Fiscal Decentralization, Economic Freedom, and Political and Civil Liberties in the Americas

Tulane Economics Working Paper Series Fiscal Decentralization, Economic Freedom, and Political and Civil Liberties in the Americas Antonio N. Bojanic Department of Economics Tulane University abojanic@tulane.edu Working Paper 1609 August 2016 Abstract This paper analyzes the impact of fiscal decentralization on economic freedom and political and civil liberties in the Americas. Regarding the latter and with the full sample of countries, the findings suggest that decentralization initially worsens but eventually improves political and civil liberties, underlining the importance of fiscal decentralization as a driver for achieving basic liberties. When Canada and the US are excluded, the evidence shows that decentralization may eventually be a detriment for political and civil liberties. With respect to the impact of fiscal decentralization on economic freedom, decentralization first hinders but eventually increases freedom when all countries are included, emphasizing the point that a decentralization regime takes time to develop and function properly. When Canada and the US are excluded, decentralization initially increases but ultimately hampers freedom, demonstrating that if the decentralization regime does not address important matters like fiscal discipline, wealth inequality, and political accountability, economic freedom like political and civil liberties will also deteriorate. Keywords: Fiscal decentralization, economic freedom, political and civil liberties. JEL codes: E62, H70, O23, P52

Fiscal Decentralization, Economic Freedom, and Political and Civil Liberties in the Americas Antonio N. Bojanic Department of Economics Tulane University abojanic@tulane.edu (504)862-8355 1

Abstract This paper analyzes the impact of fiscal decentralization on economic freedom and political and civil liberties in the Americas. Regarding the latter and with the full sample of countries, the findings suggest that decentralization initially worsens but eventually improves political and civil liberties, underlining the importance of fiscal decentralization as a driver for achieving basic liberties. When Canada and the US are excluded, the evidence shows that decentralization may eventually be a detriment for political and civil liberties. With respect to the impact of fiscal decentralization on economic freedom, decentralization first hinders but eventually increases freedom when all countries are included, emphasizing the point that a decentralization regime takes time to develop and function properly. When Canada and the US are excluded, decentralization initially increases but ultimately hampers freedom, demonstrating that if the decentralization regime does not address important matters like fiscal discipline, wealth inequality, and political accountability, economic freedom like political and civil liberties will also deteriorate. Keywords: fiscal decentralization, economic freedom, political and civil liberties JEL Classification Codes: E62, H70, O23, P52 2

I. Introduction During the last three decades, a great deal of political and economic decentralization has taken place in the developing world. The reasons for this trend are many, from the realization that greater control of resources and the power to pass legislation on a variety of issues will likely result in greater efficiencies and less waste, to the desire of many regions within countries to achieve greater control of their destinies by asserting a firmer control on their sources of income and insisting on spending this income on locally focused projects. Research in the economics literature on the specific topic of economic decentralization which will be referred to henceforth as fiscal decentralization 1 has usually been centered on the impact of this process on governance, economic growth, macroeconomic stability, and on themes related to poverty and income inequality. In this paper, the analysis is expanded by focusing on the impact of fiscal decentralization on indicators of economic freedom and of political and civil liberties in the Americas. Though it would seem natural to think of a connection between a particular region s ability to generate its own revenue and spend it as it pleases with greater economic freedom and deeper political and civil liberties, little research has been done on this topic, 2 hence the timeliness of this initial work which concentrates on a region of the world where the issue of fiscal decentralization has evolved significantly over the last few decades. The rest of the paper is organized as follows. First, a brief review of the literature on the relationship between fiscal decentralization and a variety of economic and social indicators is presented. Second, a theoretical argument for a positive relationship between fiscal decentralization and economic, political, and civil liberties is suggested. Next, the data and methodology are introduced. The results of the empirical 1 Fiscal decentralization can be understood as a more narrow component of economic decentralization, which may include the power to legislate one s own norms on aspects as varied as fiscal, monetary, trade, and industrial policies. Here, the more limited focus centers on a region s ability to tax (i.e., generate its own sources of revenue) and spend resources without the overt control of a central authority (i.e., the central or national government). 2 The issue of administrative decentralization and political power is central in the fields of Public Administration and Political Science, as evidenced, for instance, in Kaufman (1969), Rodden (2002), and Schmidt (2007), but little research on this topic is found in the economics literature. 3

estimation are presented in section five, and section six summarizes and reviews the policy implications of the findings. II. Review of the literature Three major themes have been analyzed in the economics literature from the perspective of fiscal decentralization. The first one refers to the issue of how governance is affected by devolving revenue and expenditure powers to regional institutions (i.e., state, municipal, and local governments). A second theme deals with the ways in which fiscal decentralization affects economic growth, and by extension, economic stability. Lastly, the impact of fiscal decentralization on poverty, inequality, and wealth redistribution has also been analyzed with significant detail. A brief review of some of the most important works for each theme is presented here. How fiscal decentralization affects governance has been studied for both cross country and individual cases. Representative works include Arzaghi and Henderson (2005), who analyze the more basic question of why there has been a trend for fiscal decentralization across the world; de Mello and Barenstein (2001), analyzing cross-country data for seventy eight countries and showing that the assignment of expenditure and revenue mobilization functions to subnational levels of government is associated with various indicators of governance, such as corruption, rule of law, and government effectiveness; Altunbas and Thornton (2012), also with cross-country data, and showing that countries in which a larger share of fiscal revenues and expenditures are located at the level of subnational governments appear to be less corrupt; and Kyriacou, et al. (2015) who, based on a sample of twenty four OECD countries, find that fiscal decentralization promotes regional convergence in high government quality settings but to wider regional disparities in countries with poor governance. Individual country studies include Mohapatra (2012) analyzing India; Shen, et al. (2012) for China; and Balunywa, et al. (2014) in Uganda. 4

With regards to the impact of fiscal decentralization on economic growth and economic stability, early important contributions include Davoodi and Zou (1998), working with panel data for forty six countries and finding a negative relationship between fiscal decentralization and growth in developing countries, but none in developed countries; Rodden, et al. (2003), who in the context of how budget constraints at lower levels of government may impact economic stability, provide a framework for understanding how hard and soft budget constraints impact the economy stability of eleven economies with varying degrees of political and institutional development; Martínez-Vázquez and McNab (2003), and Martínez-Vázquez and McNab (2006), who in their earlier work concluded that the impact of decentralization on economic growth remained an open question, but when a negative correlation was found for developed countries in their latter work, this negative impact might have been offset by the positive impact on macroeconomic stability; Thornton (2007), with a cross section study of nineteen OECD countries, and finding that when the measure of fiscal decentralization is limited to the revenues over which sub-national governments have full autonomy, its impact on economic growth is not statistically significant; and Neyapti (2010), working with panel data for sixteen countries and finding evidence that expenditure and revenue decentralization reduce budget deficits. More recent works include Rodríguez-Pose and Ezcurra (2011), who for a set of twenty one OECD countries find that there is a negative and significant association between fiscal decentralization and economic growth, a relationship which is robust to the inclusion of a series of control variables and to differences in expenditure preferences by sub-national governments; Gemmell, et al. (2013), also for a set of OECD countries and finding that decentralization on the expenditure side has tended to be associated with lower economic growth while decentralization on the revenue side with higher growth; and Blöchliger (2013), finding a positive association between fiscal decentralization and GDP per capita in OECD countries, with the impact of revenue decentralization greater than for spending decentralization. Representative works for individual countries include Yifu Lin and Liu (2000) on China, Xie et al. (1999) for the United States, and Samimi, et al. (2010) for Iran. 5

Concerning how fiscal decentralization affects poverty, income inequality, and the more general theme of wealth redistribution, significant contributions include Boex, et al. (2006), who in addition to surveying the literature on the ways in which fiscal decentralization affects poverty, provide a qualitative set of suggestions and recommendations directed at international development agencies for conducting fiscal decentralization reforms in a more pro-poor manner; Sepúlveda and Martínez-Vázquez (2011), finding that for a large dataset of countries fiscal decentralization appears to reduce poverty as long as the share of subnational expenditures is not greater than one third of total government expenditures, and likewise helps reduce income inequality only if the general government represents a significant share of the economy; Goerl and Seiferling (2014), utilizing a large dataset of countries and finding that decentralization of government expenditures can help achieve a more equal distribution of income conditional on several factors; Sacchi and Salotti (2014), for a set of OECD countries and finding that a higher degree of tax decentralization is associated with higher household income inequality, demonstrating that even if fiscal decentralization is attractive for efficiency reasons, it may actually have undesirable consequences on income distribution grounds; and Uchimura, ed. (2012) focusing on the experiences with fiscal decentralization and its impact on development in the Philippines, Vietnam, and Thailand. At the individual country level, a sample of contributions include Francis and James (2003) focusing on Uganda, Song (2013) on China, and Zakaria (2013) on Indonesia. The preceding survey of relevant literature shows that clear cut conclusions regarding how fiscal decentralization affects governance, economic growth and economic stability, and poverty and income inequality are still some ways off. Fiscal decentralization tends to impact a country s institutions and economic structures in seemingly uncertain ways, with the level of development of a nation which may be considered as an indicator of the quality of the institutional setting of a country and the extent to which decentralization takes place, playing a significant role in determining the effectiveness of devolving revenue and expenditure functions to sub-national government levels. 6

III. Theoretical foundations between fiscal decentralization and liberties As has been suggested in the previous section, fiscal decentralization may impact a country s institutions and economic outcomes in varying ways, conditional on a set of factors including the level of development of a nation and the extent to which decentralization takes place. Understanding the ways in which it affects economic freedom and political and civil liberties may in principle also be subject to similar factors, though it would seem more natural to argue for a positive correlation between decentralization and economic, political and civil liberties. The case for a positive relation between fiscal decentralization and political and civil liberties rests on the assumption that devolving revenue and expenditure functions to sub-national levels of government should increase the political clout and individual rights of people living in those regions. In the fields of political science and public administration, this is not a novel idea, as Kaufman (1969) already predicted that administrative decentralization which presumably includes fiscal decentralization increases the representativeness of the average citizen in government entities in charge of deciding over policy, which highlights the ways in which political and civil liberties are enhanced in a decentralized government. Likewise, Escobar-Lemmon and Ross (2014) demonstrate that, for the case of Colombia, fiscal decentralization improves perceptions of accountability among citizens, implying that local and regional politicians are more responsive and accountable to their constituents. In a similar study, Michels (2011) shows that citizen involvement increases in decentralized governments, and this has a number of positive effects on democracy: it strengthens issue knowledge, civic skills, and public engagement, and, consequently, one can infer that political and civil liberties are heightened as well. Islam (2015), focusing on Bangladesh, shows that strong local governments deepen the democratic process by ensuring participatory development, and finally, though not exhaustively, Weitz-Shapiro (2008), shows that in Argentina there is an important positive link between local government performance and citizen system support. 7

Though the ways in which fiscal decentralization affects different themes related to fundamental civil, political, and economic liberties has been studied at length in different fields in the social sciences, it is interesting to note that it has not received the same level of attention in the economics literature, with few notable exceptions, including Weingast (2009), who studies how specific political institutions, such as democracy, interact with decentralization; Faguet (2004), focusing on Bolivia and examining whether decentralization increases the responsiveness of public investment to local needs; and Fisman and Gatti (2002), analyzing the relationship between decentralization and the extent of rent extraction by private parties, with findings suggesting a strong negative relationship between expenditure decentralization and corruption. The present work adds to the economics literature by testing and analyzing the relationship between devolving revenue and expenditure functions to sub-national levels of government and political and civil liberties of the nation experiencing the decentralization process. The expected positive relationship between fiscal decentralization and economic freedom rests on many of the same principles applied to decentralization and political and civil liberties. In fact, it can reasonably be argued that economic liberties are an extension of other types of liberties, though this assumption can become tricky when analyzed in the context of countries where significant decentralization has taken place without an equal increase in political and civil liberties. 3 Regardless of the exceptions to the norm, however, it makes economic sense to think that there should be a positive association between decentralization and economic liberties just as it makes sense that a positive correlation should also exist between decentralization and political and civil liberties which motivates the present study as pertinent and relevant for a region of the world where the issue of decentralization has taken center stage in the last decades. 3 Shirk (1993) illustrates this point for the specific case of China, where significant decentralization has taken place without an equivalent increase in political and civil rights. 8

IV. Data and methodology Perhaps the most significant challenge faced when conducting a cross-country study of fiscal decentralization is how to properly measure the extent of decentralization. An optimal scenario would be one in which the decentralization data set for all countries is fully comparable and truly reflects the independent and autonomous activities of sub-national governments. Under this scenario, there would be a clear demarcation line between the activities of the central government and the activities of sub-national levels of government (regional, state and local governments), concerning both revenue collection and expenditure decisions, and hence the true extent of decentralization would be able to be measured and quantified. Constructing such a panel data set would require a thorough understanding of the nature of the tax system of a nation, particularly as it refers to the structure of revenue sharing among regions; the nature of grants and transfers between the central government and sub-national levels of government; the discretion of subnational governments to levy, collect, and modify their tax bases and rates; the degree of autonomy of subnational governments to spend public resources; and the overall level of political autonomy of sub-national governments. Predictably, it is difficult to construct a data set for a single country that meets all these ideal conditions much less for a set of countries with equally measured indicators of decentralization hence the necessity to utilize the International Monetary Fund s Government Finance Statistics Annual Yearbook (GFS), 4 as the primary data source for revenue and expenditure data for national and sub-national levels of government. Despite its limitations including the fact that it does not report on the nature of government transfers, on whether these transfers and grants are under the control of the central or sub-national governments, and on its lack of decentralized information for many developing nations the GFS represents the most widely employed data source for fiscal decentralization studies, particularly at the cross-section level, and it is the primary data source for this study as well. 4 Available at https://www.imf.org/external/pubs/ft/gfs/manual/comp.htm 9

The standard indicator of fiscal decentralization in the economics literature is the ratio of sub-national government revenues (expenditures) to general government revenues (expenditures), hence the two measures of fiscal decentralization utilized here are: (i) the ratio of total sub-national government revenues to general government revenues (expressed as a percentage), and (ii) the ratio of total sub-national government expenditures to general government expenditures (also expressed as a percentage). The data set provided by GFS reports information at the consolidated central government level and for some countries at the regional, state, and local government levels as well. Revenues (expenditures) at the regional, state, and local governments are added together in order to come up with a single figure for subnational government revenues (expenditures). Of the twenty three countries in the Americas excluding nations in the Caribbean GFS provides data disaggregated by general and sub-national government levels for eleven nations, hence this study concentrates on that sample of American countries for whom available data exists. 5 Yearly observations range from 1972 to 2012, though the dates of available data for the eleven countries do not necessarily coincide. 6 Conditional on whether the revenue or expenditure decentralization indicators are utilized as explanatory variables and on the specific methodology utilized in estimating a regression, the number of observations varies from a low of 38 to a high of 107. The end result is an unbalanced panel data set of 107 observations for 11 American countries and with observations ranging from 1972 to 2012. Though there are significant gaps in the dataset, I chose not to utilize averages or linear approximations to fill in those gaps in order to let the actual data speak for itself. The indicators of political and civil liberties and economic freedom are, respectively, a combined ratio of political and civil liberties constructed out of three indices elaborated by the non-governmental organization Freedom House, 7 and the index of economic freedom published jointly by the Heritage Foundation and the 5 The 11 countries are Bolivia, Brazil, Canada, Chile, Colombia, Costa Rica, El Salvador, Mexico, Paraguay, Peru, and the United States. 6 The appendix reports the specific time periods of available data for the countries utilized in the study. 7 The two indices of political rights and civil liberties, as well as a qualitative assessment of the degree of liberty of each nation, are available at https://freedomhouse.org/report-types/freedom-world#.vy_fwi1rhcw. Here, the two indices and the qualitative assessment have been combined into a single index of political and civil liberties. It ranges from 0.18 (most free) to 1.00 (least free). 10

Wall Street Journal. 8 Though there are other indices available that also measure political, civil, and economic liberties the Democracy Index published by the UK-based Economist Intelligence Unit, the Worldwide Press Freedom Index produced by Reporters Without Borders, and the Economic Freedom of the World Index produced by the Libertarian Fraser Institute, to mention a few these two were chosen due to their wide utilization and acceptance in the relevant literature. Control variables utilized in all estimations include growth in GDP per capita (purchasing power parity converted GDP per capita, at current prices, expressed as a percentage); 9 the inflation rate (as a percentage) to account for macroeconomic stability; gross domestic savings (percentage of GDP) utilized as a proxy for capital formation; urban population (as a percentage); internet users (per 100 people); openness to international trade ((exports + imports)/gdp, as a percentage); unemployment rate (as a percentage); and general government final consumption expenditures (percentage of GDP), to proxy for the size of government. 10 Based on the theoretical considerations of section III and the variables described above, the estimated model is thus: n I it = β 0 + β 1 D it + β 2 D 2 it + β t X it + μ it i=0 [1] where I is the index of political and civil liberties or the index of economic freedom, D is the measure of fiscal decentralization (revenues or expenditures-based), X represents the control variables described above, 8 The four broad categories that serve as benchmarks for constructing the index are rule of law, limited government this category includes the theme of fiscal freedom, which measures the tax burden imposed by government but does not consider or account for the extent of a decentralized government structure regulatory efficiency, and open markets. It ranges from 0 to 100, where lower numbers represent lower levels of economic freedom and higher scores represent higher levels of economic freedom. Specifically, 0-49.9 (repressed); 50-59.9 (mostly unfree); 60-69.9 (moderately free); 70-79.9 (mostly free); and 80-100 (free). Available at: http://www.heritage.org/index/. 9 From 1972 to 1989, the Penn World Tables is the primary source (available at: https://pwt.sas.upenn.edu/php_site/pwt_index.php). The World Bank through its World Development Indicators is the main source thereafter (available at: http://data.worldbank.org/data-catalog/world-development-indicators). 10 Source for all these data is the World Bank (World Development Indicators). 11

and μ is a standard error term. As can be observed, fiscal decentralization is deemed to have a non-linear effect on the dependent variable, indicating that the impact of fiscal decentralization may not be constant over time. The concept of non-linearity on the variables is particularly relevant and applicable to the idea of fiscal decentralization reflected in its wide utilization throughout the economics literature as this process is bound to have a learning curve and hence the expected, positive impacts of decentralization may not be immediately evident. With respect to the model specification, due to the limited amount of information and the gaps within an unbalanced panel data set, it was determined that neither fixed nor random effects were appropriate. Furthermore, the very wide variability of data for the countries studied and the likely correlation of observations within a particular cross-section were assumed to be better addressed by generalized least squares, instrumental variable and GMM specifications that account for cross-section and within a given cross-section heteroscedasticity and autocorrelation. 11 For instrumental variable estimations, the instruments selected were one-period lagged values of all explanatory variables. Generalized Method of Moments (GMM) was employed in order to analyze the dynamic nature of the relationship between fiscal decentralization and political, civil and economic liberties. V. Results of the empirical estimation: the impact of fiscal decentralization on liberties The impact of fiscal decentralization on political and civil liberties is analyzed first. Table 1 reports regression results when all eleven American countries are taken into account. As described in equation (1), the dependent variable in all regressions is the combined ratio of political and civil liberties which ranges from 0.18 (most free) to 1.0 (least free). 12 11 Specifically, for all GLS, IVs and GMM regressions, the Prais-Winsten Panel Corrected Standard Error (PCSE) Within estimator was estimated in order to correct for serial correlation (cross section and between period correlation). 12 The combined ratio of political and civil liberties was constructed as follows: Freedom House assigns a score of 1 to 7 for political rights (1 = highest degree of freedom; 7 = lowest degree of freedom) and a similar score of 1 to 7 for civil liberties (1 = highest degree of freedom; 7 = lowest degree of freedom). Additionally, it provides a qualitative 12

(Insert Table 1 about here) The five columns on table 1 report estimates for five different regression techniques. In the first one, a GLS specification is estimated correcting for both cross-section heteroscedasticity and contemporaneous correlation and with estimates of coefficient standard errors and covariances that handle cross-section correlation. The second column reports GLS estimates correcting for both cross-section heteroscedasticity and contemporaneous correlation but with conventional estimates of coefficient standard errors and covariances. The third and fourth column report results utilizing instrumental variables (2SLS) accounting for both cross-section heteroscedasticity and contemporaneous correlation, with estimates of coefficient standard errors and covariances handling cross-section correlation (column 3) and with conventional estimates of coefficient standard errors and covariances (column 4). Column 5 reports GMM specifications correcting for both cross-section heteroscedasticity and contemporaneous correlation and with estimates of coefficient standard errors and covariances that handle cross-section correlation. Additionally, each column contains two regression results: the first one reports estimates when the indicator of fiscal decentralization is revenue based, and the second one when the fiscal decentralization variable is expenditure based. From the estimation results reported, it is evident that both the revenue and the expenditure fiscal decentralization variables have similar impacts on the dependent variable. In all cases for GLS and IV estimations these variables are positive (GLS and IV estimates) and statistically significant (GLS estimates), indicating that fiscal decentralization, at least initially, has a negative impact on political and civil liberties (again, important to note that as the combined ratio of political and civil liberties increases, these liberties worsen, and viceversa). As time passes, however, and presumably as fiscal decentralization becomes more entrenched, it begins to have the expected positive impact as evidenced by the negative (GLS and IV estimates) and statistically significant (GLS estimates) coefficients of the squared variables for both assessment of the overall degree of freedom of each nation. The three broad categories that comprise this qualitative assessment were assigned a numerical value (free = 1; partly free = 2; not free = 3). By summing the individual scores of each country in these 3 categories and dividing it by the highest possible score that a country could get (17 = 7 + 7 + 3), a combined ratio of political and civil liberties is obtained for all countries of interest. The higher (lower) this combined ratio is, the lower (higher) the degree of political and civil liberties in a country. 13

fiscal decentralization indicators. Though the size of all coefficients is small and similar in absolute size regardless of whether GLS or IV are used, they do conform to the hypothesized positive relationship between fiscal decentralization and political and civil liberties, with the caveat that the positive impact may in fact take time to fall into place. The GMM fiscal decentralization coefficient estimates run in opposite directions to their counterparts on the GLS and IV specifications, but are statistically significant only when decentralization occurs on the expenditure side. This latter result may indicate that in a dynamic framework, the impact of decentralization on the dependent variable may be more subtle. The behavior of the control variables is also noteworthy. The inflation rate and unemployment rate estimates show that they are a detriment to political and civil liberties. The coefficients for both variables regardless of whether the revenue-based or expenditure-based fiscal decentralization variable is included in the specification are consistently positive (GLS, IV, and GMM) 13 and statistically significant (GLS and IV estimates for inflation, GLS for unemployment), indicating that as inflation and unemployment worsen, political and civil liberties worsen as well. The other side of the coin is that as a country becomes more urbanized and open to international trade, political and civil liberties improve evidenced by negative and statistically significant estimates for both variables with GLS estimations noting to the importance of both greater urbanization and exposure to international markets. There is some evidence that gross domestic savings hinder political and civil liberties as evidenced by the positive and significant estimate for this variable when GLS and the revenue-based fiscal decentralization variable are used which may partly be explained by resources being spent on capital infrastructure rather than on institutions that would boost basic liberties. Additionally and as evidenced by the negative and statistically significant coefficients when GLS and expenditure-based fiscal decentralization indicators are used, there is some indication that general government final consumption expenditures a proxy for the size of government may play a positive role in enhancing political and civil liberties, though the size of the coefficient and the lack of consistency with 13 The coefficient for inflation rate when IV and the revenue-based fiscal decentralization are used is negative but statistically insignificant; the coefficients for the unemployment rate when GMM is used with both revenue-based and expenditure-based fiscal decentralization indicators are negative and statistically insignificant as well. 14

alternative estimation methods draw some caution into the implications of this result. Internet usage may also dampen political and civil liberties evidenced by a positive and statistically significant coefficient when GLS and expenditure-based fiscal decentralization indicators are used but the insignificant absolute size of the coefficient and the lack of consistency with alternate estimation methods also caution on the implications of this result. Finally, the growth of GDP per capita seems to have negligible impact on the dependent variable, regardless of the method used in estimating the model. Table 2 shows the same estimations reported in table 1, but excluding Canada and the United States, the two nations with arguably highest political and civil liberties in the sample. (Insert Table 2 about here) The results utilizing the revenue-based fiscal decentralization indicator are similar to those obtained when all countries are considered. Specifically, in most cases fiscal decentralization first worsens (evidenced by a positive and statistically significant GLS and IV coefficients for revenue decentralization) but eventually improves political and civil liberties (evidenced by a negative and statistically significant GLS and IV coefficients for revenue decentralization squared). The important distinction that exists when only developing American countries are considered occurs when the expenditure-based fiscal decentralization indicator is used. The findings indicate that expenditure decentralization initially improves political and civil liberties (evidenced by a negative and statistically significant GLS and GMM coefficients for expenditure decentralization) but subsequently hampers these liberties (shown by positive and significant GLS and GMM coefficients). Though the negative impact is small as evidenced by the absolute size of the coefficients it is an interesting result that may reflect the lack of fiscal discipline most of these countries have experienced throughout their history, which not only has negatively affected their economic stability but, as the results obtained here show, the basic liberties of their people as well. 14 14 Lack of restraint in public expenditures has been a principal trigger behind fiscal imbalances in the Americas, hence it is not surprising that a lack of fiscal discipline would be reflected in the expenditure-based fiscal decentralization indicator. 15

The behavior of control variables is also noteworthy. As was the case when all countries were analyzed, the inflation rate, national savings, and the unemployment rate seem to worsen political and civil liberties, notwithstanding the indicator of fiscal decentralization used. Likewise, urban population, openness to international trade, and general government final consumption expenditures seem to play a positive role in enhancing these basic liberties. 15 However, unlike the situation when all countries are included in the analysis, growth in GDP per capita and the level of internet usage do play a positive role in improving political and civil liberties (as evidenced by negative and statistically significant GLS coefficients), pointing to the importance of economic growth and access to the worldwide web as drivers of basic liberties, especially in developing countries. The impact of fiscal decentralization on economic freedom is analyzed in tables 3 and 4. In all cases the dependent variable is the index of economic freedom described in Section IV, 16 and the explanatory variables and estimation techniques are the same as those utilized with political and civil liberties (tables 1 and 2). Table 3 reports the results of the relationship between fiscal decentralization and economic freedom when all countries are taken into consideration. (Insert Table 3 about here) The results indicate that, despite whether the fiscal decentralization indicator being used is revenue or expenditure based, decentralization first hampers (evidenced by negative and statistically significant GLS coefficients) and eventually enhances economic freedom (denoted by positive and statistically significant GLS coefficients). 17 A plausible explanation for this behavior rests on the assumption that, as was argued before, decentralization takes time to work and therefore the positive impacts of devolving fiscal 15 The GMM coefficient for general government final consumption expenditures when the expenditure-based decentralization indicator is used is positive and significant, cautioning that at a dynamic level, the impact of government activities on liberties may be counterproductive. 16 As described in Section IV, higher (lower) values of this index reflect higher (lower) levels of economic freedom. 17 The results with IV are not statistically significant, but the sign and the size of the coefficients very much mirror those obtained with GLS, hence lending credence to the GLS results. The GMM estimates, on the other hand, are difficult to interpret as they are neither statistically significant nor follow the patterns of both GLS and IV estimates. 16

responsibilities to sub-national levels of government only become evident as decentralization settles in and begins to function properly. Another way of expressing the same thing is by noting that people living in communities with increased revenue and expenditure responsibilities do not immediately realize that their individual actions will affect the decisions made by their regional governments in terms of deciding on regional sources of revenue and on expenditure decisions on regionally-focused programs, but as they do, their individual economic freedoms as well as their political and civil liberties, which, as demonstrated in table 1, tend to evolve in similar gradual manner increase as well. The behavior of the control variables is also quite interesting. As was the case with political and civil liberties when all countries were included in the analysis, inflation rate and national savings seem to have a negative impact on economic freedom (evidenced by negative and statistically significant GLS coefficients), 18 while greater urbanization tends to have the opposite, positive impact on the dependent variable (evidenced by positive and statistically significant GLS coefficients). There is also evidence that internet usage increases economic freedom (as displayed by positive and significant GLS and GMM coefficients when paired with the expenditure-based fiscal decentralization indicator), and that the size of government hampers it (evidenced by negative and statistically significant GLS coefficients for general government final consumption expenditures). A curious result occurs with how unemployment affects economic freedom. The positive and statistically significant coefficients for the unemployment rate (GLS estimates) seem to indicate that as unemployment increases economic freedom increases as well, which may be plausible under certain very limiting and debatable circumstances. If being unemployed implies greater freedom to relocate, then perhaps this result is not that surprising. An important final result is the way in which economic growth affects the dependent variable. For most specifications, the coefficient for growth in GDP per capita is negative, and it is also statistically significant within a dynamic specification 18 The GMM coefficient for inflation rate is positive and statistically significant when paired with the expenditurebased fiscal decentralization indicator, implying that on a dynamic level inflation may have a more nuanced impact on the dependent variable. When the revenue-based fiscal decentralization variable is used, however, the inflation rate coefficient is negative (though statistically insignificant). 17

and when decentralization occurs on the expenditure side. This result may be rationalized in terms of income inequality, an issue of great concern throughout the Americas. It can be sensibly argued as Feng and Hsiang do (1998) that if growth in income per capita is not accompanied by greater income equality, then freedom will suffer as well, which is the situation reflected here. Finally, openness to international trade does not show any discernible impact on the dependent variable, at least when all countries are included in the analysis. Table 4 shows the same estimations from table 3, but excluding Canada and the United States, the two developed countries in the sample and also the ones with arguably higher levels of economic freedom. (Insert Table 4 about here) As is clearly evident, decentralization regardless of whether the decentralization indicator is revenue or expenditure based first increases economic freedom (indicated by positive and statistically significant GLS coefficients for revenue and expenditure decentralization) but eventually decreases it (evidenced by negative and statistically significant GLS coefficients for revenue and expenditure decentralization squared). 19 This result may reflect the many challenges that developing nations have encountered in establishing well-functioning, long-lasting, coherent decentralization regimes. As Litvack, et al. (1998) and Fritzen and Lim (2006) point out, if the institutional capacity does not evolve properly, wealth inequality is not addressed effectively, and political accountability does not improve, then decentralization is not likely to achieve the stated goals of greater efficiency and equity in administering public resources, which should be reflected in greater levels of economic freedom. The results presented here highlight that economic freedom can be the unintended consequence of badly managed decentralization systems. 19 The GMM coefficients for expenditure and expenditure decentralization squared are, respectively, negative and positive (both statistically significant), which runs in exact opposite direction of the GLS coefficients for the same specification. A plausible explanation for this result rests on the dynamic nature of the specification and may be an indication of the subtle ways in which decentralization affects not only economic freedom but political and civil liberties as well. The behavior of IV estimates very much resemble those of GLS estimations, but are not statistically significant. 18

The behavior of the control variables tend to reinforce the results obtained with fiscal decentralization indicators. Growth in GDP per capita is once again shown to dampen economic freedom (indicated by negative and statistically significant GLS and GMM coefficients), implying that if the issue of wealth redistribution is not properly addressed, economic growth like an inefficient decentralization structure will cause economic freedom to stagnate. Likewise, internet usage seems to have a negative impact on the dependent variable (as illustrated by negative and statistically significant GLS coefficients), demonstrating that if access to internet is not widespread, then it will not increase economic freedom but will instead hamper it. The inflation rate, national savings, and the size of government have a similar, negative impact on economic freedom (evidenced by negative and statistically significant GLS coefficients), resembling the situation when all countries were included in the analysis. Similarly, there is some evidence that the unemployment rate has a positive impact on economic freedom (shown by a positive and statistically significant GLS coefficient) and that as countries become more urbanized and open to international trade (indicated by positive and statistically significant GLS coefficients), economic freedom increases. 20 Finally, it is useful to compare the behavior of the eleven countries analyzed in order to point out their perceived differences and similarities. Table 5 presents a comparison of each country s economic freedom and political and civil liberties relative to the United States. (Insert Table 5 about here) Each country s index of economic freedom and combined ratio of political and civil liberties were regressed against their corresponding decentralization indicators (revenue and expenditure indicators, in levels and squared-values). Assuming a fifty percent share of sub-national revenues (expenditures) to general government revenues (expenditures) for all countries allows a determination of the level of economic freedom and political and civil liberties for each nation. Comparing each result to the level of economic 20 The GMM coefficients for the inflation rate and internet users when the expenditure-based fiscal decentralization indicator is used run in opposite directions to the GLS and (most) IV estimations, reflecting, once again, the more subtle impact of these variables on the dependent variable in the context of a dynamic framework. 19

freedom and political and civil liberties attained in the US (as the benchmark, the values for the US are 1.0), generates the values reported in table 5. 21 For the most part, the results obtained are the expected ones. Regarding political and civil liberties, only Canada shares the same degree of liberty as that of the US, denoted by a value of 1.0. The rest of the countries have a lower perceived rank as their values are consistently above the one for the US. It is interesting to note that some of those nations faring relatively worse have also experienced longer and more repressive dictatorial regimes (Chile, Bolivia, Peru) and/or have also lived through more severe civil conflicts (Colombia, Peru), though this apparent pattern is not always true, as is plainly evident with Costa Rica. Similarly, the results for economic freedom are largely unsurprising, with most countries experiencing a lower degree of economic freedom denoted by values below 1.0 compared to the US. Interestingly, Chile and Costa Rica report values considerably higher than 1.0, perhaps reflecting the relatively liberal economic regime installed in Chile during the 1970s which has persisted through time and the significant structural reforms adopted by Costa Rica during the last two decades, including the signing of several Free Trade agreements. El Salvador and Paraguay also have high marks, though the meaning of these results may be more nuanced and linked to other factors, including the size of their informal economies, which, under some estimates, represent more than sixty percent of their GDP, among the highest in the hemisphere. 22 VI. Conclusions and policy implications This study analyzes the impact of fiscal decentralization on economic freedom and political and civil liberties in eleven countries in the Americas. Along with a host of control variables, indicators of economic 21 It is worth remembering that the combined ratio of political and civil liberties implies that as this ratio increases (decreases), political and civil liberties decrease (increase), and that higher (lower) values for the index of economic freedom imply higher (lower) degrees of economic freedom. With this clarification, the meaning of the values in table 5 are self evident. 22 A good analysis of the importance and size of the informal economy in Latin America and the Caribbean is offered by Vuletin (2008). 20

freedom and political and civil liberties were regressed against revenue-based and expenditure-based fiscal decentralization indicators for two sets of countries: one containing all eleven nations in the sample, and a second one excluding Canada and the United States. Regarding the impact of fiscal decentralization on political and civil liberties, the findings reported here generate the following conclusions. First, when all countries are included in the analysis, fiscal decentralization regardless of whether the indicator for fiscal decentralization is revenue- or expenditurebased initially worsens but eventually improves political and civil liberties, underlining the importance of fiscal decentralization as a driver for achieving basic liberties. When Canada and the US are excluded, however, there is considerable evidence that fiscal decentralization may eventually be a detriment for political and civil liberties, particularly when the fiscal decentralization indicator is expenditure-based. This result highlights the challenges faced by developing countries in the Americas aiming to establish sustainable and credible fiscal decentralization structures, and the importance of fiscal discipline as a prerequisite for a successful decentralization regime. Regardless of whether all countries or only developing American countries are considered, inflation, unemployment, and national savings seem to hinder political and civil liberties, while an urbanized population and an open economy seem to augment them. Moreover, for developing American countries, access to internet and an active public sector were found to enhance these liberties. With respect to the impact of fiscal decentralization on economic freedom, important distinctions are also evident between the situation when all countries are included in the analysis and when Canada and the US are excluded. When all countries are included, fiscal decentralization regardless of whether the fiscal decentralization indicator is revenue- or expenditure-based first hinders but eventually increases economic freedom, emphasizing the point that a decentralization regime takes time to develop and function properly. When Canada and the US are excluded, fiscal decentralization despite the type of fiscal decentralization indicator being used initially increases but ultimately hampers economic freedom, demonstrating that, as was the case with political and civil liberties, if the fiscal decentralization regime 21