Ensayo

32

Studies in Comparative International Development / Fall 2003

Decentralization: Conceptualization and Measurement*
Aaron Schneider
Decentralized government institutions are doing more of the work of government than ever before, but there is little agreement about 1) what decentralization means, or 2) how it should be measured. To overcome this confusion, this article builds on standarddefinitions of decentralization that include three core dimensions: fiscal, administrative, and political. The article offers an empirical test of that definition using factor analysis of data from 1996 for sixty-eight countries. Factor analysis confirms these three core dimensions and generates a score for each case in each dimension, allowing countries to be measured according to their type anddegree of decentralization. In future work, these scores can be used for hypothesis testing about the causes and effects of decentralization on important social outcomes. This exercise demonstrates that conceptual confusion need not hamper research when empirical tests can help verify conceptual categories.

D

ecentralized government institutions are doing more of the work of gov ernment thanever before. Recent research has responded to this development and deepened our understanding of decentralization and its links to outcomes such as growth, inequality, and political stability. Despite increased research, there remains a great deal of conceptual confusion. Researchers attach a startling diversity of definitions and measures to the decentralization concept. The availability ofcross-national statistical data has only exacerbated this proliferation. Some might consider this proliferation of measures and definitions as necessary to account for a wide variety of cases. Alternatively, others might consider, as this article does, that the proliferation of meanings and measures erodes precision and impedes our ability to assess types of decentralization. The prob-

AaronSchneider is a political scientist at the Institute of Development Studies at the University of Sussex. His research interests include comparative politics, public finance, and methodology. His current research projects include studies of federalism, decentralization, party systems, budgeting, and taxation. He has conducted research in Brazil and India, and plans to apply the measures derived in thisarticle to study the impact of decentralization. Studies in Comparative International Development, Fall 2003, Vol. 38, No. 3, pp. 32-56.

Schneider

33

lem is worsened by the evaluative nature of the decentralization concept, which leads researchers to conflate decentralization with other concepts, especially those that are also imbued with positive value, such as democracy or marketreforms. The result is that there is little agreement about what constitutes an example of decentralization, what causes decentralization, or what effects it is likely to have. To overcome this confusion, this article proposes and tests a definition of decentralization. Although there is disagreement about the meaning of decentralization, most would agree that transferring power and resources to nationalgovernments is not decentralization.1 Nevertheless, all share the assumption that decentralization includes the transfer of power and resources away from the central government. This article hypothesizes three core dimensions of the decentralization concept: fiscal, administrative, and political.2 Fiscal decentralization refers to how much central governments cede fiscal impact to non-centralgovernment entities. Administrative decentralization refers to how much autonomy non-cen, tral government entities possess relative to central control. Finally, political decentralization refers to the degree to which central governments allow noncentral government entities to undertake the political functions of governance, such as representation. Decentralized systems are those in which central...