The political benefits of decentralization: Multi-tier governments, multi-level elections, and regime stability
TABLE OF CONTENTS
iv LIST OF TABLESIX
ix LIST OF FIGURESXI
xi CHAPTER 1 POLITICAL BENEFITS OF DECENTRALIZATION 1 Rationale for Decentralization 2 Does Decentralization Return on Its Promises? 6 The Search for Political Benefits of Decentralization 10 Overview of Chapters 13 CHAPTER 2 DECENTRALIZATION, CITIZENS’ ATTITUDES
15 TOWARD GOVERNMENT, AND REGIME STABILITY 15 Defining Decentralization 16 A Theoretical Approach to the Political Benefit of Decentralization 17 Theory and Hypotheses 27 Theoretical Implications 33 Summary 36 CHAPTER 3 THE EFFECTS OF DECENTRALIZATION ON CITIZENS’
37 ATTITUDES TOWARD GOVERNMENT: AN EMPIRICAL TEST 37 Empirical Testing 38 Data and Method 43 Results and Interpretation 49 Decentralization Matters?: Canada vs. England 73 Summary 83
CHAPTER 4 PERCEPTION OF RESPONSIBILITY: ITS FORMATION
85 AND CONSEQUENCES FOR CITIZENS’ ATTITUDES 85 The Puzzle 85 The Formation of Responsibility Perception 87 How does Perception of Government Responsibility affect Citizens’ Attitudes? 110 Discussion 127 Summary 130 CHAPTER 5 THEORIES EVALUATIONS AND IMPLICATIONS 131 Evaluating the Theories 132 Implications 139 CHAPTER 6 CONCLUSION: AN EPILOGUE 147 REFERENCES 151 APPENDICES 158 Appendix A 158 Appendix B 161
LIST OF TABLES Table 3.1 Descriptive Statistics: Independent Variables 50 Table 3.2 Satisfaction with Democracy: Canada 55 Table 3.3 Predicted Probabilities of Post-Electoral Statuses on Satisfaction
57 with Democracy – Rest of Canada (ROC) 57 Table 3.4 Satisfaction with Democracy: England 59 Table 3.5 Predicted Probabilities of Post-Electoral Statuses on Satisfaction
60 with Democracy – England 60 Table 3.6 External Efficacy: Canada 63 Table 3.7 Predicted Probabilities of Post-Electoral Statuses on External Efficacy –
65 Rest of Canada (ROC) 65 Table 3.8 External Efficacy: England 67 Table 3.9 Predicted Probabilities of Post-Electoral Statuses on External Efficacy – England 69 Table 3.10 Internal Efficacy: Canada 70 Table 3.11 Predicted Probabilities of Post-Electoral Statuses on Internal Efficacy –
72 Rest of Canada (ROC) 72 Table 3.12 Internal Efficacy: England 74 Table 3.13 Predicted Probabilities of Post-Electoral Statuses on Internal Efficacy – England 76 Table 4.1 Multinomial Logit Estimates of Responsibility Perceptions
99 (with the provincial government as referent category) 99 Table 4.2 Predicted Probabilities of Responsibility Perceptions: Political Sophistication 102 Table 4.3 Predicted Probabilities of Responsibility Perceptions: Citizens’ Policy Preferences 102 Table 4.4 Logit Estimates of Responsibility Perceptions 104
Table 4.5 Predicted Probabilities of Responsibility Perceptions:
106 (Model 1 – 0 if other governments): Political Sophistication 106 Table 4.6 Predicted Probabilities of Responsibility Perceptions:
108 (Model 2 – 0 if federal government): Political Sophistication 108 Table 4.7 Predicted Probabilities of Responsibility Perceptions:
108 (Model 2 – 0 if federal government): Policy Preferences 108 Table 4.8 Satisfaction with Democracy 114 Table 4.9 Linear Combination of Multiplicative Term 115 Table 4.10 Predicted Probabilities of Satisfaction with Democracy: Rest of Canada 117 Table 4.11 External Efficacy 119 Table 4.12 Linear Combination of Multiplicative Term 120 Table 4.13 Predicted Probabilities of External Efficacy: Rest of Canada 121 Table 4.14 Internal Efficacy 124 Table 4.15 Linear Combination of Multiplicative Term 125 Table 4.16 Predicted Probabilities of Internal Efficacy: Rest of Canada 126 Table 5.1 Summary of Results of Hypothesis Testing – The Direct Effect of Institution 134 Table 5.2 Summary of Results of Hypothesis Testing – The Interaction Effect
137 of Responsibility Perception 137
LIST OF FIGURES Figure 2.1 Winners’ and Losers’ Attitude toward Government (One-Level Election) 19 Figure 2.2 Hypothesized Winners’ and Losers’ Attitude toward Government
29 (Two-Level Elections) 29 Figure 3.1 Satisfaction with Democracy among Canadians and
46 English Electorates (percentage) 46 Figure 3.2 Descriptive Statistics: Dependent Variables – Satisfaction with Democracy,
46 Internal Efficacy, and External Efficacy 46 Figure 3.3 Post-Electoral Status of Canadian Electorates (2004) and
47 English Electorates (2005) (percentage) 47 Figure 4.1 Responsibility Perceptions for Health, Education, and Social Welfare 95 Figure 4.2 Distribution of Index of Political Sophistication 95 Figure 4.3 Distribution of Index of Citizen’s Policy Preferences 97 Figure 4.4 Predicted Probabilities of Responsibility Perceptions: Political Sophistication 100 Figure 4.5 Predicted Probabilities of Responsibility Perceptions:
100 Citizens’ Policy Preferences 100 Figure 4.6 Predicted Probabilities of Responsibility Perceptions
104 (Model 1 – 0 if other governments): Political Sophistication 104 Figure 4.7 Predicted Probabilities of Responsibility Perceptions
106 (Model 2 – 0 if federal government): Political Sophistication 106 Figure 4.8 Predicted Probabilities of Responsibility Perceptions
107 (Model 2 – 0 if federal government): Citizens’ Policy Preferences 107
CHAPTER 1 POLITICAL BENEFITS OF DECENTRALIZATION
The inherent inability to take direct account of circumstances of time and place make central planning inferior to decentralized, market mechanisms. In the latter case, decisions can be left to the “man on the spot,” who is motivated to act on the “intimate knowledge of the facts of his immediate surroundings” in the pursuit of profits or individual interest. Friedrich Hayek (1945), The Use of Knowledge in Society
…, although democracy strives for equality in opportunity to participate in electoral contests, it also is unavoidably unequal in the outcomes it produces. Christopher J. Anderson et al. (2005), Losers’ Consent: Elections and Democratic Legitimacy
In recent decades, countries around the world, regardless of their developmental status, have committed themselves to a variety of decentralization reforms. Does decentralization return on its promises? Not surprisingly, many studies have been conducted to evaluate the costs and benefits of decentralization. Nevertheless, most of those studies have examined economic benefits; for example, the quality of public service delivery, the pervasiveness of corruption, and economic performance. Further, not only are the fewer studies on political benefits, but the contradictory results of those studies have made clear that there are no solid agreement on the outcomes of decentralization. As a result, this literature remains incomplete. Failure to systematically and empirically examine the political benefits of decentralization leaves political scientists and policymakers with insufficient knowledge about the consequences of decentralization. This dissertation seeks to fill these gaps by explicitly modeling the role decentralization plays in shaping citizens’ attitudes toward political system. Drawing on work in political behavior and decentralization, a theoretical framework is developed to explain the manner in
which citizens’ attitudes are shaped by election outcomes and their post-electoral win-loss status in multi-tier government. This theoretical framework will not only provide the basis for conducting the empirical work on the evidence for the political benefits of decentralization that heretofore have been understudied, but also theoretical implications and predictions on the linkage between decentralization and democratic stability – and thus enable us to understand how decentralized political structures might incur to promote greater stability of a democratic regime. To a large extent, the results produced by this dissertation can provide policy guidelines for policymakers and other interests who are in deciding whether decentralization should be pursued or disregarded as one of the options for institutional reform.
Rationale for Decentralization Decentralized systems of government have been applied in many countries around the globe for quite some time. In Latin America, subnational government officials now possess a historically unprecedented level of political and fiscal autonomy in order to make policies in their subnational jurisdictions (Montero and Samuels 2004). In Asia, decentralization has been an active policy issue in Thailand since 1992 and about 6,700 local governments were created in 1994. Decentralization in Indonesia, during the post Suharto era, has been referred as ‘Big Bang’ decentralization; it is rapidly moving the country from one of the most centralized states in the world to one of the most decentralized (Hofman and Kaiser 2004, see also Alm, Martinez- Vazquez, and Mulyani Indrawati 2004). In Africa, Ethiopia and South Africa have chosen variants of a federal system. In other African countries, the introduction and re-introduction of decentralized government is taking place along with elections for local councils (Brosio 2002, Boex and Martinez 2006). In Europe, joining Germany, Austria, and Switzerland, Spain turned
itself into a quasi-federal state in the late 1970s while Belgium transformed itself from a unitary state to a federal one. Furthermore, parliamentary assemblies have been introduced in Scotland and Wales and regional governments were created in Italy since the 1970s. Thus, it is readily apparent that decentralization is happening and is in fashion. But, what are the rationales for decentralization? At the practical level, rationales behind decentralization vary from country to country. It is safe to say that no single factor prompts decentralization in any particular country. In other words, there is no single cause that is sufficient and necessary to make decentralization happen. It is rather a combination of multiple factors (Manor 1999). Decentralization can be a result of internal reasons and external attempts. Internally, decentralization reforms derive from divisions in historical, linguistic and cultural roots. Or, it could arise for political reasons; for example, political elites who are in power see decentralization as one way in which to enhance their continued political survival (e.g., when decentralization helps them increase the probability of reelection by creating local partisanship support). On the other hand, decentralization can be demanded by subnational political elites who see decentralization as a way to gain more autonomy from the central government (see Montero and Samuels, 2004). Moreover, a shift from centralized to decentralized structures of government in many developing countries is often a result of international influence (e.g., as a condition for receiving grants in aid from international organizations such as the World Bank (WB), the United States Agency for International Development (USAID), or the European Union (EU). However, on theoretical grounds, political scientists see multi-tier government structures as a system that complements democracy. By layering the level of government, decentralization will not only increase citizen participation but also create civic virtue. In other words, multi-level
governments change the dynamic and relationship between citizens and their government. According to Downs (1957), a rational voter would not want to participate in elections (central national elections) since his/her ‘single’ vote will not likely have a decisive effect on the electoral result. However, at the subnational election level, expected utilities from participating in the elections are not the same as at the national level since voters at this level might be able to control or achieve a better outcome by coordinating and creating better voting strategies. Therefore, the chance of any given vote being pivotal is greater in subnational elections (Brock 2002). As a result, one would expect higher turnout in subnational election 1 . Moreover, at the subnational level, citizens may realize that local issues are not too trivial. In turn, this realization will give them the inspiration to attend local political meetings, file petitions, become politically active, and turn out to vote. As government size is reduced in a decentralized system, the ability of citizens to hold government accountable is increased. 2 Governments are ‘accountable’ if citizens are able to correctly evaluate the (incumbent) government so that they can decide to keep them in office or throw them out (Powell 2000). In a decentralized system, some government units are reduced in scale and therefore, become closer to citizens leading to a clarity of responsibility. In other words, citizens are able to absorb and evaluate subnational performance as a result of living in the local jurisdiction. Further, a federal form of government may be a preferred form of government in states which are multi-lingual and in which ethnic cleavages exist. According to Rodden (2006, 5), federalism is preferred in such states comes from the reason that “while a
1 However, empirically, voter turnout is lower, not higher, at subnational levels in many countries. In fact, there are two opposing forces at play: “high VS low stakes” and “lower VS higher probability of being decisive.” 2 Actually government size in terms of payroll/employment and budget size increases with decentralization, in this context the government size is reduced in terms of government units – with decentralized units, citizens will deal with smaller governments units for some of the issues that concern them.
single sovereign might be tempted to abuse its authority, federalism provides a valuable protection by dividing power among multiple, competing sovereigns.” In the eyes of economists, a multi-tiered structure of government is created not only to better respond to citizens’ preferences in public services and goods, which generally are spatially heterogeneous, but also to provide these services and goods in a more cost-effective manner. This is often called the “decentralization theorem” (Oates 1972, 35). In other words, according to Oates, “… state and local governments, being closer to the people, will be more responsive to the particular preferences of their constituencies and will be able to find new and better ways to provide these [public] services” (Oates 1999, 1120). Moreover, multi-tier governments combined with household mobility are likely to bring about competition among subnational governments in order to attract citizens that will result in a better government (Tiebout 1956). Like a private firm in the market, each subnational government is motivated to compete in providing public goods and services that are responsive to the demands of its citizens and operate in a way that is innovative, cost-effective and accountable. In short, for economists, two principle rationales for decentralization exist: it better matches public goods and services to citizen demands; and it encourages an efficient and accountable form of government. In sum, the rationales for decentralization can boil down to stability, peace, efficiency, responsiveness and accountability. In other words, multi-tier government is likely to provide citizens with public goods and services that they prefer at a lower cost; to promote competition among subnational governments in providing public services; to protect minorities with different demands and preferences; and to increase the quality of democracy.
Does Decentralization Return on Its Promises? In recent decades, countries around the world, regardless of their developmental status, have committed themselves to a variety of decentralization reforms. As time goes by, one cannot help but ask the question: Does decentralization return on its promises? As mentioned above, many rationales are behind the decision to decentralize. It could be for economic, political, social, or mixed reasons. However, when it comes to evaluating the outcomes of decentralization, the majority of these evaluations are focused on economic outcomes. The question of whether or not decentralization improves efficiency in public services provisions is among the most studied. Just to cite a few of these studies, Strumpf et al (1999) 3
examine the change in public health provisions in Uganda. They find decentralization to have a negative impact on the provision of this public good. On the other hand, a study by Robalino et al (2001) shows that greater fiscal decentralization, in terms of greater involvement of local government in managing public expenditures, is associated with lower infant mortality rates. However, they note that one should not expect a direct relationship between greater fiscal decentralization and improved health outputs. But, one should expect that if institutional capacity at subnational levels is improved, then fiscal decentralization is likely to improve health outputs. In the same vein, Schwartz et al (2002) find that the increase in public health expenditures at local government levels in the Philippines resulted in a positive impact on the use of planning and the level of child immunizations. Education is another public service sector that often is devolved to subnational governments once countries have decentralized. A number of studies focus on the allocative efficiency of decentralization in the education sector, while others focus on direct outputs, such
3 The updated version of this paper is published in Journal of Development Studies as Akin, Hutchinson and Strumpf (2005).
as enrollment ratio and attainment levels (Faguet 2004; Faguet and Sanchez 2006; Heredia-Ortiz 2005; Sole-Olle 2005; Barankay and Lockwood 2007). Habibi et al (2003) employ a ratio of students enrolled in secondary school per 1000 primary students in Argentina (as a proxy for human development indicators) and find that decentralization measures have positive and significant impacts on enrollment ratios. In addition, they find that greater fiscal decentralization is associated with continued improvements in human development at the provincial level. A study by Faguet and Sanchez (2006) also find positive and significant evidence that decentralization improves public school enrollment in Colombia. They further discover that enrollments actually increase in districts where central government has less involvement in educational finance and policy-making process. In Switzerland, Barankay and Lockwood (2007) demonstrate that decentralization, in fact, improve educational outcomes. They find that the degree of decentralization is positively associated with educational attainment. The influence of decentralization and fiscal decentralization on economic growth is another question that has received a significant amount of attention in empirical studies in the past few decades. To some extent, in many developing countries fiscal decentralization is seen as part of a country’s development strategy. Thus, determining the impact of fiscal decentralization on economic growth is not only an academic question but also an important policy issue. To determine the impact of decentralization on economic growth, many scholars have employed macroeconomic stability as a proxy of growth – since symptoms of macroeconomic instability such as higher rates of inflation and unemployment generally lead to lower rates of economic growth. However, the empirical evidence on the relationship between fiscal decentralization and macroeconomic stability is inconclusive (Martinez-Vazquez and McNab 2003). Fornasari, Webb, and Zou (2000) find a close relationship between increases in
subnational deficits and central government expenditures and deficits in the following period. In contrast, Treisman (2000) and Rodden and Wibbels (2002) find no clear relationship between the rate of inflation and decentralization. However, a study by Martinez-Vazquez and McNab (2005) shows that fiscal decentralization increases macroeconomic stability. They found that after controlling for a change in money supply, revenue decentralization is associated with lower inflation. A very interesting study of decentralization and growth in Organization for Economic Co-operation and Development (OECD) member countries by Thiessen (2003), shows a non- linear relationship between decentralization and economic growth suggesting that when a country moves from low level to medium levels in the degrees of decentralization, growth accelerates, but when a country moves to a high degree of decentralization, growth reduces (See also Qiao, Martinez-Vazquez, and Xu 2002 and Martinez-Vazquez and McNab 2003). Another crucial factor related to decentralization that affects economic growth is corruption. Since decentralized government is reduced in size and is closer to people, one would expect greater accountability of government officials to the people they represent. Consequently, decentralization should result in less corruption. The empirical studies on this relationship, however, have also found mixed results. On the one hand, Huther and Shah (1998) find a correlation between decentralization and lower corruption levels. Several other empirical studies confirm this finding: Gurglur and Shah (2002), Fisman and Gatti (2002), and Arinkan (2004). However, Bardhan and Mookherjee (2002) suggest that decentralization can result in local elite capture and thus result in more corruption. Moreover, Treisman (2007) argues that the relationship between decentralization and lower corruption found in the works of Huther and Shah (1998); Fisman and Gatti (2002); and Arinkan (2004) may be spurious because these works “… [do not] simultaneously control for two factors that are both correlated with fiscal
decentralization and strongly related to perceived corruption: Protestant religious tradition and a long experience of uninterrupted democracy” (2007, 251). The quality of democracy in decentralized countries is another aspect to which scholars of decentralization have paid attention. Huther and Shah (1998) found a positive correlation between a composite index of citizen participation (which consists of an indicator of political freedom and political stability) and decentralization (the local governments’ share of subnational expenditures). They further argue that “citizen participation and public sector accountability go hand in hand with decentralized public sector decision making” (1998, 12). Similarly, Tvinnereim (2005) demonstrates how decentralization improves the quality of democracy in subnational regions in Europe. 4 By using citizen satisfaction as a proxy for government responsiveness, he is able to find a positive relationship between electoral competitiveness (in regional government) and citizen satisfaction, mediated through accountability. He argues that electoral competitiveness in regional elections is an essential element in ensuring government accountability. In contrast, Foweraker and Landman (2002) examine the relationship between constitutional design and the democratic performance 5 of 40 countries over 29 years and found that unitary systems perform better than the federal system on measures of participation, property rights, and minority rights. Does decentralization succeed in reducing ethnic conflict? The empirical evidence on this question is also inconclusive. One group of scholars find evidence that decentralization increases ethnic conflict, arguing that it enables some majority groups to produce legislation which discriminates against other minority groups and that reinforces ethnic differences and inequalities by recognizing certain ethnic groups in countries and giving them a sense of
4 His cases were German Landers and Spanish Autonomous Communities. 5 By using measures of eight core values of liberal democratic government: accountability, representation, constraint, participation, political rights, civil rights, property rights, and minority rights.
legitimacy (Hardgrave 1994; Horowitz 1991; Kymlicka 1998; Leff 1999; Snyder 2000, Suberu 1994). Another group finds that decentralization reduces violent ethnic conflicts but not non- violent protests (Cohen 1997, Saideman et al 2002). In addition, Brancati (2006) argues that political decentralization can be used as a mechanism to reduce not only ethnic conflict but also secessionism. However, the development of regional parties may cause an adverse effect on both ethnic conflict and secessionism, since regional parties may reinforce minority identities and mobilize groups towards ethnic violence. Thus, she suggests that “the strength of regional parties must be regulated in countries through different features of decentralization or alternative institutional mechanisms, such as the type of electoral system in a country” (Brancati 2006, 681). In conclusion, these studies suggest that decentralization improves the quality of governance and democracy, may impact economic growth and improve efficiency in public service provisions. However, the results vary by study and although the economic outcomes produced by decentralization have been evaluated, the question remains whether decentralization should be pursued or disregarded as one of the options when countries consider institutional reform. Thus, providing the evaluation of political outcomes is an additional way to assess decentralization.
The Search for Political Benefits of Decentralization What are the political benefits of decentralization? In a search for the answer, this dissertation asks whether decentralization might promote greater stability of a democratic regime. With the search for answers to this question, the political benefits of decentralization will be revealed. To answer this question, this dissertation builds its argument by juxtaposing two
literatures: (1) citizens’ attitudes toward government are a function of their post-electoral win- loss status and (2) the rationales of decentralization. A number of studies have found relationships between electoral outcomes and citizens’ attitudes toward government (Anderson and Guillory 1997; Anderson et al. 2005; Fuchs et al. 1995; Norris 1999a). The principle finding of these studies is that in any election, citizens are given an opportunity to elect whoever they want in the governing body, but the electoral results will turn these citizens into either winners or losers (being or not being a part of the majority and minority parties in the legislative assembly). The status of these citizens, as winners and losers, is a crucial factor when it comes to giving their opinion on the political system. Citizens, who are a part of the majority and able to identify with the governing party, are more likely to display positive attitudes toward government, whether such attitudes might be satisfaction, trust, or support. In contrast, citizens who are a part of the minority may display lower levels of such attitudes because of the fact that the political parties which they supported are not a part of the government. Losing an election means parties they dislike are in government and they also lose a share of the policy-making process – their interests, beliefs, and ideas will not be reflected in the government for the next few years (Anderson et al. 2005). Thus, losers will be systematically less satisfied with government. Decentralization involves shifting some responsibilities, once in the hands of national government, into the hands of subnational governments. As decentralization reforms are implemented, the creation of additional tiers of government is a common by-product. This means that not only has another political arena been added where political actors will compete for governing power, but it also provides for many localized decision-making processes, which provide citizens with multiple chances to participate and give consent to policy-makers and for