Evaluation of centralization eﬃciency
The optimal federal system consists in the most eﬃcient allocation of public goods between local and central levels. As the previous section demonstrates, both these levels provide public goods with ineﬃciences. Local governments face base erosion, while the central government has to deal with a ”one size fits all” problem, as long as preferences are het-erogeneous. It follows that the optimal degree of centralization, noted c∗, allows for the right blend between the adaptability to preferences of local public good provision and the protection against base erosion of central provision.
Our criterion for assessing optimal centralization is the deadweight loss associated with public good provision, similarily to Janeba and Wilson (2011). The deadweight loss for a particular public good is defined as the decline in utility caused by an ineﬃcient provision, compared to a first-best level characterized by the Samuelson rule. Then, c∗ minimizes the total deadweight losses summed across all goods and all jurisdictions.
It is important to note that this criterion disregards other ineﬃciencies present in the model that also impact household utility. For example, the tax equilibrium aﬀects the allocation of capital and the consumption of the private good across jurisdictions. Our paper focuses on the two ineﬃciencies described above, only related to public good provision.
Finally, due to our assumption of quasi-linear utility, there are no income eﬀects in public good demands such that deadweight losses can be approximated by Harberger tri-angle formulas, as developed below.
The optimal degree of centralization
The optimal degree of centralization c∗ is the one minimizing the total deadweight loss function as defined in (35).
First, we note that absent base erosion, there would be no ineﬃciency in local provision i.e. Ll(c) = 0. It follows that the level of c minimizing the loss function would be c∗ = 0, in other words complete decentralization. Instead, with homogenous preferences, there would be no deadweight loss in central provision since the central government would provide an eﬃcient level of public goods to all jurisdictions. We would have Lc = 0 ∀i ∈ [1, n]. Full centralization would then be optimal i.e. c∗ = 1.
We present three main results. First, centralization increases provision of the goods staying at the local level. The literature on fiscal federalism identifies various costs of centralization in terms of local public finances (vertical tax externalities, administrative costs…), however we present an argument that goes against the standard thinking: cen-tralization reduces the price distortion of local provision.
Second, we show that increased base erosion favors a centralized provision of public goods. Indeed higher capital outflows mean increased deadweight loss for the local provi-sion, without aﬀecting central provision. A federal system where some taxable bases react strongly to tax changes, as it seems to be the case for CIT in Europe, benefits from cen-tralization and the optimal allocation of provision is likely to favor the central level. It also means that eﬀorts to reduce base erosion render public good centralization less relevant. In turn, a high heterogeneity of preferences does not necessarily favors decentralized pro-vision, contrary to traditional thinking, because of base erosion. Increased heterogeneity of preferences raises ineﬃciencies both at the central and local levels. More heterogeneity makes the central government increasingly unable to meet local tastes. However, if base erosion is strong, local ineﬃciencies for high taste jurisdictions also rise substantially. Cen-tral provision, although remote from their preferences, protects them against base erosion. Hence, diverging tastes for public goods in a federal system of heterogeneous jurisdictions should not necessarily translate into a lower degree of centralization, if base erosion is high.
Third, we characterize the optimal degree of centralization as the one that yields the best trade-oﬀ between the strength of base erosion and the dispersion of preferences. The fact that centralization mitigates ineﬃciencies at the local level is also what makes a fully central system non optimal. Indeed, as local distortions vanish, the cost of central provision in terms of unadaptability to preferences will always dominate for some degree of central-ization, such that at one point shifting more public goods to the central level is undesirable. However, whether some centralization is optimal depends on heterogeneity in preferences being lower than local tax distortions, so that a fully decentralized system might be optimal.
Tax competition and club formation
Tax cooperation, defined as the common setting of tax rates or ranges of rates, is well-known to face numerous obstacles. First, countries must accept to band together with allies that potentially have diﬀerent preferences and goals than their own. The benefits of cooperation have to outweight the loss of independence. Then, in a world where factor mobility erodes taxable bases, cooperation remains partial if participating is not as attractive as holding out in the rest of the world and winning the tax competition game. The possibility to benefit from cooperation elsewhere while staying out might constrain the size of alliances, or even prevent their formation in the first place if no country is willing to give up what it could potentially gain by remaining an outsider, although there is a benefit in cooperating.
In particular, if there are asymmetries in size or factor endowment, some may prefer the tax competition equilibrium to cooperation although it is Pareto improving. For example, small regions can exploit factor mobility by setting relatively low taxes to attract a lot of base (Wilson, 1991; Bucovetsky, 1991), which can explain the existence of non-cooperating tax havens. Even leaving aside structural asymmetries among competitors, policy coordi-nation itself leads to diﬀerences in tax levels and capital allocations. Diﬃculties to set-up tax agreements such as the Common Consolidated Corporate Tax Base (CCCTB) in the EU illustrate how countries can be relunctant to cooperate if they fear ending up disad-vantaged compared to other competitors. Hence, one may wonder what type of policy coordination may induce countries to join alliances in spite of tax competition, so that full cooperation is reached.
The contribution of this paper consists in including public good spillovers when ana-lyzing the formation of tax cooperation zones. In a tax competition setting, cooperating players who set their rate in common lose tax base to the rest of the world. But if cooperat-ing entails benefiting from spillovers on top of coordinating in taxes, then the net benefits of membership might be high enough to remove the incentives to remain a non-cooperating, low-tax player. Spillovers then take the form of side payments making up for the lost tax base. Provided spillovers are high enough, full cooperation may be reached. We con-nect our theoretical thinking to the financing of pan-European public goods such as single market of goods, security or defence as well as cross-border infrastructures. This analysis provides a rationale for complementing tax cooperation projects such as the CCCTB with common spending exhibiting high spillovers, such as a euro area budget to provide public goods at the European level.
In this paper, we adopt a standard model of tax competition where identical countries compete a` la Nash over a mobile tax base, capital, taxed at the source to finance a public good. Tax competition distorts downwards the amount of tax revenue the government can raise since some capital flies away following a tax increase. We introduce the public good spillovers in the following way: as long as they do not cooperate, countries restrict the access of their public good to their respective household (in that sense these public goods have the feature of publicly provided private goods). Then, we make it possible for some countries to cooperate and benefit from a share of the others’ public goods. We call this group of countries the club. Members can overcome tax competition by coordinating to set taxes and in return they benefit from a share of the others’ public good through cross-border spillovers.
Strong assumptions on functional forms are made to derive closed-form solutions for the tax equilibrium as in Bucovetsky (2009): the production function is quadratic and the utility is linear in the public good. However, the results described in this paper are representative of general mechanisms not driven by these assumptions. Through the inclu-sion of robustness checks, we show that our results are still valid with other more general functional forms.
Our work relates to two strands of literature. First, the tax competition literature.1 Specifically, greater capital mobility raises the cost of the public good since a tax increase causes capital flight, conveying a positive externality on other jurisdictions. Competing governments cannot, without cooperation, achieve an eﬃcient public good provision, which is the seminal result. It can also mean that taxation shifts to less mobile tax bases such as labor, or to the residence principle (Bucovetsky and Wilson, 1991). Moreover, an important branch of this literature investigates tax cooperation, with ambiguous welfare conclusions.2 Konrad and Schjelderup (1999) show that a partial tax harmonization is beneficial not only for cooperating countries but also for countries outside the cooperation provided tax rates are strategic complements.3 Peralta and van Ypersele (2005) present minimum capital tax and tax range as alternative forms of tax coordination policies, and find only the latter to be Pareto-improving. One should also mention matching grants, or side payments, as possible instruments to overcome tax competition (DePeter and Myers, 1994; Wildasin, 1989). Also, Burbidge et al. (1997) incorporate coalition formation in a tax competition model and show that when there are more than two regions, equilibrium can consist of partial cooperation with multiple federations. Our contribution on policy coordination among tax competitors is to consider that countries coordinating in taxes also grant access to others to a share of their own public good. In particular, we do so by studying the implications of including spillovers in terms of tax equilibrium and cooperation size.
The second strand of the literature connected to our work is that of club theory, which provides relevant insights regarding the nature of public goods we are analyzing.4 A pub-lic good is considered pure when fully non-rival (each user of the good has access to its full amount regardless of how many others use it) and non-excludable (it is impossible to prevent any user from accessing it). A club good however is non-rival but excludable to non-member of the club providing it. First identified by Buchanan (1965) and Olson (1965), clubs are created to benefit from economies of scale and to share public goods. Club theory studies congestion costs or member heterogeneity as factors determining the size of unions. In this paper, the strength of tax competition and of spillovers in public good provision ultimately determine the size that the club can reach.
Our model accounts for externalities, spillovers and tax competition. First, we calcu-late equilibrium tax rates under autarky and then when there is capital mobility but no cooperation. Subsequently, we derive the equilibrium under partial cooperation where only a subset of countries is part of the club, and finally in a case of full cooperation where all countries participate. Although we adopt a set-up with symmetric countries, partial policy coordination induces asymmetries with high and low tax regions which are analogous to having size diﬀerences. Next, we show that an increase in the size of the club benefits all, incumbent and non-cooperating countries alike. Then, we analyze the stable club size, defined as the one at which no country prefers to join or leave the club, as a function of spillovers.
We contribute to the literature by presenting a partial tax coordination framework augmented with the possibility of public good spillovers. The standard result according to which outsiders’ gains may be bigger than those of the coordinating subgroup if there is tax competition is reconsidered. We analyze the conditions for a club to emerge, and also for the last remaining country outside of the club to benefit from joining as well bringing the economy to full cooperation. We derive a condition for full cooperation to happen 4 See Sandler and Tschirhart (1980) and Sandler and Tschirhart (1997) for surveys of the literature.
according to which, in spite of tax competition, all countries participate because spillovers are high enough. We provide a numerical example to illustrate our model, as well as ro-bustness checks where our assumptions on functional forms are modified.
The paper is organized as follows. Section 2 presents the model. In Section 3, we derive the tax equilibrium under diﬀerent scenarios. Section 4 studies club formation. Section 5 concludes.
We adopt a model of international tax competition drawing on a set-up derived in Bucovet-sky (2009), itself based on a seminal model of tax competition a` la Zodrow and Mieszkowski (1986). Similar models have been presented in Peralta and van Ypersele (2005) or Keen et al. (2012). In this one-period model, n identical countries, equal in size, compete for capital that moves costlessly wherever the after-tax return is highest.5
Our take on the issue is to study how the standard tax competition framework evolves when a m-sized club of countries emerges among these n tax competitors, with 1 ≤ m ≤ n. The club consists for its members in coordination to set a harmonized tax rate, while in return benefiting from spillovers in public good provision.6 As long as 1 < m < n, some countries remain outside of the club and cooperation remains partial. That aspect connects our analysis to the one of Konrad and Schjelderup (1999). We assume that there can be only one club at the same time, and leave the study of the co-existence of multiple clubs to further research. We also characterize the no cooperation case as a world where m = 1, while full cooperation happens at m = n.
Table of contents :
I Optimal centralization and tax base mobility
2 Related literature
3 Local and central policies
3.1 Model set-up
3.2 Equilibrium levels of public goods
3.3 Capital allocation
4 Optimal centralization
4.1 Impact of centralization on public good provision
4.2 Evaluation of centralization efficiency
4.3 Does base erosion favor centralization? Does heterogeneity?
4.4 The optimal degree of centralization
II Tax competition and club formation
2 Model set-up
3 Tax equilibrium
3.1 No cooperation
3.2 Partial cooperation
3.3 Full cooperation
4 Club formation
4.1 How does the tax equilibrium evolve with m?
4.2 When does a club emerge? Can full cooperation be reached?
4.3 Illustration and robustness checks
III Unemployment insurance union
2 Literature review
3.1 Labor markets
3.4 Nash bargaining
3.5 Monetary policy
3.6 National governments
3.7 Goods and financial trade
4 Design of the European unemployment insurance
4.1 European benefits and taxes
4.2 EUI scenarios
5.1 Euro area calibration
5.2 Model steady-state and second moments
6.1 Impulse response functions : baseline
6.2 Impulse response functions : EUI
IV Euro area unemployment insurance and the ZLB
2 Related literature
3.1 Labor markets
3.6 Market clearing
4.1 Euro area calibration
4.2 Second moments of the model
5.1 Assessing the benefits of the EUBS at the zero lower bound
5.2 Extension: periphery government cut from financial markets