An Empirical Analysis of Fiscal Adjustments
Author:
Mr. C. John McDermott https://isni.org/isni/0000000404811396 International Monetary Fund

Search for other papers by Mr. C. John McDermott in
Current site
Google Scholar
Close
and
Mr. Robert F. Westcott https://isni.org/isni/0000000404811396 International Monetary Fund

Search for other papers by Mr. Robert F. Westcott in
Current site
Google Scholar
Close

This study uses the fiscal expansion and consolidation experiences of the industrial countries over the period 1970 to 1995 to examine the interplay between fiscal adjustments and economic performance. A key finding is that fiscal consolidation need not trigger an economic slowdown, especially over the medium term. Fiscal consolidation that concentrates on the expenditure side, especially transfers and government wages, is more likely to succeed in reducing the public debt rat it than tax-based consolidation. Also, the greater the magnitude of the fiscal consolidation, the more likely it is to succeed in reducing the debt ratio.

Abstract

This study uses the fiscal expansion and consolidation experiences of the industrial countries over the period 1970 to 1995 to examine the interplay between fiscal adjustments and economic performance. A key finding is that fiscal consolidation need not trigger an economic slowdown, especially over the medium term. Fiscal consolidation that concentrates on the expenditure side, especially transfers and government wages, is more likely to succeed in reducing the public debt rat it than tax-based consolidation. Also, the greater the magnitude of the fiscal consolidation, the more likely it is to succeed in reducing the debt ratio.

BUDGET DEFICITS and government debt have increased sharply in most industrial countries over the 1980s and 1990s. As their negative consequences have become more apparent, many countries have been addressing the challenge of reining them in. According to traditional Keynesian theory, reductions in budget deficits can slow the pace of economic growth in the short run and medium run by reducing aggregate demand. This theory has been a worry for policymakers. Giavazzi and Pagano (1990), however, suggest that positive expectational effects might offset or even swamp the standard negative Keynesian effect. In their view, strong actions to reduce a budget deficit may boost demand and growth, not just in the long run but even over the phase of fiscal consolidation. Some have suggested that this outcome might be particularly likely for a country with an especially large government debt and deficit, where consumer and business confidence might be heavily depressed because of the fiscal problem and where financial markets might be imposing a large premium on interest rates. This view could have great policy relevance: if politicians could be convinced that fiscal consolidation might not impose a severe economic growth penalty, they might be less inclined to delay appropriate adjustments.

This study uses the fiscal expansion and consolidation experiences of the industrial countries over the 1970–95 period to examine the interplay between fiscal changes and economic performance. The study addresses several questions: Can fiscal consolidation be good for growth? What transmission mechanisms seem to be important for these non-Keynesian effects? Are Alesina and Perotti (1995a) correct in claiming that the composition of fiscal consolidation influences whether it is likely to be successful, where success is defined as putting a country’s ratio of public debt to GDP on a sustained downward trajectory? Does the world economic environment matter for the success or failure of fiscal consolidation? Through a combination of simple statistical and econometric analysis, this paper finds some evidence of outcomes that are inconsistent with the standard Keynesian story. It is shown, for example, that episodes of fiscal consolidation need not trigger an economic slowdown, especially over the medium term. The paper also suggests that structuring fiscal consolidation primarily around spending cuts, rather than tax increases, tends to increase the chances of success. It also finds that the greater the magnitude of the fiscal consolidation, the more likely it is to succeed in reducing the debt ratio.

The paper has five main sections. The first section summarizes recent empirical work analyzing the relationship between fiscal changes and economic performance. The second describes the theoretical arguments that suggest why the standard Keynesian outcome might not occur. The third section reports the results of our empirical analysis of the relationship between changes in fiscal stance, on the one hand, and economic performance and debt reduction, on the other. The results of an econometric investigation into the factors associated with successful fiscal consolidation are reported in the fourth section. The final section concludes and looks for confirmation of our empirical results in two specific cases.

I. Review of Previous Work

Giavazzi and Pagano’s (1990) paper is one of the first to explicitly describe the possibility that a sharp fiscal contraction could be expansionary for an economy. These authors identify something that they call the consumption puzzle, which they see as evidence that fiscal contractions can have strong expectational effects. After taking into account movements in normal consumption function variables, such as income and wealth. Giavazzi and Pagano find that where governments have undertaken sharp cuts in government consumption—their focus is on Denmark and Ireland in the 1980s—private consumption did not decline as much as normal relationships would have predicted. To explain this “puzzle.” they hypothesize that large fiscal contractions could signal lower future tax burdens, which could lead to an increase in expected lifetime disposable income and, in turn, boost consumption. While the study concentrates on the expectational effects on consumers, it also points out that smaller expected corporate tax burdens can bolster investment spending. In more recent work. Giavazzi and Pagano (1995) report that positive non-Keynesian responses of private consumption are more likely when changes in fiscal policy are large and persistent. They also generalize their earlier result by finding that, in addition to cuts in government consumption, cuts in transfer payments and increases in taxes can also have non-Keynesian effects.

Alesina and Perotti (1995a) perform a cross-country analysis of fiscal adjustments in the industrial countries, concentrating mainly on the composition of the adjustments. They report that fiscal consolidations over the past 25 years have been achieved mainly through tax increases, while fiscal expansions have been driven mainly by expenditure increases. Unlike Giavazzi and Pagano (1995), they report that the size of a fiscal consolidation is not strongly related to the success or failure of the adjustment, with success again defined as putting the ratio of debt to GDP on a sustained downward path. Rather, they conclude that it is the composition of the consolidation that matters: in successful adjustments, the lion’s share of deficit cuts tends to be in spending on transfers and government wages while, in unsuccessful adjustments, there tend to be virtually no cuts in these categories but heavy reliance on either tax increases or reductions in public investment. They find some evidence that investment is “crowded in” during fiscal consolidations, and that International competitiveness, defined as the ratio of home to foreign unit labor costs, improves. Finally, they report that fiscal consolidation tends to be initiated when an industrial country is growing faster than the average for the Group of Seven (G-7) industrial countries.

In another paper. Alesina and Perotti (1995b) analyze the macroeconomic effects of fiscal adjustment and find that, in cases in which the ratio of debt to GDP is successfully put on a declining path, real GDP tends to accelerate and the unemployment rate tends to decline. When progress is not made on the debt ratio, growth tends to slow, and the unemployment rate tends to rise. In both papers, these authors emphasize that a remaining unanswered question is the complex relationship between growth and successful debt reduction.

Roseveare (1996) critiques the work of Alesina and Perotti, claiming that one year is too short a period to qualify as a substantial episode of fiscal consolidation. She also asserts that most of their fiscal contractions have been just year-after bounce backs following fiscal expansions, which are therefore not meaningful. But while a one-year rule does seem to be too short to qualify as a fiscal adjustment effort,1 these do not appear to be mainly bounce back episodes. Many of the countries examined by Alesina and Perotti spent the 1970s and early 1980s getting into debt problems, and much of the late 1980s and 1990s trying to undo the damage. Roseveare also argues that a better definition of fiscal success would he sustained improvements in cyclically adjusted primary balances, that is, cyclically adjusted balances excluding government interest payments.

Bartolini, Razin, and Symansky (1995) study the macroeconomic effects of the fiscal restructuring undertaken in the 1990s in the G-7 countries, using the IMF’s multicountry model (MULTIMOD). Their general conclusion is that fiscal consolidation leads to output losses initially, followed by recovery. More important, they find that, although those countries that rely primarily upon increases in indirect taxes and expenditure cuts face steeper short-run output losses, they can expect quicker recoveries and greater output benefits within a decade than other countries undertaking other forms of fiscal consolidation.

The present paper offers several extensions to the previous work summarized above. While it follows the general approach suggested by Alesina and Perotti, cyclically adjusted rather than unadjusted revenue and expenditure data are used to get a better picture of the underlying movements of these time series. In order to strengthen the durability criterion in identifying episodes of fiscal consolidation, a two-year rule rather than a one-year rule is used. This paper also explores more carefully the dynamics between economic growth and debt-ratio reduction by looking at experiences over various phases of the world business cycle. Finally, this paper adds econometric tests to provide more statistical rigor to the investigation of the factors explaining whether a fiscal consolidation will be successful or not.

II. Theoretical Aspects

As mentioned above, the simplest Keynesian view of fiscal consolidation is that lower government purchases or higher taxes reduce aggregate demand and income directly, and have a multiplied negative impact on output. In fuller Keynesian models, represented, for example, by standard ISLM curve analysis (or as presented in Blinder and Solow, 1973), negative multiplier effects are partially offset by crowding-in effects through lower interest rates and currency depreciation. Neoclassical models emphasize other transmission mechanisms by which reductions in government budget deficits affect an economy, especially wealth effects and expectational effects, which also may outweigh the negative Keynesian multiplier effect on demand and activity.2 A smaller budget deficit could reduce interest rates significantly by reducing the perceived risk that a government would depreciate its public debt via high inflation and thus lead to positive wealth effects on demand. In a country that has suffered from extremely large fiscal imbalances and where the action is viewed as necessary to restore government solvency, reducing the deficit could also lower the default risk premium on interest rates. This, again, could increase the market value of wealth held by the private sector and increase aggregate demand. Or consumers and businesses could view deficit reduction as signaling a permanent reduction in their future tax burden and increasing their permanent income, which could also trigger higher spending and activity.

The widening of fiscal imbalances in industrial countries over the past 25 years is attributable to growth in government expenditures in excess of the growth in either the overall economy or government revenues. While the average ratio of tax revenue to GDP in these countries increased from 28 percent in 1960 to 44 percent in 1994, the corresponding ratio for expenditures rose from 28 percent to 50 percent.3 Given the levels to which taxation has risen and the danger of exacerbating distortions by raising taxation further—to say nothing of the political resistance to such policies—it is reasonable to suppose that control of government spending offers the best means of eliminating these fiscal imbalances. Even in neoclassical models, which attribute expectational effects to deficit reduction, reducing the deficit via distortionary tax increases can clearly have negative real effects. Auerbach and Kotlikoff (1987), for example, argue that the disincentives of capital income taxation can have negative wealth effects and make output fall. Conversely, reducing deficits via spending cuts can increase output by reducing expected future taxes and the disincentives that they entail.

This idea is given further theoretical support by Bertola and Drazen (1993), who show through the use of a model of intertemporal optimizing behavior that, if government spending follows an upward-trending stochastic process and if the public believes that the resulting fiscal imbalance will be cut sharply by tax increases when a specific trigger point is reached, there will be a nonlinear negative relationship between private sector consumption and government spending.4 The basic mechanism underlying this relationship is that changes in current government spending induce expectations of future tax changes in the same direction. In this model, therefore, a spending cut can be expansionary because it lowers the present discounted value of future taxes. However, should the trigger point be reached and a deficit reduction carried out via tax increases, the probability of a spending cut is lowered and the present discounted value of future taxes increased, which could slow economic growth.5

The question of whether deficit reduction will raise or lower demand and output—and the question of how long it will take before the positive effects (if any) materialize—cannot be determined by theory alone; it is an empirical question. The instrument—tax increases or expenditure cuts—that is best suited to achieving a consolidation that reduces the ratio of debt to GDP is also an open question that can be addressed empirically.

III. Empirical Framework

We now examine the empirical evidence on episodes of fiscal adjustment in industrial countries. First, it is necessary to define what is meant by fiscal adjustment and to select the data and the rules needed to identify episodes of consolidation and expansion. In examining these episodes, structurally adjusted primary balances are used, as opposed to actual government balances. This adjustment removes two components from the government budget balance: (1) interest payments, which cannot be directly influenced in the short run by government fiscal policies: and (2) the component of the government balance that is a result of the business cycle. This structural balance is the balance that would arise if both expenditures and taxes were determined by potential rather than actual output.6

The sample comprises annual data from 1970 to 1995 for primary structural balances for 20 industrial countries: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Japan, the Netherlands, New Zealand, Norway, Portugal, Spain. Sweden, the United Kingdom, and the United States. Estimates of primary structural government balances come from the Organization for Economic Cooperation and Development (OECD) Economic Outlook database and from the IMF’s World Economic Outlook database.7 As we want to examine episodes of aggressive policy action, we use only observations with large fiscal adjustments: We define an episode of significant fiscal consolidation as one in which the fiscal balance (the ratio of the primary structural government balance to potential GDP) improves by at least 1.5 percentage points over two years and does not decrease in either of the two years. The latter part of the rule is used to avoid including episodes that have partial reversals. Table 1 lists all 74 episodes that meet this two-year criterion of fiscal consolidation.

Table 1.

Episodes of Fiscal Consolidationa

article image
Source: OECD, OECD Economic Outlook.

A primary structural government balance must increase by at least 1.5 percent of potential GDP over two years and not decrease in either of the two years to register as a fiscal consolidation.

Data before 1990 referto west Germany.

Calculations based on World Economic Outlook database.

To check the robustness of this rule, we considered two alternative rules for identifying a significant fiscal consolidation: (1) the primary structural fiscal balance increases by at least 1.5 percentage points in one year, and (2) the primary structural fiscal balance increases by at least 2.0 percentage points over three years and does not decrease in any of the three years. The one-year rule is subject to the problem of including episodes that were reversed the following year; it also has the potential to rule out episodes that achieved significant fiscal consolidation via smaller but steady improvements over more than one year. However, the three-year rule degrades the quality of a sample by sometimes registering consolidations after they were clearly over. Overall, we determined that the two-year rule offered the best compromise between excluding obvious cases of policy reversal and including cases of small but steady improvements over more than one year. The list in Table 1 includes several well-known episodes of multiyear fiscal adjustments, such as Denmark from 1983 to 1986, Ireland from 1982 to 1984 and 1986 to 1989, and New Zealand from 1985 to 1987 and in the early 1990s.

The next step is to determine which episodes of fiscal consolidation identified in Table 1 were successful. Determining what constitutes a successful fiscal consolidation is open to different interpretations, but, broadly speaking, if the ratio of public debt to GDP starts to decline and stays on a declining trend because of the discretionary change in fiscal policy, we follow Alesina and Perotti (1995a) in labeling this a successful consolidation. In a definitional sense, the change in the ratio of debt to GDP depends upon the primary balance, movements in the interest rate, and changes in GDP, the latter two of which could be caused by business cycle factors. In a subsequent subsection, we examine the complex relationship between the debt ratio, on the one hand, and business cycle factors and interest rate movements, on the other. For most of the analysis in this paper, we concentrate on changes in the cyclically adjusted primary balance and view this balance as the key determinant of success.

We examine three versions of this definition of successful fiscal consolidation: (1) a reduction of at least 3 percentage points in the ratio of gross public debt to GDP by the second year after the end of the two-year fiscal tightening; (2) the same as (1), except that GDP is replaced by potential GDP; and (3) a reduction of at least 5 percentage points in the debt ratio by the third year after the end of the two-year fiscal tightening. Successful fiscal consolidation episodes determined by these rules are shown in Table 2. Using rule (1), there are 14 successful episodes and 48 unsuccessful episodes.8 Using rule (2), there are 12 successful episodes, most of which are the same as those identified under rule (1). As might be expected, fewer cases satisfy the more stringent rule (3), but episodes that satisfy rule (3) usually also satisfy rule (1). In this study, we focus on the successful cases determined by rule (1). This criterion is satisfied by virtually all of the well-documented fiscal consolidation successes of the period, including Denmark from 1983 to 1986, Ireland from 1986 to 1989, and New Zealand in the early 1990s.

Table 2.

Episodes of Successful Fiscal Consolidation

article image
Source: OECD. OECD Economic Outlook.

Under rule (1), gross debt must be 3 percent of GDP lower two years after the fiscal consolidation to register as a success. Under rule (2), gross debt must be 3 percent of potential GDP lower two years after the fiscal consolidation to register as a success. Under rule (3), gross debt must be 5 percent of GDP lower three years after the fiscal consolidation to register as a success. The unsuccessful fiscal consolidations can be derived by comparing Table 1 with Table 2. Because the definition of success depends on the change in the gross public debt two or three years after the consolidation, 12 episodes of fiscal consolidation that occur after 1993 cannot be classified under the two-year rule, while 16 episodes cannot be classified under the three-year rule. In addition, 5 episodes are not classified because data on public debt were unavailable (Australia, 1977, 1978, 1981, 1982; New Zealand, 1986).

Calculations based on World Economic Outlook database.

We also identify episodes of fiscal stimulus. Following the methodology outlined above, an episode of significant fiscal stimulus is defined as a period in which the primary structural fiscal balance declines by at least 1.5 percentage points over two years without increasing in either year. Table 3 shows 74 such episodes of fiscal stimulus, including the well-known multi-period fiscal expansions in Japan in 1991–94 and in Sweden in 1990–93.

Table 3.

Episode of Fiscal Stimulusa

article image
Source: OECD, OECD Economic Outlook.

A primary structural government balance must decrease by at least 1.5 percent of potential GDP over two years and not increase in either of the two years to resister as a fiscal expansion.

Data before 1990 refer to west Germany.

Calculations based on World Economic Outlook database.

Fiscal Adjustments and Economic Performance

The next step in the analysis is to determine the consequences of fiscal consolidation for a series of economic performance measures, such as real GDP growth, employment growth, and changes in the unemployment rate, interest rates, exchange rates, and the shares of the major components of GDP. In order to correct for the effects of the world business cycle, real GDP growth, employment growth, and changes in the unemployment rate are measured as differences from weighted averages of all industrial countries’ rates; interest rates are measured as differences from weighted averages of the seven major industrial countries’ rates. Movements in consumption, investment, and net exports are expressed as shares of GDP.

Table 4 follows the paths of these macroeconomic variables by reporting averages for the year before the episodes of significant fiscal action, for the two-year period during which the action takes place, and for the year following. With respect to the episodes of consolidation, real GDP growth on average slightly outpaced average industrial country GDP growth in the year before fiscal contractions were undertaken but fell somewhat below average growth over the two-year fiscal consolidation phase and the year after. Employment growth in countries undergoing consolidation also tended to drop, and the unemployment rate rose faster than in the industrial countries as a whole. But in the 14 episodes in which countries were successful in reducing their ratios of public debt to GDP (as listed in rule (1) of Table 2), economic growth and job creation increased both in the adjustment phase and in the year after; the unemployment rate declined; both short-and long-term real interest rates fell; and currencies appreciated in real effective terms, following strong depreciations in a few cases prior to the consolidation efforts.9 Thus, successful fiscal consolidations tended to be associated with successful economic performance. Furthermore, in contrast to the findings of Giavazzi and Pagano (1990), most of this economic growth was from investment growth rather than consumption growth.

Table 4.

Macroeconomic Effects of Fiscal Actions, 1970–95

article image
article image
article image
Sources: OECD. OECD Economic Outlook; and IMF, World Economic Outlook database and International Financial Statistics.

Differential with respect to industrial country average.

The real interest rate differential with respect to the major industrial country average is constructed by using the short rates as defined in Table 36 of the OECD Economic Outlook less the annual percentage change in the GDP deflator.

The real interest rate differential with respect to the major industrial country average is constructed by using the long rates as defined in Table 36 of the OECD Economic Outlook less the three-year average of the annual percentage change in the GDP deflator.

For fiscal expansions, the 74 episodes are also divided into two categories in Table 4: (1) those in which debt increased significantly (45 cases), and (2) those in which it did not (23 cases). The former category identifies cases in which the ratio of debt to GDP increased by at least 3 percentage points over the two years following the fiscal stimulus. The latter category comprises cases in which the ratio of debt to GDP increased by less than 3 percentage points over the two years following the fiscal stimulus.10 On average, in cases of fiscal stimulus, real GDP growth and employment creation improved relative to average industrial country performance, but the unemployment rate remained above average and even worsened. In the 45 cases of debt-increasing fiscal expansions, the unemployment rate deteriorated markedly, employment creation was largely unchanged, and growth improved marginally, In the 23 cases in which debt was relatively stable. fiscal expansion generated more growth and employment than the industrial country average, while the unemployment rate ultimately fell below its initial position. Both short- and long-term real interest rates increased more during these periods of fiscal expansion than the average rates in the major industrial countries.

Real effective exchange rates significantly depreciated during periods of significant fiscal expansion. In the stable debt cases, these depreciations reversed themselves one year after the expansion, while currencies in the debt-increasing cases continued to depreciate. Debt-increasing fiscal expansions had very strong crowding-out effects on investment, while there was some growth in private consumption. The opposite was true in stable debt cases. Broadly speaking, there was a symmetric reaction between fiscal expansions and contractions. Keeping debt under control seems to have had a large influence on private sector investment growth and thus on growth in general.

Factors That Seem to Contribute to Success

Which factors seem to have contributed to the success of fiscal consolidation? Following procedures suggested by Alesina and Perotti (1995a and 1995b) and Giavazzi and Pagano (1995), we find that the size of the fiscal consolidation may be an important factor. The average magnitude of the two-year fiscal contraction was 4.0 percent of potential GDP for the successful cases but only 3.2 percent for unsuccessful cases.11 A timid commitment to fiscal consolidation may be more likely to fail than a strong one. This outcome may be partly due to a nonlinear relationship between fiscal policy and output growth, whereby small reductions in budget deficits may reduce aggregate demand while large adjustments may revive confidence and boost growth. Also, in many of the successful cases, the fiscal consolidation was undertaken as part of a broader adjustment and reform program that may have enhanced the overall credibility of government commitment to the consolidation.

Real short-term interest rates tended to decline in the successful consolidation cases but to increase in the unsuccessful cases (again, relative to movements in a weighted average of real short-term interest rates in the seven major industrial countries). The larger fiscal action in successful cases may have been a factor in restoring financial market confidence and allowing monetary authorities to ease monetary conditions. Real long-term interest rates decreased in both successful and unsuccessful cases, but perhaps for different reasons: improved financial confidence may have been the reason in the successful cases, where the policy initiatives were more vigorous, while weaker economic growth may have caused the decrease in the unsuccessful cases.

The composition of the fiscal consolidation also seems to have played a role. The 62 classifiable episodes of fiscal consolidation were divided into two categories: those in which the deficit was cut mainly (by at least 60 percent) through revenue increases, and those in which it was reduced mainly (by at least 60 percent) through noninvestment expenditure cuts (Table 5). The use of structurally adjusted revenue and expenditure data eliminates what would have been a bias in the selection of expenditure-cutting episodes.12 Of the 17 cases in which most of the policy action was achieved through expenditure reductions, just under half were successful, while among the 37 cases in which the consolidation was achieved mainly on the revenue side, less than one out of six had successful outcomes.13

Table 5.

Characteristics of Fiscal Consolidation, 1970–95

article image
Sources: OECD, OECD Economic Outlook; and IME, World Economic Outlook database and International Financial Statistics.

Structural expenditure cut as a percent of potential GDP (negative value denotes a cut).

As a percent of GDP (negative value denotes a cut).

The importance of the composition of fiscal consolidation is reinforced by the difference between the average structurally adjusted expenditure cut in the episodes with successful outcomes (3.7 percent of GDP) and the average expenditure cut in unsuccessful cases (2.1 percent). There was a particularly large difference in the change in government wage outlays between the two categories. In the successful outcomes, the government wage bill was cut by 0.9 percent of GDP, while the cuts averaged just 0.3 percent in the unsuccessful outcomes. In the successful outcomes, government employment was virtually constant, but it increased in the unsuccessful ones. Government consumption excluding wages was also cut much more sharply in successful than in unsuccessful episodes. Social security payments and transfers were kept in check in the successful outcomes but expanded as a share of GDP in unsuccessful ones.

A few factors that other researchers have claimed to be important for predicting the success or failure of fiscal consolidation do not appear to be significant in this study. The pace of real economic growth in the year before the fiscal contraction, for example, did not seem to be positively related to the success of the action. In fact, a larger share of fiscal contraction cases that were launched in the face of subpar growth, as opposed to above-average growth, ended up succeeding. One interpretation is that some sense of economic crisis associated with slow or negative growth might have been necessary to convince governments to make tough choices—especially larger budget cuts and cuts more focused on expenditures. Finally, real exchange rate movements do not appear on average to have played a major role in explaining why cases of fiscal consolidation succeeded.14 Between the year before the fiscal consolidation and the end of the two-year contraction, the real exchange rate on average appreciated slightly in the successful cases and depreciated slightly in the unsuccessful cases.15 These data would suggest that, on average, the successful cases did not benefit from a depreciation-driven growth in net exports. This view is supported by the data on net exports in Table 4.

Relationship Between Growth and Fiscal Consolidation

Given the interactions between economic growth and changes in public debt ratios, it is difficult to distinguish between the contribution of good growth to successful consolidations and the effect of successful consolidation in boosting demand and growth. One way to analyze this relationship is to examine fiscal adjustments over different phases of the business cycle. This objective is achieved by repeating the previous exercise, which was conducted over the full 1970–95 period, for the period 1980–82, a time of global recession and spiking interest rates, and the period 1984–89, a time of solid industrial country growth and flat or declining world interest rates. Clearly, it is difficult to reduce debt ratios in the midst of a global recession, especially if interest rates are increasing sharply at the same time.16 None of the 7 efforts at fiscal consolidation over 1980–81, for example, including the highly visible efforts in the United Kingdom, were successful in lowering debt ratios, in part because of these economic headwinds. Even in the favorable 1984–89 period, however, only 12 of the 30 cases of fiscal consolidation were successful in reducing ratios of debt to GDP. Thus, although good timing in relation to the world business cycle helps, it does not guarantee success.

The manner in which the fiscal consolidation is conducted still seems to be important, even with a strong global economy. Table 6 reports the macroeconomic effects of fiscal consolidations that took place between 1984 and 1989; it shows that the direction of the macroeconomic effects for both successful and unsuccessful episodes was the same in this subperiod as in the full 1970–95 period. More important, the same characteristics of successful fiscal consolidation cases can still be observed in the data (Table 7). Of the 11 episodes in which the fiscal consolidation was concentrated on expenditure cuts, over one half were successful, while only about one third of the 14 cases that mainly employed revenue-increasing measures were successful.17 As in the full period, successful episodes were associated with strong expenditure cuts, particularly reductions in government wages, other government consumption, and social security.

Table 6.

Macroeconomic Effects of Fiscal Consolidation, 1984–89

article image
article image
Sources: OECD, OECD Economic Outlook; and IMF, World Economic Outlook database and International Financial Statistics.

Differential with respect to industrial country average.

The real interest rate differential with respect to the major industrial country average is constructed by using the short rates as defined in Table 36 of the OECD Economic Outlook less the annual percentage change in the GDP deflator.

The real interest rate differential with respect to the major industrial country average is constructed by using the long rates as defined in Table 36 of the OECD Economic Outlook less the three-year average of the annual percentage change in the GDP deflator.

To further explore the relationship between economic growth and movements in the ratio of debt to GDP, we also search for episodes when the debt ratio declined even though there was little or no fiscal consolidation. The search is conducted by using the following rule: the ratio of gross debt to GDP must have decreased by at least 3 percentage points over a two-year period, while the primary structural balance must have tightened by less than 1.5 percent of GDP. To reduce the chances that an earlier fiscal consolidation might taint the results, episodes of debt reduction that occurred within four years after an identified consolidation have been excluded. Table 8 shows the 21 episodes—11 in the United Kingdom—when debt ratios fell despite the lack of sharp fiscal consolidation. Seven episodes occurred in the early 1970s when the debt ratio was reduced by an “inflation surprise”: inflation was accelerating rapidly, nominal GDP growth was relatively high, and real interest rates were plunging. A number of other episodes, especially in the United Kingdom, were a result of privatization programs. The three episodes in Norway resulted from substantial changes in the extent of government financial intermediation following the acquisition of financial assets in the 1970s. The reductions in debt from 1993 to 1995 in Denmark and Ireland can be attributed to the combination of impressive real GDP growth performance in the two economies and significant drops in debt interest payments because of earlier consolidations. On balance, the data suggest that, aside from some special cases, rapid GDP growth alone has not usually caused the ratio of debt to GDP to decline noticeably. In most cases, a substantial fiscal consolidation effort, as defined earlier in this paper, is a necessary condition for successful debt reduction.

Table 7.

Characteristics of Fiscal Consolidation, 1984–89

article image
Sources: OECD. OECD Economic Outlook; and IMF, World Economic Outlook database and International Financial Statistics.

Structural expenditure cut as a percent of potential GDP (negative value denotes a cut).

As a percent of GDP (negative value denotes a cut).

Table 8.

Episodes of Debt-Ratio Reductions with Little or No Fiscal Consolidationa

article image
Source: OECD, OECD Economic Outlook.

Gross debt must decrease by at least 3 percent of GDP over a two-year period following the year indicated, while the primary structural government balance must be less than 1.5 percentage points of GDP. The debt-ratio reduction must be at least four years from any fiscal consolidation period indicated in Table 1.

IV. Logistic Probability Model

Predicting Debt-Reducing Fiscal Consolidations

We report below the results of an econometric investigation into the factors associated with successful fiscal consolidations. This is a binary outcome problem, for which the logistic probability model was used. The goal is to predict the probability that the ratio of debt to GDP will decline by at least 3 percentage points within two years following an episode of fiscal consolidation, conditional upon information about the implementation of the consolidation and upon the macroeconomic environment. Using a logistic probability model allows comparisons with the previous section, as the criterion for success is identical. The model is

log ( P i 1 P i ) = β 0 + β 1 f i + β 2 D i + β 3 g i + β 4 g i w + ϵ j , ( 1 )
A03lev3sec9

where Pi is the probability that episode i will be successful,fi is the two-year fiscal impulse with sign reversed (the cumulative two-year change in the cyclically adjusted primary balance), Di is a dummy variable representing the composition of the consolidation (D = 1 if at least 60 percent of the fiscal improvement comes from current expenditure cuts and 0 otherwise), gi is domestic GDP growth minus a weighted average of industrial country growth during the period of consolidation, and gwi is the weighted average of industrial country growth during the period of consolidation.

The fiscal impulse is included as a predictor of a successful consolidation because it has a direct effect on the ratio of debt to GDP and, thus, future debt-servicing requirements, as well as an indirect effect that works through confidence and expectations channels. The dummy variable is included to test whether, because of the influence of distortionary taxes or the nonlinear relationship between government spending and economic activity, expenditure cuts are a better instrument for stabilizing the ratio of debt to GDP than revenue measures, A dummy variable is used for this purpose to avoid problems with extreme outliers: the proportion of the consolidation attributable to expenditure cuts, if used, could reach extreme values (exceeding 100 percent) if expenditure cuts are accompanied by tax cuts. Finally, growth variables are included to control for the effects of the domestic and global business cycles.

The logistic model is estimated using a maximum likelihood nonlinear estimation routine and a sample of 60 fiscal consolidation episodes.18 The individual Pis are not observed; instead, we have information on whether each consolidation episode was successful or not. Thus, the measured dependent variable is yi = 1 if the consolidation is successful and yi = 0 otherwise. We order the observations so that those associated with successful consolidations are the first n1 observations and the remaining n2 (n1 + n2 = N) observations are those associated with unsuccessful consolidations. The likelihood function to be maximized is then

L = Π i = 1 N [ 1 + exp ( x i β ) ] y i [ exp ( x i β ) ] ( 1 y l ) , ( 2 )
A03lev3sec9

where xi is a vector of independent variables for episode i, and β is a vector of coefficients to be chosen to maximize L. Standard optimization routines can be used to maximize this function to obtain estimates of the vector β. The maximum likelihood estimation procedure has a number of desirable statistical properties. All the parameter estimates are consistent. asymptotically efficient, and asymptotically normal, so that the analog of the t-tests can be applied.

The estimation results (with asymptotic standard errors in parentheses) are

log ( P i 1 P i ) = 5.698 ( 1.553 ) + .571 ( .266 ) f i + 1.721 ( .874 ) D i + .375 ( .166 ) g i + .385 ( .183 ) g i w , ( 3 ) LR=21 .2 Psudo  R 2 = .32
A03lev3sec9

where the LR test is a test of the significance of the entire logistic model and follows a chi-square distribution with 4 degrees of freedom. The results indicate that two policy factors may be important in improving the chances of success: the magnitude of the consolidation, as captured by the fiscal impulse measure, and the dummy variable indicating whether the consolidation was undertaken via higher taxes or lower expenditures. The importance of these factors can be seen in Figure 1, which shows that, regardless of the size of the two-year fiscal impulse, the probability of a successful consolidation increases dramatically if it is conducted mostly via expenditure cuts.19 Even with a fiscal improvement of 7 percent of potential GDP over a two-year period, the probability of a successful consolidation based on tax increases is about 0.5. However, a consolidation based on expenditure cuts needs just a 4 percent improvement in the structural primary balance over two years to have about a 50 percent chance of success.

Figure 1:
Figure 1:

Probability of Successful Fiscal Consolidation: Expenditure Cuts Versus Revenue Increases

Citation: IMF Staff Papers 1996, 004; 10.5089/9781451930931.024.A003

A nonpolicy factor significant in determining the probability of success is the stage of the world business cycle.20 The results imply that, for a fiscal episode that has a 0.5 probability of success, a 1 percentage point increase in average industrial country GDP growth will increase the probability of success by 0.1.21 Figure 2 shows the different probabilities of success generated by weak and strong world economic growth. The curve showing the probability of success when worid economic growth is weak is calculated under the assumption that growth is 0.8 percent a year, the average growth rate of the industrial countries over the recessionary period 1980–82. The curve showing the probability of success when growth is strong is calculated with a growth assumption of 3.7 percent a year, the average growth rate of the industrial countries over the period 1984–89. With a two-year fiscal impulse of 4.0—the average for successful cases—the probability of conducting a successful consolidation during the period of strong growth was about 0.5 greater than during the period of weak growth. The probability of success during the strong world economic growth period of 1984–89 was nearly 0.2 greater than at times of average growth. The probability of success was 0.3 less with slow growth than with average growth. In short, strong global economic growth helps to achieve a successful consolidation, and weak global growth reduces the chances that consolidation will cut the ratio of debt to GDP.

Figure 2:
Figure 2:

Probability of Successful Fiscal Consolidation: Strong Versus Weak Global Economic Growth Environment

Citation: IMF Staff Papers 1996, 004; 10.5089/9781451930931.024.A003

One potential problem is that even a large cut in a very large deficit is not likely to be successful in reducing the debt ratio. However, we have found that this factor does not strongly affect the results. For the industrial countries over 1979–94. there were 23 instances when the general government primary deficit exceeded 5 percent of nominal GDP: Italy (1979, 1981, and 1985). Belgium (1981), Denmark (1981 and 1982), Finland (1992 and 1993), Greece (1981, 1985, 1986, and 1989), Ireland (1979, 1980, 1981, 1982, and 1983), Portugal (1981), Spain (1982), and Sweden (1982, 1992, 1993, and 1994). Only three of these, however, were episodes of fiscal consolidation in which the debt-GDP ratio was not reduced: Greece (1986) and Ireland (1982 and 1983). The results of rerunning the regression reported in equation (3) with these three observations deleted are statistically equivalent to the previous results. If anything, the fit of the model is improved. In short, the results support the finding that the size and type of the fiscal adjustment affect the probability of reducing the debt-GDP ratio.

Predicting Debt-Increasing Fiscal Expansions

We repeat the exercise of the previous subsection, which investigated the factors associated with successful fiscal consolidations, in order to determine the factors associated with debt-increasing fiscal expansions. The variables in the model are defined as in equation (1), but this time Pi is the probability that the episode of fiscal expansion will increase the ratio of debt to GDP by at least 3 percentage points within two years following the expansion. The logistic regression is estimated using a maximum likelihood nonlinear estimation routine and a sample of 66 fiscal expansion episodes.22 The estimation results (with asymptotic standard errors in parentheses) are

log ( P i 1 P i ) = 1.708 ( .868 ) .267 ( .220 ) f i .354 ( .115 ) g i w , ( 4 ) LR=12 .48 Psudo  R 2 = .14
A03lev3sec10

where the LR test is a test of the significance of the entire logistic model and follows a chi-square distribution with 2 degrees of freedom.

While the regression does not fit as well as the regression in the previous subsection, it does show the importance of weak economic growth for debt dynamics. If a global recession occurs during a significant fiscal expansion, the probability increases that the ratio of debt to GDP will rise by more than the 3 percentage point threshold level. For a fiscal expansion that has a 50 percent chance of increasing the ratio of debt to GDP by at least our threshold level, a 1 percentage point drop in the industrial country growth rate will increase the probability by a further 0.9.

V. Conclusions and Some Specific Cases

The analysis presented in this paper suggests that a policy of tight fiscal consolidation does not need to cause a recession. In standard cases, fiscal consolidation often has somewhat depressing economic effects in the short run, and the positive effects may appear only after a lag. In some special cases, however, the beneficial effects may appear quite early. Why? Sharp fiscal contractions can reduce premiums on interest rates and boost investment spending, and expectations of lower future lax liabilities can encourage consumption and investment; both these forces can support aggregate demand and economic activity. The evidence also supports the Alesina and Perotti view that fiscal consolidation concentrating on the expenditure side, especially transfers and government wages, is more likely to succeed in reducing the public debt ratio than tax increases. It also appears that the greater the magnitude of the fiscal consolidation, the more likely it is to succeed in reducing the debt ratio, perhaps because it is viewed as more credible and more likely to unlock positive expectational effects. This paper’s finding that the size of the fiscal consolidation is significant is opposite to that obtained by Alesina and Perotti.

There is a relationship between the world economic growth environment and debt-ratio reduction. Evidence indicates that efforts at fiscal consolidation undertaken in adverse world growth and interest rate environments, like the 1980–82 global recession, will probably fail. However, the magnitude of the policy adjustment and its composition seem to be the dominant determinants of the outcome. Therefore, it is not necessary to wait for a particularly favorable world growth phase before initiating fiscal consolidation, but there may be a few particularly bad periods to try to avoid.

Several famous cases of fiscal consolidation that appear to have had non-Keynesian effects have been widely documented and serve to confirm the analytical indications presented in this paper. The unsuccessful efforts in Ireland in the early 1980s and the successful efforts in Ireland and Denmark in the late 1980s, for example, are examined in detail in Giavazzi and Pagano (1990); the case of Sweden in the early 1990s is described in Giavazzi and Pagano (1995). Discussed below are two additional cases that corroborate the analysis presented here: one a success and one a failure.

New Zealand is a recent success case that seems to confirm the policy message of this paper: an industrial country with a serious deficit problem should pursue a strict fiscal consolidation strategy, focusing on expenditure cuts. If policies are considered credible, interest rates can decline, economic growth can be maintained, and the net public debt ratio can fall steadily. New Zealand’s fiscal position shifted from a deficit of 5 percent of GDP in 1991/92 to a surplus of 3 percent of GDP in 1994/95, reflecting above all else structural measures that strengthened expenditure control. Expenditures as a share of GDP dropped by 10 percentage points over these years, while the revenue share remained stable. Interest rates declined noticeably. Even in the face of this fiscal contraction, real GDP growth revived—jumping from a decline of 2.5 percent in 1991/92 to a 5.4 percent increase in 1994/95—while the unemployment rate was cut in half. Meanwhile, the ratio of net public debt to GDP dropped from 52 percent to 38 percent.

The difficulties of achieving a successful fiscal consolidation are illustrated by the case of the United Kingdom in the early 1980s, which also confirms this paper’s analysis and empirical results. Over the three-year period from 1980 to 1982. the U.K. primary structural balance (including net capital outlays) improved by nearly 5 percent of potential GDP; despite this considerable effort, however, the ratio of public debt to GDP increased by 7 percentage points between 1980 and 1984. This consolidation effort faced three difficulties. First, it was attempted in the midst of a world economic recession, during which interest rates were very high. Over 1980–82. industrial country GDP growth averaged just 0.8 percent a year, as mentioned. Second, real GDP declined in the United Kingdom by nearly 4 percentage points more than real GDP in the industrial countries, and the unemployment rate rose 3.5 percentage points relative to the rate in the rest of the industrial world. In part, this outcome reflected a contractionary monetary policy in the United Kingdom over this period, which also led to a sharp appreciation of the pound sterling and reduced International competitiveness. It seems unlikely that any attempt at fiscal consolidation could have succeeded in reducing the debt ratio in such an environment.23 Third, partly because of the economic contraction, the fiscal policy mix was not favorable. Social security benefits rose by 2.5 percent of GDP over this period and accounted for the bulk of the 4.5 percent of GDP increase in overall government spending (excluding net capital outlays). This sharp increase in expenditures was matched by an increase in tax revenues of 4.5 percent of GDP, implying that the consolidation effort came wholly through a reduction in net capital outlays. In short, the world business cycle, domestic monetary policy, and the composition of fiscal policy were all working heavily against successful debt reduction in the United Kingdom in the early 1980s.24

REFERENCES

  • Alesina, Alberto, and Roberto Perotti, 1995a, “Fiscal Expansions and Adjustments in OECD Countries,” Economic Policy, Vol. 21 (October), pp. 20548.

    • Search Google Scholar
    • Export Citation
  • Alesina, Alberto, and Roberto Perotti, 1995b, “Reducing Budget Deficits,” paper presented at the Conference on Growing Government Debt: International Experiences. Economic Council of Sweden, Stockholm, Sweden, June.

    • Search Google Scholar
    • Export Citation
  • Auerbach, Alan J., and Laurence J., Kotlikoff 1987, Dynamic Fiscal Policy (Cambridge: Cambridge University Press).

  • Bartolini, Leonardo, Razin, Assaf and Symansky, Sieve 1995, “G-7 Fiscal Restructuring in the 1990s: Macroeconomic Effects,” Economic Policy, Vol. 20 (April), pp. 11146.

    • Search Google Scholar
    • Export Citation
  • Bertola, Guiseppe, and Drazen, Allan 1993, “Trigger Points and Budget Cuts: Explaining the Effects of Fiscal Austerity,” American Economic Review, Vol. 83 (March), pp. 1126.

    • Search Google Scholar
    • Export Citation
  • Blinder, Allan S., and Solow, Robert M. 1973, “Does Fiscal Policy Matter?” Journal of Public Economics, Vol. 2 (November), pp. 31937.

    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A., and Razin, Assaf 1992, Fiscal Politic and the World Economy (Cambridge, Massachusetts: MIT Press, 2nd ed.).

  • Francesco, Giavazzi, and Pagano, Marco 1990, “Can Severe Fiscal Contractions Be Expansionary? Tales of Two Small European Countries,” NBER Macroeconomics Annual 1990, ed. by Blanchard Olivier Jean and Fischer Stanley (Cambridge, Massachusetts: MIT Press), pp. 75111.

    • Search Google Scholar
    • Export Citation
  • Francesco, Giavazzi, and Pagano, Marco 1995, “Non-Keynesian Effects of Fiscal Policy Changes: International Evidence and the Swedish Experience,” NBER Working Paper No. 5332 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Roseveare, Deborah, 1996, “Role of Taxes and Expenditures in Fiscal Consolidation” (unpublished; Paris: Organization for Economic Cooperation and Development).

    • Search Google Scholar
    • Export Citation
  • Tanzi, Vito, and Schuknecht, Ludger 1995, “The Growth of Government and the Reform of the State in Industrial Countries.” IMF Working Paper 95/130 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
*

C. John McDcrmott, an Economist in the South East Asia and Pacific Department, received his Ph.D. from Yale University. He was in the Research Department when this paper was written. Robert F. Wescott, Deputy Division Chief in the Research Department, holds a Ph.D. from the University of Pennsylvania. The authors would like to thank Alberto Alesina, Stanley Fischer, Graham Hacche. Flemming Larsen. Paul Masson. Michael Mussa, Ludger Schuknecht, and Steven Symansky for helpful comments. In addition, they are grateful to Michelle Marquardt for her able research assistance.

1

Alesina and Perotti provide a durability dimension by looking at the debt ratio three years after the year of the adjustment.

2

For a useful summary, see Frenkel and Razin (1992, Chapter 7).

4

Although the extension to investment is suggested, it is not carried out formally.

5

Tax increases have little effect on consumption in a neoclassical model because consumers are assumed to know implicitly about all future tax burdens and adjust their spending behavior accordingly. Full Rieardian equivalence assumes that this adjustment is instantaneous and complete.

6

Actually, the only expenditure category that is adjusted for the structural balance is unemployment compensation. Historical elasticities are used to make these adjustments.

7

The OECD is the primary source of structural balance estimates. The mix of data provides the broadest possible country coverage and the longest possible historical perspective.

8

Of the 74 episodes of fiscal consolidation, 12 cannot be classified as successful or unsuccessful, either because debt data were unavailable or the two-year period from the time of the consolidation had not yet expired.

9

The 4.60 percent average depreciation in the year before the consolidation phase in the successful cases was heavily influenced by sharp depreciations in Australia (1987 and 1988) and Ireland (1988 and 1989). In the ten other successful cases. there was no currency depreciation on average in the year before the consolidation.

10

Two episodes cannot be classified because the two-year period from the time of the expansion has not yet expired. In addition, four episodes are not classified because data on public debt were unavailable: Australia (1976 and l984)and Denmark (1975 and 1976).

11

The significance of the size of the fiscal impulse is tested in the next section.

12

In periods of high growth, the cyclical components of expenditure will tend to fall as items such as unemployment benefits are reduced, while the cyclical components of revenue will tend to rise as the tax base increases.

13

Of the 62 episodes of fiscal consolidation, 8 are excluded because they involve a roughly equal mix of revenue and expenditure measures.

14

Notable exceptions may be Australia and Ireland in the late 1980s, and New Zealand in the late 1980s and early 1990s. In those cases, the currencies depreciated sharply in real terms before, and sometimes during, the consolidation period.

15

However, in the year after the two-year contraction, the real exchange rate appreciated sharply in the successful cases.

16

If a fiscal deficit is clearly unsustainable, however, it might be better to start correcting it as soon as possible, even if world growth is weak and the prospects of a decline in the ratio of debt to GDP arc low.

17

Five of the cases could not be classified because they involve a roughly equal mix of revenue and expenditure measures.

18

We exclude 2 episodes from our sample of 62 classifiable episodes (Portugal, 1970 and 1980) because data are not available.

19

The other variables are set to their means as follows:g# = −1. and g = 5.

20

A number of other variables were tried in the regression but proved to be statistically insignificant, including real long-term interest rates, real effective exchange rates, and the debt stock prior to the period of fiscal consolidation.

21

To interpret the effect of a change in a continuous variable χ on the probability of success, we can use the approximation ΔPi ≃ β[Pi(1-Pi)]Δχ.

22

The variable representing home country growth relative to world growth is excluded because it is statistically insignificant.

23

Even so, the United Kingdom’s fiscal consolidation efforts over this period may have been preferable to no policy action. Without these steps, it seems likely that the debt ratio would have increased much more.

24

Some of the fiscal policy actions in this period, however, did appear to set the stage for substantial debt reduction in the mid-1980s and later 1980s, when the world economy and U.K. economy enjoyed economic rebounds.

  • Collapse
  • Expand
IMF Staff papers: Volume 43 No. 4
Author:
International Monetary Fund. Research Dept.
  • Figure 1:

    Probability of Successful Fiscal Consolidation: Expenditure Cuts Versus Revenue Increases

  • Figure 2:

    Probability of Successful Fiscal Consolidation: Strong Versus Weak Global Economic Growth Environment