughart, W.F. (Eds.), The Causes and Consequences of Antitrust: The Public 
Choice Perspective. University of Chicago Press, Chicago. 
De Alessi, L., 2001. Property rights: private and political institutions, in: Shughart, W.F., 
Razzolini, L. (Eds.), The Elgar companion to public choice. Edward Elgar, 
Cheltenham. 
De Souza, F.C., Legey, L.F.L., 2010. Dynamics of risk management tools and auctions in the 
second phase of the Brazilian Electricity Market reform. Energy Policy 38, 1715-
1733. 
Defeuilley, C., 2009. Retail competition in electricity markets. Energy Policy 37, 377-386. 
233 
 
Demsetz, H., 1968. Why Regulate Utilities? Journal of Law and Economics 11, 55-65. 
Diaconu, O., Oprescu, G., Pittman, R., 2009. Electricity reform in Romania. Utilities Policy 
17, 114-124. 
Dreher, A., Lamla, M.J., Lein, S.M., Somogyi, F., 2009. The impact of political leaders' 
profession and education on reforms. Journal of Comparative Economics 37, 169-
193. 
Drury, A.C., Krieckhaus, J., Lusztig, M., 2006. Corruption, Democracy, and Economic 
Growth. International Political Science Review / Revue internationale de science 
politique 27, 121-136. 
Du, L., Mao, J., Shi, J., 2009. Assessing the impact of regulatory reforms on China's 
electricity generation industry. Energy Policy 37, 712-720. 
Duflo, E., 2005. Why political reservations? Journal of the European Economic Association 
3, 668-678. 
Duso, T., 2002. On the politics of regulatory reform: Econometric evidence from the OECD 
countries, Discussion Paper FS IV 02-07. Wissenschaftszentrum Berlin. 
Duso, T., 2005. Lobbying and regulation in a political economy: evidence from the US 
cellular industry. Public Choice 122, 251-276. 
Duso, T., Seldeslachts, J., 2010. The political economy of mobile telecommunications 
liberalization: Evidence from the OECD countries. Journal of Comparative 
Economics 38, 199-216. 
Dutra, J., Menezes, F., 2005. Lessons from the Electricity Auctions in Brazil. The Electricity 
Journal 18, 11-21. 
Duval, R., 2008. Is there a role for macroeconomic policy in fostering structural reforms? 
Panel evidence from OECD countries over the past two decades. European Journal of 
Political Economy 24, 491-502. 
234 
 
Easterly, W., Levine, R., 1997. Africa's Growth Tragedy: Policies and Ethnic Divisions. The 
Quarterly Journal of Economics 112, 1203-1250. 
EBRD, 2011. EBRD index of infrastructure reform: Electric power. 
Emerson, P., 2006. Corruption, competition and democracy. Journal of Development 
Economics 81, 193-212. 
Engoian, A., 2006. Industrial and institutional restructuring of the Russian electricity sector: 
Status and issues. Energy Policy 34, 3233-3244. 
Erdogdu, E., 2007. Regulatory reform in Turkish energy industry: An analysis. Energy Policy 
35, 984-993. 
Erdogdu, E., 2009. Some thoughts on the Turkish electricity distribution industry. Renewable 
and Sustainable Energy Reviews 13, 1485-1494. 
Erdogdu, E., 2010. A paper on the unsettled question of Turkish electricity market: Balancing 
and settlement system (Part I). Applied Energy 87, 251-258. 
Ernst & Young, 2006. Final Report Research Project: the Case for Liberalization. 
Ferrari, A., Giulietti, M., 2005. Competition in electricity markets: international experience 
and the case of Italy. Utilities Policy 13, 247-255. 
Fiorio, C.V., Florio, M., Doronzo, R., 2007. The Electricity Industry Reform Paradigm in the 
European Union: Testing the Impact on Consumers, Consumers and Utility Reforms 
in the European Union Conference, June 8-9, 2007, Milan. 
Florio, M., 2007. Electricity Prices as Signals for the Evaluation of Reforms: An Empirical 
Analysis of Four European Countries. International Review of Applied Economics 
21, 1-27. 
Fredriksson, P.G., Wollscheid, J.R., 2008. The political economy of investment: The case of 
pollution control technology. European Journal of Political Economy 24, 53-72. 
Freedom House, 2011. Map of Freedom in the World, 2008 Edition. Freedom House. 
235 
 
Gabriele, A., 2004. Policy alternatives in reforming energy utilities in developing countries. 
Energy Policy 32, 1319-1337. 
Gagliardi, F., 2008. Institutions and economic change: A critical survey of the new 
institutional approaches and empirical evidence. Journal of Socio-Economics 37, 
416-443. 
Gasmi, F., Noumba Um, P., Recuero Virto, L., 2009. Political Accountability and Regulatory 
Performance in Infrastructure Industries: An Empirical Analysis. The World Bank 
Economic Review 23, 509-531. 
Gasmi, F., Recuero Virto, L., 2010. The determinants and impact of telecommunications 
reforms in developing countries. Journal of Development Economics 93, 275-286. 
Ghertman, M., Ménard, C., 2009. Regulation, deregulation, reregulation: institutional 
perspectives. Edward Elgar, Cheltenham, UK; Northampton, MA. 
Goldberg, P.K., Pavcnik, N., 2005. Trade, wages, and the political economy of trade 
protection: evidence from the Colombian trade reforms. Journal of International 
Economics 66, 75-105. 
Göhlmann, S., Vaubel, R., 2007. The educational and professional background of central 
bankers and its effect on inflation-An empirical analysis. European Economic 
Review 51, 925-942. 
Gratwick, K.N., Eberhard, A., 2008. Demise of the standard model for power sector reform 
and the emergence of hybrid power markets. Energy Policy 36, 3948-3960. 
Green, R., Lorenzoni, A., Pérez, Y., Pollitt, M., 2009. Policy assessment and good practices, 
in: Glachant, J.-M., Lévêque, F. (Eds.), Electricity reform in Europe: towards a 
single energy market. Edward Elgar, Cheltenham, UK; Northampton, MA, pp. 172-
204. 
Haney, A.B., Pollitt, M., 2011. Exploring the determinants of “best practice” benchmarking in 
electricity network regulation. Energy Policy 39, 7739-7746. 
236 
 
Harriss, J., Hunter, J., Lewis, C.M., 1997. The new institutional economics and Third World 
development. Routledge, London; New York, pp. xiii, 360 p. 
Haselip, J., Potter, C., 2010. Post-neoliberal electricity market 're-reforms' in Argentina: 
Diverging from market prescriptions? Energy Policy 38, 1168-1176. 
Hattori, T., Tsutsui, M., 2004. Economic impact of regulatory reforms in the electricity supply 
industry: a panel data analysis for OECD countries. Energy Policy 32, 823-832. 
Helliwell, J.F., 1994. Empirical Linkages between Democracy and Economic Growth. British 
Journal of Political Science 24, 225-248. 
Heritage Foundation, 2011. The 2011 Index of Economic Freedom. Heritage Foundation. 
Hibbs, D., 1977. Political parties and macroeconomic policies. The American Political 
Science Review 71, 1467-1487. 
Hogan, W.W., 2002. Electricity Market Restructuring: Reforms of Reforms. Journal of 
Regulatory Economics 21, 103-132. 
Hrovatin, N., Pittman, R., Zoric, J., 2009. Organisation and reforms of the electricity sector in 
Slovenia. Utilities Policy 17, 134-143. 
Huang, Y., 2009. The political economy of financial reform: are Abiad and Mody right? 
Journal of Applied Econometrics 24, 1207-1213. 
Ickes, B.W., Ofer, G., 2006. The political economy of structural change in Russia. European 
Journal of Political Economy 22, 409-434. 
Ida, T., Ito, E., Kinoshita, S., 2007. Post-regulatory reform productivity gains in Japan's 
electricity industry. Applied Economics Letters 14, 975-979. 
IEA, 2010a. Energy Prices and Taxes (Edition: 2010, Quarter 1), Natural Gas Import Costs. 
International Energy Agency. 
IEA, 2010b. Energy Prices and Taxes (Edition: 2010, Quarter 1), OECD: Coal Import Costs 
and Indices by Importing Country. International Energy Agency. 
237 
 
IEA, 2010c. Energy Prices and Taxes, End-use Prices (Edition: 2009, Quarter 4). International 
Energy Agency. 
IEA, 2010d. World Energy Balances. International Energy Agency. 
IEA, 2010e. World Energy Statistics. International Energy Agency. 
Ikeme, J., Ebohon, O.J., 2005. Nigeria's electric power sector reform: what should form the 
key objectives? Energy Policy 33, 1213-1221. 
Iliadou, E.N., 2009. Electricity sector reform in Greece. Utilities Policy 17, 76-87. 
Isham, J., Kaufmann, D., Pritchett, L.H., 1997. Civil Liberties, Democracy, and the 
Performance of Government Projects. The World Bank Economic Review 11, 219-
242. 
Jamasb, T., Mota, R., Newbery, D., Pollitt, M., 2005. Electricity Sector Reform in Developing 
Countries: A Survey of Empirical Evidence on Determinants and Performance, 
World Bank Policy Research Working Paper 3549. World Bank. 
Jamasb, T., Newbery, D., Pollitt, M., 2006. Core indicators for determinants and performance 
of electricity sector in developing countries. International Journal of Regulation and 
Governance 6, 43-78. 
Jamasb, T., Pollitt, M., 2005. Electricity market reform in the European Union: review of 
progress toward liberalization and integration. The Energy Journal 26 (Special 
Issue), 11-41. 
Jamasb, T., Pollitt, M., 2007. Incentive regulation of electricity distribution networks: Lessons 
of experience from Britain. Energy Policy 35, 6163-6187. 
Jones, B.F., Olken, B.A., 2005. Do leaders matter? National leadership and growth since 
World War II. Quarterly Journal of Economics 120, 835-864. 
Jones, M.P., Sanguinetti, P., Tommasi, M., 2000. Politics, Institutions, and Fiscal 
Performance in a Federal System: An Analysis of the Argentine Provinces. Journal 
of Development Economics 61, 303-333. 
238 
 
Joseph, K.L., 2010. The politics of power: Electricity reform in India. Energy Policy 38, 503-
511. 
Joskow, P.L., 1998. Electricity Sectors in Transition. The Energy Journal 19, 25-52. 
Joskow, P.L., 2000. Chapter 4: Deregulation and Regulatory Reform in the U.S. Electric 
Power Sector, in: Peltzman, S., Winston, C. (Eds.), Deregulation of network 
industries: What's next? AEI-Brookings Joint Center for Regulatory Studies, 
Washington, D.C., pp. 113-188. 
Joskow, P.L., 2006. Markets for Power in the United States: An Interim Assessment. The 
Energy Journal 27, 1-36. 
Joskow, P.L., 2008. Lessons Learned From Electricity Market Liberalization. Energy Journal 
Special Issue, 9-42. 
Joskow, P.L., Noll, R.G., 1981. Regulation in Theory and Practice: An Overview, in: Fromm, 
G. (Ed.), Studies in Public Regulation. Cambridge: MIT Press, pp. 1-78. 
Keefer, P., 2010. Database of Political Institutions: Changes and Variable Definitions. The 
World Bank. 
Keefer, P., Knack, S., 1997. Why don't poor countries catch up? A cross-national test of an 
institutional explanation. Economic Inquiry 35, 590-602. 
Keefer, P., Knack, S., 2002. Polarization, politics and property rights: Links between 
inequality and growth. Public Choice 111, 127-154. 
Kim, B.-Y., Pirttilä, J., 2006. Political constraints and economic reform: Empirical evidence 
from the post-communist transition in the 1990s. Journal of Comparative Economics 
34, 446-466. 
Kirchner, C., 2007. Public Choice and New Institutional Economics, in: Baake, P., Borck, R. 
(Eds.), Public Economics and Public Choice. Springer Berlin Heidelberg, pp. 19-37. 
Knack, S., Keefer, P., 1997. Does social capital have an economic payoff? A cross-country 
investigation. Quarterly Journal of Economics 112, 1251. 
239 
 
Kotsogiannis, C., Schwager, R., 2006. On the incentives to experiment in federations. Journal 
of Urban Economics 60, 484-497. 
Kroszner, R.S., Strahan, P., 1999. What drives deregulation? Economics and politics of the 
relaxation of bank branching restrictions. Quarterly Journal of Economics 114, 1437-
1467. 
Kwoka, J., 2008. Restructuring the U.S. Electric Power Sector: A Review of Recent Studies. 
Review of Industrial Organization 32, 165-196. 
Lee, B.-H., Ahn, H.-H., 2006. Electricity industry restructuring revisited: the case of Korea. 
Energy Policy 34, 1115-1126. 
Levy, B., Spiller, P.T., 1996. Regulations, Institutions, and Commitment: Comparative 
Studies of Telecommunications, Political Economy of Institutions and Decisions 
Series. Cambridge/Univ. Press, Cambridge/New York/Melbourne. 
Li, W., Xu, L.C., 2002. The Political Economy of Privatization and Competition: Cross-
Country Evidence from the Telecommunications Sector. Journal of Comparative 
Economics 30, 439-462. 
Littlechild, S., 1992. Competition and regulation in the British electricity industry. Utilities 
Policy 2, 270-275. 
Littlechild, S., 2006. Competition and contracts in the Nordic residential electricity markets. 
Utilities Policy 14, 135-147. 
Littlechild, S.C., Ponzano, E.A., 2008. Transmission expansion in Argentina 5: The regional 
electricity forum of Buenos Aires province. Energy Economics 30, 1491-1526. 
Littlechild, S.C., Skerk, C.J., 2004. Regulation of transmission expansion in Argentina Part I: 
State ownership, reform and the Fourth Line., Cambridge Working Papers in 
Economics, CWPE 0464. Faculty of Economics, University of Cambridge. 
Littlechild, S.C., Skerk, C.J., 2008a. Transmission expansion in Argentina 1: The origins of 
policy. Energy Economics 30, 1367-1384. 
240 
 
Littlechild, S.C., Skerk, C.J., 2008b. Transmission expansion in Argentina 2: The Fourth Line 
revisited. Energy Economics 30, 1385-1419. 
Littlechild, S.C., Skerk, C.J., 2008c. Transmission expansion in Argentina 3: The evolution of 
policy. Energy Economics 30, 1420-1461. 
Littlechild, S.C., Skerk, C.J., 2008d. Transmission expansion in Argentina 4: A review of 
performance. Energy Economics 30, 1462-1490. 
Littlechild, S.C., Skerk, C.J., 2008e. Transmission expansion in Argentina 6: Distribution 
companies, regulation and the Public Contest method. Energy Economics 30, 1527-
1535. 
Lock, R., 2005. The New Electricity Model in Brazil: An Institutional Framework in 
Transition. The Electricity Journal 18, 52-61. 
Mauro, P., 1995. Corruption and Growth. The Quarterly Journal of Economics 110, 681-712. 
Mehta, A., Rao, H.S., Terway, A., 2007. Power sector reform in Central Asia: observations on 
the diverse experiences of some formerly Soviet Republics and Mongolia. Journal of 
Cleaner Production 15, 218-234. 
Ménard, C., Shirley, M.M., 2008. Handbook of new institutional economics. Springer, Berlin. 
Mota, R., 2003. The Restructuring and Privatisation of Electricity Distribution and Supply 
Businesses in Brazil: A Social Cost-Benefit Analysis, DAE Working Paper WP 
0309. University of Cambridge, Department of Applied Economics, Cambridge, UK. 
Mota, R., 2004. Essays on The Impact of Electricity Sector Reforms in Developing Countries, 
Faculty of Economics and Politics. University of Cambridge, Cambridge, UK. 
Mueller, D.C., 2003. Public Choice III. Cambridge University Press. 
Musole, M., 2009. Property rights, transaction costs and institutional change: Conceptual 
framework and literature review. Progress in Planning 71, 43-85. 
Nagayama, H., 2007. Effects of regulatory reforms in the electricity supply industry on 
electricity prices in developing countries. Energy Policy 35, 3440-3462. 
241 
 
Nagayama, H., 2009. Electric power sector reform liberalization models and electric power 
prices in developing countries: An empirical analysis using international panel data. 
Energy Economics 31, 463-472. 
Nagayama, H., Kashiwagi, T., 2007. Evaluating electricity sector reforms in Argentina: 
lessons for developing countries? Journal of Cleaner Production 15, 115-130. 
Nakano, M., Managi, S., 2008. Regulatory reforms and productivity: An empirical analysis of 
the Japanese electricity industry. Energy Policy 36, 201-209. 
Nepal, R., Jamasb, T., 2012. Reforming the power sector in transition: Do institutions matter? 
Energy Economics Article in press. 
Newbery, D., 2009. Electricity Investment in South Africa., EPRG Spring Seminar, 
15.05.2009. Electricity Policy Research Group, Cambridge. 
Newbery, D., Pollitt, M., 1997. The Restructuring and Privatisation of Britain's CEGB: Was It 
Worth It? Journal of Industrial Economics 45, 269-303. 
Nillesen, P.L., Pollitt, M., 2011. Ownership Unbundling in Electricity Distribution: Empirical 
Evidence from New Zealand. Review of Industrial Organization 38, 61-93. 
Niskanen, W.A.J., 1971. Bureaucracy and representative government. Aldine, Chicago. 
Noll, R.G., 1989. Chapter 22 Economic perspectives on the politics of regulation, in: Richard, 
S., Robert, W. (Eds.), Handbook of Industrial Organization. Elsevier, pp. 1253-1287. 
Noll, R.G., Owen, B.M., 1983. The political economy of deregulation: interest groups in the 
regulatory process. American Enterprise Institute for Public Policy Research, 
Washington, D.C. 
North, D.C., 1990. Institutions, institutional change, and economic performance. Cambridge 
University Press, Cambridge; New York. 
Nunn, N., 2008. Slavery, Inequality and Economic Development in the Americas: An 
Examination of the Engerman-Sokoloff Hypothesis, in: Helpman, E. (Ed.), 
Institutions and economic performance. Harvard University Press, Cambridge, Mass. 
242 
 
OLADE, 2010. Energy Statistics Report. OLADE (Latinamerican Energy Organization). 
Olper, A., 2007. Land inequality, government ideology and agricultural protection. Food 
Policy 32, 67-83. 
Olson, M., 1965. The Logic of Collective Action: Public Goods and the Theory of Groups. 
Harvard University Press, Cambridge. 
Pearce, D., 2006. The Political economy of an energy tax: The United Kingdom's climate 
change levy. Energy Economics 28, 149-158. 
Peltzman, S., 1976. Toward a More General Theory of Regulation. Journal of Law and 
Economics 19, 211-240. 
Peltzman, S., Levine, M.E., Noll, R.G., 1989. The Economic Theory of Regulation after a 
Decade of Deregulation. Brookings Papers on Economic Activity. Microeconomics 
1989, 1-59. 
Pérez-Reyes, R., Tovar, B., 2009. Measuring efficiency and productivity change (PTF) in the 
Peruvian electricity distribution companies after reforms. Energy Policy 37, 2249-
2261. 
Persson, T., Tabellini, G., 1994. Is Inequality Harmful for Growth? The American Economic 
Review 84, 600-621. 
Persson, T., Tabellini, G., 2008. The Growth Effect of Democracy, in: Helpman, E. (Ed.), 
Institutions and economic performance. Harvard University Press, Cambridge, Mass. 
Pigou, A.C., 1920. The economics of welfare. MacMillan, London. 
Pineau, P.-O., Hämäläinen, R.P., 2000. A perspective on the restructuring of the Finnish 
electricity market. Energy Policy 28, 181-192. 
Pitlik, H., 2007. A race to liberalization? Diffusion of economic policy reform among OECD-
economies. Public Choice 132, 159-178. 
Pollitt, M., 1997. The impact of liberalization on the performance of the electricity supply 
industry: an international survey. Journal of Energy Literature 3, 3-31. 
243 
 
Pollitt, M., 2004. Electricity reform in Chile: lessons for developing countries. Journal of 
Network Industries 5, 221-262. 
Pollitt, M., 2008a. Electricity reform in Argentina: Lessons for developing countries. Energy 
Economics 30, 1536-1567. 
Pollitt, M., 2008b. Liberalisation and regulation in electricity systems: how can we get the 
balance right?, in: Sioshansi, F.P. (Ed.), Competitive electricity markets: design, 
implementation and performance. Elsevier, Oxford. 
Pollitt, M., 2009. Evaluating the evidence on electricity reform: Lessons for the South East 
Europe (SEE) market. Utilities Policy 17, 13-23. 
Pollitt, M., 2012. The role of policy in energy transitions: Lessons from the energy 
liberalisation era. Energy Policy 50, 128-137. 
Posner, R.A., 1969. Natural Monopoly and its Regulation. Cato Institute, Washington, DC. 
Posner, R.A., 1972. The Appropriate Scope of Regulation in the Cable Television Industry. 
The Bell Journal of Economics and Management Science 3, 98-129. 
Potrafke, N., 2010. Does government ideology influence deregulation of product markets? 
Empirical evidence from OECD countries. Public Choice 143, 135-155. 
Rajan, R., 2004. Why are structural reforms so difficult? Finance and Development 41, 56-57. 
Romer, T., Rosenthal, H., 1987. Modern Political Economy and the Study of Regulation, in: 
Bailey, E. (Ed.), Public Regulation. MIT Press, Cambridge, MA, pp. 73-116. 
Roubini, N., Sachs, J.D., 1989. Political and economic determinants of budget deficits in the 
industrial democracies. European Economic Review 33, 903-938. 
Ruet, J., 2005. Privatising Power Cuts?: Ownership and Reform of State Electricity Boards in 
India. Academic Foundation, France. 
Schiantarelli, F., 2005. Product Market Regulation and Macroeconomic Performance: A 
Review of Cross-Country Evidence., World Bank Policy Research Working Paper 
3770. The World Bank, Washington, DC. 
244 
 
Scully, G.W., 1988. The Institutional Framework and Economic Development. The Journal of 
Political Economy 96, 652-662. 
Shughart, W.F., 2004. Regulation and Antitrust, in: Rowley, C.K., Schneider, F. (Eds.), The 
Encyclopedia of Public Choice, Volume I. Kluwer Academic Publishers. 
Singh, A., 2006. Power sector reform in India: current issues and prospects. Energy Policy 34, 
2480-2490. 
Sioshansi, F.P., 2008. Electricity market reform and ‘reform of the reforms’. Int. J. Global 
Energy Issues 29, 3-27. 
Spindler, Z.A., 1991. Liberty and Development: A Further Empirical Perspective. Public 
Choice 69, 197-210. 
Steiner, F., 2001. Regulation, industry structure and performance in the electricity supply 
industry, OECD Economics Studies. OECD. 
Stigler, G.J., 1968. The organization of industry. R. D. Irwin, Homewood, Ill.,. 
Stigler, G.J., 1971. The Theory of Economic Regulation. The Bell Journal of Economics and 
Management Science 2, 3-21. 
Stigler, G.J., 1986. The Theory of Economic Regulation, in: Leube, K.R., Moore, T.G. (Eds.), 
The Essence of Stigler. Hoover Institution Press, Stanford, CA: Stanford Univ., pp. 
243-264. 
Stigler, G.J., Friedland, C., 1962. What Can Regulators Regulate? The Case of Electricity. 
Journal of Law and Economics 5, 1-16. 
Sweeney, J.L., 2002. The California electricity crisis. Hoover Institution Press; Stanford 
Institute for Economic Policy Research, Stanford, Calif. 
Thomas, S., 2004. Evaluating the British Model of Electricity Deregulation. Annals of Public 
& Cooperative Economics 75, 367-398. 
Thomas, S., 2006a. Electricity industry reforms in smaller European countries and the Nordic 
experience. Energy 31, 788-801. 
245 
 
Thomas, S., 2006b. Recent evidence on the impact of electricity liberalisation on consumer 
prices. Public Services International Research Unit (PSIRU). 
Toba, N., 2007. Welfare impacts of electricity generation sector reform in the Philippines. 
Energy Policy 35, 6145-6162. 
Transparency International, 2011. Corruption Perceptions Index 2010. Transparency 
International. 
Trillas, F., 2010. Electricity and telecoms reforms in the EU: Insights from the economics of 
federalism. Utilities Policy 18, 66-76. 
UNU-WIDER, 2011. World Income Inequality Database. 
Upadhyaya, K.P., Raymond, J.E., Mixon Jr, F.G., 1997. The economic theory of regulation 
versus alternative theories for the electric utilities industry: A simultaneous probit 
model. Resource and Energy Economics 19, 191-202. 
US EIA, 2010a. Average Operating Heat Rate for Selected Energy Sources. U.S. Energy 
Information Administration. 
US EIA, 2010b. Electricity Restructuring by State (as of September 2010). US Energy 
Information Administration, http://www.eia.gov/cneaf/electricity/page/restructuring/ 
restructure_elect.html. 
US EIA, 2010c. International Energy Statistics. US Energy Information Administration. 
Vanssay, X.D., Spindler, Z.A., 1994. Freedom and Growth: Do Constitutions Matter? Public 
Choice 78, 359-372. 
Viscusi, W.K., Harrington, J.E., Vernon, J.M., 2005. Economics of Regulation and Antitrust. 
MIT Press, The. 
Volscho, T.W., 2007. Unions, government employment, and the political economy of income 
distribution in metropolitan areas. Research in Social Stratification and Mobility 25, 
1-12. 
246 
 
Wagner, A.F., Schneider, F., Halla, M., 2009. The quality of institutions and satisfaction with 
democracy in Western Europe -- A panel analysis. European Journal of Political 
Economy 25, 30-41. 
Wenders, J.T., 1988. The economic theory of regulation and the US telecoms industry. 
Telecommunications Policy 12, 16-26. 
White, M.W., 1996. Power Struggles: Explaining Deregulatory Reforms in Electricity 
Markets. Brookings Papers on Economic Activity. Microeconomics 1996, 201-267. 
Williams, J.H., Ghanadan, R., 2006. Electricity reform in developing and transition countries: 
A reappraisal. Energy 31, 815-844. 
Winkelmann, R., 2008. Econometric analysis of count data, 5th ed. Springer, Berlin. 
Woo, C.-K., 2001. What went wrong in California's electricity market? Energy 26, 747-758. 
Wooldridge, J.M., 2009. Introductory econometrics: a modern approach, 4th ed. South 
Western, Cengage Learning, Mason, OH. 
World Bank, 1995. Bureaucrats in Business: The Economics and Politics of Government 
Ownership, Washington, US. 
World Bank, 1997. World Development Report 1997: The State in a Changing World, 
Washington, US. 
World Bank, 2002. World Development Report 2002: Building Institutions for Markets, 
Washington, US. 
World Bank, 2004. Public and Private Sector Roles in the Supply of Electricity Services., 
Operational Guidance for World Bank Group Staff. The World Bank, Washington, 
DC. 
World Bank, 2005. The political economy of Gram Panchayats in South India: Results and 
policy conclusions from a research project, Washington DC. 
World Bank, 2010a. World Bank Country Classifications. 
World Bank, 2010b. World Development Indicators (Edition: April 2010). 
247 
 
World Bank, 2011. World Development Indicators (Edition: April 2011). World Bank. 
Yeoh, B.-S., Rajaraman, R., 2004. Electricity in China: The Latest Reforms. The Electricity 
Journal 17, 60-69. 
Yi-chong, X., 2006. The myth of the single solution: electricity reforms and the World Bank. 
Energy 31, 802-814. 
Zeckhauser, R.J., Horn, M., 1989. The control and performance of state-owned enterprises, in: 
MacAvoy, P.W., Stanbury, W.T., Yarrow, G., Zeckhauser, R.J. (Eds.), Privatization 
and stateowned enterprises: Lessons from the United States. Kluwer, Boston. 
Zhang, Y.-F., Parker, D., Kirkpatrick, C., 2008. Electricity sector reform in developing 
countries: an econometric assessment of the effects of privatization, competition and 
regulation. Journal of Regulatory Economics 33, 159-178. 
 
 
248 
 
Appendices 
 
Appendix I-1: Price-fuel cost margins when electricity reforms started 
No Countries 
The year 
electricity 
reform started 
Electricity prices for 
industry when 
electricity reform 
started (US$/MWh) 
Electricity prices for 
households when 
electricity reform 
started (US$/MWh) 
Fuel cost when 
electricity reform 
started (US$/MWh) 
Price-fuel cost margin 
for industry when 
electricity reform started 
(US$/MWh) 
Price-fuel cost margin for 
households when 
electricity reform started 
(US$/MWh) 
1 Argentina 1992 113.90 83.70 21.04 92.86 62.66 
2 Australia 1992 45.81 73.92 17.81 28.00 56.11 
3 Austria 2000 38.25 117.51 5.89 32.36 111.63 
4 Belgium 1999 55.49 152.08 15.76 39.72 136.31 
5 Bolivia 1994 71.80 72.20 19.05 52.75 53.15 
6 Brazil 1997 50.90 129.60 16.56 34.34 113.04 
7 Canada 1989 34.12 49.74 17.61 16.51 32.13 
8 Chile 1982 41.10 79.50 19.16 21.94 60.34 
9 Colombia 1994 74.10 40.20 17.29 56.81 22.91 
10 Czech Republic 2001 42.61 59.71 16.17 26.44 43.54 
11 Denmark 2000 57.69 197.44 15.15 42.54 182.29 
12 Dominican Republic 1998 118.00 95.90 14.19 103.81 81.71 
13 Finland 1995 60.22 108.86 17.87 42.35 90.99 
14 France 2000 35.76 101.66 15.81 19.95 85.85 
15 Germany 2005 83.98 212.42 29.94 54.04 182.48 
16 Greece 1999 50.28 90.08 12.68 37.60 77.40 
17 Hungary 1994 46.23 39.86 16.12 30.11 23.74 
18 India 1998 82.04 29.01 14.65 67.39 14.36 
249 
 
19 Indonesia 2000 35.92 24.62 16.53 19.39 8.09 
20 Ireland 1999 56.55 117.15 15.23 41.32 101.92 
21 Israel 1998 63.53 90.88 14.19 49.34 76.69 
22 Italy 1996 101.38 177.67 20.07 81.31 157.60 
23 Japan 1996 157.23 230.11 24.00 133.24 206.11 
24 Korea 2001 47.71 70.96 19.47 28.23 51.49 
25 Luxembourg 2000 
 
99.26 22.28 
 
76.98 
26 Mexico 1995 27.00 44.79 19.60 7.41 25.19 
27 Netherlands 1998 62.38 127.86 18.20 44.18 109.66 
28 New Zealand 2003 45.84 94.81 26.08 19.76 68.73 
29 Norway 1991 35.47 72.95 18.25 17.23 54.70 
30 Peru 1994 64.10 106.50 19.05 45.05 87.45 
31 Poland 1997 36.19 62.41 15.68 20.52 46.73 
32 Portugal 1996 108.32 175.68 16.42 91.90 159.26 
33 Romania 1998 44.66 31.71 16.07 28.59 15.64 
34 Slovak Republic 2001 43.10 62.83 19.28 23.82 43.55 
35 South Africa 2004 21.63 59.91 19.92 1.70 39.99 
36 Spain 1997 60.57 163.41 15.81 44.76 147.60 
37 Sweden 1996 44.88 109.59 14.49 30.39 95.09 
38 Switzerland 2001 67.52 108.95 28.12 39.40 80.83 
39 Taiwan (Chinese Taipei) 1999 58.47 78.54 12.81 45.65 65.73 
40 Trinidad And Tobago 1999 23.14 27.43 15.04 8.10 12.39 
41 Turkey 2001 79.31 83.61 22.59 56.72 61.02 
42 United Kingdom 1989 61.01 100.92 23.10 37.91 77.82 
43 United States 1996 46.00 83.90 15.78 30.22 68.12 
44 Venezuela 1999 30.62 47.89 15.04 15.58 32.85 
250 
 
Appendix I-2: Change in price-fuel cost margins for industry 
No Countries 
Electricity prices for industry  
when electricity reform 
started (US$/MWh) 
Change in price-fuel cost 
margin for industry since 
reform started, as of 2007 
(US$/MWh) 
1 Mexico 27.00 43.84 
2 Venezuela 30.62 -44.56 
3 Canada 34.12 12.84 
4 Norway 35.47 -21.61 
5 France 35.76 30.17 
6 Indonesia 35.92 14.36 
7 Poland 36.19 31.79 
8 Austria 38.25 53.11 
9 Chile 41.10 58.31 
10 Czech Republic 42.61 57.80 
11 Slovak Republic 43.10 77.10 
12 Sweden 44.88 2.43 
13 New Zealand 45.84 -2.00 
14 United States 46.00 6.62 
15 Hungary 46.23 56.22 
16 Korea 47.71 -2.70 
17 Brazil 50.90 64.40 
18 Ireland 56.55 53.89 
19 Denmark 57.69 23.54 
20 Taiwan (Chinese Taipei) 58.47 -20.69 
21 Finland 60.22 -0.06 
22 Spain 60.57 1.34 
23 United Kingdom 61.01 53.94 
24 Netherlands 62.38 68.47 
25 Israel 63.53 -0.82 
26 Peru 64.10 -42.89 
27 Switzerland 67.52 -12.74 
28 Colombia 74.10 -11.88 
29 Turkey 79.31 4.88 
30 Germany 83.98 18.83 
31 Italy 101.38 104.49 
32 Portugal 108.32 -6.83 
33 Argentina 113.90 -100.14 
34 Dominican Republic 118.00 88.78 
35 Japan 157.23 -60.60 
  
251 
 
Appendix I-3: Change in price-fuel cost margins for households 
No Countries 
Electricity prices for 
households when electricity 
reform started (US$/MWh) 
Change in price-fuel cost 
margin for households since 
reform started, as of 2007 
(US$/MWh) 
1 Indonesia 24.62 20.24 
2 Hungary 39.86 116.60 
3 Colombia 40.20 82.29 
4 Mexico 44.79 16.99 
5 Venezuela 47.89 -54.07 
6 Canada 49.74 22.45 
7 Czech Republic 59.71 71.26 
8 Poland 62.41 74.04 
9 Slovak Republic 62.83 108.16 
10 Korea 70.96 6.55 
11 Norway 72.95 24.47 
12 Taiwan (Chinese Taipei) 78.54 -19.44 
13 Chile 79.50 71.61 
14 Turkey 83.61 13.59 
15 Argentina 83.70 -94.17 
16 United States 83.90 11.29 
17 Israel 90.88 13.79 
18 New Zealand 94.81 42.05 
19 Dominican Republic 95.90 36.60 
20 Luxembourg 99.26 96.69 
21 United Kingdom 100.92 102.89 
22 France 101.66 28.52 
23 Peru 106.50 -34.37 
24 Finland 108.86 15.27 
25 Switzerland 108.95 -1.50 
26 Sweden 109.59 57.72 
27 Ireland 117.15 87.72 
28 Austria 117.51 53.30 
29 Netherlands 127.86 166.03 
30 Brazil 129.60 52.37 
31 Spain 163.41 -4.12 
32 Portugal 175.68 10.74 
33 Italy 177.67 48.75 
34 Denmark 197.44 127.06 
35 Germany 212.42 44.63 
36 Japan 230.11 -72.94 
 
252 
 
Appendix I-4: Estimation results for Paper One 
 
Models 
Dependent variable 
Explanatory variables Coef. Std. Err. t-stat. p value 
Number of  Number of  Hausman Test BPLM Test Preferred 
(country group) countries observations Statistic p-value Statistic p-value Specification 
Model 1.1 Log of price-fuel cost  Log of electricity consumption by industry -0.947 0.076 -12.420 0.00 31 659 77.60 0.0000 - - Fixed Effects 
 
margin for industry Electricity loses in total supply (%, 0-100) -0.094 0.013 -7.410 0.00 
       
 
(Developed countries) Log of GDP per capita 0.617 0.046 13.300 0.00 
       
  
Existence of IPPs -0.116 0.045 -2.590 0.01 
       
  
Wholesale Electricity Market -0.059 0.058 -1.020 0.31 
       
  
Choice of Supplier 0.056 0.066 0.840 0.40 
       
  
Unbundling -0.021 0.077 -0.270 0.79 
       
  
Privatization 0.188 0.074 2.530 0.01 
       
  
Existence of Market Regulator -0.238 0.100 -2.380 0.02 
       
  
Privatization and Regulator 0.193 0.103 1.860 0.06 
       
  
Privatization and Unbundling -0.280 0.105 -2.660 0.01 
       
  
Unbundling and Regulator 0.065 0.106 0.610 0.54 
           Constant 12.592 0.767 16.410 0.00               
Model 1.2 Log of price-fuel cost  Log of electricity consumption by industry -0.293 0.219 -1.340 0.18 13 241 65.84 0.0000 - - Fixed Effects 
 
margin for industry Electricity loses in total supply (%, 0-100) 0.009 0.017 0.550 0.58 
       
 
(Developing countries in  Log of GDP per capita 0.865 0.177 4.880 0.00 
       
 
America) Existence of IPPs -0.641 0.180 -3.550 0.00 
       
  
Wholesale Electricity Market 0.746 0.293 2.540 0.01 
       
  
Choice of Supplier (omitted)       
       
  
Unbundling -0.431 0.445 -0.970 0.33 
       
  
Privatization 0.575 0.681 0.840 0.40 
       
  
Existence of Market Regulator 0.309 0.288 1.070 0.28 
       
  
Privatization and Regulator -0.762 0.609 -1.250 0.21 
       
  
Privatization and Unbundling -0.250 0.943 -0.270 0.79 
       
  
Unbundling and Regulator (omitted)       
           Constant 5.470 2.052 2.670 0.01               
253 
 
Models 
Dependent variable 
Explanatory variables Coef. Std. Err. t-stat. p value 
Number of  Number of  Hausman Test BPLM Test Preferred 
(country group) countries observations Statistic p-value Statistic p-value Specification 
Model 1.3 Log of price-fuel cost  Log of electricity consumption by industry -1.295 0.179 -7.250 0.00 10 149 177.34 0.0000 - - Fixed Effects 
 
margin for industry Electricity loses in total supply (%, 0-100) 0.049 0.022 2.240 0.03 
       
 
(Other developing countries) Log of GDP per capita 1.263 0.164 7.710 0.00 
       
  
Existence of IPPs 0.124 0.142 0.870 0.38 
       
  
Wholesale Electricity Market -0.361 0.200 -1.800 0.07 
       
  
Choice of Supplier 0.286 0.221 1.290 0.20 
       
  
Unbundling 0.533 0.190 2.810 0.01 
       
  
Privatization 0.175 0.244 0.720 0.48 
       
  
Existence of Market Regulator 0.081 0.185 0.440 0.66 
       
  
Privatization and Regulator -0.019 0.526 -0.040 0.97 
       
  
Privatization and Unbundling 0.228 0.549 0.420 0.68 
       
  
Unbundling and Regulator -0.913 0.284 -3.210 0.00 
           Constant 15.555 1.861 8.360 0.00               
Model 2.1 Log of price-fuel cost  Log of electricity consumption by households -1.217 0.069 -17.550 0.00 31 714 247.20 0.0000 - - Fixed Effects 
 
margin for households Electricity loses in total supply (%, 0-100) -0.039 0.009 -4.270 0.00 
       
 
(Developed countries) Log of GDP per capita 0.925 0.041 22.440 0.00 
       
  
Existence of IPPs -0.001 0.034 -0.040 0.97 
       
  
Wholesale Electricity Market 0.020 0.043 0.460 0.64 
       
  
Choice of Supplier -0.035 0.049 -0.720 0.47 
       
  
Unbundling 0.053 0.060 0.880 0.38 
       
  
Privatization 0.172 0.057 3.010 0.00 
       
  
Existence of Market Regulator 0.016 0.079 0.210 0.84 
       
  
Privatization and Regulator 0.020 0.081 0.250 0.80 
       
  
Privatization and Unbundling -0.095 0.081 -1.170 0.24 
       
  
Unbundling and Regulator -0.018 0.083 -0.220 0.83 
           Constant 14.088 0.625 22.520 0.00               
Model 2.2 Log of price-fuel cost  Log of electricity consumption by households -0.649 0.176 -3.680 0.00 13 244 85.26 0.0000 - - Fixed Effects 
 
margin for households Electricity loses in total supply (%, 0-100) 0.030 0.012 2.540 0.01 
       
 
(Developing countries in  Log of GDP per capita 1.361 0.145 9.390 0.00 
       
254 
 
Models 
Dependent variable 
Explanatory variables Coef. Std. Err. t-stat. p value 
Number of  Number of  Hausman Test BPLM Test Preferred 
(country group) countries observations Statistic p-value Statistic p-value Specification 
 
America) Existence of IPPs -0.104 0.154 -0.680 0.50 
       
  
Wholesale Electricity Market 0.093 0.228 0.410 0.68 
       
  
Choice of Supplier (omitted)       
       
  
Unbundling 0.695 0.370 1.880 0.06 
       
  
Privatization 1.030 0.574 1.790 0.07 
       
  
Existence of Market Regulator -0.579 0.235 -2.460 0.02 
       
  
Privatization and Regulator 0.767 0.504 1.520 0.13 
       
  
Privatization and Unbundling -1.512 0.793 -1.910 0.06 
       
  
Unbundling and Regulator (omitted)       
           Constant 7.608 1.469 5.180 0.00               
Model 2.3 Log of price-fuel cost  Log of electricity consumption by households -0.806 0.144 -5.600 0.00 9 157 113.35 0.0000 - - Fixed Effects 
 
margin for households Electricity loses in total supply (%, 0-100) 0.126 0.018 6.820 0.00 
       
 
(Other developing countries) Log of GDP per capita 0.926 0.146 6.330 0.00 
       
  
Existence of IPPs 0.134 0.125 1.070 0.29 
       
  
Wholesale Electricity Market -0.474 0.152 -3.110 0.00 
       
  
Choice of Supplier 0.951 0.174 5.480 0.00 
       
  
Unbundling 0.269 0.207 1.300 0.20 
       
  
Privatization 0.121 0.216 0.560 0.58 
       
  
Existence of Market Regulator -0.249 0.172 -1.440 0.15 
       
  
Privatization and Regulator 0.109 0.470 0.230 0.82 
       
  
Privatization and Unbundling 0.037 0.488 0.080 0.94 
       
  
Unbundling and Regulator -0.119 0.281 -0.420 0.67 
           Constant 9.035 1.279 7.060 0.00               
Model 3.1 Industrial/residential Log of electricity consumption by industry 0.201 0.047 4.300 0.00 31 682 43.49 0.0000 - - Fixed Effects 
  price ratio Log of electricity consumption by households 0.040 0.056 0.710 0.48        
 (Developed countries) Electricity loses in total supply (%, 0-100) 0.004 0.006 0.650 0.52        
  Log of GDP per capita -0.107 0.026 -4.090 0.00        
  Existence of IPPs -0.022 0.021 -1.040 0.30        
  Wholesale Electricity Market -0.011 0.027 -0.420 0.68        
255 
 
Models 
Dependent variable 
Explanatory variables Coef. Std. Err. t-stat. p value 
Number of  Number of  Hausman Test BPLM Test Preferred 
(country group) countries observations Statistic p-value Statistic p-value Specification 
  Choice of Supplier 0.099 0.031 3.200 0.00        
  Unbundling -0.100 0.036 -2.800 0.01        
  Privatization -0.108 0.035 -3.090 0.00        
  Existence of Market Regulator -0.099 0.047 -2.120 0.03        
  Privatization and Regulator 0.035 0.048 0.720 0.47        
  Privatization and Unbundling 0.015 0.049 0.300 0.77        
  Unbundling and Regulator 0.026 0.049 0.540 0.59        
  Constant -1.525 0.402 -3.800 0.00        
Model 3.2 Industrial/residential Log of electricity consumption by industry 0.137 0.067 2.030 0.04 21 517 68.51 0.0000 146.41 0.0000 Fixed Effects 
  price ratio Log of electricity consumption by households 0.143 0.075 1.910 0.06        
 (Developing countries in  Electricity loses in total supply (%, 0-100) -0.001 0.004 -0.260 0.80        
 America) Log of GDP per capita -0.277 0.066 -4.210 0.00        
  Existence of IPPs 0.126 0.061 2.070 0.04        
  Wholesale Electricity Market 0.008 0.117 0.070 0.94        
  Choice of Supplier 0.245 0.143 1.710 0.09        
  Unbundling -0.059 0.390 -0.150 0.88        
  Privatization -0.095 0.223 -0.430 0.67        
  Existence of Market Regulator -0.088 0.076 -1.170 0.24        
  Privatization and Regulator 0.051 0.262 0.190 0.85        
  Privatization and Unbundling -0.058 0.190 -0.310 0.76        
  Unbundling and Regulator -0.298 0.387 -0.770 0.44        
  Constant -0.911 0.534 -1.710 0.09        
Model 3.3 Industrial/residential Log of electricity consumption by industry 0.116 0.192 0.600 0.55 9 157 -1.07 0.0000 - - Fixed Effects 
  price ratio Log of electricity consumption by households 0.273 0.175 1.560 0.12        
 (Other developing countries) Electricity loses in total supply (%, 0-100) -0.050 0.014 -3.560 0.00        
  Log of GDP per capita -0.394 0.088 -4.490 0.00        
  Existence of IPPs -0.052 0.086 -0.600 0.55        
  Wholesale Electricity Market 0.089 0.107 0.830 0.41        
  Choice of Supplier -0.367 0.125 -2.940 0.00        
256 
 
Models 
Dependent variable 
Explanatory variables Coef. Std. Err. t-stat. p value 
Number of  Number of  Hausman Test BPLM Test Preferred 
(country group) countries observations Statistic p-value Statistic p-value Specification 
  Unbundling -0.217 0.116 -1.880 0.06        
  Privatization -0.314 0.146 -2.160 0.03        
  Existence of Market Regulator -0.012 0.110 -0.110 0.92        
  Privatization and Regulator -0.204 0.306 -0.670 0.51        
  Privatization and Unbundling 0.461 0.321 1.430 0.15        
  Unbundling and Regulator 0.194 0.168 1.150 0.25        
  Constant -1.907 1.026 -1.860 0.07        
Note: The coefficients that are significant at 10% level are shown in bold.  
  
257 
 
Appendix II-1: Summary of previous econometric studies adopting a NIE approach 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Acemoglu et al. 
(2001) 
H: Settler mortality affected 
settlements; settlements 
affected early institutions; and 
early institutions persisted and 
formed the basis of current 
institutions. 
- Log GDP per capita 
(PPP) in 1995 
- Log output per worker 
in 1988 
 
- Average protection against 
expropriation risk, 1985-1995 
- Constraint on executive in 
1990 
- Constraint on executive in 
1900 
- Constraint on executive in first 
year of independence 
- Democracy in 1900 
- European settlements in 1900 
- Log of European settler 
mortality 
- Continent dummies 
- Latitude 
- Malaria in 1994 
- Life expectancy 
- Infant mortality 
- Mean temperature 
- Distance from coast 
- Yellow fever dummy 
By exploiting differences in 
European mortality rates as an 
instrument for current 
institutions, large effects of 
institutions on income per capita 
are estimated. Once the effect of 
institutions is controlled for, 
countries in Africa or those 
closer to the equator do not have 
lower incomes. 
Data Sources: World 
Bank, Political Risk 
Services, National 
Bureau of Economic 
Research (US), Atlas of 
World Population History 
Methodology: OLS 
estimation,  two-stage 
least-squares estimation 
258 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Acemoglu et al. 
(2008) 
H: There is a negative 
correlation between economic 
inequality and long-run 
economic development. 
- Secondary school 
enrolment 
- Primary school 
enrolment 
- Urbanization (1993) 
- Unsatisfied basic needs 
- Literacy rate (1937) 
- Urbanization (1937) 
- Share of buildings 
without access to public 
services 
 
- Land Gini 
- Contemporary land Gini 
- Political concentration index 
- Year of foundation of a 
municipality 
- Altitude of the municipality 
- Distance of the municipality to 
Bogota (the capital) 
- Area 
- Average rainfall 
- While the distribution of 
landed wealth in Cundinamarca 
was considerably more unequal 
than in northern U.S. states, it 
was less unequal than in the 
U.S. South. 
- There is a negative association 
between land inequality (land 
Gini) and political concentration 
across municipalities in 
Cundinamarca. 
- Land Gini (economic 
inequality) is positively 
associated with good outcomes. 
- There is a fairly robust 
negative relationship between 
political concentration (measure 
of political inequality) and good 
economic outcomes. 
Data: 
- Data on economic 
inequality in nineteenth 
century Cundinamarca 
are from  the cadastral 
(land census) data 
collected by the state of 
Cundinamarca in 1879 
and 1890 
- Data on politicians 
(mayors) are from the 
Registro del Estado and 
Gaceta de Cundinamarca, 
official newspapers 
- The contemporary data 
are from the 1993 
population census and the 
Colombian statistical 
agency DANE 
- Location and rainfall 
data from Instituto 
Geografico Agustin 
Codazzi in Bogota 
Methodology:  
- Cross-sectional ordinary 
least squares (OLS) 
estimations 
- Quantile regression 
estimation 
259 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Aghion et al. (2008) H: Democracy enhances 
economic growth. 
- 10 year output growth 
rate by sector 
- 10  year value-added 
growth rate by sector 
- 10 year employment 
growth rate by sector 
- Democracy 
- The distance to the 
technological frontier 
- Log of GDP per capita 
- Executive de facto 
independence 
- Constraints on executive power 
- Effectiveness of legislature 
- Government effectiveness 
- Competition in the legislative 
nominating process 
- Autocracy 
- Political rights average 1972-
99 
- Civil rights average 1972-99 
- Democratic institutions and 
political rights enhance growth 
of more advanced sectors. 
- An important channel of this 
effect is freedom of entry in 
markets. Political rights are 
associated with freedom of 
entry, and the latter is especially 
important for sectors close to the 
technological frontier. 
- More advanced economies 
benefit more from democratic 
institutions and therefore the 
demand for democracy should 
increase with the level of per 
capita income in a country. 
Data: 
- Industry employment 
and value-added data 
from the Industrial 
Statistics Database 
collected by the UNIDO 
(for 180 countries for the 
period 1963 to 2003) 
- Polity IV database and 
the Freedom House 
measures of civil liberties 
and political rights 
- Penn World Table 
Methodology: OLS 
estimation, fixed-effects 
model 
260 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Alesina and Rodrik 
(1994) 
H: An economy’s initial 
configuration of resources 
shapes the political struggle for 
income and wealth 
distribution, and this in turn 
affects long-run growth. 
- Average per capita 
growth rate over 1960-
1985 
- Average per capita 
growth rate over 1970-
1985 
- Gini coefficient of income 
inequality 
- Gini coefficient of land 
distribution inequality 
- Per capita GDP level in 1960 
- Primary school enrolment ratio 
in 1960 
- Dummy variable for 
democratic governments 
- There will be a strong demand 
for redistribution in societies 
where a large section of the 
population does not have access 
to the productive resources of 
the economy. Such conflict over 
distribution will generally harm 
growth. 
- The greater the inequality of 
wealth and income, the higher 
the rate of taxation, and the 
lower growth. 
- Inequality in land and income 
ownership is negatively 
correlated with subsequent 
economic growth. 
Data: 
- Cover 35 countries for 
1960-85 period 
- Heston and Summers 
dataset 
- Barro and Wolf dataset 
Methodology: OLS 
estimation 
Alesina et al. (1996) H1: There is a general 
correlation between economic 
growth and political stability. 
H2: Political stability fosters 
economic growth, and  low 
economic growth leads to 
political instability. 
- Annual rate of growth of 
per capita GDP 
- Government change 
- Major government 
change 
- Coups 
- Democracy 
- Executive adjustments 
- Number of unsuccessful 
attempts at changing the 
government 
- Log of real per capita GDP 
- World business cycle 
- In countries and time periods 
with a high degree of political 
instability, growth is 
significantly lower than 
otherwise.  
- The effect of growth on 
political instability is less clear: 
Data: 
- Cover 113 countries for 
1950-1982 period 
- Summers and Heston 
dataset 
- Jodice and Taylor 
dataset 
261 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Percentage of school age 
population enrolled in primary 
school 
- Dummy variable for countries 
in South American and Latin 
America 
- Dummy variable for countries 
in Africa 
- Dummy variable for industrial 
countries 
the effect of low economic 
growth on government collapses 
is strong for coups d’état but 
much less clear for other types 
of government change. 
- The occurrence of government 
changes in the recent past 
increases the probability of 
observing future collapses. 
- World Bank Economic 
and Social Data Base 
Methodology: OLS 
estimation 
 
Assane and Grammy 
(2003) 
H: The “quality” of the 
institutional framework 
positively affects economic 
development. 
- Level of income, 1985 - Physical capital formation 
- Labour force growth 
- Human capital formation 
- Economic freedom 
- Institutional efficiency 
- Institutional quality 
- Economic freedom-
institutional efficiency 
interaction 
- Economic freedom-
institutional quality interaction 
- “Good” institutions improve 
efficiency and accelerate 
growth. 
- The positive effect of 
institutional “quality” is more 
pronounced with mutually 
reinforcing support of economic 
freedom.  
- “Good” institutions help 
developing countries grow faster 
to achieve conditional 
convergence. 
- Economic development 
Data: 
- Cover 110 countries 
- Business International 
Corporation 
- Human development 
index 
- International Country 
Risk Guide 
Methodology: OLS 
estimation 
 
262 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
requires not only physical and 
human capital formation, but 
also freedom to choose and 
institutional support. 
Barro (1996) 
 
H: Economic freedoms, in the 
form of free markets and small 
governments that focus on the 
maintenance of property rights, 
encourage economic growth. 
- Growth rate of real per 
capita GDP over 1965-75 
period 
- Growth rate of real per 
capita GDP over 1975-85 
period 
- Growth rate of real per 
capita GDP over 1985-90 
period 
- Log of GDP 
- Male schooling 
- Female schooling 
- Log of life expectancy 
- Log of GDP x human capital 
- Log of fertility rate 
- Government consumption ratio 
- Public educational spending 
ratio 
- Black-market premium 
- Rule-of-law index 
- Terms-of-trade change 
- Investment ratio 
- Democracy index 
- Democracy index squared 
- Democracy index dummy 
- The favourable effects on 
growth include maintenance of 
the rule of law, free markets, 
small government consumption, 
and high human capital. 
- Once these kinds of variables 
and the initial level of real per 
capita GDP are held constant, 
the overall effect of democracy 
on growth is weakly negative. 
- There is a suggestion of a 
nonlinear relationship in which 
more democracy enhances 
growth at low levels of political 
freedom but depresses growth 
when a moderate level of 
freedom has already been 
attained.  
- Improvements in the standard 
Data:  
- Summers-Heston data 
set 
- World Bank 
- Barro-Lee data set 
(Economics Department, 
Harvard University) 
- Gastil measures of 
political rights 
Methodology: 
- Instrumental-variable 
technique 
- OLS estimation 
263 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
of living—measured by GDP, 
health status, and education—
substantially raise the 
probability that political 
freedoms will grow. 
Barro (2000) H: Income inequality has an 
effect on macroeconomic 
performance, as reflected in 
rates of economic growth and 
investment. 
- Average growth rate 
(1965 to 1975) 
- Average growth rate 
(1975 to 1985) 
- Average growth rate 
(1985 to 1995) 
- Ratio of real investment 
to real GDP 
- Log of per capita GDP 
- Log of per capita GDP squared 
- Government consumption/GDP 
- Rule-of-law index 
- Democracy index 
- Democracy index squared 
- Inflation rate 
- Years of schooling 
- Log of total fertility rate 
- Investment/GDP 
- Growth rate of terms of trade 
- Evidence from a broad panel 
of countries shows little overall 
relation between income 
inequality and rates of 
growth and investment. 
- For growth, higher inequality 
tends to retard growth in poor 
countries and encourage growth 
in richer places. 
- The Kuznets curve—whereby 
inequality first increases and 
later decreases during the 
process of economic 
development—emerges as a 
clear empirical regularity. 
However, this relation does not 
explain the bulk of variations in 
inequality across countries or 
over time. 
Data: 
- Deininger and Squire 
dataset 
Methodology: 
- Conditional 
convergence 
- Panel estimation (fixed 
effects model) 
 
264 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Barro (1991) H: There are some empirical 
regularities about growth, 
fertility, and investment. 
- Per capita GDP growth 
- Investment 
- School-enrolment rates at the 
secondary levels in 1960 
- School-enrolment rates at the 
primary levels in 1960 
- GDP per capita in 1960 
- The number of revolutions and 
coups per year 
- The number per million 
population of political 
assassinations per year 
- Mortality rates for children 
aged 0-4 
- Student-teacher ratio 
- Adult literacy rate 
- The total fertility rate 
- Dummy variables for Africa 
and Latin America 
- The growth rate of real per 
capita GDP is positively related 
to initial human capital (proxied 
by school-enrolment rates) and 
negatively related to the initial 
level of real per capita GDP.  
- Countries with higher human 
capital have lower fertility rates 
and higher ratios of physical 
investment to GDP. 
- Growth is inversely related to 
the share of government 
consumption in GDP, but 
insignificantly related to the 
share of public investment. 
- Growth rates are positively 
related to measures of political 
stability and inversely related to 
a proxy for market distortions. 
Data: 
- Cover 98 countries in 
the period 1960-1985 
- Summers and Heston 
dataset 
- United Nations 
- World Bank 
Methodology: OLS 
estimation 
 
Besley and 
Kudamatsu (2008) 
H1: Autocratic regimes do not 
always perform badly, at least 
as judged by economic 
indicators, such as the growth 
- Life expectancy 
- Gross primary school 
enrolment ratio 
- Economic growth 
- Per capita income 
- Ethnic fractionalization 
- European settlers’ mortality 
- French legal origin 
- Democracies can be better or 
worse than autocracies in terms 
of accountability. 
- Successful autocracies are 
Data: 
- World Bank’s World 
Development Indicators 
- UNESCO Institute for 
265 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
rate of income per capita or 
other components of the 
human development index. 
H2: Given that democracy per 
se does not guarantee good 
economic performance, some 
features of autocratic regimes 
may be conducive to good 
economic performance. 
- Socialist legal origin 
- German legal origin 
- Oil price boom 
- Decade dummies 
- Region dummies 
- Number of leadership changes 
per year 
those where poor-quality 
leadership leads to removal of 
leaders from office. 
- The forces shaping leadership 
replacement may be at work in 
successful autocracies. 
Leadership turnover is greater in 
successful compared to 
unsuccessful autocracies. 
Statistics 
Methodology: OLS 
estimation 
 
Caselli et al. (1996) H: There are two sources of 
inconsistency in existing cross-
country empirical work on 
growth: correlated individual 
effects and endogenous 
explanatory variables. 
- Change in growth rate - GDP per capita in previous 
year 
- Male education 
- Female education 
- I/GDP 
- G/GDP 
- ln(1+BMP) 
- Revolutions 
- Life expectancy 
- Assassinations 
- Terms of trade 
- Per capita incomes converge to 
their steady-state levels at a rate 
of approximately 10 percent per 
year. This result stands in sharp 
contrast to the current 
consensus, which places the 
convergence rate at 2 percent. 
- The results reject both the 
standard and the augmented 
version of the Solow model. 
Data: 
- Maddison dataset 
- Summers and Heston 
dataset 
Methodology: 
- Regressions using a 
generalized method of 
moments estimator. 
Clague et al. (1996) 
 
H1: Any incentive an autocrat 
has to respect property and 
contract rights comes from his 
- ICRG index 
- BERI index 
- Credit risk 
- The number of consecutive 
years that a country has been a 
democracy 
- There is a compelling 
empirical relationship between 
property and contract rights and 
Data: 
- The Gurr and Banks 
database (1986-90) 
266 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
interest in future tax 
collections and national 
income and increases with his 
planning horizon. 
H2: In autocracies it is the 
time horizon of the individual 
autocrat (or occasionally the 
ruling clique) that is the main 
determinant of property and 
contract rights, whereas in 
democracies these rights 
depend upon whether the 
democratic system is durable. 
- Currency depreciation 
- Black market exchange 
premium 
 
- The number of years that the 
chief executive has been in 
office in a democracy 
- The number of consecutive 
years that the chief executive in 
an autocratic nation has been in 
power 
- The duration of the ruling 
group 
- The amount of contract-
intensive money 
- Indexes from two firms 
evaluating risks to foreign 
investors 
- Credit rating variable 
- The rate of currency 
depreciation 
- Per capita income 
an autocrat’s time in power.  
- Autocrats who had been in 
power longer and who had 
reason to have longer time 
horizons were associated with 
better property and contract 
rights than autocrats who were 
in power only for a shorter time. 
- In general, democracies 
provide greater security of 
property and contractual rights 
than autocracies. But these 
benefits of democracy did not 
appear quickly: the property and 
contract rights were often poor 
in democracies that had lasted 
only a short time. 
- Gastil indexes 
- Europa Yearbook 
- The International 
Country Risk Guide 
(ICRG), published since 
1982 
- Business Environmental 
Risk Intelligence (BERI), 
published since 1972 
- Report titled 
“Institutional Investor” 
Methodology: OLS 
estimation 
 
Drury et al. (2006) H: One of democracy’s 
indirect benefits is its ability to 
mitigate the detrimental effect 
of corruption on economic 
growth. 
- Growth of GDP - Level of corruption 
- Life expectancy 
- Trade openness 
- Population growth 
- Log of GDP per capita 
- Corruption has no significant 
effect on economic growth in 
democracies, while non-
democracies suffer significant 
economic harm from corruption. 
Data: 
- Time-series cross-
section data for more than 
100 countries from 1982 
to 1997 
267 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Tropical climate 
- Government spending 
- World Bank’s World 
Development Indicators 
(2003) 
- International Country 
Risk Guide’s (ICRG) 
1982-97 assessments 
- Transparency 
International 
- Polity IV database 
- Freedom House 
measure of democracy 
Methodology: OLS 
estimation 
Easterly and Levine 
(1997) 
H1: Higher levels of ethnic 
diversity encourage poor 
policies, poor education, 
political instability, inadequate 
infrastructure, and other 
factors associated with slow 
growth. 
H2: There is a direct effect of 
ethnic diversity on economic 
growth and an indirect effect 
- Average annual growth 
rate of GDP per capita in 
the 1960s, 1970s, and 
1980s 
 
- Initial income 
- Ethnolinguistic diversity 
- School attainment 
- Political stability 
- Financial development 
- Black market premium 
- Fiscal surplus 
- Infrastructure development 
- Dummy variables for Africa 
and Latin America 
- Low school attainment, 
political instability, poorly 
developed financial systems, 
large black market exchange 
rate premiums, large 
government deficits, and 
inadequate infrastructure are 
significantly correlated with 
economic growth. 
Data: 
- Barro and Lee dataset 
- World Bank 
- IMF 
- Pick’s Currency 
Yearbook 
- Political Risk Services 
- World Resources 
Institute 
 
268 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
of ethnic diversity on public 
policy choices that influence 
long-run growth rates. 
- Assassinations 
- Financial depth 
- Fiscal surplus/GDP 
Methodology: 
- Technique of seemingly 
unrelated regressions 
Helliwell (1994) H: There are two-way linkages 
between democracy and 
economic growth. 
- Growth in real GDP per 
adult from 1960 to 1985 
- Bollen democracy index 
- Real GDP per adult in 1960 
- Investment rate 
- Schooling rate 
 
- The effects of income on 
democracy are found to be 
robust and positive. 
- It is still not possible to 
identify any systematic net 
effects of democracy on 
subsequent economic growth. 
Data: 
- Cross-sectional and 
pooled data for up to 125 
countries over the period 
from 1960 to 1985 
- Bollen index for 1960 
- Gastil index for 1976 
and 1985 
- World Bank 
Methodology: OLS 
estimation 
Isham et al. (1997) H: There is a link between 
civil liberties and democracy - 
critical determinants of how 
governments exercise public 
decisions and authority - and 
the efficacy of public 
investments. 
- Economic rate of return 
of government projects 
- The probability of a 
project being rated 
satisfactory 
- Freedom House civil liberties, 
1978-87 
- Humana, 1982-85 
- Media pluralism, 1983-87 
- Freedom to organize, 1983-87 
- There is a strong empirical link 
between civil liberties and the 
performance of government 
projects. 
- Even after controlling for other 
determinants of performance, 
countries with the strongest civil 
liberties have projects with an 
economic rate of return 8-22 
Data: 
- World Bank’s 
Operations Evaluation 
Department 
- Freedom House’s civil 
liberties index 
- UN’s Humana index 
Methodology: OLS and 
probit regressions 
269 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
percentage points higher than 
countries with the weakest civil 
liberties. 
- The strong effect of civil 
liberties holds true even when 
controlling for the level of 
democracy. 
 
Keefer and Knack 
(1997) 
H: Poor countries are falling 
back rather than catching up 
with wealthy countries. 
Deficient institutions underlie 
this divergence. 
- Average real per capita 
growth in GDP, 1960-
1989 
- GDP/Capita, 1960 
- Country Risk Index 
- Business Risk Index 
- Executive Constraints 
- Primary School Enrolment 
- Secondary School Enrolment 
- Labour Force Growth 
- Price Changes 
- Income Gap 
- Institutional Variable 
- Institution x Income Gap 
- The ability of poor countries to 
catch up is determined in large 
part by the institutional 
environment in which economic 
activity in these countries takes 
place. 
- Institutions are powerful 
determinants of the ability of 
countries to benefit from the 
“catch-up” effect. While poorer 
countries may have advantages 
because of low-cost access to 
advanced technology or the 
diminishing returns experienced 
by wealthier countries, these 
potential advantages appear to 
be squandered in countries with 
poor institutional frameworks. 
Data: 
- International Country 
Risk Guide (ICRG) 
- Business Environmental 
Risk Intelligence (BERI) 
Methodology: OLS 
estimation 
 
270 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Keefer and Knack 
(2002) 
H: Social polarization reduces 
the security of property and 
contract rights and, through 
this channel, reduces growth. 
- International Country 
Risk Guide Index (ICRG), 
1986–95 
- Annual growth in per 
capita income over the 
1970–92 period 
- Ethnic tensions (0-6 scale) 
- The percent of a country’s 
population belonging to the 
largest ethnic group 
- Log of 1985 GDP per capita 
- Per capita growth, 1980-85 
- Aggregate GDP, 1985 
- Gini: income inequality 
- Gini: land inequality 
- Ethnic homogeneity 
- Political violence 
- Regime type 
- Continent dummy 
- Log of 1970 GDP per capita 
- Mean years of education, 1970 
- Property rights index, 1982 
- Polarization makes large 
changes in current policies, 
including those guaranteeing the 
security of contract and property 
rights, more likely under a wide 
range of institutional 
arrangements. 
- Social polarization may 
directly undermine the security 
of rights. 
- If the insecurity of property 
rights slows growth in unequal 
or otherwise polarized societies, 
then governments that commit 
over the long-run to particular 
redistributive policies incur less 
risk of slowing economic 
growth. 
Data: 
- International Country 
Risk Guide (ICRG), 
published by Political 
Risk Services (from 1982 
onwards) 
Methodology: OLS 
estimation 
 
271 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Knack and Keefer 
(1997) 
H: Trust and civic norms have 
an influence on growth and 
investment rates. 
- Average annual growth 
in per capita income over 
the 1980-1992 period for 
29 countries 
- Log of output/worker 
- Log of capital/worker 
- School/worker 
- Log of total factor 
productivity 
- TRUST (the percentage of 
respondents in each nation 
replying “most people can be 
trusted”) 
- CIVIC (the strength of civic 
norms) 
- Per capita GDP in 1980 
- Labour force growth 
- (Exports + Imports )/GDP 
- M2/GDP 
- Black market premium 
- Property rights (ICRG) 
- Currency depreciation 
- Investor credit rating 
- Gini (income) 
- Confidence in government 
- Ethnic homogeneity 
- Trust and civic cooperation are 
associated with stronger 
economic performance. 
- Associational activity is not 
correlated with economic 
performance. 
- Trust and norms of civic 
cooperation are stronger in 
countries with formal 
institutions that effectively 
protect property and contract 
rights, and in countries that are 
less polarized along lines of 
class or ethnicity. 
Data: 
- The World Values 
Surveys containing data 
on thousands of 
respondents from 29 
market economies 
- International Country 
Risk Guide (ICRG) 
- Business Environmental 
Risk Intelligence (BERI) 
Methodology: OLS, 
2SLS estimation 
 
Mauro (1995) H: Efficient government 
institutions foster economic 
growth. Corruption and other 
institutional factors affect 
economic growth. 
- Total Investment/GDP 
- Equipment 
investment/GDP 
- Nonequipment 
investment/GDP 
- Equipment 
- Index of ethnolinguistic 
fractionalization 
- Bureaucratic efficiency index 
(BE) for 1980-1983 
- Political Change 
- Political Stability 
- Corruption lowers private 
investment, thereby reducing 
economic growth. 
- Bureaucratic efficiency may be 
at least as important a 
determinant of investment and 
Data: 
- Business International 
(BI) indices on 
corruption, red tape, and 
the efficiency of the 
judicial system for the 
272 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
investment/Nonequipment 
investment 
- Private investment/GDP 
- Public investment/GDP 
- Private 
investment/Public 
investment 
- Probability of Opposition 
Group Takeover 
- Stability of Labour 
- Relationship with 
Neighbouring Countries 
- Terrorism 
- Legal System, Judiciary 
- Bureaucracy and Red Tape 
- Corruption 
- Secondary education 
- Population growth 
- Primary education 
- Government expenditure 
- Revolutions and coups 
- Assassinations 
growth as political stability. period 1980-1983 for 70 
countries 
Methodology: 
Instrumental variable 
technique, OLS 
estimation 
 
 Nunn (2008) H1: Large-scale plantation 
slavery resulted in economic 
inequality. 
H2: This resulted in 
subsequent underdevelopment. 
- Per capita GDP in 2000 - Fraction slaves 
- Nonplantation slaves 
- Plantation slaves  
- Population density 
 
- Slavery was detrimental for 
economic development. 
Data: 
- Historic population data 
from a variety of sources, 
most often population 
censuses 
- Data on country-level 
per capita GDP in 2000 
are from World Bank, 
273 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Penn World Table 
- Population density and 
land data are from 
Harvard Centre for 
International 
Development’s 
Geography Database 
Methodology: OLS 
estimation 
Persson and Tabellini 
(1994) 
H: Inequality is harmful for 
growth. 
- Annual average growth 
rate of GDP per capita 
- Income Distribution 
- Political Participation 
- Average Skills 
- The Level of Development 
- Initial GDP 
- There is a significant and large 
negative relation between 
inequality and growth. 
- This relation is only present in 
democracies. 
Data: 
- Cover 1830-1985 period 
for 9 countries 
 - Summers and Heston 
dataset 
- U.S. Department of 
Commerce 
- World Bank 
- OECD 
Methodology: 
- Model building 
- OLS estimation 
- 2SLS estimation  
274 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Persson and Tabellini 
(2008) 
H: There is a positive relation 
between democracy and 
growth. 
- Per capita income - Length of sample 
- Income relative to the United 
States 
- War years 
- Domestic democratic capital 
- Foreign democratic capital 
- Initial value of polity score 
- Transitions from autocracy to 
democracy are associated with 
an average growth acceleration 
of about 1 percentage point, 
producing a gain in per capita 
income of about 13% by the end 
of the sample period. 
- The effect of transitions in the 
opposite direction is larger: a 
relapse from democracy to 
autocracy slows down growth 
by almost 2 percentage points on 
average, which implies an 
income fall of about 45% at the 
end of the sample. 
Data: 
- Annual per capita 
income data from Penn 
World Tables (1960-
2000) 
- Polity IV database 
Methodology: Semi 
parametric methods, 
difference-in-difference 
methods, propensity-
score methods, OLS 
estimation 
Scully (1988) H: The material progress 
mankind made in modern 
times has been affected 
significantly by the choice of 
the institutional framework 
designed to bring it about. 
- Economic growth over 
the period 1960-80 
- Economic efficiency 
- Change in economic 
efficiency 
- The compound growth rate of 
real per capita GDP 
- The compound growth rate in 
the capital-labour ratio 
- Politically liberal 
- Politically not liberal 
- Civil liberty 
- Limited civil liberty 
- The institutional framework 
has significant and large effects 
on the efficiency and growth 
rate of economies. 
- Politically open societies, 
which subscribe to the rule of 
law, to private property, and to 
the market allocation of 
Data: 
- Cover 115 countries for 
1960-1980 period 
- Gastil measures of 
liberty 
- Summers and Heston 
dataset 
Methodology: OLS 
275 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Economic liberty resources, grow at 3 times the 
rate and are 2.5 times as 
efficient as societies in which 
these freedoms are abridged. 
estimation 
 
Spindler (1991) H: There is appositive relation 
between economic freedom 
and economic development for 
most countries in the world. 
- Gross National Product 
per capita 
- Private GNP per capita 
- Economic freedom 
- Property freedom 
- Movement freedom 
- Association freedom 
- Information freedom 
- Civil liberties 
- Economic system 
- Dummy for oil exporting 
countries 
- Dummy for industrial countries 
- The relationship between 
economic freedom and 
economic development is strong 
and direct for such economic 
freedoms as freedom of property 
and freedom of movement but 
inverse for freedom of 
association. 
- The findings appear to be 
independent of the type of 
economic system or civil 
liberties, as measured by the 
Gastil ratings, which have their 
own important effects on 
economic development. 
Data: 
- Wright Economic 
Freedom Ratings 
covering 165 countries 
for 1982 onwards 
Methodology: OLS 
estimation 
 
Vanssay and Spindler 
(1994) 
H: There is a relationship 
between per-capita income, the 
entrenchment of various rights 
in a country’s constitution and 
- GNP per capita - Education 
- Economic freedom 
- Political structure 
- Specific protections against 
- Entrenchment of any single 
right seldom has a significant 
general economic effect, while 
the effect of economic freedom 
Data: 
- Cover 100 countries for 
1988 
- UNDP Human 
276 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
the level of economic freedom 
in a country. 
tyranny 
- Social Rights 
is significant and substantial. 
- Education, economic freedom, 
population growth and the 
saving ratio together explain 
more than 75% of the cross-
country variation in per capita 
income.  
- The entrenched elements of 
“Political structure”, 
“Protections from tyranny”, or 
“Social Charter” are not 
revealed as important 
explanatory variables. 
Development Report 
1991 
- Scully and Slottje 
dataset 
- Taubenfel dataset 
Methodology: OLS 
estimation 
 
  
277 
 
Appendix II-2: Summary of previous econometric studies based on NIE by their focus 
 
Focus of the study Major Explanatory Variable(s) Data Sources Examples 
The relationship between 
historical institutions and present 
economic performance  
- Protection against expropriation risk 
- Constraint on executive 
- Democracy  
- European settler mortality 
- Continent dummies 
- Number of slaves 
- Population density 
- Land Gini 
- Political concentration index 
World Bank, Political Risk Services, National 
Bureau of Economic Research (US), Atlas of 
World Population History, Harvard Centre 
for International Development’s Geography 
Database, the cadastral (land census) data, 
official newspapers, population census, the 
Colombian statistical agency (DANE) 
Acemoglu et al. (2001), Acemoglu et al. 
(2008), Nunn (2008) 
The relationship between political 
institutions and economic growth 
- Polity score 
- GDP per capita 
- Executive independence 
- Constraints on executive power 
- Effectiveness of legislature 
- Government effectiveness 
- Political and civil rights 
- Ethnic fractionalization 
- Legal origin 
- The duration of the ruling group 
- Domestic and foreign democratic capital 
- Bureaucratic efficiency index 
- Probability of opposition group takeover 
Industrial Statistics Database of  the UNIDO, 
Polity IV database, Freedom House, Penn 
World Table, World Bank’s World 
Development Indicators, UNESCO Institute 
for Statistics, Business International (BI), 
Heston and Summers dataset, Barro and Wolf 
dataset, Maddison dataset, International 
Country Risk Guide (ICRG), Transparency 
International, Bollen index, Gastil index, 
World Bank’s Operations Evaluation 
Department, UN’s Humana index, Summers 
and Heston dataset, Jodice and Taylor dataset 
Scully (1988), Helliwell (1994), Alesina 
and Rodrik (1994), Mauro (1995), 
Clague et al. (1996), Caselli et al. 
(1996), Alesina et al. (1996), Isham et al. 
(1997), Drury et al. (2006), Aghion et al. 
(2008), Besley and Kudamatsu (2008), 
Persson and Tabellini (2008)  
 
278 
 
Focus of the study Major Explanatory Variable(s) Data Sources Examples 
- Revolutions and coups 
- Assassinations 
- Gini coefficient of income and land 
distribution inequality 
- Primary school enrolment 
- Level of corruption 
- Life expectancy 
- Trade openness 
- Media pluralism 
- Freedom to organize  
The relationship between social 
structure and economic growth 
- Ethnic tensions 
- The percent of a country’s population 
belonging to the largest ethnic group 
- GDP per capita 
- Gini: income and land inequality 
- Ethnic homogeneity 
- Political violence 
- Regime type 
- Mean years of education 
- Property rights index 
- TRUST (the percentage of respondents in 
each nation replying “most people can be 
trusted”) 
- CIVIC (the strength of civic norms) 
International Country Risk Guide (ICRG) 
published by Political Risk Services, the 
World Values, Business Environmental Risk 
Intelligence (BERI) 
  
Knack and Keefer (1997), Easterly and 
Levine (1997), Keefer and Knack (2002) 
279 
 
Focus of the study Major Explanatory Variable(s) Data Sources Examples 
- Labour force growth 
- Currency depreciation 
- Investor credit rating 
- Confidence in government 
- Black market premium 
- Fiscal surplus 
- Infrastructure development 
- Assassinations 
- Financial depth 
The relationship between 
economic institutions (economic 
equality, protection of property 
rights etc.) and economic growth 
- GDP per capita 
- Schooling 
- Life expectancy 
- Fertility rate 
- Public educational spending  
- Rule-of-law index 
- Investment  
- Democracy index 
- Country and business risk index 
- Executive constraints 
- School enrolment 
- Labour force growth 
- The number of revolutions and coups 
- The number of political assassinations 
- Mortality rates for children aged 0-4 
Summers-Heston data set, World Bank, 
Barro-Lee data set (Economics Department, 
Harvard University), Gastil measures of 
political rights, Deininger and Squire dataset, 
International Country Risk Guide (ICRG), 
Business Environmental Risk Intelligence 
(BERI), United Nations, Wright Economic 
Freedom Ratings, Business International 
Corporation, human development index, U.S. 
Department of Commerce, OECD, Scully and 
Slottje dataset, Taubenfel dataset 
 
Barro (1991), Spindler (1991), Persson 
and Tabellini (1994), Vanssay and 
Spindler (1994), Barro (1996), Keefer 
and Knack (1997), Barro (2000), Assane 
and Grammy (2003) 
280 
 
Focus of the study Major Explanatory Variable(s) Data Sources Examples 
- Student-teacher ratio 
- Adult literacy rate 
- Economic freedom 
- Property freedom 
- Movement freedom 
- Association freedom 
- Information freedom 
- Civil liberties 
- Physical capital formation 
- Labour force growth 
- Human capital formation 
- Income Distribution 
- Political Participation 
- Average Skills 
 
  
281 
 
Appendix II-3: Classification of variables employed in previous econometric studies based on NIE by what they measure 
 
Variables measuring presence of institutions Variables measuring organization of institutions Variables measuring outcome of institutions Control Variables 
- Ethnic fractionalization 
- Confidence in government 
- Revolutions and coups 
- Educational and professional background of 
decision makers 
 
 
- Bureaucratic efficiency 
- Constraints on executive 
- Government effectiveness 
- Legislative effectiveness 
- Level of corruption 
- Political concentration 
- Regime type 
- Rule of law 
- Polity score 
- Security of property rights 
- Civil liberties 
- Political rights 
- Country and business risk  
- Economic freedom 
- GDP per capita 
- Investment level 
- Gini coefficient of income and land 
distribution inequality 
- School enrolment (education) 
- Fertility rate 
- Life expectancy 
- Literacy rate 
- Mortality rates for 
children 
- Population 
- Skills 
 
  
282 
 
Appendix III-1: Summary of previous applied econometric studies adopting a political economy approach 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Alesina et al. (2006) H: It is easier to stabilize an 
economy more decisively in 
times of crisis than in times of 
more “moderate” economic 
problems 
- Deficit/GDP ratio 
- Inflation rate 
- Number of executive 
constraints 
- Election year 
- Political orientation of the 
ruling government 
- Assembly or parliamentary 
system 
- Executive control of absolute 
majority 
- Number of years left in the 
current term for the executive 
- Total government deficit as a 
share of GDP and inflation 
- The real per capita GDP  
- The ratio of exports and 
imports to GDP 
- The dummy taking value 1 if 
the country is currently in crisis 
- Participation to IMF programs 
- Stabilizations are more likely 
to occur during crisis, at the 
beginning of term of office of a 
new government, in countries 
with “strong” governments (i.e. 
presidential systems and unified 
governments with a large 
majority of the party in office), 
and when the executive faces 
less constraints 
- The role of external 
inducements like IMF programs 
has at best a weak effect 
Data: 
- Yearly data on a large 
sample of developed and 
developing countries 
covering from 1960 to 
2003 
- Source(s): Polity IV 
project, World Bank's 
Database of Political 
Institutions, IMF's 
International Financial 
Statistics (IFS) database, 
Penn World Table 
Methodology: 
- OLS 
283 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Boschini (2006) H-1: The skills of the political 
elite and political institutions 
play a crucial role for 
industrialisation to occur 
H-2: The government 
(controlled by elite or through 
a pivotal voter) must have the 
ability as well as the incentives 
to promote the industrialisation 
process 
- Industrialisation index 
- GDP growth 1820-1913 
 
- Political regime 
- Enrolment in primary 
education 
- Index of the favourableness of 
attitudes toward 
entrepreneurship 
- Index of concentration in 
landholdings 
 
- A flat wealth distribution and 
skilled political elite enhance 
development the most in elitist 
regimes, while democracies 
perform as well as elitist 
regimes in terms of 
industrialisation 
 
Data: 
- 23 countries from 1820 
to 1913 
- Source(s): Comparative 
Patterns of Economic 
Development 1850–1914, 
John Hopkins University 
Methodology: 
- Partial sums of squares 
Chang and Berdiev 
(2011) 
H: Government ideology, 
political factors and 
globalization are crucial for 
energy regulation in electricity 
and gas industries 
 
 
- The growth rate of 
regulation indicator in 
energy industry 
- Government ideology 
- Herfindahl index to proxy for 
government fragmentation 
- Number of years that the 
incumbent government has been 
in office 
- Index of political constraints 
- Globalization index 
- Energy demand 
- Real GDP per capita (constant 
in 2000) 
- Left-wing governments 
promote regulation in gas and 
electricity sectors 
- Less politically fragmented 
institutions contribute to 
deregulation of gas and 
electricity industries 
- Long tenures of incumbent 
government have limited impact 
on regulation in electricity 
sector, while it is associated 
with an increase in regulation of 
gas sector 
Data: 
- 23 OECD countries 
over the period 1975-
2007 
- Source(s): Conway and 
Nicoletti (2006), Potrafke 
(2009), Beck et al. 
(2001), Henisz (2000), 
World Bank, BP 
Methodology: 
- The bias-corrected least 
square dummy variable 
model 
284 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Higher political constraints and 
more globalization lead to 
deregulation in electricity and 
gas sectors 
- Economic and social 
integration are the forces that 
promote deregulation in the gas 
industry, whereas political 
integration advances 
deregulation in the electricity 
industry 
 
Cubbin and Stern 
(2006) 
H: A regulatory law and 
higher quality regulatory 
governance are associated with 
superior outcomes in the 
electricity industry 
- Per capita generation 
capacity 
- Electricity (or energy) 
regulatory law 
- Autonomous or ministry 
regulator 
- License fee or government 
budget regulatory funding 
- Free or mandatory civil service 
pay scales for regulatory staff  
- Real GDP per capita 
- Debt payments as a proportion 
of national income 
- Industry value added as 
proportion of GDP 
- Controlling for privatization 
and competition and allowing 
for country-specific fixed 
effects, both regulatory law and 
higher quality regulatory 
governance are positively and 
significantly associated with 
higher per capita generation 
capacity 
Data: 
- 28 developing 
economies over 1980-
2001 
- Source(s): U.S. Energy 
Information Agency, 
World Bank 
Methodology: 
- Panel data modelling, 
error correction models 
285 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Dreher et al. (2009) H: The educational and 
professional background of a 
head of government matters 
for the implementation of 
market-liberalizing reforms 
- Composite index of 
economic freedom 
- Size of government 
index 
- Legal structure and 
security of property rights 
index 
- Access to sound money 
index 
- Exchange with 
foreigners index 
- Regulation of credit, 
labour and business index 
- Profession of heads of 
governments 
- Education of heads of 
governments 
- Economic freedom 
- Economic growth 
- Civil liberties 
- Aid 
- Linguistic fractionalization 
- Currency crises 
- Government fractionalization 
- Coalition government 
- Reforms are more likely during 
the tenure of former 
entrepreneurs 
- Entrepreneurs belonging to a 
left-wing party are more 
successful in inducing reforms 
than a member of a right-wing 
party with the same previous 
profession 
- Former professional scientists 
also promote reforms, the more 
so, the longer they stay in office 
- The impact of politicians’ 
education is not robust and 
depends on the method of 
estimation 
Data: 
- Panel data over the 
period 1970–2002 
- Profession and 
education of more than 
500 political leaders from 
72 countries 
- Source(s): Gwartney 
and Lawson (2004), 
World Bank, Alesina et 
al. (2003), Freedom 
House, Beck et al. 
(2001), Dreher (2006) 
Methodology: 
- Pooled time-series 
cross-section (panel data) 
regressions 
- Feasible generalized 
least squares 
Duso and Seldeslachts 
(2010) 
H: Differences in political, 
government and regulatory 
environments explain the 
differing speed of reforms in 
- Degree of liberalization 
in the digital mobile 
industry 
- Share of incumbent operator in 
long-distance telecom 
- Annual revenues in the mobile 
telecommunications industry 
- Majoritarian electoral systems 
are important drivers for change, 
while independent industry 
regulators slow down the 
Data: 
- 24 OECD countries  
- Source(s): OECD 
regulation database, 
286 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
the mobile telecom industries 
at the beginning of the 1990s 
- Dummy variables for 
regulatory independence 
- Number of parties in the 
opposition 
- Percentage seats in the 
legislature held by government 
parties 
- Government’s programmatic 
position: Pro market regulations 
- Government’s programmatic 
position: Pro welfare state 
limitation 
- Population 
- Share of active population aged 
between 15 and 64 years 
- Annual income per capita  
reforms 
- Powerful industry incumbents 
hold up the liberalization 
process and governing bodies 
that favour a small welfare state 
accelerate it 
Persson and Tabellini 
(1999), Woldendorp et al. 
(1998), Budge et al. 
(2001), Lijphart (1999) 
Methodology: 
- Ordered probit model 
with country random-
effects 
Duval (2008) H: Macroeconomic policies 
and ideology influence reform 
patterns in labour and product 
markets 
- Policy index - Unemployment 
- Output gap 
- Crisis years 
- Small country 
- Ideology 
- Fractionalisation 
- Degree of sustainability of 
- Sound public finances and 
fiscal expansions help foster 
reforms 
- The effect of fiscal expansion 
may also be greater for countries 
that pursue fixed exchange-rate 
regimes 
Data: 
- 21 OECD countries 
over the period 1985–
2003 
- Source(s): OECD 
Methodology: 
- Multivariate probit and 
287 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
public debt 
- Fiscal expansion 
 linear econometric 
models 
Fredriksson and 
Wollscheid (2008) 
H: Corruption and political 
instability are important 
determinants of firm 
investment in pollution control 
technology 
- Level of investment in 
clean technology in the 
steel industry 
- The respect that institutions 
and citizens use to govern their 
interactions 
- The degree to which business 
transactions involve corruption 
- The perception of the quality of 
public service provision 
- Political instability 
- Social and Institutional 
Capacity index 
- The size of the steel market 
- Per capita steel production 
- Total steel exported as a 
percentage of total steel 
produced 
- Total trade as a share of GDP 
- Gastil index 
- Government commitment 
- Per capita gross domestic 
product 
- Greater corruptibility increases 
the level of abatement 
technology investment because 
the strategic incentive to 
underinvest in pollution control 
technology declines when 
policymakers become more 
corruptible 
- Political instability raises 
abatement technology 
investment 
Data: 
- Steel-sector panel data 
from 41 countries for the 
years 1992–1998 
- Source(s): International 
Iron and Steel Institute, 
Kaufman et al. (1999), 
Banks (1995), CIESIN of 
Columbia University, 
Freedom House, World 
Bank 
Methodology: 
- Panel data estimation 
(fixed and random effects 
models) 
288 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Gasmi et al. (2009) H: There is a strong 
relationship between the 
quality of political institutions 
and the performance of 
regulation in 
telecommunications sector 
- Mainline coverage 
- Cellular subscription 
- Mainlines per employee 
- Price of monthly 
subscript to fixed-line 
service 
- Price of cellular service 
- Regulatory governance index 
- Corruption 
- Bureaucracy 
- Law and order 
- Expropriation 
- Currency risk 
- Institutional environment index 
- Checks and balances 
- Privatization 
- Competition in fixed and 
cellular line services 
- Rural population 
- Population density 
- The impact of political 
accountability on the 
performance of regulation is 
stronger in developing countries 
- Future reforms in these 
countries should give due 
attention to the development of 
politically accountable systems 
Data: 
- Panel data for 29 
developing countries and 
23 developed countries 
during 1985–99 
- Source(s): Gasmi, 
Noumba, and Recuero 
Virto (2006) 
Methodology: 
- Differenced generalized 
method-of-moments 
estimation 
Gasmi and Recuero 
Virto (2010) 
H: Sectoral, institutional and 
financial factors are important 
determinants of the reforms 
implemented in 
telecommunication industry 
- Fixed-line deployment 
- Cellular competition 
(analogue) 
- Counter (analogue) 
- Cellular competition 
(digital) 
- Counter (digital) 
- Fixed-line competition 
(local) 
- Separate regulator 
- Corruption 
- Institutional index 
- Democracy index 
- Risk index 
- Total debt service 
- Net taxes on products 
- Aid per capita 
- Population density 
- Rural population 
- Imports 
- Sectoral as well as institutional 
and financial factors are found 
to be important determinants of 
the actual reforms implemented 
- There is a positive relationship 
between the decision to 
introduce competition in the 
digital cellular segment and the 
growth of the fixed line segment 
- Countries facing increasing 
Data: 
- 1985-1999 panel data 
on 86 developing 
countries 
- Source(s): Available 
from the authors upon 
request 
Methodology: 
- Duration methodology 
- System Generalized 
289 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Privatization - Telecommunications staff 
- Checks and balances 
- English legal origin 
- French legal origin 
- Share of protestant (1980) 
- Latitude 
- Average schooling years 
(1980) 
- Ethno linguistic 
fractionalization 
- Africa 
- Crop and forest land 
- Political constraints 
- Free press 
- Ethnic tensions 
- Law and order  
institutional risk and financial 
constraints are more likely to 
introduce competition in the 
digital cellular segment and to 
privatize the fixed-line 
incumbent 
- Competition in the analogue 
cellular segment and the 
creation of a separate regulator 
seem to be relatively less 
attractive policies 
Method of Moments 
(SYS-GMM) 
Goldberg and Pavcnik 
(2005) 
H: Worker industry affiliation 
plays a crucial role in how 
trade policy affects wages in 
many trade models 
- Wage differentials - Worker characteristics 
- Occupation indicators 
- Job type indicators 
- Place of work characteristics 
 
- Without industry fixed effects, 
workers in protected sectors 
earn less than workers with 
similar observable 
characteristics in unprotected 
sectors 
- Allowing for industry fixed 
Data: 
- Data on 21 industries of 
Colombia 
- Source(s): Colombian 
National Planning 
Department 
Methodology: 
290 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
effects reverses the result: trade 
protection increases relative 
wages 
- Because tariff reductions were 
proportionately larger in sectors 
employing a high fraction of 
less-skilled workers, the 
decrease in the wage premiums 
in these sectors affected such 
workers disproportionately 
- 2SLS 
- OLS 
Huang (2009) H-1: Political structure of a 
country has a substantial 
influence on policy change in 
financial sector 
H-2: Policy change in a 
country is positively correlated 
with the initial level of 
liberalization 
- Level of financial 
liberalization 
- Balance of payments crisis 
- Banking crisis 
- Recession 
- High inflation 
- Drastic political change 
- Political orientation of ruling 
party 
- IMF program 
- Democracy 
 
- Policy change in a country is 
negatively rather than positively 
associated with the initial extent 
of liberalization level, and the 
distance behind the regional 
leader 
- Countries with highly 
repressed financial sectors have 
more potential to embark on 
reform, while countries with a 
highly liberalized financial 
sector have greater status quo 
bias 
Data: 
- 35 countries for the 
period 1973–1996 
- Source(s): IMF, World 
Bank, Polity IV project  
Methodology: 
- Common correlated 
effect pooled (CCEP) 
modelling 
291 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Economic and political 
structure and ideology can have 
a substantial influence on policy 
change, and the extent of 
democracy has a significantly 
negative effect on policy reform 
Ickes and Ofer (2006) H: Changes in the industrial 
structure of employment 
across Russian regions are 
mainly determined by legacy 
factors, political factors, and 
success factors 
- Structural change in 
industry 
- The natural resource potential 
- The initial employment share 
- The rate of urbanization 
- The specialization of industry 
- Average January (1997) 
temperature 
- Change in population 
- Change in the number of 
employed 
- Change in the number of small 
enterprises 
- Change in gross regional 
product per capita 
- FDI per 1000 employed 
- Change in the crime rate 
- Democracy index 
- Legislative quality 
- Political environment 
- Social environment 
- Initial conditions such as 
natural resource potential, 
climate, and industrial 
specialization explain more of 
the variation in industrial 
restructuring than political 
variables 
Data: 
- Data on various 
industrial sectors of 
Russia during 1990s 
- Source(s): CEFIR 
database, RSS, Russian 
Statistical Office, World 
Bank 
Methodology: 
- OLS 
292 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Kim and Pirttilä 
(2006) 
H: Both ex post and ex ante 
political constraints are 
instrumental in determining 
the extent of progress in 
welfare-enhancing reforms 
- Liberalization index - Support for reforms 
- Inflation rate 
- Unemployment rate 
- GDP growth 
- Gini coefficient 
- Government’s budget balance 
- Capital formation 
- Future loss 
- Index of political freedom 
- Progress in reform is positively 
associated with public support 
for reforms, which is affected by 
income inequality and expected 
individual performance during 
future reforms 
- Reform sequencing should 
start with a reform that is both 
popular and stimulatory to other 
reforms 
Data: 
- 14 transition countries 
for 1990-97 period 
- Source(s): EBRD, 
United Nations 
University, World 
Institute for Development 
Economics Research 
Methodology: 
- Generalized Method of 
Moments (GMM) 
- 2SLS 
- Static fixed effects 
- Dynamic fixed effects 
Li and Xu (2002) H-1: Countries with a larger 
financial sector, a higher urban 
population, and a lower 
income inequality are more 
likely to privatize and 
liberalize 
H-2: A higher government 
budget deficit makes 
privatization and liberalization 
- Non-state ownership of 
telecommunications 
sector 
- Urban/total population 
- Gini coefficient 
- Financial depth 
- Deficit/GDP 
- Profitability 
- Ideology 
- World Bank project 
- Democracy 
- Party polarization 
- Countries with stronger pro-
reform interest groups, namely 
the financial services sector and 
the urban consumers, are more 
likely to reform in more 
democratic countries 
- Less democratic countries are 
more likely to maintain the 
public sector monopoly when 
Data: 
- 50 countries over the 
period from 1990 to 1998 
- Source(s): World Bank, 
Gurr (1999) 
Methodology: 
- Fixed/random effects 
models 
- OLS 
293 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
less likely, while a larger 
government debt has the 
opposite implications 
H-3: Countries with a right-of-
centre government and 
countries that receive World 
Bank assistance in the 
telecommunications sector are 
more likely to privatize and 
liberalize 
- Number of veto players 
- The number of main lines per 
100 inhabitants 
- Real GDP per capita 
- Illiteracy rate 
- The ratio of manufacturing 
value added over GDP 
- The share of population in the 
largest city 
- The share of government debt 
in GDP 
the government benefits more 
from such a governance mode 
- Democracy affects the pace of 
reforms by magnifying the 
voices of interest groups in more 
democratic countries and by 
moderating politicians’ 
discretion in less democratic 
countries 
Olper (2007) H-1: Agricultural protection is 
influenced directly by land 
inequality and ideology 
H-2: The effect of land 
inequality is conditional to the 
ideological orientation of the 
government 
- Aggregated producer 
subsidy equivalent 
- Land inequality (land gini) 
- Ideological orientation of the 
government 
- Amount of agricultural land 
per capita 
- Share of agricultural export to 
total export 
- Agricultural share in 
employment and in GDP 
- Gastil index of political rights 
- Index of quality of institutions 
- Protection is decreasing in land 
inequality and with left-wing 
government orientation, but not 
in a linear fashion: left-wing 
governments tend to support 
agriculture in more unequal 
societies 
- The relationship holds better in 
democracies than in 
dictatorships 
 
Data: 
- 40 countries for 1982-
2000 period 
- Source(s): IFAD, 
Keefer and Knack (1995), 
FAO, Database on 
Political Institutions, 
OECD, USDA, World 
Bank, Freedom House, 
International Country 
Risk Guide 
Methodology: 
- OLS 
294 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
Volscho (2007) H: Quintile shares of size-
adjusted family income are 
impacted by union density and 
federal, state, and local 
government employment 
- Family income - Union density 
- Federal government 
employment 
- State government employment 
- Local government employment 
- Mean establishment size 
- Manufacturing employment 
- Unemployment rate 
- Female labour force 
participation 
- Female-headed families 
- Dispersion in education 
- Dispersion in age 
- Union density has a 
progressive effect that benefits 
middle and upper-middle 
income families 
- Federal government 
employment has a strong 
progressive effect on the entire 
income distribution 
- State government employment 
has a progressive effect on 
middle and upper-middle 
income families 
- Local government employment 
mainly impacts families in the 
bottom forty percent of the 
income distribution 
Data: 
- Data on 160 US 
metropolitan statistical 
areas from the 2000 
census 
- Source(s): USA 2000 
Census Data 
Methodology: 
- Seemingly unrelated 
regression estimation 
(SURE) 
Wagner et al. (2009) H: Institutional factors affect 
satisfaction with democracy 
- Average yearly 
satisfaction with 
democracy 
- Inflation 
- GDP per capita 
- Growth in GDP 
- Unemployment rate 
- BERI composite index 
- Quality of monetary policy 
- Regulatory quality 
- High-quality institutions like 
the rule of law, well-functioning 
regulation, low corruption, and 
other institutions that improve 
resource allocation have a 
positive effect on average 
satisfaction with democracy 
Data: 
- A panel of observations 
from Eurobarometer in 
the time span 1990–2000 
- Source(s): Business 
Environment Risk 
Intelligence (BERI), 
295 
 
Study Hypothesis (H) 
Dependent 
Variable(s) 
Explanatory 
Variable(s) 
Result(s) Data & Methodology 
- Rule of law 
- Control of corruption 
- Size of the shadow economy 
- Checks and balances 
- Left/right placement 
- Inequality 
Database of political 
institutions (DPI), 
Eurobarometer 
Methodology: 
- Random effects panel 
regressions 
  
296 
 
Appendix III-2: Summary of previous econometric studies based on political economy by their focus 
 
Focus of the study Major Variable(s) Primary Data Sources Examples 
Political economy of 
liberalization in electricity 
industry 
- Regulation indicator in power industry 
- Government ideology 
- Government fragmentation 
- Number of years that the incumbent 
government has been in office 
- Index of political constraints 
- Globalization index 
- Energy demand 
- Real GDP per capita 
- Per capita generation capacity 
- Debt payments as a proportion of national 
income 
- Industry value added as proportion of GDP 
- World Bank 
- US Energy Information Agency 
- BP 
Chang and Berdiev (2011), Cubbin and 
Stern (2006) 
Political economy of 
liberalization in 
telecommunications industry 
- Degree of liberalization 
- Share of incumbent operator 
- Regulatory independence 
- Government’s programmatic position 
- Share of population aged between 15-64 years 
- Mainline coverage & cellular subscription 
- Mainlines per employee 
- Price of fixed-line, cellular services 
- Regulatory governance index 
- OECD regulation database 
- World Bank 
Duso and Seldeslachts (2010), Gasmi et 
al. (2009), Gasmi and Recuero Virto 
(2010), Li and Xu (2002) 
297 
 
Focus of the study Major Variable(s) Primary Data Sources Examples 
- Corruption 
- Bureaucracy 
- Law and order 
- Expropriation 
- Currency risk 
- Institutional environment index 
- Checks and balances 
- Privatization 
- Competition in fixed and cellular 
- Democracy index 
- Total debt service 
- Aid per capita 
- Ethno linguistic fractionalization 
- Free press 
- Ownership of telecommunications sector 
- Urban/total population 
- Gini coefficient 
- Financial depth 
- Deficit/GDP 
- World Bank project 
- Real GDP per capita 
Political economy of economic 
reforms in non-infrastructure 
industries and other areas 
- Deficit/GDP ratio 
- Inflation rate 
- Number of executive constraints 
- Polity IV project 
- World Bank's Database of Political 
Institutions 
Alesina et al. (2006), Boschini (2006), 
Dreher et al. (2009), Duval (2008), 
Fredriksson and Wollscheid (2008), 
298 
 
Focus of the study Major Variable(s) Primary Data Sources Examples 
- Election year 
- Political orientation of the ruling government 
- Assembly or parliamentary system 
- Executive control of absolute majority 
- Number of years left in the current term for 
the executive 
- Total government deficit as a share of GDP 
and inflation 
- The real per capita GDP  
- The ratio of exports and imports to GDP 
- Crisis years 
- Countries’ participation to IMF programs 
- Industrialisation index 
- Index of the favourableness of attitudes 
toward entrepreneurship 
- Index of concentration in landholdings 
- Size of government index 
- Legal structure and security of property rights 
index 
- Regulation of credit, labour and business 
index 
- Profession of heads of governments 
- Education of heads of governments 
- Aid 
- Degree of sustainability of public debt 
- IMF's International Financial Statistics 
(IFS) database 
- Penn World Table 
- Freedom House 
- OECD 
- National Statistical Offices 
- EBRD 
- United Nations University 
- World Institute for Development Economics 
Research 
- International Country Risk Guide 
- Business Environment Risk Intelligence 
(BERI) 
- Eurobarometer 
Goldberg and Pavcnik (2005), Huang 
(2009), Ickes and Ofer (2006), Kim and 
Pirttilä (2006), Olper (2007), Volscho 
(2007), Wagner et al. (2009) 
299 
 
Focus of the study Major Variable(s) Primary Data Sources Examples 
- The degree to which business transactions 
involve corruption 
- The perception of the quality of public service 
- Political instability 
- Level of financial liberalization 
- The rate of urbanization 
- Support for reforms 
- Gini coefficient 
- Satisfaction with democracy 
 
  
300 
 
Appendix III-3: Estimation results for Paper Three 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
1.1.1 Entry barriers (0-6) Industry value added (% of GDP) 0.129 0.027 4.720 0.00 28 456 499.40 0.0000 - - Fixed Effects 
 
(OECD countries) Rural population (% of total population) 0.091 0.058 1.580 0.12 
       
  
Gini coefficient (0-100) -0.022 0.023 -0.930 0.35 
       
  
Polity score (-10,+10) 0.178 0.046 3.910 0.00 
       
  
EU member (0-1) -1.610 0.319 -5.040 0.00 
       
  
OECD member (0-1) 1.717 0.419 4.100 0.00 
       
  
Existence of electricity market reform idea (0-1) 1.078 0.274 3.930 0.00 
       
  
Log of population density 0.427 2.552 0.170 0.87 
       
  
Log of electricity consumption per capita (MWh) 2.566 0.796 3.220 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -5.201 0.536 -9.700 0.00 
       
  
Average number of years of adult (25+) education -0.282 0.182 -1.550 0.12 
       
  
Log of imports of goods and services (% of GDP) -2.415 0.582 -4.150 0.00 
       
  
Constant 14.124 11.049 1.280 0.20 
       1.1.2 Public Ownership (0-6) Industry value added (% of GDP) 0.076 0.015 5.200 0.00 28 456 72.18 0.0000 - - Fixed Effects 
 
(OECD countries) Rural population (% of total population) -0.015 0.031 -0.500 0.62 
       
  
Gini coefficient (0-100) -0.012 0.012 -0.980 0.33 
       
  
Polity score (-10,+10) -0.013 0.024 -0.550 0.58 
       
  
EU member (0-1) 0.235 0.170 1.380 0.17 
       
  
OECD member (0-1) 0.043 0.223 0.190 0.85 
       
  
Existence of electricity market reform idea (0-1) 0.107 0.146 0.730 0.46 
       
  
Log of population density 9.221 1.357 6.790 0.00 
       
  
Log of electricity consumption per capita (MWh) -1.188 0.423 -2.810 0.01 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -1.157 0.285 -4.060 0.00 
       
  
Average number of years of adult (25+) education -0.120 0.097 -1.240 0.22 
       
  
Log of imports of goods and services (% of GDP) -0.936 0.310 -3.020 0.00 
       
  
Constant -25.833 5.876 -4.400 0.00 
       
301 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
1.1.3 Vertical integration (0-6) Industry value added (% of GDP) 0.128 0.025 5.080 0.00 28 456 60.18 0.0000 - - Fixed Effects 
 
(OECD countries) Rural population (% of total population) -0.084 0.053 -1.570 0.12 
       
  
Gini coefficient (0-100) 0.005 0.021 0.210 0.83 
       
  
Polity score (-10,+10) 0.109 0.042 2.590 0.01 
       
  
EU member (0-1) -1.407 0.294 -4.780 0.00 
       
  
OECD member (0-1) 0.907 0.386 2.350 0.02 
       
  
Existence of electricity market reform idea (0-1) 0.521 0.252 2.060 0.04 
       
  
Log of population density -0.187 2.351 -0.080 0.94 
       
  
Log of electricity consumption per capita (MWh) 0.266 0.733 0.360 0.72 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -3.679 0.494 -7.450 0.00 
       
  
Average number of years of adult (25+) education -0.076 0.168 -0.450 0.65 
       
  
Log of imports of goods and services (% of GDP) -1.444 0.536 -2.690 0.01 
       
  
Constant 17.055 10.177 1.680 0.10 
       1.2.1 Overall indicator (0-6) Industry value added (% of GDP) 0.111 0.018 6.180 0.00 28 456 74.78 0.0000 - - Fixed Effects 
 
(OECD countries) Rural population (% of total population) -0.003 0.038 -0.070 0.94 
       
  
Gini coefficient (0-100) -0.010 0.015 -0.640 0.52 
       
  
Polity score (-10,+10) 0.091 0.030 3.050 0.00 
       
  
EU member (0-1) -0.927 0.210 -4.430 0.00 
       
  
OECD member (0-1) 0.889 0.275 3.240 0.00 
       
  
Existence of electricity market reform idea (0-1) 0.569 0.180 3.160 0.00 
       
  
Log of population density 3.153 1.675 1.880 0.06 
       
  
Log of electricity consumption per capita (MWh) 0.548 0.522 1.050 0.30 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -3.345 0.352 -9.510 0.00 
       
  
Average number of years of adult (25+) education -0.160 0.120 -1.330 0.18 
       
  
Log of imports of goods and services (% of GDP) -1.598 0.382 -4.180 0.00 
       
  
Constant 1.782 7.251 0.250 0.81 
       1.2.2 Overall indicator (0-6) Industry value added (% of GDP) 0.057 0.012 4.550 0.00 17 150 80.09 0.0000 - - Fixed Effects 
 
(Non-OECD countries) Rural population (% of total population) -0.157 0.082 -1.910 0.06 
       
  
Gini coefficient (0-100) -0.023 0.015 -1.470 0.14 
       
  
Polity score (-10,+10) -0.033 0.025 -1.300 0.20 
       
  
EU member (0-1) 0.463 0.250 1.850 0.07 
       
  
OECD member (0-1) (omitted)       
       
  
Existence of electricity market reform idea (0-1) (omitted)       
       
302 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
  
Log of population density 2.136 2.667 0.800 0.43 
       
  
Log of electricity consumption per capita (MWh) 4.001 0.568 7.050 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -3.009 0.351 -8.580 0.00 
       
  
Average number of years of adult (25+) education -0.246 0.296 -0.830 0.41 
       
  
Log of imports of goods and services (% of GDP) 0.038 0.329 0.110 0.91 
       
  
Constant 3.938 11.972 0.330 0.74 
       2.1.1 Entry barriers (0-6) Net official assistance and aid received -0.628 0.311 -2.020 0.04 30 764 83.97 0.0000 - - Fixed Effects 
 
(OECD countries) EU member (0-1) -1.060 0.234 -4.530 0.00 
       
  
OECD member (0-1) 2.136 0.287 7.430 0.00 
       
  
Existence of electricity market reform idea (0-1) 1.125 0.178 6.340 0.00 
       
  
Log of population density 0.131 1.431 0.090 0.93 
       
  
Log of electricity consumption per capita (MWh) 2.984 0.408 7.310 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -5.987 0.347 -17.280 0.00 
       
  
Average number of years of adult (25+) education -0.226 0.103 -2.200 0.03 
       
  
Log of imports of goods and services (% of GDP) -2.491 0.399 -6.250 0.00 
       
  
Constant 23.500 5.198 4.520 0.00 
       2.1.2 Public Ownership (0-6) Net official assistance and aid received -0.120 0.159 -0.760 0.45 30 764 123.73 0.0000 - - Fixed Effects 
 
(OECD countries) EU member (0-1) 0.227 0.119 1.910 0.06 
       
  
OECD member (0-1) -0.371 0.147 -2.530 0.01 
       
  
Existence of electricity market reform idea (0-1) -0.024 0.091 -0.270 0.79 
       
  
Log of population density 7.314 0.730 10.020 0.00 
       
  
Log of electricity consumption per capita (MWh) -0.354 0.208 -1.700 0.09 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -1.273 0.177 -7.200 0.00 
       
  
Average number of years of adult (25+) education -0.226 0.052 -4.310 0.00 
       
  
Log of imports of goods and services (% of GDP) -0.621 0.203 -3.050 0.00 
       
  
Constant -16.994 2.651 -6.410 0.00 
       2.1.3 Vertical integration (0-6) Net official assistance and aid received -0.252 0.284 -0.890 0.37 30 764 19.84 0.0189 - - Fixed Effects 
 
(OECD countries) EU member (0-1) -1.171 0.214 -5.480 0.00 
       
  
OECD member (0-1) 1.125 0.262 4.290 0.00 
       
  
Existence of electricity market reform idea (0-1) 0.626 0.162 3.860 0.00 
       
  
Log of population density 2.843 1.307 2.180 0.03 
       
  
Log of electricity consumption per capita (MWh) 1.297 0.373 3.480 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -4.536 0.316 -14.330 0.00 
       
303 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
  
Average number of years of adult (25+) education 0.108 0.094 1.160 0.25 
       
  
Log of imports of goods and services (% of GDP) -2.002 0.364 -5.500 0.00 
       
  
Constant 7.825 4.747 1.650 0.10 
       2.2.1 Overall indicator (0-6) Net official assistance and aid received -0.334 0.202 -1.650 0.10 30 764 855.87 0.0000 - - Fixed Effects 
 
(OECD countries) EU member (0-1) -0.668 0.152 -4.380 0.00 
       
  
OECD member (0-1) 0.964 0.187 5.140 0.00 
       
  
Existence of electricity market reform idea (0-1) 0.576 0.116 4.980 0.00 
       
  
Log of population density 3.429 0.933 3.680 0.00 
       
  
Log of electricity consumption per capita (MWh) 1.309 0.266 4.920 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -3.932 0.226 -17.410 0.00 
       
  
Average number of years of adult (25+) education -0.115 0.067 -1.710 0.09 
       
  
Log of imports of goods and services (% of GDP) -1.705 0.260 -6.560 0.00 
       
  
Constant 4.777 3.387 1.410 0.16 
       2.2.2 Overall indicator (0-6) Net official assistance and aid received -0.557 0.252 -2.210 0.03 17 271 140.97 0.0000 - - Fixed Effects 
 
(Non-OECD countries) EU member (0-1) 0.480 0.398 1.200 0.23 
       
  
OECD member (0-1) (omitted)       
       
  
Existence of electricity market reform idea (0-1) (omitted)       
       
  
Log of population density 1.823 1.354 1.350 0.18 
       
  
Log of electricity consumption per capita (MWh) 2.208 0.309 7.150 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -1.898 0.219 -8.650 0.00 
       
  
Average number of years of adult (25+) education -1.353 0.209 -6.480 0.00 
       
  
Log of imports of goods and services (% of GDP) -0.104 0.204 -0.510 0.61 
       
  
Constant 11.222 6.000 1.870 0.06 
       3.1.1 Entry barriers (0-6) Single-party government (0-1) 0.053 0.162 0.330 0.74 29 710 3.40 1.0000 326.16 0.0000 Random Effects 
 
(OECD countries) The years the chief executive has been in office 0.042 0.018 2.360 0.02 
       
  
Economic policy orientation of ruling party: Right -0.306 0.242 -1.270 0.21 
       
  
Economic policy orientation of ruling party: Left -0.423 0.246 -1.720 0.09 
       
  
Economic policy orientation of ruling party: Centre (omitted)       
       
  
Parliamentary regimes (0-1) 0.538 0.361 1.490 0.14 
       
  
Prof. bgr. of chief executive: Entrepreneur -0.326 0.309 -1.060 0.29 
       
  
Prof. bgr. of chief executive: Economics 1.333 0.389 3.430 0.00 
       
  
Prof. bgr. of chief executive: Military 0.344 0.415 0.830 0.41 
       
  
Prof. bgr. of chief executive: Politician 0.482 0.213 2.260 0.02 
       
304 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
  
Prof. bgr. of chief executive: Scientist, Other 0.446 0.243 1.840 0.07 
       
  
Prof. bgr. of chief executive: Unknown/other 0.516 0.227 2.270 0.02 
       
  
Educ. bgr. of chief executive: Economics -0.085 0.556 -0.150 0.88 
       
  
Educ. bgr. of chief executive: Natural science -0.235 0.640 -0.370 0.71 
       
  
Educ. bgr. of chief executive: Other university -0.519 0.579 -0.900 0.37 
       
  
Educ. bgr. of chief executive: Unknown/other -0.042 0.605 -0.070 0.94 
       
  
EU member (0-1) -0.829 0.233 -3.560 0.00 
       
  
OECD member (0-1) 1.697 0.327 5.190 0.00 
       
  
Existence of electricity market reform idea (0-1) 0.749 0.181 4.140 0.00 
       
  
Log of population density 0.606 0.140 4.340 0.00 
       
  
Log of electricity consumption per capita (MWh) 2.886 0.318 9.070 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -5.730 0.321 -17.860 0.00 
       
  
Average number of years of adult (25+) education -0.240 0.085 -2.830 0.01 
       
  
Log of imports of goods and services (% of GDP) -1.202 0.272 -4.420 0.00 
       
  
Constant 16.661 1.214 13.730 0.00 
       3.1.2 Public Ownership (0-6) Single-party government (0-1) -0.144 0.085 -1.690 0.09 29 710 14.85 0.8687 2929.78 0.0000 Random Effects 
 
(OECD countries) The years the chief executive has been in office 0.033 0.009 3.660 0.00 
       
  
Economic policy orientation of ruling party: Right 0.074 0.129 0.570 0.57 
       
  
Economic policy orientation of ruling party: Left -0.203 0.131 -1.550 0.12 
       
  
Economic policy orientation of ruling party: Centre (omitted)       
       
  
Parliamentary regimes (0-1) -0.407 0.231 -1.760 0.08 
       
  
Prof. bgr. of chief executive: Entrepreneur -0.457 0.161 -2.840 0.00 
       
  
Prof. bgr. of chief executive: Economics 0.056 0.204 0.270 0.79 
       
  
Prof. bgr. of chief executive: Military 0.103 0.230 0.450 0.65 
       
  
Prof. bgr. of chief executive: Politician -0.201 0.117 -1.720 0.09 
       
  
Prof. bgr. of chief executive: Scientist, Other -0.484 0.132 -3.670 0.00 
       
  
Prof. bgr. of chief executive: Unknown/other -0.302 0.124 -2.430 0.02 
       
  
Educ. bgr. of chief executive: Economics 0.163 0.285 0.570 0.57 
       
  
Educ. bgr. of chief executive: Natural science 1.123 0.330 3.410 0.00 
       
  
Educ. bgr. of chief executive: Other university 0.126 0.295 0.430 0.67 
       
  
Educ. bgr. of chief executive: Unknown/other 0.120 0.313 0.380 0.70 
       
  
EU member (0-1) -0.016 0.127 -0.130 0.90 
       
  
OECD member (0-1) -0.474 0.181 -2.610 0.01 
       
305 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
  
Existence of electricity market reform idea (0-1) 0.114 0.097 1.180 0.24 
       
  
Log of population density -0.118 0.181 -0.650 0.51 
       
  
Log of electricity consumption per capita (MWh) 0.232 0.228 1.020 0.31 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -0.638 0.187 -3.410 0.00 
       
  
Average number of years of adult (25+) education -0.333 0.055 -6.050 0.00 
       
  
Log of imports of goods and services (% of GDP) 0.196 0.191 1.030 0.31 
       
  
Constant 9.582 0.972 9.860 0.00 
       3.1.3 Vertical integration (0-6) Single-party government (0-1) -0.229 0.139 -1.650 0.10 29 710 16.93 0.7153 1075.26 0.0000 Random Effects 
 
(OECD countries) The years the chief executive has been in office 0.044 0.015 2.970 0.00 
       
  
Economic policy orientation of ruling party: Right -0.526 0.211 -2.500 0.01 
       
  
Economic policy orientation of ruling party: Left -0.380 0.214 -1.770 0.08 
       
  
Economic policy orientation of ruling party: Centre (omitted)       
       
  
Parliamentary regimes (0-1) 0.078 0.365 0.210 0.83 
       
  
Prof. bgr. of chief executive: Entrepreneur -0.591 0.264 -2.240 0.03 
       
  
Prof. bgr. of chief executive: Economics 1.982 0.335 5.920 0.00 
       
  
Prof. bgr. of chief executive: Military -0.173 0.373 -0.460 0.64 
       
  
Prof. bgr. of chief executive: Politician 0.443 0.191 2.330 0.02 
       
  
Prof. bgr. of chief executive: Scientist, Other -0.043 0.215 -0.200 0.84 
       
  
Prof. bgr. of chief executive: Unknown/other 0.725 0.202 3.590 0.00 
       
  
Educ. bgr. of chief executive: Economics 0.814 0.468 1.740 0.08 
       
  
Educ. bgr. of chief executive: Natural science 1.750 0.541 3.230 0.00 
       
  
Educ. bgr. of chief executive: Other university 0.659 0.486 1.360 0.18 
       
  
Educ. bgr. of chief executive: Unknown/other 0.507 0.513 0.990 0.32 
       
  
EU member (0-1) -1.282 0.206 -6.220 0.00 
       
  
OECD member (0-1) 0.966 0.293 3.300 0.00 
       
  
Existence of electricity market reform idea (0-1) 0.384 0.158 2.430 0.02 
       
  
Log of population density 0.470 0.218 2.150 0.03 
       
  
Log of electricity consumption per capita (MWh) 1.778 0.349 5.090 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -4.266 0.299 -14.280 0.00 
       
  
Average number of years of adult (25+) education 0.047 0.087 0.540 0.59 
       
  
Log of imports of goods and services (% of GDP) -1.484 0.296 -5.010 0.00 
       
  
Constant 14.172 1.352 10.480 0.00 
       
306 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
3.2.1 Overall indicator (0-6) Single-party government (0-1) -0.113 0.103 -1.090 0.28 29 710 14.82 0.8698 988.55 0.0000 Random Effects 
 
(OECD countries) The years the chief executive has been in office 0.038 0.011 3.400 0.00 
       
  
Economic policy orientation of ruling party: Right -0.191 0.156 -1.230 0.22 
       
  
Economic policy orientation of ruling party: Left -0.268 0.159 -1.690 0.09 
       
  
Economic policy orientation of ruling party: Centre (omitted)       
       
  
Parliamentary regimes (0-1) 0.118 0.265 0.450 0.66 
       
  
Prof. bgr. of chief executive: Entrepreneur -0.431 0.196 -2.200 0.03 
       
  
Prof. bgr. of chief executive: Economics 1.195 0.248 4.810 0.00 
       
  
Prof. bgr. of chief executive: Military 0.111 0.275 0.400 0.69 
       
  
Prof. bgr. of chief executive: Politician 0.262 0.141 1.860 0.06 
       
  
Prof. bgr. of chief executive: Scientist, Other 0.021 0.159 0.130 0.90 
       
  
Prof. bgr. of chief executive: Unknown/other 0.342 0.149 2.290 0.02 
       
  
Educ. bgr. of chief executive: Economics 0.328 0.348 0.940 0.35 
       
  
Educ. bgr. of chief executive: Natural science 0.948 0.402 2.360 0.02 
       
  
Educ. bgr. of chief executive: Other university 0.130 0.362 0.360 0.72 
       
  
Educ. bgr. of chief executive: Unknown/other 0.268 0.381 0.700 0.48 
       
  
EU member (0-1) -0.752 0.152 -4.940 0.00 
       
  
OECD member (0-1) 0.791 0.215 3.670 0.00 
       
  
Existence of electricity market reform idea (0-1) 0.465 0.117 3.980 0.00 
       
  
Log of population density 0.272 0.141 1.920 0.05 
       
  
Log of electricity consumption per capita (MWh) 1.696 0.249 6.830 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -3.628 0.218 -16.610 0.00 
       
  
Average number of years of adult (25+) education -0.169 0.063 -2.700 0.01 
       
  
Log of imports of goods and services (% of GDP) -0.954 0.212 -4.510 0.00 
           Constant 13.918 0.940 14.810 0.00               
3.2.2 Overall indicator (0-6) Single-party government (0-1) 0.137 0.156 0.880 0.38 14 194 142.00 0.0000 - - Fixed Effects 
 
(Non-OECD countries) The years the chief executive has been in office -0.002 0.023 -0.100 0.92 
       
  
Economic policy orientation of ruling party: Right (omitted)       
       
  
Economic policy orientation of ruling party: Left -0.041 0.213 -0.190 0.85 
       
  
Economic policy orientation of ruling party: Centre 0.002 0.267 0.010 1.00 
       
  
Parliamentary regimes (0-1) -0.504 0.567 -0.890 0.38 
       
  
Prof. bgr. of chief executive: Entrepreneur 0.276 0.363 0.760 0.45 
       
  
Prof. bgr. of chief executive: Economics -0.418 0.336 -1.240 0.22 
       
307 
 
Models Dependent variable Explanatory variables Coef. Std. Err. t-stat. p value 
# of # of Hausman Test BPLM Test Preferred 
countries observations Statistic p-value Statistic p-value Specification 
  
Prof. bgr. of chief executive: Military 0.080 0.414 0.190 0.85 
       
  
Prof. bgr. of chief executive: Politician -0.341 0.369 -0.920 0.36 
       
  
Prof. bgr. of chief executive: Scientist, Other 0.139 0.343 0.400 0.69 
       
  
Prof. bgr. of chief executive: Unknown/other -0.028 0.238 -0.120 0.91 
       
  
Educ. bgr. of chief executive: Economics -0.234 0.540 -0.430 0.67 
       
  
Educ. bgr. of chief executive: Natural science -0.273 0.479 -0.570 0.57 
       
  
Educ. bgr. of chief executive: Other university -0.357 0.505 -0.710 0.48 
       
  
Educ. bgr. of chief executive: Unknown/other -0.174 0.792 -0.220 0.83 
       
  
EU member (0-1) 0.515 0.393 1.310 0.19 
       
  
OECD member (0-1) (omitted)       
       
  
Existence of electricity market reform idea (0-1) (omitted)       
       
  
Log of population density 1.379 2.110 0.650 0.51 
       
  
Log of electricity consumption per capita (MWh) 2.266 0.430 5.280 0.00 
       
  
Log of GDP per capita (PPP, cur. thousand int. $) -1.245 0.370 -3.370 0.00 
       
  
Average number of years of adult (25+) education -1.613 0.292 -5.530 0.00 
       
  
Log of imports of goods and services (% of GDP) -0.098 0.228 -0.430 0.67 
           Constant 14.494 10.103 1.430 0.15               
 
  
308 
 
Appendix III-4: Correlation among explanatory and control variables 
 
Correlation among explanatory variables 
Variables ind_va aid hg_por hg_pol hg_poc hg_prep hg_prse hg_prmt hg_prpo hg_prso hg_prot hg_edec hg_edns hg_edou hg_edot 
ind_va 1 
              
aid -0.0178 1 
             
hg_por -0.0165 -0.0972 1 
            
hg_pol 0.0215 0.0223 -0.7387 1 
           
hg_poc -0.007 0.1033 -0.3572 -0.3657 1 
          
hg_prep -0.0613 0.0431 0.0702 -0.124 0.0749 1 
         
hg_prse -0.0423 0.0387 0.0612 0.0041 -0.0901 -0.0488 1 
        
hg_prmt 0.1085 0.0123 0.0529 -0.0203 -0.0449 -0.0498 -0.0469 1 
       
hg_prpo 0.0484 0.1557 -0.036 -0.0673 0.143 0.0065 -0.1628 -0.1528 1 
      
hg_prso 0.0423 0.0482 0.0124 -0.0415 0.0405 -0.0048 -0.1388 -0.1416 -0.1267 1 
     
hg_prot 0.0119 0.0575 -0.0847 0.0431 0.0572 0.0345 -0.1325 0.0483 -0.1914 -0.3846 1 
    
hg_edec -0.0497 0.0211 -0.0111 0.1632 -0.2108 -0.011 0.3683 -0.0499 -0.1227 -0.2793 0.1499 1 
   
hg_edns 0.0414 0.1493 -0.0305 -0.0115 0.058 0.1002 -0.0935 -0.0955 0.1515 0.1012 0.1263 -0.1492 1 
  
hg_edou -0.089 -0.1408 0.0956 -0.1257 0.0421 -0.0949 -0.2044 -0.057 0.0806 0.2776 -0.2642 -0.5106 -0.3633 1 
 
hg_edot 0.1349 0.0391 -0.0654 -0.0096 0.1036 0.0779 -0.0889 0.2184 -0.0357 -0.1977 0.0289 -0.2414 -0.1809 -0.3955 1 
 
  
309 
 
Correlation among control variables 
Variables hg_sp cexyrs hg_epr rural gini polity eu oecd refid pop_den elcon gdppc educ imp 
hg_sp 1 
             
cexyrs 0.0013 1 
            
hg_epr 0.1981 0.1049 1 
           
rural -0.0487 0.045 0.2544 1 
          
gini 0.3222 -0.0499 0.5225 0.1435 1 
         
polity -0.1737 -0.2988 -0.5602 -0.4218 -0.3023 1 
        
eu -0.1189 0.1124 -0.4147 -0.2192 -0.1712 0.3276 1 
       
oecd 0.0351 0.0484 -0.4485 -0.5038 -0.1968 0.5693 0.4615 1 
      
refid -0.1959 0.0342 0.0671 0.1708 0.1022 -0.048 0.0391 -0.2258 1 
     
pop_den -0.0679 0.0147 -0.0929 -0.1159 -0.1111 0.0786 0.3636 0.1497 -0.0241 1 
    
elcon -0.0409 -0.0328 -0.2662 -0.3852 -0.3395 0.3344 -0.072 0.4577 -0.0861 -0.3002 1 
   
gdppc -0.1344 0.0459 -0.4125 -0.4661 -0.2438 0.499 0.4853 0.6403 0.2057 0.1242 0.5746 1 
  
educ -0.1098 -0.0709 -0.0087 -0.3103 -0.1212 0.1908 -0.2079 0.1256 0.1341 -0.0952 0.4052 0.3607 1 
 
imp -0.3022 -0.085 -0.0529 0.3117 -0.1146 -0.1076 0.0347 -0.398 0.2892 0.0572 -0.2736 -0.1582 0.1804 1 
 
 
310 
 
Abbreviations 
Variables Abbreviation 
Explanatory Variables  
Industry value added (% of GDP) ind_va 
Net official development assistance and official aid received (current billion US$) aid 
Party orientation with respect to economic policy  
     - Right hg_por 
     - Left hg_pol 
     - Center hg_poc 
Professional background of chief executive  
     - Entrepreneur hg_prep 
     - Scientist (Economist) hg_prse 
     - Military hg_prmt 
     - Politician hg_prpo 
     - Scientist (Other) hg_prso 
     - Unknown/other hg_prot 
Educational background of chief executive  
     - Economics hg_edec 
     - Natural science hg_edns 
     - Other university hg_edou 
     - Unknown/other hg_edot 
Control Variables  
Party Structure (1: single-party, 0: coalition) hg_sp 
The years the chief executive has been in office cexyrs 
Electoral system (parliamentary regimes) hg_epr 
Rural population (% of total population) rural 
Gini coefficient (0-100) gini 
Polity score (-10,+10) polity 
EU member (0-1) eu 
OECD member (0-1) oecd 
Existence of electricity market reform idea (0-1) refid 
Population density (people per sq. km of land area) pop_den 
Electricity consumption (MWh per capita) elcon 
GDP per capita (PPP, current thousand int. $) gdppc 
Average number of years of education received by people ages 15 and older educ 
Imports of goods and services (% of GDP) imp