Journal Issue
Share
Article

Front-Loaded or Back-Loaded Fiscal Adjustments

Author(s):
Benedict Clements, Sanjeev Gupta, Emanuele Baldacci, and Carlos Mulas-Granados
Published Date:
August 2004
Share
  • ShareShare
Show Summary Details

I. Introduction

In 1981, only five countries—now referred to as emerging market economies—reported a primary fiscal balance consistent with a nonincreasing ratio of public debt-to-GDP in the medium term. In the following two decades, 25 emerging market economies, including these five, undertook more than 100 episodes of fiscal adjustment. By 2001, the number of countries with sustainable public finances had grown to 11, although none of these countries generated primary fiscal balances sufficient to thwart an increase in the debt-to-GDP ratio over the two decade period. This raises the issue of why episodes of fiscal adjustment did not lead to a durable fiscal position consistent with a nonincreasing debt-to-GDP ratio.

Previous studies have focused on the factors affecting the duration of fiscal consolidations in emerging market economies (Adam and Bevan, 2003; and Gupta and others, 2003). These papers indicate that the probability of ending a fiscal adjustment is affected by the legacy of previous fiscal failures, the size of the deficit, and the composition of public spending. Although these studies deal with the determinants of the duration of fiscal adjustment, they only partially answer the question of how governments can achieve and maintain fiscal sustainability. In particular, a key policy issue is whether countries are more likely to achieve fiscal sustainability with front-loaded adjustment strategies—where the pain is endured upfront—or more gradual approaches, which may be politically and socially more acceptable. A related issue is whether consolidations based on trimming expenditures are more likely to succeed than those based on a strengthening of the revenue effort. Finally, the role of political and institutional factors in explaining the success or failure of fiscal adjustments in these countries has not been fully explored.

This paper attempts to answer these questions by investigating the fiscal adjustment episodes experienced in 25 emerging market economies between 1980 and 2001.3 A successful fiscal adjustment is defined as one that brings a country’s primary budget balance to the level that ensures debt sustainability (Adam and Bevan, 2003). We assume that the goal of fiscal adjustment is to achieve fiscal sustainability, as defined by a nonincreasing ratio of public debt-to-GDP.4

We find that large adjustments increase the likelihood of success. Back-loaded adjustments tend to be more successful in reaching sustainability but are not influential in maintaining it. Expenditure cuts increase the probability of approaching and reaching the threshold but are insufficient to maintain that state for two years or more. Adjustment episodes launched in countries where the government has a majority in the parliament, and when elections are not imminent, are also more likely to succeed. Finally, those consolidations undertaken during IMF-supported programs tend to preserve a country’s fiscal sustainability over a longer time period.

The paper is organized as follows: Section II defines successful fiscal adjustment episodes, and Section III discusses potential economic and political institutional factors that could have an impact on fiscal adjustments. Section IV presents the model specification, and Section V reports the main empirical findings. Section VI discusses the policy implications of the econometric results, and Section VII concludes.

II. Defining Fiscal Adjustment

An episode of fiscal adjustment is defined as a year (or set of years) in which the general government primary budget balance improves by at least 0.5 percentage point of GDP per year.5 This definition, therefore, does not limit us to large fiscal adjustments. A focus on large fiscal adjustments is common in the related literature on the OECD countries (e.g., Alesina and Perotti, 1995) and emerging economies (IMF, Fiscal Affairs Department, 2004), but such an approach does not allow the inclusion of small but prolonged adjustments that are prominent in our sample of consolidation episodes. Therefore, our sample includes adjustment episodes that differ in size, as the latter, in itself, could be an important factor for explaining the likelihood of reaching and maintaining fiscal sustainability. This permits a comparison of the relative success attained by front-loaded vs. back-loaded adjustments, and for “big-bang” vs. gradual consolidations, as defined below.

Front-loaded adjustments are defined as those where more than 50 percent of the total deficit reduction was achieved in the first half of the time period covered by the episode. Otherwise, the adjustment is considered a back-loaded one. Big-bang episodes of adjustment are those in which the average annual improvement in the primary balance achieved during the episode (divided by the primary balance in the year before the episode) is greater than the sample mean. Otherwise, the episode is considered a gradual fiscal adjustment. By linking the size of adjustment to the original primary balance, this approach avoids the problems associated with the use of arbitrary definitions of “big” and “small” adjustment episodes.

The success of fiscal adjustment episodes has been traditionally identified in the literature in terms of the reduction in the public debt-to-GDP ratio in the year(s) immediately after the episode (e.g., von Hagen and others, 2001). The problem with this definition is that it does not take into account whether or not the current fiscal position is sustainable. An alternative way of judging the success of fiscal adjustment is to assess the reduction in the distance between a country’s actual and desired primary balance (consistent with debt sustainability,6 hereafter defined as the sustainability threshold) during or after the episode.

Therefore, success in approaching the threshold can be defined as an event in which the initial distance between the primary balance and the sustainability threshold is reduced by at least two-thirds during the adjustment episode and/or during the following two years. An episode of adjustment is successful in reaching the threshold if the primary balance exceeds the sustainability threshold at least for one year during the adjustment episode or during the following two years. Success in maintaining the threshold occurs when a country maintains its primary balance above the sustainability threshold for one or two consecutive years after reaching that threshold.

Finally, fiscal adjustments that are successful in persisting are those in which a country maintains its primary balance above the sustainability threshold for more than two consecutive years during the adjustment episode and/or during the following two years.

III. Brief Review of the Literature on Fiscal Adjustment

There is considerable evidence to suggest that expenditure cuts (specifically for spending on transfers and government wages) increase the likelihood of success of fiscal adjustment efforts (Alesina and Perotti, 1995, 1996; McDermott and Wescott, 1996; Alesina, Perotti, and Tavares, 1998; Alesina and Ardagna, 1998; and Alesina, Ardagna, Perotti, and Schiantarelli, 1999). These studies also found that the composition of adjustment is more important than its size for maintaining a reduction in the stock of public debt and promoting growth. On the other hand, Giavazzi and Pagano (1990), Cour and others (1996), Giavazzi and others (2000), and Purfield (2003) conclude that the size and persistence of the consolidation are relevant in determining success.7

The positive impact of longer durations of fiscal adjustment in lowering public debt has also been stressed (von Hagen and others, 2001, Maroto and Mulas-Granados, 2002, Adam and Bevan, 2003, and Gupta and others, 2003). In addition, many authors (for example, von Hagen and others, 2001, and Lambertini and Tavares, 2001) have emphasized the role of initial fiscal conditions, economic growth, monetary policy, and exchange rate changes in influencing the likelihood of achieving fiscal adjustment.

The timing, duration, size, and composition of fiscal adjustment can be influenced by institutional and political constraints. Among the considerations identified are: The cabinet’s ideology on fiscal policy (Perotti and Kontopoulus, 2002, and Mulas-Granados, 2003); the electoral system and the budget process (Persson and Tabellini, 1999, Hallerberg and von Hagen, 1997, and von Hagen and others, 2001); and the proximity of elections (Alesina, Cohen, and Roubini, 1992, Maroto and Mulas-Granados, 2002, and Buti and van den Noord, 2003). Finally, the extent to which decision making in the budgetary process is fragmented, has also been widely considered (Roubini and Sachs, 1989; Grilli and others, 1991; and Mulas-Granados, 2003). However, there are no studies that analyze both the political and economic determinants of successful fiscal adjustments (defined in terms of achieving and maintaining fiscal sustainability).8 This paper attempts to fill this gap for emerging market economies.

IV. Model Specification

We use the following probit model specification to determine the contribution of economic and political factors in successful fiscal adjustments. Our baseline specification, following the existing literature, is as follows:

where t refers to the time period during the episode of adjustment and t-1 refers to the year preceding the beginning of the consolidation.9 The dependent variable is of a discrete nature, and takes the value of one in the case of success during the episode and zero otherwise.

We estimate our baseline model using a truncated panel of adjustment years. The sample is based on the following procedure: From the original panel of 550 annual observations, we select only those years [177] where there has been an improvement in the primary balance of at least 0.5 percent of GDP—the baseline case. Adjustment years take the value of one if the adjustment episode to which they belong was successful, and the value of zero if it was unsuccessful. From this first sample, we also extract a second sample of 106 adjustment episodes10, where all variables take their average values throughout the consolidation episode. To build this second database, adjustment years are grouped together around sets of consecutive years of fiscal adjustment in every country.

A first set of variables in equation (1) controls for initial macroeconomic and fiscal conditions. These variables are similar to the ones used by others (Adam and Bevan, 2003, Gupta and others, 2003) that describe the link between the success of fiscal adjustment and initial vulnerabilities. We include the initial primary budget balance (PBBal) in the year prior to the adjustment (a measure of initial (low) fiscal vulnerability that is expected to be positively correlated with the success of adjustment), the initial change in the exchange rate (Exch), with an increase signifying a depreciation, initial real GDP growth (Growth), the initial rate of inflation Infl (to control for initial macroeconomic conditions), and the accumulated number of failures (Nfail), measured by the number of adjustment episodes that had come to an end in the year previous to the start of a new fiscal consolidation. This latter variable measures the track record of fiscal policy implementation and is expected to be negatively correlated with success. In fact, a high number of previous failures point to the government’s inability to maintain sound fiscal policy and the tendency to reverse fiscal consolidations after a few years (Gupta and others, 2003).

We also include in equation (1) variables that have not, to this point, been used in the fiscal adjustment literature. These variables capture the influence of the government’s strategy of adjustment and the political constraints within which those decisions are taken. We include a dummy variable (Exp), which takes the value of one if more that fifty percent of the deficit reduction during the episode was achieved through spending cuts, and takes the value of zero if the adjustment was revenue-based. To account for the influence of large improvements (big-bang episodes) in the budgetary position, we include a dummy (Big) that is one if the average annual change in the primary budget balance divided by the initial primary balance is above the sample mean (zero otherwise). We also include two additional variables that account for the duration and the timing of the consolidation episodes. The variable that measures duration (Dur) is the number of years that the adjustment episode lasted. The variable that deals with the timing dimension of the adjustment episode (Front) takes the value of one if more than half of the total primary balance improvement occurred during the first half of the episode (zero otherwise).

A third set of regressors attempts to capture the accompanying political conditions that have a potential impact on adjustment episodes.11 To capture the degree of fragmentation in decision making, we include a variable (Maj) that measures the fraction of seats in the legislature that are held by the government. The weaker the government is in terms of legislative support, the lower, it is assumed, is the political consensus in favor of fiscal consolidation. We also include a measure of fiscal decentralization (Decent) which takes the value of one if the state/provinces of each country have any authority over fiscal policy.12 In addition, we include a dummy variable (Elec) that captures the effect of elections taking place while the adjustment is still ongoing or scheduled for the following year.

We complement these three variables that control for domestic political conditions with a dummy variable indicating whether or not the country was being supported by an IMF-supported program. The existence of an IMF-supported program in the year immediately before the adjustment or during the adjustment is expected to have an influence on the likelihood of success in achieving fiscal adjustment. IMF-supported programs could be viewed as increasing the probability of success with sustaining fiscal adjustment because of the financial support they offer—or will trigger from other sources—on the basis of success in meeting program targets. In this light, the IMF-supported program allows countries to address macroeconomic imbalances over time through both foreign financing and domestic adjustment, rather than adjustment alone. In addition, the existence of a IMF-supported program may signal the resolve of the authorities to achieve fiscal adjustment, with a concomitantly positive reaction from financial markets. In this light, the result for the IMF-supported program dummy should be interpreted with caution, as it may, in effect, be serving as a proxy for this domestic commitment to adjustment, rather than measuring the influence of programs per se.

V. Econometric Results

Table 1 reports the descriptive statistics of the independent variables and the four definitions of the dependent variable for the 25 countries in our sample between 1980 and 2001.13 The average success rate in the sample ranges from 37 percent (based on the definition of success as persistence in adjustment) and 83 percent (based on the definition of success as approaching sustainability). A large majority of the adjustment episodes can be defined as front-loaded (69 percent), while, by definition, half of the episodes are classified as “big-bang.”

Table 1.Descriptive Statistics: Adjustment Years(In percent, unless otherwise specified)
Initial Prim.

Budget

Balance

(In percent

of GDP)
Initial Change

in the

Exchange

Rate
Initial Real

GDP

Growth
Initial

Inflation

(change

in CPI)
Expenditure-

based
Episode

Duration

(In years)
Front-loaded

Adjustment
Big-bang

Adjustment
All Countries
Mean-1.1278.8-2.47.974.62.269.451.4
St. Dev.3.12754.213.016.443.71.346.250.1
Median-1.00.4-0.42.91002.0100100
Obs.177172171172177177173173
With IMF
Mean-0.5664.0-1.710.079.72.571.454.8
St. Dev.2.64455.96.017.640.61.645.550.2
Median0.10.6-0.43.21002.0100100
Obs.6565656564656362
Without IMF
Mean-1.444.8-2.96.771.42.167.649.1
St. Dev.3.4320.315.815.645.41.147.050.2
Median-1.50.30.32.11002.01000
Obs.112107106107113112110111
Legislative

Majority
Executive

Election
Power

Decentralization
Success

Approaching
Success in

Reaching
Success in

Maintaining
Success in

Persisting
All Countries
Mean67.641.848.082.665.247.237.1
St. Dev.22.149.550.138.047.850.148.4
Median66.90010010000
Obs.175177173177177177177
With IMF
Mean63.427.737.583.176.970.858.5
St. Dev.20.145.148.837.842.445.849.7
Median63.300100100100100
Obs.65656465656565
Without IMF
Mean70.750.954.282.158.033.024.1
St. Dev.22.650.25038.449.647.243.0
Median74.410010010010000
Obs.110112109112112112112
Sources: Government Finance Statistics (GFS), World Economic Indicators, and Database of Political Institutions, and authors’ calculations.
Sources: Government Finance Statistics (GFS), World Economic Indicators, and Database of Political Institutions, and authors’ calculations.

We also report in the same table the results of the subsample averages for the periods in which the countries had IMF-supported programs. Success rates tend to be higher for periods in which countries had a program. Initial macroeconomic conditions (in terms of growth and the initial primary balance) tend to be worse for periods in which the countries did not have an IMF-supported program. Initial rates of inflation and exchange rate depreciation, however, are higher for countries with IMF-supported programs.

Table 2 shows the results of different probit model estimations for the economic and political determinants of the success in approaching, reaching, maintaining, and persisting with a fiscal stance consistent with fiscal sustainability. The results suggest that longer durations of adjustment episodes significantly increase the probability of success, whereas better initial primary balances only contribute to increasing the success in persistence and by a small margin. Expenditure-based adjustments are helpful in approaching and reaching the sustainability threshold for the primary balance, but are less clearly related to maintaining public finances in a sustainable position for more than two years. This is consistent with results derived from other studies that expenditure reforms are necessary but not sufficient to achieve sustainable positions. In some instances, revenue increases have contributed to a sustainable fiscal position (Adam and Bevan, 2003; Gupta and others, 2003).14

Table 2.Success in Fiscal Adjustments: Adjustment Years(0.5 percent of GDP threshold)
Probit EstimationsSuccess in

Approaching (a)
Success in

Reaching
Success in

Maintaining
Success in

Persisting
Initial primary budget balance0.008

(0.58)
0.017

(1.48)
0.025

(1.22)
0.022

(1.65) *
Initial change of exchange rate-0.000

(0.09)
-0.000

(0.31)
-0.000

(0.05)
-0.000

(0.20)
Initial real GDP growth-0.020

(1.45)
0.003

(1.02)
0.012

(2.33) **
0.005

(1.29)
Initial inflation rate-0.003

(1.25)
0.006

(1.18)
-0.001

(0.33)
-0.002

(0.48)
Expenditure-based15.09

(1.87) *
0.353

(2.95) ***
0.073

(0.50)
0.165

(1.83) *
Episode duration0.176

(1.55)
0.140

(2.88) ***
0.314

(3.96) ***
0.178

(3.65) ***
Accumulated no. failures-0.041

(0.99)
-0.089

(2.45) **
0.055

(0.98)
0.043

(1.17)
Front-loaded adjustments..

..
-0.230

(2.53)**
-0.116

(0.71)
-0.066

(0.60)
Big-bang adjustments25.51

(2.75)***
0.361

(4.02)***
0.490

(3.58) ***
0.118

(1.23)
Legislative majority0.145

(1.01)
0.503

(2.95) ***
1.264

(4.26) ***
0.770

(3.75) ***
Executive elections-2.13

(2.19) **
-0.343

(3.42) ***
-0.547

(4.41) ***
-0.273

(3.23) ***
Power decentralization0.011

(0.14)
-0.133

(1.48)
-0.212

(1.34)
-0.040

(0.38)
Initial IMF-standby arrangement-0.348

(1.08)
0.135

(1.71) *
0.477

(3.50) ***
0.284

(3.00) ***
Observations106152152152
Pseudo R20.800.620.530.43
Log pseudo-likelihood-10.65-37.02-49.51-57.66
Prob> χ20.0000.0000.0000.000
Absolute value of z statistics in parentheses. The reported coefficients are changes in the probability for an infinitesimal change in each independent continuous variable and a discrete change in the probability for dummy variables.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the font-oaded variable predicts perfectly the dependent variable, and thus, this variable is dropped from the equation. Also, the reported coefficients in this column have been multiplied by 1000.

Absolute value of z statistics in parentheses. The reported coefficients are changes in the probability for an infinitesimal change in each independent continuous variable and a discrete change in the probability for dummy variables.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the font-oaded variable predicts perfectly the dependent variable, and thus, this variable is dropped from the equation. Also, the reported coefficients in this column have been multiplied by 1000.

Front-loaded adjustments are negatively correlated with success in reaching sustainability, while big-bang adjustments show a systematic, large positive effect on the probability of approaching, reaching, and maintaining sustainability. As a result, those fiscal adjustments which rely on large deficit reductions and are concentrated in the second half of the adjustment episode have a higher probability of success.

The results also suggest that fiscal adjustments undertaken by governments with a majority in parliament are more likely to succeed. However, in pre-election periods, fiscal adjustment is less likely to be sustained. Decentralization does not seem to play a significant role in achieving and maintaining sustainability under any definition of success.

In addition, the results show that fiscal adjustments implemented under an IMF-supported program are more likely to succeed in reaching, maintaining, and persisting in fiscal sustainability. At the same time, adjustment under IMF-supported programs is not associated with success in approaching fiscal sustainability (that is, generating two-thirds of the improvement in the primary balance needed to achieve fiscal sustainability). This latter result is consistent with the existing empirical literature on the impact of IMF-supported programs. Buliř and Moon (2003), among others, used the Generalized Evaluation Estimator (GEE) technique to measure the impact of the IMF-supported program on the composition and the economic impact of fiscal adjustments and, in line with previous studies, did not find any effect.15

Robustness tests

The robustness of the above findings is tested by the following measures: (1) accounting for heterogeneity across countries (panel heterogeneity), (2) accounting for the endogeneity of some regressors, (3) using alternative definitions of fiscal adjustment, (4) estimating the effects of averaging observations by adjustment episode, and (5) excluding potential outliers from the sample.

Panel heterogeneity can be a problem if countries with considerable economic and institutional differences have been pooled together. The fact that these characteristics are country-specific and constant over time may result in autocorrelation in the error term. To address this, all probit regressions were rerun using nonzero between-cluster error terms. This allows us to specify that observations are independent across groups (clusters) but not necessarily independent within groups. Table 7 in the Appendix shows that in general all results are robust.

A second potential problem could come from the inclusion of the IMF dummy on the right-hand side of the equation. A country may enter into an IMF-supported program only after its macroeconomic imbalances are unsustainable. This problem is dealt with by estimating the following system of two simultaneous equations:

As can be observed in equations (2) and (3), the initial IMF-supported program is instrumented with the following variables: the success in reaching the threshold, all initial budgetary and economic conditions and all political constraints, the initial distance from the threshold, and the accumulated number of failures (which may motivate a country to seek IMF support). We apply a two-stage least squares estimation procedure with fixed effects (instead of cluster-specific error terms) to the panel of adjustment episodes. Results are reported in Appendix Table 8. These show that our baseline results hold. Nonetheless, some country-specific variables lose part of their predictive power. This is particularly the case for the accumulated number of failures, meaning that a history of poor fiscal behavior is completely captured by the country-specific effect and the history of IMF programs.

Furthermore, we test the model using alternative thresholds. Appendix Table 9 reports results using a threshold of 1.5 percent of GDP.16 Again, the results are broadly consistent with the baseline model. However, there are some changes in the statistical significance of a few variables. For example, the impact of an initial primary balance on the probability of success gains statistical significance. Also, the big-bang adjustment variable becomes significant for success in persisting in the consolidation effort. Legislative majority of the ruling party remains the dominating institutional variable but the decentralization dummy also becomes significant, reducing the probability of reaching and maintaining fiscal sustainability.17

We further tested the robustness of the baseline model on the adjustment episodes sample. While the size of most coefficients tends to be smaller (Appendix Table 6), the results of the baseline model are broadly confirmed.

We also sought to test the potential impact of large countries on the results. To this effect, we ran regressions that iteratively excluded large countries such as Argentina, Brazil, China, Indonesia, and Russia from the sample. There was no significant change in results.18

VI. Policy Implications

The following lessons can be gleaned from the above results.

First, expenditure-based adjustments are not in themselves sufficient to maintain healthy public finances over the medium term; an effort on the revenue side is also needed (Gupta and others, 2003). Second, back-loaded adjustments are more successful in achieving fiscal sustainability than front-loaded consolidations; this may owe to the fact that they spread out the social costs of adjustment over time, and thus help keep at bay the political resistance to “staying the course.” Third, majoritarian governments that implement adjustments in fiscally centralized countries after elections are the most likely to be successful, reflecting the higher degree of political consensus enjoyed by these governments in the post-election era. And finally, there is some evidence that points toward a positive effect of IMF-supported programs on the maintenance of fiscal sustainability, but not necessarily on the likelihood of moving toward a sustainable fiscal position. This suggests that IMF-supported programs matter most in cases where countries are close to the sustainability threshold and are committed to achieving it, but are less influential in countries with large fiscal imbalances.

The mechanisms that lie behind the first three results are clear and supported by evidence in other studies (Gupta and others, 2003). With respect to the impact of IMF-supported programs, this may reflect, as noted above, a proxy for the commitment of the authorities to restore fiscal sustainability that would also positively affect market sentiment. Therefore, caution is needed in interpreting these findings and drawing policy implications.

Does the composition of fiscal adjustment undertaken in IMF-supported programs differ from the sample as a whole? Table 3 shows that having a program does not affect the composition of the adjustment or its size, but rather its timing and phasing.19 Fiscal adjustment under an IMF-supported program tends to be more front-loaded than in the rest of the sample, and lasts longer. This result is of particular interest, as for the sample as a whole, back-loaded adjustments have a higher success rate. The differences in the results could owe to initial economic conditions, as countries with IMF-supported programs tend to have worse-than-average financial imbalances at the onset of the adjustment period (e.g., large pressures on the currency, lower GDP growth, and higher inflation). Under these circumstances, front-loading of the adjustment could be a result of the need to stabilize monetary aggregates and the exchange rate in a context of extreme vulnerability and limited access to financing. The composition of the adjustment is, however, not affected by the IMF program. Fiscal adjustments under IMF-supported program are not based on spending cuts any more than they are in countries without IMF-supported programs.

Table 3.The Characteristics of Fiscal Adjustments
Dependent Variable
Probit EstimationsExpenditure-

based
FrontloadedBig-bangEpisode

Duration
Expenditurebased..

..
-0.323

(1.10)
0.978

(3.92)
0.309

(1.53)
Frontloaded-0.354

(1.24)
..

..
-0.237

(0.93)
-1.147

(6.34) ***
Big-bang0.954

(3.98) ***
-0.296

(1.17)
..

..
-0.629

(3.61) ***
Episode duration0.160

(1.52)
-0.597

(5.50) ***
-0.340

(3.45) ***
..

..
Initial IMF arrangement0.230

(0.94)
0.563

(2.12) **
0.234

(1.06)
0.532

(3.06) ***
Observations165165165165
Pseudo R20.120.200.120.29
Log likelihood-82.69-82.62-100.64..
Prob>χ20.00000.00000.0000..
Absolute value of z statistics in parentheses. Constants not reported.The reported coefficients under Episode duration are from OLS regression. See notes under Table 2 for interpretation of the other coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.
Absolute value of z statistics in parentheses. Constants not reported.The reported coefficients under Episode duration are from OLS regression. See notes under Table 2 for interpretation of the other coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

These results are confirmed when the sample years are split between those under an IMF-supported program and those without, and simple regressions are run on the different definitions of success (see Table 4). For most variables, there are no statistically significant differences20 in the estimated coefficients in the sample of countries with or without IMF-supported programs.

Table 4.Success in Fiscal Adjustments and the Role of IMF Programs: Adjustment Years(0.5 percent of GDP threshold)
Success in

Approaching
Success in

Reaching
Success in

Maintaining
Success in

Persisting
With

IMF

program
Expenditurebased0.320

(2.79)***
0.301

(2.58)**
0.154

(1.24)
0.360

(2.93)***
Episode duration0.106

(3.35)***
0.133

(4.13)***
0.179

(5.20)***
0.123

(3.63)***
Frontloaded0.195

(1.73)**
0.154

(1.34)
0.178

(1.45)
-0.164

(1.36)
Big-bang0.179

(2.05)**
0.284

(3.19)***
0.168

(1.77)**
-0.069

(0.74)
Without

IMF

program
Expenditurebased0.281

(4.61)***
0.212

(2.47)**
-0.076

(0.77)
-0.026

(0.25)
Episode duration0.195

(6.22)***
0.029

(0.64)
0.053

(1.04)
0.077

(1.45)
Frontloaded-0.004

(0.08)
-0.231

(2.79)***
-0.080

(0.84)
0.017

(0.17)
Big-bang0.321

(5.75)***
0.389

(4.94)***
0.429

(4.72)***
0.255

(2.70)***
With

IMF

program
Legislative majority0.332

(1.41)
0.611

(2.55)**
0.804

(3.15)***
0.660

(2.61)**
Executive election-0.073

(0.78)
-0.205

(2.16)**
-0.246

(2.44)**
-0.199

(2.00)**
Power decentralization-0.103

(1.06)
0.087

(0.88)
-0.007

(0.06)
0.125

(1.20)
High initial target-0.102

(1.32)
-0.087

(1.10)
-0.144

(1.72)**
0.071

(0.85)
Without

IMF

program
Legislative majority0.320

(3.14)***
0.329

(2.29)**
0.521

(3.14)***
0.367

(2.13)**
Executive election-0.094

(1.83)**
-0.201

(2.77)***
-0.206

(2.46)**
-0.073

(0.84)
Power decentralization0.066

(0.90)
0.036

(0.34)
-0.141

(1.18)
-0.119

(0.95)
High initial target-0.129

(1.90)**
-0.330

(3.45)***
-0.008

(0.07)
-0.020

(0.18)
Absolute value of t statistics in parentheses. Constant included but not reported.See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.
Absolute value of t statistics in parentheses. Constant included but not reported.See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

A final issue that we investigate is the effect of ambitious fiscal consolidation targets on the probability of success. To disentangle the possible influence of large initial fiscal imbalances relative to the sustainable threshold of the primary fiscal balance, we create a new dummy variable. We first calculate the initial distance between the primary fiscal balance of a country and its debt-stabilizing primary budget balance in the year before the beginning of the adjustment episode.21 This is equivalent, under our definition of success, to the initial fiscal target of the consolidation episode. If this target is missed, the probability of reaching and maintaining fiscal sustainability is by definition zero. We then create a dummy variable, labeled high initial target, that takes the value of one when the distance from the sustainability threshold was above the sample mean (zero otherwise). The probit results are reported in Table 5 and show that the impact of high initial fiscal target on the likelihood of success is systematically negative for reaching and maintaining sustainability. For countries with a large initial fiscal imbalance, i.e., a more difficult fiscal consolidation target, the probability of reaching the sustainability threshold is reduced by almost one third, in addition to the effect of the size of their initial fiscal balance. These results show that while countries that adopt big-bang adjustments tend to be more successful than other countries, particularly when they spread the cost of the fiscal consolidation over many years, ambitious initial targets tend to make success more difficult.

Table 5.Success in Fiscal Adjustments and the Role of High Initial Targets: Adjustment Years(0.5 percent of GDP threshold)
Cluster-probit

Estimations
Success in

Approaching (a)
Success in

Reaching
Success in

Maintaining
Success in

Persisting
Initial primary budget balance0.000

(0.44)
0.019

(1.54)
0.020

(0.94)
0.020

(1.60)
Initial change of exchange rate−0.000

(0.02)
−0.000

(0.13)
−0.000

(0.14)
−0.001

(0.24)
Initial real GDP growth-0.000

(1.51)
0.003

(0.96)
0.013

(2.21) **
0.005

(1.32)
Initial inflation rate-0.000

(1.46)
0.004

(1.11)
-0.001

(0.25)
-0.002

(0.49)
Expenditurebased0.831

(1.44)
0.360

(2.74) ***
0.015

(0.10)
0.137

(1.69) *
Episode duration0.000

(1.46)
0.167

(3.27) ***
0.362

(4.31) ***
0.170

(3.83) ***
Accumulated no. of failures-0.000

(0.99)
-0.050

(1.20)
0.125

(1.96) *
0.060

(1.64)
Front-loaded adjustment..

..
-0.250

(2.48) **
-0.057

(0.33)
-0.043

(0.43)
Big-Bang adjustment5.324

(1.75) *
0.445

(4.29) ***
0.583

(3.84) ***
0.135

(1.50)
Legislative majority0.000

(1.16)
0.564

(3.12) ***
1.468

(4.31) ***
0.734

(3.85) ***
Executive elections-0.006

(2.01) **
-0.409

(3.71) ***
-0.579

(4.52) ***
-0.261

(3.35) ***
Power decentralization0.000

(0.36)
-0.051

(0.51)
-0.082

(0.48)
0.012

(0.12)
Initial IMF-standby arrangement-0.0008

(1.06)
0.132

(1.60)
0.512

(3.47) ***
0.261

(2.99) ***
High initial target-0.0006

(1.60)
-0.285

(2.55) **
-0.418

(2.65) ***
-0.133

(1.45)
Observations106152152152
Pseudo R20.840.650.570.44
Log pseudo-likelihood-8.54-33.57-45.72-56.58
Prob> χ20.0000.0000.0000.000
Absolute value of z-statistics in parentheses. See Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the front-loaded variable predicts perfectly the dependent variable. Thus. this variable is dropped from the equation. Also, coefficients in this column are multiplied by 1000.

Absolute value of z-statistics in parentheses. See Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the front-loaded variable predicts perfectly the dependent variable. Thus. this variable is dropped from the equation. Also, coefficients in this column are multiplied by 1000.

VII. Conclusions

This paper identifies factors behind successful fiscal adjustment in emerging market economies.

One important result is that initial financial vulnerabilities and institutional weaknesses may reduce the chance of success of fiscal consolidation episodes, particularly in a context of large initial deviations from a sustainable fiscal position. Furthermore, adjustments that seek to spread out the path of adjustment—that is, back-loaded adjustments—have a higher probability of success in reaching fiscal sustainability. This could be because front-loaded strategies are based on poor quality measures, including across-the-board expenditure cuts and distortionary revenue increases. Back-loaded adjustments provide the opportunity to phase-in adjustment over a longer period of time, and thus allow for the introduction of higher quality, more durable reforms, including measures to limit unproductive spending, expand the tax base, and improve tax administration. The other reason could be political: more gradual fiscal consolidations generate consensus and signal markets that fiscal discipline will be maintained over the medium term. However, there is no support for back-loaded adjustments to maintain fiscal sustainability over time.

Another important finding is that IMF-supported programs can help countries achieve fiscal sustainability. As noted in the text, the results may not indicate the impact of the IMF-supported program per se; rather, they could reflect the fact that countries with such programs are more committed to adjustment than those without. Further research is needed to quantify the effects of other factors that explain why programs may have a salutary effect on success.

A final result of our paper is that the size of the adjustment is important. This confirms previous findings in the literature. Other things being equal, large fiscal adjustments signal the authorities’ intention to put fiscal policy on a sound footing and achieve large reductions in the stock of public debt. At the same time, the results suggest those in need of fiscal adjustment are also the ones that are likely to abandon their adjustment efforts, given the significance of the “high initial target” variable. As such, the prospects for “staying the course” and achieving a steady path toward fiscal sustainability in high-debt countries should be judged with due caution.

Appendix Table 6.Success in Fiscal Adjustments by Adjustment Episodes(0.5 percent of GDP threshold)
Probit EstimationsSuccess in

Approaching (a)
Success in

Reaching
Success in

Maintaining
Success in

Persisting (a)
Initial Primary Budget

Balance
0.000*

(1.81)
0.049*

(1.86)
0.000*

(1.94)
0.000*

(1.69)
Initial Change of

Exchange Rate
-0.000

(0.52)
-0.000

(0.27)
-0.000

(0.10)
-0.000

(1.64)
Initial Real GDP

Growth
-0.000

(0.15)
0.009

(0.98)
-0.000

(0.78)
-0.000

(1.16)
Initial Inflation Rate0.000

(1.47)
0.029**

(2.27)
0.000

(1.26)
-0.000

(0.43)
Expenditure-Based-0.000

(0.57)
0.180

(0.96)
0.000

(0.62)
0.000

(1.22)
Episode Duration0.000*

(1.76)
0.197*

(1.68)
0.000**

(2.01)
0.000**

(2.36)
Accumulated No.

Failures
-0.000*

(1.80)
-0.336***

(2.90)
0.000*

(1.71)
0.000

(1.36)
Frontloaded

Adjustments
-0.000

(0.19)
-0.298

(1.57)
-0.100*

(1.82)
-0.000**

(2.05)
Big-Bang Adjustments0.000*

(1.67)
0.665***

(3.68)
0.629**

(2.34)
0.000***

(2.88)
Majority0.000

(1.43)
1.364**

(2.42)
0.000**

(2.06)
0.000**

(2.39)
Elections0.000

(1.35)
0.247

(1.27)
-0.000

(1.54)
-0.000*

(1.65)
Decentralization..

..
-0.407**

(1.96)
-0.465*

(1.94)
..

..
Initial IMF-Standby

Arrangements
-0.000

(1.35)
0.076

(0.40)
0.666**

(2.11)
0.000***

(2.97)
Observations88888888
Pseudo R20.850.640.830.69
Log pseudo-likelihood-7.64-21.46-9.97-15.45
Prob> χ20.0000.0000.0000.000
Absolute value of z-statistics in parentheses. See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent

In these specifications, the Front-loaded variable predicts perfectly the dependent variable. Thus, this variable is dropped from the equation.

Absolute value of z-statistics in parentheses. See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent

In these specifications, the Front-loaded variable predicts perfectly the dependent variable. Thus, this variable is dropped from the equation.

Appendix Table 7.Success in Fiscal Adjustments: Panel Heterogeneity Clusters(0.5 percent of GDP threshold)
By Adjustment YearsBy Adjustment Episodes
Probit EstimationsSuccess in

Approaching (a)
Success in

Reaching
Success in

Maintaining
Success in

Persisting
Success in

Approaching (b)
Success in

Reaching
Success in

Maintaining
Success in

Persisting (b)
Initial primary budget balance0.001

(1.10)
0.017

(1.10)
0.025

(0.92)
0.022

(1.30)
0.000***

(5.38)
0.049

(1.41)
0.000***

(3.33)
0.000

(1.54)
Initial change of exchange rate-0.000

(0.97)
-0.000*

(1.83)
-0.000

(0.20)
-0.000

(0.20)
-0.000***

(3.19)
-0.000***

(2.67)
-0.000

(0.17)
-0.000*

(1.67)
Initial real GDP growth-0.020

(1.56)
0.003

(1.34)
0.012***

(3.61)
0.005*

(1.77)
-0.000

(0.30)
0.009

(1.49)
-0.000

(1.31)
-0.000

(1.25)
Initial inflation rate-0.003

(1.57)
0.006

(1.50)
-0.001

(0.45)
-0.002

(0.56)
0.000***

(5.32)
0.029**

(2.45)
0.000**

(2.49)
-0.000

(0.38)
Expenditurebased15.1 **

(2.51)
0.353 ***

(2.68)
0.073

(0.39)
0.165

(1.09)
-0.000

(0.96)
0.180

(0.95)
0.000

(0.67)
0.000

(1.42)
Episode duration0.176 ***

(3.50)
0.140 ***

(2.81)
0.314 ***

(3.11)
0.178 ***

(3.10)
0.000 ***

(5.80)
0.197 *

(1.82)
0.000 ***

(3.09)
0.000 ***

(3.05)
Accumulated no. failures-0.041*

(1.73)
-0.089*

(1.90)
0.055

(0.91)
0.043

(1.11)
-0.000***

(7.07)
-0.336***

(3.43)
0.000**

(2.00)
0.000**

(2.09)
Front-loaded adjustments..-0.230**

(2.45)
-0.116

(0.56)
-0.066

(0.48)
-0.000

(0.36)
-0.298*

(1.72)
-0.100***

(2.62)
-0.000

(1.37)
Big-bang adjustments25 5***

(3.17)
0.361***

(6.09)
0.490***

(4.65)
0.118

(1.37)
0.000***

(7.53)
0.665***

(3.63)
0.629***

(3.43)
0.000***

(3.44)
Legislative majority0.145

(1.14)
0.503***

(3.02)
1.264***

(4.01)
0.770***

(4.15)
0.000***

(4.64)
1.364**

(2.45)
0.000***

(2.87)
0.000***

(2.59)
Executive elections-2.13

(1.62)
-0.343***

(2.99)
-0.547***

(3.62)
-0.273**

(2.16)
0.000***

(3.23)
0.247*

(1.80)
-0.000**

(2.43)
-0.000*

(1.87)
Power decentralization0.012

(0.19)
-0.133

(1.28)
-0.212

(1.43)
-0.040

(0.33)
..

..
-0.407**

(2.03)
-0.465***

(2.82)
..

..
Initial IMF-standby arrangement-0.348**

(1.98)
0.135

(1.63)
0.477***

(3.43)
0.284***

(2.70)
-0.000***

(2.99)
0.076

(0.39)
0.666***

(3.37)
0.000***

(3.71)
Observations10615215215288888888
Pseudo R21061521521520.850.640.830.64
Log pseudo-likelihood0.800.620.530.43-7.64-21.46-9.97-21.13
Prob> x2-10.65-37.02-49.51-57.660.0000.0000.0000.000
Absolute value of z statistics in parentheses. See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the front-loaded variable predicts perfectly the dependent variable. Thus, this variable is dropped from the equation. Also, the reported coefficients in this column have been multiplied by 1000.

In these specifications, the decentralization variable predicts perfectly the dependent variable. Thus, this variable is dropped from the equation.

Absolute value of z statistics in parentheses. See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the front-loaded variable predicts perfectly the dependent variable. Thus, this variable is dropped from the equation. Also, the reported coefficients in this column have been multiplied by 1000.

In these specifications, the decentralization variable predicts perfectly the dependent variable. Thus, this variable is dropped from the equation.

Appendix Table 8.Success in Fiscal Adjustments: Endogeneity Tests(0.5 percent of GDP threshold)
By Adjustment YearsBy Adjustment Episodes
Two Stage Least Square With

Fixed Effects
Success in

Approaching
Success in

Reaching
Success in

Maintaining
Success in

Persisting
Success in

Approaching
Success in

Reaching
Success in

Maintaining
Success in

Persisting
Initial primary budget balance0.029

(1.73)*
0.013

(0.67)
-0.000

(0.01)
0.025

(1.11)
0.012

(0.28)
0.023

(0.99)
0.042

(2.08)**
0.026

(1.22)
Initial change of exchange rate0.000

(0.44)
-0.000

(0.76)
-0.000

(0.31)
-0.000

(0.79)
-0.000

(1.12)
-0.000

(2.34)**
-0.000

(1.03)
-0.000

(0.62)
Initial real GDP growth0.003

(0.37)
0.003

(0.32)
0.001

(0.12)
0.001

(0.13)
0.001

(0.09)
0.005

(0.85)
0.004

(0.83)
0.002

(0.37)
Initial inflation rate-0.000

(0.02)
0.008

(2.38)**
0.001

(0.22)
-0.001

(0.18)
0.038

(2.31)**
0.031

(3.33)***
0.003

(0.36)
-0.002

(0.29)
Expenditurebased0.220

(2.14)**
0.356

(2.93)***
-0.015

(0.11)
0.132

(0.93)
0.340

(1.59)
0.235

(1.93)*
0.047

(0.43)
-0.025

(0.23)
Episode duration0.160

(3.77)***
0.076

(1.52)
0.144

(2.47)**
0.175

(2.98)***
0.223

(2.03)**
0.121

(1.92)*
0.140

(2.53)**
0.154

(2.71)***
Accumulated no. failures0.022

(0.37)
-0.047

(0.68)
0.063

(0.60)
0.037

(0.77)
-0.039

(0.26)
0.044

(0.49)
0.082

(0.62)


(1.13)
Front-loaded adjustments-0.080

(0.76)
-0.106

(0.86)
-0.001

(0.01)
-0.035

(0.24)
0.124

(0.42)
-0.132

(0.78)
-0.174

(1.17)
-0.301

(1.96)**
Big-bang adjustments0.207

(2.23)**
0.245

(2.23)**
0.197

(1.54)
0.022

(0.17)
0.188

(0.88)
0.408

(3.37)***
0.451

(4.22)***
0.472

(4.28)***
Legislative majority0.206

(0.72)
0.956

(2.84)***
1.295

(3.30)***
0.221

(0.56)
-0.613

(0.82)
0.223

(0.52)
0.840

(2.25)**
0.231

(0.60)
Executive elections-0.033

(0.42)
-0.188

(2.05)**
-0.400

(3.72)***
-0.344

(3.19)***
0.182

(0.78)
0.045

(0.34)
-0.161

(1.37)
-0.061

(0.50)
Power decentralization0.000

(.)
0.000

(.)
0.000

(.)
0.000

(.)
-0.772

(1.36)
-0.391

(1.21)
-0.577

(2.02)**
-0.327

(1.11)
Initial IMF-standby arrangement-0.112

(1.37)
0.082

(0.86)
0.162

(1.45)
0.135

(1.20)
-1.706

(2.44)**
-0.643

(1.61)
0.374

(1.06)
0.558

(1.53)
Constant0.191

(0.66)
-0.416

(1.21)
-0.743

(1.86)*
-0.175

(0.44)
1.074

(1.33)
0.306

(0.66)
-0.354

(0.87)
-0.217

(0.52)
Observations8888888887878787
Overall R20.400.480.480.320.130.320.540.43
Wald x2722.14363.07155.9190.68115.57232.84178.25105.06
Prob> y20.0000.0000.0000.0000.0000.0000.0000.000
1. Absolute value of z statistics in parentheses.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.
1. Absolute value of z statistics in parentheses.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.
Appendix Table 9.Success in Fiscal Adjustments by Adjustment Episodes: Large Episodes(1.5 percent of GDP threshold)
Cluster Probit EstimationsSuccess in

Approaching (a)
Success in

Reaching
Success in

Maintaining
Success in

Persisting
Initial primary budget balance0.020***

(3.92)
0.034***

(2.74)
0.019**

(2.43)
0.003

(0.82)
Initial change of exchange rate-0.003***

(2.61)
-0.001

(0.60)
-0.003***

(2.72)
-0.001

(1.62)
Initial real GDP growth-0.011

(1.62)
-0.007

(1.14)
-0.002

(0.24)
0.001

(0.26)
Initial inflation rate0.005**

(2.48)
0.015

(1.34)
-0.002

(0.70)
0.000

(0.35)
Expenditurebased0.325***

(2.65)
0.080

(0.43)
-0.070

(0.42)
0.090*

(1.88)
Episode duration0.025

(0.64)
0.164

(1.53)
0.222**

(2.28)
0.090**

(2.04)
Accumulated no. failures-0.263**

(2.29)
-0.184**

(2.49)
-0.043

(0.54)
-0.048

(1.38)
Front-loaded adjustments-0.224**

(2.46)
-0.245

(1.07)
-0.039

(0.18)
0.074*

(1.89)
Big-bang adjustments0.443***

(3.71)
0.487***

(2.82)
0.437***

(2.98)
0.170**

(2.53)
Legislative majority0.900***

(4.18)
0.518

(1.47)
0.618**

(2.01)
0.199*

(1.78)
Executive elections..

..
-0.355*

(1.72)
-0.331**

(2.55)
-0.153***

(2.58)
Power decentralization..

..
-0.259*

(1.89)
-0.265**

(2.45)
-0.004

(0.08)
Initial IMF-standby arrangement-0.137

(1.41)
0.108

(0.70)
0.352**

(2.02)
0.104

(1.57)
Observations61616161
Pseudo R20.720.470.450.43
Log pseudo-likelihood-8.77-20.84-23.10-19.36
Prob> χ20.0000.0000.0000.000
Absolute value of z statistics in parentheses. See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the executive election and power decentralization variables predict perfectly the dependent variable, and thus, both are dropped from the equation.

Absolute value of z statistics in parentheses. See notes under Table 2 for interpretation of the coefficients.* significant at 10 percent; ** significant at 5 percent; *** significant at 1 percent.

In this specification, the executive election and power decentralization variables predict perfectly the dependent variable, and thus, both are dropped from the equation.

References

    AdamChristopher and DavidBevan2003Staying the Course: Maintaining Fiscal Control in Developing Countriespaper prepared for the Brookings Institution Trade Forum Conference (WashingtonMay 17-18).

    AlesinaAlberto and SilviaArdagna1998Tales of Fiscal Contractions: Fiscal Adjustments: Why Can They Be ExpansionaryEconomic PolicyVol. 27 pp. 489543.

    AlesinaAlbertoRobertoPerotti and FabioSchiantarelli1999Fiscal Policy, Profits, and InvestmentNBER Working Paper No. 7207 (Cambridge, Massachusetts: National Bureau of Economic Research).

    AlesinaAlbertoGerald D.Cohen and NourielRoubini1992Macroeconomic Policies and Elections in OECD DemocraciesEconomics and PoliticsVol. 4 pp. 130.

    AlesinaAlberto and RobertoPerotti1995Fiscal Expansion and Fiscal Adjustments in OECD CountriesEconomic PolicyVol. 21 pp. 20548.

    AlesinaAlberto and RobertoPerotti1996Budget Deficits and Budget InstitutionsNBER Working Paper No. 5556 (Cambridge, Massachusetts: National Bureau of Economic Research).

    AlesinaAlbertoRobertoPerotti and JoséTavares1998The Political Economy of Fiscal AdjustmentsBrookings Papers on Economic ActivityVol. 1 pp. 197248.

    BeckThorstenGeorgeClarkeAlbertoGroffPhilipKeefer and PatrickWalsh2001New Tools and New Tests in Comparative Political Economy: The Database of Political InstitutionsWorld Bank Economic ReviewVol. 15 pp. 16576.

    BuliřAleš and SoojinMoon2003Do IMF-Supported Programs Help Make Fiscal Adjustments More Durable?IMF Working Paper No. 03/38 (Washington: International Monetary Fund).

    ButiMarco and Paulvan den Noord2003Discretionary Fiscal Policy and Elections: The Experience of the Early Years of EMUOECD Economics Department Working Paper No. 351 (ECO/WKP(2003)5) (Paris: Organisation for Economic Co-operation and Development).

    ConwayPatrick2000IMF Programs and External Balance: The Crisis-Participation Dynamic” (Chapel Hill: University of North Carolina).

    CourPhilippineEricDuboisSelmaMahfouz and JeanPisani-Ferry1996The Cost of Fiscal Retrenchment Revisited: How Strong is the Evidence?Série des Documents de Travail 9612 (Paris: Institut National de la Statistique et des Études Économiques).

    Dicks-MireauxLouisMauroMecagni and SusanSchadler2000Evaluating the Effect of IMF Lending to Low-Income CountriesJournal of Development EconomicsVol. 61 (April) pp. 495526.

    Fiscal Affairs Department2004Experience With Large Fiscal Adjustmentsmimeo (Washington: International Monetary Fund).

    GiavazziFrancesco and MarcoPagano1990Can Severe Fiscal Contractions be Expansionary?NBER Macroeconomics Annual pp. 75116 (Cambridge, Massachusetts: National Bureau of Economic Research).

    GiavazziFrancescoTullioJappelli2000Searching for Non-Linear Effects of Fiscal Policy: Evidence from Industrial and Developing CountriesNBER Working Paper No. 7460 (Cambridge, Massachusetts: National Bureau of Economic Research).

    GrilliVittorioDonatoMasciandaro and GuidoTabellini1991Institutions and Policies: Political and Monetary Institutions and Public Financial Policies in the Industrial CountriesEconomic PolicyVol. 13 pp. 34192.

    GuptaSanjeevEmanueleBaldacciBenedictClements and ErwinTiongson2003What Sustains Fiscal Consolidations in Emerging Market Countries?IMF Working Paper No. 03/224 (Washington: International Monetary Fund).

    HallerbergMark and Jürgenvon Hagen1997Electoral Institutions, Cabinet Negotiations and Budget Deficits in the European Union (EU)CEPR Working Paper No. 1555 (London: Centre for Economic Policy Research).

    IvanovaAnnaWolfgangMayerAlexandrosMourmouras and GeorgeAnayiotos2003What Determines the Success or Failure of Fund-Supported Programs?IMF Working Paper 03/8 (Washington: International Monetary Fund).

    KhanMohsin S.1990IMF Conditionality and Country Ownership of ProgramsIMF Staff PapersVol. 37No. 2 (June) pp. 195231 (Washington: International Monetary Fund).

    LambertiniLuisa and José TavaresJ.2001Exchange Rates and Fiscal Adjustments: Evidence from the OECD and Implications for EMUBoston College Working Paper No. 576 (Boston College, Boston: Department of Economics).

    MarotoReyes and CarlosMulas-Granados2002The Duration of Budgetary Consolidations in EuropeEuropean Economy Group (UCM) Working Paper No. 18/2002.

    McDermottC. John and Robert F.Wescott1996An Empirical Analysis of Fiscal AdjustmentsIMF Staff PapersVol. 43No. 4 (December) pp. 72553 (Washington: International Monetary Fund).

    Mulas-GranadosCarlos2003The Political and Economic Determinants of Fiscal Adjustments in EuropeEuropean Political Economy ReviewVol. 1 (1) pp. 3364.

    PerottiRoberto and YianosKontopoulus2002Fragmented Fiscal PolicyJournal of Public EconomicsVol. 82 (2) 191222.

    PerssonTorsten and GuidoTabellini1999Political Economics and Public Finance” in AuerbachA. and FeldsteinM.(eds.) Handbook of Public EconomicVol. III (Amsterdam: North Holland).

    PurfieldCatriona2003Fiscal Adjustment in Transition Countries: Evidence from the 1990sIMF Working Paper No. 03/36 (Washington: International Monetary Fund).

    RoubiniNouriel and JeffreySachs1989Political and Economic Determinants of Budget Deficits in the Industrial DemocraciesEuropean Economic ReviewVol. 33 pp. 90338.

    von HagenJürgenAndrew HughesHallett and RolfStrauch2001Budgetary Consolidation in EMUEconomics Papers148 (Brussels: European Commission).

    WeingastBarryKennethShepsle and ChristopherJohnson1981The Political Economy of Benefits and Costs: A Neoclassical Approach to Redistributive PoliticsJournal of Political EconomyVol. 89 pp. 64264.

We wish to thank Mark Flanagan, Mark Horton, Manmohan Kumar, Alex Segura, and George Tsibouris for many helpful comments on an earlier draft of the paper.

Carlos Mulas-Granados, a visiting scholar (Universidad Complutense de Madrid and Economic Office of the President) was in the Expenditure Policy Division of the Fiscal Affairs Department in early 2004.

These countries constitute the J.P. Morgan Emerging Markets Bond Index Global (EMBI), an international benchmark for the risk associated with public debt stocks.

Of course, there may be other reasons for a temporary tightening of fiscal policy, including for demand management purposes.

Alternative definitions were also used to test the robustness of the econometric results (see Section V).

A country’s fiscal position is deemed sustainable when its primary balance is such that debt stock is not increasing as a share of GDP.

Note that in all these studies, success is defined in terms of maintaining the reduction in the debt-to-GDP ratio achieved after a fiscal adjustment episode.

For a discussion of these issues, see Beck and others (2001) and Weingast and others (1981).

We omit the subscript i to indicate an episode for the sake of simplicity.

We take the sample in terms of adjustment years as the baseline, and use the sample by adjustment episodes to test robustness of our conclusions and to compare our results with past research. For example, by using the sample of adjustment years, we replicate the estimation procedures for the panel probit model used by Alesina, Perotti and Tavares (1998); by using the sample of adjustment episodes, we replicate the original specification of Alesina and Perotti (1996) and subsequent work undertaken using this approach (e.g., Purfield, 2003).

Data are drawn from several sources. Data on macroeconomic variables, such as inflation, exchange rates, real GDP growth, and fiscal deficit are from the World Economic Outlook (WEO); disaggregated data on the budget components are drawn from the Government Finance Statistics (GFS); and political variables come from the Database on Political Institutions (DPI) by Keefer and others (2002). The coverage of the government accounts is limited to central government in most cases, except where general government data are available.

The variable that measures decentralization is labeled as “Author” in the aforementioned DPI database. If the states/provinces have authority on fiscal policy, the variable takes the value of one. Authority over “cultural affairs,” or “planning” in communist systems does not qualify.

The countries are: Argentina, Bolivia, Brazil, China, Colombia, Côte d’Ivoire, Ecuador, Egypt, India, Indonesia, Jordan, Lebanon, Mexico, Nigeria, Pakistan, Peru, Philippines, Poland, Russia, South Africa, Thailand, Turkey, Ukraine, Uruguay, and Venezuela.

We could not test the relative role of tax reforms in fiscal adjustment owing to the lack of data.

For broader evaluation of the impact of IMF-supported programs, see among others, Khan (1990); Conway (2000); Dicks-Mireaux and others (2000); and Ivanova and others (2003).

Use of a threshold of 1 percent of GDP led to similar results.

We also used democracy as an alternative indicator of the quality of institutions, but it was not statistically significant in any specifications.

The results are available from the authors upon request.

This confirms the aforementioned findings of Buliř and Moon (2003).

The statistical significance of the difference between regression coefficients was assessed using t-tests (at the 5 percent level).

The average distance for the whole sample was 2.7 percentage points.

Other Resources Citing This Publication