Vertical Capital Tax Reaction Functions: Evidence from Sub-National Governments in France. September 26, 2003

Transcription

1 Vertical Capital Tax Reaction Functions: Evidence from Sub-National Governments in France September 26, 2003 Timothy J. Goodspeed Hunter College CUNY Department of Economics 695 Park Avenue New York, NY USA Telephone : tgoodspe@hunter.cuny.edu Matthieu Leprince CREREG (UMR CNRS 6585) Faculté des sciences économiques de l université de Rennes 1 7 place Hoche Rennes cedex France Téléphone : matthieu.leprince@univ-rennes1.fr Abstract: This paper investigates the extent to which the municipal or commune level of government in France interacts strategically with the département level in the setting of the local business tax, the taxe professionnelle. This is the main local tax revenue source in France giving rise to more than 50% of the total tax revenues for the communes. We use a unique cross-sectional data set on the 850 French communes (all those with population greater than 10,000) for 1995 to estimate the slope of the vertical reaction function between communes and départements. We find a significant and positive relationship between the commune business tax rate and the département business tax rate. The implied elasticity suggests that a 10 percent increase in the département business tax rate leads to a 2 percent increase in the commune tax rate. We also find that social spending mandated by the central government increases municipal business taxes. We obtained these estimates after controlling for tax base and other budgetary differences between communes, régional or cultural differences common to both départements and communes, and differences in total public spending in départements and communes. 1

2 1. Introduction Capital tax competition is an important issue with far-reaching policy implications. The literature on multinational taxation has established the importance of capital taxes for the worldwide location of capital (see e.g. Hines (1996) for a review). Furthermore, Altshuler, Grubert, and Newlon (2001) find that U.S. multinationals were more sensitive to foreign tax rates in 1992 than in A set of recent papers (see e.g. Buettner (2003) for a recent paper using European data and the comprehensive review by Brueckner, 2003) study horizontal capital tax competition within countries, that is how governments at the same level interact with each other in the face of mobile capital. Horizontal tax competition between countries has also recently been studied by Devereux, Lockwood, and Redonao (2002). Altshuler and Goodspeed (2003) find that European countries followed the lead of the United States in changing their capital tax systems after the 1986 U.S. Tax Reform Act, but not before. Capital tax competition between levels of government is also an important phenomenon in a federation. Vertical tax competition between the central and lower levels of government has been studied by Besley and Rosen (1998) for sales taxation in the United States, Goodspeed (2000, 2002) for income taxes in OECD countries, Hayashi and Boadway (2000) for capital taxes in Canada, and Esteller-Moré and Solé-Ollé (2001) for income taxes in the United States. Interactions between different tiers of sub-national governments (e.g. states and municipalities) have not been estimated. We add to the literature that attempts to estimate fiscal reaction functions by estimating a capital tax reaction function for local governments (communes) in France with respect to the higher- 2

3 tier but still sub-national département level of government. We also provide evidence on the impact of mandated social spending on local business taxes. While highly centralized in certain respects, France is a country with multiple levels of government. In addition to the central government, three other local government levels are prominent. Listed from highest level to lowest, these are the région, the département, and the commune. Our focus will be on the interaction between the commune level and the département level, the two major sub-national spenders. The commune level undertook 60 percent of sub-national spending while départements undertook 30 percent in 1995, the year of our data. The French local capital tax that we examine is the taxe professionnelle. This tax is the major revenue source for lower levels of government in addition to transfers. The base of the tax is determined through a formula that resembles the apportionment formulas used by US states. 1 The French case provides something of a clean test since, unlike the US state case, the central government determines the weights of the factors which are the same for all sub-national governments. The sub-national governments are free to set the tax rate, however. Our results indicate a positive reaction function. When the département lowers its tax rate on capital, the communal level of government also lowers its tax rate. The magnitude of the change is relatively small, however. We estimate an elasticity of approximately 0.2. The remainder of the paper is organized as follows. Section 2 provides some background concerning the French business tax that we study. Section 3 develops the empirical model. Section 4 presents the results and Section 5 concludes. 1 Weiner (1994) investigates capital s movements resulting from the formula apportionment taxations of capital by U.S. states. 3

4 2. Background on French Sub-national Governments and Business Tax (Taxe Professionnelle) The local public sector in continental France is multi-tiered with three levels of government: the municipal level (with communes or cities), the intermediate level with 94 départements and the higher level of 21 régions. There are more than municipalities, but half the 60 million population lives in the 850 municipalities of more than inhabitants. Communes (or municipalities) are responsible for urban planning, provide the main local public services (water, sewage, sanitation, public transport), finance the capital spending of nursery and primary schools, sports facilities and different public amenities. Départements provide social services to children, disabled and aged people, and training programs to unemployed people. They are also responsible for capital spending for elementary schools (collèges) and roads. Finally, départements provide some capital and economic development grants to municipalities. Régions provide régional planning, development amenities, ongoing vocational training and apprenticeship, and finance the capital spending of high schools (lycées). Régions also provide some capital and economic development grants to municipalities and départements (but the amount of régional grants is much lower than that provided by départements to municipalities). Broadly speaking, the division of responsibilities between the three levels of government doesn t vary across départements or régions. However, variation can result 4

5 from differences in population density. In rural areas, very small municipalities tend to allow the département to be responsible for some basic local public services that exhibit economies of scale (municipal roads, public transportation, economic development services) whereas in urban areas with the major municipalities (the sub-group whose tax rate is our dependent variable), the scope of départemental responsibilities remains much more in accordance to the usual division. As previously mentioned, the municipalities represent about 60% of total spending in the local public sector, départements 30% and régions 10%. Whatever the level of local government one considers, spending is roughly covered by three main revenue sources: local taxes (50%), grants from central or local governments (30%) and debt (10%). The major tax revenue source at both the commune and département levels is a business tax called the taxe professionnelle (hereafter TP). The base of this tax is determined by the central government according to a formula. In 1995 a business base included 18% of employee wages paid by the company in the jurisdiction, 100% of the value of the land in the jurisdiction used by the company in production, and 16% of the historical cost of a company s equipment used in the jurisdiction. Each level of government sets its own tax rate on this common base. The remaining fiscal revenues are mainly collected from households and come from a housing tax (the taxe d habitation, TH), a property tax (the taxe foncière sur le bâti, TFB) and a land tax (the taxe foncière sur le foncier non bâti, TFNB). Sub-national governments ability to change the business tax rate is limited by a legal requirement: the rate of the business tax may not vary more rapidly than the rates of taxes borne by 5

6 households. The shares of revenues from the four taxes for all levels combined are 50% for the TP, 22% for the TH, 25% for the TFB, and 3% for the TFNB. The mean cumulated TP rate in 1995 is 22.3% (14% communes ; 6.5% départements ; 1.8% régions). 3. Econometric Model The main purpose of the econometric model is to estimate the reaction of the commune level capital tax to a change in the département level capital tax. One might think of the underlying game between the commune and département in different ways. The communes might play Nash with respect to the département and move simultaneously. Alternatively, one might think of the département moving first as a Stackelberg leader with the communes following. We estimate the reaction function under both assumptions. Our basic econometric estimating equation is: C C D C R γ ln R + θ ln S + η ln E + µ ln D + ν ln D + C D lnttp = b0 + β lnttp + ε where t C TP and t D TP are the commune and département taxe professionnelle rates, respectively, and the control variables include other revenue sources and expenditures of the communes, R C, socio-economic characteristics of the communes, S C, expenditures by the départements, E D, a set of dummies for certain geographic and institutional characteristics of the communes, D C, and a set of régional government dummies, D R. 6

7 Sub-national data is not normally published by the French government and consequently is hard to come by. Our data set combines three data sets on sub-national governments. The first is a 1995 cross-section of the budgets of the 850 municipalities (communes) with population greater than 10,000. We also have socio-economic data on this level of government. This data is merged with a second data set at the département level and a third data set at the régional level. The dependent variable is the communal tax rate. 2 While some of the control variables are fairly standard others are somewhat particular to the case of France. We discuss below the control variables. We discuss and control for certain unique circumstances in France that could explain a correlation between the department tax rate and the commune tax rate. The commune budget constraint suggests including the commune spending and revenue variables. We include total spending, grant revenues, and revenues from three tax sources: the property tax, a land tax, and a household tax. These three taxes may have a particularly peculiar impact on the taxe professionnelle rate which does not have to do with tax competition. The household tax (taxe d habitation, TH) for instance, is a tax on the resident of a dwelling (be it the owner or renter) and is based on the value of the property in The conventional wisdom in France is that the value was underestimated in the South. Consequently, commune and département TH rates in these areas, and hence commune and département TP rates, will tend to be higher. A similar problem exists with the other tax bases. The inclusion of the revenues from these three 2 To take account of the effect of intercommunal groupings (certain services are shared in metropolitan areas), the dependent variable is measured as the sum of the municipal tax rate and, if it exits, the additional intercommunal rate. 7

8 tax sources on the right hand side helps to control for any artificially positive impact on the TP rate due to an artificially low tax base at both the commune and département levels. We also include certain fairly standard socio-economic and geographic dummy variables. These include communal population, a dummy for a center-city in the commune, a tourism dummy, a mountain dummy, and a variable reflecting the number of second homes in the commune. These variables represent factors that might influence costs or numbers of users of public services in, for instance, urban versus rural areas, and hence the communal tax rate. Per-capita income is also included to capture demand for public services. Additionally, we include dummies for the 21 régions in continental France to account for any systematic differences between régions. This could partially control for low TH bases in the south, but also is intended to control for other régional differences such as the higher reliance on private schools in the north and on public schools in the south, or simply régional differences in preferences. A second possible reason other than tax competition for a correlation between the départemental and communal tax rates is that communes are mandated to give some grants to départements for social spending. The total amount of such communal grants received by a département is called the contingent communal d aide sociale, hereafter CCAS. The more grants a commune is required to give, the higher would be the communal tax rate. To control for this, we include the level of CCAS grants (for the year 1997). 3 3 Other grants (capital grants) go from départements to communes, but we have no data to control for this. These grants mostly go to small communes that are not in our sample. 8

9 A third possible reason other than tax competition for a correlation between the départemental and communal tax rates is that the tax base at one level is positively correlated with the tax base at the other level. Hence, one might estimate a correlation in tax rates that is due to tax base correlation. To control for this, we include the communal TP base. Finally, départements may have high tax rates simply due to high département preferences or there may be complementarity or substitutability between départemental and communal public services. To control for this possibility, we include départemental spending variables in one specification. An important econometric issue arises in the estimation of equation (1). Under the assumption of Nash behavior between the commune and département levels of government, the département tax rate on the right hand side would clearly be endogenous. To obtain consistent estimates we instrument for the département tax rate under the Nash assumption. Under the assumption of Stackelberg behavior, the département level of government would move first and the département tax rate would be exogenous for the commune. Under this assumption we use the 1990 département tax rate to estimate the 1995 reaction on the part of the commune. 4. Results The results are presented in Table 1. The first two columns assume Nash competition between the commune and département levels of government while the last column assumes Stackelberg behavior. Under the Nash assumption, we use a predicted 9

10 value for the concurrent département tax rate since it is clearly endogenous. Our instruments include all the exogenous variables of the model as well as certain départemental and régional characteristics. Under the Stackelberg assumption, the commune decision comes after the département decision and we use the five-year lag for the value of the département tax rate. The key variable of interest is the département tax rate. It is positive and significant across all three specifications and the point estimate is relatively stable. The implied elasticity evaluated at the means of the sample is approximately 0.2. Several of the control variables are related to the government budget constraint. The municipal tax base is negative and highly significant. Holding other revenue sources and spending constant, the government budget constraint implies that municipalities with higher tax bases will have lower tax rates. Block grants are negative and highly significant while government spending is positive and highly significant. Again, the government budget constraint implies that holding other revenue sources and spending constant, higher grants will lead to lower tax rates. Similarly, higher spending leads to higher tax rates holding other revenue sources constant. Social spending by and large is mandated by the central government and is jointly administered by the département and municipal levels. This could lead to a positive correlation between département and municipal tax rates that is not related to tax competition. We control for this in two ways. In the first column, we include the total municipal grants that help financing départemental social spending. While positive, this coefficient is insignificant. In the second column we include départemental spending variables. The coefficient on social spending is positive and significant. Hence, 10

11 departments that have high social spending may receive higher grants from communes, inducing high commune business tax rates. While this reduces slightly the coefficient and statistical significance of the département tax rate, it remains highly significant. 5. Conclusion This paper investigates the extent to which the commune level interact strategically with the département level by using a cross-sectional data set on the 850 French communes (all those with population greater than 10,000) for 1995 to estimate the slope of the vertical reaction function between communes and départements. Data consists mainly on data on rates and bases for the local business tax, the so-called taxe professionnelle, that is the main local tax revenue source in France with more than 50% of the total tax revenues for the communes. We find a significant and positive relationship between the commune business tax rate (the main revenue source) and the département business tax rate. The implied elasticity suggests that a 10 percent increase in the département business tax rate leads to a 2 percent increase in the commune tax rate. We obtained this estimate after controlling for several other possible explanations including mandated central government spending in social spending, tax base differences between communes or other régional or cultural differences common to both départements and communes, and different preferences for public spending between départements and communes. The results indicate a significant vertical interaction between municipal and département business tax rates in France. Moreover, the results are of great significance 11

12 in the current French political context as France is at the eve of a new devolution process. The central government is currently considering devolving new tax and spending powers to local jurisdictions starting in

13 Table 1 Estimates for Nash and Stackelberg Assumptions Dependent variable: Municipal Business Tax (TP) Rate Nash-1 Nash-2 Stackelberg constant 1.208** (4.35) 0.964** (3.29) 1.193** (5.01) Départemental TP rate 0.202** (4.50) 0.189** (3.89) 0.238** (5.98) municipal TP base ** (-12.86) ** (-12.59) ** (-17.36) TH housing tax revenues (0.35) (0.42) (0.43) FB property tax revenues (1.24) (1.26) (1.24) FNB land tax revenues (-0.34) (0.28) (-0.38) Block grants ** (-4.61) ** (-4.27) ** (-5.03) Total spending 0.737** (11.27) 0.732** (11.14) 0.733** (13.99) Income ** (-6.44) ** (-6.48) ** (-8.01) Tourism dummy (-1.32) (-1.38) (-1.36) Center city dummy (-0.23) (0.26) (-0.45) Mountain dummy (-0.22) (0.44) (-0.38) Nbr of second homes ** (-4.65) ** (-4.40) ** (-5.02) Population 0.051** (3.28) 0.044** (2.82) 0.053** (3.61) Départemental roads spending (-0.91) Départemental social aid spending 0.160** (2.89) Départemental secondary schools spending (0.45) Départemental economic development spending ** (-3.07) 1997 Communal grants for départemental social spending 0.03 (1.27) Intercommunal dummies Yes Yes Yes Régional dummies Yes Yes Yes R N Notes: T-statistics in parentheses. Instruments of the TP départemental rate are the exogenous variables of the model and the following départemental variables: TP départemental base, TH revenues, grants, surface, population. Intercommunal dummies control for variation in the vertical allocation of responsibilities in the municipal and intercommunal block. We exclude Paris and Corsican communes (Ajaccio and Bastia) from the sample. 13

14 References Altshuler, Rosanne and Timothy J. Goodspeed Follow the Leader: Evidence on European and US Tax Competition. Unpublished draft. Altshuler, Rosanne, Harry Grubert, and T. Scott Newlon, 2001, "Has U.S. Investment Become More Sensitive to Tax Rates?" in: James R. Hines, Jr. (editor), International Taxation and Multinational Activity, University of Chicago Press, Besley, Timothy and Harvey Rosen, 1998, Vertical Externalities in Tax Setting: Evidence from Gasoline and Cigarettes, Journal of Public Economics 70(3), Brueckner, Jan, 2003, Strategic Interaction Among Governments: An Overview of Empirical Studies, International Régional Science Review 26, April, Buettner Thiess, Tax base effects and fiscal externalities of local capital taxation: evidence from a panel of German jurisdictions, Journal of Urban Economics 54, Devereux, Michael, Ben Lockwood, and Michela Redoano, 2002, Do Countries Compete over Corporate Tax Rates?, CEPR Discussion Paper No Esteller-Moré, Á., and A. Solé-Ollé. 2001, Vertical Income Tax Externalities and Fiscal Interdependence: Evidence from the U.S., Régional Science and Urban Economics 31, Goodspeed, Timothy, 2000, Tax Structure in a Federation, Journal of Public Economics 75, Goodspeed, Timothy, 2002, Tax Competition and Tax Structure in Open Federal Economies: Evidence from OECD Countries with Implications for the European Union, European Economic Review 46(2), Hayashi, M., and Robin Boadway, 2001, An Empirical Analysis of Intergovernmental Tax Interaction: The Case of Business Income Taxes in Canada, Canadian Journal of Economics, forthcoming. Hines, James Tax Policy and the Activities of Multinational Corporations. Leprince Matthieu and Alain Guengant, 2001, Interactions fiscales verticales et réaction des communes à la coopération intercommunale, Revue Economique 53(3), Weiner, Joann Martens, 1994, Company Taxation for the European Community : How Subnational Tax Variation Affects Business Investment in the U.S. and Canada. Ph. D. Dissertation, Harvard University. 14