While still developing, the research program on political, fiscal, and administrative decentralization has matured to the point where a thorough review of its key findings is in order. Daniel Treisman has provided perhaps the most useful and important assessment of the research program to date. Based on a thorough review of the empirical work, his own formal models, and empirical studies, Treisman concludes that there are no generalizable effects from political decentralization. Such a conclusion emerges from an exhaustive assessment of the effects of decentralization in diverse areas: Tiebout-inspired competition models, fiscal federalism, participation and representation, individual and group freedom, and ethnic conflict. The author finds that decentralization produces some predicted effects, but not always consistently. Moreover, the results do not always demonstrate that decentralized governance is any better (or worse) than centralized forms. Using formal models, he meticulously unpacks the logic for and against the effects of decentralized government to find that more often than not, the outcomes are contingent on other factors. Spurious correlations of decentralization's effects on democratization, efficiency, and redistribution are exposed throughout the book. Both the purview and the depth of the analysis are impressive, encompassing both classical arguments in political philosophy and more contemporary debates in public policy and political economy. The result is a most erudite, thorough, and insightful rethinking of the major arguments and schools of thought concerning political decentralization.

The Architecture of Government will be a staple reading in graduate schools for many reasons. First among these are the initial chapters, which discuss the conceptual and formal models that underpin much of Treisman's subsequent analysis. While the full discussion of game theoretic models in Chapter 2 (especially the section, “Modeling Politics”) can be overwhelming to the nonformal modeler, it is an essential reference for work on decentralization and other areas that are amenable to strategic interactive modeling. The organization of the presentation, facilitated by section and subsection numbering, makes these chapters (indeed, the entire book) a valuable reference for scholars of decentralization but also a broader range of political scientists looking for a useful, compact discussion of formal models and how to apply them.

In the substantive chapters, Treisman first takes aim at functionalist approaches inspired by Ronald Coase's theory of the firm and Charles Tiebout's competition arguments. The author finds no consistent logic for the degree, scope, or function of decentralized governance that guarantees Pareto optimal outcomes. Decentralization always seems to be limited and shaped by costs of communication, monitoring of agents, and heterogenous client tastes. Nor does competition among subnational units guarantee efficient distribution. After an extremely thorough assessment of Tiebout's overly restrictive assumptions, Treisman demonstrates that competition cannot overcome information costs the way the model predicts. Moreover, the effects of competition can generate additional costs through beggar-thy-neighbor behaviors and other forms of negative externalities. But the biggest failure of functionalist approaches is that they hardly compare the dynamics of decentralized to centralized government. In theory, centralized regimes can deliver policies that satisfy local tastes or they can replicate the effects of Tiebout's competition by devolving authority and resources based on migration patterns (i.e., demonstrated shifts in citizen preferences).

Turning to the well-trodden area of fiscal federalism, Treisman offers arguments that challenge core precepts of the extant literature. In Chapter 5, he turns to the subject of soft budget constraints and their effects on the fiscal probity of subnational governments. Turning the common argument that such institutions produce profligacy on its head, he points to the way in which such irresponsible behavior would work against the national political interests of subnational governments, forming its own disincentive to behaving badly. The importance of the intergovernmental game is given further attention in Chapter 6, in which the author shows that decentralizing resources and authorities does not produce efficient outcomes where there is a trade-off between empowering local governments and enabling central government to provide services and engage in cooperative policymaking.

Regarding democracy, Treisman notes that decentralization does not make citizens more participatory, nor does it make representatives more accountable to them. One problem is the role of local governance as a model for the regime as a whole. What Tocqueville praised in the small towns of New England is hardly replicable or even desirable on a larger scale. Small groups can make democratic decisions, but the procedures they use are overly cumbersome with larger numbers. The issues they address are less important than what concerns the national polity. Voters prove this true by turning out in larger numbers in national elections than in subnational ones. The democratic bona fides of subnational government are also questionable. As the American examples of Reconstruction after the Civil War, the fight over desegregation, and voting rights in the South attest, subnational governments can become bastions for authoritarianism. Citizen mobility may even enhance these antidemocratic enclaves by allowing authoritarians to take refuge in them, create their own, or force those seeking personal freedom to escape into the wilderness, saving their skins but not assuring their liberty. Decentralization in such cases weakens democracy.

This book is driven primarily by formal logic, but there is a fair amount of empirical analysis as well. Most notably in Chapter 11, Treisman examines the most prominent large-n cross-national studies of the effects of decentralization. In most cases, the author retests hypotheses attempting to replicate the results in different samples or the original set. In evaluations of the effects of decentralization on economic performance, democracy, ethnic conflict, and service provision, he finds either insufficient evidence or contradictory data for the importance of decentralization. Regarding the quality of government (measured in terms of corruption), he finds the relationship spurious by adding two heretofore omitted variables: democratic maturity and Protestant culture. The tests focusing on economic performance do show that decentralized states produce more consistent results, but this is true for both positive and negative macroeconomic performance. Otherwise, the effects of decentralization on growth and macroeconomic performance are indeterminate.

The Architecture of Government does not offer a thorough empirical evaluation of all of the areas the author examines. It is primarily a book that models intuitions concerning political decentralization. And by examining the logic of these intuitions, he shows them to be largely inadequate. At least in theory, it appears that it is not possible to design political decentralization under conditions that generate predictable functions or outcomes. This in no way destroys the research program on decentralization; instead, it refines it. Under what specific conditions or contexts will decentralization's effects hold across time and space? Treisman has usefully defined dozens of research questions revolving around this central theme. His book will inspire, instruct, and compel many of us who pursue work in this field to think more systematically about political decentralization.