Inequality, Privatization and Democratic Institutions in Developing Countries

Transcription

1 Inequality, Privatization and Democratic Institutions in Developing Countries Lidia Ceriani Simona Scabrosetti Francesco Scervini August 22, 2016 Abstract According to the existing theoretical literature there are several channels through which privatization of State-owned enterprises and assets may shape the distribution of income either increasing or decreasing the level of inequality. As a consequence, the net effect becomes an empirical matter of relative magnitudes. This paper is a first attempt to empirically investigate the relationship between privatization and income inequality adopting a new approach that focuses on the role of democratic institutions in developing countries. Using an unbalanced panel of 80 countries in the period , we find that privatization is negatively and significantly correlated with inequality when democratic institutions are well consolidated, and positively when they are not. Our results, robust to different specifications, measures of democracy and economic controls, suggest an interesting and not yet investigated policy implication for low or middle-income countries: only after having established mature representative political institutions, privatization appears to be related to a reduction in income inequality. Keywords: Inequality, Democracy, Privatization, Developing countries JEL classification: D30, O15, P5 The World Bank, Washington, DC USA. lidia.ceriani@gmail.com Department of Law, University of Pavia, Italy and DONDENA. simona.scabrosetti@unipv.it School of Advanced Studies IUSS Pavia, Italy. francesco.scervini@iusspavia.it 1

2 1 Introduction Starting in Britain in the early 1980s, privatization has then affected both developed and developing countries, although with national differences in terms of relevance, timing and methods. Since 1977, the estimated proceeds arising from divestiture programs reach US$2 trillion (Megginson, 2010). There are different reasons that can help to explain this privatization trend around the world and that have been emphasized by the theoretical literature. More specifically, thanks to divestiture programs governments can (i) reduce the national budget deficits and the stock of national debt, (ii) develop financial markets, and (iii) increase the level of firms efficiency (IMF, 2011). From 1988 to 1993 the value of privatization programs in developing countries represented between one-third and one-half of the world total (Cook and Kirkpatrick, 1997). At the beginning of the 2000s, emerging and developing economies further increased the value of their transactions thus becoming the driving forces in global privatization efforts. Moreover, in these countries proceeds from privatization have been a significant share of gdp, reaching, for instance, about 17% in Bolivia, 2007 (World Bank Privatization Database). 1 In developing countries, the decision to adopt privatization programs has been primarily determined by exogenous factors such as international emulative diffusion (see Brune et al, 2004; Doyle, 2010) and, above all, binding requests from international financial institutions, i.e. IMF and World Bank (see Stallings, 1992). Following Brune et al (2004), during the 1980s and the 1990s, on average, developing countries showed each year outstanding obligations from IMF and World Bank of respectively 3.1 and 9.2 percent of gdp. On one side, these loans are very useful to these countries to finance their development programs. On the other side, they are often related to the credible commitment on the part of receiving countries to implement specific market friendly reforms, generally starting from privatization of State owned enterprises. 2 Therefore, beyond the reasons mentioned above, exogenous international forces play an important role in explaining the adoption of privatization policies in developing countries. 3 However, even if privatization can contribute to improve firms efficiency and financial performance, 4 its distributional impact is relevant too. From a theoretical point of view, this distributional effect seems to be ambiguous. On the one hand, privatization generates a higher revenues flow which could be theoretically used by the government for redistributive aims, that is to reduce income inequality. Indeed, privatization sets free 1 data.worldbank.org/data-catalog/privatization-database 2 See Williamson (1993) for the inclusion of privatization among the policies in the Washington consensus between the US Treasury and the international financial institutions. And see Opper (2004) about the role of IMF and IBRD (International Bank for Reconstruction and Development) loans in explaining the progress in privatization. 3 Notice that public debt in developing countries is mainly hold by public institutions or international organizations and as a consequence the market pressure for debt reduction in these countries is less relevant than in developed economies (World Bank, 2016). 4 This is especially true when privatization is implemented together with deregulation or other increasing competition strategies (see Cavaliere and Scabrosetti, 2008 for a survey of the literature on privatization and efficiency). 2

3 public resources by ceasing costly transfers to inefficient public firms, and it generates higher revenues in the form of privatization proceeds and taxes from the newly higher productive private firms. On the other hand, privatization may directly and indirectly increase inequality. Directly, through capture of privatization proceeds by a powerful entrenched elite. Indirectly, through labour market effects, such as a rise in unemployment levels or an increase in wage differential among differently productive and/or skilled workers. 5 This second negative distributional impact, that makes privatization increasingly unpopular, could be a very relevant issue especially for developing countries in which, due to government or policy failures or to historical reasons, income and wealth are generally more concentrated than in developed countries (Kuznets, 1963). 6 Thus, on one hand privatization allows developing countries to improve their economic performances and in particular to receive loans from international institutions, often essential for starting their development programs. On the other hand, however, at least from a theoretical point of view the relationship between privatization and income distribution is not univocally identified and seems to be unclear. We think that political institutions may play a key role in trying to shed light on this relationship. As stated by Birdsall (1999): The risks of privatization arise because developing and transitional economies, almost by definition, are handicapped by relatively weak institutions, less well-established rules of transparency, and often, not only high concentrations of economic and political power but a high correlation between those two areas of power. On the political side, developing countries have recently experienced a democratic transition, although with some differences across countries due to their history, background, institutional, economic and social characteristics. 7 Democracies, rather than autocracies, tend to be more beneficial for the poor which can vote and express their preferences. Democratization can thus be considered as a commitment device to future redistribution to the poor (Acemoglu and Robinson, 2006). Therefore, the existence of representative political institutions might lead to identify a positive relationship between de-nationalization policies and equality of income in developing countries. In other words, thanks to democracy, the positive effects on wealth and income distribution arising from the redistribution of (some) privatization revenues could overcome the negative indirect effect of divestiture programs through labor market. If that were the case, not only the support for privatization by the poor classes, potentially more negatively affected by divestiture programs, increases, but also the support for democracy by the rich taxpayers increases too, given that privatization partially substitutes taxation in the redistribution process. 5 See Section 2 for a review of the theoretical literature on the relationship between privatization and income inequality. 6 For instance, the average Gini index, circa 2010, was 30.9 in Europe and Central Asia; 35.5 in South Asia; 36.4 in North Africa and the Middle East; 37.5 in East Asia and Pacific; 43.5 in Sub-Saharan Africa; 43.6 in North America and 47.8 in Latin America and the Caribbean (our computation from UNU-WIDER World Income Inequality Database - WIID). 7 As emphasized by a recent and growing literature, democratic and economic transitions are typically related (see, among the others, Giavazzi and Tabellini, 2005; Persson and Tabellini, 2007; Papaioannou and Siourounis, 2008). 3

4 To find empirical evidence on this we focus on a sample of 80 developing and lowincome countries in the period. We collect data on yearly privatization revenue out of gdp and net income Gini coefficients. In addition to these variables, we include democracy indicators. In particular, we refer to the Gastil index from Freedom House and to its two components, i.e. the strength of political rights and the protection of civil liberties, which capture two relevant and distinct dimensions of a democracy (Dahl, 1971), as well as the Polity2 index released by the PolityIV dataset (Marshall et al, 2016). 8 Our analysis delivers some interesting results. More specifically, we find that an increase in privatization revenues is associated to an increase in net income inequality when political institutions are not representative enough and lack accountability and legitimacy. Thus, our results suggest a not yet investigated policy implication for developing countries: the trade-off between privatization programs and reduction in income inequality that could potentially arise can be solved by the political process only in presence of well consolidated democratic institutions. The paper is organized as follows: the next section contains an overview of the related theoretical literature; Section 3 provides a description of the data; Section 4 presents our econometric method and provides an interpretation of our results and some robustness checks. Section 5 concludes. 2 Related literature Among the contributions which focus on the relationship between income inequality and privatization policies, Birdsall and Nellis (2003) underline that first of all the ownership argument has to be considered. Privatization could determine a distributional loss due to the fact that the ownership of the involved firms tends to become more concentrated. Former State assets go in the hands of few private shareholders which generally are the only to be able to capture the short-run efficiency gains of the divestiture strategy. Even in the case of vouchers, which are always labeled as the best way to equitably share the ownership of public firm among the population, the distributional impact of privatization programs has been disappointing (famous case are those of Russia and the Czech Republic). In fact the actual returns have generally been much lower than the promised ones or have been often captured by an entrenched powerful èlites. According to this, Nellis (2006) says that privatization enriches the already rich and powerful at the expense of the poor. Moreover, using a dual model of a rural-urban economy for describing a developing country, Chao et al (2006) show that an increased partial privatization may lead to wage inequality between skilled and unskilled labor, to higher goods prices and unemployment ratio, and finally to a lower social welfare. 9 Even if they agree that privatization of utilities has been a positive occurrence for the poorest individuals 8 See Section 3 for further details. 9 On the other hand, however, in their study on the effect of the privatization of utilities (primarily electricity, telecommunications, water, and gas) on income distribution in four Latin American countries, McKenzie and Mookherjee (2003) find that privatization generally has a very small effect on inequality. 4

5 all over the developing world, Birdsall and Nellis (2005) remark that privatization poses grave risks of concentrating wealth unless done well and with the full complement of regulation. In a previous work, moreover, they even conclude that, at least in the shortrun, many privatization programs have worsened the distribution of assets and income (Birdsall and Nellis, 2003). More specifically, the literature has identified different channels through which privatization would lead to higher inequality. First of all, via the labour market. In a survey of more than 300 privatized firms, Chong and de Silanes (2003) find that the number of employees declined in about 78% of cases after privatization. Dependency theory suggests another channel through which the labour market is affected by privatization: the reliance on foreign capital increases income inequality of a country through for example under-absorption of labor and sectoral disparities due to the capital intensity of foreign investments (Evans, 1979; Evans and Timberlake, 1980). In terms of prices and services access, a lot of studies, especially focusing on Latin American privatized firms, show that after privatization the urban poor population starts to have access to essential services thanks to the network expansion. This increase in access leads to a real gain for the poor if the prices of these services do not significantly increase (see, for instance, Florio and Poggi, 2010). However, higher prices for accessing a newly privatized service are often related to the elimination of the illegal connections to the grid that normally characterized low income users thus determining an increase in inequality (McKenzie and Mookherjee, 2003). Finally, the distributional outcome of privatization also depends on the effects of other reforms, involving for example market competition and regulatory regime, which are eventually carried out at the same time (Florio and Puglisi, 2005; Birdsall and Nellis, 2005). 10 A large literature has then emphasized the role of democratic institutions in shaping income inequality. In his pioneering work Cutright (1967) claims that there is a negative relationship between democracy and income inequality. 11 In fact in a democracy the poor and disadvantaged can express their preferences and this determines an increase in public policies oriented to income equalization. Moreover, contrary to what happens in autocracies, democratic institutions guarantee the free flow of information. As a consequence, the free circulation of information about the condition of the poor may be embarrassing to the government which does not take into account their needs (Sen, 1981 and Sen, 1999). Third, democracies tend to provide more public goods than nondemocracies and thus to be especially beneficial for the poor. 12 All these arguments are also in line with what found by Meltzer and Richard (1981) and more recently by Tan (2011). In a democracy people are allowed to vote and express their preferences about public policies. Thus, democratization leads to an increase in the number of voters with relatively low income. This can be considered the direct effect of democratization on inequality. This changes the position and the preferences of the median voter and 10 For example, in the European Community, the privatization strategy accounts the following steps: privatization, regulation, vertical disintegration, and liberalization (Ceriani and Florio, 2011). 11 See also Hewitt (1977), Stack (1979), Muller (1985) and Muller (1988). 12 On the relation between democracy, redistributive taxation and the private provision of public goods see also Markussen (2011) 5

6 moves public policies away from the preferences of the elites (i.e. the indirect effect). As a consequence, the more unequal the existing income distribution the stronger will be the corresponding redistributive pressure. 13 More recently, Acemoglu and Robinson (2006) theorize about democratization as a commitment device to future redistribution from the rich (the elites) to the poor (the citizens). Some empirical studies find a positive relation between democratic regimes and redistributive programs. For instance, Boix (2003) suggests that a significant share of the public sector actually depends on the political (democratic) regime in place, which also interacts with the distribution of income, citizens preferences and economic conditions. Along these lines, Aidt et al (2006) show a significant relation between the extension of the voting franchise and the size of government. Profeta et al (2013) find that democracy and civil rights protection are positively correlated with the level of tax revenue and the amount of direct taxes. Fumagalli and Narciso (2012) show a relationship between higher electoral turnout and higher total revenue, government expenditure, budget deficit and more generous welfare state outlays. However, the consensus on the positive relation between democratization and redistribution is not unanimous. Following Harms and Zink (2003), limits to redistribution can characterize both direct and representative democracies, due to limited participation in the election and electoral competition and lobbies organization respectively. Moreover, according to Milanovich (2000) even if the process of redistribution is positively associated with initial income inequality, the evidence that redistribution takes place through the median voter channel is weak. This could happen for a variety of reasons: the decisive voter is at a higher income level than the median voter; middle class gains from redistribution only in the long instead of in the short run; collective decisions are made with institutions of representative democracy and elected officials may have different redistributive preferences than their voters (see, for instance, Fishman et al, 2015). The median voter hypothesis applied to voters does not seem to describe the collective decision-making rule. Also Scervini (2012) and Georgiadis and Manning (2012) cast some doubts on the real effectiveness of the median voter theorem in explaining redistribution process in democratic countries, while Ross (2006) shows that the impact of democratic institutions on improving the condition of the poor appears not too significant. 3 Data description This paper makes use of several data sources, in order to account for the different economic and political dimensions that may affect the relationship between privatization, inequality and democratization. Data on privatization come from the World Bank Privatization Database, that covers the period for developing countries classified 13 Other contributions on the political mechanism through which greater income inequality leads to greater redistribution can be found in Alesina and Perotti (1996) and Alesina and Perotti (1997). Lindqvist and Ostling (2011) study the voters preferences for redistribution in the light of endogenous identity choices (social classes or ethnic group). They find that social class is more relevant and redistribution is higher in ethnically more homogeneous societies. 6

7 as low- or middle-income only. The database includes transactions which generate monetary proceeds to the government resulting from partial and full divestitures, concessions, management contracts, and leases. Thus, only these kinds of transactions that generate revenue for the government from privatization or private sector participation in an existing state-owned enterprise or other government assets are included. In order to make data comparable we take the share of privatization revenue in terms of gdp. Regarding economic inequality, one of the main issues is the availability of reliable data. While there are lots of country specific household surveys that allow computing inequality indices, cross-country comparability is still an open issue. The two main projects aiming at solving this issue are the Lisdatacenter (former Luxembourg Income Study, LIS (2016)), 14 that collects and harmonizes national surveys in order to guarantee the highest level of comparability, and the World Income Inequality Database (wiid) released by unu-wider (UNU-WIDER, 2015), that collects inequality indicators from dozens of different sources, classifying them according to quality, underlying measure (i.e. gross income, net income, consumption, and so on), unit of analysis, equivalence scale, population and sector coverage (i.e. urban, rural, and so on). The trade-off is between highly comparable data on a small set of high-income countries in few years and a wider dataset of barely comparable indicators. In this paper we choose a third option, namely version 5 of the Standardized wiid (swiid from now on). Swiid is a project run by Frederick Solt since 2009 (Solt, 2009) that imputes the missing data on inequality from wiid using multiple imputation techniques and validates it using the high quality Lisdatacenter dataset (see Solt, forthcoming, for more details). The great advantage of swiid is that it provides an ideally comparable panel of inequality indicators; the drawbacks are that i) it only provides Gini coefficients (while other datasets also provide quantiles and mean income); and ii) the statistical analyses must take into account the underlying multiple imputation in order to correct for the precision of the estimation of every country/year values. 15 As main indicator of the level of democracy we adopt the Gastil index of democracy, released by the Freedom House (Freedom House, 2016) on a yearly basis. the index is the average of two other indicators, one referring to Civil Liberties, the other to Political Rights. Each country receives a score from 1 to 7 in both dimensions according to several aspects, such as freedom of expression and belief, rule of law, political pluralism, and so on. The original scale has been reversed in our analysis, so that it ranks from 1 (least democratic) to 7 (most democratic). For robustness checks we also refer to an alternative measure of democracy: the Polity2 index, released by the Polity IV project (Marshall et al, 2016), 16 computed as the difference between an indicator of democracy and an indicator of autocracy. It ranges between -10 (autocracy) and 10 (full democracy). Also in this case the two constitutive dimensions summarize several characteristics of the political system. 14 lisdatacenter.org 15 On the possible issue arising from the use of secondary datasets for the analyses of cross-national inequality, we refer to Atkinson and Brandolini (2001)

8 According to the traditional literature on the determinants of inequality and to the more specific theoretical literature on the relationship between privatization and inequality, we also include in our analysis the following controls: per-capita gdp and foreign direct investments over gdp (World Bank), a household price index and the employment rate (Penn World Tables, Feenstra et al (2013)), and finally different measures of economic freedom released by the Heritage foundation (Heritage Foundation, 2016). Table 1 reports the descriptive statistics for all the country/years included in the baseline model (reported in Table 3). The working dataset consists of an unbalanced panel of 552 observations, in which 80 countries 17 are observed for about 7 points in time. Unfortunately, the coverage of the Heritage foundation data is not complete, so that the sample size is smaller when we control for these economic freedom indicators. Table 2 shows the estimated means of the Gini coefficients computed on net incomes in the whole sample and in the four regional sub-samples. As expected, inequality results to be much higher than the average in Latin America and much lower than the average in Eastern European countries Method and results In this section, we test the conditional hypothesis described before: in developing countries the relationship between privatization and income inequality may depend on the extent of democratization process. Thus, we estimate the following model: G i,t = α + βd i,t 2 + γp i,t 2 + δd i,t 2 P i,t 2 + ζx i,t 2 + η t + ε i,t (1) where G is the Gini coefficient computed on net income in country i at year t, D is the Gastil index of democracy in year t 2, P is the ratio of revenues from privatization with respect to gdp, 19 X is a set of control variables including per-capita gdp and foreign direct investments as a share of gdp, η is a set of yearly dummies, and ε is the idiosyncratic error term. Notice that the inclusion of the interaction term allows us to explicitly test our hypothesis on the relationship between income inequality and privatization in the light of democratization process in developing countries (see Brambor et al, 2006 for details about multiplicative interaction models). 20 Finally, since the error 17 The full list of countries included in the sample is the following: Albania, Algeria, Angola, Argentina, Armenia, Bangladesh, Belize, Bolivia, Bosnia and Herzegovina, Brazil, Bulgaria, Burkina Faso, Cameroon, Chile, China, Colombia, Costa Rica, Cote d Ivoire, Croatia, Czech Republic, Ecuador, Egypt, Estonia, Ethiopia, Georgia, Ghana, Guatemala, Guinea-Bissau, Honduras, Hungary, India, Indonesia, Iran, Jamaica, Jordan, Kazakhstan, Kenya, Lao, Latvia, Lesotho, Lithuania, Macedonia, Madagascar, Malawi, Malaysia, Mali, Mauritania, Mexico, Moldova, Morocco, Mozambique, Nepal, Nicaragua, Nigeria, Pakistan, Panama, Peru, Philippines, Poland, Romania, Russian Federation, Rwanda, Senegal, Serbia, Slovak Republic, South Africa, Sri Lanka, Tanzania, Thailand, Trinidad and Tobago, Tunisia, Turkey, Uganda, Ukraine, Uruguay, Uzbekistan, Venezuela, Viet Nam, Zambia, Zimbabwe. 18 Notice that World Bank data on privatization proceeds only refer to low-income countries, so that Americas does not include US and Canada and Europe only includes Eastern European countries. 19 See Doyle (2010) on the discussion of exogenous determinants of privatization. 20 Starting from our baseline model, we firstly investigate the hypothesis that the effects of our independent variables on income inequality are not instantaneous. Our preliminary results, available upon 8

9 term might be serially correlated within countries and thus wrongly inflate the precision of our estimates, we always cluster the standard errors at the country level (see Bertrand et al, 2004). While there is some within-country variation in democracy and inequality indicators over the selected time period, most of the variation is actually between-country. As a consequence, only when countries are tracked over longer time periods, within-country variation is more likely to pick up the true association, if there is one, between democratization, privatization and income inequality. For these reasons, we decide not to use country fixed effects. Our results are shown in Table 3 21 and are organized as follows. In the first (unconditional) specification we only consider our democracy measure together with the total amount of privatization proceeds out of gdp and our interaction term; in the second and third specification we simply control for per capita gdp and the share of fdi out of gdp respectively, while in the forth column we include both our controls. We find a negative and statistically significant relationship between the Gini coefficient and privatization interacted with the level of democracy, and a positive and statistically significant relationship between inequality and privatization, at least when we do not control for per-capita gdp. On the contrary, the relationship between inequality and democracy is not statistically significant. Thus, it seems that opposite forces are at play when we look at the link between ex post income inequality and both privatization and democracy. Surprisingly, there is no direct relationship between income inequality computed on net incomes (hence taking into account the redistribution process) and democracy, but for constant or increasing level of privatization proceeds, this relationship becomes negative. At the same time, there is a direct positive relationship between privatization and inequality, but for constant or increasing level of democracy, the relationship between privatization and Gini coefficient turns to be negative. As a consequence, it becomes essential to look at what we call the net relationship between net income inequality and both privatization and democracy (see the bottom part of Table 3). 22 Given that our aim consists in testing that the true relationship between income inequality and privatization may depend on the extent of democratization process in developing countries, we are specifically interested in evaluating the sign and the statistical significance of the net relationship with privatization. According to our estimates, this net relationship is negative and statistically significant, confirming that in developing countries an increase in privatization proceeds is related to a reduction in ex post income inequality only if democracy is well consolidated. On the contrary, request, confirm this reasonable assumption. Thus we decide to run our regressions with two lags in our explanatory variables and controls. Our results are virtually unaffected using one or three period lags. Notice also that the use of lagged variables can be considered as a mild way to mitigate the possible presence of reverse causality problems in our estimates. 21 Given that we use the Gini index as dependent variable, a positive (negative) relationship between our explanatory variables means that if they increase then ex post income inequality increases (reduces). 22 Notice that we compute the net relationship between ex post income inequality and privatization at the mean value of the Gastil index in our sample, while we compute the net relationship between ex post income inequality and the Gastil index at the mean value of privatization proceeds in our sample. 9

10 democracy does not seem to have any net relationship with Gini index. In order to be able to distinguish between the potential role of privatization proceeds and per-capita gdp when we are investigating their relationship with net income inequality, we re-run our regressions by using per-capita gdp as explanatory variable and privatization proceeds as control variable (Table 4). Results show an interesting pattern: per-capita gdp is positively correlated with inequality at low levels of democracy and negatively at high levels of inequality. The net relationship is not statistically different from zero, consistently with results in Table 3. The coefficient for the privatization proceeds is negative and statistically significant, again consistently with the net relationship in Table 3. Summing up, our results shed light on the theoretically ambiguous relationship between net income inequality and privatization. In our sample of developing countries, privatization proceeds are significantly associated to lower inequality only when democratic systems are well consolidated. A similar trend is found when looking at per-capita gdp. Also in this case, the correlation with inequality is negative only for consolidated democracies, and turns positive when the democracy index is particularly low. What is different is the net relationship, that is significantly negative for privatization and not statistically different from zero for per-capita gdp. 4.1 Robustness checks In this section, we want to check the robustness of the findings described in the previous section by (i) using different democracy measures; (ii) enlarging the set of control variables, and (iii) investigating the regional heterogeneity. Our results are shown in Tables First, we investigate whether both the different components of the Gastil index, i.e. civil liberties guarantee and property rights protection, and another measure of democracy usually adopted in empirical studies, i.e. the Polity2 index from the Polity IV dataset, can be differently related to inequality with respect to the Gastil index itself. Our findings do not change (see Tables 5-7). In particular, our main result on the net relation with privatization still holds, especially when we replace the Gastil index of democracy with its political rights component and the Polity2 index. 23 Second, we enlarge the set of controls (see Table 8) by including: (i) the Heritage foundation index of economic freedom and its ten components referring to rule of law, government size, regulatory efficiency and market openness; (ii) the household price index; and (iii) the employment rate. 24 Apart from the fact that economic freedom can 23 In Table 7, the coefficients of both privatization over gdp and the interaction term are negative and statistically significant. However, the coefficient of the net relation with privatization is negative and statistically significant, while the one of the net relation with the Polity2 index is not statistically significant, consistently with the results of the baseline model. 24 We re-run our regressions by adding one by one all these control variables. Our result on the net relation with privatization always holds. For space reasons, we decided to show only the estimates referred to the regressions in which our new controls show a statistically significant relationship with the Gini net index. All other results are available upon request. 10

11 facilitate effective democracy, by looking at its components we become able to control for a lot of factors that, as suggested by the theoretical literature, could affect, even indirectly, the relationship between privatization and income inequality. More specifically, the recognition and the protection of property rights; the level of freedom from corruption; the size of government measured by fiscal and government spending freedom; the individual s freedom to start and run an enterprise; labour market freedom and monetary freedom that generally refer to the regulatory environment; trade, investment and financial freedoms as measures of market openness. Finally, the household price index helps us taking into account the differences in consumption price levels across countries, and the employment rate accounts for the potential indirect effects of privatization on inequality through the labour market. Our previous result on the net relationship with privatization always holds. Moreover, we are now able to show that the relationship between the overall index of economic freedom and the Gini index is positive and statistically significant (column 2 of Table 8). Also the two specific components of the overall economic freedom index that are directly related to government size, that is fiscal and government spending freedoms, are positively correlated with net income inequality. 25 In particular, the higher coefficient of fiscal freedom seems to capture the direct effect of taxation on income inequality compared to the indirect effect of public spending (columns 3 and 4 of Table 8).We find that there is a negative and statistically significant relationship between a more open trade market and income inequality: in fact if privatization process can more easily involve foreign firms, it can potentially increase the total available resources (column 5 of Table 8). Finally, the household price index shows a positive and statistically significant relationship with the Gini net index (column 6 of Table 8). In Tables 9-10, we investigate the regional heterogeneity by taking Americas as reference group given its highest level of inequality with respect to the other areas (Africa, Asia and Europe, see Table 2). In particular, we re-estimate our model by using region fixed effects. Looking at the first column of Table 9-10, we find again that in all our developing countries an increase in privatization proceeds is related to a reduction of income inequality if democracy is high. Then, we also refer each explanatory variable to each specific region in order to be able to break up our net relationships by area. Thus, it seems that in Africa and Asia the net relationship with privatization is negative and statistically significant. At the same time, the net relationship with democracy becomes negative and statistically significant in Americas and Europe, meaning that in these regions an increase in the level of democracy is related to a reduction in net income inequality for higher privatization proceeds (see column 2 of Table 9-10). 5 Conclusions This paper is a first attempt to empirically investigate the relationship between privatization and income inequality adopting a new approach based on the interplay between 25 Notice that an increase in fiscal freedom means a reduction of taxes, and an increase in public spending freedom a reduction of public outlays. 11

12 privatization and democratization in low or middle-income countries. In particular, our analysis is devoted to explore whether representative political institutions may contribute to shed light on the relationship between privatization and income inequality. We refer to 80 developing countries belonging to four areas of the world (Asia, Africa, Europe and Americas). Our results show that an increase in privatization revenue is correlated with an increase in income inequality in countries where representative political institutions are not mature enough, thus giving empirical evidence to what stated by Birdsall (1999): The risks of privatization arise because developing and transitional economies, almost by definition, are handicapped by relatively weak institutions, less well-established rules of transparency, and often, not only high concentrations of economic and political power but a high correlation between those two areas of power. On the contrary, in consolidated democratic governments (a share of) privatization revenue can be used to satisfy the redistributive pressures from the poor and then to secure a more equal income distribution among citizens, in this way keeping under control the opposition from the rich towards a rise in the level of taxation and the opposition from the poor towards privatization programs. Our findings, robust to different specifications, measures of democracy and economic controls, suggest an interesting and not yet investigated policy implication concerning the timing of democratization and privatization in developing countries. If we care about the relationship between divestiture programs and income distribution, it seems better to start with a political transition towards democracy. Indeed, privatization is negatively correlated to income inequality only when the democratization process is well developed. References Acemoglu D, Robinson JA (2006) Economic Origins of Dictatorship and Democracy. Cambridge University Press Aidt TS, Dutta J, Loukoianova E (2006) Democracy comes to europe: franchise extension and fiscal outcomes European Economic Review 50: Alesina A, Perotti R (1996) Income distribution, political instability, and investment. European Economic Review 40(6): Alesina A, Perotti R (1997) The welfare state and competitiveness. American Economic Review 87(5): Atkinson AB, Brandolini A (2001) Promise and pitfalls in the use of secondary datasets: income inequality in OECD countries as a case study. Journal of Economic Literature XXXIX: Bertrand M, Duflo E, Mullainathan S (2004) How much should we trust differences-indifferences estimates? The Quarterly Journal of Economics 119(1):

13 Birdsall N (1999) Globalization and the developing countries: The inequality risk. In: Remarks at Overseas Development Council Conference, Making Globalization Work, International Trade Center, Carnegie Endowment for International Peace, Washington, D.C. Birdsall N, Nellis J (2003) Winners and losers: Assessing the distributional impact of privatization. World Development 31(10): Birdsall N, Nellis J (eds) (2005) Reality Check: The Distributional Impact of Privatization in Developing Countries. Center for Global Development, Washington, DC Boix C (2003) Democracy and Redistribution. Cambridge University Press Brambor T, Clark WR, Golder M (2006) Understanding interaction models: Improving empirical analyses. Political Analysis 14:63 82 Brune N, Garrett G, Kogut B (2004) The international monetary fund and the global spread of privatization. IMF Staff Papers 51(2): Cavaliere A, Scabrosetti S (2008) Privatization and efficiency: From principals and agents to political economy. Journal of Economic Surveys 22(4): Ceriani L, Florio M (2011) Consumer surplus and the reform of network industries: a primer. Journal of Economics 102: Chao CC, Hazari B, Yu E (2006) Rising wage inequality in developing economies: Privatization and competition. Journal of International Trade & Economic Development 15(3): Chong A, de Silanes FL (2003) Privatization and labor force restructuring around the world. RES Working Paper 4329, Inter-American Development Bank, Research Department Cook P, Kirkpatrick C (1997) Policy arena: Privatization and public enterprise reform in developing countries: the world bank s bureaucrats in business report. introduction and overview. Journal of International Development 9(6): Cutright P (1967) Inequality: a cross national analysis. American Sociological Review 32: Dahl R (1971) Polyarchy: Participation and Opposition. Yale, New Haven Doyle D (2010) Politics and privatization: Exogenous pressures, domestic incentives and state divestiture in latin america. Journal of Public Policy 30(3): Evans P (1979) Dependent Development. Princeton University Press Evans PB, Timberlake M (1980) Dependence, inequality, and the growth of the tertiary: A comparative analysis of less developed countries. American Sociological Review 45(4):

14 Feenstra RC, Inklaar R, Timmer M (2013) The Next Generation of the Penn World Table, NBER Working Paper no Fishman R, Jakiela P, Kariv S, Markovits D (2015), The distributional preference of an elite, Science, 349(6254) Florio M, Poggi A (2010) Energy deprivation dynamics and regulatory reforms in europe: Evidence from household panel data. Energy Policy 38: Florio M, Puglisi R (2005) Public ownership, privatization and regulation: social welfare counterfactuals for british telecom. Working Paper 13-05, Georgetown University Freedom House (2016) Freedom in the World 2016, freedomhouse.org Fumagalli E, Narciso G (2012) Political institutions, voter tournout, and policy outcomes. European Journal of Political Economy 28: Georgiadis A, Manning A (2012) Spend it like beckham? inequality and redistribution in the uk, Public Choice 151: Giavazzi F, Tabellini G (2005) Economic and political liberalizations. Journal of Monetary Economics 52(7): Harms P, Zink S (2003) Limits to redistribution in a democracy: a survey. European Journal of Political Economy 19: Heritage Foundation (2016) 2016 Index of Economic Freedom, Promoting Economic Opportunity and Prosperity Hewitt C (1977) The effect of political democracy on equality in industrial societies: A cross-national comparison. American Sociological Review 42: IMF (2011) Fiscal monitor. addressing fiscal challenges to reduce economic risks. Tech. rep., International Monetary Fund Kuznets S (1963) Quantitative aspects of the economic growth of nations. viii: The distribution of income by size. Economic Development and Cultural Change 11(1):1 80 Lindqvist E, Ostling R (2011) Identity and redistribution. Public Choice forthcoming Luxembourg Income Study Database (LIS), (multiple countries; 2016). Luxembourg: LIS. Markussen T (2011) Democracy, redistributive taxation and the private provision of public goods. European Journal of Political Economy 27: Marshall M.G., Gurr T.R., Jaggers K. (2016) POLITY IV PROJECT, Political Regime Characteristics and Transitions, , Center for Systemic Peace 14

15 McKenzie D, Mookherjee D (2003) The distributive impact of privatization in latin america: Evidence from four countries. Journal of LACEA Economia Megginson WL (2010) Privatization trends and major deals in Report 6-22, Privatization Barometer Meltzer AH, Richard SF (1981) A rational theory of the size of government. Journal of Political Economy 89(5): Milanovich B (2000) The median-voter hypothesis, income inequality and income redistribution: an empirical test with the required data. European Journal of Political Economy 16: Muller EN (1985) Income inequality, regime, repressiveness, and political violence. American Sociological Review 50:47 61 Muller EN (1988) Democracy, economic development, and income inequality. American Sociological Review 53:50 68 Nellis J (2006) Privatization, a summary assessment. Working Paper 87, Centre for Global Development Opper S (2004) The political economy of privatization: Empirical evidence from transition economies. KYKLOS 57(4): Papaioannou E, Siourounis G (2008) Democratization and growth. The Economic Journal 118: Persson T, Tabellini G (2007) The growth effect of democracy: Is it heterogeneous and how can it be estimated. Working Paper 13150, NBER Profeta P, Puglisi R, Scabrosetti S (2013) Does democracy affect taxation and public spending? evidence from developing countries. Journal of Comparative Economics 41(3): Ross M (2006) Is democracy good for the poor? American Journal of Political Science 50(4): Scervini F (2012) Empirics of the Median Voter: Democracy, Redistribution and the Role of the Middle Class, The Journal of Economic Inequality, 2012, 10(4): Sen A (1981) Poverty and Famines: An Essay on Entitlement and Deprivation. Oxford University Press, New York Sen A (1999) Development as Freedom. Alfred A. Knopf, New York Solt F (2009) Standardizing the World Income Inequality Database. Social Science Quarterly 90(2):

16 Solt F (forthcoming) The Standardized World Income Inequality Database. Social Science Quarterly Stack S (1979) The effects of political participation and socialist party strength on the degree of income inequality. American Sociological Review 43: Stallings B (1992) International Influence on Economic Policy: Debt, Stabilization and Structural Reform, in S Haggard and R Kaufman (eds), The Politics of Economic Adjustment, Princeton University Press: Princeton Tan ES (2011) Democracy, wages and inequality: political voice and public finance redistribution. Working Paper November 05, University of London - Royal Holloway College - Department of Economics UNU-WIDER (2015) World Income Inequality Database (WIID3c), September, https: // Williamson J (1993) Democracy and the Washington Consensus. World Development 21(8): World Bank (2016) International Debt Statistics Washington, DC: World Bank 16

17 Table 1: Summary Statistics Varname Obs Mean Std. Dev. Min Max Total privatization proceeds / gdp Gastil index of democracy Gastil index of civil liberties Gastil index of political rights Polity2 index of democracy Per-capita gdp (Current US$) Foreign direct investments / gdp Economic freedom (overall score) Fiscal freedom Government spending freedom Trade freedom Household price index (US$ 2005 = 1) Table 2: Gini coefficients on net incomes Area Obs Mean Std. Err. Africa Americas Asia Europe All

18 Table 3: Baseline model Dep.var.: Gini net (1) (2) (3) (4) b/se b/se b/se b/se Gastil index Privatization / GDP * * Gastil index X Privatization / GDP * * ** * Per-capita GDP FDI / GDP Constant *** *** *** *** Year FE Yes Yes Yes Yes Net relation with privatization se p-value Net relation with Gastil index se p-value Model F-test p-value Obs Countries Note: *** p 1%, ** p 5%, * p 10%. Standard errors clustered at country level. All explanatory variables are 2-periods lagged. 18

19 Table 4: Privatization revenue and per-capita gdp Dep.var.: Gini net (1) (2) (3) (4) b/se b/se b/se b/se Gastil index Per-capita GDP 0.004*** 0.004*** 0.004*** 0.004*** Gastil index X Per-capita GDP *** *** *** *** Privatization / GDP ** ** FDI / GDP Constant *** *** *** *** Year FE Yes Yes Yes Yes Net relation with GDP se p-value Net relation with Gastil index se p-value Model F-test p-value Obs Countries Note: *** p 1%, ** p 5%, * p 10%. Standard errors clustered at country level. All explanatory variables are 2-periods lagged. 19