Electoral Institutions, Cabinet Negotiations, and Budget Deficits within the European Union

Mark Hallenberg and Jürgen von Hagen

Center for German and European Studies, University of California at Berkeley

February 1997

Abstract

Large government budget deficits are a concern in most industrialized countries. Two literatures in political economy argue that differences in political institutions explain much of the variation in the success of counties in their efforts to run small deficits. One group of authors considers how differences among electoral systems affect the size of budget deficits, while the second group concentrates on the governmental institutions which structure the formation of the yearly budget. Among the "electoral institutionalists", a consensus is beginning to emerge which treats proportional representation systems as a cause of high levels of public debt. In contrast, "fiscal institutionalists" argue that the presence of certain institutions in the decision-making process at the cabinet level, such as a strong finance minister or negotiated spending targets, lead to smaller deficits than in cases where such institutions are missing. We indicate that these two literatures complement one another. Electoral institutions matter because they restrict the type of budgetary institution at the governmental phase which a state has at its disposal. A strong finance minister is feasible in states where one-party governments are the norm, and such states usually have plurality electoral systems, while negotiated targets provide an alternative in multi-party governments. In multi-party governments, which are common in states with proportional representation, the coalition members are not willing to delegate to one actor the ability to monitor and punish the others for "defections" on the budget. The empirical section of the paper indicates a strong relationship between one-party governments and strong finance minister solutions within the European Union states on the one hand and multi-party or minority governments and targets on the other. Pooled time series regression results also support our contention that it is the presence or absence of one of these budgetary institutions, rather than the plurality/proportional representation dishotomy, which has the greatest impact on debt levels.

I. Introduction

Large government budget deficits are a concern in most OECD countries. In the United States, both major political parties, while differing on how to reach the goal of a balanced budget, have nonetheless agreed to make a balanced budget a top policy priority. Within the European Union, high budget deficits may affect a member state's ability to participate in monetary union--the Maastricht Treaty stipulates that governments with excessive debt levels, defined as yearly deficits of 3% of GDP and total debt burdens of 60% of GDP, should be excluded from participation in the common currency. One reason for a renewed commitment by politicians to reduce deficits is a recognition of the negative economic effects of chronic deficits and debt levels. They lead on average to higher interest rates, lower economic growth, a depreciated currency, and a restriction on spending on valued public services. States have had varying levels of success in keeping deficits low. Some, like Germany, Great Britain, and France, have managed to maintain relatively low deficit and debt levels, while others, such as Italy, Greece, and Belgium, have suffered under chronic deficits and/or debt levels.

Two literatures in political economy argue that differences in political institutions explain much of the variation in the success of countries in their efforts to run small deficits. One group of authors considers how differences among electoral systems affect the size of budget deficits, while the second group concentrates on the governmental institutions which structure the formation of the yearly budget.

Among the "electoral institutionalists," a consensus is beginning to emerge which treats proportional representation systems as a cause of high levels of public debt. Proportional representation (PR) systems are often considered inherently more unstable than pluralist electoral systems. Government ministers who expect to lose their positions soon after they gain them do not anticipate dealing with the consequences of their actions, and they willingly increase debt levels (Persson and Svensson 1989; Roubini and Sachs 1989; Alesina and Tabellini 1990; Grilli, Masciandaro, and Tabellini 1991; Hahm 1994; for a dissenting view on the stability of policies in PR systems, see Rogowski 1987). Others emphasize that coalition governments, which are common in PR systems, are less able to deal with negative shocks to the economy. Such governments face a prisoner's dilemma of whose ministry should suffer the budget cuts. Coalition partners may have enough power to block change, but not enough leverage to effect positive change on their own (Roubini and Sachs, 1989, Alesina and Perotti 1995). PR systems also lead to greater polarization in the political system. If the party or parties in government anticipate that their opposition will someday assume power, they may seek to confine future governments by generating present debts, and the incentive to generate larger debts increases with political polarization (Tabellini and Alesina 1990). ₁ In contrast, governments that emerge under a pluralist system are more decisive, the system discourages extremist parties, and the governments stay in power longer and are more stable. For all of these reasons pluralist electoral systems lead to lower levels of government debt.

While the theoretical work has sparked interest, the empirical support for this argument has been uneven. In a reanalysis of Roubini and Sachs' dataset, Edin and Ohlsson (1991) find that minority governments, rather than PR states per se, are more likely to run large budget deficits. Alesina and Perotti (1995), while confirming a link between coalition governments and low success rates in the implementation of austerity programs in OECD countries, discover paradoxically that minority governments are the most fiscally responsible form of government, more fiscally responsible than even one party majority governments. De Haan and Sturm (1994), in a pooled time-series analysis of European Community countries from 1981 through 1989, find no statistically significant relationship at all between the form of government and budget deficits.

The "fiscal institutionalists" consider how budgetary institutions affect the size of deficits. During the formulation of the budget at the cabinet level (the governmental phase), a strong finance minister can force the decisionmakers to consider the true benefits and costs of increased spending and taxation (von Hagen 1992; Von Hagen and Harden 1994a, 1994b, 1996; Alesina, Hausmann, Hommes, and Stein 1995; Hahm, Kamlet, and Mowery 1996). Similarly, negotiated spending targets for each ministry can also lead to smaller deficits (von Hagen 1992; Von Hagen and Harden 1994a, 1994b, Forthcoming). The approach examines the structure of other parts of the budget process as well, such as how parliament deals with the government's proposed budget, how the budget is implemented, and whether there are any ex-post controls. While the statistical evidence in support of the effects of such institutions has generally been stronger, ₂ this approach does not explain why some states choose a given budgetary institution and others do not.

In this paper, we indicate that these two literatures complement one another. Electoral institutions matter because they restrict the type of budgetary institution at the governmental phase which a state has at its disposal. A strong finance minister is feasible in states where one-party governments are the norm, and such states usually have plurality electoral systems. In multi-party governments, which are common in states with proportional representation, the coalition members are not willing to delegate to one actor the ability to monitor and punish the others. Negotiated targets provide an alternative in multi-party governments. They will be credible, however, only if all the parties can monitor and punish each other. Since parties often lack the ability to provide one or the other of these functions, targets are harder to maintain successfully than a strong finance minister. This result explains why many electoral institutionalists find that PR states, on average, are more prone to run larger deficits. At the same time, since such states which do maintain negotiated spending targets will have deficits that are as low as plurality states with a strong finance minister, a general comparison of plurality states with PR states misses the effect of budgetary institutions.

We first develop a model of the budget process and show that the distinction between one-party and multi-party governments affects which institution, either a strong finance minister or negotiated targets for each ministry, a country can use to reduce spending. In the second part of the paper, we consider one distinguishing feature of electoral systems, namely their effects on the likelihood of one-party or multi-party majority governments. The existing literature indicates that plurality systems are much more likely to have one-party governments than PR states. At the same time, PR states with a low average district magnitude, which is the number of candidates per electoral district, are also more likely to have one party win a majority of votes and form a government.

The final section examines the use of such institutional constraints within the current 15 European Union states from 1981 through 1994. These states are of theoretical interest because economic shocks, which often have short-run consequences for a country's fiscal balance, should impact this group more or less equally. One can therefore provide a control for such external factors. From a policy perspective, these state are also of interest because of the Maastricht Treaty's provisions concerning yearly deficits and aggregate debt. If certain institutions have been effective in some states, they may provide a way for high debt states to bring their fiscal policies in step with the Maastricht Treaty's guidelines. This section indicates a strong relationship between one-party majority government and the use of a strong finance minister on the one hand and multi-party governments and budgetary targets on the other. Pooled time-series regressions that are presented at the end of the paper indicate that the presence or absence of these constraints, rather than the electoral system per se, is the crucial variable which affects the size of the budget deficit. Not all of the states chose one of the institutions, and those that did so registered significantly lower yearly deficits and overall debt levels than those states which chose to forego the institutional constraints.

II. The Budget Negotiating Process within Cabinet Governments

a. Introduction:

The following section contrasts one-party majority, multi-party majority, and all forms of minority government in the context of a model of the budget process. We show that the structure of the bargaining process within the cabinet affects the size of the budget. If ministers are left to determine their own budgets, they will select amounts that are larger than what is collectively optimal for the government in power. There are two reasons for this outcome. First, the budget process resembles a common pool resource problem. The respective ministers determine the spending priorities of their ministries, but they may not consider the entire tax burden in their deliberations; instead, their main worry is the burden that their constituencies must bear. An agriculture minister, for instance, will be most concerned about the services and goods she can provide to farmers and about the taxes that those farmers must pay. She therefore does not consider the full implications of each additional unit of spending on the entire government. Second, a given minister may receive an inherent benefit from increasing budgets in her ministry. The more resources she has at her disposal, the greater her ability to provide services to her constituency, and she may also receive private benefits, such as greater status and prestige, from commanding larger budgets. Once again she will want a budget for her ministry which is higher than what is collectively optimal for the government in power. A large literature has developed examining the conditions under which the players will choose to cooperate with each other in such situations (Olson 1965; Hardin 1982; Ostrom 1990; Ostrom, Gardner, and Walker 1994). All of these solutions involve the use of selective punishments or incentives and the monitoring of the actors. In some cases, the players themselves can monitor and punish each other and ensure cooperation without any formal institutions.

In the budget game, however, such solutions are usually impractical. There exist two institutions which can provide the necessary punishment/incentive and the monitoring functions: a strong finance minister, which we label "strategic dominance," or negotiated spending targets for each ministry. These institutions differ in the level of centralization that they require. Under the more centralized strategic dominance case, the actors delegate to one player, usually the finance minister, the ability to monitor and punish the other actors, and the finance minister can potentially solve both the common pool resource problem and the incentive to run larger budgets. The less centralized targets also force the players the consider the entire tax burden levied on the government, and they solve the common pool resource problem. At the same time, however, targets do not rectify the bias for larger budgets on the part of individual ministers, and the players will therefore prefer the more centralized solution if both options are available.

Whether one or both institutions are available to a government depends on whether the government is composed of one party or many parties. In the one party case, the players can delegate the monitoring and punishing roles to one central player, the finance minister. The spending ministers and the finance minister together have common ideal budgets which are consistent with their party platform, and they can trust the finance minister to use his power to safeguard the interests of the party. Within multi-party coalitions, however, the parties will not accept the high level of centralization present under strategic dominance. The parties will often have conflicting policy goals, and they can never be sure that the finance minister will detect defections within her own party and mete out fair punishments. Minority governments, which must find support in parliament to pass the budget, similarly must rely on negotiated targets if they wish to reduce spending.

b. The Model

For simplicity and without loss of generality, we assume that there are two parties in a given government, Party A and Party B. ₃ Each party has a set of ideal budgets that correspond to the different ministries. A Christian democratic party, for instance, may be interested in allocating more money for the defense ministry, while its liberal partner may want less spending in defense and a higher allocation for the ministry responsible for agriculture. These groups will also consider the revenue side of the budget, and they will be most concerned with the taxes that their constituencies pay. With these different factors in mind, the following utility equation can be written for Party A:

with xiA * equal to Party A's ideal spending level for ministry i, xi the amount of funding the ith ministry receives in the budget, mA the proportion of the total tax burden that the party's constituency is expected to pay (with mA ≤ 1, mi the tax burden on each ministry's constituency, miA the tax burden for each ministry under a party' control, and ΣnA i=1 miA =mA T the amount of tax, and α the relative weight that the group places on spending relative to taxation.

We will first assume that the players agree with each other on their respective ideal budgets, such that V _A =V _B (This assumption is relaxed later in the paper.) Under a collectively optimal budget, the ministers consider the total tax burden on the government's supporters instead of just the tax burden for their respective constituencies, and they do not value larger budgets for their own sake. If we assume without loss of generality that m_i = m/n and m= Σⁿ _i=1 m_i the total optimal budget for the parties, B_o , would then be

These parties are composed of individual ministers who may have different priorities. Spending ministers seek full funding for those programs that they consider important to reach their policy goals, and we assume that they want to minimize the difference between their ideal budgets and the budgets that they receive. Like the parties, the ministers also factor in revenue decisions into their utility equations, but they are concerned only with the tax burden on their respective constituencies. This selective treatment of the tax implications of increased spending leads to the common pool resource problem (von Hagen and Harden 1994a, forthcoming; for a Congressional context see Weingast, Shepsle, and Johnsen 1981). Increased spending creates a negative externality in the form of higher taxes which falls on all of the actors, but, since they only consider the burden on their supporters, they will request higher levels of spending than if they had considered the total tax. It should be noted that the common pool resource problem leads to excessive spending and excessive deficits simultaneously, and that, since both are aspects of the same problem, we will focus on just total spending levels. ₄

Spending ministers may also benefit simply from having larger budgets. While Niskanen (1971) discusses bureaucrats who seek to maximize their budgets, their bosses may also receive a reward from controlling more resources rather than less, and as a consequence they may request funds that are greater than the minimum needed to reach their policy goals. With the common pool problem and budget-maximizing ministers in mind, and assuming that the excess burden of taxation is quadratic, the ith minister will possess the utility function

where λ is equal to the value a given minister places on the size of her ministry's budget, x_i the amount of funding the ministry receives, x_i ^* the ideal spending level for minister i, m_i again the proportion of the total tax burden that the minister's constituency is expected to pay, and α the relative weight that the minister places on spending concerns. ₅ If spending ministers bid for funds assuming that the other ministries' budgets are given, and if the total aggregate budget, denoted as B_a , is simply the aggregation of these individual bids, the total budget will take the form

These budgets are larger than the budgets the party as a whole would want for two reasons: first, under a decentralized budget a given minister receives some utility simply from having larger budgets, and second, the minister only considers the tax burden on her constituency instead of the total tax burden on the government.

With this difference between collectively-optimal and individually-optimal levels of spending in mind, one can easily imagine that the budget process takes the form of a prisoner's dilemma game. A large literature has developed examining the conditions under which the players will choose to cooperate with each other in prisoner's dilemma situations (Olson 1965; Hardin 1982; Axelrod 1984; Ostrom, Gardner, and Walker 1994). All of these solutions involve the need for punishing the defectors and monitoring the actors. In some cases, the players themselves can monitor and punish each other and ensure cooperation without any formal institutions. In the budget game, however, such solutions are unlikely. First, not every situation will necessarily approximate a prisoner's dilemma in the strictest sense; there may be occasions where at least one player is better off if all the actors defect than if they all cooperate and propose lower budgets. A given minister, for instance, may value higher levels of spending for her department so much that she is content to see everyone receive a larger, rather than a smaller, budget, even if this increase in spending hurts everyone in aggregate because of a larger budget deficit. This implies that the punishment for defection must be stronger than a simple retaliation from the other actors. Second, it may be difficult for the actors themselves to detect defections. While they certainly know what it costs to produce a given level of output in their own ministries, they may not know the functions of the other ministries.

In budget-making situations, therefore, spending will be lower if an institution serves two functions: it collects information on the actions of all the players, and it enforces penalties on players who propose budgets which are larger than is collectively optimal. In the next two sections, we indicate that the institutions available to governments vary according to whether a majority government is composed of one party or a coalition of parties. In one-party governments, the players agree on policy priorities, and they are willing to delegate monitoring and enforcement capability to one central actor. In multi-party governments, however, the parties differ on the definition of a collectively optimal budget, and indeed if spending is to be reined in the players must negotiate the budget goals. Once these binding targets are set, the players must have the means to monitor and punish each other to assure that targets are kept. They will not delegate this power to one central person, however, because that person can come from only one of the parties.

c. One-party Majority Governments

When just one party possesses all of the cabinet positions, the definition of a collectively optimal budget is straight-forward. Members of the same political party likely hold similar views, and they are forced to unite on a common platform for elections. While party platforms can often be vague, they do provide a blueprint for the spending priorities of the government. The players therefore have more or less the same ideal budget for each respective ministry, which means that there is no disagreement on the value of x ^* _i , and the ideal budget will be one given in equation (2). ₆

This agreement on the government's optimal budget allows the players to delegate authority to a "fiscal entrepreneur," whose function is to assure that all actors choose to cooperate. To be effective, this entrepreneur must have the ability to monitor the others, possess selective incentives that he can use to punish defectors and/or reward those who cooperate, and have some motivation to bear the costs of monitoring himself (Olson 1965; Frohlich and Oppenheimer 1978; Cox and McCubbins 1993). Among the relevant cabinet members, the finance minister often plays the role of this entrepreneur. His interests generally coincide with the general interest. ₇ He has the responsibility to coordinate the formation of the budget, and, fair or not, the size of the budget deficit is often the principal indicator that others use to judge his effectiveness. He often also has a trivial budget when compared with other ministers, and he cannot "defect" in the prisoner's dilemma game being played in the cabinet. Finally, the finance minister's staff gives him the means to monitor the actions of the other ministries, and since his prestige and hence his personal benefits depend on the effectiveness of his ministry, he has a private incentive to guarantee that the monitoring occurs. The only question is whether the finance minister has the power to offer selective incentives and/or punishments to the spending ministers.

The cost a finance minister can levy depends on his relative power in cabinet, and here we return to the importance of electoral institutions and the role of the prime minister. The prime minister in one-party governments especially is the strongest member of the cabinet. The prime minister is the leader of the governing party, and this position reinforces her power within the cabinet. The prime minister also can often select cabinet members and can reshuffle her government. ₈ Even in the United Kingdom, where the norm of "first among equals" is historically strong, a prime minister dictates the shape of her cabinet. If a given spending minister consistently presents unsatisfactory budgets, the prime minister can then replace her with someone who will develop more sympathetic policies. Finally, a prime minister can call a vote of confidence on a given issue which puts the very existence of the government at issue if a given minister does not support her position (Huber 1996). If the prime minister prefers that the party's ideal budget be reached, which should usually be the case, she will have identical preferences on the budget as the finance minister. She can then delegate her power to the finance minister, and the finance minister will represent a faithful "agent" of the prime minister. ₉

Spending ministers also have reason to support a strong finance minister. If they all face a prisoner's dilemma, they prefer the solution where all players cooperate to the solution where all defect. In cases where a few ministers prefer the all defect outcome, the others have an even greater incentive to appoint a central actor to monitor their actions and punish those ministers who defect. The finance minister can use his ministry to audit the accounts of government ministries and to publicize any spending minister deviations. He can also punish intransigent ministers directly and indirectly. As long as he has the ability to modify a spending minister's budget proposal, he can punish defectors in future years. If immediate action is required, he can appeal to the prime minister to take action, and, in the most extreme case, he can insist that the prime minister relieve the spending minister of his position. Thus, a fully successful finance minister will assure that the total budget equals the figure given in equation (4).

d. Multi-Party Majority Governments

In multi-party cabinets the situation becomes more complicated. The most important feature is that they likely have different policy priorities and hence different views of what form the "optimal budget" should take for the government. A labor party, for instance, may be interested in increasing employment, while its coalition partner, a green party, wants to pass legislation which increases spending on the environment but which also leads to higher unemployment. ₁₀ This means that the parties have different ideal budgets for each of the ministries, so that x ^* _i becomes x ^* _ij. The parties then have two options for deciding the budget: they can grant each minister the authority to decide the spending level for herself; or they can negotiate fixed spending targets for each of the ministries. ₁₁

Consider first the situation where the parties simply sum their preferred budgets for the ministries they control, which is the level of spending for those ministries that maximizes V_r

Assuming again that m_i = m/n, the total budget is then

Note that this aggregate budget is simply the sum of the individual budgets. The x ^* _ij in the equation represent the ideal budgets of the jth party which controls a given ministry. In this case, both λ, which represents the players' gains from larger budgets for their own sake, and m_i , which represents the tax burden on a given players' constituency, enter the total budget equation.

The actors can reach lower levels of spending if they require the parties to negotiate binding targets for each ministry. One important factor is that the negotiation process forces the actor to consider the total tax burden on the government. There are two likely routes that the negotiations can take. First, consider a situation where the party which is awarded a given ministry also has its ideal budget figured into the joint utility equation. The budget would then take the form

This result resembles equation (5), but the denominator is larger because the negotiating process eliminates the spending bias from the common pool problem. Since λ remains in the equation, however, this budget does not prevent ministers who prefer higher spending for its own sake from receiving more funds. To the extent that ministers wnat to maximize their allocations, this budget will be larger than the one which emerges in the one-party case where a central actor can reduce or eliminate both the common pool and the public choice problems.

Next consider a situation where the parties also negotiate on which ideal budget to consider for each ministry, which we designate as x ^* _ic ; this figure then replaces x ^* _ij in equation (6), and corresponds to the Nash bargaining solution. The budget then assumes the form

To the extent that parties gain portfolios in ministries in which they would like to see higher spending, x ^* _ic \??\<x ^* _ij , and this budget will be smaller than B_d above.

Under such a solution, the individual budgets which the ministers receive take the form

but the budget they would like to receive is

Note that, even if a minister herself sets the ideal budget in the negotiations so that x ^* _jc \??\=x ^* _js , that minister will still desire a higher budget than what she actually receives because she will not consider the total tax burden. Once again, the players are faced with a collective action problem. In this case, however, the monitoring and punishment functions are not needed to determine and implement the true costs of a given public good as is the case under one-party governments, but rather to enforce the negotiated budget.

One option that is not acceptable to the players is a strong finance minister. The coalition partners have different ideal budgets for each ministry, and as such there does not exist a collectively optimal budget per se which is also in the finance minister's best interests to enforce. We expect that one's distrust of a finance minister will increase as the difference between ideal budgets increases. A finance minister can, of course, promise to safeguard the budgetary targets. Yet the "principals" can never be sure that their "agent" is not placing the interests of his party above the interests of the coalition. A finance minister can choose to devote most of his resources to monitoring the other parties in the coalition, for instance, while leaving his party colleagues unmonitored. Even if he does use his resources equally, he may decide simply to overlook any transgressions made by his party.

A second problem with the strong finance minister model is that, even if he were designated to enforce the agreement, he lacks the ability to enforce it. Unlike the one-party case, the prime minister cannot give him as much meaningful support. The distribution of portfolios is, as far as the sitting prime minister is concerned, exogenously given, since coalition negotiations determine which parties get which ministries, and a prime minister cannot easily dismiss intransigent spending ministers. This power relationship takes away an important tool finance ministers can use to punish defecting ministers. While possible punishments could in theory be negotiated as part of the coalition agreement, in fact the punishment may be difficult to enforce, especially against ministers who are not of the same party as the finance minister.

Some sort of mechanism is still needed to monitor the actions of ministers and to punish those who overstep their targets, and due to space limitations we offer only a few preliminary speculations below. ₁₂ In this case, even weak finance ministers can act as an important source of information. Yet spending ministers from other parties will presumably want other monitoring devices as well. In general, there will be a greater concern with checks on what spending ministers do during the implementation stage of the budget. One solution is to create new structures inside the cabinet itself. Cabinet committees which overlap the responsibilities of different ministries and which are composed of at least one member from each coalition party allow ministers to monitor whether their partners are following the target agreement. ₁₃

An important institution outside of the cabinet is the parliament. Parliamentary committees in particular often provide an information collection function which can also be used to oversee the government (Krehbiel 1991; Mattson and Strom 1995; Hallerberg 1996). They hold hearings on the actions of the government and, in some cases, can call different ministers to justify their spending behavior. These committees are likely to be most effective when their area of expertise matches up with a particular ministry. They also should have chairpersons who come from a different party than the minister responsible for the given spending area.

Finally, there may be a greater desire to make budgetary procedures more transparent and more public in coalition governments that have negotiated budgetary targets than under one party governments. Mandatory reporting of expenditures reduces the costs for coalition partners to monitor each other. An aggressive press can then serve as a "fire alarm" to alert other coalition members of spending excesses (McCubbins and Schwarz 1984).

The punishing mechanism is less clear, and is the most problematic part of the use of targets. The members can pledge to each other that they will force the dismissal of a given minister if she breaks the coalition agreement. Other punishments, such as automatic cuts in a given minister's budget, are also possible. ₁₄ Yet smaller punishments are unlikely to be effective if the ultimate punishment, a loss of power for the party who breaks the agreement, is not a credible threat. We theorize that there are at least two important factors which affect the ability of a party to punish its coalition partner: the existence of plausible alternative coalition partners, and, if a new coalition cannot be formed and new elections are necessary, anticipation of electoral results.

If another partner exists with whom the aggrieved party can form a coalition, the threat to leave the coalition is clearly more credible. The number of parties in parliament is one obvious limit to the number of alternative coalition partners. Even among the parties which do exist, some may be undesirable for policy reasons or may not be considered koalitionsfähig, such as the Italian Communist Party. Other parties may simply be unexcludable from the coalition formation process. A party is "strong," according to Laver and Shepsle (1996), if it can veto every potential cabinet, and coalition partners may not be able to punish a party that occupies such a dominant position. Yet, to the extent that there are several possible coalitions, reputations will be important. Parties which are known not to keep coalition agreements will have problems finding partners, and as long as parties anticipate that none of them has a reasonable chance of winning an absolute majority of seats in the future they will value the possibility of cooperation in the future.

The threat of new elections may also scare a defecting party into meeting its targets. Yet new elections may hurt the party causing the dissolution instead of helping it. ₁₅ In most cases, however, one would expect that the elections would not even be called. The parties possess relatively good information on the possibilities of other coalitions and on their future electoral chances, and threats to leave the coalition will often simply not be credible. ₁₆

e. Comparison of a Strong Finance Minister and Negotiated Targets as Institutional Solutions

This discussion of the punishment mechanism leads to another point, namely the likelihood that a strong finance minister or targets can endure: The differences in the strength and coverage of a punishment under one-party and multi-party governments influence the likelihood that a punishment will serve as a credible deterrent. In the one-party case, the punishment is heavy for the minister who overspends, but generally light for the government as a whole. It is therefore relatively easy for the prime minister to enforce, and ministers who overspend can expect to be dismissed for the good of their political party. In the multi-party case, the punishment leads to the death of the government, and, for the reasons given above, a party may be reluctant to levy the punishment. One would therefore expect strong finance minister solutions to be more durable than targets.

In fact, it is likely that, if the parties believe that they can accept the more centralized solution of a strong finance minister, they will do so. The difference in the durability of the two institutions suggests that, other things equal, a government would prefer to use strategic dominance. In cases where two parties have policy goals that are close to each other, they may be willing to accept the more centralized solution of a strong finance minister over negotiated targets. Another reason for selecting strategic dominance over targets is that a strong finance minister is a more effective solution. While the finance minister can reduce or eliminate both the common pool problem and the bias towards higher spending, targets can deal with just the latter situation. Governments will therefore choose strategic dominance over targets when both institutions are available.

f. Minority Governments

The discussion so far has focused on majority governments, be they one-party or multi-party. What institutional form is possible under minority governments? Since just under a third of all postwar European governments have taken a minority form, this question is not trivial (Tsebelis 1995a). Although minority governments have received considerably less study than majority governments, there is little consensus in the literature about how such governments affect fiscal policy. Some authors consider them to be highly unstable and incapable of passing significant legislation. Edin and Ohlsson (1991), in their re-analysis of Roubini and Sachs' (1989) dataset, find that minority governments are more likely than any type of majority government to run large budget deficits. In contrast, Strom (1990) insists that such governments are relatively stable and that they do at least as well as majority parties on the effectiveness of their policies. Tsebelis (1995a and 1995b) even argues that there is no functional difference between one-party minority and one-party majority governments.

For the purposes of this paper, however, minority governments more clearly resemble multiparty majority governments rather than one-party majority governments. A strong finance minister solution to the common pool problem is not possible under minority situations. While a finance minister may be able to keep one-party minority cabinets in line, he does not have the same ability to punish recalcitrant legislators. The government's negotiations with the opposition are instead functionally equivalent to the coalition negotiations necessary under multi-party majority governments, which suggests that budget targets may represent a feasible way to keep budget deficits low. In fact, one can speculate that such targets imposed under minority governments may even have a higher chance of success than under multi-party majority governments because of the existence of certain parliamentary institutions in states where minority governments are common. Strom (1990) indicates that parties are more likely to go into opposition and to allow a minority to continue if they have the means to influence policy in parliament. His index of parliamentary strength measures the ability of parliaments to monitor governments, and he finds that his index is strongly correlated with the existence of minority governments. This index also indicates, of course, that parties are able to monitor whether minority governments keep their budget agreements with the parliamentary opposition. The punishment mechanism if a given government reneges is also clear--the opposition can simply vote to let the government fall, although, analogous to the multi-party majority situation, the stability of the situation depends on the ability of the minority government to find other legislative partners.

III. The Role of Electoral Systems

Electoral institutions strongly influence the likelihood of one party winning a majority of legislative seats and consequently having the ability to form a one-party majority government. One important factor is the number of parties; if there are few parties, there is a higher chance that one party can win an absolute majority, and an absolute majority is a virtual certainty in two-party systems. Several studies indicate that the number of effective parties in a given system is strongly and positively correlated with the number of representatives elected from each electoral district, known as district magnitude (Duverger 1954; Taagepera and Shugart 1989, 1993). Electoral systems with low district magnitudes distribute seats less proportionately than those with large district magnitudes, and lower proportionality usually favors larger parties. In Spain, for example, where the average district magnitude is just 6.73, the Socialist party was able to win 44.3% of the popular vote in the 1986 national elections but 52.6% of the seats in the Congress of Deputies. ₁₇ At the other extreme, the Netherlands has only one electoral district composed of 150 seats for the entire country, and a party that wins less than one percent of the national vote can gain seats in parliament. Other factors which affect proportionality include legal barriers which require a party to gain a certain percentage of the national vote to win legislative seats, the method used to apportion seats, and whether or not a second allocation of seats is used to reduce disparities at the district level. ₁₈

Table 1: Comparison of Electoral Systems within the European Union

Plurality systems, which elect only one representative per district, encourage a two-party system, and they are consequently most likely to have one-party majority governments. Proportional representation (PR) systems have more variation in their district magnitudes, though the magnitudes are always larger than those found in plurality systems. They tend to have more "effective" parties in parliament and are characterized by multi-party majority or either one-party or multi-party minority governments. Empirical evidence has consistently supported this relationship--Arend Lijphart, for instance, found that from 1945 through 1980 plurality systems had on average 2.1 effective parties while PR systems had 3.8 effective parties (Lijphart 1984, 161). ₁₉ Behind these figures is a result that should be emphasized and which will appear again shortly--the stronger the relationship between the proportion of seats won and the proportion of votes, the higher the number of effective parties. Thus, Spain's PR system, which sharply discriminates against small parties with its low district magnitude, should have fewer effective parties than the Netherlands, which has a high district magnitude.

Based on the plurality/PR distinction, what is the likelihood of one-party majority governments within the European Union? One unfortunate fact for comparison's sake is that only two of the fifteen member states, Great Britain and France, have pluralist electoral systems. Yet the variation in district magnitudes in PR systems does lead to some variation in the number of parliamentary parties as well. ₂₀

Table 1 compares the political systems of the European Union countries. A few points require clarification. First, PR systems do not translate the percentage of votes directly into the percentage of seats, and smaller parties often cannot gain entry into the legislature. We noted previously that district magnitude affects the number of political parties possible, and a logical comparison would be between this figure and the likelihood of one-party government. Yet such a comparison would be somewhat misleading--as the third column in Table 1 indicates, states sometimes have different district magnitudes at different levels of the allocation process. In addition, other factors including legal thresholds (such as Germany's requirement that a party win either 5% of the nation-wide vote or three seats by plurality vote) and rules for the allocation of seats (use of D'Hondt, etc.) can also favor larger parties over smaller ones. Arend Lijphart solves our problem of how to aggregate these institutional effects with his translation of such factors into an "effective threshold," which is the percentage of the national vote a party expects it must receive to gain any legislative seats. ₂₁

Second, while France had a plurality system in all parliamentary elections but those held in 1986, its use of two rounds of voting increases the effective number of parties in parliament. Unless a given candidate wins an absolute majority in the first round, a second round of voting is held. This process encourages several parties to run candidates in the first round and to form electoral coalitions for the second round. The predicted emergence of two strong blocks facing each other under plurality does still occur, however, since the UDF (Union pour la Démocratie Française) allies almost exclusively with the RPR (Rassemblement pour la République) while the Socialist Party works equally as often with the French Communist Party. France will therefore be treated as a "one-party government" in most cases later in this paper.

Table 1 confirms the general link among electoral institutions, the number of parties, and the likelihood of a one-party majority government for the European Union countries. The correlation between the effective threshold and the number of parties has the correct sign at -.46, and it jumps to -.60 if France is excluded from the sample. The most important figure is the correlation which links the occurrence of one-party majority governments with higher effective thresholds, and the correlation of .55 (.82 if France is excluded) indicates that this relationship is relatively strong. Since states which have low district magnitudes also have higher effective thresholds, this result indicates that plurality elections or PR systems with low district magnitudes are likely to have one-party majority governments. In contrast, PR systems with high districts magnitudes usually have either multi-party majority governments or minority governments.

IV. Comparison of Institutional Solutions

This section examines the choice of budgetary institutional tools within all European Union states. The statistical comparison, while unfortunately based on only fifteen cases, nonetheless indicates a strong relationship between one-party governments and strategic dominance solutions on the one hand and multi-party or minority governments and targets on the other. Table 2 summarizes the predicted institutions based on the prevalence of one-party majority government and the actual institutions that the countries used from 1981 to 1994, which are the years for which we have data available for all the current European Union member states. We expect that strategic dominance develops in any state where one-party majority governments are the norm. We therefore code the two states which had one-party government over 90% of the time, Greece and the United Kingdom, as potential strategic dominance states, as well as France for reasons mentioned earlier in the paper. The others are presumed to be able to use binding budgetary targets.

Of the three states expected to use strategic dominance, both France and Great Britain opted for this institution while Greece did not. A similarly strong pattern emerges for the states where multiparty coalitions are common. The Federal Republic of Germany, which has never had a one-party majority government, also uses strategic dominance. After accounting for Germany, seven of the remaining twelve states predicted to use targets did so, while the remaining five opted for neither of the solutions. Five of the six states which did not use one of these institutions were predicted to choose targets. The sample size is too small to make any statistical comparison conclusive, but this high failure rate among states where one-party governments are not the norm is consistent with the argument presented here. Both the monitoring and punishment functions are presumably harder to execute in multi-party governments than in one-party governments. Thus, out of fifteen cases only one case, Germany, went against our expectations.

Yet even this result for German is not as paradoxical as it might seem, and indeed it illustrates a more general point about under what conditions a strategic dominance solution might work in some multi-party governments. During the post-war period a major party formed a coalition with a smaller partner (either the FDP or the DP) every year except during the grand coalition between the two largest Fraktionen (the SPD and the CDU/CSU) from 1966 to 1969. During elections a vote for one of the coalition members was usually considered a vote in support of the government as a whole, and coalition partners generally pledged to continue the coalition if together they received a majority of the legislative seats. To the degree that the coalition partners see their electoral fortunes as being one and the same, spending ministers regardless of party persuasion may prefer a strong finance minister who can deliver lower deficits.

Table 2: Comparison of Predicted and Actual Institutional Solutions, 1981-94

Indeed, the conduct of the coalition partners during the grand coalition provides evidence that parties do act differently when they anticipate running against each other in the next election. When the SPD and the CDU/CSU formed the coalition, they both anticipated that it would last no longer than through the national elections in 1969, and during the elections they campaigned vigorously against each other. The Chancellor, Kurt Georg Kiesinger, was weak relative to other post-war chancellors, and the coalition parties negotiated most major decisions in the smaller Kressbonner Kreis composed of senior cabinet members from both sides. The finance minister during the time period, the Christian Social Union leader Franz-Josef Strauss, consequently did not have the freedom of action his predecessors had nor which his successors would have, and the coalition forced him to coordinate budget policy with the Social Democratic Economics Minister, Karl Schiller. Under such circumstances only budget targets were politically practical as a device to combat budget deficits, and indeed that is exactly what the coalition partners used. Schiller and Strauss together formulated the so-called "Mifrifi" (Mittelfristige Finanzplanung), which, among other things, detailed explicit spending targets for the coalition (Hildebrand 1984, 290) ₂₂ .

Further details from other states which either did or did not use one of the institutions make the connection between electoral institutions and budgetary institutions clearer. Compared to Germany, the United Kingdom and Greece represent more standard cases where we expect strategic dominance to be the likely budgetary institution of choice. The United Kingdom is the only state in the sample which uses a pure plurality system in its parliamentary elections, and according to the theoretical discussion it is a good candidate for a strategic dominance solution. Indeed, the structure of the budget process at the governmental level follows this form. The prime minister is exceptionally strong and can reshuffle the cabinet as well as appoint ministers almost at will. The chancellor of the exchequer is generally regarded as second in power only to the prime minister, and she is given the power to negotiate one-on-one with spending ministers about their budget allocations. If there is a dispute between the finance minister and a given spending minister, it goes to a committee composed of senior ministers without portfolio for consideration and not to the full cabinet for resolution. These ministers do not have budgets of their own, and a logrolling situation in favor of the spending minister is not possible. Since the senior ministers are appointed to consider the general interests of the cabinet as a well, they usually support the finance minister (Von Hagen and Harden, forthcoming).

In Greece one-party majority governments were also the rule, but in this case the government did not opt for a strategic dominance solution. The country uses a "preferential" form of proportional representation, which is designed to favor the largest parties in elections, and in practice only three parties usually win seats. The greatest distortion between the number of votes and number of seats occurred in the 1974 elections, when New Democracy received 54.4% of the votes but 72% of the seats (Alivizatos 1990, 137). Except during the short period when the conservatives allied with the small communist party, either the conservatives or the socialists have controlled the government with a one-party majority in parliament. Yet Greece has not developed framework in which the finance minister has significant power over the spending ministers. The full cabinet makes all major decisions related to the budget (von Hagen 1992, 63, 66).

The Netherlands and Italy illustrate well the use or non-use of budgetary targets. The Dutch have the most representative system in the European Union, and as a consequence have never had a one-party majority government during the post-war period. In contrast to her counterpart in the United Kingdom, the prime minister has little power. Negotiations among the parties during the formation of the coalition determine most items of importance, including the distribution of portfolios, and the prime minister consequently lacks the ability to remove defiant ministers. The prime minister also does not have the power to settle any disputes, and she votes in cabinet meetings only in cases of a tie. Instead of using a strong finance minister, the coalition negotiations inscribe into the coalition agreement explicit budgetary targets which constitute the fiscal contract among the parties. As expected, there are several institutional devices which promote the ability of parties to monitor each other's behavior. The legislature in particular serves an important oversight role. Committee jurisdictions are matched with specific ministries, and the committee chair is required to come from a different party than the one which provides the minister (Andeweg and Irwin 1993, 141). Parties also have the means to punish defectors. The same parties are likely to be potential coalition partners again, and, since there is little likelihood that any of them could win an absolute majority, the parties anticipate the need for a multiparty coalition government in the future. Unlike in the Italian case which follows shortly, there is also competition among them for positions in the government: with the exception of a few extremist parties which receive almost no parliamentary seats, all of the parties are potential coalition partners, and a given party which breaks a coalition agreement can be excluded from future governments. It is therefore in a party's best interest to assure that it cultivates a reputation as a party that keeps its coalition agreements.

In contrast, Italy before 1994 similarly had a proportional representation electoral system, but no such target agreements were negotiated among the coalition partners. Even if such agreements had existed, however, it is doubtful that they would have been credible. The one cabinet mechanism which could serve a monitoring function, the prime minister's cabinet secretariat, "has some monitoring responsibility, but this is neither very developed nor very effective" (Burch 1993, 125). Strom (1990, 75) indicates that the parliament did have oversight powers, but the Italian coalition partners lacked an effective punishment mechanism to use even if they did detect defections. While no party could expect to gain an absolute majority under the proportional electoral system, there was also a low probability that a given party could be excluded from a future coalition. With many parliamentary seats going to extremist parties, whose share of seats reached a high of 40% in the early 1970s, there were fewer coalition possibilities (Ibid., 160). Every government had to include the largest party in the center, the Christian Democrats, in order to have any chance of passing legislation, and coalitions often were composed of the same parties as their predecessors. No party could credibly threaten the Christian Democrats as long as the political system remained in force.

V. Budgetary Institutions and Deficit and Debt Levels

The use of these institutions also contributed to sounder fiscal policies, although debt level comparisons should be treated with some caution. Von Hagen and Harden (Forthcoming) indicate that if states lack an institutional solution to the common pool problem for any one of four characteristics of the budget process (the governmental, legislative, and implementation stages, as well as the informativeness of the budget draft) then budget deficits will be comparatively high. This paper examines the process for just one of the characteristics, the governmental stage, and hence the solutions discussed here will not by themselves always lead to lower deficits.

Nonetheless, even with these caveats Figure 1 indicates a striking difference between the states which used either targets or strategic dominance and those which chose neither of the institutions. The graph displays yearly deficit data for the period 1990-94, which are the five most recent years for which we have data. The states with the institutions had a much lower average yearly budget deficit of -2.7, whose difference was statistically significant at the .01 level from the much higher average budget deficit of -7.6 for the states without the institutions. All of the "no institutional constraint" states had an average yearly deficit that was larger than the highest average yearly deficit among the states with one of the institutions, the United Kingdom.

Similar figures also exist for the net total debt burden that the states carried. Once again one must be careful with interpreting the figures, since the debt levels indicate cumulative fiscal policy decisions that extend beyond the period 1990-94. Yet one may also anticipate that, everything else being equal, over time the states with either targets or strategic dominance will be able to adjust more readily to fiscal shocks than states that lack such institutions, and that the total debt figures will reflect this tendency. Indeed, the average size of net debt burden in states without the institutions is almost twice as high at 87.0% of GDP in comparison to 53.5% of states with one of the two budget mechanisms, with the difference in means significant at the .06 level.

Figure 1: Comparison of Average Yearly Budget Deficits, 1990-94

Of course, these comparisons are based on only five years, and they do not take into account other potentially important variables, such as changes in the economic health of a country or political variables such as possible partisanship effects or governmental instability. A more general claim of this paper is that the plurality/PR dichotomy is important because it affects the form of budgetary institutions which are politically feasible, but that it is the presence or absence of these institutions, rather than the plurality/PR distinction itself, which affects the budget balance. How does the model presented here compare to other explanations?

Table 3 presents preliminary ordinary least square results from a panel dataset for the fifteen current members of the European Union from 1981 to 1994. Our list of variables generally follows those provided in Roubini and Sachs (1989) as well as in DeHaan and Sturm (1994) with the important distinction that we add dummy variables for the presence or absence of a strong finance minister or budgetary targets. ₂₃ The first set of variables measure fluctuations in a given country's economy, and they are expected to have some impact in budget deficits regardless of the presence or absence of government policies meant to reduce public debt levels. Changes in Gross Domestic Product should improve the budget situation, while increases in the unemployment rate are likely to add to the size of the deficit due to automatic payments of unemployment compensation. In addition, changes in real interest rates affect the size of interest payments on the debt, and, if the real interest rate is higher than the real growth rate, interest payments will generally cause an increase in the total debt level. We therefore include a variable for the net change in debt servicing costs. ₂₄ A lag for our dependent variable, which is the change in the debt level, is also included to reduce autocorrelation in the model.

The second set of variables covers some of the most frequently cited political explanations. Consistent with Grilli, Masciandaro, and Tabellini (1991), a change in government, which is defined as either one or more changes in coalition partners or the occurrence of an election, is expected to increase the size of government debt. Roubini and Sachs (1989) also argue that multi-party majority and all minority governments face a prisoner's dilemma of whose constituency should bear the brunt of budget cuts, with the dilemma becoming worse as the number of parties in the coalition increases. Following Edin and Ohlsson (1991), we include dummy variables for the number of parties in a majority coalition government (either 2-3 parties or 4-5 parties) and for the presence of minority governments, with a one-party government equal to the case where the 2-3 party, 4-5 party, and minority government dummies all equal zero. One would therefore anticipate that the presence of any of these dummies would positively affect the debt level, with coefficients that increase as one moves from 2-3 party majority government to minority government.

The partisan hue of the government may also be important. The general expectation is that left-wing governments are more tolerant of larger budget deficits than right-wing governments. Yet previous empirical studies offer little guidance--Roubini and Sachs(1989) indicate that left-wing parties are associated with larger deficits. Alesina and Perotti (1995) pin the blame on center parties, while Borrelli and Royed (1995) consider right wing governments the least able to control deficits. To keep this study comparable with Roubini and Sachs (1989) and DeHaan and Sturm (1994), we code this variable as the percentage of cabinet seats held by left-wing parties in a given year. Finally, we include a dummy variable for the presence or absence of a strong finance minister or of targets.

The results of the regression are encouraging. The variables for the two budgetary institutions are both significant and have the correct sign. The more negative coefficient for a strong finance minister than for the budgetary targets also confirms the intuition that a strong finance minister is more effective in keeping the deficit lower than targets, other things being equal. Section 2 argues that a strong FM reduces both the common pool resource problem and budget-maximizing cabinet members while targets combat just the former, and these results fit that argument. All of the economic variables but the change in debt servicing costs also significant and have the anticipated sign. ₂₅ This is to be expected--if fluctuations in the economy did not have some sort of impact on the budget one would have reason to doubt the empirical results.

With the exception of changes to the government, other strictly political variables do not fair so well. The dummy variables for the contention that the form of government impacts the size of the budget are all insignificant, and the dummy for minority governments even has the wrong sign. The measurement for partisanship is also not significant, although its sign indicates that left-wing parties were more likely to reduce the size of the debt burden. Only a change in government has a significant impact on the growth of debt.

Table 3: Comparison of Alternative Explanations for the Growth or Decline of the Gross Debt Burden, 1981-94

One interesting possibility is that there is an interactive effect among the budgetary institutions and a change in government. Countries with the proper institutions may be able to isolate their fiscal health from political instability. Table 4 presents results which include interactive terms for a change in government with a strong finance minister and with negotiated targets. The significance of the two institutions on their own disappears, although they continue to have an impact in the expected direction. More importantly, however, the interactive terms are both negative and significant. The regression indicates that, in years where there was a change in government, the aggregate debt burden grew by almost three percentage points of GDP. If those states also had one of the budgetary institutions, however, that effect was eliminated. Thus, the negative consequences of political instability can be neutralized if a country puts in place either a strong finance minister or negotiated budget targets.

Table 4: Consideration of the Interaction of the Change in Government with Either a Strong Finance Minister or Budgetary Targets

Conclusion

The Maastricht Treaty's debt and deficit guidelines for states which want to join a future common currency may help to create a common interest in lower deficits in states where such a consensus has so far been lacking. Our statistical evidence indicates that the use of either a strong finance minister or negotiated budget targets can have a significant impact on the growth of the budget deficit. Such institutions are especially effective in keeping deficits down in countries were there is some political instability.

States which want to reduce their deficits should choose one of these budgetary institutions based on the form of government that they commonly experience, either one-party majority government or multi-party majority (or any minority) government. One-party governments are most suitable for strong finance minister solution, while multi-party governments have reason to rely on budgetary targets. The comparison of the various systems and solutions which are now used indicates that, under certain circumstances, the use of a strong finance minister can be expanded to multi-party governments. The key is that all the parties in the government see their electoral fortunes as one, as in France and Germany. This indicates that strategic dominance may soon be a viable solution for Italy. The new electoral system introduced in 1994, which relies on the plurality method for 75% of the seats in parliament, has led to two distinct constellations of parties. The presence of a center-left minority government indicates that targets may be the only feasible short-term solution, but if the electoral system continues to evolve and one of the two blocs can expect to win a majority of seats in future elections a strong finance minister may become a better choice.

In other problem states with multi-party governments, such as in Belgium and Portugal, a target-based approach will likely be the most practical route to solving the common pool problem. Coalition partners have little reason to support a strong finance minister because they doubt that the finance minister will safeguard the collective interests of all. In such states it is important that targets be made credible, and further research is needed to determine how targets can be made credible when the threat of a coalition collapse is not a realistic deterrent.

Bibliography

Alesina, Alberto, and Ricardo Hausman, Rudolf Hommes, and Ernesto Stein. "Budget Institutions and Fiscal Performance in Latin America." Unpublished Manuscript, September 1995.

Alesina, Alberto, and Roberto Perotti. "Fiscal Expansions and Adjustments in OECD Countries." Economic Policy (October 1995): 207-248.

Alivizatos, Nikos. "The Difficulties of `Rationalization' in a Polarized Political System: the Greek Chamber of Deputies." in Ulrike Liebert and Maurizio Cotta, eds. Parliament and Democratic Consolidation in Southern Europe: Greece Italy, Portugal, Spain and Turkey. London and New York: Pinter Publishers, 1990: 131-153.

Andeweg, Rudy, and Galen Irwin. Dutch Government and Politics. London: MacMillan, 1993.

Andeweg, Ruby. "The Netherlands: Cabinet Committees in a Coalition Government." in Thomas T. Mackie and Brian W. Hogwood, eds., Unlocking the Cabinet: Cabinet Structures in Comparative Perspective. London: Sage Publications, 1985: 138-154.

Bingham Powell, G. Contemporary Democracies--Participation, Stability and Violence. Cambridge: Harvard University, 1982.

Borrelli, Stephen A., and Terry J. Royed. "Government `Strength' and Budget Deficits in Advanced Democracies.' European Journal of Political Research 28 (1995): 225-260.

Christensen, Grønnegârd Jørgen. "In Search of Unity: Cabinet Committees in Denmark." in Thomas T. Mackie and Brian W. Hogwood, eds., Unlocking the Cabinet: Cabinet Structures in Comparative Perspective. London: Sage Publications, 1985: 114-137.

Cox, Gary W., and Mathew D. McCubbins. Legislative Leviathan. Berkeley and Los Angeles: University of California Press, 1993.

DeHaan, Jakob, and Jan-Egbert Sturm. "Political and Institutional Determinants of Fiscal Policy in the European Community." Public Choice. 80 (1994): 157-172.

Duverger, Maurice. Political Parties. Their Organization and Activity in the Modern State. New York: Wiley, 1954.

Edin, Per-Anders, and Henry Ohlsson. "Political Determinants of Budget Deficits: Coalition Effects Versus Minority Effects." European Economic Review 35 (1991): 1597-1603.

European Commission. Statistical Annex of European Economy. November 1995.

Franzese, Robert J. "The Political Economy of Public Debt: An Empirical Examination of the OECD Post-war Experience. "Paper Presented at the 1996 Midwest Political Science Association Meeting, 1996.

Frohlich, Norman, and Joe A. Oppenheimer. Modern Political Economy. Eaglewood Cliffs, New Jersey: Prentice-Hall, 1978.

Gallagher, Michael, Michael Laver, and Peter Mair. Representative Government in Western Europe. New York: McGraw Hill, 1992.

Grilli, Vittorio, Donato Masciandaro, and Guido Tabellini. "Institutions and Policies." Economic Policy. 6 (October 1991): 341-391.

Hahm, Sung Deuk, Mark S. Kamlet, and David C. Mowery. "The Political Economy of Deficit Spending in Nine Industrialized Parliamentary Democracies. The Role of Fiscal Institutions." Comparative Political Studies 29, no. 1 (February 1996): 52-77.

Hahm, Sung Deuk. "The Political Economy of Deficit Spending: A Cross Comparison of Industrialized Democracies, 1955-1990." Unpublished Manuscript, September 1994.

Hallerberg, Mark. "The Role of Parliamentary Committees in Monitoring Governmental Agreements on the Budget." Paper Presented at the International Conference on the Changing Roles of Parliamentary Committees, Budapest, Hungary, June 21, 1996.

Hallerberg, Mark, and Jürgen von Hagen. "The Budgetary Process and the Size of the Budget: A Reexamination." Unpublished Manuscript, July 1996.

Hildebrand, Klaus. Von Erhard zur Groβen Koalition 1963-1969. Stuttgart: Deutsche Verlags-Anstalt, 1984.

Huber, John. "The Vote of No Confidence in Parliamentary Democracies." American Political Science Review 90, no. 2 (June 1996): 269-282.

Jesse, Neal. "The Single Transferable Vote and Duverger's Law: Consequences for Party Systems and Elections. "Ph.D. Dissertation. The University of California, Los Angeles, 1996.

Katz, Richard. A Theory of Parties and Electoral Systems. Baltimore and London: The Johns Hopkins University Press, 1980.

Krehbiel, Keith. Information and Legislative Organization. 1991.

Laver, Michael, and Kenneth A. Shepsle. Making and Breaking Governments. Cambridge: Cambridge University Press, 1996.

Laver, Michael, and Kenneth A. Shepsle. "Cabinet Ministers and Government Formation in Parliamentary Democracies," in Michael Laver and Kenneth A. Shepsle, eds. Cabinet Ministers and Parliamentary Government. Cambridge: Cambridge University Press, 1994: 3-14.

Laver, Michael, and Norman Schofield. Multiparty Government. Comparative European Politics. Oxford: Oxford University Press, 1990.

Lijphart, Arend. Electoral Systems and Party Systems: A Study of Twenty-Seven Democracies 1945-1990. Oxford: Oxford University Press, 1994.

Lijphart, Arend. Democracies: Patterns of Majoritarian and Consensus Government in Twenty-one Countries, New Haven: Yale University Press, 1984.

Lupia, Arthur, and Mathew D. McCubbins. "Who Controls? Information and the Structure of Legislative Decision Making." Legislative Studies Quarterly 19, no. 3 (August 1994): 361-384.

Mackie, Thomas T., and Richard Rose. The International Almanac of Electoral History. Third Edition. London: MacMillan, 1991.

Mattson, Ingvar, and Kaare Strom. "Parliamentary Committees." in Herbert Döring, ed. Parliaments and Majority Rule in Western Europe. New York: St. Martin's Press, 1995: 249-307.

McCubbins, Matthew, and Thomas Schwarz. "Congressional Oversight Overlooked: Police Patrols vs. Fire Alarms." American Journal of Political Science. 28 (1984): 165-79.

Müller, Wolfgang C. "Models of Government and the Austrian Cabinet," in Michael Laver and Kenneth A. Shepsle, eds. Cabinet Ministers and Parliamentary Government. Cambridge: Cambridge University Press, 1994: 15-34.

OECD Economic Surveys 1994-95: Belgium/Luxembourg. Paris: OECD, 1995.

OECD Outlook 57 (June 1995).

Ostrom, Elinor. Governing the Commons. Cambridge: Cambridge University Press, 1990.

Persson, Torsten, and Lars Svennson. "Why a Stubborn Conservative Would Run a Deficit: Policy with Time Inconsistent Preferences." Quarterly Journal of Economics. (1989): 325-346.

Rogowski, Ronald. "Trade and the Variety of Democratic Institutions." International Organization. 41, no. 2 (Spring 1987): 202-223.

Roubini, Nouriel, and Jeffrey D. Sachs. "Political and Economic Determinants of Budget Deficits in the Industrial Democracies." European Economic Review. 33 (1989): 903-938.

Strom, Kaare. Minority Governments and Majority Rule. Cambridge: Cambridge University Press, 1990.

Taagepera, Rein, and Matthew Soberg Shugart. "Predicting the Number of Parties: A Quantitative Model of Duverger's Mechanical Effect." American Political Science Review. 87, no. 2 (June 1993): 455-464.

Taagepera, Rein, and Matthew Soberg Shugart. Seats and Votes: The Effects and Determinants of Electoral Systems. New Haven: Yale University Press, 1989.

Tabellini, Guido, and Alberto Alesina. "Voting on the Budget Deficit." The American Economic Review. 80, no. 1 (March 1990): 37-49.

Tsebelis, George. "Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism. Multicameralism and Multipartyism." British Journal of Political Science 25 (1995a): 289-325.

Tsebelis, George. "Veto Players and Law Production in Parliamentary Democracies." in Herbert Döring, ed. Parliaments and Majority Rule in Western Europe. New York: St. Martin's Press, 1995b: 83-111.

Velasco, A. "The Common Property Approach to the Political Economy of Fiscal Policy." Paper Presented to the Research Training Group in Positive Political Economy, MIT/Harvard, Cambridge, MA, 1995 (as cited in Franzese 1996).

Von Hagen, Jürgen, and Ian Harden. "Budget Processes and Commitment to Fiscal Discipline." IMF Staff Papers. Forthcoming.

Von Hagen, Jürgen, and Ian Harden. "Budget Processes and Commitment to Fiscal Discipline. European Economic Review. 39 (1994a): 771-779.

Von Hagen, Jürgen, and Ian Harden. "National Budget Processes and Fiscal Performance." In European Economy. Reports and Studies 3 (1994b): 315-418.

Von Hagen, Jürgen. "Budgeting Procedures and Fiscal Performance in the European Communities. Economic Papers 96 (October 1992).

Weingast, Barry, Kenneth Shepsle, and Christopher Johnsen. "The Political Economy of Costs and Benefits." Journal of Political Economy. 89 (1981): 642-664.

Woldendorp, Keman, and Ian Budge. "Political Data 1945-1990." European Journal of Political Research. 24 (1993): 1-120.

Notes

Note 1: Grilli, Masciandaro, and Tabellini also theorize that the existence of extremist parties in proportional representation systems should lead to higher levels of debt, but they do not find polarization to be statistically significant in their regressions. Back.

Note 2: De Haan and Sturm (1994), in their comparison of different explanations for the size of the deficit, find von Hagen's (1992) institutional variable statistically significant. Back.

Note 3: These "parties" can also correspond to different factions within one political party. Back.

Note 4: For a formal proof of this proposition see the appendix in von Hagen and Harden (Forthcoming). Franzese (1996) indicates that Velasco (1995) makes the same argument. Back.

Note 5: Von Hagen and Harden (Forthcoming). Back.

Note 6: Laver and Shepsle (1994, 9-10), for instance, in summarizing the findings of the case studies in their edited volume. note that the distribution of portfolios among members of the same political party has little effect on the policies which the government adopts; much more important is the distribution of portfolios among different parties. Back.

Note 7: In order to keep actors straight, the finance minister will be referred to as `him,' while the other ministers will be referred to as "her." Back.

Note 8: The prime minister does not have unlimited freedom, since the formation of a cabinet under a one-party government involves intra-party negotiations and agreements. Yet the prime minister does generally have some flexibility in deciding which faction will acquire which portfolio, as well as who will represent that faction in cabinet. Back.

Note 9: Lupia and McCubbins (1994) indicate that an agent will choose the principal's optimal policy if two conditions are met: the principal understands the implications of maintaining the current policy or accepting the agent's proposal, and the policy that is most favorable for the principal is the one that the agent proposes. Especially in cases where spending cuts are needed, the prime minister can clearly see the implications of continuing spending at current levels or accepting the finance minister's negotiated settlement, and both principal and agent alike have the same interest to reduce the budget deficit. With both conditions met, the finance minister makes the same proposal the prime minister would have had she had better information. Back.

Note 10: The "Red-Green Coalition" formed in 1995 in the German state of Northrhine-Westphalia, composed of the Social Democrats and the Greens, faces exactly this dilemma. Back.

Note 11: Of course, a solution which falls between the two extremes is also possible. The coalition partners could decide to negotiate targets for a subset of ministries which have the largest budgets while granting autonomy to those ministers whose demands on the total budget are minimal. Back.

Note 12: A future paper will present a developed model of monitoring and punishment mechanisms in multi-party governments that use spending targets. Back.

Note 13: Christensen (1985, 115), for instance, indicates that cabinet committees are more relevant under coalition governments than under one-party governments in Denmark. Back.

Note 14: Further research is needed to determine what punishments coalition partners have at their disposal. Back.

Note 15: For instance, the leader of the Austrian People's Party, Wolfgang Schüssel, forced the dissolution of his grand coalition with the Austrian Social Democrats in the fall of 1995 when his coalition partners refused to back spending cuts in social programs. To his regret, the December elections only strengthened the Social Democrat's representation in parliament. Back.

Note 16: Past coalitions and the general compatibility of party platforms provide guidelines for what sort of coalitions are feasible, while polls of the electorate indicate one's chances in an election if it is called immediately. Back.

Note 17: Thomas T. Mackie and Richard Rose (1991, 397, 399). The average district magnitude figures is reported in Lijphart (1994, 22). Back.

Note 18: A succinct summary is found in Gallagher, Laver, and Mair (1992, 153-159). Back.

Note 19: Other empirical studies that confirm this link include Lijphart (1994) and Taagepera and Shugart (1989 and 1993). Back.

Note 20: A measurement for number of parties considers the strength of parties as well as their absolute number. The measure that will be used here is for the effective number of parties in parliament and is taken from Mark Laakso and Rein Taagepera, as quoted in Lijphart (1994, 68). It is calculated as N_s = 1/Σ S² _i where N_s equals the effective number of parties in parliament and s_i equals the proportion of seats party i possesses in the legislature. Back.

Note 21: More specifically, the "effective threshold" is the mean of the lower threshold, which is the minimum proportion of votes needed to gain representation, and the upper threshold, which is the proportion of the votes where one is guaranteed to win a seat. Lijphart (1994: 25-30) provides further details on the creation of the index. Back.

Note 22: Similarly, the use of negotiated targets assumed a prominent role in Austria after the creation of a Grand Coalition between the People's Party and the Social Democrats in 1987. For a discussion of the need for negotiated policy goals in Austrian grand coalition governments see Müller (1994). Back.

Note 23: There are also some differences in the countries and years covered in the respective studies. DeHaan and Sturm base their regressions on the EC 12 from 1981 to 1989, while Roubini and Sachs consider 14 OECD countries from 1960 to 1985. Back.

Note 24: Following DeHaan and Sturm (1994, fn. 8, 169) we code this variable as (Nominal Long Term Interest Rate - Inflation Rate - Real GDP Growth Rate)^*Debt Level (t-1). Back.

Note 25: This is the possibility that there is a simultaneity problem because the GDP term appears in some form on both sides of the equation. We therefore redid the regressions with a new indicator for the change in GDP as follows. We regressed the change in GDP in time t on GDP at t-1 as well as on the average change in GDP at time t within Europe. We then used the fitted values as our independent variable to measure the effects of economic growth. The results were virtually identical to those reported here. Back.