Elsevier

Journal of Public Economics

Volume 90, Issue 3, February 2006, Pages 429-458
Journal of Public Economics

Income segregation and local progressive taxation: Empirical evidence from Switzerland

https://doi.org/10.1016/j.jpubeco.2005.09.003 Get rights and content

Abstract

This study investigates spatial income segregation in fiscally decentralized urban areas. The theoretical part proposes the progressivity of local income taxes as a new theoretical explanation for income segregation. The empirical part studies how income tax differentials across municipalities affect the households' location decisions. I use data from the Swiss metropolitan area of Basel that contains tax information on all moving households in 1997. The location choice of the households is investigated within the framework of the random utility maximization model. Different econometric specifications of the error term structure, such as conditional logit, nested logit and multinomial probit, are compared. The empirical results show that rich households are significantly and substantially more likely to move to low-tax municipalities than poor households. This result holds after controlling for alternative explanations of segregation. Social interactions and distance from the central business district are established as other major factors for income segregation. Households in general tend to choose locations close to other households like themselves.

Introduction

“To secure an efficient outcome, the provision of public services should be determined and paid for by those who benefit”, as Musgrave (Buchanan and Musgrave, 1999, p. 156) pointedly remarks. Oates (1972) also argued that local units deciding upon public programs are more likely to trade off costs against benefits if these programs are financed by local taxes.

While the virtues of decentralized financial responsibility are uncontested, the resulting tax differentials are highly disputed. Tax differentials are often seen as consequences of different preferences for the locally provided public goods. However, different tax rates can also be the result of different economic resources of the local population, since rich local jurisdictions can raise the same revenue with lower tax rates than poor ones. While the effect of the tax base on tax rates is trivial, the opposite effect is less evident. This paper addresses the question whether tax differentials across local jurisdictions is not just the consequence, but also the cause of differences in local average income.

The theoretical part of this paper proposes the progressivity of a local income tax as a new theoretical explanation for income segregation of the population. The empirical part studies the community choice of households in Switzerland. Swiss metropolitan areas are a laboratory for federal systems as they are divided into a multitude of municipalities with extensive political and fiscal autonomy. Moreover, the main local tax is on income rather than on property. The estimated multinomial response models show that rich households are significantly and substantially more likely to move to low-tax municipalities than poor households. This result holds after controlling for alternative explanations of segregation. Social interactions and distance from the central business district are established as other major factors for income segregation.

The theoretical literature on the local provision of local public goods goes back to Tiebout (1956). Tiebout showed that fiscal decentralization leads to an efficient provision of local public goods because people with similar preferences would settle in particular municipalities and vote for their desired level of public goods provision. Tiebout's result rests on the assumption that households have equal incomes. The location of households and the local provision of public goods when households differ in incomes were studied by Ellickson (1971), Westhoff (1977) and in the literature surveyed in Ross and Yinger (1999).

The segregation hypothesis is one of the central propositions in multi-community models in the tradition of Tiebout. Endogenous segregation means that different people choose different locations in equilibrium. While the Tiebout model focuses on preference heterogeneity, Ellickson and Westhoff turned their attention to income as the main dimension of difference. Several mechanisms have been proposed to explain why rich households make different choices than poor households (see Ross and Yinger, 1999, for property tax models, and Schmidheiny, 2002, for income tax models). The nature of the local public good, ranging from a monetary transfer to a non-substitutable pure public good, induces a self-sorting of the population when rich households esteem public goods relatively more than poor households. Another mechanism draws on the income elasticity of housing: If housing expenditures become relatively less important with increasing income, rich households are less concerned about high housing prices than poor households.

The segregation mechanism in this paper builds on the empirical fact that most income tax schedules are progressive and that local jurisdictions can often only set the tax level within a given federal tax schedule. The high priority of tax rates in rich households' decisions is explained by the progressivity of the tax schedule.

The segregation hypothesis of Tiebout type models has been challenged by a series of empirical studies.1 A first strand of research investigates the equilibrium predictions of multi-community models using data on aggregate community characteristics. Epple and Sieg (1999) and Epple et al. (2001) develop a strategy for estimating the household preference parameters in an equilibrium model where the local income distribution and local policy variables are simultaneously determined. They show that the differing income quantiles across 92 local jurisdictions in the Boston area are well explained by the model predictions. Feld and Kirchgässner (2001) regress the share of various income classes in Swiss cantons and main cities on income tax rates. They find a strong negative relationship between the tax rate and the share of rich households. However, their treatment of the generic endogeneity of tax rates by instrumental variables from mainly lagged observations may not be sufficient to solve the endogeneity problem, as the general equilibrium of tax rates and income segregation is most likely a long-run phenomenon. Rhode and Strumpf (2003) assess the importance of the segregation mechanism in Tiebout type models from a historical perspective. They collected an impressive data set with measures of heterogeneity in the population over a period of 140 years. Given that the costs of moving dramatically declined during this time, multi-community models predict that the population within local units should have become more homogeneous while the differences across local units should have aggravated. They conclude that their data do not support the model predictions on a national scale. For metropolitan areas, however, the observed pattern does not contradict the segregation hypothesis.

The second empirical approach – also used in this paper – directly targets the location choice of individual households using a multinomial response framework. This approach circumvents the endogeneity problem because, from the perspective of a single household, the community characteristics can be taken as given. Friedman (1981) used a conditional logit model to study the location choice of 682 households among nine residential areas in the San Francisco area. Nechyba and Strauss (1998) use the same model to study the choice of over 22,000 households among six school districts in the suburbs of Philadelphia. Both studies show that public expenditures are an important locational factor. The segregation hypothesis needs explicit consideration as household specific variables are not identified in linear conditional logit models (see Section 5.1). In need of a variable that depends on both household and community characteristics, Nechyba and Strauss calculate the households' hypothetical consumption of private goods for all school districts. This variable depends on after-tax local housing prices assuming that households consume the same amount of housing in all school districts. They therefore implicitly assume that the price elasticity of housing is zero.2

Bayer et al. (2004) attempt a combination of the two empirical approaches. Following Berry et al. (1995), they first estimate the households' choice of a neighborhood, using neighborhood fixed effects and a multitude of interaction effects between household and neighborhood characteristics. In a second step they explain the neighborhood fixed effects by neighborhood characteristics using instrumental variables which make use of an explicit general equilibrium model. The predictions of the estimated model therefore adequately take into consideration the (long-run) adjustment of the endogenous aggregate neighborhood characteristics.

This study follows Friedman (1981) and Nechyba and Strauss (1998) but shifts the focus to assessing the (income) segregation hypothesis. The general locational attractiveness of a community is considered in community specific constants. This is equivalent to the first stage estimation in Bayer et al. (2004). The specification of the model is drawn on an explicit theoretical multi-community model.

The paper is organized as follows: Section 2 describes fiscal decentralization in Switzerland. A theoretical model of location choice based on progressive income taxation is proposed in Section 3. Alternative explanations of income segregation are discussed in Section 4. The econometric model is outlined in Section 5, while Section 6 describes the data. The empirical results are presented in Section 7. Section 8 draws conclusions.

Section snippets

Fiscal decentralization and progressive taxation in Switzerland (and elsewhere)

Switzerland is an exemplary federal fiscal system. The Swiss federation comprises 26 states, the so-called cantons. The cantons are divided into roughly 3000 municipalities of varying size and population. All three state levels finance their expenditures essentially by their own taxes and fees. The total tax revenue of all three levels was 77 billion CHF in 1997, of which 45% is imposed by the federation, 32% by the cantons and 23% by the municipalities.3

A model of location choice and local progressive income taxation

The theoretical model describes a metropolitan area with a fixed number J of distinct local jurisdictions, called communities. The political borders of the communities are the outcome of a historical process and thus taken as given. The area is populated by a continuum of heterogeneous households, which differ in incomes. Income is distributed according to a distribution function f(y) > 0 with support [y, ], y > 0,  < ∞. There are three goods in the economy: private consumption b, housing h and a

Alternative explanations for income segregation

The theoretical model in the Previous section relates income segregation solely to institutions (in the case of taxes) and markets (in the case of housing prices). There are other mechanisms which potentially explain income segregation in metropolitan areas. This section discusses two leading alternative explanations which are considered in the empirical implementation.

The econometric model

If spatial income segregation is the result of individual households' decisions it can be studied by looking at the decisions when they are made. Households do not daily decide upon their place of residence. There are specific moments in any individual's life when the decision about where to live becomes urgent. Such moments occur for example when an individual leaves college to start a first job, when a couple decides to move together or when the family size changes. Career opportunities later

Data

The empirical investigation is based on non-public household data from the Tax Administration of the Canton of Basel-Stadt. The data contain information of all households in the city of Basel that moved within the city or from the city to a municipality in the periphery in the year 1997.

The decision maker in the theoretical model is a household.15

Estimating the base model

This section presents the estimates of the base model reported in Table 2. The base model is a direct implementation of the theoretical model in Section 3. In this section, I model the location choice of moving households as a choice among 17 alternative communities: the city of Basel and 16 peripheral communities. Three specifications of the error term structure are reported in the respectively labelled columns: the conditional logit model, the nested logit model and the multinomial probit

Conclusions

This study investigates spatial segregation of the population in fiscally decentralized urban areas. The theoretical part proposes the progressivity of local income taxes as a new theoretical explanation for income segregation.

The empirical part studies how income tax differentials across municipalities affect the location decisions of households in the metropolitan area of Basel. The estimation results show that rich households are substantially and significantly more likely to move to low-tax

Acknowledgments

The discussions with Klaus Neusser, Mike Gerfin, and Veronika Aegerter, greatly improved this paper. I would like to thank my anonymous referees and the editor, Hans-Werner Sinn, for their constructive comments and suggestions. I am grateful to the Tax Administration and the Statistical Office of the Canton of Basel-Stadt for the provision of the household data and to Wüest und Partner for the provision of the housing price data, my thanks to Martin Sandtner and Jürg Mosimann in particular.

References (53)

  • Patrick Bayer et al.

    An Equilibrium Model of Sorting in an Urban Housing Market, NBER Working paper 10865

    (2004)
  • Roland Bénabou

    Workings of a city: location, education, and production

    Quarterly Journal of Economics

    (1993)
  • Roland Bénabou

    Heterogeneity, stratification, and growth: macroeconomic implications of community structure and school finance

    American Economic Review

    (1996)
  • Roland Bénabou

    Equity and efficiency in human capital investment: the local connection

    Review of Economic Studies

    (1996)
  • Moshe Ben-Akiva et al.

    Discrete Choice Analysis: Theory and Application to Travel Demand

    (1985)
  • Abraham Berman et al.

    Nonnegative Matrices in the Mathematical Sciences

    (1994)
  • Steven Berry et al.

    Automobile prices in market equilibrium

    Econometrica

    (1995)
  • Robin W. Boadway et al.

    Canadian Tax Policy, Toronto

    (1980)
  • Denis Bolduc et al.

    The impact of incentives policies on the practice location of doctors: a multinomial probit analysis

    Journal of Labor Economics

    (1996)
  • Denis Bolduc et al.

    Multinomial probit estimation of spatially interdependent choices: an empirical comparison of two new techniques

    International Regional Science Review

    (1997)
  • James M. Buchanan et al.

    Public Finance and Public Choice: Two Contrasting Visions of the State

    (1999)
  • Steven N. Durlauf

    A theory of persistent income inequality

    Journal of Economic Growth

    (1996)
  • Steven N. Durlauf

    Neighborhood feedbacks, endogenous stratification, and income inequality

  • Bryan Ellickson

    Jurisdictional fragmentation and residential choice

    American Economic Review

    (1971)
  • Dennis Epple et al.

    Mobility and redistribution

    Journal of Political Economy

    (1991)
  • Dennis Epple et al.

    Estimating equilibrium models of local jurisdictions

    Journal of Political Economy

    (1999)
  • Cited by (0)

    View full text