Fiscal Politics
Chapter

Chapter 9. Fragmented Politics and Public Debt

Author(s):
Vitor Gaspar, Sanjeev Gupta, and Carlos Mulas-Granados
Published Date:
April 2017
Share
  • ShareShare
Show Summary Details

Introduction

Rising public-debt-to-GDP ratios can be attributed to either large fiscal deficits or weak economic activity. Standard economic wisdom advocates following a countercyclical fiscal policy during recessions and letting public debt grow, and lowering it during economic expansions (Barro 1979; Lucas and Stokey 1983). However, data since the 1970s show that debt reductions in good times rarely compensate for debt accumulation in bad times.1 Perhaps other factors are at play, possibly of a political nature (Alesina and Passalacqua 2015)—the incentives to overspend tend to increase with the number of political actors involved in budget decisions.

This chapter uses data for 92 advanced and emerging market and developing economies during 1975–2015 to study the relationship between the key indicators of political fragmentation and changes in public debt.2 More precisely, the questions addressed are the following: Is higher political fragmentation associated with debt increases? Does the presence of veto players make it more difficult to lower debt?

Some scholars have focused on explaining political factors behind large cross-country differences in debt levels, while others have focused on short-term variations in debt ratios in a small sample of countries. One weakness with both approaches is that they do not align debt and political dynamics, which typically change every four to five years with the change of government. This analysis adopts instead a unique approach in two dimensions: first, the time frames are legislative periods (the span between two consecutive elections); and second, a large panel data set with ample variation across space and time is used. The advantage of following this approach is that it allows the effects that divided governments, fragmented legislatures, and ruling coalitions have on debt dynamics during their entire tenure to be encapsulated.

The analysis also focuses on the quality of institutions given that earlier studies have found corruption to be positively associated with the accumulation of public debt. Political fragmentation can thus have a distinctive impact on public debt dynamics in societies in which corruption is perceived to be high.

This analysis finds strong evidence that political fragmentation plays a prominent role in explaining public debt dynamics. The results are consistent with the hypotheses underlying both common pool and veto players theories. In addition, the chapter shows that the prevalence of corruption magnifies the effect of political fragmentation. The impact of political fragmentation on debt dynamics appears to be asymmetric, with larger and more significant effects during periods of debt decrease.

The chapter is structured as follows: The next section reviews the relevant literature and discusses the two principal theories in this area. The third section presents the empirical model and the data, and the fourth section discusses the main results. The fifth section explores the differential impact of political fragmentation on public debt dynamics based on the level of perceived corruption and the overall prevailing level of public debt. The final section summarizes and concludes.

Literature Review

Earlier literature (Barro 1979; Lucas and Stokey 1983; Aiyagari and others 2002) focused on explaining how the observed pattern of debt accumulation differs from the normative prescription. The school of public choice has argued that “fiscal illusion” and Keynesian policies were behind excessive deficits and resulting debt accumulation (Buchanan and Wagner 1977). Voters suffer from fiscal illusion in that they do not understand the notion of the intertemporal budget constraint and overestimate the benefits of current spending relative to the costs of future taxation. Keynesian policies prescribe spending and deficits during recessions, but the political process creates an asymmetry during expansions, not allowing for spending cuts and higher taxes, ultimately leading to an increase in the size of government and persistent deficits (Alesina and Passalacqua 2015).

Another strand of recent research focuses on the role of rational actors—voters, lobbyists, politicians, and bureaucrats—in causing fiscal outcomes to deviate from the optimal level. Political economy models that assume rational voters show that politicians may exploit only temporarily a certain degree of information asymmetry. Empirically, political budget cycles explain only a small departure from optimal policy around election times, especially in new democracies (Persson and Tabellini 2002; Brender and Drazen 2005; Drazen and Eslava 2010; Alesina and Paradisi 2014).

In contrast, the literature that studies public debt dynamics irrespective of the electoral calendar focuses on how the number of political actors may affect spending, deficits, and debt accumulation. Two main theories have been advanced to explain suboptimal behavior: common pool and veto players.

Common Pool

Weingast, Shepsle, and Johnsen (1981) argue that representative legislatures often pass budgets that give priority to local projects in districts they represent. Often referred to as pork-barrel spending, it is an increasing function of the number of electoral districts. Presented as the law of 1/n, where total public revenue is a common pool, 1, available to n representatives (policymakers or districts), which they overuse proportionally to n in distributing benefits. The deviations of fiscal policy from the optimal—which would maximize social welfare—will be greater when the number of actors who represent subsets of the national purse (that is, spending ministers and parties in government) increases. A larger number of actors who thus fail to fully internalize the costs of raising additional revenue will lead to higher-than-optimal levels of spending and deficit financing (Wehner 2010).

Veto Players

A government system with a large number of veto players and sharp ideological differences among them on policy options enhances policy stability—that is, it is difficult to change the status quo (Tsebelis 1995, 2000, 2002). The status quo then becomes the preferred policy choice of those involved. Changes will only materialize once a certain number of institutional or partisan actors agree. This stasis makes it difficult to adapt policy to changing circumstances. As the number of veto players increases, fiscal adjustment becomes slower, leading to suboptimal public debt accumulation (Roubini and Sachs 1989; Alesina and Drazen 1991; Spolaore 2004). Similarly, as the ideological distance between the government players increases, the likelihood of any policy change from the status quo decreases (Franzese 2005; Tsebelis and Chang 2004). The presence of a large number of veto players and sharp ideological polarization among them reduces the chances of agreeing on policy changes and stabilizing the magnitude of excessive public debt (Cox and McCubbins 2001; MacIntyre 2001; Mian, Sufi, and Trebbi 2014). In contrast to the common pool, the veto players model explains the changes in public debt rather than the actual level of public debt.

Empirical Model and Data

The econometric approach followed here relates cross-country variation in public debt to multiple aspects of political fragmentation, including common pool considerations and the influence of veto players (Franzese 2002, 2005; Battaglini 2011). Unlike in previous empirical models that test the impact of alternative political outcomes on annual changes in public debt (Kagan 2015), this analysis focuses, in addition, on the period between elections for a national legislative body, thus restricting the sample to countries and periods in which competitive elections have taken place. The exercise thus tests the impact of political fragmentation on changes in public debt that occurred in years between legislative elections. The reason for focusing on multiyear legislatures (which typically last four or five years between two consecutive elections) is that debt creation is ultimately a decision of parliaments. In countries where there is a debt ceiling (for example, the United States), Congress has to explicitly approve any new debt limit. In other countries, debt issuance is decided by the executive but is usually the result of parliaments not passing revenue-raising measures or approving excessive spending.

First, debt episodes (the change in government debt) between two legislative election periods are defined, using data from the World Bank’s Database of Political Indicators (DPI), which provides years in which those elections were held.3

Second, cross-country variations in public debt as a function of political fragmentation are modeled, controlling for the structure of the economy:

where ΔDit denotes the change in public debt, expressed relative to GDP, in country i = 1,. . .,N at period t = 1,. . .,L, as defined above; PF denotes political fragmentation variables; and X is a vector of controls. For the baseline results, equation (9.1) is estimated using ordinary least squares. The analysis covers 806 episodes of changes in public debt between legislative elections for 92 advanced and emerging market and developing economies for the period 1975–2015.

A concern in this literature is the existence of reverse causality. It is possible that at least some of the correlations uncovered in this chapter are instead generated by (1) an omitted driving variable (such as an economic crisis or stagnant growth) causing political fragmentation, an increase in public debt, and high levels of unemployment that make it difficult to reduce public debt without significant social cost; or (2) reverse causation whereby the need for fiscal consolidation engenders political polarization and fragmentation. By focusing on multiyear debt episodes, the econometric approach used in this chapter greatly reduces the likelihood of reversed causality. In addition, for robustness, an instrumental variable approach is subsequently used in estimating equation (9.1) using annual changes in public debt, with instruments based on lagged values of the political fragmentation variables (one electoral period back in time), and also including country fixed effects.

Data on gross general government debt, expressed relative to GDP, are drawn from the IMF’s historical public debt database.4Figure 9.1 shows the evolution of public debt over time and across advanced and emerging market and developing economies in the sample. Figure 9.1 indicates large accumulations of public debt in the 1970s and 1980s in both advanced and emerging market and developing economies, with debt accumulating at a pace of more than 2 percent of GDP on a yearly basis over that period. Alesina and Passalacqua (2015) discuss alternative hypotheses of political distortions behind this sharp increase in public debt among advanced economies during a peace period.5 This debt accumulation was followed by fiscal consolidation in the 1990s and a large part of the early 2000s that generally slowed or reduced debt accumulation. The financial crisis of 2008–09 again triggered accumulation of government debt, in particular among advanced economies.6

Figure 9.1.Change in Public Debt

(Annual averages, percent)

Sources: IMF, World Economic Outlook and International Financial Statistics databases; and Mauro and others 2013.

Data on political fragmentation relating to the common pool considerations are also drawn from DPI. Four alternative indicators from this database are considered to test the hypothesis. First, government terms characterized by larger parliamentary majorities are expected to react faster to the need for fiscal adjustment. To account for this, an indicator for margin of majority is included, which is defined as the fraction of parliamentary seats held by the government as a share of total seats. Second, the extreme situation is represented by the case in which the government party has an absolute majority (more than 50 percent of the seats) in the houses that have lawmaking powers, which is tested using a dummy variable control of parliament that takes the value one if this is the case and zero otherwise.7 Alternatively, a third indicator for executive polarization is tested, which measures the ideological distance between the executive’s party (left, right, or center orientation) and the other three largest parties’ orientation.8 Finally, an additional predictor of political fragmentation is cabinet fragmentation within the executive branch of government. To account for this, the analysis follows Perotti and Kontopoulos (2002) in considering an indicator for the size of the cabinet, measured by the number of ministries, from Seki and Williams (2014).

Data on political fragmentation relating to the veto players theory are drawn from several sources. The actual number of veto players in a given country considers “individual or collective decision-makers whose agreement is required for the change of the status quo” (Tsebelis 1995, 289). Similarly, the analysis includes checks and balances, from DPI, which measures the number of political players influencing the government’s decision making. In addition, the number of working days lost due to strike9 from the International Labour Organization’s Social Dialogue Database are considered as a proxy for social tensions, making policy changes more difficult to pass, and thus translating into larger public debt accumulation or slower debt reductions. The analysis also considers popular support, from the International Country Risk Guide, measuring the level of support for government (and its leaders), which facilitates the implementation of reforms. Finally, the old-age dependency ratio from the World Bank’s World Development Indicators (WDI) is included. This indicator is defined as the ratio of older dependents—people older than 64—to the working-age population (those ages 15–64), to account for possible rigidities in the speed of public debt adjustment, related to a growing share of age-specific public spending on health and pensions.

Figure 9.2 illustrates the impact of selected political fragmentation indicators on debt dynamics, suggesting that indeed larger fragmentation, related to both the common pool and the veto players theories, is generally associated with higher increases in public debt (or smaller reductions). Among advanced economies, the increase in public debt on average over legislative periods has been about 3 percentage points of GDP higher in countries with below-average margins of majority in the parliament. Among emerging and developing countries, public debt has decreased about 1 percentage point of GDP faster in countries with above-average majority in the parliament, or in countries where political polarization is low. Similar magnitudes are found when considering the number of ministries.10 For the selected indicators for the veto players theory, the increase in public debt has been about 2 percentage points higher in advanced economies facing a large number of days of strike while the decrease has been about 1 percentage point lower in emerging and developing countries. Also, the number of veto players leads to faster accumulation of public debt (about 1½ percentage points of GDP) among advanced economies, the only group for which this indicator is available. Finally, for both advanced and emerging market and developing economies, a clear positive correlation emerges between the average accumulation of public debt (horizontal axis) and the average old-age dependency ratio (vertical axis). The analysis also looks at the role of the traditional control variables, such as changes in tax revenue and government spending, that affect government debt dynamics.11 Data are drawn from the World Bank’s WDI and the IMF’s World Economic Outlook. The data include per capita GDP in constant U.S. dollars (and the change in per capita GDP); the share of agriculture in value added; the degree of trade openness, measured as the sum of the shares of imports and exports in GDP; consumer price index inflation (and the change in consumer price index inflation); the change in natural resources rents;12 the change in the nominal exchange rate; the change in the unemployment rate; and the share of social spending in GDP.

Figure 9.2.Indicators of Political Fragmentation and Changes in Public Debt

(Average change in public debt over periods, percent)

Sources: Escolano and others 2014; International Labour Organization; World Bank, World Development Indicators; and authors’ calculations.

Finally, the quality of institutions is also controlled for using the International Country Risk Guide corruption index, as well as institutional strength and quality of the bureaucracy indicators. The final sample size varies depending on the specification. Table 9.1 provides descriptive statistics. The list of countries included in the sample is provided in Annex 9.1.

Table 9.1.Descriptive Statistics
VariableAverageMinimumMaximumStandard Deviation
Debt to GDP (percent)50.200.97289.5537.47
Advanced Economies50.961.60283.9636.51
Emerging Market and Developing Economies49.560.97289.5538.25
Change in Debt to GDP0.15−117.25118.9110.64
Advanced Economies1.00−84.9793.436.10
Emerging Market and Developing Economies−0.57−117.25118.9113.30
Margin of Majority0.680.0310.21
Control of Parliament0.52010.50
Polarization0.52020.83
Number of Ministries26.04110113.08
Number of Opposition Parties3.16016811.27
Number of Veto Players2.29161.23
Checks and Balances2.931181.81
Popular Support2.2603.910.57
Days of Strike381.33012,765869.98
Old-Age Dependency Ratio12.223.7441.906.82
Legislative Election0.23010.42
Executive Election0.09010.29
Change in Natural Resource Rents0.04−20.5734.552.39
Inflation0.40−0.31156.064.41
Change in Nominal Exchange Rate0.85−1.002,626.7739.36
Trade Openness65.561.33809.2250.94
Per Capita GDP (log)10.895.8017.372.38
Social Spending to GDP (percent)25.33055.5114.62
Control of Corruption (percentile rank)53.53010030.15
Quality of Bureaucracy2.15041.17
Change in Unemployment Rate0.01−12171.14
Source: Authors’ calculations.
Source: Authors’ calculations.

Main Results

This section reports the results of estimating equation (9.1) for all episodes of changes in public debt between legislative elections during 1975–2015. Table 9.2 provides the basic results using ordinary least squares for political fragmentation related to the common pool theory; variables capturing the veto players theory are presented in Table 9.3. For robustness and ease of comparison with earlier literature, results based on annual changes in public debt, using an instrumental variable estimation approach, are presented in Annex 9.2 (Annex Tables 9.2.1 and 9.2.2). For brevity, the full set of control variables is only shown in the annex tables and omitted in tables in the main text. These control variables are generally significant in explaining changes in public debt and present the expected sign.

Table 9.2.Common Pool Theory: Legislative Periods
(1)(2)(3)(4)(5)
Margin of Majority−8.160**
(1.231)
Control of Parliament−6.300***
(2.077)
Polarization−1.071
(1.581)
Number of Ministries0.146***
(0.054)
Number of Opposition Parties0.025
(0.036)
R20.2860.3030.3030.5420.286
F-statistic7.748.258.2210.617.61
P-value0.0000.0000.0000.0000.000
Observations359360344238360
Number of Countries6161593261
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Table 9.3.Veto Players Theory: Legislative Periods
(1)(2)(3)(4)(5)
Number of Veto Players1.451*
(0.896)
Days of Strike0.002**
(0.0006)
Checks and Balances1.057**
(0.482)
Popular Support−5.964***
(2.221)
Old-Age Dependency Ratio0.397**
(0.184)
R20.4470.2530.1120.4000.274
F-statistic7.137.023.3810.428.59
P-value0.0000.0000.0000.0000.000
Observations99202763257348
Number of Countries1657616160
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.

The regressions reveal that the effect of political fragmentation on changes in public debt is generally significant and can be large. The estimated coefficients from the common pool theory indicators (Table 9.2) suggest, for instance, that less fragmentation in the parliament facilitates fiscal consolidation. For each additional 10 percentage points of parliamentary majority, there is an average public debt reduction of about three-quarters of a percentage point of GDP, with full control of the parliament leading to a reduction in public debt of 6 percentage points of GDP. In contrast, a more polarized political system (measured by the lack of majority and divergent political preferences) can induce larger debt accumulation: the estimated coefficient suggests that the maximum level of polarization creates an average differential in public debt increase of about 2 percentage points of GDP vis-à-vis minimum polarization, even though the estimated coefficient is not statistically significant. Finally, a more fragmented government, measured by the number of ministries, creates scope for faster debt accumulation as the size of the cabinet increases (although the estimated effect seems relatively small).

The indicators on the veto players theory (Table 9.3) are significantly correlated with changes in public debt during legislative periods, and the magnitude of their effect is strong. According to the estimated coefficients, each additional veto player generates an average increase in public debt during a legislative period of about 1.5 percentage points of GDP. Also, each additional 100 days of strike13 explain an increase in public debt of about 0.2 percentage point of GDP. In addition, each additional political actor influencing government decision making (an increase in the variable “checks and balances” by one-half standard deviation) leads to a faster accumulation of public debt by 1 percentage point of GDP. Interestingly, a decrease in popular support for the government (by 1 standard deviation) also leads to faster debt accumulation by about 3½ percentage points of GDP during a legislative period.

Finally, each 10 percentage point increase in the old-age dependency ratio contributes to an average increase in public debt of 4 percentage points of GDP.

Robustness Tests

The previous section showed that political fragmentation can have a sizable impact on public debt dynamics. It also showed that these results are robust to alternative definitions of the period under analysis (that is, public debt changes during multi-year periods between legislative elections and during annual changes in debt). This section assesses the robustness of the results to alternative specifications. It first looks at the sensitivity of the results to the simultaneous inclusion of both the common pool and veto players variables. It then explores the potentially differential impact of political fragmentation on public debt dynamics based on the level of perceived corruption and the prevailing level of public debt.14 Further robustness tests consist of isolating periods of debt increase and decrease, separately, and looking at the role of independent fiscal institutions in mitigating the impact of fragmentation. To address potential endogeneity concerns, all results presented in this section use an instrumental variable estimation approach, as discussed above.

Other checks were performed on the results of the previous section. In particular, to control for countries that have received debt relief under the Heavily Indebted Poor Countries (HIPC) initiative, HIPC countries were excluded from the sample. Also, given the importance of nominal GDP in driving public debt levels (especially in developing countries), the analysis considers the level of public debt, rather than the public-debt-to-GDP ratio, and controls for the change in GDP in the regressions. Relatedly, because increases in public-debt-to-GDP ratios capture many other factors beyond fiscal profligacy (including, for example, increases due to stock-flow adjustments), an alternative measure was considered using the general government’s primary balance. The results for these additional checks are qualitatively identical to those presented above and thus have been omitted to preserve space.

The first robustness test consists in exploring the relative importance of the different fragmentation hypotheses more closely. Table 9.4 reports the results of including both the common pool and veto players variables in the regression. For this exercise, only variables that are not highly correlated within each group are included.15 Results are qualitatively similar to those presented in the previous section, and the estimated coefficients are similar in magnitude, which further reinforces the importance of considering both aspects of political fragmentation.

Table 9.4.Common Pool and Veto Player Variables
(1)
Change in Debtt − 10.145***
(0.036)
Margin of Majority−4.775**
(2.477)
Number of Ministries0.037**
(0.020)
Days of Strike−0.001
(0.001)
Checks and Balances0.198*
(0.124)
Old-Age Dependency Ratio0.154**
(0.043)
Time Fixed EffectsYes
R20.517
F-statistic2.99
P-value0.000
Observations706
Number of Countries35
Source: Authors’ calculations.Note: Dependent variable is change in debt-to-GDP ratio. All control variables are included in all regressions. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable is change in debt-to-GDP ratio. All control variables are included in all regressions. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.

The next robustness test consists in assessing whether the effect of political fragmentation on public debt dynamics is influenced by the level of corruption. Earlier evidence has shown that the level of corruption can be positively associated with the level of public debt (IMF 2016; Cooray and Schneider 2013), either through a direct increase in public spending (Kaufmann 2010; Tanzi and Davoodi 2002), or indirectly by affecting its composition (Gupta, De Mello, and Sharan 2001; Mauro 1998), and by reducing the ability of a government to raise tax revenues (IMF 2016; Schneider, Buehn, and Montenegro 2010; Kaufmann 2010).

This analysis splits the sample on the basis of the World Bank corruption indicator, which reflects perceptions of the extent to which public power is exercised for private gain. Table 9.5 presents the results for countries belonging to the upper 50th percentile of perceived level of control of corruption (that is, those with the lowest level of corruption); Annex Table 9.2.3 presents results for countries in the lower 50th percentile (high corruption).

Table 9.5.Low Perceived Corruption
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Change in Debtt − 10.0590.0680.0700.0900.0010.353***0.170***0.0260.119***0.029
(0.102)(0.096)(0.096)(0.177)(0.021)(0.047)(0.026)(0.022)(0.046)(0.020)
Margin of Majority−4.266**
(2.142)
Control of Parliament−1.560**
(0.780)
Polarization0.223
(0.430)
Number of Ministries0.020
(0.019)
Number of Opposition Parties0.042
(0.068)
Number of Veto Players0.370*
(0.222)
Days of Strike0.001
(0.002)
Checks and Balances0.477**
(0.208)
Popular Support−0.713*
(0.515)
Old-Age Dependency Ratio0.052
(0.124)
Time Fixed EffectsYesYesYesYesYesYesYesYesYesYes
R20.1700.1700.2000.3720.1960.5190.2390.2070.3910.196
F-statistic2.322.592.381.522.841.803.582.341.433.27
P-value0.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
Observations1,5131,6111,5511,2581,5614391,2151,6055331,849
Number of Countries90918941922058918183
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.

A simple comparison of the results of indicators for the common pool theory shows that in countries with low perceived corruption (Table 9.5), less political fragmentation—as measured by a larger margin of majority or control of parliament—is negatively and significantly associated with changes in public debt. The opposite can be observed in countries with high perceived corruption (Annex Table 9.2.3), where a higher margin of majority or even the full control of parliament are not necessarily associated with reductions in public debt. Interestingly, a more fragmented government, measured by the number of ministries, is associated with much faster debt accumulation in countries with high perceived corruption: the estimated coefficient is some 40 times higher in countries with high perceived corruption as compared with countries with low perceived corruption and is highly significant.

The indicators on the veto players theory show a similar pattern. The estimated coefficients for the number of days of strike and the old-age dependency ratio are larger and more significant in Annex Table 9.2.3, implying a stronger link between political fragmentation and accumulation of public debt in countries with high corruption. Interestingly, higher popular support is associated with slower debt accumulation in countries with low perceived corruption but faster debt accumulation in countries with high perceived corruption (although this last coefficient is not statistically significant). The only exception is checks and balances, which is positively associated with increases in public debt only in countries with low perceived corruption.

A further robustness test consists in assessing the differential impact of political fragmentation on public debt dynamics while separately considering periods of decreasing public debt (Table 9.6) and periods of increasing public debt (Annex Table 9.2.4). A comparison of the results shows an apparent asymmetry. For periods in which public debt decreased, the results are largely as presented earlier, that is, more political fragmentation—through both common pool and veto players theory—is associated with slower reduction in public debt; less political fragmentation—as, for example, measured by a higher margin of majority or attaining control of parliament—is associated with faster debt reduction. However, most of the estimated coefficients are statistically insignificant for the common pool theory, meaning these results do not necessarily hold in periods in which public debt increased.

Table 9.6.Debt Decreases
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Change in Debtt − 10.0070.0100.0080.160***−0.060**0.129**0.070*−0.038*0.202***0.009
(0.092)(0.084)(0.084)(0.021)(0.020)(0.069)(0.027)(0.021)(0.046)(0.084)
Margin of Majority−4.778***
(1.735)
Control of Parliament−2.579**
(0.644)
Polarization1.494***
(0.358)
Number of Ministries0.034*
(0.018)
Number of Opposition Parties0.052*
(0.039)
Number of Veto Players0.466*
(0.275)
Days of Strike0.001
(0.001)
Checks and Balances0.412*
(0.228)
Popular Support0.116
(0.885)
Old-Age Dependency Ratio0.272***
(0.048)
Time Fixed EffectsYesYesYesYesYesYesYesYesYesYes
R20.1230.1290.1220.4190.1720.3910.1170.1740.2350.153
F-statistic4.004.073.792.453.893.695.693.751.473.73
P-value0.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
Observations1,1091,0801,0516191,1221656591,0664151,172
Number of Countries90928941922059918192
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.

Results for indicators on the veto players theory are similar for periods of both debt increase and decrease, except that as in the previous case, the estimated coefficients are smaller for periods of debt increase. Only the indicator for popular support behaves differently, and it is negatively and significantly correlated with periods of debt increase only, suggesting that weaker popular support can lead to faster debt accumulation.

An additional robustness check is performed by analyzing the differential impact of political fragmentation on public debt dynamics once initial conditions are accounted for. For this purpose, a debt-to-GDP threshold is defined using the sample average of 50 percent. This level of public debt is used as a threshold above which the economy becomes vulnerable to shocks, and with further increases in public debt potentially affecting economic growth more significantly.16 The underlying hypothesis is that countries with a debt-to-GDP ratio above 50 percent face a hard constraint that may prevent political fragmentation from affecting debt dynamics. The results for countries that are below the defined threshold are almost identical to those presented in the main results section and are not repeated here to preserve space. When focusing on periods for which countries are above the defined threshold, however, the results become much weaker (Annex Table 9.2.5), which confirms the underlying hypothesis. Only the interactive coefficient for higher fragmentation measured as the number of ministries is positively and significantly correlated with changes in public debt. Among the veto players indicators, only the interactive coefficient for popular support is significant but positively correlated with public debt changes, suggesting that weaker popular support at high debt levels leads to lower debt accumulation rates. In sum, higher political fragmentation has little impact on further increasing public debt once a high level of public debt has been accumulated. Interestingly, however, the opposite is also true, that is, less political fragmentation does not appear to be effective in accelerating public debt consolidation once that level of high public debt has been accumulated.

A final robustness check consists in analyzing how the existence of independent fiscal institutions affects the impact of fragmentation on public debt dynamics. A small but growing literature has argued that independent fiscal institutions, such as fiscal councils, could improve policymakers’ incentives to opt for sound fiscal policies even in the presence of political fragmentation (IMF 2013). First, by fostering transparency over the political cycle, a fiscal council can improve democratic accountability and discourage opportunistic shifts in fiscal policy (for example, preelection spending sprees). Second, through independent analysis, assessments, and forecasts, such bodies can raise public awareness of the consequences of unsustainable policy paths resulting from the presence of veto players, or contribute to a stability culture that directly addresses fiscal illusion linked to the common pool problem. Hence, a fiscal council can raise the reputational and electoral costs of unsound fiscal policies associated with political fragmentation. Third and finally, a fiscal council can provide direct inputs to the budget process—for example, forecasts or assessments of structural positions—thereby closing technical loopholes that allow governments to circumvent numerical fiscal rules.

Using the IMF Fiscal Council Dataset,17 a dummy variable is defined that takes the value one if the fiscal council in a given country has a score that is above the sample average and takes the value zero otherwise (or in the absence of a fiscal council). Annex Tables 9.2.6 and 9.2.7 present the results for strong and weak fiscal councils, respectively. The results suggest that the impact of fragmentation on the accumulation of public debt is stronger and more significant in countries without fiscal councils or in countries where fiscal councils are weaker than the average.18

Concluding Remarks and Policy Implications

This chapter focuses on the political determinants behind public debt dynamics. Using an empirical approach, it tests the role of traditional indicators of political fragmentation in explaining changes in public debt. The analysis both looks at annual data and introduces a selection of periods between consecutive legislative elections that is novel to the literature.

The results show that political fragmentation plays a prominent role in explaining public debt dynamics. The main theoretical hypotheses are confirmed—both common pool theory and veto players theory indicators show a positive association between political fragmentation and changes in public debt. In addition, we show that corruption magnifies these effects: in societies perceived to be corrupt, high political fragmentation has a sizable impact on debt increases. In contrast, low political fragmentation is not effective at reducing public debt in the presence of high corruption

Finally, the impact of political fragmentation on debt dynamics appears to be somewhat asymmetric, with larger and more significant effects of fragmentation in periods of debt decline. This finding only applies, however, to normal times, that is, when public debt is relatively low (less than 50 percent of GDP). For countries with high levels of public debt, political fragmentation cannot explain further increases in public debt. In addition, low political fragmentation appears to be ineffective in reducing public debt above that threshold.

The findings of this chapter are relevant for policymakers. An environment of political fragmentation is likely to be associated with excessive spending, deficits, and debt, regardless of whether such a policy stance is good or bad for the economy. This points to the need to strengthen fiscal institutions (fiscal rules and fiscal councils, in particular) to limit the impact that political fragmentation has on government spending. The use of a binding medium-term budget or fiscal framework could be considered, which sets, for instance, binding expenditure ceilings for a number of years, thereby constraining the ability of political players to influence fiscal policy. It points to the need for greater transparency in the decision-making process so that the public can better understand how fiscal and economic decisions are made in the short term and what their implications are in the long term.

Annex 9.1. Countries in the Sample

Advanced economies: Australia, Austria, Belgium, Canada, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Iceland, Ireland, Israel, Italy, Korea, Luxembourg, Netherlands, New Zealand, Norway, Portugal, Singapore, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, United Kingdom, United States.

Developing countries:19 Albania, Argentina, Armenia, Bolivia,* Brazil, Bulgaria, Burkina Faso,* Cambodia, Cameroon, Chad,* Chile, China, Colombia, Republic of Congo,* Croatia, Côte d’Ivoire,* Ethiopia,* Georgia, Ghana,* Haiti,* Honduras,* Hungary, India, Indonesia, Islamic Republic of Iran, Jordan, Kazakhstan, Kenya, Lao P.D.R., Latvia, Lithuania, Madagascar,* Mali,* Mexico, Moldova, Morocco, Mozambique,* Myanmar, Nepal, Nicaragua,* Nigeria, Pakistan, Peru, Philippines, Poland, Romania, Saudi Arabia, Senegal,* Sudan,* Tanzania,* Thailand, Turkey, Uganda,* Ukraine, Uzbekistan, Vietnam, Yemen, Zambia.*

Annex 9.2. Other Results
Annex Table 9.2.1.Common Pool Theory: Annual Data
(1)(2)(3)(4)(5)
Change in Debtt − 10.122***0.115***0.115***0.166***0.116***
(0.027)(0.026)(0.027)(0.029)(0.027)
Margin of Majority−5.871**
(2.659)
Control of Parliament−1.730**
(0.816)
Polarization0.092
(0.413)
Number of Ministries0.032**
(0.015)
Number of Opposition Parties0.002
(0.035)
Legislative Election0.0370.0140.0120.3370.061
(0.514)(0.522)(0.546)(0.440)(0.520)
Executive Election0.9470.8370.6101.302*0.971
(0.799)(0.805)(0.851)(0.789)(0.805)
Change in Oil Rents−0.291**−0.281*−0.266*1.029**−0.276*
(0.152)(0.155)(0.158)(0.413)(0.154)
Change in Unemployment Rate0.1990.1440.239−0.0290.235
(0.191)(0.184)(0.202)(0.199)(0.190)
Share of Social Spending0.1140.158**0.188**0.056**0.122
(0.082)(0.083)(0.087)(0.025)(0.084)
Quality of Bureaucracy−0.428−0.532−0.247−0.400−0.458
(0.894)(0.896)(0.897)(0.458)(0.921)
Corruption−0.373−0.374−0.407−0.389*−0.333
(0.473)(0.481)(0.483)(0.250)(0.480)
Inflation−0.281***−0.294***−0.287***−0.065−0.277***
(0.058)(0.056)(0.057)(0.051)(0.058)
Change in Inflation0.135***0.161***0.147***0.659***0.150***
(0.060)(0.057)(0.059)(0.087)(0.060)
Change in Nominal Exchange Rate0.116***0.117***0.123***0.113***0.113***
(0.014)(0.015)(0.015)(0.015)(0.015)
Trade Openness0.006−0.013−0.0100.003−0.013
(0.017)(0.017)(0.018)(0.004)(0.017)
Per Capita GDP0.071***0.072***0.086***0.0150.081***
(0.025)(0.026)(0.027)(0.111)(0.026)
Change in per Capita GDP−0.679***−0.700***−0.759***−0.948***−0.679***
(0.086)(0.084)(0.089)(0.103)(0.086)
Constant−75.256***−77.952***−95.313***−0.568−87.781***
(28.977)(29.329)(30.228)(2.103)(29.152)
Time Fixed EffectsYesYesYesYesYes
R20.2970.2900.2960.4930.283
F-statistic1.831.881.842.111.76
P-value0.0000.0000.0000.0000.000
Observations1,1531,2051,1477501,193
Number of Countries6161593261
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Annex Table 9.2.2.Veto Players Theory: Annual Data
(1)(2)(3)(4)(5)
Change in Debtt − 10.316***0.213***0.160***0.215***0.162***
(0.065)(0.027)(0.028)(0.037)(0.028)
Number of Veto Players0.837***
(0.306)
Days of Strike0.0005*
(0.0003)
Checks and Balances0.619***
(0.225)
Popular Support−0.932*
(0.533)
Old-Age Dependency Ratio0.112***
(0.040)
Legislative Election−0.0520.855−0.018−0.3640.397
(0.475)(0.611)(0.501)(0.578)(0.570)
Executive Election1.1940.4520.5401.931**0.744
(1.240)(1.038)(0.796)(0.827)(0.852)
Change in Oil Rents−1.2270.193−0.361**−0.419***−0.378**
(1.081)(0.337)(0.161)(0.138)(0.169)
Change in Unemployment Rate0.966***0.874***0.1640.449**0.321*
(0.292)(0.267)(0.182)(0.221)(0.202)
Share of Social Spending0.0340.0570.0890.0240.175
(0.157)(0.421)(0.085)(0.113)(0.137)
Quality of Bureaucracy−2.5611.165***−0.1861.342***0.896**
(1.870)(0.407)(0.865)(0.542)(0.411)
Corruption0.064−0.995***−0.266−1.200−1.078***
(0.465)(0.307)(0.452)(0.869)(0.310)
Inflation−0.333*−0.260***−0.207***−0.098−0.263***
(0.195)(0.057)(0.057)(0.097)(0.051)
Change in Inflation0.048***0.017*0.097*0.0150.195***
(0.017)(0.010)(0.061)(0.090)(0.068)
Change in Nominal Exchange Rate0.0070.028***0.137***−0.4350.139***
(0.016)(0.064)(0.017)(2.960)(0.019)
Trade Openness0.0140.008*−0.0060.015−0.001
(0.038)(0.005)(0.016)(0.024)(0.004)
Per Capita GDP0.0580.0990.064***0.093***−0.002
(5.504)(0.131)(0.025)(0.036)(0.119)
Change in per Capita GDP−0.586***−0.645***−0.708***−0.406***−0.549***
(0.186)(0.094)(0.082)(0.093)(0.089)
Constant4.109−4.641−72.019***−134.25***1.064
(57.826)(8.466)(28.100)(42.839)(1.886)
Time Fixed EffectsYesYesYesYesYes
R20.6750.5460.3160.3770.466
F-statistic1.061.251.861.261.38
P-value0.0000.0000.0000.0000.000
Observations2101,0631,1496551,119
Number of Countries1657616160
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Annex Table 9.2.3.High Perceived Corruption
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)
Change in Debtt − 1−0.003−0.0120.082**0.006−0.0240.009−0.0830.102**−0.019
(0.042)(0.095)(0.040)(0.060)(0.095)(0.095)(0.040)(0.054)(0.039)
Margin of Majority2.805
(3.471)
Control of Parliament−1.328
(1.837)
Polarization1.348
(1.168)
Number of Ministries0.748**
(0.415)
Number of Opposition Parties−0.0697
(0.0565)
Days of Strike0.011*
(0.006)
Checks and Balances0.128
(0.370)
Popular Support0.202
(0.438)
Old-Age Dependency Ratio0.434***
(0.127)
Time Fixed EffectsYesYesYesYesYesYesYesYesYes
R20.3800.2710.2620.9030.2560.5520.2870.3570.227
F-statistic1.812.102.091.291.931.162.002.051.64
P-value0.0000.0000.0000.0000.0000.0000.0000.0000.000
Observations54559158676589104558259611
Number of Countries50515075122514151
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Annex Table 9.2.4.Debt Increases
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Change in Debtt − 10.0120.065*0.085***0.207**0.078***0.222***0.189**0.076**0.057*0.091***
(0.089)(0.027)(0.028)(0.042)(0.030)(0.063)(0.084)(0.028)(0.038)(0.027)
Margin of Majority−2.593*
(1.862)
Control of Parliament0.750
(0.704)
Polarization−0.010
(0.418)
Number of Ministries−0.029
(0.026)
Number of Opposition Parties−0.070
(0.064)
Number of Veto Players0.206*
(0.014)
Days of Strike−0.001
(0.001)
Checks and Balances−0.132
(0.197)
Popular Support−0.656*
(0.422)
Old-Age Dependency Ratio0.245**
(0.113)
Time Fixed EffectsYesYesYesYesYesYesYesYesYesYes
R20.2240.2020.1950.3150.2130.4090.2300.2300.3270.187
F-statistic11.438.776.8025.0910.694.515.143.362.165.12
P-value0.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
Observations1,0521,1251,0867271,0853136581,0974831,288
Number of Countries91929042922060928292
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Annex Table 9.2.5.High Level of Public Debt
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Change in Debtt − 10.122***−0.0220.113***0.113***0.050*0.310***0.296***0.155***0.0090.115***
(0.027)(0.029)(0.027)(0.030)(0.027)(0.066)(0.047)(0.028)(0.061)(0.028)
Margin of Majority−0.027
(1.081)
Control of Parliament−0.814
(1.436)
Polarization0.506
(0.489)
Number of Ministries0.041**
(0.019)
Number of Opposition Parties0.188
(0.126)
Number of Veto Players0.152
(0.300)
Days of Strike−0.001
(0.002)
Checks and Balances0.209
(0.148)
Popular Support0.812**
(0.398)
Old-Age Dependency Ratio0.040
(0.041)
Time Fixed EffectsYesYesYesYesYesYesYesYesYesYes
R20.2590.2560.2570.3640.2580.7050.2600.3340.2580.217
Observations1,1531,0321,1477501,0032105831,1494781,156
Number of Countries61725932721641686161
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Coefficients represent the differential impact (the interactive term) for public-debt-to-GDP ratio above 50 percent. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = annual change in debt-to-GDP ratio. Coefficients represent the differential impact (the interactive term) for public-debt-to-GDP ratio above 50 percent. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Annex Table 9.2.6.Strong Fiscal Council
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Change in Debtt − 10.160***0.179***0.179***0.180***0.138***0.430***0.151***0.137***0.181*0.177***
(0.053)(0.056)(0.055)(0.073)(0.036)(0.080)(0.063)(0.035)(0.086)(0.056)
Margin of Majority−1.729*
(1.366)
Control of Parliament−0.676**
(0.389)
Polarization0.138
(0.247)
Number of Ministries0.034
(0.45)
Number of Opposition Parties−0.032
(0.063)
Number of Veto Players0.127
(0.225)
Days of Strike0.0002
(0.0003)
Checks and Balances0.103 (0.123)
Popular Support−0.851**
(0.457)
Old-Age Dependency Ratio0.004
(0.032)
Time Fixed EffectsYesYesYesYesYesYesYesYesYesYes
R20.4570.4620.4550.5630.5260.7270.4880.5280.6250.456
F-statistic11.7811.2310.926.413.1512.068.102.991.7011.61
P-value0.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
Observations589600589420501124426500120588
Number of Countries1616161116514161516
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Annex Table 9.2.7.Weakor No Fiscal Council
Common Pool TheoryVeto Players Theory
(1)(2)(3)(4)(5)(6)(7)(8)(9)(10)
Change in Debtt − 10.0170.0040.0140.190***0.0330.305***0.211**0.046*0.233***0.033
(0.023)(0.022)(0.082)(0.028)(0.080)(0.057)(0.092)(0.023)(0.035)(0.077)
Margin of Majority−2.059
(3.447)
Control of Parliament−1.617*
(0.965)
Polarization−0.062
(0.598)
Number of Ministries0.048*
(0.033)
Number of Opposition Parties−0.002
(0.018)
Number of Veto Players0.503*
(0.319)
Days of Strike0.0004*
(0.0002)
Checks and Balances0.429*
(0.290)
Popular Support−0.387
(0.585)
Old-Age Dependency Ratio0.122***
(0.033)
Time Fixed EffectsYesYesYesYesYesYesYesYesYesYes
R20.1800.1510.1510.3800.1350.5240.2400.1580.3560.145
F-statistic1.991.681.591.311.701.951.731.641.201.42
P-value0.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
Observations1,5701,7031,6449511,8613231,0461,7627331,875
Number of Countries75767430761549766776
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
Source: Authors’ calculations.Note: Dependent variable = change in debt-to-GDP ratio. Instrumental variables approach with instruments is based on lagged values of the dependent variable, including country fixed effects. All control variables are included in all regressions. Robust standard errors are in parentheses.*p < .1; **p < .05; ***p < .01.
References

    AiyagariS.A.MarcetT. J.Sargent and J.Seppala.2002. “Optimal Taxation without State-Contingent Debt.Journal of Political Economy110 (6): 122054.

    AlesinaA. and A.Drazen.1991. “Why Are Stabilizations Delayed?American Economic Review81 (5): 117088.

    AlesinaA. and M.Paradisi.2014. “Political Budget Cycles: Evidence from Italian Cities.Working Paper 20570National Bureau of Economic ResearchCambridge, Massachusetts.

    AlesinaA. and A.Passalacqua.2015. “The Political Economy of Government Debt.Working Paper 21821National Bureau of Economic ResearchCambridge, Massachusetts.

    BaldacciE.B.ClementsS.Gupta and Q.Cui.2008. “Social Spending, Human Capital, and Growth in Developing Countries.World Development36 (8): 131741.

    BarroR. J.1979. “On the Determination of the Public Debt.Journal of Political Economy87 (5): 94071.

    BattagliniM.2011. “The Political Economy of Public Debt.Annual Review of Economics3 (September): 16189.

    BaunsgaardT. and M.Keen.2010. “Tax Revenue and (or?) Trade Liberalization.Journal of Public Economics94 (9-10): 56377.

    BrenderA. and A.Drazen.2005. “Political Budget Cycles in New Versus Established Democracies.Journal of Monetary Economics52 (7): 127195.

    BuchananJ. M. and R. E.Wagner.1977. Democracy in Deficit: The Political Legacy of Lord Keynes.Indianapolis, Indiana: Liberty Fund, Inc.

    CoorayA. and F.Schneider.2013. “How Does Corruption Affect Public Debt? An Empirical Analysis.WP 1322 Department of Economics Johannes Kepler University of Linz.

    CoxG. and M.McCubbins.2001. “The Institutional Determinants of Economic Policy Outcomes.” In Presidents and Parliaments edited by S.Haggard and M.McCubbins.Cambridge, U.K.: Cambridge University Press.

    CrivelliE. and S.Gupta.2016. “Does Conditionality in IMF Programs Promote Revenue Reform?International Tax and Public Finance23 (3): 55079.

    DebrunX. and T.Kinda.2014. “Strengthening Post-Crisis Fiscal Credibility—Fiscal Councils on the Rise. A New Dataset.Working Paper 14/58International Monetary FundWashington.

    DrazenA. and M.Eslava.2010. “Electoral Manipulation via Voter-Friendly Spending: Theory and Evidence.Journal of Development Economics92 (1): 3952.

    EasterlyW.2014. The Tyranny of Experts. New York: Basic Books.

    EscolanoJ. and V.Gaspar.2016. “Optimal Debt Policy under Asymmetric Uncertainty.Working Paper 16/178International Monetary FundWashington.

    EscolanoJ.L.JaramilloC.Mulas-Granados and G.Terrier.2014. “How Much Is a Lot? Historical Evidence on the Size of Fiscal Adjustments.Working Paper 14/179International Monetary FundWashington.

    FranzeseR.2002. “Electoral and Partisan Cycles in Economic Policies and Outcomes.Annual Review of Political Science 5: 369421.

    FranzeseR..2005. “Empirical Strategies for Various Manifestations of Multilevel Data.Political Analysis13 (4): 43046.

    GuptaS.L.De Mello and R.Sharan.2001. “Corruption and Military Spending.European Journal of Political Economy17 (4): 74977.

    International Monetary Fund (IMF). 2013. “The Functions and Impact of Fiscal Councils.Staff Discussion Note 16/5 May Washington DC. https://www.imf.org/external/np/pp/eng/2013/071613.pdf.

    International Monetary Fund (IMF). 2016. “Corruption: Cost and Strategies.Staff Discussion Note 16/5 Washington DC. http://www.imf.org/external/pubs/cat/longres.aspx?sk=43888.

    KaganE.2015. “Testing Competing Interpretations of Public Debt: An Empirical Analysis of the Veto Player and Common Pool Theory in a Transnational Context.Paper presented at the Achievements and Challenges for the Emerging Economies of Central Europe Economics and Business Conference London June2224.

    KaufmannD.2010. “Can Corruption Adversely Affect Public Finances in Industrialized Countries?Brookings Institution Washington DC. http://www.brookings.edu/research/opinions/2010/04/19-corruption-kaufmann.

    LucasR. E. and N. L.Stokey.1983. “Optimal Fiscal and Monetary Policy in an Economy without Capital.Journal of Monetary Economics12 (1): 5593.

    MacIntyreA.2001. “Institutions and Investors: The Politics of the Economic Crisis in Southeast Asia.International Organizations55 (1): 81122.

    MauroP.1998. “Corruption and the Composition of Government Expenditure.Journal of Public Economics69 (2): 26379.

    MauroP.R.RomeuA.Binder and A.Zaman.2013. “A Modern History of Fiscal Prudence and Profligacy.Working Paper 13/05International Monetary FundWashington.

    MianA.A.Sufi and F.Trebbi.2014. “Resolving Debt Overhang: Political Constraints in the Aftermath of Financial Crises.American Economic Journal: Macroeconomics6 (2): 128.

    PerottiR. and Y.Kontopoulos.2002. “Fragmented Fiscal Policy.Journal of Public Economics86 (2): 191222.

    PerssonT. and G.Tabellini.2002. “Political Economics and Public Finance.” In Handbook of Public Economics edited by J. B.Taylor and M.Woodford.Amsterdam: Elsevier.

    ReinhartC. and K.Rogoff.2010. “Growth in a Time of Debt.American Economic Review: Papers and Proceedings100 (May): 57378.

    RodrikD.1998. “Why Do More Open Economies Have Bigger Governments?Journal of Political Economy106 (5): 9971032.

    RoubiniN. and J. D.Sachs.1989. “Political and Economic Determinants of Budget Deficit in the Industrial Democracies.European Economic Review33 (5): 90338.

    SchneiderF.A.Buehn and C.Montenegro.2010. “Shadow Economies All over the World: New Estimates for 162 Countries from 1999–2007.Policy Research Working Paper 5356World BankWashington.

    SekiK. and L. K.Williams.2014. “Updating the Party Government Data Set.Electoral Studies34: 27079.

    SpolaoreE.2004. “Adjustments in Different Government Systems.Economics and Politics16 (2): 11746.

    TanziV. and H.Davoodi.2002. “Corruption, Public Investment, and Growth.” In Governance Corruption and Economic Performance edited by G.Abed and S.Gupta28099. Washington: International Monetary Fund.

    TsebelisG.1995. “Decision Making in Political Systems: Veto Players in Presidentialism, Parliamentarism, Multicameralism, and Multipartyism.British Journal of Political Science25 (3): 289326.

    TsebelisG.2000. “Veto Players and Institutional Analysis.Governance13 (4): 44174.

    TsebelisG.2002. Veto Players: How Political Institutions Work. Princeton NJ: Princeton University Press.

    TsebelisG. and E. C. C.Chang.2004. “Veto Players and the Structure of Budgets in Advanced Industrialized Countries.European Journal of Political Research43 (3): 44976.

    WehnerJ.2010. “Cabinet Structure and Fiscal Policy Outcomes.European Journal of Political Research49 (5): 63153.

    WeingastB. R.K. A.Shepsle and C.Johnsen.1981. “The Political Economy of Benefits and Costs: A Neoclassical Approach to Distributive Politics.Journal of Political Economy89 (4): 64264.

The authors are grateful to Suman Basu, Era Dabla-Norris, Vitor Gaspar, Fabien Gonguet, Laura Jaramillo, Joao Jalles, Javier Kapsoli, Constant Lonkeng, Sanaa Nadeem, Tigran Poghosyan, John Ralyea, Christiane Roehler, Manrique Saenz, and participants at a FAD seminar for many helpful suggestions.

According to Escolano and Gaspar (2016), this debt accumulation bias is a relatively recent development, starting after the 1970s. Before that, debt spikes were typically followed by similar periods of debt decline. In their words, “Over the past two and a quarter centuries, the path of government debt ratios in the United Kingdom and the United States were characterized by occasional large increases prompted by national emergencies such as wars and large economic crises followed by long periods of sustained reductions in debt ratios” (Escolano and Gaspar 2016, 6).

Comprising emerging and low-income economies. Annex 9.1 lists the countries.

Results are broadly similar for electoral periods that exclude instances of early or repeated elections.

Originally compiled in Mauro and others (2013) and updated using the IMF’s World Economic Outlook and International Financial Statistics data.

Easterly (2014) suggests that in the early 1970s, many countries did not internalize a secular growth downturn requiring a reduction of government spending to keep the size of government constant, which ultimately led to an accumulation of debt.

Descriptive statistics in Figures 9.1 and 9.2 do not differ significantly once countries that have received debt relief under the Heavily Indebted Poor Countries (HIPC) initiative are excluded.

In addition, a fifth indicator for the number of opposition parties has been considered.

The variable takes the value zero if the legislative index of political competitiveness or the executive index of political competitiveness—both from DPI—are less than 6 (elections are not competitive) and if the chief executive’s party has an absolute majority in the legislature.

This indicator counts the days of strike in major economic sectors. If two or more economic sectors conduct strikes in a given day, then the indicator adds these sectors together, which may result in more than 365 days of strike in a given year.

The groups (low/high) for the number of ministries and the number of veto players are computed using the average, plus or minus one standard deviation.

This variable from the World Bank’s World Development Indicators captures the sum of natural resource rents from oil, gas, coal (hard and soft), minerals, and forests, expressed as a percentage of GDP.

Note that the average number of strike days in the most recent legislature period under study was 100 days, including all sectors in the economy, as reported by the International Labour Organization.

In addition to controlling for the level of GDP per capita in all regressions, a further test consisted in exploring the potentially differential effect based on the level of development of the country, by splitting the sample into OECD versus non-OECD countries. The results were inconclusive and have been omitted to preserve space.

We also exclude variables with only a limited number of observations, such as the number of veto players, which is only available for advanced economies.

Although below the 60 percent threshold included as a criterion (upper limit) in the European Union’s Stability and Growth Pact (and also used in Reinhart and Rogoff [2010]), the 50 percent threshold is likely more relevant for developing countries. Results using a 60 percent threshold, however, do not differ significantly.

See http://www.imf.org/external/np/fad/council. The IMF Fiscal Council Dataset describes key features of 39 institutions identified as fiscal councils as of 2014 across the IMF membership. The data set includes general information such as the name and acronym of the council and its date of creation, the main features of the council’s remit, their specific tasks and instruments to influence the conduct of fiscal policy, as well as key institutional characteristics such as the existence of formal guarantees of independence, accountability requirements, and human resources. Debrun and Kinda (2014) provide the list and definition of variables included in the Fiscal Council Dataset. They also describe the variety of sources used to assemble the data.

The establishment of fiscal councils is usually preceded by the adoption of fiscal rules, in many instances to ensure that these rules are followed. As such, the identified impact could be driven by the existence of fiscal rules. We also considered the potential impact of fiscal rules in mitigating the impact of political fragmentation. Results are qualitatively similar but significance generally weakens, potentially suggesting that fiscal rules can indeed mitigate the impact of political fragmentation and help the development of sound fiscal frameworks.

Countries denoted with an asterisk are those that have benefited from the Highly Indebted Poor Countries initiative.

    Other Resources Citing This Publication