Wallace E. Oates University of Maryland. College Park. MD and Resources for the FulUle. Washington. DC On the Welfare Gains from Fiscal Decentralization 1. - Introduction The traditional economic case for decentralised fiscal choice is based on the welfare gains from an improved allocation of resources within the public sector. Fiscal decentralizationallows individualjurisdictions(I shall refer to them as «local governments») to tailor the levels of local public outputs to the particular tastes and other circumstances that characterizetheir respective communities. In this paper, I want to explore in more depth the precise nature of these gains, the determinants of their magnitude, and a further dynamic source of welfare gains from so-called laboratory federalism. But first it is necessary to address the issue of why fiscal decentralization is needed to realize these gains. In a world of perfectinformation,a benevolent central planner could presumably introduce the differentiation in local outputs required to maximize social welfare. The traditionalresponse to this point is to argue that, in reality, there exist important imperfections in infonnation. More specifically, individual local governments are much closer to the people and geography of their respective jurisdictions; they possess knowledge of local preferences and cost conditions that are hard to come by for a more distant central agency. There is, in addition, a second (and I think also persuasive) point here. There are, typically, basic political pressures for a certain degree of uniformitY of treatment under central directives. It is not easy for a central government, for example, to provide higher levels of services in some jurisdictions than others. Such political constraints generally prevent central government from introducing the sort of local fiscal differentiation that is requiredfor optimization in a multi-level public sector. There are thus important information and politicallimitations on central government that typically prevent centralized programs from mimicking an optimal pattern of local outputs.The welfare gains, in short, from fiscal decentralization can hardly be expected to be realized through centralized provision of all public goods. Economia delle see/Ie pubbliehe. 2-3. 1997 84 Wallace E. Oates 2. -Welfare gains from variations in demand: the decentralization theorem A more systematic treatment of the welfare gains from fiscal decentralization produces some interesting and useful insights. In this section, we focus our attention on divergences in demand for local public goods and, in the next part, we address the issue of cost differentials across jurisdictions. I will assume throughout the discussion that there are no interjurisdictional spillover effects: the benefits and costs of the provision of local public goods and services are taken to be limited to local residents. '. Figure 1 depicts the demand curves for a local public good of the representative residents of jurisdictions one and two. The assumption here is that the local public good can be provided at a constant cost per unit per resident of Me. The good is thus taken to be subject to congestion in the same way as a private good.2 We see in the figure that the optimal outputs of the local public good are EJ in jurisdiction one and E2in jurisdiction two. Suppose, however, that instead FIGURE1 s MC (J Eo l!c: OutpUtof Local Public Good I There is a large literatUre on the implications of such spillover effectS. One obvious solution in principle (but not so easy in practice) is a system of intergovermental matching grantS (or unit taxes) to the jurisdictions that internalize the spillover benefitS (costS). Other forms of central inrervention or, alternatively, inteljurisdictional bargaining are also possible. 1ThiS. incidentally, is a fairly standard assumption in the local finance literatUre. It is a basic featUre of the Tiebout model. There is. moreover, considerable empirical support for this assumption [OATES, 1996b). It will be relaxed in the next section. On the Welfare Gainsfrom Fiscal Decentrali::ation 85 of this decentralised outcome, the central governmentdetermines that a unifonn level of output of Ec is to be provided in all jurisdictions. It is straightforward to measure the loss in social welfare from centralized provision: the loss is triangle DCE for each resident of jurisdiction two and triangle ABC for each resident of jurisdiction one. It is thus clear that. in this setting, a uniform, centrally determined level of local public outputs will result in a lower level of social welfare than an outcome in which each local jurisdiction provides its own optimal output. This result. incidentally, is the so-called Decentralization Theorem [OATES,1972, ch. 2]. In addition, we can see in Figure I just what determines the magnitude of the gain in social welfare from fiscal decentralization.First.it is clear that the greater the divergence between DJand D2,the larger will be the trianglesABC and CDE. Fiscal decentralization is thus more important where the demand for local pub.lic goods has greater variation across jurisdictions. In countries where the population is quite homogeneous in terms of the demand for local public goods, the potential gains from fiscal decentralization are correspondingly smaller; we would thus expect the political and social forces pushing for devolution to be somewhat weaker than in countries characterized by greater divergences in demands for local outputs. One aspect of this result requires special note. In order for fiscal decentralization to take advantageof differing demands for localpublic goods, people with similar demands must (at least to some degree) be grouped together in jurisdictions. Even if there is wide variation in demand, if people are located randomly across localities, there will be no systematic way to address these differences in demands by varying local outputs. Thus, the extent of the welfare gains from fiscal decentralization depends importantly on the grouping of populations according to individual demands for local public goods. Such segregation can be facilitated to some extent by the mobility of households in response to local fiscal differentials. In the limiting case, the famous Tiebout model [1956] envisions' an outcome in which perfectly mobile households locate injurisdictions that satisfy precisely their demands for local public goods. A Tiebout equilibrium is thus one in which localities are perfectly homogeneous in tenns of demands for the local public good. Figure 1 depicts just such an outcome. This kind of sorting process maximizes the capacity of fiscal decentralization to promote social welfare. At the same time, it is important to recognize that the kind of mobility postulated in the Tiebout model is not necessary to the gains from fiscal decentralization. Indeed, there has been a tendency in some of the literature to identify (at least implicitly) the welfare gains from local finance with the Tiebout model. But this is a serious overstatement. Even if there were no mobility of households across jurisdictions, we would still expect to find interjurisdictional differences in demands (and, perhaps, in costs) for public goods. And these differences would create potential gains in social welfare from decentralised public choice. Tiebout mobility enhances these gains, but should not be viewed as their sole source. - -------- 86 Wallace E. Oates Figure I provides a second insight into the determinants of the welfare gains from fiscal decenn-alization.We see that the size of the welfare-gain-triangles ABC and CDE depends on the slope of the demand curves. More specifically, the steeper are the demand curves, the larger are the triangles and hence the greater is the gain in social welfare from differentiationin local outputs. This follows because where demand is less price responsive, individual valuations of marginal units change relatively rapidly as we move away from the optimum.3 This second result is of more than purely theoretical interest, because we have some estimates of the price elasticity of demand for local public goods. There is a substantial econometric literature (going back to the seminal paper Bergsn-om and Goodman [1973]) that has provided estimates of the demand for local public goods. And this literature suggests that the demand for most local public goods is typically highly price inelastic on the order of O.3 to O.4:~The implication is that in the presence of significant differences in demands for local services, the potential welfare gains from fiscal decenttalization may be large. There is one study [BRADFORD and OATES,1974] that produced actual estimates of these gains and found them to be quite sizeable. - 3. -lnterjurisdictiorial decentralization cost differences and the welfare gains from fiscal Of course, cost differentials across jurisdictions (as well as differences in demands) can be a source of welfare gains from fiscal decentralization. Figure 2 depicts such a case. Suppose that everyone has the same demand for local public goods, namely demand curve DD. but that the marginal cost of providing a unit per person differs as between the twojurisdictions (MCI, injurisdiction one and MCz injurisdiction two). In Figure 2 the Pareto efficientoutcomes are EI and Ez respectively. In this case, centralized provision of a uniform level of output, E.. results in welfare losses per resident of triangle ABC injurisdiction one and triangle CDE in jurisdiction two. We see two results from the diagram. First, it is obvious that the size of the welfare loss trianglesvaries directly with the magnitudeof the interjurisdictional cost differential. The greater the distance between MC, and MCz, the farther will the optimal outputs in the twojurisdictions divergefrom one another and the larger will be the social loss in welfare from a centtally determined, uniform level of output. )This result. incidentally. is just the opposite from taXanalysis. We find in the theory of taXation that the welfare (or deadweight) loss from a taX varies directly with the price elasticity of demand. Here the welfare toss from fiscal cenualization varies inversely with the price responsiveness of demand. This happens because in the case of taXation we are introducing the source of the distortion along the price axis. while here the distortion has its source on the quantity axis. For surveys of this econometric literarure. see Rubinfeld [1987] and Oates [1996b]. · 87 On the Welfare Gainsfrom Fiscal Decentralization FIGURE2 $ E'I Output of Local Public Good Second, in contrast to the variation in the demand case, we find that, for the case of cost differences, the welfare gain from fiscal decentralization varies inversely with the absolute value of the slope of the demand curve. In this case, the less steep are the demand curves (i.e., the more price-responsive is the demand for local outputs), the more divergent will be the efficient outputs in the two jurisdictions and the greater the loss in social welfare associated with a centrally prescribed and uniform level of local outputs. In Figure 2, we see that for the more price-elastic demand curve D'D', the welfare-gain triangles from fiscal decentralization increase to CBF in jurisdiction one and CDG in jurisdiction two. So we find that the effect of the price responsivenessof demand on the p0tential welfare gains from fiscal decentralization depends on whether the divergence in Pareto-efficient local outputs has its source in interjurisdictional variation in demand or variation in costs. S Interjurisdictional cost differentials can result from two quite different sources. First, it may simply require more in the way of inputs to provide a given level of output in one place than another. For example, keeping the roads clear of snow in the winter will require more effort in an area that gets lots of snow than in one with a milder winter season. Thus, the difference between MC/> and MC2 in Fig- , This interesting asymmetry. incidentally, has a basic rationale similar to that for the famous Weitz. man Theorem [1974] on the choice betWeen price and quantity insttUments. - -- ---- --- 88 Wallace E. Oates ure 2 may simply result from differences in the production functions such that one jurisdiction requires more inputs per unit of output than another. But there is a second and more intriguing source of cost differences. Public goods may exhibit quite different congestion characteristics.In the TIebout model. for example, an extreme assumption is made: local public goods are postulated to be fully congestible (like private goods) in the sense that a doubling of the number of people who consume the good requires a doubling of the quantities of inputs in order to keep public output per resident the same. Suppose, in contrast, that local public goods exhibit some degree of «publicness». To take the extreme case, let us assume that they are pure public goods in the sense that a given unit of output can be consumed by all residents of the jurisdiction with no congestion effects. Or, put slightlydifferently,an additional reso'ident in a jurisdiction will not reduce the levels of consumption of local public outputs of existing residents. In this instance, the cost per person of providing incremental units of the local public good will fall with the size of the population. In terms of Figure 2, the lower marginal cost of public outputs injurisdiction two, in this instance, would be the result of a larger population in two than in jurisdiction one.6 An interesting and important example of interjurisdictionalcost differentials that have their source both in differences in costsof production and in economies of scale is the provision of safe drinking water.The quality of the natural waters varies significantlyacross geographical areas.--Incertainplaces, for instance, contaminants are present that are absent elsewhere - and the degree of contamination by a particular pollutant can vary widely from place to place. But even more important are the enormous economies of scale available in treatment. It is far more costly per capita to treat drinking water in a system that serves only a small number of people than one which services a large population. The result is that there exist very large local differences in the costsper person of reducing the levels of various drinking-water contaminants. As a recent study [U.S. CONGRESSIONAL BUDGET OFFICE,1997] notes, the benefits and costs of achieving various standards of purity differ dramatically across jurisdictions. As a result, «Standards that pass an overall [i.e.. from the perspectiveof the nation as a whole] costbenefit-test [...] may not be efficient for small systems» [poxii]. Since the benefits from safe drinking water are limited primarily to local residents (as it is long-term exposure that is important for most of the major contaminants), this appears to be a case where decentralisedprovision offers significant welfare gains. These potential gains were ignored in the U.S. in some ear- 6This case of economies of scale with respect to group size. incidentally, raises some interesting efficiency issues in its own right. In their seminal paper. Flatters. Henderson. and Mieszkowski (1974] demonstrated that costless mobility of individuals can lead [0 inefficient outcomes. The basic problem is that individuals seek [0 locate in relatively populous jurisdictions in order to reduce the tax-price of public outputs. This can lead to excessive concentrations of population in cases where there are diminishing recums to labor in private production. ---- - - - - - On the Welfare Gains/rom Fiscal Decentraliz.ation 89 ly legislation; the Safe Drinking Water Act of 1974 prescribed a set of federal standards for drinking water quality to be met in alljurisdictions. But it has become clear that especially for smaller treatment systems, some of the standards result in costs that far exceed any reasonable estimatesof the benefits. In response to vehement protests from many parts of the country,morerecent legislation now allows a range of exceptions to the national standards. 4. - Welfare gains from innovation in the provision of public services: laboratory federalism The analysis to this point has focused on the welfare gains from fiscal decentr.Uization in a setting of a given «technology» of public provision of local outputs. But there is another way in which decentralization can contribute to an improved perfonnance of the public sector. It can encourage experimentation with a variety of kinds of public programs that can lead to the discovery of better ways of organizing and providing public services. As James Bryce [1888] observed in the last century, «Federalism enables a people to try experiments which could not safely be tried in a large centralized country» [Vol. I, p. 353]. 1bis was echoed in 1932 by Justice Louis Brandeis, who wrote: «There must be power in the States and the Nation to remould. through experimentation, our economic practices and institutions to meet changing social and economic needs [...]. It is one of the happy incidents of the federal system that a single courageous State may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country [OSBORNE, 19~8]». There have, in fact, been a number of cases in U.S. federalism where policies that proved successful at state and local levels have later been adopted at the nationallevel. To take one example, unemployment insurance existed in a number of states before the federal government made it effectively a mandatory national program during the 1930s. A more recent and intriguing example involves the use of a novel technique for environmental management. In the 1990 Amendments to the Clean Air Act, the U.S. adopted a national system of tradeable emissions allowances to address the acid-rain problem. Under this program, the federal government issued a limited number of permits for emissions of sulfur into the atmosphere. Holders of these permits are free to buy and sell them so that a market in emissions allowances has emerged. Estimates suggest that the program has resulted in large cost savings relative to the more traditional commandand-control approach. Such a program would, I feel sure, never have been enacted without the precedent of Emissions Trading at state and local levels. Under a variety of Emissions Trading programs encompassing several air pollutants, state and local governments have, over the past twenty years, made effective use of market forces for air quality management. It was this experience at the state and local levels that provided the basis for the introduction of the sulfur trading program at the national level. - ---------- --- --- 90 Wallace E. Oates While laboratory federalism offers an interesting perspective on the welfareenhancing potential of fiscal decentralization, there is little in the way of a systematic theoretical framework for exploring its properties. There have been a number of studies by political scientists that examine the pattern of diffusion of successful policy innovations.7But there is a real need for some basic conceptual (as well as empirical) work to deepen our understanding of the role of fiscal decentralization in policy innovation. 5. - A concluding note on efficiency in decentralized finance In this paper, the discussion has focused on the nature of the gains from fiscal decentralization. We have reviewed the sources of these gains and the determinants of their magnitude. While there is a compellingpresumption that these gains are unlikely to be realized in a system of wholly centralized public deci. sion-making, there remains the issue of the extent to which decentralised government can exploit these opportunities to increase social welfare. The implicit assumption that runs through the analysis here is that local governments will, in fact, provide (at least roughly) the Pareto-efficientlevels of outputs for their respective jurisdictions. However, we are weU aware of the kinds of problems that arise in collective decisions. Niskanen-types of public agents may have their own agendas with some scope to implement them so that certain kinds of «monopolistic~ outcomes occur in the public sector. But even where public decision-makers respond readily to voter preferences, we find that such things as the median-voter outcome are fully efficient only under special conditions. In a setting of decentralized finance, things can become even more complicated. There is now a large literature that explores the implications of interjurisdictonal competition. The claim here is that in their eagerness to promote economic growth and create new jobs, «local» officials will keep tax rates below their efficient levels and will adopt excessively lax regulatory measures on such things as the environment in order to attract new firms. As a result, we can expect subcentral governments to underprovide public outputs. This is a serious indictment of decentralised finance. Taken at face value, it implies that provincial and local governments cannot be expected to perform well- they will not provideefficientlevels of public outputs. But the theory on this matter is not at all clear.It is straightforward, for example, to constrUctmodels of decentralised fiscaldecision making in which interjurisdictional competition leads to fully efficient outcomes [e.g., OATESand SCHWAB, 1988]. In such models, competition provides the right sorts of signals for allocative decisionsjust as it does in a competitivemarket. Adam Smith's in- 7 - See Oates (1998] for a brief survey of this work. -- --- On rh~ W~lfar~ Gainsfrom Fiscal D~c~nrrali:.arion 91 visible hand works here too. On the other hand, it is not difficult to introduce a variety of realistic elementS into these models (e.g.. limitations on fiscal instrumentS, strategic behavior, etc.) that can introducedistortions,including, in some cases, underprovision of public services.8The real question, it seems to me, is an empirical one. How large are these distortions(if any) and bow do they compare with the welfare losses associated with centralizedprovision? My sens~ is that decentralised finance comes off reasonablywell when seen in these terms. But we really do not have much firm empirical evidence with which to answer this question [COURANT, 1994 and OATES,1996b]. : REFERENCES BERGSTROM, T. C., GooDMAN,R. P., «Private Demands for Public Goods», Anurican Economic Review. 63, 1973, pp. 280-96. BRADFORD, D. F., OATES,W. E., «Suburban Exploitationof Central Cities and Govermental Structure», in HOCHMAN, H. and PETERsON, G. (cds.), Redistribution Through Public Choice, New York. Columbia UniversityPres~, 1974,pp. 43-90. BRYCE.J., The American Commonwealth, London, Macmillan, 1901 [first published in 1888]. COURANT, P. N., «How Would You Know a GoodEconomicPolicyIf You Tripped Over One? Hint: Don't Just Count Jobs», National TaxJournal, 47,1994, pp. 863-881. FLATTERS, F., HENDERSON, V. and MlEszKOWSKl, P., «PublicGoods,Efficiency, and Regional Fiscal Equalization», Journal of Public Economics.3, 1974, pp. 99-112. OATES,W. E., Fiscal Federalism. New York, HarcourtBrace Jovanovich. 1972. OATES,W. E., «Estimating the Demand for Public Goods: The Collective Choice and Contingent Valuation Approaches», in D. BJORNSTAD, and 1. KAHN,(eds.), The Contingent Valuation of Environmental Resources, Aldershot, U.K.. Edward Elgar, 1996a. pp. 211-232. OATES,W. E., «The Invisible Hand in the Public Sector:Interjurisdictional Competition in Theory and Practice», unpublished paper, 1996b. OATES,W. E., «An Essay on Fiscal Federalism», 1998. OATES,W. E., SCHWAB, R. M., «Economic CompetitionAmongJurisdictions», Journal of Public Economics. 35, 1988, pp. 333-354. OSBORNE. D., Laboratories of Democracy. Boston, HarvardBusinessSchool Press, 1988. RUBINFELD. D. L., «The Economics of the Local Public Sector». in A. AUERBACH and M. FELDSTEIN (eds.), Handbook of Public Economics. Amsterdam, Nonh Holland, 1987. pp. 571-645. 'For an excellent survey of this literatUre. see lohn Wilson (1996]. ------- --- 92 Wallace E. Oates TIEBOUT, C. M., «A Pure lbeory of Local Expenditures»,Journal of Political Economy, 64, 1956, pp. 416-424. U.S. CONGRESSIONAL BUDGET OFFICE.Federalism and Environmental Protection: Case Studies/or Drinking Water and Ground-Level Ozone, Washington D.C., 1997. WEITZMAN, M. L., «Prices vs. Quantities», Review of Economic Studies, 41, 1974, pp. 477-491. Wo.sON,J. D., «Capital Mobility and Environmental Standards: Is There a Theoretical Basis for a Race to the Bottom?» in J. BHAGWATI, and R. HUDEC,(eds.). Fair Trade and Harmonization: Prerequisites for Free Trade? Vol. I. Cambridge, Mass., MIT Press, 1996, pp. 393-427. -- -