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.
-- -