Decentralizing the State
Cambridge University Press
0521846943 - Decentralizing the State - Elections, Parties, and Local Power in the Andes - by Kathleen O'Neill
ELECTORAL THEORY AND COMPARATIVE EVIDENCE
From Bolivia to India, and from the United Kingdom and Spain to Uganda, national governments are giving away their power. A revolution of local empowerment has quietly swept both developed and developing nations alike in the closing decades of the twentieth century.
Perhaps nowhere has this trend reshaped the political landscape more dramatically than in Latin America - a region more likely to call to mind highly centralized governments run by military strongmen, civilian dictators, or one-party rule (Véliz 1980). In 1980, only half of Latin America's governments held democratic elections at the national level; by 1997, all but one elected both national and regional and/or local governments (Inter-American Development Bank 1997: 99). Even more stunning, many countries have bolstered local political empowerment by increasing the financial independence of subnational governments by apportioning real fiscal resources to elected officials.
Have national politicians all gone mad? Surprisingly, few scholars1 have paused to probe the puzzling question of why politicians are giving power away. Instead, a great deal of intellectual activity has been devoted to two other questions: Has fiscal decentralization increased economic gains?2 And have more elections improved democratic quality?3 Determining what motivates decentralization seems a necessary and prior question to determining what its consequences may be; it is a missed step that can throw the best analysis of consequences awry. Exploring the why and the when of decentralization is the heart of this project. Unfortunately, the answer is not obvious. In most cases, decentralization has not been adopted as a response to pressures from below; rather, it has been initiated by the national governments most likely to be hurt by decentralizing reforms. For these reasons, decentralization poses a thorny puzzle for political scientists who assume that political actors seek to gain and hold on to power.
In some countries, especially the Southern Cone countries of Argentina and Brazil, but also Chile to some extent, decentralization might be written off as merely a return to the pre-authoritarian power-sharing relationship that had been established between central and subnational governments.4 The more fascinating - and puzzling - cases are those in which both political and fiscal powers were devolved simultaneously and for the first time. In the Andean region of Latin America, comprised of Bolivia, Colombia, Ecuador, Peru, and Venezuela, decentralization generally meant an unprecedented devolution of power from national to local governments.
In addition to being unitary states (although Venezuela is nominally federal), these countries also provide an ideal set of comparative cases because they share the same cultural and regional context, have achieved broadly similar levels of economic development, and yet differ widely in the pace and extent of decentralization adopted. They also differ substantially along the range of explanatory factors expected to account for differences in decentralization's adoption.
Understanding why administrations in the Andes chose to decentralize and why they chose to decentralize at particular moments in their democratic trajectories is the main task of Decentralizing the State. Decentralizing reforms are complex and highly varied policy initiatives, making it highly unlikely that a single logic can explain their adoption; however, I make the case that political - and particularly electoral - motivations play a critical and overlooked role in many decentralizing reforms. By focusing on the political party as the key decision-making unit and by examining its electoral motives, this project finds a solution that reconciles seemingly irrational political action with the drive for political survival. The central argument of Decentralizing the State is that administrations are more likely to favor decentralization when their party is likely to benefit from electoral contests for subnational positions. This is most likely when the party in power believes it cannot hold on to power that is centralized in the national government but believes it has a good chance of winning a substantial portion of decentralized power through subnational elections. Decentralization distributes power at one moment in time to the venues where a party's political allies are most likely to win it in future contests. Thus, decentralization can be seen as an electoral strategy to empower political parties with reasonably long time horizons.
While Chapter 2 devotes a good deal of attention to deriving and developing this theory, the underlying logic can be gleaned from a simple thought experiment. Imagine that you are the head of a party that controls the presidency in a highly centralized governing system. Thinking ahead to the next presidential contest, you can compete under the same centralized system or you might choose to use your power to decentralize the system prior to the election. A win yields another term of centralized rule; a loss may mean nearly complete exclusion from power for several years. A decentralized system, in contrast, offers contesting parties a somewhat less attractive presidency but several opportunities to gain footholds of power at subnational levels. Before you cast your lot with the decentralizers, consider that decentralization comes at a cost. To compete in a decentralized system in the future, the administration must give up some of its power now, when its hold on national power is already assured for the term. The benefits of competing under a decentralized system in the future must outweigh the cost of lost power in the present. In this calculation, decentralization will be most attractive when a party's national support is weak (it is unlikely to win the next presidency), when subnational electoral chances look good (it is likely to win subnational contests under a decentralized system), and when the party values the future (since costs are incurred now and benefits accrue in the future).
How does a party have weak national support and strong subnational support? Another thought experiment could illustrate. Imagine a country with 100 states, where the president is selected by majority rule and population is spread evenly across the states. The current president won 60 percent of the vote. In scenario one, this 60 percent was evenly spread across the 100 states, so 60 percent of voters in each state cast their votes for the winning candidate. In scenario two, this 60 percent came from 100 percent of the voters in sixty states voting for the victor and 100 percent of the voters in the remaining forty states voting against the victor. If national support is expected to fall to 40 percent in the next election and if the ruling party expects support to be distributed as in the past, then the party will win 40 percent of the vote in each of the states in scenario one and 100 percent of the vote in forty states in scenario two. If decentralization means that each of the states would elect a governor based on majority rule, the ruling party could expect to lose not only the national election but also all of the state elections in scenario one. Decentralization in scenario two would mean winning state elections in forty states. National support for the party as a whole declines in both scenarios (the party loses the presidency), but subnational electoral prospects look remarkably better in the second scenario. To find an example of a party that is nationally weak but subnationally strong, one need look no further than the Republican party in the United States in 1996; though its candidate faced insuperable odds of winning the presidency, many of its members won or maintained gubernatorial positions throughout the country. Thus, decentralization will not become attractive to a party whose national support is falling, if its electoral possibilities at the subnational level are also in precipitous decline. Likewise, a party that is highly uncertain about its future electoral prospects or that is under-institutionalized may heavily discount the future, also making decentralization unlikely.
While the body of scholarship on decentralization remains incipient, political scientists have developed large bodies of scholarship on federalism and delegation, two areas that would seem to be rich sources of theoretical insight into this wave of reform. No mere extension of the theories developed to explain federalism or delegation seems to explain recent decentralizing reforms in Latin America, however; instead, the study of decentralization promises to add to both literatures.
Federalism and decentralization are closely related: Both refer to systems in which subnational units enjoy a degree of autonomy from central policy makers. Federalism traditionally differs from decentralization in that states or regions play the primary role in federal systems, whereas decentralization often skips the regional level altogether, directing resources toward local governments; in addition, (successful) federal systems usually include a bicameral legislature in which one house is elected to represent territorial interests. Despite these differences, federal and decentralized systems have much in common. Perhaps theories of federalism may shed some light on decentralization, as well.
The most oft-cited scholar of federalism, William Riker, theorizes that federalism results from "a bargain between prospective national leaders and officials of constituent governments for the purpose of aggregating territory, the better to lay taxes and raise armies" (Riker 1964: 11). Prospective national leaders seek to expand territorial control in the face of military or diplomatic threats (or opportunities) and seek confederation because either they cannot conquer the territory or they find conquest distasteful. Although he relies heavily on the U.S. case, his analysis includes the federal states of Switzerland, Germany, the USSR, India, Pakistan, Argentina, and Brazil. As the instances of decentralization studied here begin with a unitary state devolving powers to its constituent units, this theory does not seem appropriate.
After the collapse of communism, a new wave of scholarship on federalism emerged, exploring the role of federal institutions in forestalling or encouraging the decomposition of states (Roeder 1991; Bunce 1999; Solnick 1999). This literature has spread beyond Eastern Europe to include Indonesia (Ferrazzi 2000), Nigeria (Suberu 2001), and Spain (Moreno 2002), to name a few. Though geographically and culturally disparate, these studies have in common a primary focus on the tensions between a central state and its ethno-territorial components.
Mindful of this wave of case-specific explorations of federalism, Stepan (1999, 2000a, 2000b, 2001) has attempted to amend Riker's theory, arguing that Riker's theory of confederation describes only one of several paths through which countries adopt a federal system. Stepan adds two other categories to the "coming-together" federalism discussed by Riker: "holding-together" federalism and "putting-together" federalism. The latter involves the coercion of units into a federal arrangement (as with some states in the formation of the USSR), while the former occurs when a unified state trades increased autonomy to subunits for their acquiescence in remaining within the broader union of states. This latter is particularly interesting from the perspective of decentralization, as it involves the transfer of power from the center to constituent units; however, "holding-together" federalism is a bargain struck by the center with subunits that are threatening to secede.5 Even this amended federal literature cannot account for the Latin American cases of decentralization: devolution of power to regions or localities in the absence of anything like secessionist demands from subunits. An exploration of decentralization thus goes beyond the literature on federalism, seeking reasons for the empowerment of subnational units in the absence of secessionist threats. While federal and decentralized governments look similar in practice, the literature addressing federalism's adoption provides little guidance when one attempts to understand the adoption of decentralization in Latin America.
Perhaps decentralization better approximates delegation than federalism; certainly, the literature on delegation begins with the same motivating question: If politicians seek access to political and fiscal resources, why do they give power away? A rich literature probes this question, focusing on delegation of authority by legislatures to the bureaucracy, by parties to legislative committees, and by elected governments to independent agencies.
Lowi (1969) pioneered in this subject matter when he observed an increasing bureaucratization of policy making in the United States. He argued that Congress was abdicating its duties to the bureaucracy, leaving policy open to the influence of special interests that could, and often did, "capture" their regulators. Later, Fiorina (1977) argued a slightly different form of the abdication hypothesis: that members of Congress delegated authority knowing there would be problems, expecting to intervene to address egregious errors and, thereby, boost their support.
An opposing theoretical view argued that delegation, whether by parties to committees (Cox and McCubbins 1993) or by the Congress to the bureaucracy (Kiewiet and McCubbins 1991), did not represent a significant abdication of authority after all. They argued that delegators retained a great deal of control over the committees or bureaucracy through powers of appointment in the latter case and various forms of oversight in the former.
A third theoretical approach draws on the insights of transactions cost analysis in economics. Epstein and O'Halloran (1999) hypothesize that legislators compare the likely policy outcome on each issue from not delegating (i.e., the relevant committee's most preferred outcome) with the likely outcome from delegation and choose the method most likely to get them re-elected. This leads to several testable hypotheses: On issues where the relevant committee's preferences closely approximate the legislature's preferences, delegation is not likely; when the executive's preferences closely approximate the legislature's preferences, delegation is more likely; and when the issue to be decided is more informationally intensive and/or less distributive in nature, delegation is more likely.
Finally, scholars exploring why administrations sometimes attempt to institutionalize their policy preferences by delegating important policy areas to independent agencies (i.e., central banks) provide yet another perspective on the question of giving power away. McCubbins, Noll, and Weingast (1987), Horn and Shepsle (1989), Moe (1990), and Boylan (2001) theorize that outgoing administrations face incentives to institutionalize their policies to protect their interests from incoming opposition.
None of these theories of delegation, which focus on the delegation of authority to appointed officials, quite explains decentralization, which allows for popular elections at subnational levels. Still, several insights can be gleaned from this body of work. In contrast to what the shirking hypothesis would predict, the instances of decentralization studied here6 involved a substantial devolution of real fiscal resources along with broad policy-making authority. Likewise, the idea that institutional constraints effectively limit the degree of delegation occurring in these cases also does not fit with the nature of these decentralizing reforms, which allow for the popular election of subnational officials and give them broad policy making authority backed by significant fiscal resources that are not distributed purely at the central government's discretion.7
The logic of the transactions cost and institutional insulation arguments suggests that decentralization should occur when the delegator's policy preferences are more closely approximated by the workings of a decentralized rather than a centralized system. This basic insight is also the bedrock of this volume's analysis; however, the transactions cost and institutional insulation hypotheses are insufficient to give a full explanation of decentralization. The transactions cost approach, for example, suggests that certain policy areas are more susceptible to delegation than others: those that are more informationally intense and less distributive in nature. This does not seem to fit these cases of decentralization, where policy making in such highly distributive areas as education, health, and local infrastructure was devolved to subnational governments. The work on institutional insulation adds the important component of comparing the preferences of the current administration with an incoming administration, which will prove essential to explaining decentralization, but this comparison remains rooted at the central government level. Because decentralization leads to changes in policy-making authority not just between pieces of the national government, but also across levels of government, one must also look at the likely composition of incoming administrations throughout the country at subnational levels. Furthermore, because decentralization is costly in the current period, one is forced to consider the extent to which potential decentralizers value the future since the benefits will occur in future time periods, though the costs will be incurred in the current administration. While the delegation literature does provide some insight into the incentives to decentralize, it provides an incomplete framework for analysis. Exploring decentralization, a policy with clear affinities to both federalism and delegation, may lead to a better understanding of these other fields as well.
© Cambridge University Press