Discipline with Lessons from U.S. Federalism
2.1 Ensuring Local Fiscal Discipline
Efficient resource allocations by governments require that all benefits and all costs of the public action be fully internalized, that is, accounted for, by public officials when making their policy choices. The failure to account for all social benefits of a public action will typically mean that too little of that activity is provided. Conversely, the failure to account for all social costs will mean that too much of the chosen service or reg- ulation is provided. Called benefit or cost spillovers, these failures can be significant for subnational governments in economies with mobile residents, workers, and capital. In some cases, these policy failures are an unavoidable consequence of using subnational governments to provide public services; for example, children educated in one com- munity will provide benefits to other communities when they relocate as adults. In these instances, there are central government policies—
typically, intergovernmental grants-in-aid—that can induce local and provincial governments to provide the efficient level of the affected service (see Inman 1999). Our concern here, however, is not with these resource, or allocative spillovers, but rather with what might be called fiscal spillovers, created when local or state governments shift the budgetary costs of their own expenditures onto nonresidents, current or future. Again, there will be a failure by subnational governments to account for all the social consequences of their fiscal choices. An eco- nomically inefficient provision of local services will result. Controlling the ability of subnational governments to shift the costs of their fiscal choices becomes important.
Cost shifting by local governments can occur in three ways. First, local governments, working on behalf of resident taxpayers, may shift the production costs of local services onto nonresidents through feder- ally funded transfers or by tax exporting through local taxes whose burdens fall primarily on nonresidents. Second, local governments may borrow money for current-period expenditures through deficit financ- ing and then refuse to repay those debts, thereby shifting costs onto current-period lenders if the debt defaults or onto current-period national taxpayers if the debts are covered by a central government
bailout. Third, local governments again deficit-finance current-period expenditures, but now use deficit rollovers year after year until current local taxpayers have left the local jurisdiction, leaving future residents to finance aggregate debt repayment through higher future taxes. Each of these cost-shifting strategies, if successful, subsidizes the provision of current public services to current local residents with the subsidy paid by current nonresidents or by future residents and nonresidents.
What are the economic consequences of allowing local cost shifting?
Figure 2.1 describes how a typical local government might set its local budget and illustrates the economic inefficiencies that may arise under each cost-shifting strategy. The MB curve measures the marginal ben- efits to a typical local resident from another unit of the local public service consumed in the current period; the local service provides no benefits outside the local jurisdiction. The MC curve measures the social marginal costs of producing each unit of the local service in the current period. Efficient local government allocations occur at point Xe
where MB =MC; here local taxpayers receive area [A+B+E] in ben- efits and pay area [B+E] in costs earning a net fiscal surplus of area [A]. Successful local cost shifting breaks this equality at MB =MC by introducing a subsidy of F· MC between social marginal costs, MC, and
$
MC MB
A
B
E
C
F D
(1-F)MC
Local public services
Xe Xl
Figure 2.1
Local budgeting with cost shifting.
the marginal costs actually paid by local residents, (1 - F) · MC (see figure 2.1). With tax exporting, Fequals the fraction of social costs paid by taxing nonresidents. With transfers, Fequals the fraction of social costs paid by the national government through monetary or in-kind grants. With deficit spending, F equals the fraction of current social costs paid by borrowing, later financed by defaults or bailouts or through the taxation of future residents. In each case, local residents now find it optimal to increase local spending until MB equals their after-subsidy local marginal costs: MB =(1 - F) · MC. Local residents now purchase Xlin services, receiving area [A+B+C+E+F] in tax- payer benefits, but now paying only area [E+F] in local taxes, thereby earning a net fiscal surplus of area [A+B+C] (see figure 2.1). Clearly, local taxpayers prefer the subsidy. National taxpayers, bondholders, or future taxpayers pay the cost of the subsidy—area [B+C+D] in Figure 2.1. For the economy as whole, this is an inefficient outcome; subtract- ing national taxpayers’ costs from local taxpayers’ net benefits yields a measure of the full economy’s aggregate net benefits equal to only area [A-D], less than what economy as whole had earned—area [A]—when the local government bought services fully on its own. Area [D] in Figure 2.1 measures the resulting economic inefficiency.2Local bailouts and transfers are privately preferred but, without significant extra- jurisdiction allocative spillovers, socially inefficient.
To control these inefficiencies from local cost shifting, Fmust equal 0; local taxpayers must be responsible for the full marginal costs of the local services they purchase. Yet Fis itself endogenous and determined within the context of a wider federal system, a context within which local taxpayers may have important influence. If so, a richer institu- tional analysis than local choice alone (Figure 2.1) is required for an understanding of local fiscal discipline. Ensuring local fiscal discipline must be done with an understanding of the full political economy of local finance. The analysis that follows illustrates how narrow, local interests might come to undermine a national interest in an efficient local public sector. In the process, we learn what wider political and market institutions will be needed to check inefficient cost shifting by the local public sector.
2.1.1 Controlling Local Transfers and Tax Exporting
Both transfers from the central government and approval by the central government for local governments to tax nonresidents’ income or con-
sumption offer an implicit subsidy to the local sector for the purchase local public goods. Transfers may be paid as direct price subsidies (matching grants), as in-kind services such as federally financed in- frastructure construction, or as targeted lump-sum grants for new services or activities (closed-end grants). Absent significant intergov- ernmental spillovers, each such transfer creates a wedge between the social marginal costs (MC) and benefits (MB) of local services; an inef- ficiency of area [D] results.3So too will allowing local governments to tax nonresidents, for example, through the taxation of firm capital, natural resources, or other export goods (see Inman and Rubinfeld 1996 for an overview). Both strategies require the cooperation of the central government, either directly through the payment of intergovernmen- tal transfers or indirectly by allowing local governments to tax non- resident incomes and consumption. Why would national taxpayers, those who pay the subsidy, allow such inefficiencies to stand?
Exhibit 2.1 illustrates the core difficulty facing any democratically elected national government wishing to remove inefficient local trans- fers or the taxation of nonresidents. The problem arises when central government policies are set by a national legislature composed of inde- pendent representatives elected from each of the country’s many local governments. In such a legislature, if local interests are to shift local costs, then cost shifting wins, despite a national interest in checking such behaviors. The problem, as exhibit 2.1 makes clear, is the pris- oner’s dilemma character of legislative politics dominated by inde- pendent local interests.4While all local representatives might agree that a national policy favoring local cost shifting is collectively inefficient—
each local government gets its own subsidy but then must pay for the subsidies given to all other local governments—no single local repre- sentative can afford unilaterally to sacrifice his own government’s subsidy for the benefits of improved national fiscal policy. In Figure 2.1, for example, all subnational governments would prefer allocation Xeto allocation Xl, but only if all other local governments joined a coalition and also supported the efficient outcome.5
To control local cost shifting, it is essential to control the prisoner’s dilemma game played by these representatives when setting national policy. To do so, the incentives of the independent local legislators must be changed. Either penalties must be imposed when they vote for a local cost-shifting policy or extra rewards must be offered when they do not. Either way, the cooperative no-cost-shifting strategy must become the best choice for every locally elected representative no
Exhibit 2.1
The Central Legislature’s Cost-Shifting Game
The central legislature’s decision to allow cost shifting by local govern- ments is an example of a prisoner’s dilemma game in which the locally elected representative chooses either to support the legislative action s (“shift”), which provides benefits to its citizens while shifting a fraction of the costs of those benefits to the citizens of all other subnational gov- ernments, or the legislative action g(“no shift”), which provides the ben- efits without shifting any of the associated costs. The shifting strategy s can be thought of as proposing and voting for a local tax on nonresident tax bases (tax exporting) or submitting a local subsidy for national financing (pork-barrel spending or deficit bailout). The no-shifting strat- egy gcan be thought of as not submitting a local subsidy for the national budget. If all local governments adopt the strategy of no shifting, then each of their citizens receives a payoff of Pgg. (For simplicity, but without loss of generality, assume all governments are identical.) If, however, all local governments adopt the strategy of shifting, then the citizens of each of those governments receive a payoff of Pss. If one government adopts the strategy gof no shifting, but all others choose sand support to shift local costs, then the payoff to the citizens of the “honest” no-shifting gov- ernment will be Pgs, while the citizens in a typical government adopting the shifting strategy will receive Psg. Conversely, if one government adopts the shifting strategy s, and all others adopt the honest no- shifting strategy g, then the citizens of the shifting government receive Psgand the citizens of a typical no-shifting government receive Pgs. The game presents each subnational government with the following payoff matrix (with the local government’s payoffs reported first):
All other subnational governments
No shift (g) Shift (s) No shift (g) Pgg; Pgg Pgs; Psg
One local government
Shift (s) Psg; Pgs Pss; Pss
For the game to qualify as a prisoner’s dilemma game, the payoffs for each local government must follow the sequence:
The other payoffs in the matrix, Psgand Pgs, are the payoffs to other gov- ernments and are not relevant to each individual government’s making its own choices. To rule out a randomized strategy as preferred, we also require that Pgg>[Psg+ Pgs]/2. In words, each local government will most prefer to cost-shift (s) and have all other local governments pay for their own services from own fiscal resources. If that outcome cannot occur, then all local governments’ agreeing not to shift their costs is
Psg>Pgg>Pss>Pgs.
matter what the other representatives do.6With appropriate penalties and rewards, the cooperative outcome can be enforced. The needed penalties and rewards must come from some organization or individ- ual with sufficient extralegislative resources, however, and it must be an organization or individual rewarded for encouraging the nationally efficient—no local cost shifting—allocations. Within the structure of democratic politics, either of two commonly mentioned alternatives will meet these two conditions:
1. Nationally elected political parties with the ability to control the election prospects of local representatives (Wittman 1989), or nation- ally elected presidents with the ability to grant or deny favors valued by local representatives and their constituents (Fitts and Inman 1992, Chari, Jones, and Marimon 1997).
Paradoxically perhaps, to ensure efficient local government finance, we must first establish institutions to ensure a strong central government.
2.1.2 Controlling Local Defaults and Central Bailouts
However successful an economy might be in finding extralegislative institutions to check the inefficient prisoner’s dilemma behaviors of local representatives in decentralized legislatures (exhibit 2.1), there remains a second route through which local governments can extract socially inefficient subsidies from the national government: demand and receive a bailout for a locally created fiscal crisis. Even strong central governments capable of resisting local demands for current period subsidies remain susceptible to this strategy. Here the local gov- ernment spends borrowed money to subsidize current services.7 The
preferred. Having all governments adopt the shifting strategy is each government’s third best outcome. In last place is the outcome from the naive strategy of not cost shifting while all other governments do.
With this ordering of outcomes, each local government prefers the cost-shifting strategy knowing that all other local governments also prefer that strategy. The outcome of the game is all governments’ adopt- ing the cost-shifting strategy and receiving the socially inefficient payoff of Pss.
Exhibit 2.1 (continued)
Exhibit 2.2
The Default-Bailout Game
The default-bailout game is a special case of a sequential game in which the local or state government—more generally, the subordinate division in an organization’s hierarchy—adopts an action D (unfunded deficit) that provides benefits to the local government of Bl(>0) and benefits to central government—more generally, the center in an organization’s hierarchy—of Bc(≥or £0). Following the adoption of action D, the central government responds by adopting either an action b (bailout) or an action h(no bailout). If action bis chosen by the central government, it costs the center Cbc(>0) and the local government Clb(≥0), while if action h is selected, then the central government bears a cost of Cch (>0) and the local government Chl (≥0). If the local government chooses not to adopt action D, then the status quo remains in place and the local and central governments receive the payoffs in the status quo, Ql and Qc, respectively.
The game has the following payoff structure:
CURRENT PERIOD: Local Government Adopts Action:
D OR T
FUTURE PERIOD: Center Adopts Action: Center Adopts Action:
b OR h Status Quo
Center Payoff: Bc-Ccb Bc-Cch Qc Local Payoff: Bl-Cbl Bl-Clh Ql
This sequential game becomes the default-bailout game when the local government prefers action D, the central government prefers action b given that the local government has chosen D, and together these actions reduce aggregate social welfare measured by the combined net benefits to the citizens of the local and central governments. For this to be the case, three restrictions on the values of benefits and costs must apply:
Condition 1: Central government prefers b, given D. That is, or
Condition 2: Local government prefers D, given b, but would prefer the status quo, given h. That is,
In words, the local government (i.e., division) adopts an action Dthat is beneficial to its residents (i.e., employees) but prompts a second-best Condition3:Aggregate inefficiency That is. ,[Bc-Cbc]+[B1-C1b]<Qc+Q1 Condition b2 :B1-C1h£Q1
Condition a2 :B1-Cb1>Q1 Chc>Cbc
Bc Cc Bc c
- b> -Ch
local government then declares a “fiscal crisis” and defaults on its local debts. The hope is that the central government will step in to cover the debt through a fiscal bailout. Will it?
Exhibit 2.2 illustrates when the central government might offer a fiscal bailout to a local government threatening default. The essential problem here is one of time inconsistency, or what has become known as the problem of the soft budget constraint: today’s promises by the central government of no bailouts are not credible against tomorrow’s circumstances (see Kornai 1986). First, the central government prom- ises there will be no bailouts, perhaps because of a constitutional pro- vision requiring balanced local budgets and no national government deficit relief. Second, the local government then decides how to finance its current local services, either by borrowing some or all of the needed resources (action D in exhibit 2.2) or, alternatively, by not borrowing and paying for services through local taxes only (action T in exhibit 2.2). If only local taxes are used to finance local services, then the effi- cient allocation at Xein Figure 2.1 results. If the local government has borrowed D today, then tomorrow it declares default and asks the central government for a debt bailout. Third, the central government can decide to offer the bailout (action bin exhibit 2.2) or not (action h in exhibit 2.2). If no bailout is offered, then the burden of debt repay- ment is shifted back onto the local government, which either repays the debt from local taxes or defaults, shifting the final burden onto bondholders.
Which decisions are best for the local and central governments turns on the benefits and costs associated with each of the three possible out- comes in exhibit 2.2: the default-bailout (D,b) path, the default–no bailout (D,h) path, or the tax-financing path (T). If the local government has borrowed to finance local public services—either path (D,b) or Exhibit 2.1
(continued)
b response from the central government (i.e., center), which is damag- ing to the country (i.e., organization) as a whole. The problem arises because the central government bears a large cost if it does not respond with b(condition 1), and knowing this, the local government prefers the privately beneficial (condition 2) but socially inefficient (condition 3) strategy D.
(D,h)—then a level of local public services will be chosen by the local government today, resulting in benefits for the central and local deci- sion makers of Bc and Bl, respectively. If there are no spillovers from the local budget to citizens outside the community, then Bc=0;8 local services will be provided only if there are local benefits, so Bl>0. Once the local government has chosen its service level, I also assume the resulting benefits are then fixed and not affected by the subsequent bailout decision of the central government.
The central government’s bailout decision—either path (D,b) or (D,h)—does affect the costs for both levels of government, however.
Because a bailout is a subsidy, a local government’s net fiscal costs with bailouts (D,b) will always be less than its net costs without bailout (D,h)—that is, C1b<C1h. If it chooses bailouts, then the central govern- ment’s costs may be larger, equal to, or smaller than its costs from choosing the no-bailout strategy: Ccb>, =, < Cch. Bailouts will cost the central government national tax dollars, but not to bail-out a distressed local government may lead to costly financial or political crises. It will be the magnitude of the costs of not bailing out a “failing” local gov- ernment that will prove decisive in this analysis. Finally, if the local government chooses the tax financing path (T), the central government accepts the resulting allocation, and net benefits are simply Qcfor the central government and Q1 for the local government. As (I assume) there are no national spillovers from the locally provided public serv- ices (Bc=0), and no central government tax payments when T is chosen, Qc=0 as well.9Local government net benefits from tax-financing local services are assumed to be positive when the public service is provided;
thus Q1>0.
For this default-bailout game to be a source of local government fiscal inefficiency, the local government must prefer action D, the central government must prefer action bgiven that the local government has chosen D, and together these actions must reduce aggregate social welfare measured by the combined net benefits to the citizens of the local and central government decisions. For this to be the case, three conditions must hold; they are listed in exhibit 2.2. First, the central government must prefer to use the bailout strategy. This occurs when the no-bailout choice is more costly than the bailout itself; thus, condi- tion 1: Chc>Ccb. Second, knowing this, the local government must prefer to use local debt for the financing of local services; thus, condition 2a:
B1- C1b>Q1. To be sure that it is central government bailout policies that are really creating the economic inefficiencies, we also require that