Phoenix Rising From the Ashes: New Evidence on National Fiscal Rules in the EU *

Transcription

1 Phoenix Rising From the Ashes: New Evidence on National Fiscal Rules in the EU * By U. Michael Bergman Department of Economics, University of Copenhagen Øster Farimagsgade 5, Building 26, DK-1353 Copenhagen K, Denmark Michael.Bergman@econ.ku.dk Michael M. Hutchison Department of Economics, University of California, Santa Cruz Santa Cruz, CA USA hutch@ucsc.edu Svend E. Hougaard Jensen Department of Economics, Copenhagen Business School Porcelaenshaven 16A, 2000 Frederiksberg C, Denmark shj.eco@cbs.dk September 2014 Abstract. Supranational fiscal rules in the EU have had limited effectiveness due to enforceability problems. In this paper, we use a panel data set covering the EU countries to investigate the effectiveness of national fiscal rules. We also examine whether the effectiveness of fiscal rules is affected by the level of governmental bureaucratic efficiency, and whether the enforcement of rules may change from normal times to a period of financial distress. We show that countries with tight national fiscal rules have larger cyclically adjusted primary balance surpluses compared to counties without such rules. Our estimates suggest that fiscal rules applied at the national level have improved fiscal outcomes in EU countries both during normal periods and periods of financial distress, even when controlling for a broad spectrum of economic and political factors. Finally, we find that the effects of fiscal rules are reinforced in countries with high quality of government. Keywords: National vs. supranational fiscal rules; government efficiency; euro area; sovereign debt. JEL codes: E62, F36, F44, G14, H63 An earlier version of this paper was presented at the 16th annual conference of the Swedish Network for European Studies in Economics and Business (SNEE) on European Integration, held at Mölle, Sweden, May We are grateful for helpful comments from our discussant, Iain Begg, as well as from other conference participants. 1

2 1. INTRODUCTION In the aftermath of the financial crisis, budget deficits and debt ratios have increased to levels that would have been unthinkable just a few years ago. Market interest rates and sovereign credit default swaps (CDS) spreads have increased sharply in many countries, signalling lack of confidence in fiscal (and other) policies and forcing the political system to respond with programmes designed to restore long-term sustainable public finances and to a renewed emphasis on designing and implementing new fiscal rules and institutions such as fiscal councils. The architects behind EMU recognized the need to complement market pressure with explicit fiscal rules designed to ensure sound public finances. However, the track record of the euro area s fiscal rule, the Stability and Growth Pact (SGP), is rather poor. While remaining operative on paper - complete with its excessive deficit procedure, sanctions for violators and other enforcement procedures - the SGP has had clear shortcomings in practice (Schuknecht, Moutot, Rother and Stark, 2011). However, the poor track record of fiscal rules in Europe refers to rules implemented at the supranational level, not at the national level. While supranational fiscal rules have largely failed, and thereby may have given fiscal rules a bad name, a number of European countries have found national fiscal rules to be helpful in achieving greater budgetary discipline. For example, Sweden, Finland and the Netherlands all realized improvements in their fiscal situations after adopting rules that limit spending (Ayuso-i-Casals, 2012). More generally, a growing literature focusing on the effects of national fiscal rules and institutions suggest that stronger fiscal rules limit the deficit bias. 1 Against that background, the leading question addressed in this paper is whether fiscal rules in fact may have a much better track record than we think. Using a sample of 27 EU countries, the focus of the paper is an empirical assessment of fiscal rules implemented at the national level in the EU over An important aspect of our analysis is to determine what extent the effectiveness of fiscal rules in generating sound public finances is conditional upon the effectiveness of government administration. Our objective is to measure whether fiscal rules are significant determinants of public policy once controlling for institutional quality, and whether the effectiveness of fiscal rules is in fact enhanced when employed in environments associated with strong institutions. To this end, we develop a unique fiscal 1 See, for example, Debrun, Moulin, Turrini, Ayuso-i-Casals and Kumar (2008), Afonso and Hauptmeier (2009), Dahan and Strawczynski (2010), Holm-Hadulla, Hauptmeier and Rother (2012) and Nerlich and Reuter (2013). 2

3 rules index, varying across countries and over time, based on 28 specific characteristics of rules using the IMF FAD data base (Schaechter, Kinda, Budina, and Weber, 2012) as our underlying data source on legislative and practical constraints on government fiscal policy. As an alternative measure to test the robustness of our index we use the EU Commission numerical fiscal rule index. We also study which types of fiscal rules appear to be effective (balance budget, expenditure, revenue or debt limits) and whether fiscal rules are effective during financial crises. Our focus here is on (1) whether the effectiveness of fiscal rules in reducing deficit bias is dependent on the specific type of rule implement and (2) whether the effectiveness of rules is related to the state of the economy ( normal times vs. financial crisis) and how well they are enforced, with enforcement being measured by the efficiency of government bureaucracy. 2 Pursuing this avenue of research, we employ the World Bank efficiency of government bureaucracy index, which is part of the World Bank Worldwide Governance Indicators, 2013 Update (WGI) project research dataset. This indicator measures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies. The road map of the paper is as follows. Section 2 briefly discusses the rationale for introducing fiscal rules and it reviews the track record of fiscal rules operating at the supranational level in the EU. Section 3 provides description how our new fiscal rule index is constructed and how it is related to the EU Commission index. This section also introduces our measure of government efficiency. This is followed, in Section 4, by an econometric assessment of the link between national fiscal rules, government efficiency and the actual conduct of fiscal policy during normal times and during the financial crisis. Finally, Section 5 concludes. 2 There is a related literature focusing on the cyclicality of fiscal policy and government efficiency. In a recent paper, Calderón, Duncan and Schmidt-Hebbel (2012) consider the role of government administrative quality in shaping fiscal outcomes. Using a world sample of 115 advanced, emerging and developing countries for , they find that the level of institutional quality plays a key role in countries' ability to implement counter-cyclical macroeconomic policies. The results show that countries with strong (weak) institutions adopt counter- (pro-) cyclical macroeconomic policies, reflected in extended monetary policy and fiscal policy rules. The threshold level of institutional quality at which monetary and fiscal policies are a-cyclical is found to be similar. Bergman and Hutchison (2014) study whether government efficiency interacts with the strength of fiscal rules in 81 countries and find that the pro-cyclicality of fiscal policy is reduced in countries with high government efficiency that also adopt stronger fiscal frameworks. 3

4 2. FISCAL RULES: SCOPE AND EXPERIENCES Fiscal rules are generally legislative agreements intended to promote fiscal discipline by tying the hands of policy makers in order to constrain decisions about spending and revenue programs (Bernanke, 2010). For example, fiscal rules may impose constraints on total government expenditures, deficits or debt. Whatever target is chosen, fiscal rules are best used to address longer-term budget problems. A fiscal rule clearly does not guarantee improved budget outcomes-- any rule imposed by a legislature can be circumvented by the same legislature. According to IMF (2009), fiscal rules have been introduced in about 80 countries, and while the experiences are mixed, the message from the empirical literature is relatively clear, namely that well-designed rules can help promoting a better fiscal performance. In the EU, fiscal rules have been in place for several years. The Maastricht Treaty imposed numerical limits on the fiscal debt and deficits of those joining the euro area (EA). Those limits became permanent with the SGP in 1997, prior to the creation of the euro, in order to ensure that fiscal discipline is maintained in the member states. On balance, however, the SGP was not effective (see Hughes Hallett and Jensen, 2012). The SGP was followed, in March 2012 by the Treaty on Stability, Coordination and Governance in the Economic and Monetary Union (TSCG). The most important component of the TSCG is the fiscal compact, which states that the budgetary position of the general government shall be balanced or in surplus. This rule is not new, and is part of the SGP, but the difference is that the TSCG requires member states to adopt the medium-term objective in national law. Thus, a major development of the Fiscal Compact is the expectation that the rules will be introduced into national legislation. Many countries have implemented national fiscal rules and institutions in order to obtain long-run sustainable public finances. Schaechter et al. (2012) have constructed a database containing accounts of both supranational and national fiscal rules in a large number of countries and provide three main findings: (i) the number of countries adopting fiscal rules is increasing and, in countries already having fiscal rules, these are strengthened; (ii) emerging economies have implemented fiscal rules rapidly, and have caught up with advanced countries; and (iii) fiscal rules have generally become more complex by incorporating formulations of explicit objectives or targets, as well as more regulations on implementation, communication and monitoring. In contrast to the poor performance of supranational fiscal rules in the EU, national fiscal rules may be associated with stronger fiscal performance. There is also empirical evidence 4

5 suggesting that countries with more effective institutions also have lower volatility of government spending, see Albuquerque (2011). Schaechter et al. (2012) find in their comprehensive study of both advanced and emerging countries that tighter and more encompassing fiscal rules are correlated with stronger cyclically adjusted primary balances in EU countries; budget balance and debt rules have contributed to better budgetary outcomes than expenditure and revenue rules; and rules covering wider levels of government have been associated with more fiscal discipline. Some features of the rules such as a strong legal basis and strict enforcement appear to have a beneficial impact on fiscal performance. These results suggest that fiscal rules may be more efficient in countries with more efficient government and legal systems. This is the focus of our empirical study in the remainder of the paper. 3. MEASURING THE STRENGTH OF FISCAL RULES 3.1 DATA SOURCES We use two sources for the fiscal rules: (A) the IMF Fiscal Rules Dataset, as described in Schaechter et al. (2012) and (B) the EU Commission numerical fiscal rule index. 3 The IMF database covers 81 countries for the sample 1985 to 2012 and includes detailed information on four types of rules: budget balance rules, debt rules, expenditure rules, and revenue rules. The IMF database does not include rules implemented at the subnational level, only at the level of the central or general government. The dataset includes descriptions of the rules as well as information about the type of rule, year of implementation, number of rules, legal basis, coverage, monitoring, enforcement, institutional supporting features such as for example multi-year expenditure ceilings, and stabilization features such as budget balance rules accounting for the state of the economy. The underlying data is collected from many different sources; fiscal framework legislations, published and unpublished country documents, IMF staff reports and other IMF papers, information provided by national authorities, the EU Commission etc. In the accompanying working paper Schaechter et al. (2012) show how an overall index reflecting the strength of fiscal rules in a country can be constructed. We follow their method and construct indices reflecting five main characteristics of the fiscal rules, viz. monitoring, enforcement, coverage, the legal basis and escape clauses. Even though it is difficult to know whether monitoring is effective and to what extent escape clauses are used, as also argued by 3 More detailed information on how the EU Commission fiscal rule index is constructed can be found on the EU Commission website: 5

6 Schaechter et al. (2012), we include this measure as well when constructing our overall index. In addition to information on these four types of rules, each described by five key characteristics, we also use information concerning supporting procedures and institutions (multi-year expenditure ceilings implemented at the aggregate level, by ministry of by line item), whether there is an independent body setting budget assumptions, information about transparency and accountability, whether a balanced budget target is defined and whether there are rules excluding public investments or other priority items from the ceiling. This adds eight institutional regulations to the four fiscal rules. In total we then have 28 different characteristics describing national fiscal rules in each country. 4 In order to construct an overall index describing the fiscal framework in each country we simply add all 28 characteristics and renormalize the resulting index to be in the range between 0 and 4. We use equal weights even though it could be argued that some rules or features may be more important than others. However, there is no clear-cut empirical evidence to rely on when defining the weights and therefore we assume equal weights of all features of the fiscal regulation. The resulting index is denoted NFRI. The IMF database also includes information about supranational fiscal rules. Such rules usually apply to members of monetary unions. The Maastricht Treaty, signed in February 1992, defines the supranational fiscal rules applied to EU countries. Both budget deficit and debt rules were introduced in the Treaty. In 1997, these rules where strengthened by the Amsterdam Treaty signed in 1997 where the Stability and Growth Pact (SGP) was approved. This pact was then reformed in The recent changes in the fiscal framework, i.e., the fiscal compact, six-pack and other agreements are all approved or ratified outside our sample which ends in It should be noted that these supranational rules apply to a country when it becomes member of the EU. The IMF database allows us to construct an index of supranational fiscal rules. This is an alternative to using simple dummy variables. We construct an index capturing the strength of the supranational fiscal rules by taking the sum of sub-indices measuring monitoring, enforcement coverage, legal basis and whether there are well-defined escape clauses for each type of fiscal rules (budget balance rules, debt rules, expenditure rules, and revenue rules), a total of 22 characteristics using equal weights and rescaling the total index to lie between 0 and 4. 4 Many countries have national balanced budget rules and debt rules that also are supranational rules. In some countries, such as Austria and Spain, supranational rules in the form of balanced budget and debt rules were introduced first and then, at a later stage, these rules were implemented at the national level as well. When constructing our national fiscal rule strength indices, we take this into account. 6

7 When looking more closely at the constructed measure we find that it is almost constant over the period 1995 until 2005 and then somewhat lower reflecting the revision of the SGP in There are some minor differences between countries but in general a dummy variable captures the main changes in the SGP. Even though a dummy variable would capture the existence of supranational fiscal rules we use our constructed index in our empirical application. As for the national index we renormalize such that it will be in the range between 0 and 4. The second database on fiscal rules used in this paper is the EU Commission numerical fiscal rule index. The EU Commission constructs a composite index of the strength of the national fiscal framework in all EU countries covering the period The EU Commission data is somewhat broader in scope than the IMF data since it also includes information on general government, central government and local government frameworks. While the IMF database focuses on central or general government rules, the EU Commission data also includes rules applied at the subnational level. Another difference is that while the EU Commission constructs and publishes a single overall measure of the strength of fiscal rules in each country, the IMF database only reports information on each type of rule. Furthermore, the IMF database also includes detailed descriptions of supranational rules, thereby allowing us to distinguish between rules designed and implemented on a national level and rules implemented on a supranational level. The EU Commission has compiled a data set of national fiscal rules in force in all current member countries. The data was first collected in a survey conducted by the Commission services in 2006 and the database has since then been updated annually. After the latest update, the database covers the period 1990 to Each member country provides information on four types of numerical fiscal rules (budget balance rules, debt rules, expenditure rules and revenue rules) in place at all levels of government (central, regional, local and social security). The provided data from each member country also include the description and definition of the fiscal rule and its coverage, its statutory base, monitoring and enforcement mechanisms, as well as experience with the extent to which the rule has been respected. Using this information, DG ECFIN has constructed an index representing the strength of these fiscal rules where information on the statutory base of the rule, room for setting or revising its objectives, the body in charge of monitoring respect and enforcement of the rule, the enforcement mechanisms relating to the rule, and the media visibility of the rule has been 7

8 used. The resulting Fiscal Rule Strength Index is a measure of the strength of each fiscal rule for each level of government. With this underlying data, the EU Commission constructs an overall time-varying index summarizing the strength of fiscal rules implemented in each member country. To construct this index, the different fiscal rule strength indices are weighted by the coverage of general government finances of the respective rule and if there is more than one rule covering the same government sub-sector, the second, third and fourth rules obtain decreasing weights reflecting decreasing marginal benefits of multiple rules applied to the same government subsector. The resulting overall index reflecting the strength of fiscal rules in force in each country is named Fiscal Rule Index (FRI). The IMF database and its information about supranational fiscal rules has, to the best of our knowledge, not been used in empirical studies of the effects of rules on public finances, apart from the analysis in Schaechter et al. (2012). The FRI index has previously been used in several publications by the EU Commission, for example in Public Finances in EMU Debrun et al. (2008) used this index to analyze the effects of fiscal rules on public finances using the sample Nerlich and Reuter (2012) also use the same data but instead of using the FRI index they define an aggregate measure of fiscal rules using dummy variables, the dummy is equal to one if there is a numerical fiscal rule in place, zero otherwise. In Table 1 we report some descriptive statistics of our two national fiscal rules indicators, as well as the number of fiscal rules implemented in all countries in our sample. 5 Table 1 reveals that the variation among national rules is substantial, while the variation in supranational rules depends whether the EU country is a member of the EMU (where rules apply) or not. Looking at the developments over time (not shown in Table 1) there were no supranational rules in place in 1990, and only a few EU countries had any fiscal rule. From 1995, the average number of national fiscal rules has remained fairly constant, around one national fiscal rule per EU member country. At the same time two supranational rules (the balanced budget rule and debt rule stipulated by the SGP) have been introduced and applied to all EU member countries. For non-eu countries, the trend has been to implement new national fiscal rules. In fact, only balanced budget and debt rules have been implemented. This is in contrast to EU member countries where both expenditure and revenue rules have been in force, expenditure rules more frequently than revenue rules. 5 Table A2 in Appendix A2 provides further descriptive statistics on the number of fiscal rules and the fiscal strength in groups of European countries. 8

9 - Table 1 about here COMPARISON OF THE TWO DATA SOURCES Even though the two indices measure the strength of fiscal rules in the EU countries, they differ in a distinct way. The EU Commission index (FRI) is broader in scope than the IMF index since it covers rules implemented at the sub-national level and that unequal weights are used when constructing the overall index. The IMF based index is constructed treating all features of the rules as equally important. Some examples serve to highlight differences between the two data sources. First, the Czech Republic has a medium-term expenditure framework but according to the IMF database, there have been numerous changes in the framework and therefore it is not regarded as a fiscal rule. There is no other national fiscal rule in place and hence there are no changes in the IMF fiscal rule indicator for this country. However, the EU Commission regards the same regulation as a fiscal rule. Also, according to the EU Commission, there are both a debt ceiling and an indicator of total liquidity in place that also constitute a fiscal rule. Second, while Portugal since 2002 has had a balanced budget rule, the rule only covers 13 percent of the general government finances. The IMF does not therefore consider this a fiscal rule. The EU Commission, on the other hand, views this as a fiscal rule implying that their measure of the fiscal rule index increased in There is a new budgetary framework approved in 2011 but this will not be implemented until The EU Commission also identifies two additional rules for Portugal, a debt ceiling and a rule concerning structural balance as a percentage of GDP. Third, according to the IMF, Slovakia did not have a fiscal rule in place before 2012 (a debt rule was implemented in March 2012). By contrast, the EU Commission views Slovakia as having a fiscal framework, a debt ceiling and a nominal expenditure ceiling earlier. Fourth, in Ireland there is no fiscal framework according to IMF whereas EU Commission lists three regulations, allocation of expenditure, a nominal expenditure ceiling and budget balance in nominal terms. Fifth, for Italy there are even larger differences between the two data sources. According to IMF there is no fiscal regulation at all in Italy. The EU Commission agrees but since they also include sub-governmental rules, they regard for example nominal expenditure ceilings at the regional and local levels of government as fiscal rules. Finally, both IMF and the EU Commission find that there is no fiscal framework or fiscal rules in Cyprus, Greece, Latvia and Malta. 9

10 Even though the two indices to some extent measure different aspects of the national fiscal rule strength they are highly correlated. Indeed, the correlation coefficient among the aggregate indices is 0.80 over the sample period. At the individual country level, the correlations between FRI and NFRI ranges between 0.22 (for Slovakia) and 0.99 (for Austria). For most countries the correlation coefficient is very high (above 0.9), suggesting that the EU Commission and the IMF data to a large extent provide similar measurements of the strength of fiscal rules regardless of the different treatments of local and regional subgovernment regulations. For countries with low correlation either the EU Commission or the IMF data finds that fiscal rule strength is constant or that there are no fiscal rules in a certain country. Since the EU Commission also includes rules enforced on sub-governments there could be cases when the IMF database suggests the absence of a rule whereas a regulation is interpreted as a fiscal rule by the EU Commission. For a broader perspective on the two measures of fiscal rules in Europe, Figure 1 shows the aggregated IMF fiscal rule strength index (upper graph) and the EU Commission numerical fiscal rule index (lower graph) in 2000 and in For comparison, we have rescaled both measures to be [0,1] variables using their minimum and maximum values, respectively. Both measures provide similar conclusions. - Figure 1 about here - Another perspective on the differences between EA and non-ea countries is the development of the fiscal strength measures over time. In Figure 2 we show the simple averages of NFRI and FRI for EU27, the initial 12 EA countries (EU12), the 17 EA countries (EU17) and the 10 non-ea member states (non-emu). 6 Apparently, there is a common pattern suggesting that all countries make efforts to strengthen their fiscal framework. The average fiscal strength among the non-emu countries exceeds the average of EA countries from 2004 when using the IMF based measure and from 1996 when using the EU Commission data. Countries remaining outside the monetary union seem to put more emphasis on designing and implementing stronger fiscal rules than countries that also have adopted the euro. This may be surprising given the focus of the EU on fiscal rules. Note also that the supranational fiscal rules in the EU apply to all member states so it is not the case that EA countries have replaced national fiscal rules with supranational rules. - Figure 2 about here - 6 Latvia joined the monetary union in 2014 and is considered to be among the 10 non EA states in our sample. 10

11 Judging from this graph, the average strength of the fiscal framework in EA countries lag behind the non-ea countries and even lag behind the EU average. In Figure 3 we show how both our measures of fiscal strength have changed over time for a selection of countries. In the upper graph we show (left) the IMF based measure NFRI for the so-called GIIPS countries (Greece, Italy, Ireland, Portugal and Spain) and (right) for a selection of countries with tight fiscal rules. A similar presentation is made in the lower graph using the EU Commission-based measure FRI. What stands out clearly is that the fiscal framework in Spain has been strengthened considerably from the early 1990s (when Spain had relatively loose fiscal rules compared to other GIIPS countries) to among the tightest fiscal rules in Europe. For the other GIIPS countries only minor changes have occurred in their fiscal frameworks over time. By contrast, countries with strong fiscal rules tend to increases the tightness of their rules since Figure 3 about here IS FISCAL STRENGTH IMPROVED BY QUALITY OF GOVERNMENT? We next turn to the question of whether government efficiency enhances the effects of fiscal rules in preventing unsustainable developments of public finances. The government efficiency index used in what follows is taken from the World Bank s Worldwide Governance Indicators (WGI), 2013 Update (see This dataset consists of data on the quality of governance provided by a large number of enterprise, citizen and expert survey respondents in industrial and developing countries. These data are gathered from a number of survey institutes, think tanks, non-governmental organizations, international organizations, and private sector firms. The WGI consists of aggregate indicators of six broad dimensions of governance: (i) Voice and Accountability, (ii) Political Stability and Absence of Violence/Terrorism, (iii) Government Effectiveness, (iv) Regulatory Quality, (v) Rule of Law, and (vi) Control of Corruption. 7 For our research, we employ the Government Efficiency indicator that reflects perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies. The government efficiency indicator ranges from -2.5 to 2.5, with higher values indicating greater government efficiency. The data on government efficiency is biannual from 1996 until 2002 and then 7 Charron, Lapuente and Rothstein (2010) find in their comparison of alternative measures of quality of government that the World Bank data are both empirically and conceptually superior and provide the best measurement for reliable and meaningful comparisons of quality of government in the EU. 11

12 annual. We use linear interpolation to add observations in 1997, 1999 and 2001 and in order to extend the data back to 1990 we make use of available data on an alternative measure of government efficiency, the International Country Risk Guide (ICRG), see Appendix A1 for more details on the interpolation. In the next section we analyse whether the fiscal strength as measured by the indices discussed above affect public finances and the role played by government efficiency. 4. DO NATIONAL FISCAL RULES IMPROVE FISCAL OUTCOMES? AN ECONOMETRIC INVESTIGATION 4.1 A FIRST LOOK AT THE DATA In this subsection we present empirical evidence on the link between fiscal rules and fiscal outcomes, by using the numerical fiscal rule indices described above. Initially, we focus on national fiscal rules and later we add supranational rules. Both indices cover national fiscal rules for the period and should be interpreted as measures of the quality of the institutional design, and not as a measure of the effectiveness of the rules. 8 We first consider cross-sectional linkages between quality of government, strength of fiscal rules and public finances. We illustrate the relationships between qualities of government, how quality of government interacts with fiscal rules and public finance outcomes (both cyclically adjusted primary balance and the change in government debt). Regression lines are included in all graphs to illustrate a linear relation. 9 In Figure 4 we show the relationships between our two measures of fiscal rule strength (NFRI and FRI) and two measures of fiscal positions: average cyclically adjusted primary balance (first column) and the change in government debt as a ratio of GDP between 1984 and 2012 (second column). As can be seen in the graphs, there is a positive correlation between quality of government and the cyclically adjusted primary balance surplus: a higher degree of quality of government is associated with larger primary balance surpluses. Turning to the change in government debt, there is a clear negative relation: larger falls in the debt ratio is associated with stronger fiscal rules. These results (although based on cross-sections) are consistent with the earlier literature suggesting that countries having stronger fiscal rules also tend to have larger primary balance surpluses and smaller increases in debt ratios. 8 Since the EU Commission data only covers the period we simply assume that the fiscal framework has not changed in This assumption is consistent with the IMF data we use which suggest no change in fiscal rules in The slopes of these regression lines are always significant at least at the 5% level. These results are not shown for brevity but are available from the authors upon request. 12

13 - Figure 4 about here - Is there a difference in these relationships between fiscal rule strength and public finances depending on government efficiency? To answer this question we run regressions using the cross section data where we interact fiscal rule strength with government efficiency and then we use these estimates to compute the predicted value of primary balance conditional on high, good and moderate government efficiency. Specifically, the cross-section regression is: PBCA GE FRI GE FRI, where i denotes country, PBCA is the cyclically i 1 i 2 i 3 i* i i adjusted primary balance, GE is government efficiency and FRI is either the IMF based measure of the EU Commission measure. Using the estimates, we then assume fixed values for GE to compute PBCA for a range of values for FRI (from its min to its max value in the sample). We use the mean values of our measure of government efficiency for each country and then we decompose the countries into three groups of equal size (quintiles) and then we compute the predicted value of primary balance conditional on the average value of government efficiency for each of the three groups. The three graphs in the first row of Figure 5 show the implied relationship between the IMF based measure of fiscal rule strength and primary balance whereas the three graphs in the second row illustrate the relationship between primary balance and the EU Commission measure. The results are consistent with the relations found in Figure 4: an increase in the strength of the fiscal rule is associated with larger primary balance surpluses. However, there is a distinct difference across the groups of countries when taking parameter estimation uncertainty into account. There is a significant effect from fiscal rules in countries with more efficient governments, but no individual effect of government efficiency on the primary balance in countries with high government efficiency but no fiscal rules. Higher government efficiency reinforces the effects of fiscal rules on primary balance surpluses, i.e., the effect of fiscal rules on primary balance are increasing in government efficiency. For countries with good or moderate government efficiency, there is no statistically significant effect of fiscal rules on primary balance, seen in the last two columns of Figure 5. Designing and imposing stronger fiscal frameworks in these countries does not appear to be effective according to our cross-section estimates. Instead, it is more warranted to increase government efficiency in these countries because our estimates suggest that the effects of existing frameworks will increase. - Figure 5 about here - The graphs above do suggest that countries having less strict fiscal rules also have larger primary balance deficits and higher debt ratios. There also seems to be an additional effect 13

14 when combining fiscal rule strength and quality of government. High quality of government tends to reinforce the effect of fiscal rules leading to a better performance of public finances. The above analysis only considers the cross-sectional linkages and we have not controlled for other factors that may be of importance for the soundness of public finances. In the next section we will estimate panel data regressions adding a number of control variables to study the role played by fiscal rules and quality of government. 4.2 NATIONAL FISCAL RULES: BASELINE MODEL To analyse whether there is a relation between fiscal rules, government efficiency and fiscal performance we follow the standard practice in the literature, by estimating the following dynamic panel regression 10 FP FP FR GE GE FR X (1) ' it, 0 it, 1 2 it, 3 it, 4 it, it, it, it. where is the fiscal policy variable (the cyclically adjusted primary balance, primary expenditures and primary revenues); is the numerical national fiscal rule index; is the government efficiency index; and, is a vector of control variables. Our primary interest in this empirical investigation is whether stricter fiscal rules, combined with higher levels of government efficiency, lead to better fiscal performance as measured by some combination of higher primary balances, lower primary expenditures and higher primary revenues. However, it is important to control for other factors that might also lead to differences in fiscal performance across countries and across time. As control variables, downloaded from AMECO, we follow the extant literature and employ the lagged output gap, the lagged debt level, the dependency ratio, the degree of openness, the natural logarithm of population, and the rate of consumer price inflation. In line with the literature, we also include a number of political variables to control for differences in preferences across countries to fiscal institutions. Here we use the fragmentation of government measured as the sum of the squared seat shares of all parties in the government, number of years left in current term, plurality dummy variable which is equal to 1 if the country has a plurality system, and an election year dummy variable which is equal to 1 if parliamentary election took place. These variables are taken from the Database of Political Institutions at the World Bank. Equation (1) with these control variables, and excluding the terms FR and GE terms, is often referred to as the deficit bias regression, i.e., the response function of policy makers to economic as well as political variables. 10 See, for example, Debrun et al. (2008), Nerlich and Reuter (2012) and De Haan, Jong-A-Pin and Mierau (2013). 14

15 Following Hallerberg and von Hagen (1999) we add dummy variables representing the budget process. They show that European governments have developed different types of budget processes in order to promote fiscal discipline. Under delegation the minister of finance has been delegated agenda-setting powers in the preparation of the budget whereas under commitment, the budget process hinges on jointly negotiated and pre-established fiscal targets. 11 We use the classifications in Hallerberg, Strauch and von Hagen (2007) for EU15, the classification for Central and Eastern European countries suggested by Yläoutinen (2004) and the classification of Gregor (2004) for Malta and Cyprus. In addition, following Debrun et al. (2008), we introduce a dummy variable representing the run-up to EMU (RUNUP) equal to 1 for EU-15 countries and years between years 1994 and 1998 and a dummy variable representing enlargement (Enlargement) which is equal to 1 for EU-10 countries after year Descriptive statistics of all control variables are provided in Table A3 in Appendix A2. In order to remove fixed effects it is common to use the Arellano-Bond difference transform (Arellano and Bond, 1991). However, our panel is unbalanced taking first difference implies loss of observations, so we instead employ the orthogonal deviation method, as suggested by Arellano and Bover (1995). This method allows us to limit the number of unused observations in the estimations. Rather than subtracting the previous observation from the current one, this method subtracts the average of all future available observations of a variable and, as a consequence, only the last observation for each individual is missing which minimizes data loss. We also report tests of autocorrelation of both first and second order, Sargan test statistic and Hansen J test statistic for overidentifying restrictions. A potential problem when implementing GMM methods is that the number of instruments is quadratic in T. This is also a potential problem in our panel. Roodman (2009b) discusses many of the potential pitfalls of instrument proliferation and its consequences, including over fitting of endogenous variables, bias in estimates and the weakening of Sargan tests. These issues have not been fully analyzed in the literature and there exists very little guidance on how to handle this problem in GMM estimation of dynamic panel data models, see the discussions in Hall and Peixe (2003), Roodman (2009b) and Bontempi and Mammi (2012). Roodman (2009b) suggests either that the number of instruments is limited to certain lags or a method of collapsing the instruments by having separate moments for each lag instead of 11 Hallerberg (2004) suggests two additional types of budget processes, hybrid for countries with minority governments and fiefdom for countries with unstable party systems. We will not include these types of budget processes in our regressions. 15

16 for each lag and time period. We will use the latter approach in our empirical application. 12 An additional issue is the potential endogeneity issues. The Arellano-Bover GMM estimator allows us to handle endogeneity using internal instruments. We will assume that our dependent variable, cyclically adjusted primary balance is endogenous and include this variable as a GMM-style instrument. All other explanatory variables are assumed to be exogenous and are therefore included as iv-style instruments. Table 2 reports the results of panel data estimations including various combinations of the national fiscal rule indicator, government efficiency and interaction terms, together with a common set of control variables typically employed in the literature. Consider first the deficit bias regression (economic and political control variables alone) results reported in the first column of Table 2. Parameters have the expected signs and the majority significant at the 10 percent level. Focusing on the significant parameters we find that cyclically adjusted primary balance is strongly positively autocorrelated, a finding consistent with earlier evidence. Lagged output gap is negatively related to primary balance implying that an increase in the output gap is associated with falling primary balance surpluses. Government debt has a positive significant effect; larger government debt leads to larger primary balance surpluses. The dependency ratio is negatively related to primary balance surpluses such that countries with higher dependency ratio also tend to increase the primary balance. The effect of openness is positive and the effect of population (used as a proxy for the size of the economy) is negative. More open economies tend to have larger primary balance surpluses and larger countries smaller surpluses. Increasing inflation tends to increase primary balance surpluses. Only two of the political variables are significant. Years in office have a positive effect, newly appointed governments tend to produce larger surpluses and as election is closing in, surpluses are reduced. This is consistent with a political business cycle view where governments tend to expand the budget close to elections. The commitment dummy variable is positive implying that this type of budget process leads to higher primary balance surpluses than countries adopting other budget processes. - Table 2 about here - Our focus is on columns (2)-(8) of Table 2. Here we find that both fiscal rule indices in columns (2) and (3) are statistically significant and the parameters are positive indicating that a country with stricter national fiscal rules also is likely to have larger cyclically adjusted primary surpluses. A one unit in the fiscal rule index increases the structural budget position 12 The Stata command xtbond2 written by Roodman (2009a) implements both these methods. 16

17 by percentage points. What is surprising, given the earlier literature, is that we do not find evidence that government efficiency alone improves public finances (column 4), nor does it appear to have a significant impact when interacted with fiscal rules (columns 5 and 6). Government efficiency does not seem to affect primary balances, either directly or as an interaction term with the fiscal rule strength. This is different from what we found in the cross-sectional analysis in the previous subsection. This may be attributable to the inclusion of a broad set of control variables in Table 2, unlike our cross-section regressions reported in Figure 4. Economic and political factors therefore appear to affect the behaviour of public finances, and capture the effects correlated with the quality of the government bureaucracy. (In addition, the panel regressions allow time variation while the cross-section regressions employ sample averages). The empirical evidence reported in Table 2 does not suggest a strong effect of government efficiency on fiscal performance, either as a direct effect or as an factor interacting with the strength of fiscal rules. However, an alternative explanation for this finding may be that the measure that we are using -- the World Bank government efficiency measure -- is not refined enough to distinguish nuances in government efficiency across countries and time. In particular, it may be that the World Bank measure does not reliably can distinguish between the quality of government in similar countries, e.g. among the Nordic countries or among the former Eastern European countries. On the other hand, the World Bank data allow us to compare the relative standing of a group of countries against other countries. Shedding light on this issue, Charron et al. (2010) perform a detailed analysis of the World Bank data incorporating not only the particular measure we use but also all other aspects of quality of government measured by the World Bank. Using a number of different methods they combine five World Bank indices measuring different aspects of efficiency and quality of government in the EU countries they find that the EU countries can be decomposed into three groups depending on their relative quality of government (QoG). They also perform sensitivity analysis showing that the clustering is robust to a number of alterations of the method used to compute the overall measure of quality of government and how these changes affect the clustering. Their conclusion is that the clustering is robust. We will follow their suggestion and use their clusters in our regressions instead of using the measure of government efficiency. Table 3 shows the three clusters. - Table 3 about here - It is also interesting to compare the means of the two aggregate measures of the strength of fiscal rules across clusters. As expected we find that the fiscal rule strength is higher in 17

18 cluster 1 but there seems to be only small differences between clusters 2 and The strength of supranational rules, which goes hand in hand with EU membership, declines with QoG. The same holds for individual fiscal rules except for debt rules where countries in cluster 3 have stronger debt rules than countries in cluster 2. Interaction terms also confirm this pattern, as the interaction terms are consistently higher the higher is the quality of government. Table 4 uses these clusters, representing different levels of the quality of government instead of the World Bank indicator of government efficiency, in regressions where we replace government efficiency with the interaction of national fiscal rules and cluster dummy variables. Interaction terms are explicitly included for good (cluster 2; QoG2) and moderate (cluster 3; QoG3) government quality countries; high (cluster 1; QoG1) quality government is the baseline result interpreted from the coefficient associated with the fiscal rule (FR) variable. As usual, clustered and robust standard errors are reported within parenthesis below the point estimates. The total effect of fiscal rules for each cluster is shown within brackets below the standard errors. (Asterisks denote significance levels; see footnotes). All control variables used in our previous regressions reported in Table 2 are included in these regressions but not reported for brevity. There are only marginal changes in the point estimates of the parameters associated with these control variables and very minor changes in their significance (detail results available upon request). The results reported in Table 4 indicate that fiscal rules, when combined with a high quality of government (QoG1), are again positive and significant for both the IMF and EU Commission indices (point estimate range ). However, the total effects of fiscal - Table 4 about here - rules (terms in brackets) are not statistically significant in cases with either moderate (1.03) or good quality (-0.21) of institutions. Indeed, the marginal effect of the interaction terms are negative and statistically significant for good quality governments (-0.42 and -0.95), but the estimated total effects of fiscal rules are also insignificant in this case, i.e. total effects estimated are (0.73 plus -0.95) and 0.11 (0.53 plus -0.42). This empirical evidence suggests that, for fiscal rules to be effective, a high level of government efficiency is a necessary precondition. 13 Table A4 in Appendix A2 shows the average fiscal strength in the three clusters of countries. 18

19 4.3 WHICH FISCAL RULES ARE MOST EFFECTIVE? Another question of interest in our study is which specific fiscal rules are most effective in promoting fiscal balance balanced budget rules, expenditure rules, revenue rules or debt rules? In the analysis above we used a general index of the strength of fiscal rules. The effects we have measured can be viewed as an average effect of the different rules implemented by the European countries, but the analysis does not reveal whether some rules are more or less effective than other types of rules. In this section we therefore decompose the IMF based measure into indices measuring each type of fiscal rule, a balanced budget rule, expenditure rule, revenue rule and debt rule. Using the database we construct an index of the strength of each of each of these four types of rules using a procedure similar to the one we used when constructing the overall index above. For each type of rule we add using equal weights the characteristics of the specific rule into an index and then we normalize such that the index is in the range of 0 to 4. In Table 5 we report estimates of our basic model specification with these four specific types of rules (rather than the aggregate rule indices). The top panel of Table 5 reports the rules separately (columns 1-4) and together (column 5), with all of the regressions including the standard control variables. Panel B reports the regressions using specific rules combined with the quality of government interaction terms. - Table 5 about here - Panel A suggests that the type of rule matters: balanced budget rules and debt rules appear to be effective in limiting the rise in the fiscal deficit, with both coefficients significant at the 1% level of confidence, while expenditure rules and revenue rules have no discernible effect. Panel B reports similar results as Panel A, results consistent with our earlier findings fiscal rules are effective in countries with the highest quality governments, at least fiscal rules based on balanced budgets (column 1) or debt limits (column 4). In addition, we find that balanced budget rules are effective for countries with moderate (but not good) quality governments, with the total effect in this case estimated as 0.96 (significant at 5% level). Oddly, it appears that revenue rules may be counter-productive for countries with moderatequality governments, i.e. revenues rules for this group appear to lower fiscal balances (total effect of ). A possible reason for this may be that even though a country manages to implement a revenue rule, nothing in principle constrains expenditures. This could lead to larger deficits in cases with less efficient governments. The results in Table 5 suggest that balanced budget rules and debt rules seem to have the desired effect of increasing primary balance surpluses in countries with highly efficient 19

20 governments. For countries with less efficient governments there is no significant effect from having implemented national fiscal rules or the rules may even be counterproductive. 4.4 IS THE RESPONSE OF PRIMARY BALANCE DIFFERENT DURING THE FINANCIAL CRISIS? This section investigates how the effect of fiscal rules may change during the financial crisis, and whether these differential responses to the extent that they are evident-- differ by the leve3l of government efficiency. To do this end we define a dummy variable that takes on the value of unity for the period for all countries and then interact this dummy with the fiscal rule indicators for each cluster. The base model in equation (1) is then rewritten as ' FP i,t 0 FP i,t 1 X i,t 2 FR i,t 3 D 1,t FR i,t i,t (2) where D1, t 1 for t and zero otherwise. The parameter 3 represents the change in the effect of national fiscal rules on cyclically adjusted primary balance during the financial crisis. The total effect during the crisis period is equal to 2 3. We then extend this equation by distinguishing between clusters of countries. The regression equation is FP FP X FR FR * QoG ' it, 0 it, 1 it, 2 it, 3 it, 2 FR * QoG D * FR D * FR * QoG D * FR * QoG 4 it, 3 5 1, t it, 6 1, t it, 2 7 1, t it, 3 it, (3) where 2 3 and 2 4 are the total effects of national fiscal rules during normal times for good and moderate QoG, respectively. The total effects of fiscal rules during crisis periods are given by 2 5, and for high, good and moderate QoG respectively. We run these two regression equations using first the IMF based measure (NFRI) and then using the EU Commission measure (FRI) as our indicator of the strength of fiscal rules. The results are reported in Table 6. All regressions also include the same set of control variables as in Table 2 and Table 4, but Table 6 only reports the estimates of the effects of fiscal rules interacted with government efficiency and distinguishing between normal times and the financial crisis period. The first two columns report the results when running regression equation (2) and the other two columns show the results when estimating equation (3). Looking first at columns (1) and (2), we find that the strong and significant effect from national fiscal rules on primary surplus remain regardless of our measure of fiscal rule strength and the point estimates are very close to those reported in Table 4. The interaction term between fiscal strength and the crisis dummy is significantly negative using the NFRI 20

21 measure (IMF-based index), giving an insignificant total effect of rules during the crisis (total effect is term in brackets, 0.343; insignificant at 10% level). Moreover, even though crisis dummy parameter ( 3 of equation 2) is insignificant when using FRI (EU Commission-based measure) as the fiscal rules indicator, and the normal period effect is significantly positive (0.536), the total effect during crisis (term in brackets, 0.264) is positive but statistically insignificant at the 10% level. - Table 6 about here - Columns (3) and (4) report the results from estimating equation (3). Column (3), with the NFRI measure, again indicates that effect of rules on primary balance is positive and significant only for high quality governments. In addition, there is no effect of rules during the crisis period regardless of government efficiency. In particular, the parameter estimate for the high quality government during the crisis period is -0.71, giving a total effect (0.26) that is positive but statistically insignificant. Fiscal rules therefore do not moderate primary balance deficits during the crisis at moderate, good or high levels of government efficiency. Column (4), with the FRI measure, are very similar for the moderate and high levels of government efficiency rules do not appear effective in either tranquil or periods of crisis. But for the highest quality governments, the results indicate that fiscal rules moderate fiscal imbalances both during tranquil and crisis periods. Indeed, there is no significant difference between the effects of fiscal rules during crisis and non-crisis periods. The results from column (3) for the high quality governments are particularly interesting fiscal rules are effective during normal periods but do not moderate deficits during crisis periods. This may reflect the ability of high quality governments to follow fiscally sustainable policies during normal times, reinforced by fiscal rules, but the explicit desire to make these rules conditional on certain measures of economic performance. During a period of crisis, optimal policy may well be to pursue much more expansionary policies to stimulate economies, with associated larger fiscal imbalances, when these policies are set against a backdrop of otherwise sustainable fiscal policies. By contrast, moderate and good governments do not rely upon fiscal rules during either normal or crisis periods. 5. CONCLUSIONS The experiences with fiscal rules operating at the supranational level are not very encouraging. This is particularly the case in the EA where several attempts to enforce the union-wide Stability and Growth Pact (SGP) have not been successful. Indeed, while 21

22 complete on paper with detailed excessive deficit procedures, sanctions for violators, and other enforcement procedures, the SGP has essentially not been enforced in practice. In this paper we have taken a closer look at the track record of fiscal rules implemented at the national level in the EU. We investigate whether fiscal rules help to increase the cyclically adjusted primary balance in a panel of 27 EU countries over the period Based on the IMF Fiscal Affairs Division database, we develop a unique fiscal rule index based on distinct characteristics of actions, legislative or procedural, that constrain fiscal policy actions in each country at each point in time. In addition to this measure we also use the EU Commission numerical fiscal rule index covering the same sample. We investigate whether these aggregate rules are effective in limiting fiscal deficits, measured by the cyclically-adjusted primary budget balance, and whether the effects of national fiscal rules depend on the degree of government efficiency. We also explore which specific types of rules balance budget, expenditure, revenue or debt rules are effective in limiting budget imbalance. Finally, we investigate the enforcement of fiscal rules differs between normal or tranquil periods and crisis periods. In all of these investigations, we measure the marginal effects of national fiscal rules while conditioning on a range of economic and political factors. Our main finding is that fiscal rules have a significant effect on the primary balance, whether in normal or crisis periods but these effects are primary concentrated in countries with quite high levels of government efficiency and primary because they implement balanced budget and debt rules. (Expenditure and revenue rules do not appear effective in promoting fiscal solvency). Countries with tight fiscal rules have in fact larger primary balance surpluses as a percentage of GDP compared to countries with softer fiscal frameworks. This may be comforting news for policymaking in the EA. Indeed, a key component of the fiscal compact, as recently agreed among a large majority of members of the EA, is that these rules shall take effect in the national law of all EA countries through provisions of binding force and permanent character, preferably constitutional, or otherwise guaranteed to be fully respected and adhered to throughout the national budgetary processes. However, our empirical results also suggest that it is not enough to implement new stronger fiscal rules in order to attain sound public finances in the long-run. Only the most efficient governments in the EU appear able to effectively enforce national fiscal rules. We find that countries with highly efficient governments are also those countries where primary balance surpluses tend to be larger for a given strength of the fiscal framework. This implies 22

23 that it would be helpful, for purposes of promoting fiscal sustainability, for countries that already have implemented fiscal rules to also increase government efficiency. Our empirical results suggest that government reform of this nature would have the increasing the effectiveness of the existing fiscal rules, at least once a high threshold of government quality has been attained. Overall, the findings of this paper suggest that fiscal rules with a national anchor may be able to foster more sustainable public finances. The fiscal rules Phoenix may indeed be rising from the ashes. References Afonso, A. and S. Hauptmeier (2009),. "Fiscal Behaviour in the European Union: Rules, Fiscal Decentralization and Government Indebtedness," Working Paper Series 1054, European Central Bank. Albuquerque, B. (2011), Fiscal Institutions and Public Spending Volatility in Europe, Economic Modelling, 28: Arellano, M. and S.R. Bond (1991), Some Tests of Specification for Panel Data: Monte Carlo Evidence and an Application to Employment Equations, Review of Economic Studies, 58: Arellano, M. and O. Bover (1995), Another Look at the Instrumental Variables Estimation of Error- Components Models, Journal of Econometrics, 68: Ayuso-i-Casals, J. (2012), National Expenditure Rules: Why, How and When, European Commission, Economic Papers 473. Bergman, U. M., M. M. Hutchison (2014), Economic Stabilization in the Post-Crisis World: Are Fiscal Rules the Answer? working paper. Bergman, U. M., M. M. Hutchison and S. H. Jensen (2013), Policy Actions in the Euro Area and Market Perceptions of Sovereign Default Risk, working paper Bernanke, B. (2010), Fiscal Sustainability and Fiscal Rules, Boards of Governors of the Federal Reserve System, mimeo. Bontempi, M.E. and I. Mammi (2012), A Strategy to Reduce the Count of Moment Conditions in Panel Data GMM, MPRA Paper No Calderón, César, Roberto Duncan and Klaus Schmidt-Hebbel (2012). Do Good Institutions Promote Counter-Cyclical Macroeconomic Policies? Federal Reserve Bank of Dallas Globalization and Monetary Policy Institute, Working Paper No. 118 (June). Charron, N., V. Lapuente and B. Rothstein (2010), "Measuring the Quality of Government and Subnational Variation", Report for the European Commission Directorate-General Regional Policy Directorate Policy Development. Dahan, M and M. Strawczynski (2010), "Fiscal Rules and Composition Bias in OECD Countries," CESifo Working Paper Series 3088, CESifo Group Munich. De Haan, J., R. Jong-A-Pin, J.O. Mierau (2013), Do Budgetary Institutions Mitigate the Common Pool Problem? New Empirical Evidence for the EU, Public Choice, 156: Debrun, X., Moulin, L., Turrini, A., Ayuso-i-Casals, J. and Kumar, M. S. (2008), Tied to the Mast? National Fiscal Rules in the European Union, Economic Policy, 23:

24 Gregor, M. (2004), Governing Fiscal Commons in the Enlarged EU, Working Papers IES 56, Charles University Prague, Faculty of Social Sciences, Institute of Economic Studies. Hall A.R. and F.P.M. Peixe (2003), A Consistent Method for the Selection of Relevant Instruments, Econometric Reviews, 22: Hallerberg, M. (2004). Domestic Budgets in a United Europe: Fiscal Governance from the End of Bretton Woods to EMU. Ithaca, NY: Cornell University Press. Hallerberg, M and J. von Hagen (1999), "Electoral Institutions, Cabinet Negotiations, and Budget Deficits in the European Union," in J. Poterba and J. von Hagen (eds.), Fiscal Institutions and Fiscal Performance, University of Chicago Press Hallerberg, M., R. Strauch and J. von Hagen (2007), The Design of Fiscal Rules and Forms of Governance in European Union Countries, European Journal of Political Economy, 23: Holm-Hadulla, F., S. Hauptmeier and P. Rother (2010), "The Impact of Numerical Expenditure Rules on Budgetary Discipline over the Cycle," Working Paper Series 1169, European Central Bank. Hughes Hallett, A. and S. H. Jensen (2012), Fiscal Governance in the Euro Area: Institutions vs. Rules, Journal of European Public Policy, 19: International Monetary Fund (2009), Fiscal Rules Anchoring Expectations for Sustainable Public Finances, prepared by the Fiscal Affairs Department, IMF, Washington DC. Nerlich, C. and W.H. Reuter (2013), The Design of National Fiscal Frameworks and their Budgetary Impact, ECB Working Paper Series, No Roodman, David, 2009a, How to do xtabond2: An instroduction to difference and system GMM in Stata, Stata Journal, Vol. 9, No. 1, pp Roodman, D.M. (2009b), A Note on the Theme of Too Many Instruments, Oxford Bulletin of Economics and Statistics, 71: Schaechter, A., T. Kinda, N. Budina, and A. Weber, Fiscal Rules in Response to the Crisis Toward the Next-Generation Rules. A New Dataset, IMF Working Paper 12/187. Schuknecht, L., P. Moutot, P. Rother and J. Stark (2011), The Stability and Growth Pact: Crisis and Reform, Occasional Paper Series No. 129, European Central Bank, Frankfurt. von Hagen, J. (1991), A Note on the Empirical Effectiveness of Formal Fiscal Constraints, Journal of Public Economics, 44: Yläoutinen, S. (2004), The role of electoral and party systems in the development of fiscal institutions in the Central and Eastern European countries, ZEI Working Papers B , ZEI - Center for European Integration Studies, University of Bonn. 24

25 Table 1: Descriptive statistics of fiscal strength indicators in EU countries All countries Mean Median Std dev Min Max FRI NFRI # of national rules Note: FRI is the EU Commission numerical fiscal rule indicator and NFRI is the IMF based index of fiscal strength. Number of national fiscal rules refer to the average number of rules per country over the sample as defined in the IMF database. 25

26 Table 2: The effect of government efficiency and fiscal rules on the cyclically adjusted primary balance. (1) (2) (3) (4) (5) (6) PBCA(-1) 0.695*** 0.687*** 0.683*** 0.690*** 0.680*** 0.676*** (0.048) (0.047) (0.047) (0.046) (0.044) (0.043) GAP(-1) ** *** *** ** *** *** (0.024) (0.022) (0.022) (0.024) (0.023) (0.022) Debt(-1) 0.036*** 0.037*** 0.039*** 0.037*** 0.038*** 0.040*** (0.007) (0.006) (0.005) (0.007) (0.005) Dependency * * * * * * (0.046 (0.048) (0.047) (0.047) (0.049) (0.049) Openness 1.206* * (0.606) (0.614) (0.586) (0.621) (0.633) (0.602) Population *** *** *** -8,975*** *** *** (2.750) (3.114) (3.060) (0.621) (2.829) (3.054) Inflation 0.060* 0.062* 0.063* 0.060* 0.064* 0.064* (0.033) (0.033) (0.034) (0.033) (0.034) (0.034) Years in office 0.141* 0.139* 0.140* 0.143* 0.140* 0.142* (0.077) (0.076) (0.075) (0.077) (0.075) (0.075) Election (0.227) (0.232) (0.229) (0.230) (0.233) (0.231) Plurality * * (0.407) (0.370) (0.422) (0.416) (0.353) (0.434) Gov. fragmentation (0.595) (0.681) (0.684) (0.594) (0.722) (0.703) Commitment 0.626* ** 0.660* ** (0.360) (0.390) (0.314) (0.374) (0.409) (0.320) Delegation (0.430) (0.453) (0.427) (0.457) (0.461) (0.437) Run-up (0.235) (0.278) (0.291) (0.233) (0.269) (0.293) Enlargement (0.327) (0.313) (0.322) (0.368) (0.320) (0.364) NFRI 0.593* 0.651* (0.325) (0.366) FRI 0.471*** 0.486*** (0.137) (0.174) QoG (0.640) (0.665) (0.673) NFRI*QoG (0.388) FRI*QoG (0.211) #observations #instruments AR(1) AR(2) Sargan Note: AR(1) and AR(2) are tests of autocorrelation of the first and second order, respectively. Only p-values are reported for Sargan and Hansen J over-identifying tests and the two tests for autocorrelation. Clustered and robust standard errors reported below each estimate. *** denotes significant at the 1% level, ** at the 5% level and * at the 10% level. 26

27 Table 3: Clusters of EU countries with respect to quality of government (QoG). Cluster 1: High QoG Cluster 2: Good QoG Cluster 3: Moderate QoG Denmark France Czech Republic Sweden Belgium Lithuania Finland Malta Hungary Netherlands Spain Slovakia Austria Portugal Poland Germany Cyprus Latvia Unite Kingdom Estonia Greece Ireland Slovenia Italy Luxembourg Bulgaria Romania Source: Charron et al. (2010), table 18 on page

28 Table 4: Effect of quality of government and fiscal rules on the cyclically adjusted primary balance. (1) (2) NFRI FRI FR 0.738* 0.527*** (0.375) (0.131) FR*QoG ** 0.419* (0.447) (0.231) [ 0.213] [0.108] FR*QoG (0.873) (0.420) [1.037] [0.650] #obs #instruments AR(1) AR(2) Sargan Note: QoG2 is a dummy variable equal to one for countries included in cluster 2 and zero otherwise and QoG3 is a dummy variable equal to one for countries included in cluster 3 and zero otherwise. Clustered and robust standard errors reported within parentheses below each estimate. The total effect of fiscal rules for each cluster of countries is shown within brackets below the standard error. Only p-values are reported for Sargan and Hansen over-identifying tests and the two tests for autocorrelation. ** denotes significant at the 1% level, * at the 5% level and + at the 10% level. All regressions also include the same control variables as reported in Table 2 but they are not reported for brevity. 28

29 Table 5: Response of the cyclically adjusted primary balance to specific national fiscal rules. Panel A: Response of cyclically adjusted primary balance to specific fiscal rules (1) (2) (3) (4) (5) Balanced budget rule 0.549*** 0.471*** (0.157) (0.156) Expenditure rule (0.174) (0.206) Revenue rule (0.121) (0.169) Debt rule 0.387*** (0.135) (0.143) #obs #instruments AR(1) AR(2) Sargan Panel B: Response of cyclically adjusted primary balance to specific fiscal rules interacted with quality of government. (1) (2) (3) (4) Balanced Expenditure Revenue Debt rule budget rule rule rule FR 0.474*** *** (0.100) (0.142) (0.359) (0.083) FR Good QoG *** (0.189) (0.234) (0.375) (0.157) [0.296] [-0.186] [0.007] [0.139] FR Moderate QoG ** (0.392) (0.412) (0.396) (0.210) [0.964**] [0.433] [-0.527***] [0.230] #obs #instruments AR(1) AR(2) Sargan Note: Clustered and robust standard errors reported below each estimate. Only p-values are reported for the Sargan over-identifying tests and the two tests for autocorrelation. In Panel B we also report the total effect of fiscal rules for each cluster of countries is shown within brackets below the standard error. ** denotes significant at the 1% level, * at the 5% level and + at the 10% level. All regressions also include the same control variables as reported in Table 2 but are not reported for brevity. 29

30 Table 6: Responses of the cyclically adjusted primary balance to national fiscal rules during normal times and during the financial crisis. (1) (2) (3) (4) NFRI FRI NFRI FRI FR 0.790** 0.536*** 0.974** 0.544*** (0.310) (0.137) (0.445) (0.133) FR Good QoG (0.571) (0.345) [0.023] [0.308] FR Moderate QoG (0.721) (0.467) [0.987] [0.814*] FR Crisis 0.448* *** (0.244) (0.178) (0.225) (0.221) [0.343] [0.264] [0.263] [0.537**] FR Crisis Good QoG (0.661) (0.457) [ 0.793] [ 0.054] FR Crisis Moderate QoG (0.436) (0.399) [0.981] [0.238] #obs #instruments AR(1) AR(2) Sargan Note: Clustered and robust standard errors reported below each estimate. Only p-values are reported for the Sargan overidentifying tests and the two tests for autocorrelation. The total effect of fiscal rules for each cluster of countries during normal times and during crisis is shown within brackets below the standard error. ** denotes significant at the 1% level, * at the 5% level and + at the 10% level. All regressions also include the same control variables as reported in Table 2 but are not reported for brevity. 30

31 Figure 1: Fiscal Rule Strength for EU countries in 2000 and IMF fiscal rule index (NFRI) SE ES BG PL UK FR LT DK EE LU DE NL AT SI FI SK CZ HU BE LV IT PT RO IE CY EL MT EU Commission numerical fiscal rule index (FRI) SE ES BG PL UK FR LT DK EE LU DE NL AT SI FI SK CZ HU BE LV IT PT RO IE CY EL MT Source: IMF database and European Commission. Both indices have been rescaled to [0,1] variables. 31