1、Fiscal competition between decentralized jurisdictions,theoretical and empirical evidenceClement CarbonnierTHEMA - Universite de Cergy-PontoiseAdress: THEMA - Universite de Cergy-Pontoise, 33 boulevard du port, F 95011 Cergy-Pontoise cedex, FrancePhone: +33 1 34 25 63 21Fax: +33 1 34 25 62 33E-mail:
2、 carbonnierjourdan.ens.frFebruary 29, 2008AbstractThis article provides theoretical and empirical evidence that local fiscal competition generates a bias toward lowbusiness tax rates. Furthermore, it is shown that this bias is stronger for smaller jurisdictions. First, a theoreticalmodel is settled
3、with private and public capital and a fixed factor. The fixed factor allows to consider differencesbetween the jurisdictions. The results show that there exists a bias toward low tax rates due to tax competition.This bias generates an underprovision of public capital, and therefore production is sma
4、ller with tax competitionthan with cooperation. Moreover, the bias toward low tax rates is stronger for jurisdictions with less fixed factor.That means that tax competition generates a larger production decrease for smaller jurisdictions. The empiricalpart aims at estimating the bias toward low tax
5、rates and its dependency with respect to the fixed factor. Panelregressions with temporal and individual fixed effects of the tax rates are implemented with French local data, usingthe creation of intercity communities. The results indicate that the bias toward low local tax rates is strong: up to23
6、% decrease for the smaller cities. It is also significantly decreasing with respect to the city size: there is no taxrate decrease due to tax competition for the biggest cities.Key words: Optimal taxation; Business taxes ; Tax competition ; Public capital; Firm location.JEL classification: H21; H25
7、; H73; R12; R30.11 IntroductionThe level of decentralization is a crucial issue in state organization. For an example, France hasbegun a decentralization second run since the year 2003. The first run occurred during years 1982and 1983. However, some local jurisdictions seem to be too much small to t
8、ake profit from thedecentralization. Thus, city unions also occurred since the year 1999. Hence, both decentralizationand centralization are processing. All developed countries have a decentralized authority. Thereare several reasons to give local jurisdiction authority. Historical development is on
9、e of these.Another important reason is that decentralized authority may make better choices concerningpublic investments. The nearer to the investment the decision is taken, the more it fits the needs.Therefore, decentralization allows public investments to be more productive. Theoretically, Alesina
10、& Spolaore (1997) use this to build a model explaining the number of countries, with peoplemobility and preference heterogeneity. Two forces act in this model. In the one hand, creatingmore jurisdictions has a fixed cost by administration. In the other hand, creating more jurisdictions- and therefor
11、e jurisdictions with less people - allows local governments to take decisions closer tothe inhabitant preferences. Empirically, this result is confirmed by Barankay & Lockwood (2007).They study the case of education in Switzerland, financed locally by the cantons.However, decentralization is not fre
12、e. Multiplying decision levels also multiplies the adminis-tration costs. Furthermore, there may be costs due to local tax competition. Indeed, the differentlocal administrations may enter a fiscal competition with their neighbors, which may result in taxrate decreases and then in public investment
13、decreases.The aim of the present paper is to point out local tax competition. First, theoretical evidenceare presented. A local tax model is settled. It is a very classical tax competition model, whichdemonstrates that local tax competition generates a bias toward low tax rates. The novelty is that,
14、due to decreasing factor returns, the bias toward low tax rates is stronger for smaller jurisdictions.Then, empirical evidence confirms the model results. The bias is estimated through French inter-city agreement modifications. It is found strong (up to 23% for small cities) and significantlydecreas
15、ing with the city size.Some papers have already studied tax competition with a local point of view. Theoretically first,Zodrow & Mieszkowski (1986) build a model of local jurisdictions. Each local government choosesthe business tax rate as the best response to the neighbor rates and the national rat
16、e. Private2capital elasticity is fixed exogenously as a model parameter. It results in a business tax rates Nashequilibrium. The authors find that local tax rates are strategic complements. However, they donot determine if local tax rates and national tax rates are strategic complements or substitut
17、es.Benassy-Quere et al. (2005) introduce at an international level the idea of positive relationshipbetween tax and base. They explain it by public investment arguments.Haughwout et al. (2004), Haughwout & Inman (2001) and Mutti et al. (1989) study calibratedmodels. Haughwout et al. (2004) and Haugh
18、wout & Inman (2001) conclude that local taxes onbusiness activities are allready too much high in Philadelphia and New-York. Mutti et al. (1989)calibrate a six regions model, and conclude that the relationship between rates and bases of localbusiness taxes may be ambiguous.From an empirical point of
19、 view, Boadway & Hayashi (2001) study the case of Canada. Theyestimate the tax rate decision interaction in a three provinces model: Ontario, Quebec and therest of Canada. Buettner (2001) does the same thing with a panel of German jurisdictions. Bothconfirm the fact that local business tax rates are
20、 strategic complements the ones to the others.Furthermore, both find that local and central business tax rates are strategic substitutes. Buettner(2003) tests the impact of local business tax rates on local business tax bases in Germany and findsit negative. Bell & Gabe (2004) measure the policy imp
21、act on new establishment location and findthat additional public spending and higher taxes may be good to attract firms.Thornton (2007) studies the link between fiscal decentralization and growth from a macroe-conomical point of view. Measuring the decentralization by the full-autonomy fiscal revenu
22、e ofsub-national administration, he finds that there is no significant relationship between fiscal decen-tralization and growth.The rest of the article is organized as follows. In Section 2, the theoretical model is presented.Subsection 2.1 presents the model without fiscal competition. It results i
23、n the first best optimallocal business tax rate. Subsection 2.2 introduces fiscal competition and results in the second bestoptimal business tax rate. The model finds a bias toward low local business tax rates generatedby fiscal competition, this bias is decreasing with respect to the jurisdiction s
24、ize. In Section 3, thedata used for the empirical study is presented. The French inter-city reform is explained. Since1999, some cities choose to sign an inter-city agreement to build a new administrative level: a cityunion. Then, four data bases are presented. There is an inter-city union data base
25、, a local tax data3base, a local social properties data base and a geographical data base. In Section 4, the empiricalstudy is presented. The model parameters are estimated. With this estimates, the bias towardlow rate is calculated. It is found strong and decreasing with respect to the jurisdiction
26、 size. Italso shows that resolving the fiscal competition issue allows increasing the business local tax rateswithout any negative impact on private capital settlement. In Section 5, conclusions are presented.2 Theoretical frameworkIn order to understand the impact of tax competition on local corpor
27、ate tax rates, a standard modelis settled. The novelty is to consider an asymetrical allocation of the fixed production factor. Itallows to catch the city size impact on local corporate tax rate. The model considers a countrywith a fixed number of n cities. At each period t in each city i (i = 1n),
28、there is lit inhabitants,kit private capital and pit public capital. These production factors allow private firms to produceyit with the production function yit = F(kit,lit,pit). The production function used for this model isa Cobb-Douglas production function yit = Akitlitpit, with two kinds of capi
29、tal, private and public.No hypothesis is assumed on return to scales.In order to focus on capital only, lit is supposed exogenous. There is a total amount of fixedfactor Lt distributed irregularly among the cities. The point is to understand the impact of lit onthe local corporate tax rate it. lit m
30、ay be interpreted as land, that is really a fixed productionfactor. However, the number of inhabitants, considered in its static point of view, may be a goodproxy of the city size.The public capital is financed by local business taxation. City i taxes private capital kit at rateit and invest the rev
31、enue itkit as public capital for the following period t+1. As public capital isdepreciating at rate , the public capital quantity pit at time t in city i is pit = (1)pit1+it1kit1.In each city, entrepreneurs borrow private capital and organise production. In order to maximizeemployement and income fo
32、r inhabitants, the objective of the city is to maximize production.When a city tax rate varies, two phenomenon impact private capital. The total quantity in thecountry K varies, and remaining private capital is reallocated betweens cities. Total private capitalK is the result of inter-temporal consu
33、mption optimization of agent utility u(ct,.,ct+k,.), wherect is the consumption at period t, as much as the result of international capital partial mobility.Therefore, the total amount K of private capital depends on the private capital returns. At period4t, the impact of tax it on public capital pi
34、t has not occurred yet. Therefore, the elasticity of Kwith respect to do not depend on p. To measure this variation, the total capital elasticity withrespect to local business tax rate epsilon1K = 1iki Ki is used.Then, the private capital quantity kit in each city is the result of the total capital
35、Kt allocationbetween cities. This allocation is done in order to equalize the capital returns between cities.Equation (1) is the condition for the capital returns to be equal in each city.yiki = g1 = Ak1i (1i)li pi (1)Where g1 is equal for all cities i. Equation (2) gives the resulting private capit
36、al allocation ki,as a function of pi, li, i and K.ki = f(i)summationtextnj=1 f(j)Kf(i) = (1i) 1p1i l1i(2)Equation (2) gives ki as a fraction of K. Moreover, as summationtextnj=1 f(j) is not depending on cityi, the fraction of K is higher when f(i) is higher. The cities with more fixed production fac
37、tor,more public capital or less taxes attract more private capital. The intensity of this attractivenessis increasing with the productivity parameters , and . The private capital productivity impacts the low tax rate attractiveness. The public capital productivity impacts the publiccapital attractiv
38、itness. The fixed factor productivity impacts the city size attractivitness.Two ways of resolving this model are implemented. First, the optimization process is done inorder to maximise the overall production. It is the case of cooperation between cities, with no fiscalcompetition. Second, fiscal co
39、mpetition may occur and each city maximizes its own production,with using its own rate. The model is solved in Nash equilibrium.2.1 Resolution with cooperationThis first Subsection consists in resolving the model with cooperation between cities. This is athree steps problem. First, cities choose a t
40、ax rate. Second, private capital owners choose wherethey invest their savings. Finally, the production process is settled. The optimisation problemwithout fiscal competition consists in determining the set tax rates for each city that maximizesthe overall production, under private capital settlement
41、 constraints. Resolution is done at the5permanent equilibrium. Therefore, equation (3) gives the permanent equilibrium public capital, asa function of permanent equilibrium tax rate and private capital.pi = i ki (3)Thus, production in city i may be given as a function of ki and i, as presented in eq
42、uation (4).yi = A(1i)klbracketleftBigi kibracketrightBig= A lk+i (1i) (4)As there is no fiscal competition, the goal of the optimization problem is to maximise Y =summationtextni=1 yi, controling with the tax rates (i)i=1n. First order conditions depending on i are givenby equation (5).A k+j lj (1j)
43、jbracketleftbiggj 1jbracketrightbigg= nsummationdisplayi=1( + )Akijk+1i (1i)li i (5)The left hand term of equation (5) is decreasing from + to when j goes from 0 to 1.The right hand term of this equation is positive and finite. Hence, there exists a solution strictlybetween 0 and 1. This solution is
44、 a maximum because yi is positive for i between 0 and 1 and yiis equal to zero for i = 0 and i = 1.In order to calculate the optimal tax rates i , condition (5) has to be simplified. Left handterm of equation (5) is equal to right hand term of formula (6). Right hand term of equation (5) isequal to
45、formula (7) term. According to these two simplifications, equation (8) gives the value ofthe optimal tax rates in each city.A k+j lj (1j)jbracketleftbiggj 1jbracketrightbigg= kj yikibracketleftbiggj 1jbracketrightbigg(6)nsummationdisplayi=1( + )Akijk+1i (1i)li i + yikikj1j epsilon1K (7)i = + 11 + ep
46、silon1K(8)The main property of that first best optimum is that all cities have the same optimal tax rate.This tax rate does not depend on the number of cities and not either on the city sizes. The optimalrate formula is composed of two different terms.6The first term is + and reflects the optimal ra
47、tio between ki and pi. This term comes fromthe maximization of i (1i). Therefore, is decreasing with respect to because it representsthe private capital productivity in the Cobb-Douglas production function. The more productiveprivate capital is, the higher is the cost of taxing it. In addition, is i
48、ncreasing with respectto because it represents the productivity of the public capital in the Cobb-Douglas productionfunction. The more productive public capital is, the higher are the benefits of taxation. As thisfirst term represents an optimal ratio between private and public capital, it does not
49、depend onthe city size.The second term is 11+epsilon1K, and represents the classical fiscal arbitrage between tax rate and taxbase. If base elasticity with respect to tax rate is high, optimal tax rate is low, and vice versa.2.2 Resolution with fiscal competitionIn this second Subsection, fiscal competition is introduced. The optimization problem consists foreach city in maximizing its own p
