1. Introduction

The move towards greater decentralization of fiscal policy is one of the most important political trends in recent decades (see ^{Garman Haggarg, and Willis, 2001}; and ^{Hankla, 2008}). Among the expected benefits of fiscal decentralization is that local governments can accommodate the heterogeneous preferences of households across jurisdictions. The normative theory of federalism identifies conditions in which a shift from a Pareto efficient but uniform provision of local public goods (LPGs) by a central government towards a Pareto efficient and differentiated provision of LPGs by a system of sub-national governments entails a nationwide gain in the welfare of residents (see ^{Oates 1972}, ¹⁹⁹⁹).

However, critics of fiscal decentralization usually stress that, under interregional spillovers, the decentralized provision of LPGs will not be Pareto efficient. Some of the solutions provided in the literature for solving the externality problem in the provision of local public goods include the following alternatives: ^{Oates (1972}, ¹⁹⁹⁹⁾ identifies conditions in which LPGs with inter-regional spillovers should be fiscally centralized (instead of decentralized).¹ Another solution is to preserve a fiscally decentralized provision of LPGs but a benevo-lent central government could implement Pigouvian intergovernmen-tal transfers to internalize spillovers from LPGs, see ^{Oates (1999)}.

^{Myers (1990)} develops a model of decentralized provision with benevolent governments, free mobility, and inter-regional transfers that sustain Pareto efficient local public goods without the intervention of the central government.² Other applications of this type of model include ^{Silva and Caplan (1997)}, ^{Hoel and Shapiro (2003)}, and ^{Mansoorian and Myers, (1993)}. However, ^{Wellisch (1994)} finds that central government intervention might still be needed if households are imperfectly mobile because the decentralized provision would be inefficient.

Inefficiencies in the decentralized provision of LPGs can also arise as a result of coordination failures among sub-national governments. A straightforward application of the Folk Theorem suggests that, in an inter-temporal setting, a decentralized Pareto efficient provision could be achieved if local governments are patient enough, and the cost of deviating from Pareto improving strategies is high enough.³

However, voting and legislative models cast doubts on whether the solutions to the problem of spillovers that depend on benevolent governments are compatible with weakly dominant strategies of candidates in elections, and of public officials in the legislature. For instance, the analyses of ^{Lockwood (2002)} and ^{Besley and Coate (2003)} suggest that the sub-national provision of LPGs with and with-out inter-regional spillovers is not necessarily Pareto efficient if policy choices are determined by a legislature.^4,5

Recent analysis on political economy has focused on interest groups and fiscal federalism: ^{Lockwood (2008)} studies whether the decentralization theorem holds (see ^{Oates, 1972}) when collective choices are made by majority rule and lobbying. ^{Bordignon, Colombo and Galmarini (2008)} characterize conditions in which lobbying induces a decentralized provision with spillovers that is Pareto efficient when there is some optimal weight attached to social welfare (as opposed to money) but inefficient if politicians become too greedy. ^{Guriev, Yakovlev and Zhuravskaya (2010)} find that different measures of performance of firms improve for an economy with multi-state lobbies in comparison with local lobbies.

In this paper, I develop a political economy model with interest groups in a system of sub-national governments in which the provision of local public goods with and without spillovers is Pareto efficient. In my analysis ideological parties face a tradeoff between campaign contributions and the design of local public spending. In my model, families can contribute to parties participating in the local elections of different jurisdictions in exchange for a more favorable policy.⁶ Parties realize that they can attract campaign contributions from voters of different regions and, therefore, parties have incentives to recognize the full distribution of benefits from LPGs with regional spillovers. As a result, the decentralized provision of LPGs is Pareto efficient.

My paper is different from the previous literature, since the efficiency in the provision of LPGs does not require benevolent governments, perfect mobility of families, or the intervention of the central government. My analysis is also different from the literature on interest groups mentioned above, since I focus on the process of preference aggregation associated with the provision of LPGs and campaign contributions.⁷

My paper is also related to the literature on how different structures of governments can (or cannot) accommodate an inter-regional heterogeneous distribution of preferences into policy. In his seminal work, ^{Oates (1972)} considers that the decentralized provision of local public goods maximizes the society’s gains associated with matching the heterogeneous preferences of individuals with the government’s policies. However, recent models of political economy suggest that this might not necessarily be the case. For instance, the political process might produce the ideal policy of a minority coalition within the electorate (see ^{Wittman, 1973}, ¹⁹⁸³; ^{Myerson, 1993}; and ^{Roemer, 2001}) instead of maximizing the wellbeing of the society, resulting in a suboptimal degree of policy differentiation.

In this paper, I also study whether local governments maximize the gains attributed by ^{Oates (1972)} to policy differentiation once I relax the assumption that governments are benevolent social planners. In particular, I compare the relative merits of decentralization in matching the heterogeneous preferences of voters with economic policies both with and without campaign contributions. To the best of my knowledge, I am the first to provide such analysis. In this paper, I show that an increase in the heterogeneity of preferences over local spending does not necessarily lead to a more differentiated provision of LPGs. In particular, if local public goods do not show spillovers or if the spillovers are moderate, then the more heterogeneous the preferences of voters for public goods across districts, the more heterogeneous is the inter-regional provision of public goods. In this case, a system of local governments can successfully accommodate different preferences of households across jurisdictions.

However, if spillovers of LPGs for districts with a high demand for public spending are sufficiently large or if heterogeneity of preferences is significant, then the more heterogeneous the inter-regional preferences of voters, the higher the political incentives for the low-demand district to free ride. In this case, even if local preferences are heterogeneous, only the local government with a sufficiently high demand for public spending provides a public good while the low demand district free rides. This result means that the inter-regional distribution of voters’ preferences could change without a corresponding change in the provision from local governments. This outcome also means that local governments do not necessarily maximize the society’s gains attached to matching the preferences of voters with the provision of local public goods.⁸

Finally, in this paper I develop a comparative analysis of the incentives to differentiate the supply of LPGs according to inter-regional differences in preferences for economies with and without campaign contributions. I show that for an economy in which campaign contributions are not allowed, the decentralized provision of LPGs does not maximize the society’s welfare gains attached to policy differentiation because parties select a policy to maximize the welfare of a minority coalition in the electorate (parties do not select policy to maximize the welfare of the society).

However, in an economy with campaign contributions, local governments select the ideal policy of a weighted average voter (that is to say, parties select middle-of-the-road policies) and therefore, a system of local governments provides public goods that maximize the society’s welfare gains from matching local spending with the heterogeneous preferences of voters.

This paper is structured as follows: the baseline voting model with ideological parties and without campaign contributions is developed in section 2. I develop a model with ideological parties and campaign contributions in section 3. Sections 2 and 3 show the relative efficiency outcomes in the provision of LPG’s for economies with and without campaign contributions. The comparative analysis of LPGs in the political equilibriums with and without campaign contributions is also provided in section 3. Section 4 concludes.

2. The model

Consider an economy with districts i and −i and n ⁱ = 1, 2 . . . . . . N ⁱ individuals in each district such that N ⁱ =6 N ⁻ⁱ . Preferences of a household j living in district i are μjixji,gi,g-i,αji=xji+αjiIng2+k-ig-i where xji is a private good, the overall consumption of local public goods by a resident of district i is Gi=gi+k-ig-i where g ⁱ is a local public good provided by district i and g ⁻ⁱ is provided by district −i.⁹ The parameter k ⁻ⁱ ∈ [0, 1) measures the extent of inter-regional spillovers of g ⁻ⁱ over residents of district i. For local public goods without spillovers k ⁻ⁱ = 0.

The parameter αji∈R+:αji∈αmini,αmaxi∀i,∀j measures the intensity of the individual’s preferences for public goods. I asume that the individuals’ preferences over public goods are heterogeneous across districts. The budget constraint of household j living in district i is xji=eji-ti, where eji is the household’s endowment and t ⁱ is a head tax on residents of district i.

3. Political equilibrium with campaign contributions

In this section, I characterize an economy in which voters can make campaign contributions to parties participating in local elections of districts i and −i. The timing of the political process is as follows: In the first stage, voters offer a platform-campaign package to parties in which different campaign contributions correspond to different levels of local public spending. Political contributions seek to influence the design of public spending. Each voter makes an offer knowing that the rest of voters are, simultaneously and non-cooperatively, offering their corresponding platform-contribution packages. Also, in the first stage, parties announce their policy platforms¹⁴

In the second stage, voters observe the candidates’ platforms and vote sincerely for the policy that maximizes the voters’ own preferences. Voters are sequentially rational and recognize that after the election takes place (in the third stage), the party winning the election takes all and implements the policy that maximizes the party’s preferences over public spending and the size of campaign contributions.

Following the literature, including ^{Grossman and Helpman (2001)}, and ^{Bernheim and Whinston (1986)}, and for simplicity of analysis, I ignore the issue of the dynamic inconsistency of the parties’ policies since campaign contribution take place at the first stage and the implementation of policy in the third stage. In a model with repeated interactions between voters and parties, through sequential elections, there is an equilibrium in which parties commit to their promises to voters to change policy for contributions to guarantee future contributions from voters. However, such analysis is left for future work.

For the analysis that follows, I characterize the indirect preferences over public goods of a voter type αji in district i by:¹⁵

The voter’s budget constraint is xji=cji-ti-cjmiαji where cjmiαji, cjm-iαji are the contributions of voter type αji to parties m in districts i and −i.

Moreover, the parties’ preferences over local public spending are given by:

Where Cmi=∑∀i,-i∑j=1Nicjmiαji ∀m,∀i is the amount of contributions-policy offers from all voters in districts i and −i to a party m = {L, R} in the election of district i.

DEFINITION 3. A subgame perfect Nash equilibrium for the economy is characterized by the parties’ policy announcements g^∗mi and g^∗m−i, by the voters’ contributing choices c^∗ _j ^mi ∀ m = {L , R} and voting choices in districts i and −i such that:¹⁶

3.i) In the first stage, voters in all districts offer platform-campaign contributions schedules cj*mi gji, αji ∀i,∀m to parties such that ¹⁷

 cj*migji,αji∈argmax vji(gmi,gm-i,cjmi,cjm-i,αji)s.t vmi(g*mi,g*m-i,C*mi)

Where Maxvmig˘mi,g˘m-i,C˘¨mi∀m,∀i  vmi(g*mi,g*m-i,C*mi)≥axvmig˘mi,g˘m-i,C˘¨mi∀m,∀i is an incentive compatibility constraint for parties where C^∗mi corresponds to the sum of all contributions of voters in which voter type αji contributes with a strictly positive amount to party m such that C*mi=∑∀i,-i∑j=1Ni-1ch*mig*mi,αhi+cj*mig*mi,αjiand cj*mig*mi,αji>0 while C˘*mi=∑∀i,-i∑j=1Ni-1c˘h*mig˘ji, αhi corresponds to the sum of all contributions from voters and, in this case, voter type αji does not make a campaign contribution and sets c˘j*mig˘ji, αji=0∀m,∀i

Moreover, c∀h≠j*mig*mi,αhi, c˘∀h≠j*mig˘hi, αhi are best response offers of all voters h≠j in each district, and g˘mi∈argmax vmi gmi,gm-i,C˘*mi,αmi∀m,∀i

Also, in the first stage, parties announce policies g^*mi ∀m, ∀i that maximize the parties’ preferences over local public goods and the size of campaign contributions C^*mi such that:

g*mi∈argmax vmi gmi,gm-i,C*mi,αmi∀m,∀i

3.ii) In the second stage, voters type αji in districts i, ∀i,−i vote for Party L if xjLi=vjig*Li.g*m-i,cj*mi,cj*m-i,αji-vjig*Ri.g*m-i,cj*mi,cj*m-i,αji>0.

L if xjLi<0 they vote for party R.

3.iii) In the third stage, the elected party in each district implements g^*mi ∀m, ∀i.

g*mi∈argmax vmi gmi,gm-i,C*mi,αmi∀m,∀i

The main difference between my political equilibrium with campaign contributions in this section, and the equilibrium of the last section is that in this section, voters of all districts provide a platform-campaign contributions offer cj*migji, αhi>0 ∀m, ∀ithat seeks to maximize the voters’ utilities subject to an incentive compatibility constraint in which the government’s payoff under the distribution of political contributions is given by C*mi=∑∀i,-i∑j=1Ni-1Ch*mig*mi, αhi+cj*mi(g*mi,αji).

The incentive compatibility constraint says that a voter type αⁱ makes a campaign contribution to some party m = L or R (that is cj*mig*mi, αhi>0) if the utility associated with the package of campaign contributions and the government’s corresponding policy, g^∗mi, that is selected as the government’s response to the voter’s contribution, is at least as high as the utility of the voter under the state of the economy in which the voter does not contribute, that is c˘j*mig˘mi, αji=0 ∀m, ∀i and in this case the government selects the alternative policy g˘mi∈argmax vmi gmi,g˘m-i,C˘*mi,αmi∀m,∀i

Proposition 3, below, characterizes the subgame perfect Nash equilibrium for an economy with campaign contributions and shows that the decentralized provision of local public goods with and with-out spillovers is Pareto efficient.

PROPOSITION 3. In the SPNE, assume some party m = {L ∨ R} satisfies Ωi∃αji∈αmini,αmaxi:xjmiαji>0>12∀i, -i. Then, in each district some party wins the local election, and Pareto efficient local public goods with and without inter-regional spillovers are provided in each district. θji∈R++ ∀j, ∀i, at the political equilibrium, g^∗mi ∀ i, ∀m is given by

∀i is the weight that the elected candidate in district i assigns to the preferences of voters of district i and −i such that the weights are allocated as follows:

Where ∑∀i,-i∑j=1NiθjiMRSjiαji is the weighted sum of the marginal rates of substitution MRSji αji=∂μji∂Gi/∂μji∂xji ∀j, ∀i and θji∈R++ ∀j, ∀i is the weight that the elected candidate in district i assigns to the preferences of voters of district i and −i such that the weights are allocated as follows:

PROOF. The problem is solved by backward induction. In the third stage of the game Ωi∃αji∈αmini,αmaxi:xjmiαji>0>12∀i, -i, which means that party m = {L ∨ R} wins the local election in each district. In the third stage, party m selects g*mi∈argmax vmi gmi,gm-i,C*mi,αmi/∂gi=0 for g*mi>0 ∀m,∀i implies

Where MRSmi=∂vmi∂Gi/∂vmi∂xmi=∂μmi∂Gi/∂μmi∂xmi.

In the first stage of the game, voters type αji in districts i and −i provide an offer of local spending-campaign contributions cj*migmi, αji ∀m, ∀i,-i to parties to be elected in districts i and −i in the third stage such that

Parties will accept campaign contributions in exchange for changes in public spending if ¹⁸

From (12), it follows that the government’s tradeoff between political contributions of voters of type αⁱ _j ∀m, ∀i and changes in local public spending is given by

The voters’ political contribution problem in (11) can be solved by selecting the ideal spending policy of voter type αji, ∀m, ∀i, given by gj*i, which maximizes

State the problem in (14) as follows:

The first order condition for the problem of voters’ type αji implies

And

Where MRSjiαji=∂μji∂Gi/∂μji∂xji∀j,∀i ¹⁹

The size of contributions of voter type αji evaluated at some feasible level of g~j*mi and g~j*m,-i are given by

cj*mig~j*m,-i,αji=∫0g~j*miMRSjiαji-1NIdgi

and

cj*m-ig~j*m,-i,αji=∫0g~j*m,-ik-iMRSjiαjidgi

Substitute (16) and (17) into (10) to show that

Define θji∈R++ ∀j, ∀i,-i as the weight that the elected party

in district i assigns to preferences of voters of district i and −i such that

Therefore condition (18) is equivalent to

As I mentioned before, parties realize that they can attract campaign contributions from voters of different regions by offering changes in the parties’ policy positions. In fact, parties may face a tradeoff between campaign contributions and the design of local public spending: On the one hand, parties would like to design policy to benefit their local supporters, but on the other hand, they recognize that they can offer changes in local policy in exchange of campaign contributions from voters living in other jurisdictions.

Hence, proposition 3 says that the tradeoff between policy platforms and campaign contributions induces the party controlling the local government in each district to recognize the distribution of marginal benefits of g^∗mi over all residents in district i plus its external benefits over residents of district −i against the marginal costs of providing g^∗mi. As a result, the SPNE in (8) leads to a fiscally decentralized provision of local public goods with and without inter-regional spillovers that is Pareto efficient. Condition (8) is a modified Samuelsonian condition of the equilibrium provision of LPG’s. In particular, the left hand side of condition (8) ∑∀i,-i∑j=1NiθjiMRSjiαjiis a weighted sum of marginal rates of substitution MRSjiαji=∂μji∂Gi/∂μji∂xji∀j,∀i,-i and θji∈R++ ∀j, ∀i,-i is the weight that the elected candidate in district i assigns to the preferences of voters of district i and −i while the right hand side of (8) is the social marginal cost of providing g^∗mi.

In what follows, proposition 4 shows that the inter-regional heterogeneity of preferences is linked with the incentives of local governments to provide (and to free ride in the provision of ) a local public good in their corresponding districts.

PROPOSITION 4. The distribution of local public goods for districts i and −i for the political equilibrium with campaign contributions is given by:

4.i) For ki = k⁻ⁱ= 0, g^*mi > 0 and g^*m−i > 0 : g^*mi ≠ g^*m−i satisfying g¨mi=Ni∑j=1Niθji ∀m,∀i

4.ii) For k-i∈0,1,  ∃Θ~i=γ-i/γi<1 with γi=∑j=1Niθjiαji/1Ni-k-iN-i and γ-i=∑j=1N-iθj-iαj-i/1Ni-kiNi such that 0 ≤ kⁱ < Θi < 1 and g^*mi ≠ g^*m−i with g^*mi > 0 and g^*m−i > 0 satisfying g^*mi = γⁱ − k⁻ⁱ γ⁻ⁱ ∀m, ∀i.

4.iii) For k-i∈0,1,  ∃Θ~i=γ-i/γi:Θ~i<1, and 0< Θi ≤ ki:g*mi=Ni∑j=1Niθjiαji>0 and g^*m−i = 0. In this case, district i is the only provider of a local public good (since g^*mi > 0) while district −i free rides and sets g^*m−i = 0.

PROOF. Fromproposition (3), g^*mi ∀i,−i is given by ∑∀i,-i∑j=1NiθjiMRSjiαji=1NI ∀i,-i which is equivalent to

Condition (21) leads to a system of two equations that correspond to the equilibrium conditions for g^*mi and g^*m−i. Solve the system and reduce terms to show that g^*mi is given by

The Cournot-Nash equilibrium is

Define γi=∑j=1Niθjiαji/1Ni-k-iN-i and γ-i=∑j=1N-iθj-iαj-i/1Ni-k-iN-i outcomes of 4.i ), 4.ii) and 4.iii) follow by assigning the values of ki and k−i identified in the proposition.

Proposition 4 has a similar interpretation to that given in proposition 2, and it shows that an increase in the heterogeneity of voters’ preferences over public spending does not necessarily lead to greater policy differentiation in the local provision of public goods for an economy with campaign contributions. In particular, if local public goods do not show spillovers, then their provision is differentiated according to g*mi=Ni∑j=1Niθjiαji>0 and g*mi=N-i∑j=1N-iθj-iαj-i>0 In this equilibrium g^*mi depends on the intensity of the party’s preferences for LPGs.

If the heterogeneity of preferences is moderate or the spillovers are also moderate (see conditions 4.ii)), then the supply of public goods is also differentiated with g*mi=∑j=1N-iθj-iαj-i1N-i-kiNi -ki∑j=1Niθjiαji1Ni-k-iN-i >0In this case, the size of g^*mi depends not only on the distribution of preferences of voters in district i (note that g^*mi depends positively on ∑j=1Niθjiαji) but also on the whole distribution of preferences of the electorate since g^∗mi also depends on the preferences of voters of district −i (since g^*mi is a function of ∑j=1Niθj-iαj-i) In the last two cases, if preferences become more heterogeneous, then local public spending reflects this heterogeneity into a set of differentiated local public goods with g^∗mi > 0 and g^∗m−i > 0.

However, if the heterogeneity of preferences is sufficiently high or if spillovers are sufficiently high (see condition 4.iii)) then the party’s marginal benefit of producing a local public good in the low demand district will be below its marginal cost and therefore only the high demand district will produce a public good with γi=Ni∑j=1Niθjiαji>0 and g^∗m−i = 0. In this case, the low demand district free rides and sets g^∗m−i = 0.

This outcome is explained by the strategic interaction between local governments in districts i and −i. When there is too much heterogeneity, the local government with high demand for local public goods has incentives to provide a local public good in district i that is g^∗m−i > 0. Through the effect of spillovers of g^∗mi in district −i, the marginal benefit for parties in district i of providing a local public is driven below its marginal cost. As a result, the local government of district −i free rides and sets g^∗m−i = 0.

Therefore, there could be a change in the heterogeneity of preferences of voters for LPG’s of districts i and −i without a corresponding change in the local provision of public goods. This outcome also high-lights the limitations of a system of local governments in maximizing the society’s gains associated with matching the inter-regional preferences of voters with the provision of LPGs.

One interesting question is whether the issue of free riding from the low demand district is more prevalent in an economy in which campaign contributions are not allowed (see proposition 2), or in which they are allowed (see proposition 4)? Proposition 5 provides a simple comparison between the results in propositions (2) and (4) and identifies sufficient conditions under which the provision of local public goods with campaign contributions might reduce the incentives of the low demand district to free ride. This proposition shows that a system of local governments in which campaign contributions are allowed is more likely to maximize the society’s gains associated with matching the inter-regional preferences of voters with the provision of LPGs. Formally:

PROPOSITION 5. For θji∈R++ ∀j, ∀i,-i and Eαji=∑j=1Niθjiαji∀i is a weighted average of the preferences for local public goods of residents in district i. If

Where

Then the strategic interaction between elected local governments and the special interests groups for an economy with campaign contributions reduces the incentives of the district with the low demand for local public goods to free ride relative to the incentives to free ride in the political equilibrium without campaign contributions.

PROOF. The electoral-economic equilibrium for an economy without campaign contributions implies, from proposition 2.iii), that ∃ k⁻ⁱ ∈ [0 , 1) , Θⁱ = N ⁻ⁱα^m−i /N ⁱ α^mi: Θⁱ < 1 and 0 < Θⁱ ≤ kⁱ ⇒ ĝ^mi = N ⁱ α^mi > 0 and ĝ^m−i = 0. Similarly, from the equilibrium with campaign contributions (in proposition 4.ii)), ∃ k⁻ⁱ ∈ [0 , 1) , Θⁱ = γ^-i/γⁱ<1 with γi=∑j=1Niθjiαji/1Ni-k-iN-i and γ-i=∑j=1Niθj-iαj-i1N-i-kiNi:0≤ki<Θ~i<1 and g^*mi ≠ g^*m−i with g^*mi > 0 and g^*m−i > 0 satisfying g^*mi = γⁱ− k⁻ⁱ γ⁻ⁱ ∀ m, ∀i. Therefore, if 0 < Θⁱ < Θⁱ ≤ ki proposition 5 holds.

Let ∃Θ~i=∑j=1N-iθj-iαj-i1N-i-kiNi /∑j=1N-iθjiαji1N-i-k-iN-i >1 and Θⁱ =N^−im,α^m,−i/Ni^mi: Θⁱ < 1:

For Nⁱ, N⁻ⁱ ∊ R_++,NiN-iΓ>0⟺Γ>0 and the expresión NiN-iΓ is equivalent to Ni/N-iΓ=Eαj-iEαji1-Ni/N-ik-i1-N-i/Niki-αm,-iαmi . Define Φ=1-N-i/Niki1-Ni/N-ik-i then Eαj-iEαji>Φαm,-iαmi⇒Ni/N-iΓ>0

Under the political equilibrium without campaign contributions (see proposition 1), the provision of local public goods reflects the ideal policies of the minority coalition of voters that is in control of the elected party in each district. If parties take into account only the preferences of a minority coalition of voters in the electorate then local public policy could be extreme or polarized (implying that the provision of public goods could be too high or too low) if the preferences of voters in control of the party are not moderate.

Therefore, in this type of equilibrium, the heterogeneity of preferences across districts could be high leading to the outcome in which the district with low demand for local public goods free rides and only the district with the high demand for public spending provides a LPG. Also, recall that in this case, there could be an increase in the inter-regional heterogeneity of preferences of parties for LPGs and yet the provision from local governments does not respond to changes in the inter-regional preferences of parties.

In the political equilibrium with campaign contributions (see proposition 3), local public goods reflect the ideal policies of residents of all districts of the economy, and therefore, the provision of LPGs with campaign contributions leads to more moderate policies relative the distribution of public goods under the equilibrium with-out campaign contributions, (see conditions 24 and 25).²⁰ It follows that if policies are moderate then the district with the low demand for local public spending has less incentives to free ride. Hence, it is more likely that the provision of local public goods is positive in each district and differentiated according to the inter-regional differences in the voters preferences.

Hence, proposition 5 identifies conditions in which a system of local governments in an economy with campaign contributions maximizes the welfare gains associated with the differentiation of local public goods while, in an economy without campaign contributions, a system of local governments is likely to produce suboptimal inter-regional policy differentiation. This outcome is explained by the fact that, for an economy with campaign contributions, local public goods are supplied as if local governments select public spending to maximize a weighted social welfare function.

In addition, in the equilibrium with campaign contributions, policies are more moderate than the corresponding policies adopted in the equilibrium with no campaign contributions. As a result, for an economy without campaign contributions, policies could be more extreme and then the difference between gˆ^mi − gˆ^m−i > 0 is larger which makes more likely that the marginal political benefits of producing gˆ^m−i in the low demand district falls below its marginal cost, in which case, district −i free rides and the supply of local public goods in the federation is gˆ^mi > 0 and gˆ^m−i = 0.