- Shopping Bag ( 0 items )
This book reexamines the arguments that underlie the modern faith in decentralization. Using logical analysis and formal modeling and appealing to numerous examples, it shows that most such arguments are based on vague intuitions or partial views that, do not withstand scrutiny. A review of empirical studies of decentralization finds these as inconclusive and mutually contradictory as the theories they set out to test. The book's conclusion - that one cannot generalize about when decentralizing will be beneficial and when harmful - promises to prompt a rethinking of both the theory of political decentralization and current rationales for development aid.
About the Author:
Daniel Treisman is a professor of political science at the University of California, Los Angeles
The Mexican novelist Carlos Fuentes believes that federalism may be the only way to preserve local cultures in a world of increasing economic integration. The Federalist Papers, he has argued, “should be distributed in the millions.” When British Prime Minister Tony Blair set out to modernize his country, he made devolving power outside Westminster a key element in the campaign. This was necessary, he said, to protect Britons’ “fundamental rights and freedoms” and to “develop their sense of citizenship.” In the 1990s, the diplomat and historian George Kennan confessed to dreaming of a United States reconstituted as a confederation of twelve regional republics, each of which would be small enough to provide “intimacy between the rulers and the ruled.”1
For anyone who might not yet have noticed, political decentralization is in fashion. Along with democracy, competitive markets, and the rule of law, decentralized government has come to be seen as a cure for a remarkable range of political and social ills. Enthusiasm extends across geographical and ideological boundaries, uniting left and right, East and West, and North and South. It ishard to think of any other constitutional feature – except perhaps democracy itself – that could win praise from both Bill Clinton and George W. Bush, Newt Gingrich and Jerry Brown, François Mitterrand and Jacques Chirac, Ernesto Zedillo and Vicente Fox, Mikhail Gorbachev and Boris Yeltsin.2
Political decentralization means different things to different people, and I will discuss definitions in Chapter 2. But most would agree that a political system is more decentralized to the extent that local officials are chosen locally and have the right to make final decisions on important policy issues. Political decentralization differs from mere administrative decentralization, under which the central government delegates some policy responsibilities to its appointed local agents but retains the right to overrule its agents’ decisions. Complete political and administrative centralization – found only in small, unitary states such as Monaco – exists if all policy decisions are made and implemented by a single central government and its centrally located agents.
The belief that political decentralization is a good thing has been reshaping government across the globe. In Western Europe, Italy, Spain, and France created directly elected regional legislatures in recent decades, and Belgium turned itself into a federal state. The United Kingdom introduced parliamentary assemblies in Scotland and Wales, reversing centuries of precedent, and revived one in Northern Ireland.3 In the postcommunist East, countries from Poland to Kyrgyzstan have been strengthening local governments. (“Before the Decentralization Program, we were like camels,” one Kyrgyz village leader enthused to a visiting aid worker, “but now we are horses.”) In Latin America, Argentina, Bolivia, Brazil, Chile, Colombia, Paraguay, Peru, Venezuela, and many of their Central American neighbors introduced local or provincial elections, and most devolved responsibilities to subnational bodies. In Africa, Ethiopia adopted a federal constitution, and post-apartheid South Africa gave its provinces considerable autonomy. Seventeen other countries – from Benin to Zimbabwe – introduced elections for local councils. In Asia, India rewrote its constitution to empower rural panchayat governments, and post-Suharto Indonesia shifted functions and resources to its subnational units. The Philippines transferred responsibilities for health care, education, social services, and the environment to localities. Even China, not to be left out, began holding village elections in the late 1980s and authorized the elected committees to arbitrate civil disputes and provide local services.4
Although the impetus has come from many directions, international development agencies have been energetic – and generous – cheerleaders. Calculating how much such agencies have spent promoting decentralization is difficult, but in recent decades it has surely run into the billions. According to its Web site, the Inter-American Development Bank approved $671 million in loans to support “decentralization and subnational government” in Latin America between 1961 and 2005.5 In 1997–2003, the World Bank was allocating about $300 million–500 million a year on loans to projects with a decentralization component.6 Besides this, the Bank claimed to have 180 members “holding meetings, sponsoring seminars and workshops, and developing Web sites to ensure that the latest thinking on [decentralization] is widely available” (Ayres 2003, p. 74). According to the same report, half the education projects supported by the Bank recently included “decentralization strategies in their design” (ibid., p. 76).
Various United Nations agencies have also done their part. As of 2000, the UN Development Program was supporting decentralization programs in ninety-five countries (UNDP 2002). The UN’s Capital Development Fund and Food and Agriculture Organization were both financing decentralization and local government in Africa (Morell 2004). The European Union provides grants, as do many of its member countries.7 The Asian Development Bank allocates loans and grants for decentralization in Asia. The U.S. Agency for International Development said in 2000 it was “supporting decentralization and democratic local governance initiatives in some 50 countries” (Dininio 2000, p. 2). Its funding for “democracy and local governance” averaged $141 million a year recently, some of which went to decentralization projects.8 In part, such flows probably aimed to make already decentralized structures more effective rather than to stimulate further decentralization. But for a developing country short of money, devolving power must look like an easy way to cash in on the rich world’s desire to help.
Behind these aid dollars stands a series of arguments. Many have a familiar, common-sensical feel to them. Decentralization brings government “closer to the people.” It focuses authority at a level at which governments must compete against one another – like firms in a market – to please footloose voters or investors. It makes it easier for citizens to hold their representatives accountable. Devolving power to local governments makes better use of local knowledge, protects individual liberty, encourages citizen participation, nurtures civic virtue, and alleviates ethnic grievances. Decentralized units can serve as “laboratories” of democracy, hosting parallel policy experiments. Besides being intuitive, these arguments have a distinguished provenance. Many date to the work of political thinkers such as Montesquieu, Rousseau, Tocqueville, or John Stuart Mill. Others are associated with great twentieth-century economists such as Friedrich Hayek and James Buchanan.
Not all scholars are sure that decentralization is always beneficial. Skeptics have argued that empowering local governments can undermine macroeconomic discipline because of “common pool” problems or “soft budget constraints” that lead governments to overspend (Prud’homme 1996, Tanzi 1996). Others have been disappointed by the record of decentralization reforms in practice. Sophisticated advocates of decentralization have taken such reservations in stride. While remaining enthusiastic about the objective, they have sought simple rules to guide how countries should decentralize, as well as remedies for the inconveniences decentralization might generate.
The goal of this book is to reexamine these arguments – both those of decentralization’s advocates and those of the skeptics. In the chapters that follow, I consider a dozen or so of the most common and influential arguments about how decentralization affects economic and political outcomes. I use formal modeling where useful to clarify the logic underlying each of these, to test for consistency, and to see what conditions or assumptions each presupposes. Are there reliable reasons to think that decentralizing government will in general have the effects the arguments predict? Even if the effects are not fully general, can one identify precise, observable conditions under which the arguments do hold?
The short answer to both these questions turns out to be no. When examined closely, neither the arguments about benefits of decentralization nor those about macroeconomic dangers are general at all. Some are simply invalid. Others do hold given certain conditions. But the conditions are so complicated and difficult to observe that the results provide little basis for empirical work or policy advice. One argument withstands scrutiny a little better, but even this implies nothing general about when decentralization is beneficial and when it is harmful. Although it might seem a waste of time given this conclusion even to review empirical studies of the consequences of decentralization, I do so briefly in Chapter 11. I find there that, as one would expect given the uncertain and conditional results of theory, almost no robust empirical findings have been reported about the consequences of decentralization.
To be clear, I do not find that decentralization is generally bad. In fact, the arguments against decentralization appear to be as partial and inconclusive as those for it, and the empirical evidence for them is just as weak and inconsistent. Rather, decentralization’s consequences are complex and obscure. Many effects pull in different directions, leaving the net result indeterminate. To choose to decentralize, in most settings, requires a leap of faith rather than the application of science. To devote hundreds of millions of dollars to persuading others to decentralize, given the current state of knowledge, seems odd to say the least. The emperor may not be completely naked. But he is dressed in little other than his underwear.
This conclusion will seem controversial and unappealing to many readers, and I do not expect it to be readily accepted. Critics will find points to question in my treatment of the arguments. I think it is harder to question the general picture that emerges. At the least, the analysis should cast doubt on certain widely shared assumptions and challenge advocates of decentralization to develop a more systematic and compelling theoretical case.
1.1 A Quick Look Back
The question of how governments should be organized must be as old as the study of politics.9 From Aristotle to Polybius and Cicero, classical authors debated whether public authority should be entrusted to a monarch, a senate of aristocrats, a popular assembly, or some mixture of the three. The advantages of different constitutions were scrupulously examined.
Almost all the ancient scholarship focused on the institutions of central government. It is striking how little, by comparison, classical thinkers had to say about the vertical structure of government – the division of states into several tiers and the distribution of responsibilities among them. Local government did not entirely escape notice. Cicero, who served himself as governor of Cilicia, expounded on the duties of provincial governors, and Plutarch offered memorable advice to Greek municipal councilors on how to deal with their Roman overlords.10 But both – like other writers – seem to have taken the existing vertical structure for granted. How functions should be divided among central and local organs does not appear to have struck them as an interesting or relevant question.11
This is strange because multi-tier states have been common since the beginning of recorded history. The Sumerians, who left some of the oldest writing, inhabited a network of twelve or so theocratic city-states, each of which was divided into villages or rural communities, which were in turn subdivided into hamlets (Diakonoff 1974, pp. 8–10; Crawford 1991; Finer 1997, vol. 1, pp. 104–27). Every unit had its own priest-ruler. All the great empires – from Egypt to Persia – were administered by territorial governors, viziers, satraps, or other agents. The Israelite tribes and the Greek city leagues compete for the credit of having invented the confederation (Larsen 1968, Elazar 1987). The Roman republic and its provinces were integrated by an innovative system of administrative law, in which centrally appointed provincial governors could be sued, after leaving office, by those they had governed. Even the polis, that symbol of unitary, direct democracy, was not as flat as might be thought. Athens, after Cleisthenes’ reforms of 508–507 B.C., was divided into 139 demes – city wards or rural villages – that served as both administrative subdivisions and “self-contained and self-determining units of local government in their own right” (Whitehead 1986).12
Scholarly analysis of multilevel systems seems to have begun with a few paragraphs in Aristotle’s Politics (1996).13 Aristotle begins by deconstructing the Greek city into a three-tier hierarchy of households, villages, and the polis, each of which aims at a different good. In fact, only the third of these is political. The household exists to “supply men’s everyday wants” – material provision and procreation (ibid., pp. 12, 1252a–b). The polis is the setting in which a citizen can participate in self-government, realizing his nature as a being capable of practical wisdom. What purpose there is for something in between is not spelled out. Aristotle says only that the village aims at “something more than the supply of daily needs.” He does not mention the demes (the word used for “village” is kômê).14 In short, the passage does not so much justify multilevel government as explain why some nonpolitical communities are needed to supplement the uniquely political one.
Medieval Christian scholars, starting from this passage, stretched Aristotle’s city into a five- or six-tier hierarchy – rising from the domus (household) to the imperium (Empire) – that more closely approximated their own world (Gierke 1966, p. 277). The particular function of each tier was often left vague. However, Dante, in De Monarchia, provided a reason why multiple levels were necessary. Only in a pyramid of different-sized, nested communities could the full multiplicity of human potential be realized all at once:
There is, then, some distinct function for which humanity as a whole is ordained, a function which neither an individual nor a household, neither a village, nor a city, nor a particular kingdom, has power to perform.…[This function is] to actualize continually the entire capacity of the possible intellect, primarily in speculation, then, through its extension and for its sake, secondarily in action. (Dante 1904 [c.1314–20], Book 1, chs. 3–4)
God created the multi-tiered Empire as the stage on which the different dramas of human life could be simultaneously and peacefully enacted.
Aristotle’s and Dante’s images of social organization may seem foreign. But the notion that different public functions should be assigned to different-sized units in a hierarchy, so that multiple goals can be achieved simultaneously, continues to inform constitutional thinking. Another source of modern ideas about decentralization was the medieval – and, before that, Roman – association of law with custom. As the Institutes of Justinian put it, “immemorial custom approved by consent of those who use it supplies the place of law” (quoted in McIlwain 1932, p. 128). In the fragmented world of medieval Christendom, reverence for custom led naturally to the empowerment of local groups and individuals. Because customs were clearly rooted in particular places, judging what was and was not customary required consulting the locals. In medieval society “the normal way to prove custom was to have it stated by a body of people who represented the community within which it applied” (Reynolds 1984, pp. 42–3). In England, as described by Blackstone, that meant asking a twelve-man jury of local citizens.15 Local tribunals such as the hundred courts, gathering together freemen to interpret custom, established a tradition of local assemblies and popular participation in the administration of justice that extended across the feudal world (Bloch 1961, vol. 2; Reynolds 1984, p. 19). This history implanted a close association between local government, freedom, and democracy into the subconscious of Western societies.
As nation-states solidified in the early modern period, the interaction between central authorities and local communities became a more common preoccupation of political thinkers. One can trace the emergence of three counterposed, ideal-type representations of this relationship. The first casts the state as a top-down hierarchy, in which local officials are subordinate agents of an all-powerful sovereign. The purest exponent is Hobbes, for whom state officials are mere mechanical devices, resembling “the Nerves, and Tendons that move the severall limbs of a body naturall” (Hobbes 1968 , p. 290). So long as these agents faithfully implement the sovereign’s orders, the subjects have an obligation to obey them. Bodin (1992 ) also placed complete authority with the crown but argued for some enlightened – always reversible – delegation.16
A second image conceives the state as the creation of freely associating, self-governing local units, which covenant among themselves to delegate authority upward while retaining their individual sovereignty. This is Montesquieu’s “federal republic” – really a confederation – a “society of societies that make a new one” (Montesquieu 1989 , Part Ⅱ, Book 9, ch. 1). The image, taken up by American Antifederalists such as Melancton Smith, reaches its extreme expression in Proudhon’s ideal state, in which higher governments are the strictly accountable and highly constrained agents of local governments, subject to recall at any time, and limited to contractually pre-specified tasks (Vernon 1979, p. xxiv).17
The third ideal type – that of the “compound republic,” later the “federal state” – appears first in the writing of Harrington (1992 ), Milton, Hugo, and Leibniz, and then in practice in the U.S. Constitution of 1789.18 This conception seems at first to fall between the first two, but in fact it is distinct from both. In a compound republic, neither central nor local governments command the other; they act in parallel, deriving separate grants of authority from a common sovereign. The supreme power, James Wilson declared, “resides in the PEOPLE, as the fountain of government…The power both of the general government, and the State governments, under this system, are acknowledged to be so many emanations of power from the people” (Wood 1969, pp. 530–1). While sovereignty remained inalienable and undivided, it could be exercised simultaneously by several mechanisms. These separate mechanisms – state and federal governments – were equally legitimate, authorized by the sovereign’s consent and by the constitutional compact that defined them and their powers.19