theories of tax competition - national tax journal

Forum on Mobility and Tax Analysis

269

Abstract - A central message of the tax competition literature isthat independent governments engage in wasteful competition forscarce capital through reductions in tax rates and public expendi-ture levels. This paper discusses many of the contributions to thisliterature, ranging from early demonstrations of wasteful tax com-petition to more recent contributions that identify efficiency-enhancing roles for competition among governments. Such rolesinvolve considerations not present in earlier models, including im-perfectly-competitive market structures, government commitmentproblems, and political economy considerations.

INTRODUCTION

The modern literature on tax competition began with anattempt to understand the potential efficiency problems

associated with competition for capital by local governments.Oates (1972, p. 143) describes this problem as follows:

“The result of tax competition may well be a tendency towardless than efficient levels of output of local services. In an at-tempt to keep taxes low to attract business investment, localofficials may hold spending below those levels for which mar-ginal benefits equal marginal costs, particularly for those pro-grams that do not offer direct benefits to local business.”

In other words, local officials will supplement the conven-tional measures of marginal costs with those costs arisingfrom the negative impact of taxation on business investment.These additional costs might include lower wages and em-ployment levels, capital losses on homes or other assets, andreduced tax bases. Their presence will reduce public spend-ing and taxes to levels where the marginal benefits equal thehigher marginal costs. Oates’s conclusion that this behavioris inefficient rests on the idea that when all governments be-have this way, none gain a competitive advantage, and con-sequently communities are all worse off than they would havebeen if local officials had simply used the conventional mea-sures of marginal costs in their decision rules.

Since the mid-1980s, there has been an outpouring of aca-demic research on tax competition, and this research contin-ues unabated. Interest in this area has been stimulated byhighly publicized instances where U.S. states and localitiesdo seem to have engaged in tax competition, including themany cases where they have offered large subsidies to for-eign and domestic automobile companies in an attempt to

Theories of Tax Competition

John DouglasWilsonDepartment ofEconomics,Michigan StateUniversity, EastLansing, MI 48824

NATIONAL TAX JOURNAL

270

influence plant location decisions. In ad-dition, researchers and policymakers havefound that Oates’s (1972) description of taxcompetition can be applied more broadlyto a host of important policy concerns, suchas competition for investment throughweaker environmental standards or reduc-tions in welfare payments by states tryingto avoid attracting poor households.

More generally, the view that intergov-ernmental competition is wasteful hasraised fundamental questions about theappropriate roles of the central govern-ment and lower level governments. Inparticular, this view runs counter to thehighly influential “Tiebout Hypothesis.”Tiebout (1956) argues that competition formobile households is welfare enhancing,and subsequent work has applied similarideas to competition for mobile firms, indirect conflict with much of the tax com-petition literature. This conflict is reflectedin the ongoing “devolution debate” overthe desirability of “devolving” federal re-sponsibilities to lower levels of govern-ment within the United States.1

Moreover, this conflict has surfaced indebates concerning the appropriate de-gree of economic integration betweencountries. In Europe, integration hasproceeded to the point where barriers tolabor and capital mobility have beenofficially dismantled (although culturalbarriers remain). The Tiebout literaturesupports policies that allow free factormobility and enable national governmentsto function independently in most policyareas, and the Treaty on the EuropeanUnion reflects a presumption in favor ofthis independence. In contrast, the taxcompetition literature draws attention tothe potentially adverse effects of this in-dependence. Similar concerns about inte-gration surface in Sinn’s (1997) discussionof the “selection principle,” which identi-fies a “fundamental selection bias towards[government] activities that have proved

to be unsuitable for private markets” (p.248). Through a series of examples, Sinnessentially argues that if private marketsfail to efficiently provide particular goodsand services, then introducing competi-tion among governments that seek to pro-vide them will generate similar problems.

Given the central importance of tax com-petition in these ongoing policy debates,the time seems ripe to assess what we havelearned from the large and growing theo-retical literature on this phenomenon. Thisis the objective of the present paper.

The scope of my discussion is broaderthan Oates’s (1972) original description oftax competition. Thus, I discuss modelsin which the governments may be inter-preted as local, state, or provincial gov-ernments within a country, or as countrieswithin the world economy. I often use theterm “region” to encompass these variousinterpretations. A common feature of mostof the models considered here is thateach government independently (or“noncooperatively”) chooses its tax orsubsidy policies to maximize the welfareof residents within the region, and itschoice affects the size of the tax basesavailable to other governments. These taxbases often consist of mobile capital, but Iconsider alternatives, such as competitionfor cross-border shoppers. While the mainfocus of my discussion is on the nonco-operative choice of taxes or subsidies, Ialso discuss cases in which governmentscompete by using nontax instrumentssuch as expenditure or regulatory policies.

My survey shows that the initial formalmodels of tax competition largely confirmOates’s (1972) conjecture, since they stickmost closely to the situation he envi-sioned. Subsequent extensions of thesemodels produce a variety of inefficienciesthat differ from the original conclusionthat taxes and expenditures are ineffi-ciently low. Toward the end of the paper,I discuss recent work that identifies some

1 See Tannenwald (1998) for an overview of this debate.


271

beneficial effects of competition amonggovernments that seem not to have beenpreviously recognized, in part because theyrepresent departures from both the stan-dard tax competition models and Tieboutmodels. My concluding remarks call formore work on the potentially importanttrade-offs between the good and bad as-pects of intergovernmental competition.

This paper is organized as follows. I be-gin in the next section by discussing theidealized Tiebout world and how tax com-petition models typically depart from it.The third section then surveys a large num-ber of models in which governments usetax policies to compete for mobile capital.Throughout this discussion, firms are as-sumed to be competitive, and only taxes onmobile capital are considered. The fourthsection discusses several attempts to ex-pand the basic model to include multipletax instruments, including models whereboth capital and labor are mobile. In thefifth section, I change the focus to “com-modity tax competition,” where regionscompete for cross-border shoppers, a situ-ation that is relevant not only for regions,states, or provinces within a single coun-try, but also for countries with few bordercontrols, such as the European Union. Thesixth section briefly discusses competitioninvolving nontax instruments. In the sev-enth section, I turn to two areas of researchwhere the concern has been that taxes aretoo high, not too low: competition betweendifferent levels of government (“vertical taxcompetition”) and the use of double taxa-tion conventions. The eighth section thendiscuses several lines of research that haveidentified some possible efficiency-enhanc-ing roles for tax competition. The ninth sec-tion provides conclusions.

To keep this survey manageable, I re-strict its focus in several respects. In keep-ing with common practice in the literature,I limit the discussion to full employment

models. As Huang (1992) argues, unem-ployment may provide an additional in-centive for wasteful tax competition, sincegovernments now benefit from the em-ployment generated by additional capital.2

My discussion of competition for capitalis also restricted to “industrial capital,”rather than “residential capital.” The lat-ter type of competition has been investi-gated using what Mieszkowski andZodrow (1989) refer to as “metropolitanmodels.” Such models contain residentialhousing and assume interjurisdictionallymobile residents, making them less appli-cable to tax competition between countries.Although I briefly discuss multinationals,interested readers should consult morespecialized surveys on this topic, such asGresik (1998) and Hines (1997, 1999). Fi-nally, my focus on theories of tax competi-tion precludes a detailed examination ofthe empirical work on its existence; seeBrueckner and Saavedra (1998) for someempirical evidence and related references.

THE TIEBOUT MODEL

Tiebout’s (1956) theory of local publicgood provision also provides a theory ofefficient tax competition. In modern for-mulations of the theory, it is often as-sumed that each region’s government iscontrolled by its landowners, who seek tomaximize the after-tax value of theregion’s land by attracting individuals toreside on this land. To do so, the govern-ment offers public goods that are financedby local taxes. A critical assumption is thatthere are many “utility-taking” regions, inthe sense that no single region can alterthe utilities that must be offered to indi-viduals to induce them to reside there.Thus, the model is quite similar to mod-els of competitive markets for privategoods, with land rents serving as profitsand utilities serving as prices. Therefore,

2 Haaparanta (1996) also examines a model with unemployment, using the common agency approach dis-cussed below.


272

it is not surprising that the equilibria forsuch models are found to be efficient un-der the usual definition of efficiency: acentral authority cannot feasibly reallocategoods and resources in a way that makessome individuals better off without mak-ing anyone worse off. There is tax compe-tition here in the sense that a region’s taxesmust be kept low enough to induce indi-viduals to reside in the region, given thepublic goods that are being provided.These taxes are collected from residentsin the form of efficient “head taxes,” andthey are chosen so that each resident’s taxpayment equals the cost of providing himwith the chosen levels of public goods andservices. This marginal-cost-pricing ruleresults in efficient migration decisions.

There is now a vast literature extendingthese efficiency results in various direc-tions. As originally suggested by Fischel(1975) and White (1975), the theory maybe easily extended to include mobile firms.These firms are modeled in effectively thesame way as mobile residents by assum-ing that they are in infinitely elastic sup-ply to any given region and that the re-gion supplies them with various “publicinputs.” In equilibrium, each firm is as-sessed a tax equal to the marginal cost ofsupplying it with these public inputs, andthe equilibrium is efficient. See Richter andWellisch (1996) for a detailed developmentof this extension. Wellisch (forthcoming)discusses this and several other exten-sions. Note, in particular, that the modelmay be extended to include not only mo-bile firms, but also mobility of both thelabor and capital that they employ.

Wasteful tax competition involves sometype of departure from the idealized set-tings of “Tiebout models.” The mainsource of departure is the existence of“interregional externalities,” whereby the

actions that one region’s governmenttakes to increase the welfare of its ownresidents leads to reductions in the wel-fare of residents in other regions. In thetax competition literature, this external-ity is often described as a “fiscal external-ity,” which occurs through the effects ofone region’s public policies on the gov-ernment budgets in another region(Wildasin, 1989). For example, when a re-gion lowers its tax rate on mobile capital,it gains capital at the expense of other re-gions, causing their tax bases to fall and,hence, their tax revenues to decline. Suchexternalities are often present in tax com-petition models because governments areassumed not to possess unlimited taxingpowers. In particular, the assumption ofefficient taxes on residents or firms isclearly violated in practice.3

Another type of externality is the “pe-cuniary externality” that exists whenregions are large enough to affect theproduct or factor prices confronting otherregions. I show in the “Large Regions”section that these externalities lead to in-efficient policy differences across regions,causing a misallocation of factors. Theydo not, however, allow us to conclude thattaxes and public spending are too low insome overall sense.

Other interregional externalities arise asa result of inefficiencies in private markets,coupled with the failure of governmentsto correct them. In particular, this studybriefly addresses the issue of tax compe-tition under imperfect competition.

Finally, regional governments maymake policy choices that are not in the bestinterests of their residents. Such behaviormay give rise to various interregional ex-ternalities. However, we cannot say apriori whether the interregional mobilityof factors improves or worsens the behav-

3 Hamilton (1975, 1976) argues that the zoning policies employed by local governments can turn property taxesinto efficient “user fees” for local public goods, but the availability and use of efficient zoning policies is opento question. See Mieszkowski and Zodrow (1989) and Ross and Yinger (1998) for critical assessments ofHamilton’s view, and see Hoyt (1991a), Krelove (1993), and Wilson (1997) for analyses of the residentialproperty tax as an “inefficient user fee.”


273

ior of these governments. The “PoliticalEconomy” section discusses a model inwhich government behavior is made moreefficient by the presence of free factor mo-bility.

CAPITAL TAX COMPETITION

Following Oates’s (1972) discussion oftax competition, it was not until the mid-1980s that economists began to build for-mal models based on his ideas. Two of theearliest papers were Zodrow andMieszkowski (1986) and Wilson (1986).4

Since the production structure in Zodrowand Mieszkowski’s model is simpler thanWilson’s structure, I let their model serveas the “basic tax competition model” inthis survey. The model is next presentedin its simplest form, and then it is ex-tended in various directions.

Competitive Regions

Consider a system of many regions,which may be variously interpreted as cit-ies, states, provinces, or countries. Withineach region, competitive firms produce asingle output, using two factors of produc-tion: mobile “capital” and an immobilefactor that I will call “labor.”5 The immo-bile factor is inelastically supplied by theregion’s residents, who are themselves im-mobile (an assumption that I drop in the“Labor Mobility” section). These residentsalso own fixed endowments of capital,and the assumption of perfect capitalmobility means that they are free to in-vest their capital anywhere.6 Once invest-ment and production take place, theoutput is sold to residents as a final con-sumption good, and to the government asan intermediate good, which it then trans-

forms into a public good. Income distri-bution issues are ignored by assumingeither that each region’s residents areidentical or that their aggregate welfarecan be depicted by the preferences of a“representative consumer.” In particular,these preferences are represented by awell-behaved utility function, U(C, G),where C is private consumption and G isconsumption of the public good. The rep-resentative consumer finances C with thewage and capital income from her endow-ments of labor and capital. Summing thecapital endowments across the residentsin all regions gives the fixed supply ofcapital in the “world economy.”

A critical assumption in this model isthat each region’s public good supply isfinanced by a tax on the capital employedwithin its borders. One simple justifica-tion for this assumption is that the gov-ernment finds it administratively conve-nient to tax both capital and land at thesame rate, as in the case of local propertytaxation in the United States. I later dis-cuss a three-factor model that explicitlyincludes a uniform property tax on mo-bile capital and immobile land, and I alsodiscuss models where governmentschoose among multiple tax instruments.But for the “basic model,” I followZodrow and Mieszkowski (1986) by as-suming that only capital is taxed, althougheach region’s residents would be better offpaying a head tax.

The problem confronting a region’sgovernment is to choose the unit tax rateon capital, t, to maximize the representa-tive consumer’s utility, U(C, G), subject toa budget constraint requiring that tax rev-enue equal public good expenditures:

[1] tK(r + t) = G

4 See Beck (1983) for another early, but less general, treatment of tax competition.5 Zodrow and Mieszkowsk (1986) interpret the immobile factor as land and assume no absentee ownership, an

assumption that I address below. They also allow regions to possess market power, an extension I leave to the“Large Regions” section.

6 The tax competition literature has devoted little attention to the intermediate case of “imperfect” capitalmobility. For an exception, see Lee (1997).


274

where r is the after-tax return on capital,which is equated across regions by capitalmobility but treated as fixed by any smallregion; and K(r + t) is a function relatingthe demand for capital in the region to thecost of capital, r + t.7 As more capital isinvested in the region, its marginal prod-uct falls and the marginal product of laborrises. Firms invest to the point where themarginal product of capital equals r + t.

Consider now the government’soptimality condition for G. To finance aunit rise in G, the government must raiset. As a result, the cost of capital rises, caus-ing the demand for capital to change bysome negative amount, ∆K. Since r is fixedfrom the region’s viewpoint, however, thehigher tax rate does not reduce the resi-dents’ capital incomes. Instead, they in-directly pay the tax through a decline intheir wages. In particular, if a unit rise inG requires that t be raised by ∆t to bal-ance the budget, then wage income willfall by the resulting rise in the cost of capi-tal, K∆t, to prevent firms from earningnegative profits. Unfortunately, this taxincrease must be high enough to not onlypay for the “marginal resource cost” of G,denoted MC, but also to offset the nega-tive impact of the capital outflow on taxrevenue, t∆K. Thus, the resulting fall inthe residents’ wage income will exceedMC by the positive amount, –t∆K. At theregion’s optimal level of G, the sum of theresidents’ marginal willingness to pay foranother unit of G, or “marginal benefit”of G, is equated to this wage reduction:

[2] MB = MC – t∆K

Alternatively, it is possible to use the de-pendence of ∆K on the elasticity of the

demand for capital with respect to the costof capital to rewrite equation 2 as follows:8

[3] MB = 1 – τε

where ε denotes this demand elasticity(measured positively) and τ is the ad valo-rem tax rate, τ = t/(r + t).

To conclude, both of these rules dem-onstrate that the marginal benefit ofG exceeds the marginal resource cost tocompensate for the tax-induced capitaloutflow. In other words, we have a “modi-fied” Samuelson rule for public good pro-vision, since the actual Samuelson rule forefficient public good provision would re-quire that MB = MC.

The tax rate t in these rules has an im-portant interpretation. It represents the dis-crepancy between the social value of anadditional unit of capital and the socialopportunity cost of this unit, measuredfrom the single region’s viewpoint. Thissocial value is the marginal product of capi-tal, MPK, which firms equate tor + t when they choose their profit-maxi-mizing investment levels. In contrast, thesocial opportunity cost is only r, since thetax provides revenue for the governmentand is therefore not a social cost. Thus,t = MPK – r. It is this discrepancy betweenthe value and opportunity cost of capitalat the margin that implies that the regionbenefits from a capital inflow and isharmed by a capital outflow. In a more gen-eral model, we would recognize that capi-tal investments impose various burdens onthe public sector, such as the increased de-mands for public infrastructure and vari-ous public services. If capital were taxedto cover the marginal costs of these publicgoods and services, then the region would

MC

7 For simplicity, the literature often specifies capital taxes as tax rates on each unit of capital. The distinctionbetween unit and ad valorem taxes is irrelevant for the current model. When regions with market power arelater considered, the equilibrium can depend on the form of taxation, but its qualitative properties do not change.

8 Using calculus notation, ∆K = (dK/dt)(dt/dG), where dt/dG is the rise in t needed to finance a unit increase inG. As noted in the text, wages fall by K(dt/dG) to offset the higher capital costs, and this wage reduction is setequal to MB at the optimum. Thus, we also have ∆K = (dK/dt)(MB/K). Substituting this relation into equation2 and rearranging gives equation 3.


275

be indifferent about a marginal inflow oroutflow of capital, and so the terms involv-ing t in equations 2 and 3 would disappear.But if capital is not taxed efficiently and weview the term t more generally as the dif-ference between the social value and so-cial opportunity cost of capital at the mar-gin, then equations 2 and 3 remain valid.The subsequent analysis therefore can begiven this more general interpretation.9

Having described a single region’s op-timal public good supply, let us now con-sider whether it is efficient for the systemof regions as a whole. A critical insightfrom the tax competition literature is thata rise in one region’s tax rate benefits otherregions by increasing their capital supplies.Under the assumed fixity of the total capi-tal stock, the tax-induced outflow of capi-tal from the given region represents an in-flow of capital to other regions, and thevalue of this inflow will depend on the taxrates being set in these regions. In particu-lar, another region j benefits by the amounttj∆Kj, where tj is its tax rate and ∆Kj is thecapital inflow that it experiences.10 In otherwords, a rise in the given region’s tax ratecreates a positive externality. The region’sgovernment fails to account for such ex-ternal benefits because it is concerned onlywith the welfare of its own residents. Con-sequently, it sets its tax rates and publicgood levels at inefficiently low levels.

The tax competition literature has fo-cused heavily on the case where all regionsare identical and therefore choose the sametax rates. This case nicely isolates ineffi-ciencies in the overall level of public goodprovision from the efficiency and equityissues concerning differences in tax ratesand public good levels across regions. Inthis case, the cost of a capital outflow fromone region is exactly offset by the benefitsfrom the accompanying capital inflows toother regions. Hence, ∆K disappears fromrule 2 and we are left with the rule for effi-

cient public good provision, MB = MC.Starting from the equilibrium level of pub-lic good provision, where MB > MC, wecan satisfy the efficiency condition by in-creasing all regions’ tax rates and publicgood levels by identical amounts. Thesechanges raise welfare in all regions.

Another way to satisfy the efficiencycondition would be to close each region’sborders to factor mobility, so that theregion’s firms could use only the capitalsupplied by its residents. In this case, theinterregional externalities described abovewould disappear, and the capital taxwould become equivalent to an efficientlump-sum tax. Thus, each governmentwould independently choose to set itspublic good supply where MB = MC. Withthe borders open, however, no single gov-ernment has an incentive to raise G to thepoint where MB = MC, given the costs as-sociated with the resulting capital outflow.There must be some type of coordinationamong regions, suggesting a role for a cen-tral authority. This role is discussed belowin the context of nonidentical regions.

Finally, two considerations may miti-gate the tendency of regions to under-provide public goods. First, supposethat the model is modified to include avariable supply of capital for the systemof regions as a whole, due to the savingsbehavior of residents. If a subset of regionsincreases its tax rates, then total savingsmay decline, dampening the amount ofcapital that is redirected to other regions.The interregional externalities remain,but their importance is somewhat reduced.We may conclude that allowing a variablesupply of capital reduces, but does noteliminate, the tax competition problem.

Second, extending the basic modelto include absentee ownership of theimmobile factor, reinterpreted as “land,”introduces “tax exporting” into the analy-sis. The capital tax is now capitalized into

9 This discussion assumes an exogenous level of public infrastructure. See footnote 27 on this issue.10 I use subscripts to identify regions only in cases where they are needed to avoid confusion.


276

the return on land, passing part of itsburden on to nonresidents. This form oftax exporting counteracts the effects oftax competition, thereby raising the sup-ply of public goods.11 As an empirical mat-ter, it seems unlikely that householdscome close to owning the diversified port-folio of land needed to fully eliminate theunderprovision problem in the basic taxcompetition model.

Corrective Policies

Regions may choose different tax ratesbecause they possess different productiontechnologies or their residents possess dif-ferent preferences or factor endowments.If the tax rates do differ, then two types ofinefficiencies exist in the economy. First,public good levels are set inefficiently, be-cause regions fail to account for the inter-regional externalities discussed above.Second, capital is misallocated across re-gions, so that the marginal product of capi-tal is relatively high in high-tax regions. Afully efficient allocation cannot beachieved if tax rates differ across regions,and identical tax rates are usually not con-sistent with efficient differences in publicgood levels across regions, unless a cen-tral authority also redistributes revenueacross the government treasuries.

The central authority can achieve thisefficient allocation by providing each re-gion with a “corrective subsidy” on therevenue it raises, while at the same timeengaging in lump-sum transfers of incomebetween the regional governments. Insymbols, a region-i government faces agrant schedule of the form

[4] Si = ai + si(tK)

where the ai’s are positive or negative, re-flecting the interregional transfers, and thesi’s denote the corrective subsidies. Thesesubsidies have been suggested by Wildasin(1989) and analyzed further by DePaterand Myers (1994). For a system of Nregions, the ai’s and si’s represent 2N vari-ables that are chosen to satisfy the 2N con-ditions for an efficient allocation: the N – 1equations requiring that all regions choosethe same tax rates, the N equations requir-ing that each region’s public good level beset where MB = MC, and the requirementthat the central authority’s budget balance.In equilibrium, the corrective subsidieswill generally differ across regions to in-duce them to choose the same tax rates.But the critical point here is that all correc-tive subsidies should be positive inequilbrium. Only then will regions be com-pensated for the positive externalities as-sociated with increases in their tax rates.12

In practice, however, it is doubtful thatthe policy intervention described by equa-tion 4 would be feasible, since it wouldrequire that the grant functions be tailoredto the individual characteristics of regions,which might be difficult for a central au-thority to observe. A recent paper byBucovetsky, Marchand, and Pestieau(1998) designs an alternative form ofpolicy intervention that takes into accountthe central authority’s information prob-lem. Their model contains two types ofregions, distinguished only by the resi-dents’ demands for public goods.13 Thecentral authority has two policy instru-ments: a “national capital tax,” T, and a

11 See Lee (1998) for a fuller treatment of the case of absentee landowners. Burbidge and Myers (1994) providelimited efficiency results for a three-factor model where capital and labor are mobile between two nonidenticalregions and each individual owns equal amounts of capital, labor, and land in each region. But Bucovetsky(1995) emphasizes the inefficiencies that emerge from this setup when factor endowments are allowed to differ.

12 By focusing solely on efficiency issues, this analysis ignores issues involving the interregional distribution ofincome. Given the limited policy instruments available to a central authority, it might want to deviate from anefficient allocation to reduce inequities in the income distribution.

13 Following the basic tax competition model, Bucovetsky et al. (1998) assume many regions of each type. Forthe difficult problem of regulating large regions that possess informational advantages, see Dhillon, Perroni,and Scharf (forthcoming).


277

single grant schedule, which determinesthe total grant provided to a region as afunction of the region’s chosen tax rate. Insymbols, the grant schedule may be de-scribed by a function, Si = f(ti). The criticaldifference between this type of policy andthe one given by equation 4 is that bothtypes of regions must face the same func-tion f, in contrast to the different linearfunctions in equation 4. Given this restric-tion, we should not expect the optimalgrant function to induce the two types ofregions to choose the same tax rates, sincethey would then receive the same grants,despite their preference differences.

The problem confronting the centralauthority is to choose T and the grantschedule to maximize the total welfare ofresidents, measured by summing the wel-fare levels of residents across all regions.Using solution techniques from principal-agent theory, Bucovetsky, Marchand, andPestieau (1998) demonstrate that the op-timal grant function induces the high-demand regions to choose a higher taxrate than low-demand regions. Thus, thecapital market is necessarily distorted atthe optimum as a means of inducing thedifferent types of regions to select differ-ent grant levels. Moreover, the regions re-spond to the grant function by choosinginefficient public good levels. For the spe-cial case where public good demands areindependent of income levels (quasi-lin-ear preferences), Bucovetsky, Marchand,and Pestieau show that the high-demandregions underprovide the public good(i.e., MB > MC), but the low-demand re-gions are induced to overprovide the pub-lic good (i.e., MB < MC). This makes sense.In order to lessen the capital market dis-tortions caused by raising the high-demand tax rate above the low-demandtax rate, the central authority accepts poli-cies where the interregional difference be-tween public good levels is less than itwould be under an efficient allocation(where MB = MC in all regions). It is strik-ing, however, that information asymme-

tries can cause the central authority to“overcorrect” the tax competition problemby inducing some regions to go fromundersupplying public goods to oversup-plying them.

To advance our understanding of “op-timal fiscal federalism,” future researchwill increasingly need to take informationproblems into account. However, theinformation-based approach has twolimitations. First, we do not have a goodunderstanding of how information asym-metries occur between different levels ofgovernment, and what exact form theseasymmetries take. Rather, we have vagueideas, such as the understanding that lo-cal officials know more because they are“closer to the people.” Formal models thattry to capture such ideas must usuallyassume such a simple economic environ-ment that it seems difficult to justify whythe central authority cannot easily obtainthe information that is assumed absent.In Bucovetsky, Marchand, and Pestieau(1998), for example, it would seem to be asimple matter to check whether the de-mand for the single public good is higherin one region than in another. Future re-search should try to “endogenize” thislack of information, rather than simplyassume it.

Second, this information approach tofiscal federalism is largely a normativeexercise, in the sense that government of-ficials at all levels of government seek onlyto raise the welfare of those individualsthey represent. This ignores self-interestedbehavior and the constraints imposed byexisting political institutions, both ofwhich would be useful to model.

At the international level, there do notexist strong institutions for coordinatingthe activities of sovereign nations. Theneed to induce cooperation among suchnations severely restricts the set of feasiblepolicies. Fortunately, the tax competitionliterature often identifies forms of inter-vention that leave all regions better off,suggesting some scope for cooperation.


278

Large Regions

Suppose now that regions are largeenough to influence the equilibrium after-tax return on mobile capital. For this large-region case, a Nash equilibrium is theaccepted equilibrium concept. Specifi-cally, the economy is in equilibrium wheneach region’s strategy maximizes its ob-jective function, given the strategies pur-sued by the other regions. The literaturetypically treats tax rates as the strategyvariables, in which case public good lev-els adjust to satisfy each region’s govern-ment budget constraint once all of the taxrates have been chosen. A single regionchooses its tax rate, treating as fixed thetax rates chosen by other regions.14 It thenrecognizes that the equilibrium after-taxreturn on capital depends on its chosentax rate and that of every other region:r = r(t1, . . . tN), where ti is the tax rate cho-sen by region i in a system of N regions.In particular, if region i raises its tax rate aunit, then its cost of capital, r + ti, will riseby 1 + ∆r, which is less than one, becausethe resulting change in r, ∆r, must be nega-tive to clear the capital market. As a re-sult, K(r + ti) is less sensitive to changes inti than in the case of many small regions,where r is fixed from a single region’sviewpoint. An increase in one region’s taxrate continues to create a positive exter-nality through a capital outflow, but nowthis outflow is less severe, due to the par-tial capitalization of higher tax rates intothe after-tax return on capital.

Consider first the simpler case of iden-tical regions. Rule 2 remains valid, recog-nizing that the capital outflow, ∆K, islower than before, and rule 3 is modifiedto account for the capitalization effect:

[5] MB = 1 – τε(1 + ∆r)

.

Thus, market power is actually beneficialfrom the viewpoint of welfare in theentire system of regions, since it lowersthe perceived marginal cost of public goodprovision, thereby stimulating publicgood provision. Hoyt (1991b) carries thisidea further by showing that public goodlevels and tax rates rise as the number ofcompeting regions drops. Of course, thereis no longer any tax competition problemonce the economy contains only one re-gion. In this case, there are no capital out-flows, and a unit rise in ti is fully capital-ized into r, so that there is no change inthe cost of capital, r + ti.

New considerations arise when regionsdiffer in size. Bucovetsky (1991) and Wil-son (1991) analyze “asymmetric tax com-petition” between a “large” region and a“small” region, as distinguished by thenumber of residents, each possessing thesame endowments of capital and labor.Since the large region is the relatively largedemander in the capital market, an in-crease in its tax rate depresses the after-tax return on capital, r, by a relatively largeamount. Thus, the cost of capital, r + t, isless sensitive to tax changes in the largeregion than in the small region. This con-sideration suggests that the large regionwill compete less vigorously for capitalthrough tax rate reductions and thereforeend up with the higher tax rate.Bucovetsky and Wilson demonstrate thatthis is indeed the case.

This finding leads to interesting conclu-sions about the advantage of “smallness”in a tax competition model. Because thesmall region possesses the lower cost ofcapital in equilibrium, firms there employmore capital per unit of labor and there-fore offer higher wage rates than in thelarge region. As a result, the residents ofthe small region can be shown to be bet-

MC

14 Wildasin (1988, 1991) analyzes the alternative formulation in which the public good levels serve as the strat-egy variables, with the tax rates adjusting to satisfy the government budget constraints. For a system ofidentical regions, this change in strategy variables reduces equilibrium public good levels further below theefficient level.


279

ter off than the residents of the large re-gion. In fact, Wilson (1991) demonstratesthat if the difference between regionalsizes is sufficiently large, then the smallregion will be better off than it would bein the absence of tax competition, wherethe capital tax is replaced with head taxeson residents. Thus, we find that althoughtax competition is inefficient, it can actu-ally benefit some regions.

This comparison between small andlarge regions shows that tax competitionis quite different from the analysis of “tar-iff wars.” In the latter case, it is generallybelieved that sufficiently large countrieswin tariff wars in the sense that they arebetter off than they would be under freetrade. The basic idea is that they havemore ability to manipulate the terms oftrade through their use of tariffs and willtherefore employ higher tariffs than smallcountries. Kennan and Riezman (1988)present a formal analysis of a tariff warbetween two countries, modeled as aNash equilibrium in tariff rates, and theyfind that the larger country does “win” ifthe size difference is sufficiently great. Thereason for this difference in results can betraced back to interregional externalities.In the case of tax competition, we haveseen that one region’s tax creates a posi-tive externality through the flow of capi-tal to other regions. Since the smaller re-gion has the lower tax rate, it is thereforethe net beneficiary of these interregionalexternalities. In the case of tariff wars,however, a country’s tariff creates a nega-tive externality by changing the terms oftrade in an unfavorable way from theother country’s viewpoint. This differencein the sign of the externalities is respon-sible for the different welfare results.15

Another interesting implication ofasymmetric tax competition between tworegions is that one region may choose to

overprovide the public good relative tothe rule for efficient provision. Differencesin regional size, production technologies,or consumer preferences can be expectedto cause one region to export capital to theother, and the capital-importing regionhas an incentive to restrict such imports,thereby driving down the required after-tax return on capital. The tax-inducedchange in r, again denoted ∆r < 0, nowenters the rule for equilibrium public goodprovision as follows:

[6] MB = MC{[1 – (k/k)]∆r + 1}

.

where k is the capital owned by theregion’s residents and k is the capital usedby the region’s firms, making k – k im-ports of capital. Comparing this rule withthe rule given by equation 5 for the sym-metric case, we see that the new term,[1 – (k/k)]∆r, acts to reduce the marginalcost of the public good for capital-import-ing regions. In other words, such regionshave an extra incentive to increase the taxrate in order to achieve desirable “terms-of-trade effects.” Of course, such effectscome at the expense of capital-exportingregions, which are directly harmed by thedrop in r. DePater and Myers (1994) referto these terms-of-trade effects as pecuni-ary externalities, in contrast to the fiscalexternality associated with the capitalelasticity in equation 6. It is the existenceof a pecuniary externality that may leadto overprovision of the public good incapital-importing regions. On the otherhand, its presence works in the oppositedirection in capital-exporting regions,thereby aggravating the underprovisionproblem.

This overprovision possibility has im-plications for the optimal form of centralauthority intervention, an issue thatDePater and Myers (1994) also address.

1 – τε(1 + ∆r)

15 In a paper that combines elements of trade and tax competition, Haufler and Wooton (1999) consider competi-tion between two countries for a foreign-owned monopolist and conclude the large country wins the competi-tion, because the monopolist benefits from a larger market, due to the assumed existence of transport costs.


280

Once again, the first-best allocation can beachieved with a subsidy function of theform given by equation 4. But now thecorrective subsidy may be negative incapital-importing regions, to offset thepecuniary externality. Thus, we can nolonger say that all regions engage inwasteful tax competition, in the sense thatthey set their tax rates inefficiently low toattract capital. On the other hand, the mis-allocation of capital resulting from thediffering incentives faced by capital-im-porting and capital-exporting regions is anew and potentially important type ofinefficiency.

Finally, the issue of large versus smallregions raises interesting questions aboutthe incentives that groups of regionsmight possess to form into coalitions (or“federations”) for the purpose of compet-ing with other regions for scarce capital.Burbidge et al. (1997) analyze this issueby combining the basic tax competitionmodel with a model of coalition forma-tion. If the number of regions exceeds two,then the equilibrium coalition structurecan involve more than one independentcoalition. Thus, the tax competition prob-lem does not necessarily disappear whenthere exists endogenous coalition forma-tion.

Trade

We have seen how different regionsmay choose different tax rates, creating amisallocation of capital. A natural ques-tion is whether the pattern of interregionaltrade in private goods is also distorted bytaxation. The basic tax competition modeldoes not address this issue, because pri-vate consumption is aggregated into asingle good. Wilson (1987) considers in-stead a system of many regions with twoprivate goods, one labor intensive and theother capital intensive. The implicationsof this change are surprising: even if thereexist no innate differences between re-gions, such as the usual trade-creating

differences in comparative advantage, dif-ferent regions choose different tax ratesand trade goods with each other. In fact,the low-tax regions produce only the capi-tal-intensive good, and the high-tax re-gions produce only the labor-intensivegood, with the share of regions produc-ing each good determined by the require-ment that demand equal supply in thegoods markets. To see this, observe that ifa region produced both goods, then bothindustries would earn zero economic prof-its, as required for a competitive equilib-rium. But then a tiny reduction in the taxrate would lower the cost of capital andraise the wage rate so that only the capi-tal-intensive industry could break even,thereby driving labor-intensive firms outof the region and leaving the region spe-cialized in the capital-intensive good. Theresult would be a large jump in theregion’s capital stock, which would raisetax revenue and thereby provide residentswith more public good provision. Thus,it can never be optimal for a region to pro-duce both goods.

Stated differently, some regions chooseto compete vigorously for capital, therebyending up with capital-intensive firmsand high wages, but low public good lev-els, whereas others forego vigorous com-petition and settle for labor-intensivefirms and low wages, but high publicgood levels. Any single region is indiffer-ent between the two tax policies, since allregions are innately identical. Since themodel assumes that all individuals areidentical, it is inefficient for the residentsof different regions to consume differentbundles of private and public goods.Thus, the analysis identifies another po-tential type of inefficiency in tax competi-tion models.

MULTIPLE TAX INSTRUMENTS

There would not exist a tax competitionproblem if the regional governmentscould utilize head taxes or other forms of


281

lump-sum taxation, i.e., taxes that do notdistort private sector behavior becausethey are collected in fixed amounts, inde-pendent of consumption and factor sup-ply decisions. For example, a tax on laborincome would be a lump-sum tax in thebasic tax competition model, since thesupply of labor is fixed. As such, it wouldbe definitely preferred to the capital tax.The modeling practice of much of the taxcompetition literature is to exclude suchtaxes in order to analyze taxes that gov-ernments typically levy on mobile factors.The original tax competition models fo-cused on capital taxes, because such taxesare a component of property taxes in theUnited States, given the common practiceof not distinguishing between the valuesof capital and land at a given site.

One reason given for not relying heavilyon lump-sum taxes is that administra-tively feasible forms of lump-sum taxationwould not be equitable or politically fea-sible. For example, Margaret Thatcher’simplementation of a poll tax in Great Brit-ain is widely viewed as having helpeddrive her from office. If it is not possibleor desirable to generate tax revenuethrough the use of lump-sum taxes, weare still left with the question of whetherthe inefficiencies from tax competition inthe basic model remain if capital taxationis supplemented with other forms of taxa-tion that distort consumer or producerdecisions. This section discusses a few at-tempts to address this issue.

The next two subsections discuss somealternative assumptions about availabletaxes, using extensions of the basic taxcompetition model. I then amend themodel in a fundamental way by allowingboth labor and capital to be mobile. In bothcases, I identify assumptions under whichpublic goods are underprovided, but someof these assumptions imply that only la-bor is taxed, not capital. In contrast, theSamuelson rule for efficient public good

provision is found to hold in a case wheregovernments choose to tax capital. Takentogether, these results demonstrate that theunderprovision of public goods is not tiedto the taxation of mobile capital when othertax instruments are available. In a finalsubsection, I discuss the “common-agencyapproach,” which explicitly models theinformation problems that governmentsface when they design their tax policies.

Optimal Commodity Taxation

The analysis of alternative tax instru-ments must confront a fundamental resulton optimal taxation in an open economy:if a government can satisfy its revenuerequirement using a system of optimalcommodity taxes, then it should not usetax instruments that distort the pattern ofgoods trade or factor trade with other re-gions. In particular, it should not use thetype of “source-based” capital tax em-ployed in the basic tax competition model,which is levied on only the capital incomeearned within a region’s borders. Instead,it should use a “residence-based” capitaltax, which is levied on each resident’sworldwide capital income. This resultapplies to regions that are small in thesense that they cannot manipulate theterms of trade, including the required af-ter-tax return on capital, and it also as-sumes the absence of untaxed profits.Gordon (1986) provides a proof within thecontext of a two-period model of a singleregion, in which residents choose howmuch labor to supply to competitive firmsin the first period and how much to savefor consumption in the second period. Theresidence-based tax on capital is basicallya tax on the residents’ income from sav-ings, which is a tax on future consump-tion. By also taxing labor income, thegovernment implements an optimal com-modity tax system, leaving no room for abeneficial source-based tax.16 Although

16 Current consumption serves the role of an untaxed numeraire commodity in this model.


282

Gordon assumes away income distribu-tion problems, the desirability of com-modity taxation extends to regions withheterogeneous populations.

This negative conclusion about the useof source-based capital taxation essen-tially extends a theorem by Diamond andMirrlees (1971) to an open economy con-text. They show that the governmentshould keep the economy on the frontierof its aggregate production possibility setif it can employ an optimal commodity taxsystem (where factors such as labor areamong the commodities). For a small re-gion, “international trade” in goods orfactors may be viewed as another produc-tion sector, where exports are used to“produce” imports via a linear technology.Viewed this way, aggregate productionefficiency requires that government poli-cies not distort trade (Dixit and Norman,1980). In particular, source-based capitaltaxes should not be employed. In the ab-sence of income distribution problems,this particular conclusion is quite intui-tive. A small region faces an infinitely elas-tic supply of capital at the after-tax returnrequired by investors in the worldeconomy. In contrast, the region’s resi-dents have savings and labor supplycurves with finite elasticities. Thus, tax-ing the income from savings and laborminimizes the “deadweight loss,” or “ex-cess burden,” from taxation.

The availability of a residence-basedcapital tax has important implications fortax competition. Bucovetsky and Wilson(1991) examine this issue within the con-text of a system of identical regions. They

employ the two-period setup used byGordon (1986) to examine a single region’stax policy, and they allow the number ofregions to be either small or large. If theseregions have access to both source- andresidence-based capital taxes, then theequilibrium is efficient, given the availabletax instruments. Thus, the tax competitionproblem disappears when a residence-based capital tax is available.17

In practice, it is quite difficult to taxcapital income on a residence basis, inde-pendently of where it is owned. The ad-ministrative and tax compliance problemsinvolved in taxing foreign-source incomeare much more severe than those associ-ated with taxing domestic income.18 As aresult, researchers have investigated mod-els where residence-based taxation is ei-ther limited or not available.19

Taxes on Labor and Capital

The absence of a residence-based capi-tal tax does not justify taxing capital atsource. In fact, Bucovetsky and Wilson(1991) also demonstrate that a small re-gion should meet all of its revenue needsby taxing only labor income, although la-bor-leisure decisions are distorted. Again,the intuition is that capital investment isin infinitely elastic supply for a small openeconomy, whereas the labor supply elas-ticity is finite. If, instead, regions are largeenough to have influence over the equi-librium after-tax return on capital, then aregion’s optimal tax system again includesa source-based tax on capital income.

17 In contrast, Razin and Sadka (1991) and Frenkel, Razin, and Sadka (1991) suggest that tax competition isefficient even in the absence of residence-based taxation. In particular, Frenkel, Bazin, and Sadka conclude,“tax competition leads to a constrained optimum, relative to the set of tax instruments that is available” incases where the countries “cannot effectively tax their residents on their income from capital invested in therest of the world” (p. 206). The apparent conflict is resolved by noting that they consider only two smallcountries that face a fixed world interest rate determined in the rest of the world. There is no such rest of theworld in Bucovetsky and Wilson (1991).

18 See Blumenthal and Slemrod (1995) for some estimate of these costs.19 Gordon and MacKie-Mason (1995) and Gordon and Bovenberg (1996) provide additional explanations for

why a government might desire to use source-based taxes or subsidies.


283

For both cases, Bucovetsky and Wilson(1991) demonstrate that regions under-provide the public good in equilibrium. Butnow the argument centers on a mechanismfor interregional externalities that differsfrom the one in the basic tax competitionmodel. Suppose that a single region raisesits labor tax to finance additional publicgood provision. As a result, the supply oflabor declines, and firms respond by reduc-ing their demands for capital. More capi-tal is then available to other regions, rais-ing the marginal products of labor in theseregions. The resulting increase in the wagerates in other regions encourages workersthere to supply more labor, thereby par-tially offsetting the distortionary effect ofthe labor tax on labor supplies. Thus, a risein one region’s labor tax creates positiveinterregional externalities, but now theseexternalities occur through interactionsbetween the undistorted capital marketand the distorted labor markets. Their pres-ence implies that tax rates and public goodlevels are set too low. In particular, every-one could be made better off if a centralauthority forced governments to tax wageincome at a higher rate to finance greaterpublic good provision.

The availability of multiple tax instru-ments introduces a second source of inef-ficiencies involving government behavior:the choice among tax instruments. For ex-ample, suppose that the variable supplyof savings in the Bucovetsky–Wilsonmodel is replaced by the assumption offixed capital endowments, as in the basicmodel. It remains true that small regionswill continue to tax only labor income, butthe efficient tax policy for the system ofregions now consists of taxing only capi-tal, given its fixed supply. Thus, the de-centralized choice of tax instruments getsthings completely wrong from the view-point of efficiency. We see once again thatextending the basic tax competition modelin various directions can produce newtypes of inefficiencies.

Huber (1999) investigates tax competi-tion in a model that essentially merges thebasic tax competition model with a finite-type version of the Mirrlees (1971) modelof optimal income taxation. Each of manyidentical regions contains two types ofresidents, distinguished by the type of la-bor they provide, “high skilled” and “lowskilled.” Each government uses a nonlin-ear tax on wage income and a source-based capital tax to finance public goodprovision and redistribute income. As inthe Mirrlees model, a government seeksto maximize an objective function thatdepends on the welfare levels of all resi-dents. But unlike the Mirrlees model, thetwo types of labor enter production func-tions as separate complementary factors.As a result, they should be viewed as sepa-rate commodities, and thus an optimalincome tax system does not constitute anoptimal commodity tax system, becausethe latter would require that the two typesof labor income face different tax sched-ules. The absence of optimal commoditytaxation creates a role for a source-basedcapital tax. In particular, it should now beused to distort investment decisions in away that reduces the spread between theskilled and unskilled before-tax wagerates. Doing so partially compensates forthe government’s inability to apply dif-ferent rate schedules to these two typesof labor income. But the equilibrium capi-tal tax may be positive or negative, de-pending on the way in which a rise in thecapital stock affects the relative marginalproducts of the two types of labor.

In either case, Huber’s analysis suggeststhat the equilibrium capital tax will beinefficiently low under reasonable condi-tions. To identify the interregional exter-nality, suppose that the capital tax is in-creased above its equilibrium level in oneregion. To clear the capital market, we canexpect this higher tax rate to be partiallycapitalized into the after-tax return oncapital, i.e., the return falls by some small


284

amount.20 In the empirically reasonablecase where high-skilled workers ownmore capital than low-skilled workers, thefall in the after-tax return will create amore equal distribution of income in allregions. Thus, the capital tax continues tocreate a positive interregional externality,but now this externality consists of ben-eficial equity effects. It follows that theequilibrium tax rate on capital is ineffi-ciently low, as in the basic tax competi-tion model. The analysis nicely demon-strates that whether the equilibrium taxon capital is positive or negative does nottell us whether it is too low from the view-point of the entire system of regions.

To conclude, it appears that competi-tion for capital can lead to inefficiently lowlevels of taxation even if the tax instru-ments available to the government extendsignificantly beyond a source-based capi-tal tax. An important task for future re-search is to explicitly model the reasonsfor why particular sets of tax instrumentsare utilized by governments. The informa-tion and political economy approachesdiscussed below represent some initialefforts along this line.

Labor Mobility

Since individuals are normally free tochoose where to reside within their coun-try of citizenship, the models developedabove are perhaps more suitable for taxcompetition between countries (except forthose groups of countries with free migra-tion, such as the European Union). How-ever, labor mobility can be added to thebasic tax competition model without

changing the results, provided we con-tinue to assume that only capital is taxed.To set the stage for the analysis of mul-tiple tax instruments, I begin by describ-ing two such models.

Brueckner (1999) retains the two-factorsetup of the basic model but allows eachindividual not only to choose where toinvest capital, but also to pick the com-munity in which to work and consume.He assumes a large number of competi-tive “developers,” who choose publicgood levels and tax rates on mobile capi-tal to maximize the “profits” from com-munity development (which equal zero inequilibrium). The public good has theproperties of a private good, meaning thatthe per capita cost of providing a givenamount to each resident is independentof the number of residents, i.e., no scaleeconomies in public good consumption.The mobile individuals differ only in theirpreferences for the public good. In equi-librium, communities offer different taxrates and public good levels, and indi-viduals sort themselves across communi-ties according to their preferences.21 Butthe capital tax continues to create a posi-tive externality, resulting in inefficientlylow tax rates and public good levels. Infact, the rules for equilibrium public goodprovision given by equations 2 and 3 re-main unaltered. As found in the “Correc-tive Policies” section, however, not every-one must lose from tax competition. In thepresent case, individuals with relatively“low” preferences for the public good maybe better off under tax competition thanthey would be in a fully efficient equilib-rium with head taxes used to finance pub-

20 Huber does not show that the after-tax return must fall in all cases. Assuming that it does fall, the assumptionthat the region is small implies that its residents’ capital income falls by a negligible amount. But when wesum the changes in capital income across the residents in the many small regions, we obtain a non-negligiblechange.

21 If the public good were pure (i.e., no congestion), then its per capita costs would fall with the number ofresidents, creating incentives for regions to grow in size. Employing this assumption, Perroni and Scharf(1997) analyze a model in which each region contains many types of residents and majority rule serves as themechanism by which taxes and public good levels are chosen. They argue that increased capital mobility maymake everyone better off in equilibrium, but that tax competition reduces efficiency “in less extreme sce-narios.”


285

lic good provision. Intuitively, these indi-viduals benefit from living where capitaltaxes are relatively low and, therefore,capital-labor ratios are relatively high.

The model developed by Wilson (1995)yields similar conclusions in the absenceof scale economies. There exists a large butfixed number of identical regions, eachpossessing a fixed amount of a third fac-tor, land. Each region competes for bothmobile capital and identical mobile work-ers, with the objective of maximizing thevalue of the region’s land. Capital is nowtaxed through the use of a “property tax,”levied uniformly on the values of bothcapital and land. Assuming this is the onlytax, the rules given by equations 2 and 3remain valid, but with the capital outflowreplaced by a reduction in the per capitavalue of the region’s capital and land, sincethe latter now serves as the per capita taxbase. As before, a rise in the property taxshifts some of this base to other regions,creating the usual positive externalities.

If we were to add a head tax to thismodel while retaining the assumption ofno scale economies, then the responsewould be exactly the same as in the basictax competition model: governmentswould abandon the property tax and useonly the head tax. But now the head taxwould be used to efficiently control mi-gration. In particular, each governmentwould make each resident pay a head taxequal to the marginal cost of supplyingthe public good to another individual. Noother taxes would be needed to balancethe government budget because the ab-sence of scale economies implies that thismarginal cost equals the per capita costof public good provision. With efficienthead taxation, governments would choosethe efficient level of public good provision,which satisfies the Samuelson rule.

Wilson’s (1995) surprising conclusion isthat this rule for efficient provision con-tinues to hold when there exist scaleeconomies, although the head tax nolonger satisfies the marginal-cost-pricingrule. The presence of scale economies in-troduces a need for other taxes, since theper capita cost of public good provisionthen exceeds the marginal cost. Govern-ments respond by employing the propertytax and manipulating the head tax to par-tially compensate for the distortionaryeffects of the property tax. But they do notdeviate from efficient public good provi-sion. We see, then, that the taxation ofmobile capital need not imply that publicgoods are underprovided in equilibrium.

This last conclusion becomes even morepronounced if the head tax is replacedwith a labor tax that distorts labor-leisurechoices. If scale economies are absent, thenthe results are similar to those in theBucovetsky–Wilson (1991) model of a sys-tem of many regions: only labor is taxedbut public goods are underprovided.Adding scale economies creates a role forproperty taxation, but the public goodneed no longer be underprovided. In fact,the Samuelson rule is now satisfied in thespecial case of a pure public good (an ex-treme case of scale economies).22 Onceagain, we cannot identify cases ofunderprovision by examining whetherregions choose to tax capital.

Yet another consideration is the man-ner in which labor is mobile. In the mod-els discussed so far, individuals choose theregion in which to reside and work. Incontrast, Braid (1996) examines a modelwith commuting. The world economyconsists of many metropolitan areas, eachcontaining a fixed number of identical lo-cal jurisdictions with fixed residentialpopulations. As in Wilson (1995), competi-

22 In contrast to the head-tax case, however, this equilibrium is inefficient. Wilson (1995) provides conditionsunder which a rise in every region’s public good supply beyond the equilibrium level reduces welfare if thefinancing comes from the labor tax, but increases welfare if the property tax is used instead (Prop. 4, p. 349).These results suggest that the primary inefficiency in this case is an inefficient tax mix, not an inefficient levelof public good provision.


286

tive firms in each jurisdiction use immo-bile land and mobile capital and labor toproduce output, but now labor is mobilebecause individuals can costlessly com-mute to other jurisdictions. The publicgood in this model is assumed to benefitresidents, not commuters. To finance it,jurisdictions can employ source-basedtaxes on wage and capital income, andalso a property tax, levied uniformly onboth capital and land. Jurisdictions pre-fer the property tax over a tax on mobilecapital alone, since the former includes thenondistortionary land tax. To compete forcommuters, they employ both the prop-erty tax and the wage tax, rather thansolely relying on the wage tax. Thus,Braid’s model also endogenizes thetaxation of property. He ties the level oftaxation to the degree of competition forcommuters, as indicated by the numberof jurisdictions. As this number rises, thewage tax declines, the property tax rises,and the public good provision declines.23

Thus, Braid’s work shows that compe-tition for commuters increases the degreeto which public goods are underprovided,while causing greater reliance on the prop-erty tax. In contrast, recall that Wilson(1995) finds no such positive relation be-tween public good underprovision andreliance on the property tax. One impor-tant difference between the two modelsis that scale economies in public good pro-vision are central to Wilson’s results butirrelevant in Braid’s model, since he as-sumes that only residents consume thepublic good, and their number is fixed ineach jurisdiction. It would be interestingto see whether the two models yield moresimilar conclusions if commuters also ben-efit from a jurisdiction’s public expendi-tures.

The Common-Agency Approach

The research strategy reported in theprevious subsections involves limiting thegovernment’s power to tax in interestingways and then investigating the resultingimplications for tax competition. An alter-native approach is to derive such limitsas the outcome of specific aspects of theeconomic environment. One such aspectis incompleteness in a government’s in-formation about the firms that it is at-tempting to tax. For example, firms maydiffer in the degree to which they areinterregionally mobile, but such differ-ences may be difficult for the governmentto observe. In this case, the governmentcannot tax a firm in a way that directlydepends on its unobserved mobility char-acteristics. Instead, the government mustbase its tax on observable aspects of firmbehavior that may serve as signals of thesecharacteristics, such as the firm’s invest-ment decisions. This is a principal-agentproblem, with the government serving asthe principal and the firm as the agent. SeeOsmundsen, Hagen, and Schjelderup(1998) for an example of such a model.

When two or more governments com-pete for a share of a mobile firm’s profits,this problem becomes a common-agencyproblem, with the governments now serv-ing as multiple principals. A Nash equi-librium is established when each govern-ment chooses its optimal policy, given thepolicy chosen by the other government.The firm’s private information about itsattributes allows it to earn an “informa-tion rent,” adding a new dimension to thewelfare analysis of tax competition. InMezzetti (1997), for example, tax compe-tition (i.e., his “independent contracting”)raises the information rent earned by the

23 This decline in public good provision also occurs when only the property tax is available. In another paper,Braid (1998) obtains a similar relation between commuting and public good provision, using a spatial modelto capture the costs of commuting. In particular, public good provision falls as commuting costs fall. Braid(1997) adds absentee ownership of land to the model developed by Braid (1996) and finds that overprovisionmay result, but this is due to “tax exporting,” which I have discussed in the “Capital Tax Competition”section.


287

firm relative to its level when the two prin-cipals cooperate, and the result is loweraggregate welfare for the principals. Thus,tax competition again worsens welfare,but for new reasons.

The common-agency approach seemsespecially useful for analyzing the taxa-tion of multinationals. Many countriesattempt to tax the income of foreign sub-sidiaries, suggesting a common agencyproblem in which the home and hostcountries are the principals. Bond andGresik (1996) present an interesting modelin which the home and host countries in-dependently confront the multinationalwith subsidies and trade taxes. Relativeto setting these policy instruments coop-eratively, tax competition is shown tolower the countries’ aggregate welfareand leave the multinational worse off.24

In other words, the inefficiencies associ-ated with tax competition turn out to bedetrimental to all parties.

While there is thus some evidence thatthe detrimental effects of tax competitionin the basic model carry over to the morecomplex common agency problem, muchremains to be done in this exciting area ofresearch. As increasingly complex govern-ment policies are considered, however, weare led to increasingly question the basicassumption that government officials seekto maximize welfare, rather than engagein self-interested behavior. A model thatcenters around such behavior is discussedin the “Political Economy” section.

COMMODITY TAX COMPETITION

In addition to the literature on compe-tition for scarce capital, a literature on“commodity tax competition” has alsobeen developed. Mintz and Tulkens (1986)introduce a model where this form of com-petition occurs between two regions thatare linked by cross-border shopping. Each

region contains a fixed number of identi-cal residents, whose utility depends posi-tively on the consumption of a privategood and a public good, and negativelyon the supply of labor. Public good expen-ditures are financed by a tax on privategood consumption, levied on an originbasis, and Mintz and Tulkens examine theNash equilibrium in these tax rates. Theuse of origin-based taxes means that eachregion’s government collects a uniformtax on only the output of domestic firms,regardless of where this output is ulti-mately consumed. As a result, the region’sresidents can escape the tax by incurringthe transport costs necessary to cross theborder and purchase the private good inthe other region. In contrast, a destination-based commodity tax would enable a re-gion to collect a tax on all of its residents’private good consumption. This could bedone through the use of border adjust-ments, under which the tax is collectedfrom domestic firms, but a tax rebate isgiven for exports of these goods and a taxis collected on imports. (Imports and ex-ports take place through cross-bordershopping in the Mintz–Tulkens setup.) Inthis case, a region’s residents would notescape their government’s tax by crossingthe border. But border adjustments areadministratively difficult to enforce, andin some areas, they have been effectivelyeliminated, most notably in the EuropeanCommunity. See Lockwood (1993) for adetailed comparison of commodity taxcompetition under the destination andorigin principles.

In contrast to the models of competitionfor capital, Mintz and Tulkens (1986) andde Crombrugghe and Tulkens (1990)describe cases where the equilibrium isfully efficient. However, these cases occurwhen transport costs are so high that nocross-border shopping occurs, either inequilibrium or in response to small tax

24 See also Bond and Gresik (1998), who examine the case where the two governments are asymmetrically in-formed about the firm’s production costs.


288

changes.25 In such cases, there are noneof the interregional externalities describedpreviously. It seems difficult, however, todescribe these cases as “tax competition,”because the governments are not reallycompeting over a tax base.

Instead, tax competition in these mod-els is associated with tax rate differencesthat are sufficiently large to overcometransport costs, so that the low-tax region“exports” the good to cross-border shop-pers from the high-tax region. In this case,a rise in the high-tax region’s tax rateraises the amount of shopping done by itsresidents in the low-tax region, therebyincreasing the latter ’s tax base. This taxbase change is called the “public con-sumption effect.” As in the basic tax com-petition model, it represents a positiveexternality, implying that the high-taxregion’s tax rate is inefficiently low. Inparticular, Mintz and Tulkens (1986) showthat if a central authority were to force thetwo regions to change their tax rates indirections that made both of them betteroff, then any such tax changes would in-volve a rise in the high-tax region’s taxrate.

But would this efficiency-improving taxchange also involve a rise in the low-taxregion’s tax rate? The recent work byHaufler (1998) concludes “not necessar-ily.” When the low-tax region increases itstax rate, it not only creates the public con-sumption effect mentioned above, but alsoa “private consumption effect,” consistingof the welfare loss that cross-border shop-pers experience from the increased priceof the private consumption good. As aresult of these conflicting effects, it is pos-

sible for the rise in the low-tax region’stax rate to harm residents of the high-taxregion, i.e., create a negative externality.Hence, it is not possible to show that bothregions set their tax rates “too low” inequilibrium.

To conclude, the pattern of interregionalexternalities is more complex than in thebasic tax competition model, producingless clear-cut results. It is interesting tonote, however, that the literatures on capi-tal tax competition and commodity taxcompetition do seem to obtain similar re-sults concerning the advantage of small-ness. See, in particular, Kanbur and Keen’s(1993) analysis of a spatial model of cross-border shopping.26

OTHER FORMS OF COMPETITION

There are many ways in which govern-ments can compete for mobile factors orshoppers other than through the use oftaxes or subsidies. One possibility is tocompete through the use of public inputsthat improve the productivity of capital.Keen and Marchand (1997) argue, for ex-ample, that the equilibrium pattern of ex-penditures is inefficiently weighted to-ward too much public input provision andtoo little public good provision, since thelatter benefits residents but does not at-tract capital.27 There is also a literaturesuggesting that regions compete too ag-gressively for mobile firms through theuse of inefficiently lax environmental poli-cies, creating a “race to the bottom.” Wil-son (1996a) reviews this literature andconcludes that the possibility of a “race”depends critically on assumptions about

25 See Mintz and Tulkens (Prop. 9b) and also de Crombrugghe and Tulkens (Prop. 2).26 A limitation of this model is that governments care only about maximizing tax revenue. Trandel (1994) and

Haufler (1996) analyze spatial models with less extreme objective functions, but they do not directly addressthe welfare comparisons in Kanbur and Keen (1993). See also Braid (1993). Unlike the other spatial models, heallows lump-sum taxation in addition to commodity taxes.

27 Noiset (1995) and Bayindir-Upmann (1998) obtain less clear-cut results, and Sinn (1997) claims that “publicinfrastructure,” modeled as reducing the “cost” of investment in a region, is efficiently provided. But neitherpaper also models the endogenous provision of public goods. If they did, then public expenditures would beinefficiently weighted away from public good provision and toward public inputs or infrastructure


289

the available tax and subsidy instru-ments.28 As mentioned in the Introduc-tion, policymakers have also express-ed concern about the possibility of a“race to the bottom” in welfare benefits,brought on by the mobility of welfarerecipients.

Other government regulatory programscan be subject to wasteful competitionbetween governments. See, for example,Sinn’s (1997) demonstration that indepen-dent governments choose inefficientlylow product quality standards, given thattheir products are interregionally traded.In addition, he discusses the failure of in-dependent governments to provide ad-equate amounts of social insurance in thepresence of factor mobility.29

The degree to which existing tax laws areenforced may also serve as a strategy vari-able. Cremer and Gahvari (1996) considertwo countries that compete with each otherfor cross-border shoppers, using both taxrates and tax audit probabilities as strate-gic variables. In this case, any coordinationbetween the two countries in the setting oftax rates will be offset to some extent bycompetition for shoppers through reduc-tions in the audit probabilities.

In some cases, governments may designnontax policies that will commit them tochange the degree to which they competein taxes with other governments. See,for example, Jensen and Toma’s (1991)analysis of the use of national debt as astrategic variable in a tax competitionmodel. By committing to debt now, agovernment can signal to other govern-ments that it will have to impose highertaxes on capital in the future. If theseother governments respond by raisingtheir own capital taxes in the future, thenthe first government will have benefitedfrom this reduction in competition forcapital.

These various examples illustrate themultifaceted nature of competition amonggovernments, which is likely to create dif-ficulties for the design of cooperativeagreements to reduce the wasteful aspectsof this competition. It may be possible tocoordinate one policy variable, but doingso may simply cause governments to com-pete more vigorously by means of anotherpolicy variable.

THE POSSIBILITY OF EXCESSIVETAXATION

The general thrust of much of the taxcompetition literature is that tax compe-tition leads to inefficiently low taxes. Incontrast, two types of tax competitionhave been found to produce inefficientlyhigh taxes: vertical tax competition andtax competition with double taxation con-ventions. By examining recent work, how-ever, I find that this conclusion may be toohasty.

Vertical Tax Competition

The types of tax competition discussedto this point can be characterized as “hori-zontal tax competition,” in the sense thatthe governments doing the competing areall at the same level. There also exists aliterature on “vertical tax competition”between different levels of government,such as the federal government and stategovernments within the United States.The basic problem is that each level ofgovernment imposes a tax on the same taxbase. Whereas one state’s tax increases thetax base available to another state underhorizontal competition, now the tax im-posed by one level of government dimin-ishes the size of the tax base available tothe other level of government. In the caseof capital taxation, for example, a rise in

28 For formal models of a race, see Kim and Wilson (1997), Markusen, Morey, and Olewiler (1996), Oates andSchwab (1988), and Rauscher (1995).

29 On the other hand, the existence of factor mobility may alter voters’ preferences and make them elect govern-ments with more liberal views about redistributive policies. See Persson and Tabellini (1992).


290

the federal government’s tax rate reducesnational savings, thereby lowering theamount of capital available to each stategovernment. A rise in a single state’s taxrate has a similar, but smaller, effect, re-ducing the tax base available to the fed-eral government.

Since tax increases now create negativeexternalities, rather than positive exter-nalities, we might expect conclusions fromthe basic tax competition model to be re-versed: taxes are now set too high. Thisconclusion is too hasty, however, becauseother aspects of the political-economicenvironment must be specified. First,there is the issue of what objective func-tion is employed by the federal govern-ment. A benevolent federal governmentdesires to maximize some measure of theaggregate welfare of all residents, whereasa benevolent state government is con-cerned with maximizing the welfare of thestate’s own residents. Unlike horizontaltax competition, then, the objectives of thefederal government and state govern-ments overlap to some extent, reducingconflict. If the federal government pursuesobjectives that are independent of residentwelfare, then we might want to look notonly at ways to facilitate more efficientinteractions between the federal govern-ment and state governments, but also atways to improve the internal functioningof the federal government.

The other issue to consider is thetiming of the actions undertaken by thefederal and state governments. The “Cor-rective Policies” section discusses optimalfiscal federalism under the standard as-sumption that the federal government“moves” first, committing itself to a set ofpolicies that state governments then treatas fixed when choosing their own policies.In contrast, the models of horizontal taxcompetition assume that all governmentschoose their policies simultaneously. Insome cases, institutional features argue forthe higher-level government having thefirst-mover advantage; see Hoyt and

Jensen (1996). But this is not the onlyreasonable assumption. The sequentialmove models described in the “CorrectivePolicies” section simplify matters by as-suming that each government chooses itspolicies only once. In a more complicatedsetting, state governments might both re-act to the past decisions of the federal gov-ernment and also make new decisions thatinfluence the federal government’s futurebehavior. Models with simultaneousmoves sweep all such complications awayby putting the state and federal govern-ments on an equal footing in terms of thetiming of their moves.

Clearly, the best case for efficiency willoccur when the federal government isbenevolent and is able to move first, sothat it can influence the behavior of thestate governments. Boadway, Marchand,and Vigneault (1998) consider this case ina model where the federal governmentand identical state governments utilize anincome tax that redistributes incomeamong a diverse set of residents. They dis-tinguish between cases where migrationis possible or impossible, with the formercase allowing horizontal tax competitionto exist. For both cases, the equilibrium isefficient, given the available policy instru-ments. In other words, the federal govern-ment could do no better if it directly con-trolled the states’ policy instruments. Thisoutcome can be readily understood. Anindividual state engages in vertical taxcompetition by not accounting for thenegative effects of its tax on the federalbudget. But the federal government hassufficient tax tools and foresight to undoany inefficiencies in state government be-havior. In other words, vertical tax com-petition seems to be happening at the statelevel but not the federal level, because thefederal government “sees through” thestate budget constraints when it makes itsown policy choices. The result is an effi-cient equilibrium.

If the federal and state governmentsset their policies simultaneously, then the


291

federal government can no longer alterstate policy choices by changing its ownpolicies. An efficient equilibrium isunachievable in most cases, but a benevo-lent federal government can still manipu-late its policy instruments to at least par-tially offset inefficiencies at the state level.See Hoyt (1996) for a recent analysis of thiscase.30

Keen and Kotsogiannis (1996) providea good example of the inefficiencies thatcan result when governments are nolonger benevolent.31 In their model, thefederal and state governments care onlyabout maximizing tax revenue. Althoughthere exists both horizontal tax competi-tion between states and vertical tax com-petition between the federal governmentand states, the vertical tax competitionproblem dominates in the case where allgovernments move simultaneously. Theequilibrium tax rates are found to be toohigh, even relative to those that maximizetax revenue.

To conclude, vertical tax competitionseems to be a more slippery concept thanhorizontal tax competition. It seems mostlikely to create inefficiencies in modelswhere the federal government is unableto optimally influence the choice of policyinstruments by local governments, due tocommitment problems, information prob-lems, or objectives other than welfaremaximization. Future work should seekto develop more realistic political pro-cesses within and between the differentlevels of government.

Double Taxation Conventions

An important issue in the taxation ofmultinationals is the use of double taxa-tion conventions. Given that a home coun-try attempts to tax foreign-source income,

we have a situation where a firm’s incomeis being taxed by two different countries,home and host. As a result, double taxa-tion is viewed as a potential problem, andthree methods have been used to alleviatethe problem. First, the home governmentcan provide a tax credit for taxes paid tothe host government. Second, it can allowforeign investors to deduct these taxesfrom their taxable income. Finally, it canexempt foreign-source income from taxa-tion. Of these three methods, the deduc-tion method is the least commonly used.

The formal analysis of how indepen-dent governments choose their tax poli-cies under these methods begins withBond and Samuelson (1989). They use atwo-country model in which a capital-ex-porting (home) country and capital-im-porting (host) country use their tax rateson foreign-source income as strategy vari-ables. The surprising result is that if thehome country provides tax credits, thenthe Nash equilibrium involves taxes sohigh that all international capital flowscease. In striking contrast to the basic taxcompetition model, the problem here isnot that taxes are too low, but rather thatthey are too high. The basic intuition isthat the host country always has an incen-tive to raise its tax rate at least to the levellevied by the home country, since thelatter ’s government treasury effectivelypays the tax by providing tax credits toforeign investors. But as long as capitalexports occur, the home country shouldkeep its tax rate on foreign-source incomeabove the host country’s rate, since doingso allows it to exercise its market poweron the world capital market, i.e., to driveup the equilibrium after-tax return oncapital by reducing incentives to investabroad. As a result, capital tax rates are sohigh in equilibrium that all capital exports

30 Hoyt (1996) also shows that a uniform matching grant can be used to ensure that the combined policies of thetwo levels of government are efficient, given the restrictions on the tax instruments. This result is limited to thecase where the “localities” controlled by lower-level governments are identical. For other recent work on theuse of intergovernmental grants to offset fiscal externalities, see Boadway and Keen (1996) and Dahlby (1996).

31 See also Keen’s (1998) review of the literature on “vertical tax externalities.”


292

cease. In contrast, capital flows do occurunder tax deductions. Thus, we are leftwith the empirical puzzle of why creditsare so prevalent in the world economy.

Subsequent research has attempted toreconcile theory with practice. Davies andGresik (1998) allow foreign subsidiaries tofinance their investments by borrowingadditional capital from the host country’sdomestically owned firms, subject to col-lateral requirements. This additional sourceof investment funds dramatically changesthe effects of tax credits. In particular, thecountries no longer set their tax rates sohigh that all capital flows are eliminated.Thus, the use of credits is no longer as badas suggested by Bond and Samuelson(1989). This outcome does not fully explainthe use of credits, however, since Daviesand Gresik still find that the home countryweakly prefers deductions to credits.

Whereas Bond and Samuelson (1989)allow the home country to tax at differentrates the income its residents earn at homeand abroad, Janeba (1995) considers thecase where no such discrimination is per-mitted. A surprising implication of this as-sumption is that the home country sets itstax rate equal to zero under both the taxcredit and exemption methods. The basicidea is that a positive tax rate would in-crease capital exports, creating undesir-able terms-of-trade effects. But Janeba alsofinds that each country’s equilibrium levelof national income is independent ofwhich method is used (credits, deduc-tions, or exemptions). Thus, his modelprovides a solution to the empirical puzzleraised by Bond and Samuelson. Under allmethods, however, the equilibrium taxrates continue to be inefficiently set. Theproblem is not that tax rates are too low,as in the basic tax competition model, butthat the relative rates induce the homecountry to export too little capital.

In a recent paper, Davies (1998) consid-ers two-way capital flows, where eachcountry exports capital to the other coun-try. Again, nondiscriminatory tax policies

are assumed, and there always exist posi-tive capital flows in equilibrium, regard-less of which double taxation conventionis employed. However, the choice amongthem is no longer a matter of indifference.Davies endogenizes this choice by allow-ing each country to independently choosewhich method to employ, prior to thechoice of tax rates. His results providesome indication that deductions will bethe preferred method. In particular, theyare always used in the special case wherethe two countries have identical charac-teristics. While this conclusion once againseems at odds with the popularity of taxcredits, further results demonstrate thedesirability of cooperative agreementsthat eliminate the use of deductions. Per-haps the most striking conclusion is thatin the case of identical countries, eliminat-ing the use of deductions results in a fullyefficient equilibrium. Although inefficien-cies reappear in the asymmetric case, thislimited efficiency result offers an intrigu-ing contrast to the inefficiencies found inthe basic tax competition model.

To conclude, while Bond andSamuelson’s (1989) conclusion that taxcredits eliminate all capital flows remainsa striking result, subsequent research hasgenerated less extreme results, ranging allthe way to the efficiency results obtainedby Davies (1998). In all cases reviewed here,two-country models are employed, imply-ing that terms-of-trade effects are centralto the analysis. As seen in the “Large Re-gions” section, extending the basic taxcompetition model to the case of asymmet-ric tax competition between large countriesis itself enough to significantly change thewelfare implications. Thus, there is per-haps not as much inconsistency betweenthe two types of models as might appear.

EFFICIENCY-ENHANCINGCOMPETITION

Whereas the Tiebout model produces aform of “efficient tax competition,” we


293

have seen that departures from the ideal-ized setting of this model can producevarious forms of wasteful tax competition.Recently, researchers have turned theirattention to the possibility that such de-partures may also create an efficiency-en-hancing role for competition involving taxor nontax policy instruments. This sectiondescribes some work in this direction.

Bidding for Firms

The discussion so far has assumed thatcapital investments may be made in smallincrements. In many cases, however, in-vestments come only in large increments,i.e., they are lumpy. Regions then findthemselves competing for these lumps,often taking the form of large firms. Com-petition for automobile plants is one ex-ample. It is often assumed that regions areable to target subsidies and other taxbreaks to these mobile firms, rather thanaltering the entire tax system. This com-petition may be referred to as “bidding forfirms.” In this case, the effect of tax com-petition on the public goods provided toresidents is no longer a central issue. Butwe can still ask whether the subsidies ortax breaks provided to firms are efficientin the sense that they lead to efficient firmlocation decisions while not creating anyunnecessary costs for the system of re-gions as a whole. Additional efficiencyissues arise in cases where firms receivethe benefits of public services and infra-structure.

Two well-known papers produce mod-els in which the bidding for a large firmenhances efficiency. In Black and Hoyt(1989), two regions compete for a largefirm because its presence attracts moreresidents, which lowers the average costof providing a public good to existing resi-dents. The Tiebout assumption of efficienthead taxes is absent from this model. Inparticular, each resident pays a tax equal

to the per capita cost of public good pro-vision, which is below the marginal cost.Nevertheless, bidding for firms never re-duces the social efficiency of firm location,and in some cases, this bidding causesfirms to locate more efficiently. Black andHoyt conjecture, however, that the use ofpublic services to compete for firms willnot produce an efficient outcome. Theyalso demonstrate that firm location deci-sions are inefficient in cases where the firmpossesses private information about howits production costs differ between the tworegions, which it is unable to reveal to thetwo regional governments.32

The dynamic model developed byKing, McAfee, and Welling (1993) also de-parts from the Tiebout world by introduc-ing uncertainties about firm productivity.In particular, the social value of a firm isgiven by the “surplus” that it generatesby producing in a region, but this surplusis uncertain to both the firm and regionalgovernments prior to actual production.Two regions compete for the firm over twoperiods. After choosing a location in thefirst period, the firm is free to relocate (ata cost) in the second period. The firm’slocation in each period is determined byan auction mechanism, and this locationis found to be efficient. The second partof the paper allows each region to investin “infrastructure.” Before the auctiontakes place, the two regions play a Nashgame in investment levels, under whicheach region sets its investment level opti-mally, given the level chosen by the otherregion. The authors demonstrate that onlyan asymmetric Nash equilibrium exists,where the equilibrium investment levelsdiffer. In the first period, the firm locateswhere investment is highest. However,the losing region may choose a positive(but lower) investment level, because thisraises the probability that the firm willswitch locations in the second period. Thispossibility of relocation implies that the

32 Martin (1997) focuses on the problems associated with bidding for firms that possess private information.


294

losing region’s investment is not sociallywasteful. In fact, the authors show that theequilibrium is efficient.

Both of these models follow the previ-ous literature by assuming that each gov-ernment is concerned with the welfare ofits own citizens. In contrast, Biglaiser andMezzetti (1997) investigate a model inwhich attracting mobile firms provides astate governor with the opportunity toengage in activities that imperfectly sig-nal his “ability” to voters. When two ormore governors with re-election concernscompete for firms, the resulting locationof the firm will not necessarily be efficient.However, such inefficiencies are not aproblem associated with large projects orfirms per se, but rather with the imper-fect political institutions.

It is also important to recognize that in-terregional externalities may play an im-portant efficiency-reducing role, even incases where subsidies can be targeted toindividual firms. But now such externali-ties can easily work in favor of too littlecompetition for firms. For example, if onestate in the United States attracts a foreignfirm, then all states may benefit in theform of lower prices, due to reduced trans-port costs. This example has similaritiesto the pecuniary externalities discussed inthe context of large regions in the “LargeRegions” section: by “importing” the firm,a region creates desirable price effects forother regions.33 But such price effects canwork the other way. If all states in theUnited States are competing for a foreignfirm that faces limited opportunities forlocating its plant outside the country,then they will possess market power thatcan be exercised by competing less vigor-ously. Competition for foreign firms

through the provision of subsidies maybe better than no subsidies, but the equi-librium levels of these subsidies are notlikely to be optimal from the nation’sviewpoint.

Imperfect Competition

If firms are large, then the issue of im-perfect competition becomes potentiallyimportant. The international trade litera-ture on “strategic trade policy” has alreadyexhaustively explored the effects of imper-fect competition on a country’s optimaltrade policies. In particular, this literaturehas justified the use of output subsidies toencourage the country’s firms to competemore aggressively on world markets,thereby increasing profits at the expenseof foreign firms. Janeba (1998a) combinesthese strategic trade motives with a modelof tax competition. He first follows thestrategic trade literature by specifying amodel with two countries, each contain-ing a single firm that sells output in a thirdmarket. In this case, the two governmentscompete by offering subsidies to theirfirms. But then Janeba allows each firm tobe mobile between the two countries,meaning that it locates where its after-taxprofits are highest. The governments nowrecognize that their subsidies will affectnot only firm output decisions, but alsolocation decisions. In particular, each gov-ernment may seek to attract the othercountry’s firm and thereby capture someof its profits through taxation. Janeba as-sumes that the tax system must be non-discriminatory, meaning that a countryimposes the same tax rate on the outputsof all firms that operate within its borders,whether domestic or foreign.34

33 Rauscher (1995) identifies this type of externality as a possible reason for why regions might not seek toattract a polluting firm, even though it is desirable to do so (a phenomenon called “not in my backyard”).

34 Although countries do typically discriminate to some extent, they are restricted in this practice by the nondis-crimination rules established by international agreements such as GATT or the laws of the European Union.Janeba and Peters (1999) focus on the different implications of nondiscriminatory taxes and taxes that dis-criminate between domestic and foreign firms operating within a country, but they abstract from marketstructure issues by assuming that each country has access to an “immobile tax base” and competes for a“mobile tax base.”


295

Janeba’s (1998a) surprising conclusionis that competition for mobile firms causesthe countries to compete their tax ratesdown to zero. No country offers a tax ratebelow zero, because it would then attractboth of the firms but be hurt by the trans-fer of subsidy revenue to the foreign firm.Janeba is also able to generalize the zero-tax result to include cases where the firms’outputs are sold to the consumers in oneof the two countries, rather than in a thirdcountry. In these cases, the country con-taining these consumers cares about con-sumers’ surplus, along with tax revenueand its firm’s profits. In neither case is theequilibrium fully efficient, since the inef-ficiencies associated with imperfect com-petition are still present. But tax competi-tion does improve welfare.

Thus, imperfect competition seems todramatically alter the welfare implicationsof tax competition.35 Recall, however, thatregions would choose zero taxes on mobilecapital in the basic tax competition modelif they had access to head taxes. Capitaltaxes are avoided because each region facesan infinitely elastic supply of capital. Simi-larly, each country’s production capacity isinfinitely elastic in Janeba’s model; it willlose all of its capacity if its tax rate is in-creased slightly above the other country’srate. But the two models differ in how gov-ernments behave when the borders areclosed. The equilibrium is fully efficient inthe basic model, whereas wasteful subsi-dies emerge when there is imperfect com-petition. Thus, tax competition is able toplay an efficiency-enhancing role underimperfect competition, but not in the com-petitive environment of the basic model.

This comparison suggests the basic taxcompetition model could be accused of“stacking the deck” against the possible

welfare-improving effects of tax competi-tion by failing to recognize possiblesources of inefficiency that might exist inthe absence of capital mobility. In the “Po-litical Economy” section, inefficienciesfrom the functioning of the political pro-cess are considered.

Commitment Problems

The basic tax competition model as-sumes that governments commit to a taxsystem, and then capital owners choosewhere to invest their capital. In practice,commitment issues arise because firms orcapital become partially immobile oncelocation decisions are made. One way todeal with this problem is to commit to ini-tial subsidies or “tax holidays” for newfirms, thereby shortening the period oftime in which commitment is required. Butsuch policies have the disadvantage of in-creasing firm turnover. In other words,some fraction of firms may choose to leavea region after the initial tax break has ex-pired, perhaps seeking tax breaks in otherregions. Wilson (1996b) models the use ofinitial subsidies to attract new firms, find-ing that excessive firm turnover is indeedthe equilibrium outcome. Bond (1981)finds empirical evidence of the problem.

Janeba (1998b) demonstrates that taxcompetition may actually help solve com-mitment problems. He investigates a one-firm, two-region model in which decisionsare made in three stages. First, the firmundertakes a single project consisting ofinvestment in “capacity” in each country.Oil pipelines are a possible example. Sec-ond, each government chooses the rate atwhich to tax the project output within itsborders. Finally, the firm chooses its out-puts.36 A commitment problem arises here

35 Note also the importance of the nondiscrimination assumption. If discrimination in taxes were possible, thenthe two governments would continue to provide wasteful subsidies to their own firms.

36 In contrast, Kehoe (1989) considers a two-country model in which savings, but not the location of investment,is chosen prior to tax rate decisions. If the two countries collude in setting their tax policies, then they taxaway savings, since it is fixed at the time taxes are chosen. As a result, nobody chooses to save. Tax compe-tition is preferred, because governments forgo capital taxation in an effort to attract investment.


296

because the governments are able tochoose tax rates after the firm has fixedits capacity levels. If there were a singlegovernment, then it would have an incen-tive to tax all profits away. The firm wouldrecognize this incentive at the time of itsinitial investment decision and choose notto invest in capacity. When there are twogovernments, they will “compete” in taxrates if the firm has excess capacity and istherefore able to reallocate output be-tween the two regions in response to dif-ferences in the tax rates.37 Provided invest-ment costs are sufficiently low, the firmthen chooses to undertake the project byinvesting in excess capacity as a means ofcreating tax competition. To conclude,commitment problems provide anotherpossible role for tax competition as an ef-ficiency-enhancing activity.

Political Economy

A common assumption in the literaturereviewed so far is that each regional gov-ernment seeks to act in the best interestsof some set of residents or factor owners.Indeed, the basic tax competition modeland many of its extensions assume awaydifferences between residents, so thatthere is no conflict about which residents’preferences are given the most weight. Inthe basic model, the only potential sourceof inefficiency is tax competition, makingit relatively easy for this competition toturn out to be a bad thing. A rather differ-ent perspective is taken by the publicchoice literature. Brennan and Buchanan(1980) argue that tax competition im-proves welfare, because the size of gov-ernment would be excessive in the ab-sence of this competition. Rauscher (1996,1998) and Edwards and Keen (1996) ex-

amine this view formally in various “Le-viathan models,” where governments areconcerned in part with maximizing thesize of the public sector. Their conclusionsabout the welfare implications of tax com-petition are mixed, but all three papersassume that governments retain somedegree of “benevolence,” perhaps causedby re-election concerns that are not for-mally modeled.

It is difficult to ascertain empiricallywhether the welfare-improving or wel-fare-worsening view of tax competition ismore accurate, since both views seem topredict that an increase in the number ofcompeting governments should reducethe total size of government. Moreover, theempirical tests of the Leviathan modelconducted by Oates (1985, 1989) and oth-ers have encountered difficulties in evenconfirming that there is a relation betweenaggregate government size and the decen-tralization of fiscal decisions among inde-pendent governments, let alone identify-ing the welfare implications of such a re-lation.38 It therefore appears useful to ex-plore alternative models of governmentdecision making that do not necessarilycreate this relation.

Wilson and Gordon (1998) depart fromthe assumption in Leviathan models thata single monolithic entity exercises con-trol over the entire range of tax and pub-lic expenditure instruments. Their modelrecognizes that some policy instrumentsmight be more accurately modeled as ef-fectively under the control of electoratesor their representatives, whereas othersare largely delegated to self-interestedgovernment officials, leaving the elector-ate with only rudimentary methods ofcontrol. In particular, government officialsare assumed to choose public expenditure

37 The concept of excess capacity is well-defined here, because the model assumes that the demand for thefirm’s output is completely inelastic at prices below the consumers’ reservation level.

38 After reviewing his and other work, Oates (1989) concludes, “The empirical literature on fiscal centralizationand government size thus contains a number of puzzles and inconsistent findings” (p. 582). More recently,Anderson and Van den Berg (1998) find no evidence of a relation between fiscal decentralization and govern-ment size.


297

policies but not tax rates, which are cho-sen to maximize welfare. In another pa-per, Gordon and Wilson (1998) argue atlength that this asymmetric treatment oftax and expenditure policies appears to bea good description of the situation of mostgovernment bureaucrats. For example,while legislatures have substantial controlover the choice of tax rates, they cannotdeal adequately with the innumerablespecific expenditure and regulatory deci-sions that affect the tax base, and theymust therefore delegate these decisions toothers. Moreover, the electorate can moni-tor easily what happens to tax rates buthas a harder time monitoring the qualityof the many different expenditure deci-sions.

Following Niskanen (1971) and the sub-sequent Leviathan models, Wilson andGordon (1998) assume that governmentofficials benefit personally from the bud-get they control and, as a result, face in-centives to pursue activities that increasethe size of the budget. A positive connec-tion between the tax base and public goodlevels is modeled by assuming a systemof many identical regions, each consistingof a fixed amount of land, and allowinglabor to be perfectly mobile across theseregions. Each region employs a linear in-come tax, consisting of a head tax or sub-sidy and a uniform tax rate on all laborand land income earned within theregion’s borders. Given this tax system,the government officials in each regionengage in “expenditure competition” byincreasing their provision of public goodsto attract more labor, thereby expandingthe tax base.

To isolate the efficiency-enhancing ef-fects of expenditure competition, Wilsonand Gordon (1998) compare the equilib-rium in this “open economy” with theequilibrium in a “closed economy,” inwhich the lack of interregional factor mo-bility means that all factor supplies arefixed. In particular, each resident pos-sesses fixed endowments of land and la-

bor, and thus there are no opportunitiesfor residents to confront public officialswith a tax base that can be expandedthrough additional public good provision.Rather, other incentive devices must berelied upon. The model assumes that of-ficials can be replaced with a probabilitythat is related to their job performance,and that the effective penalty from beingreplaced can be directly controlledthrough the choice of the officials’ sala-ries.

By creating expenditure competition,opening the economy increases the “effi-ciency” with which government officialsutilize tax revenues; there is less “waste”in government. Consequently, expendi-ture competition improves welfare in allregions. But regions also engage in a formof tax competition similar to what occursin the basic tax competition, causing taxrates to be too low. In the present case,each region fails to design its tax systemto fully exploit the potential incentive ef-fects created by labor mobility. Higher taxrates would strengthen the connectionbetween the level of tax revenue and pub-lic good provision, thereby providinggreater incentives for public officials toincrease public good supplies. But the re-gions compete for mobile labor throughreductions in tax rates. If they were to allraise their tax rates simultaneously, thenthe incentive would be strengthened with-out any outflows of labor.

Despite the efficiency losses from taxcompetition, welfare in the open economyexceeds welfare in the closed economybecause of the existence of expenditurecompetition. The potential tax incentivesassociated with mobile labor may not befully exploited, but at least such incentivesare present, in contrast to their completeabsence in the closed economy.

By reducing waste in government,opening the economy may reduce theeffective “price” of the public goodenough to induce residents to provide thepublic sector with more tax revenue than


298

is available in the closed economy. Thus,opening the economy may raise or lowerthe total “size” of government, althoughwelfare unambiguously rises. This conclu-sion is consistent with the empirical am-biguities identified by Oates (1985, 1989)and others.

Although Wilson and Gordon (1998) donot model capital mobility, it is likely toplay a similar efficiency-enhancing role.In this case, government officials will en-gage in expenditure competition by in-creasing those public inputs that enhancethe productivity of capital. But it is alsopossible that this competition for capitalmight inefficiently distort the pattern ofpublic expenditures away from expendi-tures on public goods or inputs that donot enhance capital productivity, an out-come similar to the Keen–Marchand(1997) finding mentioned above.

CONCLUDING REMARKS

The intellectual history of tax competi-tion seems to have taken a normal route,going from simple models yieldingstraightforward results to more compli-cated and less clear-cut conclusions. Theoriginal insight that tax competition canlead to inefficiently low taxes and publicgood levels has been shown to hold inmore general settings than originally in-vestigated. However, the literature hasalso identified circumstances under whichother inefficiencies occur, and it has alsoinvestigated competition involving a va-riety of nontax instruments. Competitionamong governments is now seen as a lessstraightforward phenomenon than per-haps originally envisioned. In fact, recentwork has begun to examine models inwhich this competition has beneficial as-pects.

The literature has also begun to inves-tigate models in which political processesinvolving self-interested government of-ficials take center stage. Here, intergov-

ernmental competition for mobile factorshas been shown to play a beneficial role.Briefly stated, this competition may in-duce government officials to reduce wastein government. This possibility brings usback to the contrast between Tiebout mod-els and tax competition models madein my introductory remarks. Tieboutmodels are motivated by the view thatcompetition among independent govern-ments is similar to competition in theprivate sector and therefore has desirableefficiency properties. In contrast, taxcompetition models often take the viewthat intergovernmental competition de-parts from the assumptions of the stan-dard competitive model in ways thatnegate its efficiency properties. This pa-per has stressed the role of interregionalexternalities in this regard. As mentionedin the Introduction, Sinn (1997) stressesthe nature of the goods and services pro-vided by governments: since they tend tobe those goods and services for whichcompetitive markets do not performwell, reintroducing competition amonggovernments in their provision is likelyto reintroduce market failures. The politi-cal approach to modeling intergovern-mental competition takes a middleground. On the one hand, it follows theTiebout approach by recognizing that thiscompetition introduces efficiency-enhanc-ing incentives similar to the profit motivesfacing competitive firms. On the otherhand, it departs from Tiebout models byrecognizing that such incentives operatein an environment characterized by mar-ket failures that make a fully efficient equi-librium unattainable. As such, competi-tion among governments has both goodand bad aspects, the importance of whichvary across the attributes of the goods andservices that the governments provide.This assessment suggests a role for inter-vention by a central authority, but bothpolitical considerations and informationproblems should be carefully addressed.


299

Acknowledgments

This paper grew out of a series of lec-tures on tax competition that I presentedduring a visit to the Center for EconomicStudies, University of Munich, in Decem-ber 1996. I am grateful to Hans-WernerSinn and other members of CES for astimulating visit, and I thank Ralph Braid,Sam Bucovetsky, William Hoyt, EckhardJaneba, Patricia Wilson, and John Yingerfor helpful comments and suggestions onthis paper.

References

Anderson, John E., and Hendrik Van den Berg.“Fiscal Decentralization and GovernmentSize: An International Test for LeviathanAccounting for Unmeasured Economic Ac-tivity.” International Tax and Public Finance 5(1998): 171–86.

Bayindir-Upmann, Thorsten.“Two Games of Interjurisdictional Compe-tition When Local Governments Provide In-dustrial Public Goods.” International Tax andPublic Finance 5 (1998): 471–87.

Beck, John H.“Tax Competition, Uniform Assessment,and the Benefit Principle. Journal of UrbanEconomics 13 (1983): 127–46.

Biglaiser, Gary, and Claudio Mezzetti.“Politicians’ Decision Making with Re-Elec-tion Concerns.” Journal of Public Economics66 (1997): 425–47.

Black, Dan A., and William H. Hoyt.“Bidding for Firms.” American Economic Re-view 79 (1989): 1249–56.

Blumenthal, Marsha, and Joel B. Slemrod.“The Compliance Cost of Taxing Foreign-Source Income: Its Magnitude, Determi-nants, and Policy Implications.” InternationalTax and Public Finance 2 (1995): 37–54.

Boadway, Robin, and Michael Keen.“Efficiency and the Optimal Direction ofFederal-State Transfers.” International Taxand Public Finance 3 (1996): 137–55.

Boadway, Robin, Maurice Marchand, andMarianne Vigneault.

“The Consequences of Overlapping TaxBases for Redistribution and Public Spend-ing in a Federation.” Journal of Public Eco-nomics 68 (1998): 453–78.

Bond, Eric W.“Tax Holidays and Industry Behavior.” Re-view of Economics and Statistics 63 (1981): 88–95.

Bond, Eric W., and Thomas A. Gresik.“Regulation of Multinational Firms withTwo Active Governments: A CommonAgency Approach.” Journal of Public Econom-ics 59 No. 1 (January, 1996): 33–53.

Bond, Eric W., and Thomas A Gresik.“Incentive Compatible Information TransferBetween Asymetrically Informed Princi-pals.” Department of Economics, Pennsyl-vania State University. Mimeo, 1998.

Bond, Eric W., and Larry Samuelson.“Strategic Behavior and the Rules for Inter-national Taxation of Capital.” Economic Jour-nal 99 (1989): 1099–1111.

Braid, Ralph M.“Spatial Competition Between JurisdictionsWhich Tax Perfectly Competitive Retail (orProduction) Centers.” Journal of Urban Eco-nomics 34 (1993): 75–95.

Braid, Ralph M.“Symmetric Tax Competition with MultipleJurisdictions in Each Metropolitan Area.”American Economic Review 86 No. 5 (Decem-ber 1996): 1279–90.

Braid, Ralph. M.“Property Taxes, Wage Taxes, and Tax Ex-porting with Multiple Jurisdictions in EachMetropolitan Area.” Department of Econom-ics, Wayne State University. Mimeo, 1997.

Braid, Ralph M.“A Spatial Model of Tax Competition withMultiple Tax Instruments.” Department ofEconomics, Wayne State University. Mimeo,1998.

Brennan, Geoffrey, and James Buchanan.The Power to Tax: Analytical Foundations of aFiscal Constitution. New York: CambridgeUniversity Press, 1980.


300

Brueckner, Jan K.“Tiebout/Tax-Competition Models.” Work-ing Paper No. 69. Champaign: The Instituteof Government and Public Affairs, Univer-sity of Illinois at Urbana–Champaign, 1999.

Brueckner, Jan K., and Luz A. Saavedra.“Do Local Governments Engage in Strate-gic Tax Competition.” Department of Eco-nomics, University of Illinois at Urbana–Champaign. Mimeo, 1998.

Bucovetsky, Sam.“Asymmetric Tax Competition.” Journal ofUrban Economics 30 (1991): 67–181.

Bucovetsky, Sam.“Rent Seeking and Tax Competition,” Jour-nal of Public Economics 58 (1995): 337–63.

Bucovetsky, Sam, and John D. Wilson.“Tax Competition with Two Tax Instru-ments” Regional Science and Urban Econom-ics 21 (1991): 333–50.

Bucovetsky, Sam, Maurice Marchand, andPierre Pestieau.

“Tax Competition and Revelation of Prefer-ences for Public Expenditure.” Journal ofUrban Economics 44 (1998): 367–90.

Burbidge, John B., and Gordon M. Myers.“Population Mobility and Capital Compe-tition.” Regional Science and Urban Econom-ics 24 (1994): 441–59.

Burbidge, John B., James A. DePater, GordonM. Myers, and Abhijit Sengupta.

“A Coalition-Formation Approach to Equi-librium Federations and Trading Blocs.”American Economic Review 87 (1997): 940–56.

Cremer, Helmuth, and Firouz Gahvari.“Tax Evasion, Tax Policy and EconomicIntegration.” Department of Economics,University of Illinois at Urbana–Champaign. Mimeo, 1996.

Dahlby, Bev.“Fiscal Externalities and the Design of Gov-ernment Grants.” International Tax and Pub-lic Finance 3 (1996): 397–412.

Davies, Ronald B.“The OECD Model Tax Treaty: Tax Compe-tition and Two-Way Capital Flows.” Depart-ment of Economics, Pennsylvania StateUniversity. Mimeo, 1998.

Davies, Ronald B., and Thomas A. Gresik.“Tax Competition and Foreign Debt.” Work-ing Paper No. 4-98-1. University Park: De-partment of Economics, Pennsylvania StateUniversity, 1998.

de Crombrugghe, Alain, and Henry Tulkens.“On Pareto Improving Commodity TaxChanges Under Fiscal Competiton.” Journalof Public Economics 41 (1990): 335–50.

DePater, James A., and Gordon M. Myers.“Strategic Capital Tax Competition: A Pecu-niary Externality and a Corrective Device.”Journal of Urban Economics 36 (1994): 66–78.

Dhillon, Amitra, Carlo Perroni, and KimberleyScharf.“Implementing Tax Coordination.” Journalof Public Economics (forthcoming).

Diamond, Peter A., and James Mirrlees.“Optimal Taxation and Public Production.”American Economic Review 61 (1971): 8–27and 261–78.

Dixit, Avinash, and Victor Norman.Theory of International Trade. London: Cam-bridge University Press, 1980.

Edwards, Jeremy, and Michael Keen.“Tax Competition and Leviathan.” EuropeanEconomic Review 40 (1996): 113–34.

Fischel, William A.“Fiscal and Environmental Considerations inthe Location of Firms in Suburban Commu-nities.” In Fiscal Zoning and Land Use Controls,edited by Edwin Mills and Wallace Oates,119–74. Lexington, MA: D.C. Heath, 1975.

Frenkel, Jacob A., Assaf Razin, and EfraimSadka.

International Taxation in an Integrated World.Cambridge, MA: MIT Press, 1991.

Gordon, Roger H.“Taxation of Investment and Savings in theWorld Economy.” American Economic Review76 (1986): 1086–1102.

Gordon, Roger H., and Jeffrey K. MacKie-Mason.

“Why is There Corporate Taxation in a SmallOpen Economy? The Role of Transfer Pric-ing and Income Shifting.” In The Effects ofTaxation on Multinational Corporations, editedby M. Feldstein, J. Hines, Jr., and R. Hubbard.Chicago: University of Chicago Press, 1995.


301

Gordon, Roger H., and A. Lans Bovenberg.“Why is Capital so Immobile Internation-ally? Possible Explanations and Implica-tions for Capital Income Taxation.” Ameri-can Economic Review 86 (1996): 1057–75.

Gordon, Roger H., and John D. Wilson.“Tax Structure and Government Behavior:A Principal-Agent Model of Government.”Department of Economics, Michigan StateUniversity. Mimeo, 1998.

Gresik, Thomas.“The Taxing Task of Taxing Transnationals.”Department of Economics, PennsylvaniaState University. Mimeo, 1998.

Haaparanta, Pertti.“Competition for Foreign Direct Invest-ments.” Journal of Public Economics 63 (1996):141–53.

Hamilton, Bruce W.“Zoning and Property Taxation in a Systemof Local Governments.” Urban Studies 12(1975): 205–11.

Hamilton, Bruce W.“Capitalization of Intrajurisdictional Differ-ences in Local Tax Prices.” American Eco-nomic Review 66 (1976): 743–53.

Haufler, Andreas.“Tax Coordination with Different Prefer-ences for Public Goods: Conflict or Har-mony of Interest?” International Tax andPublic Finance 3 (1996): 5–28.

Haufler, Andreas.“Asymmetric Commodity Tax Competition—Comment on de Crombrugghe and Tulkens.Journal of Public Economics 67 (1998): 135–44.

Haufler, Andreas, and Ian Wooton.“Country Size and Tax Competition for For-eign Direct Investment.” Journal of PublicEconomics 71 (1999): 121–39.

Hines, James R., Jr.“Tax Policy and the Activities of Multina-tional Corporations.” In Fiscal Policy: Les-sons from Economic Research, edited byAlan J. Auerbach, 401–45. Cambridge, MA:MIT Press, 1997.

Hines, James R., Jr.“Lessons from Behavioral Responses to In-ternational Taxation.” National Tax Journal52 No. 2 (June, 1999): 305–22.

Hoyt, William H.“Competitive Jurisdictions, Congestion, andthe Henry George Theorem: When ShouldProperty Be Taxed Instead of Land?” Re-gional Science and Urban Economics 21(1991a): 351–70.

Hoyt, William H.“Property Taxation, Nash Equilibrium, andMarket Power.” Journal of Urban Economics34 (1991b): 123–31.

Hoyt, William H.“Tax Policy Coordination and Optimal Taxa-tion in a System of Hierarchical Govern-ments.” Department of Economics, Univer-sity of Kentucky. Mimeo, 1996.

Hoyt, William H., and Richard A. Jensen,“Precommitment in a System of Hierarchi-cal Governments.” Regional Science and Ur-ban Economics 26 (1996): 481–504.

Huang, Yophy.Tax Competition with Involuntary Unemploy-ment. Ph.D. diss. Bloomington: Indiana Uni-versity, 1992.

Huber, Bernd.“Tax Competition and Tax Coordination inan Optimum Income Tax Model.” Journal ofPublic Economics 71 (1999): 441–58.

Janeba, Eckhard.“Corporate Income Tax Competition,Double Taxation Treaties, and Foreign Di-rect Investment.” Journal of Public Econom-ics 56 (1995): 311–26.

Janeba, Eckhard.“Tax Competition in Imperfectly Competi-tive Markets.” Journal of International Eco-nomics 44 (1998a): 135–53.

Janeba, Eckhard.“Tax Competition When Governments LackCommitment: Excess Capacity as aCountervailing Threat.” Department of Eco-nomics, Indiana University. Mimeo, 1998b.

Janeba, Eckhard, and Wolfgang Peters.“Tax Evasion, Tax Competition and theGains from Nondiscrimination: The Caseof Interest Taxation in Europe.” EconomicJournal 109 (1999): 93–101.

Jensen, Richard, and Eugenia F. Toma.“Debt in a Model of Tax Competition.” RegionalScience and Urban Economics 21 (1991): 371–92.


302

Kanbur, Ravi, and Michael Keen.“Jeux Sans Frontieres: Tax Competition andTax Coordination When Countries Differ inSize.” American Economic Review 83 (1993):877–92.

Keen, Michael.“Vertical Tax Externalities in the Theory ofFiscal Federalism.” International MonetaryFund Staff Papers 45 (1998): 454–84.

Keen, Michael, and Christos Kotsogiannis.“Federalism and Tax Competition,” Depart-ment of Economics, University of Essex.Mimeo, 1996.

Keen, Michael, and Maurice Marchand.“Fiscal Competition and the Pattern of Pub-lic Spending.” Journal of Public Economics 66(1997): 33–53.

Kehoe, Patrick J.“Policy Cooperation Among BenevolentGovernments May Be Undesirable.” Reviewof Economic Studies 56 (1989): 289–96.

Kennan, John, and Raymond Riezman.“Do Big Countries Win Tariff Wars?”International Economic Review 29 (1988):81–5.

Kim, Jinyoung, and John D. Wilson.“Racing to the Bottom with Multiple Tax In-struments.” Japan and the World Economy 9(1997): 537–51.

King, Ian, R. Preston McAfee, and LindaWelling.

“Industrial Blackmail: Dynamic TaxCompetition and Public Investment.”Canadian Journal of Economics 26 (1993): 590–608.

Krelove, Russell.“The Persistence and Inefficiency of Prop-erty Tax Finance of Local Public Expendi-tures.” Journal of Public Economics 51 (1993):415–35.

Lee, Kangoh.“Tax Competition with Imperfectly MobileCapital.” Journal of Urban Economics 42(1997): 222–42.

Lee, Kangoh.“Factor Ownership and Governmental Stra-tegic Interaction: An Efficient Nash Equi-librium.” Department of Economics,Towson University. Mimeo, 1998.

Lockwood, Ben.“Commodity Tax Competition Under Des-tination and Origin Principles.” Journal ofPublic Economics 52 (1993): 141–62.

Markusen, James R., Edward R. Morey, andNancy Olewiler.

“Competition in Regional Environmental Poli-cies When Plant Locations Are Endogenous.”Journal of Public Economics 56 (1996): 55–77.

Martin, Laurent.“Bidding for Firms: An Asymmetric Auc-tion Model of Interjurisdictional Competi-tion.” Department of Economics, Universityof Maryland. Mimeo, 1997.

Mezzetti, Claudio.“Common Agency with Horizontally Differ-entiated Principals.” Rand Journal of Econom-ics 28 (1997): 323–45.

Mieszkowski, Peter, and George Zodrow.“Taxation in the Tiebout Model.” Journal ofEconomic Literature 27 (1989): 1089–1146.

Mintz, Jack, and Henry Tulkens.“Commodity Tax Competition BetweenMember States of a Federation: Equilibriumand Efficiency.” Journal of Public Economics29 (1986): 133–72.

Mirrlees, James A.“An Exploration in the Theory of OptimumIncome Taxation.” Review of Economic Stud-ies 38 (1971): 175–208.

Niskanen, William A., Jr.Bureaucracy and Representative Government.Chicago: Aldine, 1971.

Noiset, Luc.“Pigou, Tiebout, Property Taxation, and theUnderprovision of Local Public Goods:Comment.” Journal of Urban Economics 38(1995): 312–16.

Oates,Wallace E.Fiscal Federalism. New York: Harcourt BraceJovanovich, 1972.

Oates, Wallace E.“Searching for Leviathan: An EmpiricalAnalysis.” American Economic Review 75(1985): 748–57.

Oates, Wallace E.“Searching for Leviathan: A Reply and SomeFurther Reflections.” American EconomicReview 79 (1989): 578–83.


303

Oates, Wallace E., and Robert M. Schwab.“Economic Competition Among Jurisdic-tions: Efficiency Enhancing or Distortion In-ducing?” Journal of Public Economics 35(1988): 333–54.

Osmundsen, Petter, Kare P. Hagen, andGuttorm Schjelderup.

“Internationally Mobile Firms and TaxPolicy.” Journal of International Economics 45(1998): 97–113.

Perroni, Carlo, and Kimberley A. Scharf.“Tiebout with Politics: Capital Tax Compe-tition and Constitutional Choices.” WorkingPaper No. 1997–05. Copenhagen: EconomicPolicy Research Unit, Copenhagen BusinessSchool, 1997.

Persson, Torsten, and Guido Tabellini.“The Politics of 1992: Fiscal Policy and Eu-ropean Integration.” Review of EconomicStudies 59 (1992): 689–701.

Rauscher, Michael.“Environmental Regulation and the Loca-tion of Polluting Industries.” InternationalTax and Public Finance 2 (1995): 229–44.

Rauscher, Michael.“Interjurisdictional Competition and the Ef-ficiency of the Pubic Sector: The Triumphof the Market over the State?” Working Pa-per 732. Kiel: Institute of World Econom-ics, 1996.

Rauscher, Michael.“Leviathan and Competition Among Juris-dictions: The Case of Benefit Taxation.”Journal of Urban Econonomics 44 (1998): 59–67.

Razin, Assaf, and Efraim Sadka.“International Tax Competition and Gainsfrom Tax Harmonization.” Economics Letters37 (1991): 69–76.

Richter, Wolfram F., and Dietmar Wellisch.“The Provision of Local Public Goods andFactors in the Presence of Firm and House-hold Mobility.” Journal of Public Economics60 (1996): 73–93.

Ross, Stephen, and John Yinger.“Sorting and Voting: A Review of the Lit-erature on Urban Public Finance.” Centerfor Policy Research, The Maxwell School,Syracuse University. Mimeo, 1998.

Sinn, Hans-Werner.“The Selection Principle and Market Failurein Systems Competition.” Journal of PublicEconomics 66 (1997): 247–74.

Tannenwald, Robert.“Devolution: The New Federalism—AnOverview.” New England Economic Review(1998): 1–12.

Tiebout, Charles M.“A Pure Theory of Local Expenditures.” Jour-nal of Political Economy 64 (1956): 416–24.

Trandel, Gregory A.“Interstate Commodity Tax Differentials andthe Distribution of Residents.” Journal ofPublic Economics 53 (1994): 435–57.

Wellisch, Dietmar.The Theory of Public Finance in a Federal State.New York: Cambridge University Press(forthcoming).

White, Michelle J.“Firm Location in a Zoned MetropolitanArea.” In Fiscal Zoning and Land UseControls, edited by Edwin Mills and WallaceE. Oates. Lexington, MA: D.C. Heath,1975.

Wildasin, David E.“Nash Equilibria in Models of Fiscal Com-petition.” Journal of Public Economics (1988).

Wildasin, David E.“Interjurisdictional Capital Mobility: FiscalExternality and a Corrective Subsidy.” Jour-nal of Urban Economics 25 (1989): 193–212.

Wildasin, David E.“Some Rudimentary Duopolity Theory.” Re-gional Science and Urban Economics 21 (1991):393–421.

Wilson, John D.“A Theory of Interregional Tax Competi-tion.” Journal of Urban Economics 19 (1986):296–315.

Wilson, John D.“Trade, Capital Mobility, and Tax Competi-tion.” Journal of Political Economy 95 (1987):835–56.

Wilson, John D.“Tax Competition with Interregional Differ-ences in Factor Endowments.” RegionalScience and Urban Economics 21 (1991): 423–52.

.


304

Wilson, John D.“Mobile Labor, Multiple Tax Instruments,and Tax Competition.” Journal of Urban Eco-nomics 38 (1995): 333–56.

Wilson, John D.“Capital Mobility and Environmental Stan-dards: Is There a Theoretical Basis for a Raceto the Bottom?” In Harmonization and FairTrade, Vol. 1, edited by Jagdish Bhagwati andRobert Hudec, 395–427. Cambridge, MA:MIT Press, 1996a.

Wilson, John D.“The Tax Treatment of Imperfectly MobileFirms: Rent-Seeking, Rent-Protection, andRent-Destruction.” In The Political Economyof Trade Policy: Papers in Honor of Jagdish

Bhagwati, edited by Robert C. Feenstra, GeneM. Grossman, and Douglas A. Irwin, 225–44. Cambridge, MA: MIT Press, 1996b.

Wilson, John D.“Property Taxation, Congestion, and LocalPublic Goods.” Journal of Public Economics64 (1997): 202–17.

Wilson, John D., and Roger H. Gordon.“Expenditure Competition.” Department ofEconomics, Michigan State University.Mimeo, 1998.

Zodrow, George R., and Peter Mieszkowski.“Pigou, Tiebout, Property Taxation, and theUnderprovision of Local Public Goods.”Journal of Urban Economics 19 (1986): 356–70.

theories of tax competition - national tax journal

Documents