Decentralization and Recentralization in the Developing World

In the 1980s and 1990s, much of the developing world experienced transitions to democracy accompanied by economic liberalization and decentralization of power to subnational governmental bodies. The process of decentralization has been studied intensively, but little attention has been paid so far to the recentralization that has occurred in some countries in the past decade. In this book, J. Tyler Dickovick seeks to illuminate how the processes of decentralization and recentralization are interrelated and what the dynamics of each is. He argues that decentralization occurs as a result of the decline in the power of the presidency, whereas recentralization occurs when the president resolves an extraordinary economic crisis. The processes of decentralization and recentralization, Dickovick further argues, have the same dynamics whether they occur in federal or unitary states. To test the theory, Dickovick compares a strong federal system, Brazil, with a weak one, South Africa, and compares these in turn with two unitary regimes, Peru and Senegal. Decentralization and Recentralization in the Developing World provides a much more nuanced understanding of when and why decentralization and recentralization happen, and what their importance is to intergovernmental shifts in power.

“J. Tyler Dickovick has written a pathbreaking work built on an insightful analytical framework and sustained by excellent fieldwork in several countries. He makes several notable contributions to the field of comparative studies of decentralization and federalism. First, he broadens the way that decentralization is measured in cross-national perspective. Second, he applies the theoretical framework in varied policy settings, such as health care and education, proving that it is broadly applicable. Third, in his analysis, he seriously considers differences in the sequencing of crises and policy responses, thereby adding to an emerging interest in these issues in the decentralization field. And finally, his meticulous empirical work is cross-regional as well as cross-national, making it a valuable resource for future studies in a variety of locations. Decentralization and Recentralization in the Developing World is a good example of the finest scholarship in comparative politics.”—Alfred P. Montero, Carleton College

“Why does decentralization stick in some countries but not in others? How do presidents try to recentralize authority? Under what conditions do they succeed? In the first major book to appear on the politics of recentralization, J. Tyler Dickovick offers compelling answers to these questions. By insisting that we move beyond decentralization—and by showing how we should apply a common analytical framework to the study of decentralizing and recentralizing changes—Dickovick significantly broadens the scope of the literature on intergovernmental relations. Based on extensive primary research in an unusually disparate set of African and Latin American cases, Decentralization and Recentralization in the Developing World also innovates by focusing on dynamics that had yet to be integrated into the political science literature on decentralization, including what takes place within the bureaucracy, who controls labor markets, and why subnational governments either proliferate or amalgamate in the wake of decentralization. This is a must-read for students of subnational politics.”—Kent Eaton, University of California, Santa Cruz

“J. Tyler Dickovick has made an important contribution to in-depth studies of decentralization. His detailed focus on Brazil and South Africa—a choice that will stimulate much-needed comparative research among scholars in these countries—during a period of intense institutional development and social change, with additional discussion of Peru and Senegal, will help all who are trying to follow and understand the pendulum swings that take place in national-local relations and in different policy areas. For too long there has been a tendency to view the centralization-decentralization dynamic as one in which the latter is somehow more democratic than the former and in which subnational governments are the cornerstone of democratic development. In this perspective, recentralization would normally carry negative overtones. Dickovick not only points out that things are not as simple as they might seem, but also helps us to recognize the importance of the fiscal arena, with its negotiations and bargains between very different actors at different levels, as the place in which to look for the unfolding history of the present.”—Peter Spink, Fundação Getúlio Vargas–São Paulo

“Dickovick provides extensive conceptual and theoretical discussion . . . ultimately emphasizing historical institutional dynamics and political economy.”—K. Staudt, Choice

J. Tyler Dickovick is Associate Professor of Politics at Washington and Lee University.

Contents

List of Figures and Tables

Acknowledgments

List of Abbreviations

1 Decentralization and Recentralization in Developing Countries

2. Historical Trajectories in Subnational Autonomy

3. Subnational Revenue Autonomy

4. Subnational Expenditure Autonomy

5. Subnational Contractual Autonomy

6. Subnational Autonomy in Unitary States

7. When the Center Holds: Conclusions and Implications

Appendix: Interviewees (by Country)

References

Index

Decentralization and Recentralization in Developing Countries

In the 1980s and 1990s, a wave of decentralization swept across much of the developing world. Along with democratization and economic liberalization, decentralization became a major theme in the relocation of political and economic power. The wave affected a wide range of states: historically centralized command economies and federal countries; developed and undeveloped economies; and in regions around the world, from Eastern Europe to East Asia to South America to sub-Saharan Africa. This impressive sweep of decentralization has increasingly captured the attention of scholars. In the developing world, the literature is on the front burner in Latin America, where major federations such as Brazil, Argentina, and Mexico decentralized significantly as they democratized in the 1980s and 1990s. In Africa too, democratically inclined governments in small, unitary states have increasingly looked to decentralize power to overcome the pathologies of centralized rule. Political scientists have long recognized that the power to govern depends on the control of fiscal resources, and increasingly we have needed to emphasize that this is true of both central and subnational governments: questions about which level of government controls fiscal resources have profound implications for our most fundamental theories about where power lies.

For all the recent excitement, the wave of devolution has crested in some countries, and indeed in others it has reversed. Increasingly in recent years, decentralization processes have been overturned by newly powerful central governments. This demands explanation as well. Literature on decentralization and federalism, which necessarily focus on the devolution of power to subnational levels, has not given these processes sufficient attention. Only in very recent work have scholars begun to address the pressing issues of recentralization and the consolidation of central power in the period after decentralizing reforms. We need to examine the processes of decentralization and recentralization in a single framework. It is crucial that studies of decentralization address not only when, in the words of W. B. Yeats, “the center cannot hold,” but also when the center can hold or regain political power, fiscal resources, and administrative control.

Brazil and South Africa are two countries that illustrate the phenomena of decentralization and recentralization in different ways. This study explains when and why central governments decentralize and centralize fiscal power, with a central comparison based on Brazil, a robust federal country, and South Africa, a country with a much weaker form of federalism. Both initiated political and fiscal decentralization processes as they democratized in the 1980s and 1990s, respectively. By 1996 subnational governments in both countries were run by newly empowered elected officials whose tax revenues and expenditures totaled nearly 60 percent of national totals. Subnational governments (SNGs) received constitutionally guaranteed transfers and gained the right to contract debt in capital markets. The extent of decentralization in the two countries is well documented, even if arguments about causes and consequences remain contentious. These were, in short, two of the most astonishing cases of decentralization in the developing world in the 1980s and 1990s.

Yet both countries have also experienced significant limitations on decentralization. In Brazil, overt recentralization has occurred, following economic crises that gave presidents unique opportunities to reduce subnational power. This has defied the expectations of many Brazilianists who view the country as a case of decentralism run amok, a federation where the states (estados) dominate politics and the central government is chronically weak. By contrast, South Africa’s recentralization has been more limited, but the central government there has consolidated several institutions that call into question the very significance of the decentralization process. Whereas the country’s provinces continue to manage a very high percentage of revenues, the center has held onto and reinforced its control over SNGs through tight monitoring and control of spending and borrowing. In both countries, then, central governments have succeeded in limiting and reducing the fiscal powers of subnational governments: they have trimmed borrowing powers, reduced abilities to provide patronage, managed expenditures with increasing technocratic efficiency, and passed increasingly strict statutes that criminalize subnational fiscal irresponsibility.

The importance of these questions extends far beyond these two cases, however, and even to countries with unitary systems of governance. Accordingly, I also examine two unitary countries—Peru and Senegal—in order to show that a common set of dynamics informs both federal and unitary states; I examine the period from the wave of decentralization in the 1980s to the present. Answering these questions about decentralization and recentralization gives insight into broader issues of public finance and state power by shining a light on moments when fiscal resources are redistributed.

I organize this opening chapter as follows. First, I explain how electoral decline and economic crisis alter executive power relative to actors representing subnational interests. The weakening of executives’ partisan powers drives decentralizing change. Arguments linking decentralization to partisan decline have been formulated by Kathleen O’Neill (2003, 2005) as a largely top-down process, and while O’Neill’s work underpins part of the analysis here, I also account for decentralization processes that occur “from below” in federal states. The resolution of economic crisis, on the other hand, serves as an impetus for recentralizing change, as I illustrate in a subsequent section. I show thereafter how presidential power plays out in three institutional arenas to shape subnational governments’ autonomy over their revenues, their expenditures, and their ability to set and enforce contracts. After a brief examination of alternative arguments, I discuss how historical-institutional and rational-institutional arguments must be combined to account for the various forms of subnational fiscal autonomy in each country case.

Explaining Subnational Fiscal Autonomy

As is clear from the outset, I do not deem decentralization to be an adequate analytical outcome but also seek explanations for its converse: (re)centralization. This means exploring when SNG autonomy is increased, when it is decreased, and when it remains the same. I thus conceptualize the dependent variable as subnational fiscal autonomy, or the set of fiscal relationships between the autonomous levels of government in a single country. This includes various elements of intergovernmental fiscal relations, such as revenue and tax distribution, expenditure responsibilities, and the independence of the different levels of government with respect to contracting in labor and capital markets. In the simplest terms: which levels of government control tax revenues, who is responsible for which policies, and who can borrow and hire and fire at will? In federal states, fiscal relationships will arise among three levels of government: the national (federal) government, local governments, and intermediate-level (i.e., state or provincial) governments. In unitary states, fiscal relations are usually between only two levels—the national and the local—though changes in the political and fiscal importance of regional governments also occur. This study is not about relationships between the different vertical layers of a single central state bureaucracy.

To understand subnational fiscal autonomy better, I consider three components, of which revenue autonomy is the first and most prominent. Revenue autonomy refers to how independent SNGs are from national governments for their revenues. One source of revenues for SNGs is intergovernmental transfers (IGTs). We can measure SNG autonomy by assessing whether these transfers are “automatic” or not. Do central governments have the right to impose conditions on the SNG receiving transfers? If so, this means lower revenue autonomy. On the other hand, where substantial unconditional transfers are guaranteed in the constitution, a country would have higher subnational revenue autonomy. SNGs also procure revenues from own-sources, that is, from their own taxation. Here, we can measure what taxes SNGs are allowed to collect. Can domineering central governments ensure subnational dependence by refusing to devolve tax authority? Or might SNGs have the right to collect valuable income or value-added taxes? Together, the degree of “automaticity” of IGTs and the independence over taxation give a reasonable estimate of how much revenue autonomy an SNG possesses.

Expenditure autonomy is the second element. Provincial and local governments are deemed to have greater autonomy where they have control over their spending. If provinces and municipalities have the freedom to spend their revenues on whatever they want (including patronage), they are more autonomous than their counterparts in other countries with no such discretion. Are SNGs legally assigned specific responsibilities, and are these assignments enforced? Are SNGs required to assume public goods responsibilities de jure, and do they de facto? I focus in particular on two major policy areas—education and health care—where decentralization has proved salient, thereby ensuring variation on the dependent variable. SNGs that have high expenditure autonomy are less compelled by the central government to spend their resources on specific policy areas. The measure of expenditure autonomy, then, will be taken by looking at the rules in place governing expenditures by SNGs. As discussed in chapter 4, this contrasts with the use of aggregate spending figures to measure expenditure decentralization; the reason for this is that subnational spending will correlate strongly (though not perfectly) with subnational revenues. Given a certain extent of devolution in revenues/expenditures, the question becomes how subnational spending is controlled. As we shall see, this choice becomes especially important in the context of the Brazil–South Africa comparison.

The third and final component of SNG autonomy may be termed contractual autonomy. In this category, the constitutional or legal independence of SNGs is the sine qua non for other forms of autonomy. Beyond this, there are two main types of contractual authority: capital market autonomy and labor market autonomy. SNGs can reasonably be said to have more authority over fiscal matters if they are empowered to enter into independent contracts with employees and creditors, without the explicit oversight of national governments. SNGs that control their own wage bills and debt burdens are more autonomous than those who must look to central governments to set these parameters. Can SNGs borrow on the open market? Do they have their own banks that make soft loans? Can they hire and fire, and change wages, at will? Each of these rights would be indicative of greater subnational contractual autonomy. Those with constitutional guarantees of authority in their jurisdiction are less susceptible to being dissolved or otherwise disbanded by national governments; their contractual autonomy is obviously higher in this case. Table 1.1 encapsulates several of the key questions that may be asked in assessing the measures of subnational autonomy.

The conceptual framework of subnational autonomy allows me to capture the essence of both decentralization and recentralization in a single dependent variable by highlighting increases and decreases in SNG autonomy over time. Decentralization and (re)centralization are not simply questions of cross-national comparative statics, but rather involve calculations and bargains between several sets of actors, including executives and legislators at both the national and subnational levels, with political parties and party systems playing important mediating roles and shaping the calculations of actors in the intergovernmental fiscal system. Taken together, these two sides of the coin constitute the ebb and flow of fiscal power in an intergovernmental system.

The Argument: Decline, Crisis, and the Exercise of Executive Power

Why do central governments choose to decentralize power, and under what conditions can central governments reverse the decentralization of power? I argue that subnational autonomy varies inversely with the power of chief executives, since presidents (or chief executives representing heads of state in parliamentary systems) typically wish to limit subnational autonomy, while SNGs wish to increase their own autonomy. Causally, this presidential power in turn depends on changes in the national political economy, at least in the short run where political institutions such as electoral rules and party systems remain relatively stable. Shifts in executive power occur specifically at moments of electoral decline and economic crisis. The national electoral decline of the chief executive’s party reduces presidential power; building on O’Neill’s (2005) argument, this often forces presidents to increase subnational autonomy by decentralizing resources and powers. Presidents can reverse these increases and centralize only under extraordinary circumstances. Economic crisis—specifically, the resolution of crisis—creates such an opportunity, strengthening the hands of presidents and enabling centralization efforts that would not otherwise succeed.

Presidents typically prefer not to decentralize power. In the area of intergovernmental relations, they prefer to limit subnational actors and maximize their own power and discretion over national resources. This is true in both presidential systems (such as in much of Latin America and Africa) as well as parliamentary systems (of which South Africa is a representative case here). While history suggests presidents (especially in countries such as Brazil) respond to the needs of their subnational bases of support, we may make the initial assumption that presidents by virtue of their political position prefer to limit SNG autonomy on fiscal questions. Under this assumption, decentralization is seen as a presidential initiative only in the sense that presidents act in accordance with institutionally generated outcomes. Presidents decentralize when under political duress, and even top-down decentralization must be seen as a form of constrained optimization, not a reflection of underlying presidential preferences. Presidents are more often eager to recentralize, but their ability to do so is contingent on relatively uncommon governing opportunities, as noted below. In short, I argue that decline leads to decentralization and crisis to centralization. I consider these two independent variables—electoral decline and macroeconomic crisis—and their consequences in turn.

Executive Partisan Decline and Decentralization

Governments decentralize power, and increase subnational autonomy, when presidential partisan powers weaken. This is true when presidents face national electoral defeat or require subnational support to hold together a governing coalition. At these moments of weakness, presidents are vulnerable to bottom-up pressures from subnational actors within their own governing coalitions, where these subnational interests are institutionally powerful, as in Brazil (see Eaton 2004a; Samuels 2003). Decline of the president’s national party strength also increases the attractiveness of top-down solutions whereby presidents strengthen SNGs in order to develop or strengthen regional bases for the future (cf. O’Neill 2005). From O’Neill’s perspective, presidential partisan weakness is especially salient when subnational copartisans are well positioned to win elections and have strong representation in national legislatures. That is, national governing parties decentralize when they are waning nationally, but are strong subnationally.

Democratization may be seen as a special case of governmental decline—wherein regimes themselves decline—and democratization is thus likely to bring decentralization in its wake. This may occur because “democrats are decentralizers” or because there is some “elective affinity” between democratization and decentralization, insofar as both are predicated on dispersing rather than concentrating power. While many decentralizing processes occur alongside transitions from authoritarian rule, I do not argue that democratization alone causes decentralization. I show that declining regimes facing Constituent National Assemblies (CNAs) are likely to increase SNG autonomy, just like declining governments awaiting elections. In the unitary cases (Peru and Senegal), decentralization occurred when parties anticipated defeat in regularly scheduled national elections, while in federal Brazil and South Africa, decentralization occurred during regime changes at CNAs. Decline drove decentralization in all four case countries, despite these variations in regime politics. Taking as given that the likelihood of decentralized governance increases with democracy and decreases with authoritarianism, I conduct my analysis by looking principally at variation within nominally democratic regimes and at moments of transition.

Cases where governing parties decentralized from the top down before losing national elections support a rational expectations model of intergovernmental relations. In such cases, to include South Africa and Peru in the late 1980s, decentralization not only preceded national electoral defeat, but also became increasingly salient in direct relation to the certainty of that defeat. The National Party in South Africa and the Alianza Popular Revolucionária Americana in Peru in particular sought aggressive decentralization at an increasing pace as their electoral defeats at the national level became increasingly certain.

Central governments may have electoral reasons to decentralize in the face of decline even when they are not strongest at the subnational level, however. Victoria Rodriguez argues that decentralization in Mexico was a scheme concocted out of tactical decisions to raise public support of the regime and to enhance the government’s legitimacy (Rodriguez 1997). The benefits to governments of decentralization may thus outweigh the costs, even if the national party stands to lose regional elections. Admitting this possibility means we retain the analytical emphasis on the calculations of central government actors, while relaxing the assumption that the center devolves power only when the governing party is strong subnationally. The Senegalese case examined in chapter 6 seems to blend the two top-down motives outlined by O’Neill and Rodriguez, respectively: the Parti Socialiste in the 1990s sought to strengthen its subnational political base while also engaging in a popular reform; the combination may be seen as the party “diversifying its portfolio of holdings” as electoral uncertainty increased.

Also of great import to the study here, however, are processes of decentralization that do not conform to top-down, electoralist models, but rather depend on deeply entrenched political institutions that shape calculations by political players. Kent Eaton’s (2004a) analyses of decentralization “from below” blend rational-institutional and historical-institutional arguments to evaluate decentralization across different time periods. Brazil, for instance, is best viewed as a country whose decentralization processes have been driven by powerful subnational actors, with shifts in the political economy triggering changes in subnational autonomy; political parties, legislatures, and states serve as mediating factors or intervening variables in such processes. In chapter 2, I discuss at length the historical trajectories that give rise to top-down dynamics in unitary and weakly federal countries, and the bottom-up dynamics that prevail in robustly federal states; this highlights the need to offer rationalist accounts that are historically informed rather than ahistorical.

Crisis Resolution and Recentralization

Centralization and recentralization (or, more generally, reductions in subnational autonomy) have received less attention than decentralization processes in recent years. There are good reasons for this. The greater emphasis on decentralization is due to several asymmetries. Recentralization is conditional upon some prior instance of decentralized governance, whereas decentralization processes are conditional upon prior centralized governance; given the nature of the centralized state in the developing world, it is understandable that decentralization would take precedence. Moreover, decentralization is examined in a variety of issue areas, including fiscal decentralization, administrative decentralization, and political decentralization (cf. Falleti 2005), whereas studies of recentralization generally focus on central control of revenue authority. Related to both of these facts, and perhaps most importantly, recentralization is likely to be quite contested. Decentralization invests power and authority in subnational officials, and these officials once empowered are likely to strongly resist recentralization with the resources at their disposal (O’Neill 2005, 60).

Yet the political-economic approach that offers explanations of decentralization “from above” and “from below” leaves several questions unanswered. Most importantly, why would incoming governments not reverse decentralization? If declining parties can decentralize without friction at the end of their mandates, why can’t newly inaugurated governments (often at the peak of their legitimacy, unlike their “outgoing” predecessors) simply reverse the legislation? Why does decentralization “stick” in some countries, but get reversed elsewhere?

The comparative political economy literature on economic crisis provides guidance in understanding recentralizing changes. Economic crisis can shape important political changes, with some scholars finding that such a crisis leads to less central government control over a variety of decisions, at least at initial stages (Grindle 1996; Haggard and Kaufman 1995), while others highlight the leverage crises provide to central governments (e.g., Weyland 1998). The argument here modifies these two positions by arguing that crisis resolution, rather than the mere existence of crisis, drives the most significant processes of recentralization. Presidents that resolve crises have the political power to reduce subnational autonomy. Crisis actually triggers changes that favor the national executive in negotiations with subnational actors, by providing extraordinary governing opportunities for those presidents who succeed in stabilizing the economy. Whereas decline leads to decentralization, macroeconomic crisis—or more specifically, its resolution—leads to centralization.

Two points must be elaborated with respect to the concept of economic crisis and its use as a variable. First is the question of conceptual precision, since crisis as a variable may be used sloppily to signify a range of economic phenomena of the analyst’s choosing. As recent work by Reinhart and Rogoff (2009) has shown, economic crises may be defined in several ways, ranging from crises based on quantitative macroeconomic indicators (such as inflation) to those understood as “events” (such as bank failures). Crises take a variety of forms, including fiscal crises (and government defaults), financial sector or banking collapses, exchange rate collapses, balance of payment crises, and inflationary crises. Each is associated with its own indicators, such as government deficits and indebtedness, asset price collapses, or inflation rates. Even more loosely, scholars have referred to economic “crises” in the developing world that are chronic rather than acute; these may include extended periods of no growth or slow growth (as in “lost decades” or “Africa’s economic crisis”), dramatic slowdowns in real economic activity, or simple economic contractions or reductions in capital investment. Without a firm grasp on the subset of crises purported to have causal impact, we are left with an analytical risk: observing the predicted outcome on a dependent variable, and inferring an impact of crisis whenever any of a number of economic conditions turns sour. We must define which crises matter for politics.

I expect significant effects on recentralization from economic crises that have a particular characteristic: they significantly affect purchasing power in the present. Crises in which citizens are seeing their purchasing power actively eroded will have distinct political consequences relative to other “crises” in which repercussions on the citizenry are still ambiguous or forthcoming. In open economies, the purchasing power of the money base can collapse in two principal ways: through inflation or through exchange rate collapse (i.e., the loss of purchasing power for imports). Of course, where both inflationary crises and currency crises occur simultaneously, the anticipated effects would be even stronger. These crises in particular should matter because they shape citizens’ willingness to allow dramatic market reform. Under the conditions of purchasing power crises, citizens operate “in the domain of losses,” to use Weyland’s (2002) term. From the perspective of prospect theory, an aversion to further losses—more simply, a desire to “just make it stop”—allows governments to implement strategies with higher risks and major implications for the institutional balance of power. This includes painful economic adjustments and—in the present case—recentralizing changes.

By the purchasing power criterion, many other forms of so-called crises should not lead to recentralization. Extreme fiscal imbalances, for instance, may constitute a “fiscal crisis,” but these are not posited to have their political impact until purchasing power is affected. Fiscal deficits may of course eventually lead to collapses in purchasing power through the intervening variables of indebtedness, inflation, balance of payments crises, and exchange rate depreciation, but if these purchasing power declines have not yet been realized, then we would not anticipate a willingness to confer immediate authority to the central government executive on the basis of fiscal deficits alone. Similarly, trade deficits or imbalances in trade would not constitute a crisis for the purposes of the present argument until purchasing power was affected through a collapse in the exchange rate. Banking sector crises and government defaults are also significant insofar as they engender actual purchasing power losses for the citizenry. Finally, the more chronic forms of crisis—especially long-term economic stagnation and underperformance—are not considered likely contributors here because they do not place citizens in loss-averse situations.

To a large extent, then, I follow Weyland’s (1998, 2002) emphasis on hyperinflation, which is less susceptible to slipperiness than other indicators, although the exact threshold is debatable, often being defined at 100 percent per annum. Using this definition of hyperinflation has the advantage of conceptual clarity, but the choice is also informed by a theoretical choice, since hyperinflation may be seen as economic crisis in its most acute form, and as qualitatively different from other forms of economic crisis, in part because it gives citizens a greater propensity to accept risk; the psychological effect of hyperinflation on citizens and on politicians alike makes these crises ripe for bold policy prescription and offers the potential of extraordinary political leverage for those who tame it. In the cases examined here, hyperinflation was the key indicator of a crisis of purchasing power, though this was partially accompanied by collapses in exchange rates as well. The in extremis nature of crisis also provides crucial political cover for centralizing reforms that might otherwise be construed as antidemocratic power grabs. Importantly, the South African case further shows the importance of a clear understanding of crisis: recentralization there was much more circumscribed, due to the much more localized and limited nature of the crisis in government finances.

A second major issue with respect to the concept of crisis is the need to specify the path-dependent causal sequence linking crisis and recentralization. As noted above, crisis resolution (rather than crisis itself) is key to the complex chain of events that initiates and sustains recentralization. Crisis may be a necessary condition for presidents to initiate successful recentralization processes, but is not sufficient, since consummating a recentralization process depends on presidential leverage. The cases here leave clear that presidential administrations that take blame for causing an economic crisis will have great difficulty recentralizing; presidents who governed during the onset of crisis are likely to see declines in presidential leverage, which will have not a recentralizing impact, but rather the decentralizing impact noted above. By contrast, presidencies that can initiate and sustain a resolution to hyperinflationary crises will increase their intergovernmental advantage and will have unique centralizing opportunities. Among the four case countries, two presidents (Fernando Henrique Cardoso in Brazil and Alberto Fujimori in Peru) resolved major hyperinflationary crises and used their resulting political leverage to engineer substantial reductions in SNG autonomy. Absent such a crisis, would-be centralizers in other cases (Thabo Mbeki in South Africa and Abdoulaye Wade in Senegal) had less scope to reverse decentralizing moves that had occurred to varying degrees in their countries, even though they had considerably greater partisan powers.

Recentralization processes were successfully initiated during a hyperinflationary moment in Brazil (1994–95), but these processes also had an even more significant path-dependent quality. The fiscal recentralization that occurred in Brazil built upon itself once initiated, with a presidential administration that was surprisingly successful—especially over the years from 1995 to 2000—at institutionalizing central control over larger proportions of government revenues, expenditures, and borrowing. In Peru, recentralization was largely subsequent to the resolution of the crisis, occurring mainly in 1992, by which time inflation had subsided considerably.

The preponderance of recentralizing moves in both countries occurred after economic crisis resolution, as path-dependent sequences shaped continued recentralization throughout the 1990s. Yet it should be noted how sequences differed somewhat between the federal case of Brazil and the unitary case of Peru, largely as a function of the countries’ prior degrees of decentralization. In Brazil, recentralization was in large part a necessary response to SNG profligacy (since the SNGs were among the actors primarily responsible of the economic crisis), whereas in Peru (where SNGs had little or no role in causing the crisis), the process of recentralization came largely after crisis resolution. This accelerated recentralizing sequence under Brazil’s federal system was more endogenous and reciprocal, since crisis resolution and recentralization were mutually contingent in ways that did not apply in unitary Peru. The consequence was the more simultaneous interplay between crisis resolution and recentralization in the former case, and the more sequential path in the latter. The particular sequences, and their relationships to the federal–unitary distinction, are further elaborated in the case analyses below.

Taken from a comparative static perspective, it will come as a surprise to observers of Brazilian and/or African politics that Brazil centralized more than the African cases. To be clear from the outset, Brazil remains considerably more decentralized (in the comparative static sense) than any country in Africa. The reasons for Brazil’s continued high levels of decentralism are well documented and relate strongly to the historical development of legislative institutions that favor SNGs in national decision making (see, e.g., Ames 2001; Garman, Haggard, and Willis 2001; Mainwaring 1997a, 1997b; Samuels 2003). On the other hand, once power was devolved in each of these cases, Brazil’s central government has been successful in clawing back some of the fiscal authority of SNGs. Given Brazil’s weak party system, this cannot be explained by partisan powers of the president alone. Despite clear efforts to recentralize, South Africa’s African National Congress (ANC) and Senegal’s Parti Démocratique Sénégalais (PDS) presidents—with their stronger partisan powers—left revenue provisions and subnational responsibilities largely intact, while the Cardoso government in Brazil reformed that country’s fiscal federal system in the wake of economic crisis.

Additional observable implications reinforce the argument about the importance of economic crisis: I show that even small, localized fiscal crises in selected SNGs also generate a centralizing dynamic, but one that is correspondingly weaker than the dynamic created by larger systemic crises characterized by full-blown hyperinflation. Localized fiscal problems of this sort occurred in South Africa, where the central government resolved provincial budget crises, and in so doing intervened directly in provincial fiscal affairs. The key observation is that the extent of recentralization depended on the extent of fiscal crisis: whereas crisis was generalized to the macroeconomic level in the Latin American cases, it was limited to specific provinces in South Africa, and the corresponding recentralization was limited as well. Obviously, limited recentralization in South Africa involved direct inter-executive negotiations between the levels of government. Other reductions in autonomy came through the exercise of executive power in other arenas, most notably within the state and in the legislature.

Political Science

           Comparative Politics