Received wisdom and beyond: lessons from fiscal Consolidation in the EU.

Larch, Matin ; Turrini, Alessandro

Restoring sustainable public finances in the aftermath of the Great Recession is a key challenge in most EU countries. In order to learn from history, our paper examines consolidation episodes in the EU since 1970. We shed light on the factors that favour the start of a consolidation episode and determine its success. Compared to the existing literature, we add a number of new dimensions n the analysis. First, we explore a broader set of potential ingredients of the 'recipe for success', including the quality and strength of fiscal governance and the implementation of structural reforms. Secondly, we check whether the 'recipe for success' changed over time. Our analysis broadly confirms received wisdom concerning the conditions triggering a consolidation episode and the role of the composition of adjustment for success, with some qualifications related to the role played by government wages, In addition it provides evidence that well-designed fiscal governance as well as structural reforms improve the odds of both starting a consolidation episode and achieving a lasting fiscal correction. We also show that, over time, successful and unsuccessful consolidation episodes have become more similar in terms of adjustment corn position.

Keywords: Fiscal consolidations; fiscal rules; budgetary procedures; structural reforms

JEL Classifications: E63; H30; H63

1. Introduction

The Great Recession has put enormous strain on public finances in the EU. Government deficit and debt ratios surged on the back of both collapsing revenues and government action aimed at averting the meltdown of financial systems and dampening the sharp contraction of economic activity. Between 2007 and 2010 the average deficit ratio in the EU jumped from 0.9 per cent of GDP to close to 7 per cent of GDP and the average government debt level increased from slightly below 60 per cent of GDP to around 80 per cent of GDP. In the majority of EU member states public finances are on an overtly unsustainable path.

A strong rebound in economic activity that would simply reabsorb the fiscal imbalances is not on the cards. On the contrary, the post-2007 financial and economic crisis is expected to have lowered medium-term growth prospects. Moreover, most, if not all, EU member states will soon be facing the budgetary impact of demographic ageing, a process which has been in the offing for a long time. In the absence of reforms it will soon produce significant claims on the public purse. Available projections indicate that, on unchanged policies, public support for the elderly in terms of provision of pension and other old age benefits, healthcare, and long-term care is projected to increase by, on average, some 4 1/2 percentage points of GDP in the EU over the next 50 years.

In 2010 most EU countries switched from fiscal stabilisation to fiscal consolidation. The single most important fiscal policy challenge going forward is to sustain the adjustment process and to achieve a lasting correction of public finances. With a view to drawing lessons from history, this paper reviews the experience with fiscal consolidation in the EU since 1970. It sheds light on the factors that trigger the start of fiscal consolidation episodes and that determine the success of consolidation. Following common practice in the literature, the notion of success used in our analysis refers to a more lasting, as opposed to a merely short-lived, correction of the budgetary position.

Analogous empirical studies that look into fiscal consolidation episodes achieve a series of broadly common findings, in spite of differences in the definitions of consolidation and 'success' used, and notwithstanding the inevitable robustness issues associated with discrete-dependent-variable econometrics performed on relatively small samples (for example Alesina and Perotti, 1995; Ardagna, 2004; Guichard et al., 2007). In particular, consolidations based on expenditure adjustment appear to be more long lasting in their achievements compared with those based on tax increases.

In our paper, we aim to take a step forward from existing analyses and investigate not only the role of the characteristics of consolidation episodes but also that of those framework conditions that may affect the political capital of policymakers or the control of fiscal authorities on budgetary outcomes. To that end, we explore the role of a broader set of control variables, including indicators of the political strength of the ruling coalition, the quality and strength of fiscal governance, and the implementation of structural reforms. These additional controls better qualify the recipes for successful consolidations.

As well as exploring these additional determinants of consolidations and their success, we are interested in checking whether the recipe for successful fiscal adjustment has remained broadly unchanged over time or whether some significant changes have taken place.

The paper is organised as follows. Section 2 prepares the ground for the empirical analysis of fiscal consolidations in the EU by reviewing the existing literature. It provides the definitions of fiscal consolidation and success used in the subsequent empirical analysis, and illustrates some stylised facts in the EU countries. Section 3 explores the conditions that trigger episodes of fiscal adjustment and determine whether consolidations are gradual or of a 'cold shower' type. Section 4 goes a step further and examines the odds for success. It focuses on the features and elements of fiscal consolidation that on average are likely to give rise to a lasting correction. Section 5 summarises and concludes.

2. Preliminaries: definitions, received wisdom, and basic facts

This section sets the scene for our empirical work. It first establishes operational definitions of a consolidation episode and of when such an episode is considered to be successful. It then briefly reviews the main findings of the existing empirical literature on the determinants of successful fiscal consolidation. Finally, the section also provides some basic statistics for the fiscal consolidation episodes in our sample.

2.1 Defining episodes of fiscal consolidation

A definition of successful consolidation involves at least three different elements: (i) a measure of fiscal consolidation; (ii) a reference period over which a given size of consolidation is implemented; and (iii) a criterion discriminating between success and failure.

As regards the measure of fiscal consolidation, we use improvements in the cyclically-adjusted primary budget balance (CAPB), derived as the difference between the nominal primary balance and the cyclical component of the budget. (1) Interest expenditure is excluded because it is generally not considered discretionary, unless exceptional measures are taken to reduce debt.

The most commonly used measure in the literature is an indicator proposed by Blanchard (1990). It attempts to isolate discretionary components by calculating the primary budget balance that would have prevailed if the rate of unemployment had remained unchanged with respect to the previous year. In the 1990s measures of this type were clearly preferred over more complex indicators involving potential output estimates such as the CAPB (see Alesina and Perotti, 1995; Alesina and Perotti, 1996 and Alesina and Ardagna, 1998). In the meantime, the CAPB has become the main reference for purging the budget of its temporary cyclical components. It is used by all major international economic organisations including the IMF and the OECD. The cyclically-adjusted budget balance is also the official indicator in the EU fiscal surveillance framework to capture the budgetary effects of discretionary fiscal policy.

Neither the Blanchard-type indicator nor changes in the CAPB are perfect measures of discretionary fiscal policymaking. Their limitations are well known and documented. Apart from capturing the effect of discretionary fiscal policy measures they can also reflect other elements such as one-offs and accounting distortions, autonomous revenue fluctuations and growth surprises. Koen and Van den Noord (2005) examine the distortions arising from one-off measures and accounting issues. Girouard and Price (2004) highlight the role of autonomous fluctuation in revenues that are generally not netted-off when adjusting the budget for the effect of the cycle. Larch and Salto (2005) point to the impact of economic growth surprises.

In our analysis we try to address these measurement issues (i) by choosing sufficiently large changes in the CAPB and (ii) by using specific variables that may control for at least one of these other factors, notably data on apparent tax elasticities.

As regards the size and timing of consolidation, we allow for two different types of consolidation episodes. The first is characterised by a sharp fiscal adjustment effort concentrated in one single year. The second type is one in which the fiscal correction is implemented over a longer period.

Definition 1--Consolidation: A consolidation is an improvement of the CAPB of at least 1.5 per cent of GDP which is either achieved (i) in one single year or (ii) over a period of three years, where in each single year the improvement of the CAPB is less than 1.5 per cent of G DP and the CAPB does not deteriorate by more than 0.5 per cent of GDP compared with the year before. (2)

The relatively high threshold of a 1.5 per cent of GDP improvement in the CAPB was chosen for three reasons. First and foremost, it was chosen because of what we noted above about the 'noise' included in observed changes in the CAPB. Large adjustments are unlikely to result from other factors than discretionary fiscal policy. Second, it is easier to discern differences in the composition of fiscal adjustment if the overall correction is larger. Third, a 1.5 per cent of GDP improvement in the CAPB is in line with that used in most existing empirical studies (e.g., Alesina, Perotti, and Tavares, 1998; for a survey of thresholds used in previous studies see also Giudice, Turrini, and in't Veld, 2003).

Episodes satisfying at least one of the two conditions in Definition 1 are consolidation episodes for the purpose of our analysis. Episodes of the first type will be referred to as 'cold shower' consolidations, to highlight the relatively strong tightening over a period of one calendar year. Episodes of the second type will be called gradual consolidations. It is important to note that the two definitions are mutually exclusive but a 'cold shower' adjustment could be adjacent to a gradual episode. Moreover, the definition of a 'gradual' adjustment formally excludes consolidations of more than 4.5 per cent of GDP over three years. Consolidations of this type are treated as successive 'cold shower' episodes.

The reason for discriminating between these two types of consolidation episodes is straightforward. They can be taken to represent polar cases. In a 'cold shower' adjustment the fiscal correction is concentrated over a short period of time and may potentially reflect a completely different economic environment as well as different institutional arrangements from those of a gradual consolidation episode.

Definition 2--Success: A consolidation in line with Definition 1 is deemed successful if the following condition applies; in the three years after the end of the consolidation episode the CAPB does not deteriorate by more than 0.75 per cent of GDP in cumulative terms compared with the level recorded in the last year of the consolidation period.

In other words, at least half of the overall minimum fiscal correction required to qualify as consolidation has to be safeguarded for the following three years. A consolidation is deemed unsuccessful if this condition is not met. Our definition of success departs from previous work in one important respect; it is not linked to the evolution of the government debt ratio. This was a deliberate choice so as to avoid the clear head start of high debt countries to reduce the debt ratio for a given rate of nominal GDP growth, which is generally positive except in severe recessions. (3)

2.2 Findings from the existing literature

A fairly rich literature has emerged on the determinants and the economic effects of successful fiscal consolidation. In some cases success and economic effects are covered at the same time. In this section we review the results concerning the factors that determine the success of fiscal consolidation. A synthetic overview is provided in table 1.

The first comprehensive empirical analysis of fiscal adjustments is by Alesina and Perotti (1995). It focuses on OECD countries and lays the foundation for the by now familiar notion that the composition of adjustment is crucial for success. In particular, Alesina and Perotti (1995) find that successful adjustments are mainly expenditure based, with a focus on primary current expenditure. This result has been replicated and confirmed by a series of later studies (for instance, Mesina and Perotti, 1997; Alesina and Ardagna, 1998; von Hagen et al., 2002; Briotti, 2004; Lambertini and Tavares, 2005) and is by now accepted as received wisdom.

Part of this received wisdom is also the channel through which expenditure-based consolidations promote the effectiveness of consolidation. Alesina and Perotti (1997) were among the first to point out that the 'right' composition of fiscal adjustment can produce beneficial effects on labour cost developments which, in turn, spur economic activity producing a positive feedback on the consolidation effort.

A further well established fact is the importance of initial conditions. Successful consolidations are typically triggered by a strained fiscal situation during economic bad times. The gravity of fiscal and economic conditions makes it easier for fiscal policymakers to launch a decisive adjustment programme, as the electorate understands the need for action.

Fiscal issues aside, over the years increasing attention has been paid to political factors: for an early overview see Alesina et al. (1998). Single party governments are generally found to be more effective in achieving fiscal consolidation than coalitions, while the political alignment of governments hardly matters. Potentially painful consolidation measures are found to be best implemented during the period soon after an election, when popular support for the government is still running at a high level. The role of political leadership (by the Prime Minister and Finance Minister in particular) in promoting fiscal consolidation and the way consolidation is communicated to the public is often discerned as relevant too.

As to monetary conditions and exchange rates, they have also been identified as factors that determine the likelihood of success of fiscal consolidations, for instance by Alesina and Ardagna (1998) and Lambertini and Tavares (2005). However, the findings are less robust than those concerning the composition of adjustment.

2.3 Fiscal consolidations in the EU: some basic statistics

Our sample covers all 27 EU member states. The time period depends on the availability of the data and is not the same for all countries. The exact sample length by country is reported in table 2 below. Overall, our dataset contains 634 observations of which 146, close to a quarter, qualify as years of consolidation in line with Definition 1. One third of the 146 years of consolidation were crowned with success in line with Definition 2.

Table 2 also summarises some basic statistics. It shows a clear prevalence of 'cold showers', which account for around two thirds of the total number of years in which fiscal consolidations have taken place. Gradual adjustments are significantly less frequent. The clear prevalence of abrupt and sizeable fiscal corrections is also evidenced by the frequency distribution of the annual changes of the CAPB. In close to 70 per cent of the years featuring a consolidation episode in line with Definition 1, the CAPB improved by 1.5 per cent of GDP or more. The high frequency of annual corrections of 3 per cent or more largely reflects the experience of the member states that joined the EU in 2004 or later. With a view to their EU accession, these countries implemented

at times impressive fiscal adjustments. Almost 30 per cent of the consolidation years recorded for the 'new' EU member states gave rise to an annual improvement of the CAPB of 3 per cent of GDP or more.

As regards the distribution over time, the first half of the 1980s, more specifically the years following the second oil price shock, hosted more than a quarter of the overall number of years of consolidation in our sample (see figure 2). After this first wave, the number of episodes dropped significantly in the second half of the same decade, in spite of the fact that only a small share of the corrections implemented in the first half had turned out to be successful.

Against this backdrop, and also in view of the convergence process towards the common currency, which required EU member states to bring the deficit and the debt ratio in line with the provisions of the EU Treaty, fiscal consolidation episodes boomed again in the second half of the 1990s, this time with greater success. More than half of the years of consolidation gave rise to improvements that were at least in part safeguarded in the three years after the end of the consolidation period. The occurrence of fiscal consolidations remained invariably high in the first six years of the 2000s, but the success rate dropped significantly.

2.4 Analysing the data: new dimensions and some issues

In addition to traditional factors, such as the macroeconomic and fiscal conditions prevailing ahead of the consolidation episode, we also explore the role played by fiscal governance and structural reforms in triggering fiscal consolidations and determining their success. Compared with the few existing studies taking into account fiscal governance elements that make use of time-invariant dummy variables (e.g., Guichard et al., 2007), our assessment is based on time-varying indicators of numerical fiscal rules (see European Commission, 2006, and Debrun et al., 2008) and country-specific indexes summarising characteristics of the budgetary process with relevance for fiscal outcomes (e.g., von Hagen and Harden, 1995). A detailed description of the datasets underlying our work is provided in table A1 in the Appendix.

The dependent variables are discrete indicators. When analysing the start of a consolidation process, the dependent variable takes value 1 if an episode that qualifies as consolidation starts in a particular year, in a particular country, and zero otherwise. The dependent variable is also discrete when analysing the type of consolidation, i.e., 'cold shower' versus gradual. When the analysis focuses on successful consolidations, the variable takes value 1 if a consolidation qualifies as successful, zero if the same consolidation is instead classified as non-successful.

The number of data points available for the analysis will be limited not only by the availability of data for all the explanatory variables but also by the dependent variable. When analysing the success of consolidations, the sample will be limited to country/year combinations where consolidations take place.

The use of discrete dependent variable techniques such as probit and logit pose an issue in samples of limited size, as this could lead to likelihood functions lacking the necessary properties for well-behaved maximisation. For this reason, specifications will be kept parsimonious in the econometric analysis. In particular, fixed effects will not be included (although coefficient standard errors will be robust with respect to clustering within countries to address the impact of country-specific factors on statistical inference) and the simultaneous inclusion of many regression variables will have to be avoided. Our empirical strategy therefore consists in augmenting a baseline specification for each of the issues investigated, with additional control factors, which will be included one by one sequentially. Although this strategy has some limitations (in particular, it does not allow assessment of the interaction among several additional control factors), it does allow benchmarking of the performance of alternative explanatory factors when augmenting baseline specifications.

3. The determinants of fiscal consolidations

The aim of this section is twofold: to shed light on the factors that have an impact on the probability of starting an episode of fiscal consolidation and to identify the factors that determine the type of fiscal adjustment, i.e. gradual versus 'cold shower'.

3.1 What triggers fiscal consolidations?

Table 3 below displays the results of probit regressions. In addition to a baseline specification which includes the initial headline deficit and the output gap, the probit regressions were, as mentioned before, run sequentially by adding individually different types of explanatory variables. In the table, the output of the estimations is reported by thematic groups of explanatory variables: political factors, fiscal governance and structural reforms.

In line with the findings of previous studies, years of consolidation are preceded by fiscal hardship. The estimated coefficient of the headline deficit has the expected positive sign and is statistically significant. This means that a fiscal correction becomes more likely as the state of public finances deteriorates. The political economy behind this result is intuitive and can be taken to reflect the mechanics of a war of attrition (Alesina and Drazen, 1991). Changes to the prevailing course of fiscal policy are put off until there is the necessary awareness that further delays may be more costly than taking corrective measures.

The results related to initial economic conditions (as measured by the output gap) contrast with those found in a series of previous papers, where it is found that the chances to engage in a fiscal correction increase as economic conditions turn sour. Our results would seem to support the opposite conclusion; the estimated coefficient of the output gap is positive and significant. There are two possible explanations for this result. First, the behaviour of fiscal authorities can be genuinely countercyclical. If this is the case, major consolidations are not to be expected when economic conditions are weak. Second, the output gap estimates used in the probit regressions are not those available in real time, i.e. those available to fiscal authorities at the moment decisions are taken. They represent ex post estimates based on the information at hand in 2007. As shown by Forni and Momigliano (2004), the behaviour of fiscal policymaking, as measured by the annual changes in the cyclically-adjusted budget balance, is better explained by the output gap estimates available in real time. This second interpretation is corroborated by proxying cyclical conditions by the rate of unemployment, a variable not subject to ex post revisions like the output gap. The probability of starting a fiscal correction increases with the number of employees searching for a job. This finding--not reported in table 3--is confirmed by a simple mean comparison, which shows that years preceding a fiscal consolidation are characterised by a higher rate of unemployment than 'normal' years.

The role of political factors in shaping the occurrence of fiscal corrections is captured by three variables: parliamentary elections, the strength of the ruling coalition in parliament and the Herfindahl index measuring the fragmentation of national parliaments. (1) The expectation is that the likelihood of fiscal consolidations is higher immediately after parliamentary elections, i.e. at the beginning of a new political term, as well as with the strength of the political backing of the government in parliament or the concentration of political power. The estimated impact of the election dummy, after controlling for fiscal and cyclical conditions, has the expected sign but is not statistically significant at conventional levels. The same holds for the size of the ruling coalition in parliament. Somewhat surprising is the result for the variable capturing the degree of concentration or fragmentation of political parties. On the face of it, the estimate suggests that after controlling for the state of public finances and the economy, the likelihood of fiscal correction occurring increases if the parliament is more fragmented, a result for which we have no clear interpretation.

Regarding aspects of fiscal governance, in this paper we focus on two aspects: numerical fiscal rules and budgetary procedures. A comprehensive database covering features of fiscal governance has been built by the Directorate-General for Economic Financial Affairs of the European Commission over the past few years. It covers the EU-25 member states over the period 1990-2005. The most developed section of the database is on fiscal rules. It formed the basis of the analysis presented in European Commission (2006) and in Debrun et al. (2008). (2) In the database, we use, in line with Kopits and Symansky (1998), a numerical fiscal rule defined as a permanent constraint on fiscal policy, expressed in terms of a summary indicator of fiscal performance, such as the government budget deficit, borrowing, debt or a major component thereof. As for budgetary procedures, we consider all arrangements related to the planning, implementation and monitoring of the annual government budget. The assumption that well-designed fiscal rules and procedures help trigger and carry through consolidations in case the fiscal performance deteriorates is supported by the results of our probit regressions.

The likelihood of starting a fiscal consolidation increases with the coverage and strength of numerical fiscal rules and with the quality of budgetary procedures. In the case of fiscal rules, the statistical significance is relatively weak, but very close to the 10 per cent level. The role played by expenditure rules, a specific category of fiscal rules, turns out to be rather unimportant. This evidence is consistent with recent findings pointing to a less significant budgetary impact of expenditure rules compared with budget balance rules (Debrun et al. 2008).

Episodes of fiscal consolidation may or may not be coupled with other economic policy measures, notably structural reforms. A priori, there could be trade-offs as well as complementarities. The trade-off would reflect the fact that some reforms have a direct budgetary cost. By the same token, reforms and consolidation could also be considered to be complementary on the grounds that some reforms release weight from the expenditure side of the budget, such as a reform of social transfers. The evidence provided in Deroose and Turrini (2005) and European Commission (2005) supports the view that structural reforms are not necessarily implying a looser fiscal stance in the years when they take place, as their statistical significance in a fiscal reaction function is weak.

Our analysis relies on indicators of labour and product market reforms used in IMF (2004) and dummies reporting the year of adoption of reforms of employment protection, unemployment benefits, and pensions from the Rodolfo De Benedetti Foundation (FRDB) database. Our regression analysis actually provides some evidence that specific structural reforms can increase the probability of beginning fiscal adjustment. In particular, the approval by parliament of reforms that reduce the level of unemployment benefits turns out to be a significant factor. The impact of the aggregate indicators of labour and product market reforms from IMF (2004) is also positive on the probability of consolidation, albeit not significantly. The causality implied by these results is not clear cut. It might simply signal that consolidations tend to go along with changes in some expenditure categories that improve the budget (e.g., unemployment benefits) or that reforms in general are a good predictor of the general willingness of fiscal policymakers to bring public finances in order. The evidence for the second conjecture is mixed. The estimated impact of pension reforms, as measured by the indicators provided by FRDB, is also not statistically significant. This is not surprising, as the pension reform indicator used in our regression refers to the approval and not the implementation of reforms, and to the fact that the implementation of pension reforms is often phased out in time.

3.2 What explains the difference between 'cold shower' and gradual consolidations?

Consolidations can be concentrated in a short time-span or distributed over time. What determines the probability of having a 'cold shower' rather than a gradual consolidation? The results are reported in table 4.

While the likelihood of engaging in a gradual rather than a 'cold shower' type of adjustment is negatively, but not strongly, associated with the size of the initial deficit, it significantly increases if the adjustment comes close after an earlier episode of consolidation, specifically within a period of three years. This result is quite intuitive, and suggests that it is politically costly to overburden the electorate with a sequence of episodes of large fiscal corrections.

The probability of a gradual adjustment is also positively, but only weakly, linked to the gravity of the initial cyclical conditions, as measured by the output gap in the year preceding the correction. The interpretation is that fiscal policymakers are more likely to favour a fiscal therapy that spreads out the impact on economic activity so as to make it more palatable to the electorate.

A third interesting result refers to the composition of the expenditure restraints. Fiscal consolidations that rely strongly on a reduction of politically sensitive items such as government wages are more likely to be of the gradual type. Fiscal policymakers will generally face the resistance of public employees, a relatively homogeneous and generally well organised interest group, to abrupt and large reductions in their salaries or to a reduction of staff. A similar reasoning applies to cuts in subsidies or transfers; large savings in this type of expenditure are more likely to be achieved in the context of a gradual fiscal adjustment)

By contrast, larger cuts in expenditure items that do not relate to well defined constituencies, typically investment expenditure, are more likely to be implemented during 'cold shower' episodes. The estimated coefficient relating to changes in government investment expenditure in the respective probit regression reported in table 4 is positive and highly significant, implying that cuts in investment expenditure are more characteristic for short and sharp consolidation episodes.

Among the political factors examined in our regression analysis, the size of the majority in parliament seems to matter. In line with expectations, a larger majority decreases the likelihood of a gradual adjustment as a strong political backing ceteris paribus provides the basis for more decisive action.

The role of fiscal governance and structural reform does not appear as neat as in the case of the analysis of the determinants of the start of fiscal consolidation episodes. The estimated coefficients have very low statistical significance (with the possible exception of expenditure rules, which appear to be positively linked to the probability of gradual consolidations) and the sign of the estimated coefficients does not lend itself to a straightforward interpretation.

4. Determinants of success

In our sample, roughly one out of three consolidation episodes turns out to be successful on the basis of Definition 2. What determines the probability of success? The baseline specification of our probit regressions includes variables gauging the initial economic and fiscal conditions as well as the size and expenditure content of the fiscal adjustment. Additional variables are added individually and are arranged in four different groups: the composition of the fiscal adjustment, other fiscal factors, political factors, fiscal governance and structural reforms. A summary of the estimation results is provided in table 5.

4. I Initial conditions, size and composition of adjustment

Results indicate that the deficit ratio is a statistically significant determinant of success: the worse the initial public finance situation, the higher the probability of implementing a lasting fiscal correction. Large deficits seem to heighten the awareness that significant and sustained correction measures are required to change the status quo.

The estimated coefficient of the change in the CAPB is not statistically significant, and the negative sign suggests that more contained adjustments are not more likely to be reversed. This is not in line with the results of a series of previous analyses, although the findings in the literature concerning the link between the size of adjustment and the likelihood of success are not clear cut (results being driven also by the specific definition of success used in the empirical analysis). (4)

Instead, what appears to make a clear difference for success is the composition of the fiscal adjustment, as measured by the size of the change in cyclically-adjusted primary expenditure. In particular, the likelihood of success increases significantly with savings in primary expenditure net of cyclical factors.

4.2 The composition of expenditure adjustment

We augmented the baseline specification in table 5 with explanatory variables adding information on the composition of the expenditure adjustment during consolidation periods.

A common finding in the literature is that successful consolidations are those that focus on social security and, in particular, on government wages. Our findings suggest instead that there is no individual item of current primary expenditure that stands out as particularly instrumental for the likelihood of success.

A possible reason for the difference between our results and those in preceding analyses may be linked to the sample used. Previous studies largely focus attention on OECD economies, i.e. including a number of non-EU countries, notably the US, Canada and Australia, where cuts in government wages during episodes of fiscal consolidation may have been particularly important. (5) Other possible explanations for the weak link between cuts in wages and success in our EU sample are: (i) wage cuts are likely to be implemented in a gradual way and hence do not produce their full effect in three years after the end of the consolidation period; (ii) wage cuts are politically costly and are coupled with compensatory measures like tax cuts in the short run.

The indirect and probably equally important channel through which fiscal consolidation impacts on wage developments, which in turn contributes to the success of a fiscal correction via stronger economic growth, is the one suggested by Alesina and Perotti (1997). In the framework of a country's wage setting mechanism, wage claims will generally be more moderate if fiscal consolidation does not affect after-tax wages in the economy as a whole. This will typically be the case for expenditure as compared to revenue-based fiscal corrections. Hence, the right composition of adjustment can induce wage moderation in the economy as a whole, including the government sector. Moreover, wage moderation is conducive to sustained economic growth which will feed back to the government sector via revenues. (6)

A second relevant result emerging from our analysis is that strong reliance on cuts in government investment expenditure during consolidation periods reduces significantly the chances of carrying out a successful adjustment. This evidence confirms that government investment is easily cut but also easily raised, so that investment expenditure cuts carried out during consolidation periods are more likely to be reversed over time compared with other expenditure items.

4.3 Political factors

Of the three political indicators considered in our work, two turn out to have a positive, yet statistically not significant, effect on the likelihood for success: the beginning of a political term and the size of the majority of the ruling coalition in parliament. While these two factors seem to matter less for a consolidation to occur, they tend to raise the odds on achieving a more lasting fiscal correction once the decision for a fiscal consolidation is taken. This result does not come as a surprise, as governments with a strong majority in parliament can be assumed to stand a greater chance of implementing more effective fiscal corrections. Similarly, 'serious' or politically more costly measures can also be expected to be taken shortly after elections rather than towards the end of a political term ahead of an election.

4.4 Fiscal governance and other fiscal factors

Results clearly show that, after controlling for initial conditions as well as for the size and the composition of the fiscal adjustment, the presence of fiscal rules and budgetary procedures are conducive to successful consolidation. Somewhat surprisingly, the link between success and fiscal governance is weak when considering expenditure rules alone. The estimated coefficient is not statistically significant and has the wrong sign. One possible reading of this result could be that expenditure rules may impose an excessive focus on expenditure thereby affecting investment expenditure, which by experience is likely to rebound. By contrast, deficit and debt rules provide leeway for combining expenditure cuts with some revenue increases. Alternatively, the weaker role of expenditure rules in explaining the success of consolidations could simply reflect the fact that in practice this type of rule is generally limited to central government, whereas deficit and debt rules cover a larger area of general government public finances.

The link between fiscal governance and the success of fiscal adjustment is likely to work via at least two different channels. First, comprehensive and strong fiscal rules favour discipline-oriented budgets. They provide incentives for designing adjustment measures that stand a higher chance of being effective and lasting, not least in view of the possible costs associated with the risk of running foul of the rules. Second, well designed budgetary procedures favour good planning, a balanced composition and an effective implementation of consolidation measures, as opposed to a situation in which measures are drawn up over a short period of time, in an uncoordinated way and potentially based on not very prudent assumptions.

In addition to the size and the composition of the fiscal adjustment, we have examined two other fiscal factors: the type of adjustment and the behaviour of tax elasticities. Whereas there is no clear a priori evidence concerning the type of adjustment, the expectations linked to tax elasticities are as follows. Empirically, the tax elasticity with respect to GDP, i.e. the relative change of total current taxes with respect to the relative change in nominal GDP, can be subject to significant autonomous fluctuations. Such fluctuations are due to changes in the composition of aggregate demand or changes in the primary distribution of income towards more or less tax rich components. For instance, private consumption expenditure is markedly more tax rich than exports or investment expenditure, and compensations of employees are generally more tax rich than gross operating surplus. If a fiscal consolidation relied on a temporary increase in tax elasticities the chances for success should be lower, as revenues would at some point move back to normal levels.

This conjecture is not confirmed by our regression analysis. The estimated coefficient of the variable controlling for fluctuations of tax elasticity has not the expected negative sign and is not statistically significant. This result is likely to reflect a caveat of the type of tax elasticities used in our regression. As there are no consistent estimates of discretionary fiscal policy measures, neither across time nor across countries, we resorted to apparent tax elasticities. The evident price to pay for using such a crude proxy is that it does not necessarily capture revenue windfalls or shortfalls alone. It also includes revenue measures that may actually have been part of a consolidation package. In such a case an increase in the tax elasticity should a priori not weigh on the probability of success.

4.5 Structural reforms

The last group of potential determinants of success examined in our regression analysis is structural reforms. Ex ante it could be argued that many types of reforms have the potential to improve the chances of success as they should typically result in durable changes in the way public money is spent. For instance, labour market reforms or pension reforms translating into a reduction of benefit levels should, ceteris paribus, produce direct and lasting effects on expenditure. In addition, some structural reforms can also be expected to have a positive impact on economic growth and hence support the success of consolidation via the denominator of the deficit ratio.

A positive link between the probability of success and certain types of structural reform is confirmed for our sample. A first interesting and clear point emerging from the analysis is that the likelihood of success is significantly increased when the fiscal consolidation is linked to or falls in years in which labour and/or product market reforms--as measured by the indicators used in IMF (2004)--are enacted.

The RDBF database of structural reforms allows for a further differentiation between two kinds of labour market reforms: modifications of the employment protection legislation and of unemployment benefits. Both reforms turn out to have a positive impact on the likelihood of producing a lasting correction of the underlying fiscal position, yet only the reform of unemployment benefits seems to produce a statistically significant effect. (7)

The positive effect of labour market reforms in our regression results confirms the findings about the importance of the labour market channel emphasised by Alesina and Perotti (1996) and confirmed by Alesina and Ardagna (1998). Notably, reforms that help to boost price competitiveness and employment creation can also be conducive to a better and sustained fiscal performance.

The final type of reforms explored in our regression analysis are expenditure reforms as defined by Hauptmeier et al. (2006) who examine expenditure reforms enacted in 21 OECD countries over the period 1960-2007. Depending on the degree of effort, expenditure reforms are divided into two categories: 'ambitious' and 'timid' expenditure reforms. Ambitious reforms are identified as episodes in which the primary spending ratio is reduced by at least 5 per cent of GDP over a period of seven years. Timid reforms are episodes in which the reduction is less than 5 per cent of GDP, again over a seven-year period. (8) Although the estimated coefficient for expenditure reforms has the expected positive sign, statistics fall short of conventional significance levels.

4.6 Is the recipe for successful consolidation changing over time?

The issue arises as to what extent the 'recipe' for successful fiscal consolidations emerging from the previous analysis is valid irrespective of the specific time period chosen for the analysis. To provide an answer to this question, we repeat the analysis shown in table 5 for a sample starting after 1989.

Separate probit regressions for this sub-period show that savings in primary expenditure still have a positive impact on the probability of success but the evidence is much weaker (table 6). The estimated coefficient is not statistically significant compared with the regression for the period as a whole.

This result is quite important as it points to a possible shift in the 'recipe' for success over time. Conventional wisdom on the importance of expenditure-based consolidations seems to have lost some of its bearings and there has been some convergence in the composition of adjustment for successful and unsuccessful consolidation episodes.

Looking at the specific expenditure components, it appears that while the explanatory power of government consumption rises when limiting the analysis to years after 1989 (cuts in this component are associated with a higher probability of success), that of government investment is slightly reduced. How then to interpret this shift in the role of expenditure adjustment in driving success?

A first explanation is that, since the early 1990s, the size of government in many EU member states has been smaller compared with previous decades, which reduced the likelihood for further expenditure cuts being sustained over a protracted period. Another important element in this context is certainly the experience ahead of the inception of the Economic and Monetary Union (EMU), when a number of EU member states made relatively large efforts to qualify for the euro. In several cases those efforts involved both a significant increase in government revenues and significant savings on the expenditure side. What made the difference between success and failure was the ability to safeguard the corrections over time, independently of the composition of the adjustment. Structural reforms and numerical fiscal rules might have played an important role in this respect. The consolidation episodes in Italy and Spain in the 1990s are telling examples. In the case of Italy, the heavy reliance on higher revenues was accompanied by measures aimed at capping existing expenditure trends. Such measures did not translate into measurable expenditure savings in per cent of GDP but contributed to setting a ceiling on expenditure dynamics. In the case of Spain, the success of a revenue based consolidation was probably helped by the fact that the overall tax burden was comparatively low. A common feature of both cases is that fiscal consolidations were accompanied by the strengthening of fiscal governance and the implementation of structural reforms.

5. Summary and conclusions

Our empirical analysis of successful fiscal consolidations in the EU highlights a number of important lessons. Some match up with the findings in the existing literature; others are new or somewhat different from conventional wisdom.

In line with previous findings, we show that the likelihood of success increases significantly if initial fiscal conditions are difficult. We also show, again consistent with previous analyses, that cuts in current primary expenditure are more likely to produce a lasting effect than higher revenues or large cuts in government investment. However, our analysis also suggests that the validity of this by now familiar notion needs to be qualified somewhat; since the beginning of the 1990s the composition of adjustment per se seems to have lost some of its weight in securing success.

A number of factors may explain this finding. Firstly, starting from the relatively large size of government in the 1980s, many EU member states have embarked on a path that has measurably reduced the weight of the public sector in the economy. Along this path, the leeway for further expenditure cuts was gradually reduced unless embedded in a structural overhaul of specific functions of government. Secondly, the motivation to participate in the common currency induced member states to implement sizeable consolidation packages that did not necessarily follow the conventional recipe for success. Cuts in primary expenditure still played a role but were complemented by additional elements that improved the sustainability of fiscal corrections even when revenue-based. Improvements in fiscal governance and structural reforms may have played a prominent role in this respect.

On the basis of our empirical analysis, the quality of fiscal governance turns out to be conducive to the success of fiscal consolidation. After controlling for initial conditions and the composition of adjustment, the probability of producing a lasting correction is increased when public finances are covered by numerical fiscal rules and/or effective budgetary procedures. The link between the quality of fiscal governance and the odds for the success of fiscal consolidation is likely to be complex and needs to be examined in more detail. The main point is certainly that effective fiscal governance fosters discipline-oriented budgets and an effective implementation of budgetary plans including fiscal corrections.

The chances of achieving a lasting fiscal correction also increase significantly if consolidation efforts are coupled with structural reforms. Apart from pension reforms, for which the statistical evidence is weak, reforms that aim at improving the functioning of labour and product markets turn out to be conducive to success. The channels through which structural reforms help fiscal consolidation could be twofold: directly by capping or flattening existing expenditure trends and indirectly by spurring economic activity.

A second notable qualification of received wisdom emerging from our analysis relates to the more detailed composition of expenditure cuts. According to our analysis, the main contributors to savings in primary expenditure of successful consolidations are transfers and non-wage government consumption, while cuts in the government wage bill play only a relatively minor role.

Appendix

Table A1. Indicators of the determinates of fiscal consolidation

 Source Description of data from which
 indicators have been constructed

Fiscal governance

Fiscal rule Commission Time-varying index, constructed
coverage and services by combining the information
strength contained in the 'fiscal rule
 coverage index' and the
 information contained in 'the
 fiscal rule strength index'. For
 details see European Commission
 (2006).

Fiscal rule Commission Index summarising, for each
coverage services member state, the information on
 what part of general government
 finance is covered by numerical
 rules (measured as the share of
 government expenditure of the
 general government subsector to
 which the rule applies in total
 general government expenditure).
 All numerical rules are
 aggregated, in case of overlap
 different weights are applied.
 The index hence may exceed 1.

Fiscal rule Commission Index, calculated taking into
strength services account 5 criteria: the statutory
 base of the rule; whether there
 is an independent monitoring of
 the rule; the nature of the
 institution responsible for the
 enforcement of the rule; the
 existence of pre-defined
 enforcement mechanisms; and the
 media visibility of the rule.

Expenditure Commission Index, constructed following
rule coverage/ services exactly the same methodology (but
strength restricting the sample to
 numerical expenditure rules),
 measuring the share of government
 finances covered by expenditure
 rules.

Budgetary Commission Index capturing features national
procedures services budget procedures. It covers 6
 dimensions: transparency, multi/
 annual planning horizon,
 centralisation of the budget
 process, top/down budgeting,
 prudent economic assumptions and
 performance budgeting. Data
 source: OECD/World Bank Budget
 Practices and Procedures Database
 (2003).

Structural reforms

Expenditure Hauptmeier The index consists in the average
reform et al. (2006) annual reduction in primary
 expenditure over a period of 7
 years, starting in the peak year
 with the highest primary
 expenditure-GDP ratio. Original
 data source: Commission services.

Labour market IMF (2004) Labour market index consisting of
reform the unweighted ave. of indicators
 of employment restriction,
 unemployment benefit replacement
 rate and benefit duration. The
 index is normalised in such a way
 as to be between 0 and 1 and to
 increase as labour market
 restrictions are reduced.
 Original data source: Nickell and
 Nunziata (2001), Labour Market
 Institutions Database and data
 used in OECD (2003), World
 Economic Outlook, April 2004.

Product market IMF (2004) Index measuring entry barriers,
reform public ownership, market
 structure, vertical integration
 and price controls in public
 utilities and transport services.
 The index is normalised in such a
 way as to be between 0 and I and
 to increase as product market
 restrictions are reduced.
 Original data source: Nicoletti
 and Scarpetta (2003).

Pension reforms Rodolfo de Data indicating the years in
 Benedetti which reforms in pension systems
 Foundation were implemented and the major
 (FRDB) characteristics of reforms

Unemployment Rodolfo de Data indicating the years in
benefit reform Beneditti which reforms in unemployment
 Foundation benefit systems were implemented
 (FRDB) and the major characteristics of
 reforms.

Employment Rodolfo de Data indicating the years in
protection Beneditti which reforms in EPL were
reform Foundation implemented and the major
 (FRDB) characteristics of reforms.

 Country Year Indicator
 coverage coverage

Fiscal governance

Fiscal rule EU-25 1990- The size of the
coverage and 2005 index is larger
strength than the median

Fiscal rule EU-25 1990- The size of the
coverage 2005 index is larger
 than the median

Fiscal rule EU-25 1990- The size of the
strength 2005 index is larger
 than the median.

Expenditure EU-25 1990- The size of the
rule coverage/ 2005 index is larger
strength than the median.

Budgetary 18 EU 2003 The size of the
procedures member index is larger
 states than the median.

Structural reforms

Expenditure 25 1981- If the index is
reform industrial 2004 larger than 5%
 countries the expenditure
 reform is ambit-
 ious, if it is
 between 0 and
 5%, it is timid.

Labour market EU-14 1970- The yearly
reform (EU-15 1998 * change in the
 countries labour market
 before the index is positive
 2004 en- and larger than
 largement the median
 excl. Lux.) positive change.
 except
 Greece

Product market EU-14 1975- The yearly
reform except 1998 change in the
 Greece product market
 index is positive
 and larger than
 the median
 positive change.

Pension reforms EU-14 1985- A reform making
 2001 the system less
 generous was
 approved by
 parliament in the
 year.

Unemployment EU-14 1985- A reform making
benefit reform 2001 the system less
 generous was
 approved by
 parliament in the
 year.

Employment EU-14 1985- A reform making
protection 2001 the system less
reform generous was
 approved by
 parliament in the
 year

REFERENCES

Ahrend, R., Catte, P. and Price, R. (2006), 'Interactions between monetary and fiscal policy: how monetary conditions affect fiscal consolidation', OECD Economics Department Working Paper no. 521.

Alesina, A. and Ardagna, S. (1998), 'Tales of fiscal adjustment', Economic Policy, 13, pp. 487-545.

Alesina, A. and Drazen, A. (1991), 'Why are stabilizations delayed?', American Economic Review, 81, pp. 1170-88.

Alesina, A. and Perotti, R. (1995), 'Fiscal expansions and fiscal adjustments in OECD countries', NBER Working Paper No. W5214.

--(1996), 'Fiscal adjustment in the OECD countries: composition and macroeconomic effects', NBER Working Paper No. 5730.

--(1997), 'Fiscal adjustments in OECD countries: composition and macroeconomic effects', IMF Staff Papers, 44, pp. 210-48.

Alesina, A., Perotti, R. and Tavares, J. (1998), 'The political economy of fiscal adjustments', Braokings Papers on Economic Activity, I, pp. 197-266.

Ardagna, S. (2004), 'Fiscal stabilizations: when do they work and why?', European Economic Review, 48, pp. 1047-74. Bertelsmann Stiftung (2006), Erfolgreiche Budgetkonsolidierungen im Internationalen Vergleich, Gutersloh.

Blanchard, O. (1990), 'Suggestions for a new set of fiscal indicators', OECD Department of Economics and Statistics Working Paper No. 79.

Blanchard, O. and Wolfers, J. (1999), 'The role of shocks and institutions in the rise of European unemployment: the aggregate evidence', NBER Working Paper No. 7282.

Briotti, G. (2004), 'Fiscal adjustment between 1991 and 2002: stylised facts and policy implications', ECB Occasional Paper No. 9.

Drazen, A. (2000), Political Economy in Macroeconomics, Princeton, Princeton University Press.

Debrun, X., Moulin, L., Turrini, A., Ayuso, J. and Kumar, M. (2008), 'Tied to the mast? National fiscal rules in the European Union', Economic Policy, 23, pp. 297-362.

Deroose, S. and Turrini, A. (2005), 'The short-term budgetary implications of structural reforms: evidence from a panel of EU countries', CEPR Discussion Paper No. 5217.

European Commission (2005), Directorate-General for Economic and Financial Affairs, 'Public Finances in EMU--2005', European Economy, No. 3/2005.

--(2006), Directorate-General for Economic and Financial Affairs, 'Public Finances in EMU--2006', European Economy, No. 3/ 2006.

Forni, L. and Momigliano, S. (2004), 'Cyclical sensitivity of fiscal policies based on real-time data', Applied Economics Quarterly, 3, pp. 299-326.

Giudice, G., Turrini, A. and in't Veld, J. (2003), 'Can fiscal consolidations be expansionary in the EU? Ex-post evidence and ex-ante analysis', European Economy--Economic Papers 195, Directorate General Economic and Monetary Affairs, European Commission.

Giruoard, N. and Price, R. (2004), 'Asset price cycles, 'one-off' factors and structural budget balance', OECD Economic Department Working Paper No. 391.

Guichard, S., Kennedy, M., Wurzel, E. and Andre, C. (2007), 'What promotes fiscal consolidation: OECD country experiences', OECD Economics Department Working Papers No. 553.

Hauptmeier, S., Heipertz, M. and Schuhknecht, L. (2006), 'Expenditure reform in industrialised countries, a case study approach', ECB Working Paper No. 634.

IMF (2004) Euro Area Policies: Selected Issues. IMF Country Report No. 04/235, Washington DC.

Jackman, R., Pissarides, C. and Savouri, S. (1990), 'Labour market policies and unemployment in the OECD', Economic Policy, 5, pp. 449-90.

Koen, V. and van den Noord, P. (2005), 'Fiscal gimmickry in Europe: one-off measures and creative accounting', Economics Department Working Paper No. 417.

Kopits, G. and Symansky, S. (1998), 'Fiscal policy rules', IMF Occasional Papers No. 162.

Lambertini, L and Tavares, J. (2005), 'Exchange rates and fiscal adjustments: evidence from the OECD and implications for the EMU', Contributions to Macroeconomics, 5, Article II.

Larch, M. and Salto, M. (2005), 'Fiscal rules, inertia and discretionary fiscal policy', Applied Economics, 37, pp. 1135-46.

Mourre, G. (2004), 'Did the pattern of aggregate employment growth change in the euro area in the late 1990s?', Applied Economics, 38, pp. 1783-807.

Nickell, S. and Nunziata, L. (2001), 'Labour market institutions database', http://www.econ.upf.edu/-reiter/webbcui/ combineddata/LMIDB.pdf.

Nicoletti, G. and Scarpetta, S. (2003), 'Regulation, productivity and growth: OECD evidence', OECD Economics Department Working paper, No. 347.

von Hagen, J. and Harden, I.J. (1995), 'Budget processes and commitment to fiscal discipline', European Economic Review, 39, pp. 771-9.

von Hagen, J., Hughes Hallett, A. and Strauch, R. (2002), 'Budgetary consolidation in Europe: quality, economic conditions and persistence', Journal of the Japanese and International Economies, 16, pp. 512-35.

Zaghini, A. (1999), 'The economic policy of fiscal consolidations: the European experience', Banca d'ltalia, Temi di discussione, No. 355.

NOTES

(1) The Herfindahl index measures the degree of concentration in parliament with respect to the number of political parties. It takes a large value when the number of parties is low and lower values if the number of parties is high.

(2) The indicators of fiscal rules are relatively complex. They are constructed in such a way as to capture not only the existence of rules, but also their strength and the fraction of general government finances covered by the rule. Similarly, the indicators of budgetary procedures encompass a number of dimensions such as transparency, level of centralisation and prudence. A brief description of the database and its content is in table AI in the Appendix.

(3) The political economy dimensions underlying these results are well known and have been extensively explored in the literature under the heading of vested interests (see for instance Drazen, 2000). Strong policy measures, in our specific case expenditure cuts, that affect well defined or powerful constituencies will encounter fiercer opposition as opposed to measures that concern a broader and heterogeneous group.

(4) The definition used in our analysis is based on a fixed deterioration of the CAPB compared to the last year of the adjustment. As long as this criterion is met, there is no difference between very large adjustments and smaller ones. Our definition may even penalise very large adjustments which, while not meeting the condition of Definition 2, give rise to a larger net improvement of the CAPB. In a number of other studies, in addition to a deficit criterion the stabilisation of the debt ratio is used as a complementary condition for success. In that case success is correlated with the initial debt level; i.e. the larger the fiscal correction the higher the likelihood of stabilising the debt.

(5) This is partly confirmed by Alesina and Ardagna (1998) who provide country-specific information on a number of consolidation episodes in the OECD.

(6) For a discussion of the link between wages, employment and economic activity see for instance Blanchard and Wolfers (1999) and Mourres (2004).

(7) The statistically weaker effect of changes to the employment protection legislation can be reconciled with labour market theory according to which lower protection produces ambiguous effects on unemployment (see for instance Jackman et al., 1990) with no clear repercussions on the budget.

(8) Clearly, such definitions potentially overlap our definition of consolidation with one big exception; they refer to expenditures alone and, hence, depending on what happens on the revenue side, may more or less intersect with fiscal consolidation as measured by the improvement of the primary budget balance.

Matin Larch, Alessandro Turrini *

* European Commission, Directorate General Economic and Financial Affairs. e-mail: martin.larch@ec.europa.eu. We would like to thank S. Deroose, R. Eisenberg, E. Flores, V. Sorebo and D. Prammer for helpful comments. The opinions expressed in this paper are those of the authors and do not necessarily reflect those of the Directorate-General for Economic and Financial Affairs of the European Commission.

doi: 10.1177/0027950111420917

Table 1. Main determinants of successful fiscal consolidations:
a synthetic overview of the evidence from previous empirical
studies

Determinants Main findings

Composition of Cuts in expenditure are more
fiscal effective than tax increases
adjustment in making consolidation
 successful. Reductions in
 public sector employment and
 wages, and in transfers are
 found to be particularly
 conducive. Thus far, this
 result represents
 'conventional wisdom'. More
 recent studies, focusing on
 country cases, provide
 evidence that both expenditure
 and revenue-based
 consolidation can be
 successful.

Size of fiscal The size of fiscal adjustment
adjustment is found to be relevant as it
 may make a consolidation
 harder to reverse. The result
 is not robust across
 alternative studies and seems
 to depend on the definition of
 success.

Initial The gravity of initial
conditions macroeconomic and fiscal
macroeconomic conditions plays a role,
and fiscal especially in triggering an
conditions episode of consolidation. It
 is also found to influence the
 success rate of consolidation.

Monetary A number of studies conclude
stance that the monetary policy
 stance is relevant for success
 as it may accommodate
 consolidation. This conclusion
 is not corroborated in
 general.

Exchange rate As for the monetary stance,
 the exchange rate is found to
 matter as it may accommodate
 consolidation. In particular,
 depreciations increase the
 chances of success.

Rate of GDP The findings concerning
growth economic growth are not clear
 cut. There is evidence that
 accelerating growth benefits
 the rate of success.

Political Single-party governments are
factors generally more effective than
 coalitions, while the
 political alignment hardly
 matters.

Determinants References

Composition of Alesina and Perotti (1995),
fiscal Zaghini (1999), von Hagen et al.
adjustment (2002), Briotti (2004),
 Lambertini and Tavares (2005),
 Ahrend et al. (2006),
 Alesina and Perotti (1997),
 Alesina and Ardagna (1998),
 Bertelsmann Foundation (2006),
 Guichard et al. (2007)..

Size of fiscal Ardagna (2004), von Hagen et al.
adjustment (2002), Briotti (2004),
 Lambertini and Tavares (2005),
 Alesina and Ardagna (1998),
 Zaghini (1999).

Initial Ardagna (2004), Zaghini (1999),
conditions von Hagen et al. (2002), Briotti
macroeconomic (2004), Lambertini and Tavares
and fiscal (2005), Ahrend et al. (2006),
conditions Alesina and Ardagna (1998).

Monetary Ahrend et al. (2006),
stance Bertelsmann Foundation (2006),
 von Hagen et al. (2002), Ardagna
 (2004), Lambertini and Tavares
 (2005).

Exchange rate Alesina and Perotti (1997),
 Alesina and Ardagna (1998),
 Lambertini and Tavares (2005).

Rate of GDP Alesina and Perotti (1995),
growth Bertelsmann Foundation (2006)
 and Ardagna (2004).

Political Alesina and Perotti (1995),
factors Bertelsmann Foundation (2006).

Table 2. Overview of episodes of fiscal consolidation in the EU

 Type of consolidation

 No. No.
 Cold shower of Gradual of
 years years

BE 1977, 1982, 1984, 1993, 5 1985, 1986, 1987, 1996, 6
 2006 1997, 1998
BG - -
CZ 2004 1 -
DK 1983, 1984, 1986 3 2003, 2004, 2005 3
DE 1982, 1989, 2000 3 1983, 1984, 1985, 1992, 6
EE 2000, 2003 2 1993, 1994 -
IE 1976, 1983, 1988, 2004 4 1991 , 1992, 1993, 4
EL 1974, 1982, 1986, 1987, 1994
 1991, 1994, 1996, 9 -
 2005, 2006
ES 1986,1992,1996 3 -
FR 1996 1 2004, 2005, 2006 3
IT 1976, 1982, 1983, 1991,
 1992, 1993, 1997 7 -
CY 2000, 2004, 2005 3 -
LV 1996, 2000 2 2003, 2004, 2005 3
LT 1998, 1999 2 -
LU 1983, 1985, 1993, 1997 4 1994, 1995, 1996 3
 1991-2006
HU 1999.2003 2
MT 1999, 2004, 2005 3 2000, 2001, 2002 3
NL 1985, 1991, 1993, 1996, 5 1971, 1972, 1973, 1981, 7
 2005 1982, 1983, 1984
AT 1984, 1996, 1997, 2001 4 -
PL 2005 1 -
PT 1977, 1982, 1983, 1984,
 1986, 1992, 2002, 2006 8 -
RO 1997, 1998, 1999 3 -
SI 2002 I -
SK 1998, 2001 , 2003 3 -
FI 1976,1981,1984, 1988, 7 -
 1996,1998, 2000
SE 1971 , 1976, 1983, 6 1980, 1981 , 1982,
 1987, 1995, 1996 1984, 1985, 1986, 9
 2003, 2004, 2005
UK 1974, 1980, 1982, 7 -
 1996, 1997, 1998,
 2000
Total 99 47

 Total
 No. of Sample
 years period

BE 11 1971-2006
BG 0 2003-2006
CZ 1 1998-2006
DK 6 1971-2006
DE 9 1971-2006
EE 2 1996-2006
IE 8 1971-2006
EL 9 1971-2006
ES 3 1971-2006
FR 4 1971-2006
IT 7 1971-2006
CY 3 1999-2006
LV 5 1996-2006
LT 2 1996-2006
LU 7 1983-1987
HU 2 1998-2006
MT 6 1999-2006
NL 12 1971-2006
AT 4 1971-2006
PL 1 1996-2006
PT 8 1971-2006
RO 3 1996-2006
SI 1 2001-2006
SK 3 1997-2006
FI 7 1971-2006
SE 15 1971-2006
UK 7 1971-2006

Total 146

Table 3. Probability of starting a fiscal consolidation

 Explanatory variables Estimated z statistic
 coefficient

Baseline

Initial conditions Headline deficit 0.02 3.84
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.01 2.32
 cent of GDP

Political factors Elections (t-1), 0.07 1.47
 dummy

 Size of majority in 0.12 0.45
 parliament

 Party concentration -0.39 -2.31
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules, index 0.07 2.26

 Expenditure rules, 0.03 1.32
 index

 Budgetary 0.03 2.09
 procedures, index

Structural reforms Pensions (RDBF), -0.01 -0.25
 dummy

 Employment 0.01 0.30
 protection
 legislation (RDBF),
 dummy

 Unemployment 0.09 2.96
 benefits (RDBF),
 dummy

 Labour market (IMF, 0.03 0.80
 2004), dummy

 Product market (IMF, 0.06 1.26
 2004), dummy

 Explanatory variables p-value No. of
 obs.

Baseline

Initial conditions Headline deficit 0.00 466
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.02 466
 cent of GDP

Political factors Elections (t-1), 0.14 425
 dummy

 Size of majority in 0.65 427
 parliament

 Party concentration 0.02 427
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules, index 0.02 230

 Expenditure rules, 0.18 230
 index

 Budgetary 0.04 213
 procedures, index

Structural reforms Pensions (RDBF), 0.80 216
 dummy

 Employment 0.76 216
 protection
 legislation (RDBF),
 dummy

 Unemployment 0.03 216
 benefits (RDBF),
 dummy

 Labour market (IMF, 0.42 297
 2004), dummy

 Product market (IMF, 0.20 263
 2004), dummy

Notes: See table A1 in the Appendix for detailed description of the
indicators. Estimation method: probit regression on panel data,
standard errors adjusted for intra-panel error correlation. In
addition to the baseline specification, regressions were run
sequentially by adding each individual additional variable. Estimated
coefficients represent the marginal contribution of the explanatory
variable (measured at sample mean) to the probabiliy of starting a
fiscal consolidation.

Table 4. Probability of a gradual fiscal consolidation

 Explanatory Estimated z statistic
 variables coefficient

Baseline

Initial conditions Headline deficit -0.03 -1.51
 (t-1) per cent of
 GDP

 Output gap (t-1) per -0.03 -1.13
 cent of GDP

 Consolidation in 0.43 4.37
 three preceding
 years (dummy)

Composition of
adjustment
(change, of GDP) Cyclically adjusted 0.15 3.69
 primary expenditure

Expenditure Government wage bill -0.36 -2.45
composition of
adjustment Government 0.29 2.27
 investment
 expenditure

 Government final -0.06 -0.82
 consumption
 expenditure

 Subsidies -0.3 -2.80

 Transfers other than -0.21 -1.40
 in kind

 Transfers in kind -0.25 -1.78

Political factors Elections, dummy 0.04 0.40

 Size of majority in -0.58 -1.38
 parliament

 Party concentration -0.22 -0.30
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules, index -0.02 -0.42

 Expenditure rules, 0.05 1.33
 index

 Budgetary -0.005 -0.25
 procedures, index

Expenditure Expenditure reforms, 0.067 0.76
reforms dummy

 Explanatory p-value No. of
 variables obs.

Baseline

Initial conditions Headline deficit 0.13 127
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.25 127
 cent of GDP

 Consolidation in 0.00 127
 three preceding
 years (dummy)

Composition of 127
adjustment
(change, of GDP) Cyclically adjusted 0.00 127
 primary expenditure

Expenditure Government wage bill 0.01 127
composition of
adjustment Government 0.02 127
 investment
 expenditure

 Government final 0.41 107
 consumption
 expenditure

 Subsidies 0.05 124

 Transfers other than 0.16 104
 in kind

 Transfers in kind 0.07 98

Political factors Elections, dummy 0.68 117

 Size of majority in 0.16 116
 parliament

 Party concentration 0.76 116
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules, index 0.67 63

 Expenditure rules, 0.18 63
 index

 Budgetary 0.80 53
 procedures, index

Expenditure Expenditure reforms, 0.44 105
reforms dummy

Notes: See table A1 in the Appendix for detailed description of the
indicators. Estimation method: probit regression on panel data,
standard errors adjusted for intra-panel error correlation. On top pf
the baseline specification, regressions were run sequentially by
adding individually each one of the additional variables.
Coefficients represent marginal contribution of the explanatory
variable (measured at sample mean) to the probabiliy of enacting a
gradual rather than a 'cold shower' adjustment.

Table 5. Probability of successful fiscal consolidation

 Explanatory variables Estimated z statistic
 coefficient

Baseline

Initial conditions Headline deficit 0.07 4.03
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.05 1.51
 cent of GDP

Size and Cyclically adjusted -0.08 -1.12
composition of primary balance
adjustment
(change, of GDP) Cyclically adjusted -0.12 -2.77
 primary expenditure

Expenditure Government wage bill 0.15 0.94
composition
of adjustment Government 0.38 2.01
 investment
 expenditure

 Government final -0.08 -0.84
 consumption

 Subsidies -0.02 -0.12

 Transfers other than -0.15 -1.05
 in kind

 Transfers in kind 0.09 0.56

Other fiscal Gradual -0.14 -0.63
factors consolidation, dummy

 Tax elasticities 0.01 0.54

Political factors Elections (t-1), 0.13 1.27
 dummy

 Size of majority in 0.31 0.49
 parliament

 Party concentration -1.13 -1.21
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules, index 0.51 2.92

 Expenditure rules, 0.01 0.18
 index

 Budgetary 0.27 2.55
 procedures, index

Structural reforms Pensions (RDBF), 0.25 1.36
 dummy

 Employment 0.18 1.10
 protection
 legislation (RDBF),
 dummy

 Unemployment 0.38 2.41
 benefits (RDBF),
 dummy

 Labour market (IMF, 0.57 4.00
 2004), dummy

 Product market (IMF, 0.54 3.02
 2004), dummy

 Expenditure reforms, 0.21 1.33
 dummy

 Explanatory variables p-value No. of
 obs.

Baseline

Initial conditions Headline deficit 0.00 113
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.13 113
 cent of GDP

Size and Cyclically adjusted 0.26 113
composition of primary balance
adjustment
(change, of GDP) Cyclically adjusted 0.01 113
 primary expenditure

Expenditure Government wage bill 0.34 113
composition
of adjustment Government 0.04 113
 investment
 expenditure

 Government final 0.40 113
 consumption

 Subsidies 0.90 113

 Transfers other than 0.29 94
 in kind

 Transfers in kind 0.57 88

Other fiscal Gradual 0.53 113
factors consolidation, dummy

 Tax elasticities 0.58 87

Political factors Elections (t-1), 0.20 108
 dummy

 Size of majority in 0.62 107
 parliament

 Party concentration 0.22 107
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules, index 0.00 52

 Expenditure rules, 0.80 52
 index

 Budgetary 0.01 44
 procedures, index

Structural reforms Pensions (RDBF), 0.17 54
 dummy

 Employment 0.20 54
 protection
 legislation (RDBF),
 dummy

 Unemployment 0.02 54
 benefits (RDBF),
 dummy

 Labour market (IMF, 0.00 81
 2004), dummy

 Product market (IMF, 0.00 74
 2004), dummy

 Expenditure reforms, 0.18 96
 dummy

Notes: See table A1 in the Appendix for a detailed description of the
indicators. Estimation method: probit regression on panel data,
standard errors adjusted for intra-panel error correlation. Starting
from the basesline specification additional variables were added
individually in turn. Coefficients represent the marginal
contribution of the explanatory variable (measured at sample mean) to
the probability of consolidation being successful.

Table 6. Probability of successful fiscal consolidation (from
1990 onwards)

 Explanatory Estimated z statistic
 variables coefficient

Baseline

Initial conditions Headline deficit 0.08 3.14
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.01 0.52
 cent of GDP

Size and Cyclically adjusted -0.08 -0.93
composition of primary balance
adjustment
(change, Cyclically adjusted -0.06 -1.20
of GDP) primary expenditure

Expenditure Government wage bill -0.11 -0.55
composition
of adjustment Government 0.23 1.07
 investment
 expenditure

 Government final -0.2 -1.93
 consumption

 Subsidies -0.25 -0.94

 Transfers other than -0.11 -0.77
 in kind

 Transfers in kind -0.15 -0.51

Other fiscal Gradual -0.01 -0.05
factors consolidation, dummy

 Tax elasticities 0.04 1.61

Political factors Elections (t-1), 0.16 1.14
 dummy

 Size of majority in 0.012 0.01
 parliament

 Party concentration -0.96 -0.91
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules 0.51 2.92
 (average), index

 Expenditure rules 0.01 0.18
 (average), index

 Budgetary 0.27 2.55
 procedures, index

Structural reforms Pensions (RDBF), 0.07 0.42
 dummy

 Employment 0.01 0.06
 protection
 legislation (RDBF),
 dummy

 Unemployment 0.14 0.92
 benefits (RDBF),
 dummy

 Labour market (IMF, 0.47 2.21
 2004), dummy

 Product market (IMF, 0.53 1.55
 2004), dummy

 Expenditure reforms, 0.17 0.64
 dummy

 Explanatory p-value No. of
 variables obs.

Baseline

Initial conditions Headline deficit 0.02 58
 (t-1) per cent of
 GDP

 Output gap (t-1) per 0.60 58
 cent of GDP

Size and Cyclically adjusted 0.35 58
composition of primary balance
adjustment
(change, Cyclically adjusted 0.22 58
of GDP) primary expenditure

Expenditure Government wage bill 0.57 58
composition
of adjustment Government 0.28 58
 investment
 expenditure

 Government final 0.05 57
 consumption

 Subsidies 0.34 58

 Transfers other than 0.44 57
 in kind

 Transfers in kind 0.60 53

Other fiscal Gradual 0.96 58
factors consolidation, dummy

 Tax elasticities 0.11 52

Political factors Elections (t-1), 0.25 58
 dummy

 Size of majority in 0.99 58
 parliament

 Party concentration 0.36 58
 in parliament
 (Herfindahl index)

Fiscal governance Fiscal rules 0.00 52
 (average), index

 Expenditure rules 0.80 52
 (average), index

 Budgetary 0.01 44
 procedures, index

Structural reforms Pensions (RDBF), 0.67 38
 dummy

 Employment 0.95 38
 protection
 legislation (RDBF),
 dummy

 Unemployment 0.35 38
 benefits (RDBF),
 dummy

 Labour market (IMF, 0.03 31
 2004), dummy

 Product market (IMF, 0.12 31
 2004), dummy

 Expenditure reforms, 0.52 41
 dummy

Notes: See table A1 in the Appendix for a detailed description of the
indicators. Estimation method: probit regression on panel data,
standard errors adjusted for intra-panel error correlation. Starting
from the basesline specification additional variables were added
individually in turn. Coefficients represent the marginal contribution
of the explanatory variable (measured at sample mean) to the
probabiliy of consolidation being successful.

Figure 1. Distribution of the change in the cyclically-adjusted
primary budget balance (CAPB) during consolidation
years

[DELTA]CAPB<0 2.7
0<[DELTA]CAPB<0.5 8.2
0.5<[DELTA]CAPB<1 8.2
1<[DELTA]CAPB<1.5 13.7
1.5<[DELTA]CAPB<2 26.0
2<[DELTA]CAPB<2.5 17.1
2.5<[DELTA]CAPB<3 7.5
3<[DELTA]CAPB 16.4

Note: Table made from bar graph.

Figure 2. Distribution of consolidation episodes over time

 Years of consolidation of which successful

-1974 6 1
1975-1979 8 2
1980-1984 32 6
1985-1989 12 5
1990-1994 21 13
1995-1999 34 18
2000- 33 4

Note: Table made from bar graph.