Leseprobe
Contents
1 Introduction
2 Theoretical Considerations
2.1 The Model
2.2 German Reunification
3 Institutional Background and Data Description
3.1 German Federal System
3.2 The Data
4 Empirical Investigation
4.1 Spatial Weighting Matrix
4.2 Empirical Approach
4.3 Results
5 Summary and Conclusion
6 Appendix
A Sources
B Table 2 (complete)
References
1 Introduction
Since the beginning of the 1970’s, closer economic integration has continuously aug- mented the importance of strategic interaction amongst governments in the field of public economics. One of the predominant subjects with regards to inter-governmental strategic interaction is the analysis of tax competition which manifests in competition for a scarce and mobile factor — capital.1 Whilst the early approaches to analyze the effects of local capital taxation as, for instance, by Hamilton (1975), rather promoted beneficial effects of inter-governmental competition, the contrary is true for the more recent works — also known as the new view — which emphasize the distortive effects of capital taxation and the resulting suboptimal provision of public goods (Zodrow and Mieszkowski 1986, p. 309f). Thereby, the latter theoretical works build on a model developed by Mieszkowski (1972) which was further elaborated and extended to allow for inter-governmental competition by Zodrow and Mieszkowski (1986). Henceforth, many renowned economists such as Wilson (1986), Hoyt (1991), Bucovetsky (1991) and Kanbur and Keen (1993), who build on the basic model of Zodrow and Mieszkowski (1986), have devoted substantial theoretical work to examine various aspects of tax competition. These aspects ranged from welfare implications to the effect of size differ- entials between competitors as well as differences between horizontal and vertical tax competition.2
Following the classification of Brueckner (2003), models of strategic interaction amongst governments can be differentiated in resource-flow models on the one hand and spillover models on the other, where both use a similar game-theoretic framework — a simultaneous Nash game. In the former, the interaction originates from a common resource, for example capital or labor, which is supposed to be mobile between multiple local governments and whose allocation decision is influenced by political instruments such as tax rates (Brueckner 2003, pp. 175f). Obviously, tax competition models belong to this type of strategic interaction. In contrast to the resource-flow models, the interaction in spillover models arises from externalities which can occur, for example, from public goods that exert positive or negative spillovers across regional borders.
One of the most famous models belonging to this category of strategic interaction is the ”Yardstick Competition” model developed by Besley and Case (1995).3
Despite its political importance, the number of empirical attempts to identify the existence and quantify the intensity of tax competition is still small compared to its theoretical counterpart (Buettner 2001, p. 216). The early empirical approaches to identify tax coordination focused on the international level. Yet, the international scope entailed several difficulties as to find a large enough set of countries which pro- vides information for a sufficiently long time.4 Furthermore, various differences between countries as well as the existence of borders caused — and still cause today — substan- tial transaction costs, thereby undermining the factor mobility which is an essential requirement for resource-flow models. Besides, nations feature different definitions of their tax bases and taxable incomes rendering a comparison problematic. As a con- sequence, the more recent empirical literature on tax competition moved away from the international perspective and focused on the analysis of federal states that feature local tax setting autonomy.5 Such a framework provides a larger database compared to the international perspective and features further advantageous characteristics such as short distances between competing local authorities, similar cultures and the same language and currency.
The purpose of this paper is to analyze the presence and, in particular, the economic relevance of local tax competition in Germany. To this end, I rely on a large panel of communities in two major German states Bayern and Niedersachsen that covers the period from 1982 to 2007.
The German federal system exhibits numerous characteristics that make an analysis of tax competition in its context very promising. First and foremost, this is the close resemblance of the local taxation of business income to the rationale of source-based capital taxation which is widely promoted in literature on international and local tax competition (Buettner 2001, p. 216). In particular, communities are assigned the competence to decide upon the collection rate for business earnings and property.6 I will provide further convenient attributes within the course of this paper.
Apart from analyzing strategic interaction in the usual manner, the present dataset allows to investigate the impact of a historical shock, namely the German reunification, on both the tax rates and the intensity of strategic interaction. To derive a prediction for the evolution of tax rates under the special circumstances of reunification, I rely on the basic model of tax competition as introduced by Zodrow and Mieszkowski (1986) and Wilson (1986), whereby — for the sake of simplification — assuming that the effects of reunification can be reduced to an increase in the number of (asymmetric) competitors for communities in close proximity to the border. Indeed, the model only suffices to make some a priori predictions regarding the reaction of tax rates but remains silent about the evolution of strategic interaction. For this reason, the present analysis is explorative regarding the impact of reunification on strategic interaction.
In the course of the empirical investigation, I perform a regression of the business tax rate on several exogenous determinants including revenue and expenditure as well as demographic variables. My objective is twofold: to identify a spatial and/or temporal pattern in the observed tax rates and to verify the existence of strategic interaction. Regarding the first, I am interested to find out whether the empirical results confirm what basic theory predicts, namely a downward trend of tax rates in those communities which face more intense competition for capital due to reunification. Concerning the second, my aim is not only to identify strategic interaction, but also to assess its economic relevance and possible changes after 1989.
The remainder of this paper is structured as follows. Section 2 introduces the standard model of tax competition and provides a discussion of possible reunification effects regarding tax rates and strategic interaction. Section 3 will describe the German institutional background and the dataset. Section 4 turns to the empirical investigation before section 5 summarizes the results and concludes.
2 Theoretical Considerations
This section begins with the introduction of the basic model of tax competition which constitutes the foundation of my empirical analysis. I emphasize the main assump- tions and implications of the model before applying it to analyze the effects of the German reunification on tax rates. At last, I provide a few ad hoc scenarios to outline potential reactions regarding the intensity of strategic interaction due to the shock of reunification, something the tax competition model remains silent about.
2.1 The Model
The standard model of tax competition is a two-factor model using a mobile factor capital K and an immobile factor labor L to produce a certain private good by em- ( ) ( ) Ki ploying a neoclassical production function: [Abbildung in dieser Leseprobe nicht enthalten] with ki being the capital-labor ratio in the ith jurisdiction and fi (ki) the per capita output. From neoclassical assumptions, it follows that f′′i (ki) < 0 < f′i (ki). Jurisdictions are assumed to be symmetric in every aspect and, particularly, to pos-sess an identical initial endowment of capital and labor. The overall supply of capital in the economy is fixed, but it is perfectly mobile between local governments.7 Speak-ing in terms of transaction costs, the previous assumption is equivalent to supposing transaction costs being prohibitively high for moving outside the economy and absent for moving between local governments within an economy. Moreover, both factors are assumed to be perfectly mobile within each local jurisdiction (Wilson 1986, p. 298). In the strategic layout of tax competition, each local government is large in economic terms, which implies that each jurisdiction’s tax choice alters the optimal tax decision of all other jurisdictions.8
The basic model further assumes a benevolent government which aims at maximizing the income of a representative consumer. As the endowment with labor is fix and the production functions identical, this income is solely affected by the amount of capital being available in a particular jurisdiction. In order to impinge the capital availability, a local government is supposed to have only a restricted number of tax instruments at its disposal and, in particular, uses a source-based tax on capital.9 For the sake of simplification, it is assumed that local governments use the hereby raised tax revenues to provide a public good at constant marginal costs. One might notice that the term ”public good” is not quite accurate in this context because at its most basic definition non-rivalry and non-excludability would assure the costs of providing a public good not to depend on the number of consumers and thus not to increase at constant marginal costs. In the specific case, where the latter are supposed, an interpretation as a ”publicly provided private good” or a ”public good with private characteristics” rather takes the biscuit (Brueckner 2003, p. 180).
As product and factor markets are competitive and capital is perfectly mobile, the latter moves in response to tax differentials between local governments to equalize the net return to capital, whereby[Abbildung in dieser Leseprobe nicht enthalten]. This is formalized by the arbitrage dti dtj condition: f′ (ki) − ti = R (with R being the economy-wide net return to capital and ti the capital tax rate in the ith jurisdiction) and the capital market clearing condition: [Abbildung in dieser Leseprobe nicht enthalten] which, taken together, determine the capital allocation and tax rates in equilibrium. The capital market clearing condition thereby assures that all capital is employed.10
By imposing the symmetry assumption, it can be easily seen that the equilibrium outcome is characterized by a uniform tax rate and an identical capital-labor ratio in all jurisdictions.
The capital flow which results from tax differentials between local governments causes an externality affecting other local authorities in the economy either in a positive — in case, the government causes a capital outflow by increasing its tax rate — or in a negative way — in case, the same government lowers its tax rate causing a capital inflow and hence outflow from the neighboring governments. The former (positive) externality is also known as tax base flight whereas the latter is referred to as tax exportation. In tax competition models, these externalities are usually subsumed under the term fiscal externality. Against this backdrop, the economic consequences of tax competition are straightforward: each local government fails to internalize the positive externality of an increase in its own tax rate and thus finds it optimal to set an equilibrium tax rate which is lower than the efficient level (Keen and Kotsogiannis 2002, p. 365). More specifically, in reaction to the perceived elasticity of capital supply — which amplifies the marginal cost of raising tax rates — local governments set too low a tax rate in order to preserve their tax bases. Eventually, this leads to a ”race to the bottom” over time. As a result, the public good is underprovided compared to the efficient level as predicted by the Samuelson rule for the optimal level of public good provision.11
At this point, it is expedient to recall the aforementioned distinction between hor- izontal and vertical tax competition. The external effects as described in the pre- ceding paragraph correspond to a model of horizontal tax competition, i.e. the in- teraction takes place among authorities on the same governmental level. The effects turn out to be exactly opposite for vertical tax competition where competition takes place between different governmental levels taxing the same tax base. In this case, strategic interaction arises because each local government’s tax decision alters the tax revenues of all competitors and not the capital allocation across jurisdictions (Wilson and Wildasin 2004, pp. 1066f). Furthermore, it is assumed that the federal government knows about the local government’s budget constraint but not vice versa which implies an inefficiency arising from a ”bottom-up” externality. The economic consequences in this situation are less intuitive than in the horizontal one: in contrast to the federal gov- ernment, a local authority does not internalize the external effect its own tax decision imposes on the federal level by means of altering the federal government’s tax revenues.
More specifically, an increase in the tax rate at the local level imposes an external cost to the federal level through a degenerated tax base and hence truncated tax revenues. As a result, the local authority sets an excessively high tax rate in equilibrium. For exemplification of this suboptimal behavior, suppose a situation where a community (as being the local authority in Germany) increases its tax rate to generate additional revenues of e 1. Correspondingly, the tax revenues at the federal level decrease by the same amount. By assuming that the federal revenues are equally spent among the lower-level governments, each of the latter faces a reduction in its apportionment by e 1/n (where n is the number of communities), leaving the community which raised its taxes with a surplus of [Abbildung in dieser Leseprobe nicht enthalten]. Therefore, it is optimal for each local government n to set too high a tax rate at the expense of all other local jurisdictions.12
In short, the standard model of tax competition has revealed the following implications for the equilibrium tax rates and capital allocation: First, all local governments set equal tax rates, thereby yielding capital to allocate evenly between them and second, all tax rates are set too low compared to the efficient level as proposed by the Samuelson rule, hence leading to a suboptimal provision of public goods.
2.2 German Reunification
Within the previously introduced standard model of tax competition, it is possible to hypothetically analyze the effects of German reunification on the equilibrium tax rates and capital allocation. The shock of reunification can thereby be curtailed to an increase in the number of competitors for communities located in close proximity to the former Inner-German border. Against this background, my first objective is to examine the effect of an increase in n, the number of communities, on the equilibrium tax rates. Considering the ample disparities between East and West Germany with regards to fiscal capacity and infrastructure at that time, it is reasonable to withdraw the assumption of identical communities in order to render the hypothetical analysis of an increase in n more realistic.
The first attempts to allow for asymmetric competition within the basic model were made by Bucovetsky (1991) and further elaborated by Wilson (1991). Supposing that competitors differ only with respect to their population size, their prediction is straightforward: a large government perceives a less elastic capital supply than a small government and hence sets a higher tax rate in equilibrium. This induces a relatively higher capital-labor ratio in the smaller government. In fact, it is possible to show that — for a large enough size differential between competitors — the smaller government can be even better off in the presence of tax competition than in its absence.13 However, for my current purpose it suffices to concentrate on the main result of asymmetric tax competition in the basic model which states that the basic prediction of the symmetric model, namely an underprovision of public goods due to horizontal externalities, carries over to an asymmetric framework (Bruelhart and Jametti 2006, p. 2039).
For simplification and to render the results unambiguous, I suppose that commu- nities in East Germany do not contribute any additional capital to the total amount of capital in the economy, i.e. n increases but K stays constant. I further assume eastern communities to be smaller with respect to population size. In resemblance to the standard model, the new equilibrium allocation is determined by both the arbitrage condition and the capital market clearing condition. However, because of an infinitely large marginal product of capital in East German communities, the arbitrage condition leads to a relocation of capital to the east until the net return on capital is equalized again. Apparently, from a westerly perspective the reunification causes an increase in the individually perceived elasticity of capital supply and hence induces a downward pressure on capital tax rates.14 Due to the supposed absence of transaction costs, all communities are equally affected by an increase in n.
These results, however, do not indicate a certain reaction concerning the intensity of strategic interaction. In fact, the model does not provide any implications regarding changes in strategic interaction, therefore making it impossible to tell whether the local governments would react by being more or rather less sensitive to their neighbor’s tax decisions if the number of competitors rises. In this, the standard model does not differ from other models of tax competition which focus on changes in tax rates and factor allocation, but neglect to analyze the determinants and the degree of strategic interaction between competitors. It is out of the scope of this paper and hence left for further research to develop a model which takes into consideration the intensity of strategic interaction. At this point, I only want to briefly depict feasible scenarios regarding the effects of German reunification on the intensity of strategic interaction. Before doing so, it is, yet, sensible to make some fundamental considerations with regards to premises that justify a change in strategic interaction in response to an increase in the number of competitors. Strategic interaction as such measures the degree to which the behavior of a competitor influences the tax decision of a local government. Thus, in order to induce a change in strategic interaction, reunification has to alter the local governments’ tax setting behavior. The following scenarios do not claim to be comprehensive and their implications regarding the effects on strategic interaction are certainly not deterministic. Instead, they are provided to give the interested reader a starting point for further reasoning.
1. Scenario
In the simplest possible case, the western communities do not perceive the eastern neighbors as serious competitors for capital. One can justify this by arguing that local governments might consider the discrepancy between West and East Germany regarding e.g. infrastructure to be too large as for a firm to think of moving their business across the border. In such a case, the strategic interaction is simply unaffected by the border removal.
2. Scenario
The communities in the west do apprehend a measurable threat emerging from the east, but do not take it too seriously and consider the east to be such a bad substitute for allocation that they hardly take into account the tax rates of eastern communities when deciding upon their own tax rates. In other words: for a firm choosing to allocate in East instead of West Germany, it takes more than a slight difference in collection rates. In this scenario, the strategic interaction might be supposed to rise, but the increase is likely to be disproportionately lower than the number of additional eastern competitors would indicate.
3. Scenario
Western communities do not differentiate whether competition arises from western or eastern neighbors, therefore taking every competitor equally serious. A feasible reaction in this case would be a rise in strategic interaction proportionately to the increase in the number of competitors.
4. Scenario
Due to vertical payments and subsidies for firms that locate in East Germany, western communities perceive the new competition as even more threatening than the competition they already had to face before reunification. In such a scenario, one would expect the influence exerted by eastern communities’ tax rates on the tax decision of western communities to be disproportionately larger. Yet, it remains uncertain what happens to the total degree of strategic interaction: western communities might have the tendency to divert their focus from compe- tition amongst western to eastern communities, considering eastern tax setting behavior to a larger and western competition to a lower extent.
5. Scenario
Finally and in continuation of the previous scenario, western communities might collude to form one large community which decreases the perceived elasticity of capital supply. The tax rates in western communities would then be perfectly correlated and strategic interaction at its maximum amongst communities in the west but zero with regards to communities in the east.
3 Institutional Background and Data Description
3.1 German Federal System
The German federal system exhibits a number of advantageous characteristics rendering an empirical analysis of tax competition in its context very promising.
Tax Principle: Taxation of capital income in Germany is very close to the rationale of source-based capital taxation. This type of taxation is very prominent in the theoretical literature on tax competition not least because of the familiar dif- ficulties with regards to levying taxes on foreign-source income (Wilson 1991, p. 425). Moreover, the allocation of capital across communities does not influ- ence the communities’ tax income in a setting with residence-based taxation. In such a case, a tax on capital income would be similar to a non-distorting head tax, rendering the analysis obsolete. However, under a system of source-based taxation, communities use the distorting tax rate on the mobile factor capital and by doing so, induce an in- or outflow of capital which causes strategic behavior.
Fiscal Autonomy: The rights and obligations of local authorities are laid down in Article 28 of the German constitution. It restricts the tax instruments being at the communities’ discretion to the so-called Hebesatzrecht which constitutes their right to determine the collection rate for business income and property. In addition to this, the revenues from taxation of business income amount to a substantial fraction of total tax revenues on the community level. In particular, it generates approximately one third of total tax revenues at the community level and, with a share of approximately 10 percent, contributes substantially to the total revenues of an average community.15
Large Number of Communities: Current figures reveal that the number of com- munities in Germany amounts to 11,993.16 Bayern and Niedersachsen cover more than 25 percent of the total number of German communities. The great major- ity of these communities are obliged to similar rights and duties by the German constitution.17 Yet, they show various differences regarding population size and structure as well as fiscal capacity. In light of the estimation, I appreciate the cross-sectional asymmetries between the municipalities because they allow for a better identification of the partial effects of the incorporated variables.
Infrastructure: Germany provides a comprehensive and well-developed public infrastructure. In particular, the quality of the street network and the expansion of the communication system both class among high level. With regards to the the- oretical model, these attributes are advantageous as they keep transaction costs of moving business activities between several communities low and hence amplify capital mobility.18
Despite several advantageous aspects, the German setting still exhibits some char- acteristics which appear to be precarious. My first concern regards the German fiscal equalization system which has shown to change the tax setting behavior of communi- ties.19 In particular, the availability of alternative funds decreases the extent to which a community’s budget depends on locally raised revenues. Eventually, the marginal cost of taxation turns out to be lower if alternative funds are available as the impact of a diminishing tax base on the community’s total budget decreases. Whereas tax rates are thus expected to exhibit a general upward trend in response to an increasing pertinence of the fiscal equalization scheme, it is ambiguous what happens to the inten- sity of strategic interaction. With regards to the subsequent estimation, this issue is, however, only a minor threat as the fiscal equalization system affects all communities in a similar manner.
My second concern refers to a certain attribute of the federal structure in Ger- many. In the course of the local government reform in the late 1960’s and early 1970’s, most German states introduced a new administrative unit ranging between the commu- nity and the state level. These administrative units are formed by several individual communities which have to meet certain requirements — for instance, with regards to population size and geographic proximity. In Niedersachsen these institutions are called Samtgemeinden, whereas in Bayern they are referred to as Verwaltungsgemein- schaften. Although these institutions exhibit differences in the precise scope of their tasks, their underlying conception is the same, namely to increase the efficiency of public good provision at the community level by centralizing parts of their local ad- ministration (Bogner 2007, p. 247 and pp. 251-258).
[...]
1 Albeit capital is most notably referred to as the subject of competition, the rationale of tax competition can be applied in a similar manner to other factors such as labor, firms or shoppers, contingent on the specific tax base.(Wilson and Wildasin 2004, p. 1067)
2 See Brueckner (2003), Wilson (1999), Wilson and Wildasin (2004) or Gaboury and Vaillancourt (2003) for a review of the literature on tax competition.
3 Besley and Case (1995) elaborate a theoretical model that describes the simultaneous tax-setting behavior of jurisdictions in a political economy approach. They provide empirical evidence for their theory that tax-setting and vote-seeking are closely intertwined by means of yardstick competition, i.e. local voters observe the outcome in neighboring jurisdictions, thus being able to judge their local politicians’ actions.
4 In order to analyze dynamic effects of strategic interaction, it is necessary to observe the tax setting behavior of governments for several periods. Moreover, it enables the use of lagged values of local characteristics within the empirical investigation, thereby taking into account possible reaction lags in the tax decision due to rules and regulations, which prohibit immediate changes in tax rates, and long-winded bureaucracy.
5 See Case, Rosen, and Hines (1989) for an early empirical investigation of fiscal interaction between states in the USA. Brett and Pinkse (2000) as well as Hayashi and Boadway (2001) provide an empirical analysis of business tax rates in Canada; Buettner (2001) investigates tax competition in the German state Baden-Wuerttemberg; Bruelhart and Jametti (2006) compare horizontal and vertical tax externalities in Switzerland.
6 Strictly speaking, the terms collection rate on business earnings and business tax rate are not equivalent. The collection rate is a factor which is multiplied with the basic federal rate on business income (Buettner 2001, p. 224). The basic federal rate is a uniform tax rate for all municipalities that is determined by the federal government. Until 2008, the basic federal rate used to be staggered in 5 steps for business partnerships and sole proprietorships, ranging from 1 percent for business income below e 12,000 to 5 percent for business income above e 48,000. At the same time, corporations faced a uniform rate of 5 percent, regardless of the size of taxable profits. For illustration: A firm with taxable profits of e 100,000 located in a community with a collection rate of 350 percent faces a tax rate of 0.05*3.5=17.5 percent. In the subsequent paper, I will not further distinguish between the two terms and treat the collection rate on business earnings of each community as the local business tax rate.
7 Mobility and immobility of capital and labor can alternatively be grasped as perfectly elastic and perfectly inelastic supply of the respective factors.
8 See Hindriks and Myles (2006), Intermediate Public Economics, pp. 569-576
9 In the hypothetical case where a lump-sum tax was available, a benevolent government would always prefer to use this kind of taxation because it circumvents a distortion in consumption and production decisions. Therefore, a lump-sum tax is oftentimes referred to as ”first-best”. An equivalent result is obtained by raising a tax on the immobile factor labor. As the immobile factor cannot evade, the usual distortions are avoided and taxation is ”costless” in terms of a deadweight loss. Bucovetsky and Wilson (1991) show that in case of a large number of identical competitors a government would not even choose to levy a tax on capital if labor supply decisions were distorted. However, if a local government has the power to influence the equilibrium net return to capital, they predicate that governments may decide on taxing both.
10 Note that∑ni=1 ki =nki holdsbecauseofthesymmetryassumption.
11 See Wilson (1986) and Zodrow and Mieszkowski (1986) who also provide the formal derivations. Note that the results largely depend on the hypothesized objective function of the government. Keen and Kotsogiannis (2003) and Bruelhart and Jametti (2007), for instance, show that tax competition can even be welfare-enhancing in cases where this assumption is exchanged for a ”Leviathan” govern- ment that acts as a monopolist which seeks to maximize tax revenues. In such a case, tax competition can impose disciplining effects on the politicians’ behavior. Similar implications are proposed by Sato (2003) who develops a model where tax competition can act as a constraint to rent-seeking private interest groups.
12 See Keen and Kotsogiannis (2002, p. 365), who also provide the formal results of this interpreta- tion.
13 See Bucovetsky and Wilson (1991) for the formal derivation of this result.
14 In a similar manner, Hoyt (1991), Wilson (1991) and Bucovetsky (1991) demonstrate that even under the assumption of an increase in identical communities the underprovision of public goods is amplified.
15 Own computations based on pooled cross-sectional data on total revenues and tax revenues for Niedersachsen and Bayern.
16 See Statistical Yearbook 2010 for the Federal Republic of Germany, table 2.3 published by the Office (2010). Figures include inhabited district-free areas.
17 Some minor exceptions exist for district-free areas which belong to either the state they are situated in or the federal government. Contemporary figures by the Federal Statistical Office (2010) reveal a total number of 77 district-free areas whereby only three of them are inhabited. Bayern reports 43 district-free areas, all of which are unoccupied and Niedersachsen reports 25 district-free areas, with two being occupied.
18 This contrasts other empirical studies which focus on the international level where the actual capital mobility is oftentimes fairly low due to high transaction costs. The latter can emerge from long distances, different currencies and cultures as well as further border effects.
19 See Buettner (2006) for a detailed analysis. By employing a panel of communities in BadenWuerttemberg, he shows that communities set significantly higher business tax rates under a system of fiscal equalization compared to a situation where such a system is absent.