The standard economic model of government debt posits a benevolent government that uses debt to finance capital accumulation or smooth the impact of natural and financial disasters or economic fluctuations. But in fact, decisions on debt and fiscal policy are made by politicians who may have in mind other issues, such as the result of the next election and the interests of their constituencies. If these considerations are important factors in decision making, one needs to formulate a different model of why governments go into debt and what determines debt levels and the evolution of debt over time.
A look at debt levels across countries reveals a wide dispersion that is not easily traceable to the need to smooth the impact of economic shocks. Figure 9.1 plots the distribution of debt over GDP during the 1995–2005 period and shows that most countries have debt-to-GDP ratios of around 50 percent, but the range goes from 0 to 200 percent. In the advanced economies, average public debt during the 1995–2005 period ranged between 3 and 140 percent of GDP and was characterized by a twin-peaked distribution of public debt (the light green line in Figure 9.1). This group includes three countries (Australia, Luxembourg, and Norway) with public debt below 25 percent of GDP and four countries (Belgium, Greece, Italy, and Japan) with public debt above 100 percent of GDP. In the group of developing and emerging market economies, average public debt over the 1995–2005 period was about 60 percent of GDP. But here also the dispersion was very large, with levels of public debt going from 0 to 200 percent of GDP. In fact, this group includes 16 countries with public debt lower than 25 percent of GDP and 13 countries with debt-to-GDP ratios higher than 100 percent.
The relationship between the level of development (proxied by GDP per capita) and debt ratios appears to be U-shaped rather than linear. At first, debt decreases with the level of GDP per capita, reaching a minimum at an income per capita of approximately $6,000, and then starts increasing again (Figure 9.2). The fit of the line to the data, however, is rather poor. Latin American and Caribbean middle-income countries have levels of per capita income that range between $3,000 and $7,000. Hence, they should be characterized by low levels of debt. However, the region includes countries with low levels of debt (like Chile, Colombia, and Mexico) and countries with high levels of debt (like Jamaica, Argentina, and Uruguay).
As it is difficult to reconcile the enormous dispersion in debt-to-GDP ratios with standard economic theories of public debt, economists have looked for explanations in the political arena (Alesina and Perotti, 1994). It is possible to organize the literature in this area into three groups of theories: (1) theories based on the opportunistic behavior of policymakers whose fiscal choices are intended to maximize voters’ support; (2) theories that emphasize the conflict among different politicians, or distributional conflicts between different groups in society; and (3) theories that highlight the importance of budget institutions. Clearly, the three potential determinants of fiscal choices emphasized by these three strands in the literature interact with each other. In fact, distributive conflicts between groups of voters affect fiscal choices partly because officials face opportunistic incentives and are not constrained by budget institutions that work well.
The Fiscal Choices of Opportunistic Policymakers
Early theories that emphasized the role of opportunistic policymakers relied on a mechanism in which voters value public spending but consistently underestimate its costs in terms of the tax burden, especially if those costs are postponed.[1] Thus, voters support policymakers who provide high levels of deficit-financed expenditures and oust incumbents who are fiscally conservative (Buchanan and Wagner, 1977). This literature has been criticized because of the assumption that voters make systematic mistakes (i.e., they are fooled over and over again by politicians). Opportunistic politicians, however, could be consistent with rational voters if the latter have imperfect information about the competence level of each politician and extract information about the competence of an incumbent running for re-election from his past fiscal choices. According to this class of models, an incumbent who has provided more government programs is inferred to be more competent, and this creates incentives for politicians to run deficits to finance larger expenditures (Rogoff and Sibert, 1988).
This literature has three empirical implications. First, voters should prefer high levels of public expenditure. Second, debt accumulation should be negatively correlated with the transparency of the budget. Third, in countries where fiscal outcomes cannot be observed by voters, electoral periods should be characterized by fiscal expansions. The empirical evidence, however, does not fully support the idea that voters like high levels of public expenditure. Alesina et al. (1998) find that governments that follow tight policies on expenditure are no more likely to be replaced than others. If anything, the opposite seems to be true. Voters’ attitude towards expenditure, however, is not independent of the type of government spending. Using data on local elections in Colombia, Eslava (2006) and Drazen and Eslava (2005) show that the share of votes received by an incumbent increases with capital expenditures.
There is, in contrast, clear evidence of a relationship between budget transparency and fiscal outcomes (Alt and Lassen, 2006) and of an interaction between lack of transparency and the possibility of opportunistic use of deficits during election times. Eslava (2006) uses a sample of developing and developed countries and shows a strong negative correlation between accountability and budget deficits. The relationship between accountability and deficit is consistent with the fact that in developing countries there are substantial pre-election increases in public expenditures (Schuknecht, 1994) and that the association between elections and deficits is due to the behavior of “new democracies” (Brender and Drazen, 2005). This indicates that political deficit cycles emerge only in contexts in which voters and the media have not yet developed the ability to efficiently monitor fiscal policy. However, the relationship between deficit and accountability is not purely driven by differences between developing and developed countries; it is also present when the sample is restricted to Latin American and Caribbean countries (Figure 9.3).
Distributional Conflicts, Electoral Systems, and Fiscal Policy
In a system with two parties with preferences for different publicly provided goods, an incumbent will find that there are at least two advantages to running a deficit. First, she will be able to devote resources to the types of public goods she prefers. Second, she will “tie the hands” of her successor because, if she is replaced, the cost of the deficit will fall disproportionately on the goods she values less (Tabellini and Alesina, 1990). A related argument arises in the case in which politicians differ in their preferences regarding the optimal size of the government. If faced with a high probability of being replaced, high-spending incumbents may run surpluses to force their successors into high expenditure levels, while low-spending incumbents may do the opposite (Persson and Svensson, 1989).[2]
Excessive fiscal deficits can also result from the presence of heterogeneous interests across groups of voters. If legislators making budget decisions represent geographic units interested in different government-funded projects and government revenues are centralized, each district internalizes the full benefit of specific projects, but only part of the cost. As a result, there is demand for overprovision of government projects (Weingast, Shepsle, and Johnsen, 1981). Similar common-pool problems have been studied by those who emphasize voracity effects (see Chapter 3).
A testable implication of these models is that deficits and debt accumulation should be positively related to the number of groups or districts that are effectively represented in the process of choosing the budget. Empirical research seems to confirm that electoral systems that result in more political cohesion and stability generate more fiscal discipline. Stein, Talvi, and Grisanti (1998) examine the relationship between different electoral systems and fiscal performance in a sample of Latin American countries, finding that electoral systems with more proportionality and a larger number of parties are associated with larger deficits. Figure 9.4 focuses on 17 Latin American and Caribbean countries and shows a negative correlation between a country’s deficit and the fragmentation of its legislature.[3]
If the distortionary costs of taxation increase in the level of debt, debt accumulates up to the point at which each group perceives that a new deficit will imply higher costs than benefits. Hence, high levels of debt can be a solution to the common-pool problem (Velasco, 1999). An implication of this model is that higher levels of debt should be associated with a higher probability of a country’s undergoing a fiscal adjustment. However, the evidence on the relationship between debt and fiscal adjustments is mixed. Stein, Talvi, and Grisanti (1998) find that debt accumulation in a given period is actually increasing in the initial level of debt. Alesina et al. (1998) and Gupta et al. (2004) find that, conditional on a fiscal stabilization’s being under way, the probability that the adjustment is successful is increasing in the initial level of debt. The results of the studies cited above also seem to suggest that no other political or institutional variable has a significant effect on the probability of a country’s undergoing an adjustment.
Budget Institutions
The way in which the factors discussed above end up shaping deficit and debt accumulation will depend on the constraints policymakers face when deciding on the budget. Some of those constraints are given by the political environment, as discussed above, while others relate to the set of rules, procedures, and practices according to which budgets are crafted. There are three types of rules that can be used to constrain politicians: numerical targets, procedural rules, and transparency rules. Filc and Scartascini (2006) study the evolution of budget reforms in Latin America and show that procedural rules continuously improved (according to scores on an index constructed by these authors) over the 1992–2004 period and that the budget process became more transparent from 1997 on (Figure 9.5).
Numerical targets may take the form of simple or cyclically adjusted balanced budget constraints. The advantage of a balanced budget constraint is its transparency, and the disadvantage is that it limits a country’s ability to conduct countercyclical policies. It is also possible to establish numerical rules that limit the government’s ability to borrow (see below). However, the government can often circumvent such rules by borrowing through state-run agencies not included in the main government’s budget (Poterba, 1994). While numerical targets are frequently imposed on subnational governments (for instance, most U.S. states have balanced budget rules), they are less common for national governments.[4]
Procedural rules are used to establish the functions and rights of the policymakers that participate in the budget negotiations. Existing procedural restrictions can be divided into those that prevent the legislative branch from increasing the total amount of expenditure and those that prevent changes in the levels of deficit. An important distinction is that between “hierarchical” and “collegial” rules. Hierarchical rules often concentrate budgetary power in the finance ministry inside the cabinet, and in the executive vis-à-vis the legislature. Collegial rules, in contrast, are those that allow greater representation of different interests in the budgetary process but risk generating overspending problems. Institutions can be more or less hierarchical at different stages of the budget process. In the drafting of a budget, hierarchical institutions limit the power of spending ministers and centralize drafting power in the minister of finance. At the voting stage, hierarchical institutions limit the legislature’s ability to modify the size of the budget proposed by the executive.[5] At the implementation stage, hierarchical institutions impose limits on the congress’s ability to propose ex post amendments aimed at modifying the size of the budget.
While more hierarchical procedural rules are likely to increase fiscal discipline, rules can be circumvented through creative accounting, and they will be effective only in the presence of a high degree of transparency. Transparency rules increase information flows and thus enhance other rules. For instance, transparency can increase the effectiveness of numerical and/or procedural rules by limiting the scope for creative accounting. Transparency laws often focus on making publicly available the maximum amount of data covering details on contingent liabilities, a clear explanation of the methodology used to construct projections for fiscal figures, and information on the level and composition of the stock of public debt at all levels of government.
Alesina et al. (1999) study the importance of budget institutions for 20 countries in Latin America and the Caribbean in the 1980s and early 1990s and find that countries with more stringent numerical targets, more hierarchical institutions, and more transparency exhibit lower deficits. Stein, Talvi, and Grisanti (1998) corroborate this finding, after controlling for the fragmentation of the electoral system. Their results indicate that both electoral systems and budget institutions have significant effects on fiscal performance. Recent work by Filc and Scartascini (2006) corroborates this result by showing a strong positive correlation between a country’s fiscal balance and (1) the presence of hierarchical rules (Figure 9.6) and (2) an overall indicator that measures the quality of the country’s budget institutions (Figure 9.7).
The Role of the Courts
A country’s judiciary is a key player in the fiscal policy arena because it can rule some elements of the country’s budget unconstitutional and hence effectively modify the government’s fiscal policy. Although this institutional feature has received limited attention in the formal literature on the political economy of fiscal policy, it has become a key issue in many countries (IDB, 2005b). In Colombia, for instance, there is an intense debate on the fiscal role of the constitutional court’s rulings (Box 9.1).
Figure 9.8 depicts judicial activism in eight Latin American countries and shows that Costa Rica, Colombia, Guatemala, and Brazil are the countries with the highest degree of judicial activism, Argentina is in an intermediate position, and Mexico, Chile, and Paraguay have the lowest degree of judicial activism. Eslava (2006) examines how judicial activism affects fiscal policymaking in Latin America, and her results strongly support the hypothesis that judicial activism is correlated with larger deficits. Not only are the results statistically significant, but they are also quantitatively important. Eslava’s point estimates suggest that, other things being equal, if Costa Rica and Colombia had the same levels of judicial activism as Mexico and Chile, they would observe an improvement in their fiscal balance of close to 3 percent of GDP. While these are exploratory results based on a small sample of countries, they suggest that a country’s courts do play a key role in its implementation of fiscal policy.
The Interaction between Political Failures and Financial Market Imperfections
While the literature surveyed above focuses on political failures, another strand of the literature on the behavior of emerging market debt has focused on the role of market failures.[6] Rochet (2006) discusses a simple but illuminating theoretical model that unifies these two strands of the literature and shows how financial market and political failures complement one another.
Rochet models an economy in which shortsighted politicians try to borrow as much as they can, while international financiers agree to lend a large amount because they anticipate that other lenders will lend again in the future (making repayment of the original loans more likely). Thus debt accumulates until the country cannot repay, at which point investors refuse to lend any more. There are thus three sources of imperfections: (1) governments are unable to commit to their future borrowing policies (government imperfection), (2) governments care only about the short-term consequences of their decisions (short-termism), and (3) financial markets are unable to provide complete contingent contracts (market incompleteness). Rochet begins by showing that the borrowing policy that would be implemented in the absence of these imperfections is characterized by constant government expenditure, higher levels of borrowing (or smaller debt repayments) during periods of low growth, and lower levels of borrowing (or larger debt repayments) during periods of high growth.[7] In other words, the optimal level of debt displays behavior consistent with a countercyclical fiscal policy. Next, Rochet shows that the joint presence of the three imperfections listed above leads to suboptimal procyclical debt policies in which governments borrow in good times and repay in bad times. Interestingly, he shows that the presence of politicians with short-term objectives is not a necessary condition for the suboptimal behavior described in the paper. This is because the optimal policy is not time consistent and, in the absence of a commitment mechanism, even politicians with long-term objectives will have an incentive to reoptimize in each period, borrow more in good times and less in bad times, and hence deviate from the ex ante first-best policy (Box 9.2 discusses the concept of time inconsistency). One possible solution would be to find a commitment mechanism (like the fiscal rule and fiscal responsibility laws described in Boxes 9.3 and 9.4). Another solution would be to write a state-contingent debt contract. For instance, Rochet (2006) shows that a GDP-indexed bond that makes higher payments in good times and lower payments in bad times can replicate the optimal policy even in the absence of a commitment device. In this sense, removing a financial imperfection (i.e., allowing for state-contingent debt contracts) can solve the problems that arise from a political imperfection. Summing up, Rochet’s (2006) results are that (1) if the government could commit to its future borrowing policy, it could implement the first-best policy through perfectly countercyclical borrowing and standard debt contracts, and conversely, (2) if contingent debt contracts were available, even a government without commitment power could implement the first-best policy.
Next, Rochet (2006) studies a case of extreme political instability, in which governments last for only one period and maximize the current level of public consumption without any consideration of future outcomes. He shows that under this extreme assumption, governments always borrow as much as they can, and hence indebtedness is completely determined by the market’s willingness to lend. This willingness to lend in turn depends on the market expectations of the government’s future income, but also on markets’ willingness to lend again in the future. Rochet shows that these simple assumptions lead to a situation in which debt grows continuously until the country becomes unable to pay and defaults. However, as ability to pay is partly determined by investors’ willingness to lend, the maximum amount of sustainable debt is determined not only by a country’s level of income, but also by the characteristics of the country in terms of growth and volatility. Hence, the model does not generate a situation in which the maximum amount of debt is just a fraction, constant for all countries, of the country’s GDP. This heterogeneity is an important component of the model and is consistent with the real-world observations that countries tend to have debt crises at very different debt levels (and with the fact that there is a very poor correlation between debt ratios and sovereign ratings, as shown in Figure 1.1). Using a simulation, Rochet (2006) shows that there are two key determinants of the probability of default and the maximum level of sustainable debt. The first is long-run GDP growth, and the second is the volatility of GDP growth. The effect of GDP growth is straightforward, as countries with higher levels of long-run growth can sustain higher levels of debt. The effect of volatility is less obvious. Contrary to the classic result of Eaton and Gersovitz (1981), in which higher volatility is associated with a higher level of sustainable debt (because volatility increases the insurance value of debt), Rochet finds a U-shaped relationship between volatility and the maximum amount of sustainable debt, a finding that seems to be in line with the empirical evidence presented by Catão and Kapur (2004).
One other interesting result of Rochet’s (2006) relatively simple setup is that political and financial imperfections may lead to a situation in which a country’s ability to borrow may reduce welfare because it increases (instead of decreasing) the volatility of public expenditure and generates a constant probability of debt crisis.[8] One reason that it is not clear whether preventing a country from borrowing may increase welfare is that in Rochet’s model, there are two justifications for sovereign debt: income smoothing and “front loading,” or benefiting in advance from future growth of government income. Political and financial failures prevent government from achieving the first objective (and generate a situation in which the ability to borrow increases volatility), but the second objective is partially attained.[9] He suggests that, even if one were to find that the first effect dominates the second, welfare could be improved through less extreme forms of policy intervention than preventing the government from borrowing. One such intervention is a constitutional reform that prevents the government from borrowing more than a certain fraction of current income. As welfare is presumably increased by the reform, at least up to the point at which the front-loading motive becomes dominant, such a reform would require an evaluation of the “optimal” level of debt.[10]
Rochet (2006) argues that a country could reach an even higher level of welfare through a second type of intervention: by combining a cap on borrowing and an insurance policy from the international financial institutions. With such a setup, a country that agrees to put a cap of this type into its constitution would then benefit from contingent credit lines (or equivalently, credit risk insurance) financed ex ante by actuarial premiums. The country could then borrow at a constant rate, face a lower probability of crisis, and pay lower spreads on its debt.[11]
