American Political Science Review Vol. 99, No. 3 August 2005 A Sequential Theory of Decentralization: Latin American Cases in Comparative Perspective TULIA G. FALLETI University of Pennsylvania oth advocates and critics of decentralization assume that decentralization invariably increases the power of subnational governments. However, a closer examination of the consequences of B decentralization across countries reveals that the magnitude of such change can range from substantial to insignificant. In this article, I propose a sequential theory of decentralization that has three main characteristics: (1) it defines decentralization as a process, (2) it takes into account the territorial interests of bargaining actors, and (3) it incorporates policy feedback effects. I argue that the sequencing of different types of decentralization (fiscal, administrative, and political) is a key determinant of the evolution of intergovernmental balance of power. I measure this evolution in the four largest Latin American countries and apply the theory to the two extreme cases (Colombia and Argentina). I show that, contrary to commonly held opinion, decentralization does not necessarily increase the power of governors and mayors. oes decentralization always increase the power Moving beyond the fiscal arena, the decentralization of governors and mayors? If so, what explains movement has seen major public services such as ed- Dthe different degrees of change observed in the ucation and health transferred to subnational govern- intergovernmental balance of power? Over the last ments. Moreover, political and electoral reforms have 30 years, decentralization reforms have swept across left governors and mayors more accountable to their the world, changing decades of centralized political and constituencies. This large-scale transfer of resources, economic practices as well as the way in which we study responsibilities, and authority has brought subnational politics. As James Manor writes, “Nearly all countries governments to the forefront of politics. Recent in- worldwide are now experimenting with decentraliza- ternational news’ headlines testify to the importance tion . seen as a solution to many different kinds of of subnational elections and local governance issues.2 problems” (1999, vii). One need only look as far as the The decentralization movement has also highlighted fiscal data to observe this trend. In 1980, subnational the relevance of intergovernmental relations, once de- governments around the world collected on average scribed as the “hidden” or “fourth branch of govern- 15% of revenues and spent 20% of expenditures. By ment” (Edmund Muskie 1962, cited in Wright 1978, 5), the late 1990s, those figures had risen to 19% and 25%, in comparative politics. Increasingly, political scientists respectively, and had even doubled in some regions.1 are shifting the locus of their analyses from the national to the subnational levels (Snyder 2001) and from the horizontal relations among branches of government Tulia G. Falleti is Assistant Professor, Department of Political Sci- to the vertical relations between levels of government ence, University of Pennsylvania, 208 South 37th Street, 202 Stiteler Hall, Philadelphia, PA, 19104-6215 ([email protected]). (Gibson 2004). Despite this ostensible change in the The author thanks Isabella Alcaniz,˜ Dawn Brancati, Ruth Ben- political and analytic landscapes, the question remains, Artzi, Ernesto Calvo, Maxwell Cameron, Edward Gibson, Gerald has decentralization led to the expected shift in the McDermott, Richard O. Moore, Ben R. Schneider, Rogers Smith, balance of power among presidents, governors, and Kathleen Thelen, and the editor of the American Political Science Re- view and three anonymous reviewers for very helpful comments on mayors? previous drafts. The author is also grateful to Daniel Brinks, Katrina A substantial body of work on the consequences of Burgess, Barbara Geddes, John Gerring, Merilee Grindle, Wendy decentralization hinges on the answer to this question; Hunter, Jose´ Itzigsohn, James Mahoney, Paul Pierson, Dietrich nevertheless, little attention is paid in the literature to a Rueschemeyer, Theda Skocpol, Benjamin Smith, Martin Tanaka, critical assumption that could very well be unjustified. and the seminar participants at Northwestern University, Brown University, the Central European University in Budapest, the David Political scientists who draw from the liberal tradition Rockefeller Center for Latin American Studies at Harvard Univer- argue that decentralization helps to deepen and consol- sity, the Kellogg Institute at the University of Notre Dame, and the idate democracy by devolving power to local govern- migration of authority conference at the University of Michigan- ments (Diamond and Tsalik 1999). Economists who Ann Arbor for their comments at earlier stages of the project; and to David Samuels and Matthew Shugart for sharing their congressional draw from a market theory of local expenditures argue data on Brazil and Colombia. The Social Science Research Council, that decentralization helps to improve resource allo- the United States Institute of Peace, the Center for International and cation through better knowledge of local preferences Comparative Studies and the Alumnae Association at Northwestern and competition among localities (Oates 1972). Other University supported the fieldwork on which this article is based. scholars, meanwhile, warn against the devolution of The Kellogg Institute for International Studies at the University of Notre Dame and the Centre of International Relations at the Univer- sity of British Columbia provided financial support and outstanding environments to write this article. to 29% in 2000. Source data available at: http://www1.worldbank. 1 In the large Latin American countries (Argentina, Bolivia, Brazil, org/publicsector/decentralization/fiscalindicators.htm. Chile, Colombia, Mexico, Paraguay, and Peru, for which comparable 2 “The waiters’ revolt. State elections in Mexico,”The Economist, data is available) the subnational shares of revenues and expendi- February 12, 2005; “Conservatives Claim to Carry German State in tures increased from averages of 14% and 16% in 1980, respectively, Close Vote,” The New York Times, February 21, 2005; among others. 327 Sequential Theory of Decentralization August 2005 power to subnational officials and show that it can aug- policies are important determinants of the evolution of ment distributional conflicts (Treisman 1999), foster intergovernmental balance of power. Previous studies subnational authoritarianism (Cornelius, Eisenstadt, have successfully accounted for varying degrees of fis- and Hindley 1999), and exacerbate patronage (Samuels cal decentralization at one point in time (e.g., Garman, 2003). Recent studies also suggest that, in the absence Haggard, and Willis 2001), but have fallen short of of proper fiscal and political mechanisms, the trans- explaining the effects of decentralization policies on fer of resources to subnational governments may lead the evolution of intergovernmental relations. I will not to higher levels of inflation (Treisman 2000), larger only measure the absolute level of decentralization at deficits (Rodden 2002), and poorer overall macroeco- different points in time but also trace the effects of nomic performance (Wibbels 2000). Interestingly, de- earlier reforms on later ones. spite their disagreements on the effects of decentraliza- The article also brings subnational actors and inter- tion for democratization and economic reform, all of ests to the center of the analysis. The puzzle of why the aforementioned studies share an assumption that national politicians choose or agree to give power away decentralization increases the power of subnational has led scholars to focus largely on the interests of na- officials. This power increase is generally used as the tional politicians toward decentralization, either in the intervening variable connecting decentralization poli- executive branch (Grindle 2000; O’Neill 2003) or in cies and either positive or negative outcomes, without the relations between the national executive and the questioning the existence of such a power increase in legislature (Escobar-Lemmon 2003; Willis, Garman, the first place. and Haggard 1999). I show that a wide array of social If we conceive of decentralization as a multidimen- and political actors, including the governors and their sional process (Montero and Samuels 2004, 8) that ministers, the mayors, the governors’ and mayors’ asso- entails political bargaining over the content and im- ciations, the unions of the sectors to be decentralized, plementation of different types of policies, we find that and other sectors of civil society are also the makers of certain forms of decentralization in fact decrease the decentralization. power of subnational officials. In order to evaluate Finally, the article emphasizes the territorial com- the consequences of decentralization on broader pro- ponent of interest representation. A large part of the cesses of democratization and economic reform, we literature on decentralization has focused on the parti- need to establish first when and how decentralization san or electoral incentives that move decentralization policies increase or decrease the power of subnational forward. Although very important, such emphasis on officials. This article advances a definition of decentral- electoral incentives overlooks the territorial aspects of ization that distinguishes among