Address (work): Nettelbosje 2, Groningen E-mail addresses: bramgootjes@hotmail.com; b.gootjes@student.rug.nl Telephone: +316 23175950

(1)

1

Procyclicality of fiscal policy in European Union countries

Bram Gootjes S2382466 University of Groningen Master’s Thesis RM E&B Supervisor: Prof. Dr. Jakob de Haan

Version 23 June 2019

Abstract:

This thesis sheds light on the cyclicality of fiscal policy for a panel of 27 European Union (EU) member states over the period 2000-2015. Using a real-time data approach, I examine whether institutional factors – like fiscal rules and government quality – improve the stabilization function of fiscal policy. The results indicate that ex-post fiscal policy is procyclical on average, but planned fiscal policy appears to be acyclical. Government efficiency and fiscal rules (and, in particular expenditure rules) reduce fiscal procyclicality. However, the effect of government efficiency weakens as fiscal rules become more stringent, indicating that government efficiency and fiscal rules are substitutes for improving the cyclical behaviour of fiscal policy.

JEL classifications: E32; E62; H6

Key words: Fiscal policy; Procyclicality; Fiscal rules; Institutional quality

Address (work): Nettelbosje 2, Groningen

E-mail addresses: bramgootjes@hotmail.com; b.gootjes@student.rug.nl

(2)

2

1. Introduction

The global financial crisis of 2007-2008 had a profound impact on the world economy. Several countries had to bail-out highly indebted banks to prevent a collapse of their banking systems, whereas at the same time a contraction in economic activity led to a sharp reduction of tax revenues. European countries in particular experienced a severe downturn, as in the aftermath of the financial crisis the European Union (EU) slid into a sovereign debt crisis. Some countries – Greece, Portugal, Ireland, Spain, and Cyprus – were unable to repay or refinance their debts and needed the assistance of other EU countries, the European Central Bank (ECB) and the International Monetary Fund (IMF) to avoid a possible sovereign default. The concern for the sustainability of countries’ public finances increased all over the world. As a consequence, there has been a renewed interest in the role of fiscal policy as a stabilization tool for the economy (Feldstein, 2009; Fatás and Mihov, 2011; IMF, 2017).

Early public finance theories already highlighted the importance of fiscal policy as an instrument for macroeconomic stabilization (see, e.g., Musgrave, 1959). The conventional wisdom is that optimal fiscal policy should be countercyclical – that is, fiscal policy should be contractionary in good times and expansionary in bad times to smooth out business cycle fluctuations in output. Fiscal procyclicality in turn increases output volatility and might have a negative impact on long-term economic growth (Woo, 2009). Although implementing countercyclical fiscal policy seems straightforward, in practice fiscal policy is often procyclical. In the past, this was argued to be a phenomenon of mainly the developing world, but recent studies show that procyclicality is also evident in European countries (e.g., Eyraud et al., 2017), despite the fact that for member states of the European monetary union fiscal policy is the only stabilization tool available at the national level (Wyplosz, 2005). Procyclical fiscal policy could therefore have far-reaching consequences for the stability of national output and the sustainability of public finances of EU countries.

(3)

3 the EU’s economic governance and in response the European Commission (EC) strengthened the SGP. Legislative packages known as the “Six Pack”, “Fiscal Compact”, and the “Two Pack” were introduced to reinforce monitoring and enforcement procedures of the European fiscal rules and to develop ever-closer coordination of economic policies within the euro area. Although fiscal rules are widely accepted as an instrument to promote budgetary discipline, relatively little is known about their impact on the stabilization function of fiscal policy. Recent papers found that well-designed fiscal rules reduce procyclicality, but Bergman and Hutchison (2015) suggest that this result only holds when these rules are supported by sufficiently high levels of government efficiency. These authors find that government efficiency alone is not sufficient to reduce procyclicality, but other studies report that high institutional quality alleviates procyclical fiscal policy (e.g., Calderón et al., 2016 and Bova et al., 2018).

A drawback of these studies is that they only consider ex-post fiscal data, which has often been revised and almost certainly gives a misleading inference about policymaking (Croushore, 2011). The decision-making process of fiscal policy contains long and often unpredictable time lags and governments have inaccurate estimates of the state of the economy at the time of their decisions. Therefore, budgetary outcomes of actual fiscal policies may be significantly different from the ex-ante intentions of policymakers (Cimadomo, 2016). Furthermore, institutional variables may affect plannend and implemented fiscal policy (i.e. real-time fiscal policy) in different ways.

(4)

4 My results indicate that over the period 2000-2015 fiscal policy plans are acyclical on average in EU countries. However, the cyclical stance weakens during the implementation of fiscal policy, whilst ex-post fiscal outcomes appear to be procyclical. Fiscal rules improve the cyclical behaviour of fiscal policy and result into acyclical fiscal policies once they are sufficiently stringent. In particular, expenditure rules are highly effective in reducing fiscal procyclicality. Similarly, I also find that the level of government’s administrative quality is of great importance for improving the cyclicality of fiscal policy. In contrast to Bergman and Hutchison (2015), I do not find that the institutional factors are complements. Instead, it seems more likely that fiscal rules and government efficiency are institutional substitutes for improving the cyclical behaviour of fiscal policy.

The remainder of this thesis is structured as follows: Section 2 describes the related literature and the hypotheses. Section 3 offers the conceptual framework and the data. Section 4 presents the results and section 5 provides a robustness analysis. Section 6 concludes.

2. Related literature and hypotheses

2.1 Cyclical stance of fiscal policy

Why is fiscal policy often procyclical? Two main explanations have been provided to account for the procyclicality bias of fiscal policy: borrowing constraints during bad times and political agency

problems that explain how distorted political incentives result into a deterioration of the budget in times

of economic prosperity (see Box 1). Several empirical studies argue that fiscal policy is procyclical in emerging markets and developing countries, whereas it is acyclical in advanced economies.1_Ilzetzki

and Végh (2008), however, show that procyclicality is not solely confined to developing countries, as they find evidence of procyclical fiscal policy in high-income countries as well. Similarly, Galí and

1_{Gavin and Perotti (1997) find that fiscal policy has been procyclical in Latin American countries (over the period 1968-1995),}

(5)

5 Perotti (2003) show that fiscal policy in Economic and Monetary Union (EMU) countries was significantly procyclical, considering a sample period of 1980-2002. Their finding, however, only holds for the period before the Maastricht Treaty, as significance largely disappeared in the post-Maastricht period. Examining other Organisation for Economic Co-operation and Development (OECD) countries, this seemingly appeared to be a detection of a global trend towards more countercyclical fiscal policies. More studies report results on the graduation of fiscal policy towards a more countercyclical stance (see, e.g., Wyplosz (2006) and Bénétrix and Lane (2013) for EMU member countries; Frankel et al. (2013) for developing economies), but not all research supports this finding. For instance, Bova et al. (2018)

Box 1. Procyclical fiscal policy

A common answer that explains the procyclicality bias of fiscal policy is the supply of credit. Gavin and Perotti (1997) discuss stylized facts of fiscal policy in Latin America and find that fiscal policies are particularly procyclical in bad macroeconomic times. They argue that borrowing constraints intensify in bad times and force countries to cut spending. Kaminsky et al. (2004) study the cyclical properties of credit supply and fiscal policy for 104 countries and find that for developing countries – and particularly for middle high income countries – the capital flow cycle and the macroeconomic cycle amplify each other (i.e. “when it rains, it pours”).

(6)

6 do not find that the procyclicality bias had declined over time, covering a sample of 48 resource-rich countries over the period 1970-2013. Furthermore, a recent study of Eyraud et al. (2017) shows that fiscal policy was procyclical in euro area countries over the 1999-2015 period.

2.2 Fiscal rules and government quality

As a reaction to the ever-growing pressures on public finances in the last few decades, many countries adopted fiscal rules to promote more budgetary discipline. Fiscal rules impose long-lasting constraints on fiscal policy through numerical limits on budgetary aggregates (Schaechter el al., 2012). Four main type of fiscal rules can be distinguished: debt rules, balanced budget rules, expenditure rules, and revenue rules. Although each type of rule serves a different purpose, they are generally aimed at correcting budgetary distortions and maintaining sound fiscal policies. Recent empirical studies have found that fiscal rules lower public debt (Azzimonti et al., 2016), lower budget deficits (Caselli and Reynaud, 2019), constrain political budget cycles (Gootjes et al., 2019) and reduce the probability of experiencing a sovereign debt crisis (Asatryan et al. 2018). However, the role of fiscal rules as a stabilization tool for the macroeconomic cycle is ambiguous at forehand. Fiscal rules intend to reduce excessive deficits and control the public debt level, but when they do not sufficiently consider cyclical developments they can potentially trigger procyclical polices: in good times they can allow for too large of a margin to spend windfall revenues, whereas in bad times they can force governments to cut spending. In spite of this, most research finds that these rules reduce fiscal procyclicality, but the design of fiscal rules seems to play a crucial role. For instance, flexible rules can be associated with enhanced countercyclicality of fiscal policy (Guerguil et al., 2017).2_{Moreover, not every type of fiscal rule seems}

to be equally efficient in improving the cyclical fiscal stance, but especially the combination of multiple fiscal rules appears to be fruitful (Nerlich and Reuter, 2015; Combes et al., 2017; Bergman and Hutchison, 2018).

2_{Note that flexibility in fiscal rules comes with a price, as flexibility makes fiscal rules less transparent and more complicated.}

(7)

7 Not all empirical evidence is in favour of fiscal rules. As fiscal rules are generally adopted to promote fiscal discipline, countries with non-sustainable public finances may be more likely to adopt fiscal rules. Most research argues that causality runs from fiscal rules to fiscal behaviour, but recently Heinemann et al. (2018) showed with a meta-regression analysis of 30 empirical studies that fiscal rules cannot be treated as exogenous and that the budgetary impact of fiscal rules tends to lose statistical significance once endogeneity is properly accounted for. Another issue with fiscal rules is that they may encourage the use creative accounting. Gilbert and de Jong (2017), for instance, show that in the euro area the European fiscal rules cause an upward bias in the EC’s fiscal forecasts when national governments expect the budget deficit to exceed the critical value of 3% of GDP. Also not all research supports the hypothesis that fiscal rules reduce procyclicality. For example, Bova et al. (2014) show that the adoption of fiscal rules did not help emerging market and developing economies escape the procyclicality trap.

Another institutional factor considered in the literature is the quality/efficiency of government administration (i.e. the institutional quality). Governments with a high institutional quality are independent from political pressures, clearly formulate fiscal plans, and credibly commit to these plans. Calderón et al. (2016) find that the quality of the institutional framework plays an important role in countries’ ability and willingness to implement countercyclical fiscal policy. Similar results are reported by Frankel et al. (2013), who argue that institutional quality is pivotal for a country’s ability to graduate from procyclical to countercyclical policies.

(8)

8 Whether assessing fiscal rules and institutional quality simultaneously affects the results remains an empirical question, but so far, only few studies addressed it. Bergman and Hutchison (2015) find that government efficiency is not enough to reduce procyclicality of fiscal policy, using a panel of 81 advanced, emerging and developing countries. They show that national fiscal rules are very effective in reducing procyclicality of fiscal policy when government efficiency is sufficiently high. Furthermore, they find that a high quality of government administration and strong fiscal rules is a potent combination that enables governments to follow countercyclical fiscal policies. However, Bova et al. (2018) find that the adoption of fiscal rules does not reduce procyclicality in a significant way, but instead they show that institutional quality of the government helps to limit procyclicality. So, the literature has not found consistent results when the role of different types of institutional factors are examined simultaneously, but for EU countries it is important to find out the actual impact of the these factors. Fiscal policy is the main policy tool to stabilize the macroeconomic cycle for EU member states and the political and fiscal institutional frameworks – at the national and supranational levels – form the guardrails of national fiscal policies in EU economies. To shed more light on the effect of institutional factors on the cyclical behaviour of governments in the EU, I consider the following hypotheses:

Hypothesis 1 Fiscal policy is procyclical in EU countries.

Hypothesis 2 Institutional factors – e.g. fiscal rules and government efficiency – improve the

cyclical stance of fiscal policy.

Hypothesis 3 Fiscal rules and government efficiency act as complements in their effect on the

cyclical stance of fiscal policy.

2.3 Real-time analysis

(9)

9 tend to be large. Differences between the first release and final outcome of fiscal variables are driven by several factors. It may be the consequence of methodological changes in calculating and collecting the data. As governments do not have full control over the budget, uncertainty about the state of the economy when planning the budget and during implementation of fiscal measures may also play a role. Furthermore, policymakers may intervene in the production of statistics out of strategic reasons (Cimadomo, 2016). As data revisions would clearly matter, fiscal policy analysis considering only the latest available data almost certainly gives a misleading representation of the actual intentions of fiscal policymakers. Real-time data analysis therefore uses data containing the information that comes closest to the information set available to policymakers when constructing and implementing their fiscal plans (Croushore, 2011).

There seems to be a general agreement in the literature that the assessment of fiscal policy with real-time data tends to signal a more countercyclical stance (see Cimadomo (2016) for a survey of the literature). For instance, Golinelli and Momigliano (2009) find a weak indication of countercyclicality in the policy intentions of EMU countries over the 1994-2008 period. However, compared to the ex-post fiscal outcomes, this is completely off-set in the actual fiscal policies. Beetsma and Giuliodori (2010) find that planned fiscal policy is acyclical for EU countries and countercyclical for other OECD countries. Furthermore, they show that EU countries tend to react procyclical to unexpected deviations in the output gap, whereas the response of other OECD countries is acyclical.

(10)

10 Improving fiscal forecasts and reducing the wishful thinking bias are absolutely crucial, but whether this can solve the procyclicality puzzle remains debatable. Using a simple framework (see Box 2) and data of 101 countries, Avellan and Vuletin (2015) show that fiscal policy responds procyclical to both predictable components and unanticipated cyclical developments. As a consequence, fiscal policy ends up more procyclical ex-post, but they do not find any evidence that this is driven by the degree of over-optimism. The authors argue that “one should not readily accept the interpretation that forecast errors have ‘unfortunate’ effects on fiscal procyclicality” and provide some preliminary results that traditional political economy explanations could help solve the procyclicality puzzle.

Box 2. Conceptual framework to explain fiscal procyclicality

Avellan and Vuletin (2015) try to explain the interrelationship between fiscal procyclicality and output forecast errors. Using a simple conceptual framework based on the correlation between the change in government expenditures (∆𝑔) and the ex-post change in output (∆𝑦) – denoted as 𝑐𝑜𝑟𝑟(∆𝑔, ∆𝑦) – they rationalize the cyclical stance of government expenditures. In addition, the ex-post observed degree of cyclicality can be decomposed into a predicted component (∆𝑦𝑃𝑟𝑒𝑑_;

representing policymakers’ intentions) and an unanticipated driver (∆𝑦𝑈𝑛𝑝𝑟𝑒𝑑_{), i.e. ∆𝑦 = ∆𝑦}𝑃𝑟𝑒𝑑₊

∆𝑦𝑈𝑛𝑝𝑟𝑒𝑑. A positive correlation (𝑐𝑜𝑟𝑟(∆𝑔, ∆𝑦) > 0) implies procyclicality and a negative correlation (𝑐𝑜𝑟𝑟(∆𝑔, ∆𝑦) < 0) points to countercyclicality:

𝑐𝑜𝑟𝑟(∆𝑔, ∆𝑦) = 𝑐𝑜𝑣(∆𝑔, ∆𝑦) 𝑠𝑑(∆𝑔)𝑠𝑑(∆𝑦)−

𝑐𝑜𝑣(∆𝑔, ∆𝑦𝑃𝑟𝑒𝑑) + 𝑐𝑜𝑣(∆𝑔, ∆𝑦𝑈𝑛𝑝𝑟𝑒𝑑)

𝑠𝑑(∆𝑔)𝑠𝑑(∆𝑦)

Using this conceptual framework, over-optimism in output forecasts (i.e., ∆𝑦𝑈𝑛𝑝𝑟𝑒𝑑= ∆𝑦 − ∆𝑦𝑃𝑟𝑒𝑑< 0) is neither necessary nor sufficient to explain fiscal procyclicality. What matters for procyclical fiscal policy is not whether countries are over-optimistic, but whether there is a systematic pattern of correlation between the change in government expenditures and the forecast error; a positive association points to procyclicality and leads to an increase in government expenditures when ex-post change in output is larger than the predicted change (∆𝑦 > ∆𝑦𝑃𝑟𝑒𝑑) and to a decrease in government spending for the reverse situation (∆𝑦 < ∆𝑦𝑃𝑟𝑒𝑑). In principle, even when a country has no forecast bias (∆𝑦𝑈𝑛𝑝𝑟𝑒𝑑_{= 0), there could still be a systematic pattern of}

(11)

11 Holm-Hadulla et al. (2012) find that over the 2002-2008 period, the presence of domestic expenditure rules in EU member states reduced the procyclical reaction of government spending to unexpected deviations in the output gap. However, they only study the impact of expenditure rules on surprising cyclical developments and do not consider whether these rules lead to more countercyclical policy intentions. In general, the literature has paid only scant attention to the context-conditionality of fiscal procyclicality in earlier stages of fiscal policymaking. One of the exceptions is the work of Eyraud et al. (2017). They examine whether political economy factors affected fiscal outcomes in the euro area during the planning, implementation and ex-post phase and show that policy distortions at both the national and supranational levels lead to excessive spending during good times in euro area countries. National and supranational fiscal rules have not successfully alleviated this procyclicality bias, arguably because of low ex-post compliance to the fiscal rules. Eyraud et al. (2017), however, do not test the impact of fiscal rules on the government budget in a formal econometric framework. Furthermore, they do not consider the quality of government administration.

To consider the cyclical behaviour of fiscal policy and the conditional effect of institutional factors in earlier stages of fiscal policymaking, I will test hypotheses 1-3 also for the fiscal policy plans and real-time implementation of the budget.

3. Empirical strategy and data

3.1 Baseline model

To consider the cyclical reaction of fiscal policy across its various stages, I test a standard fiscal reaction function for a sample of 27 EU countries overt the period 2000-2015.3_{The baseline specification is a}

dynamic panel data model taking the following form:

∆𝐶𝐴𝑃𝐵𝑖,𝑡𝑡−1+𝑠= 𝛼 + 𝛿𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠+ 𝛾𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠+ 𝛽′𝑋𝑖+ 𝜇𝑖+ 𝜏𝑡+ 𝜀𝑖,𝑡, for 𝑠 = 0, 2 | 𝑒𝑥 𝑝𝑜𝑠𝑡, (1)

(12)

12 where the superscript 𝑡 − 1 + 𝑠 indicates the year of publication of the fiscal and economic variables. Fiscal plans are usually formed a year ahead, i.e. policymakers use information of economic and fiscal forecasts constructed in period 𝑡 − 1 to make fiscal plans for period 𝑡; the implementation phase of the budget and the actual realization of fiscal policy can be assessed using data of period 𝑡 + 1 and the latest vintage available, respectively. To assess the planning and implementation stage of fiscal policy, I use a real-time dataset constructed by Gilbert and de Jong (2017). This dataset consists of official forecasts of budgetary aggregates and macroeconomic variables published by the EC. Since 1998, every spring and autumn the EC provides economic forecasts of the current year and the year ahead for all EU members. Furthermore, the EC provides realization data of the past four years. I consider the spring forecasts, as these come closest to the information available to policymakers when they make their budgetary plans (de Jong and Gilbert, 2018).4_{I have collected data on the ex-post realization of}

macroeconomic and fiscal variables from the annual macro-economic (AMECO) database of the EC. See Table A.1 in the appendix for the descriptive statistics of all variables.

I employ the change in the cyclically-adjusted primary budget balance (∆𝐶𝐴𝑃𝐵𝑖.𝑡𝑡−1+𝑠) as the left-hand

side variable to capture the discretionary fiscal action (in country 𝑖 for period 𝑡 in percentage of GDP). The CAPB corrects for interest payments on outstanding government debt, as these payments do not reflect government policies in the current period. Furthermore, the CAPB is adjusted for fluctuations in the business cycle and thereby gives a better reflection of the underlying budgetary stance of the government. In addition, the lag of the government budget balance (𝐶𝐴𝑃𝐵_{𝑖.𝑡−1}𝑡−1+𝑠₎_{controls for path}

dependence.5

I include the level of the output gap (𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠) to capture the cyclical reaction of fiscal policy; a negative

sign points to a procyclical reaction, whilst a positive value indicates countercyclicality. 𝜇𝑖is a

country-4_{Most often, fiscal plans for the Stability and Convergence Programmes have already been approved in the autumn forecasts.}

Therefore, the one year ahead fiscal forecasts in the autumn publications might have used these programs as strategic device.

(13)

13 specific constant eliminating unobserved heterogeneity, 𝜏𝑡is an annual time constant capturing common

time effects, and 𝜀𝑖,𝑡is the idiosyncratic error term.

𝑋𝑖 is a vector of control variables. It consists of 𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁𝑖,𝑡𝑡−1+𝑠, which captures the influence of

inflationary dynamics on the budget balance. As pointed out by Tujula and Wolswijk (2004), the overall effect of inflation on budget balances is not clear a priori. It may have an automatic effect through a nominal progression in tax receipts and governments may welcome inflation as it erodes the real value of outstanding nominal government debt. It might, however, also increase long-term interest rates and have a negative effect on investment and economic growth. 𝑋𝑖 also includes political variables

considered in the literature:6_{variables capturing the impact of election years (}_{𝐸𝐿𝐸𝐶}

𝑖,𝑡), ideological

orientation (𝐼𝐷𝐸𝑂𝐿𝑂𝐺𝑌𝑖,𝑡), and the number of checks and balances in the political system (𝐶𝐻𝐸𝐶𝐾𝑆𝑖,𝑡).

See Table A.2 in the appendix for a description of the political variables and its data sources.

3.2 Institutional factors: fiscal rules and government quality

The baseline model incorporates economic and political covariates, but does not consider institutional variables that might affect fiscal policy. To assess whether institutional factors constrain fiscal policymaking, I consider the following equation:

∆𝐶𝐴𝑃𝐵𝑖,𝑡𝑡−1+𝑠= 𝛼 + 𝛿𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠+ 𝛾𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠+ 𝜗𝑌𝑖,𝑡+ 𝜃𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝑌𝑖,𝑡+ 𝛽′𝑋𝑖+ 𝜇𝑖+ 𝜏𝑡+ 𝜀𝑖,𝑡, (2) for 𝑠 = 0, 2 | 𝑒𝑥 𝑝𝑜𝑠𝑡,

where 𝑌𝑖,𝑡is one of the institutional barriers – fiscal rules or institutional quality – and 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝑌𝑖,𝑡

captures the interaction between the output gap and the institutional variable. The impact of the institutional factors on the fiscal stance might, however, change when they are considered simultaneously. Therefore, I consider the following three-way interaction model:

(14)

14

∆𝐶𝐴𝑃𝐵𝑖,𝑡𝑡−1+𝑠= 𝛼 + 𝛿𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠+ 𝛾𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠+ 𝜗𝐹𝑅𝐼𝑖,𝑡+ 𝜑𝐼𝑄𝑖,𝑡+ 𝜃𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝐼𝑖,𝑡+ 𝜔𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐼𝑄𝑖,𝑡+ 𝜓𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝐼𝑖,𝑡∗ 𝐼𝑄𝑖,𝑡+ 𝛽′𝑋𝑖+ 𝜇𝑖+ 𝜏𝑡+ 𝜀𝑖,𝑡, for 𝑠 = 0, 2 | 𝑒𝑥 𝑝𝑜𝑠𝑡. (3)

To measure the effect of fiscal rules, I have constructed a stringency index (𝐹𝑅𝐼𝑖,𝑡) using information

from the fiscal rules database of the IMF. The database provides information on the use and design of national and supranational fiscal rules and covers all types of rules (budget balance rules, debt rules, expenditure rules, and revenue rules). Information is provided on key characteristics of these rules, such as their legal basis, monitoring mechanisms, and enforcement procedures. I have constructed the fiscal rules index following instructions of Schaechter et al. (2012).7_{I use information on the legal basis,}8

government coverage,9_{supporting procedures,}10_{enforcement mechanisms,}11_{and flexibility}12_{for each}

type of fiscal rule. All indicator terms are normalized to unity, such that the index ranges from 0 to 5.

To capture the quality of government administration, I use data from the Worldwide Governance Indicators 2018 project of The World Bank (Kaufmann et al., 2011). The institutional quality measure (𝐼𝑄𝑖,𝑡) captures “the 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 index takes on a value between -2.5 to 2.5, where a higher value of the index reflects more institutional quality.

Figure 1 shows the average value of the fiscal rules index and institutional quality measure. The average value for the institutional quality measure remains virtually unchanged and fluctuates around its average value of 1.18 over the sample period. The average value for the fiscal rules index gradually increases from 1.13 to 3.21 over the period 2000-2015. The fiscal rules index increased as the EC reinforced the

7_{In contrast to Schaechter et al. (2012), I also include information about the flexibility of the fiscal rule into measure. Recent}

literature shows that flexible fiscal rules are desirable for macro-economic stabilization purposes (Guerguil et al. 2017).

8_{Political commitment = 1, Coalition agreement = 2, Statutory rule = 3, International treaty = 4, Constitutional rule = 5.} 9_{No coverage = 0, Central government = 1, General government = 2. The number may be adjusted upward by 0.5 to account}

for similar rules applying to different levels.

10_{0-1 dummies for the presence of multi-year expenditure ceilings, a fiscal responsibility law and an independent fiscal body}

setting budget assumptions and monitoring the implementation of the budget.

11_{0-1 dummies for the presence of a formal enforcement procedure and whether there is monitoring body outside the}

government.

12_{0-1 dummies for the presence of well-specified escape clause, whether a balanced budget target is cyclically-adjusted, and}

(15)

15 SGP rules over the sample period, but it also reflects an improvement of national fiscal governance as several EU member states adopted new ‘national’ fiscal rules and strengthened their fiscal frameworks.

[Insert Figure 1 about here]

4. Results

4.1 Cyclical stance of fiscal policy

Table 1 shows the results of eq. (1) estimated using panel fixed effects (FE). The estimates in column (1) refer to the planning stage of fiscal policy; column (2) shows the results for the implementation stage and column (3) offers the findings for the ex-post fiscal outcomes. The coefficient of the lagged dependent variable is negative and significant in all columns, indicating high path dependence of the government budget balance. The lag of the public debt level measure shows a positive and significant result in all columns, suggesting that a high public debt level leads to more contractionary fiscal policies. The coefficient on inflation is positive and significant in column (1), but insignificant in the subsequent columns. This indicates that high inflation induces a more prudent construction of fiscal plans, which disappears in the actual fiscal policies. The coefficients of the political variables predominantly have a negative sign, but are insignificant. Only in the planning stage of fiscal policy the presence of checks and balances in the political system seems to significantly improve the government budget balance.

[Insert Table 1 about here]

(16)

16 be in line with Eyraud et al. (2017), but is in contrast to the majority of empirical studies that consider the cyclicality of fiscal policy in EU member countries (e.g,, Galí and Perotti, 2003; Wyplosz, 2006; Bénétrix and Lane, 2013). Furthermore, the pattern of cyclicality across the various stages of fiscal policy seems to be in line with the finding of Cimadomo (2016), i.e. that the cyclical stance of fiscal policies is estimated to be more ‘countercyclical’ when real-time data are used instead of ex post data.

4.2 Fiscal rules

Do fiscal rules matter? Table 2 shows the results for eq. (2) when fiscal rules are considered. Columns (1), (3), and (5) show the results excluding the interaction term. The coefficient for the output gap measure in all columns shows a similar pattern to the results of eq. (1) presented in Table 1. The coefficient of the fiscal rules index is positive in all columns, but only significant at the five-percent level for the ex-post fiscal outcomes. This suggests that fiscal rules successfully improve the government budget balance, but do not necessarily result into lower deficits or higher surpluses during real-time fiscal policy.

(17)

17 rules index. Instead, a simple figure can be used to illustrate the marginal effect of the output gap on the government budget balance and the corresponding standard errors conditional to the strength of fiscal rules.

Figure 2 shows that budgetary plans in EU countries are acyclical irrespective of the presence and strength of fiscal rules. However, results for later stages of fiscal policy are different. When fiscal rules are absent during the implementation phase of the government budget, the marginal effect of the output gap variable is negative and significant given a 95% confidence interval. That is, if the economy is operating one percentage point above potential the government budget will deteriorate with 0.56% of GDP. The marginal effect gradually diminishes as the fiscal rules index increases and after passing a minimum threshold (approximately 2.5) the significance of the output gap disappears, indicating that fiscal policy becomes acyclical. Similar results are found for the ex-post fiscal outcomes. In the absence of fiscal rules, a positive (negative) output gap of one percentage point leads to a deterioration (improvement) of the government budget equal to 0.42% of GDP. Fiscal rules reduce the procyclical reaction of fiscal policy and when the fiscal rules index takes on a value of 3 or higher, fiscal policy is acyclical.

4.3 Government quality

Does government quality play a role? Table 3 shows the results for eq. (2) for the quality of government administration. Similar to Table 2, columns (1), (3), and (5) show the results without the interaction term; columns (2), (4) and (6) show the results including the interaction term. All columns show a positive and significant result for the institutional quality measure. This suggests that countries with a high quality of government administration construct fiscal plans and implement fiscal policies with less deficits (higher surpluses). The interaction term with the output gap variable is positive and significant in the planning stage and actual fiscal policies, but is insignificant in the implementation stage.

(18)

18 Figure 3 shows the marginal effect of the output gap on the government budget conditional to government efficiency. The graphs show that fiscal policy is procyclical for countries with low institutional quality. In all phases of policymaking, a positive output gap equal to one percentage point results into a deterioration of the government budget of approximately 0.4% of GDP when the institutional quality measure is negative. The cyclical reaction of fiscal policy improves as institutional quality increases and adopts an acyclical stance after passing a minimum threshold. In the planning phase, fiscal policy even becomes countercyclical when the quality of government administration is considerably high. The conditional relationship between the output gap and the institutional quality measure weakens as fiscal policy moves from the planning phase to the eventual budgetary outcomes, but overall, fiscal policy seems to be highly dependent on the quality of government administration.

4.4 Fiscal rules and government quality

The cyclicality of fiscal policy conditional to fiscal rules and government quality displays a similar pattern (i.e. fiscal rules and government quality improve the cyclical behavior of fiscal policymakers). The results, however, might change when both institutional variables are considered simultaneously. Table 4 presents the results for eq. (3) without interaction terms. The estimation output is very similar to the results presented in Table 2 and 3 (i.e. fiscal rules and institutional quality significantly improve the government budget). Column (3) shows that the coefficients of the fiscal rules index and institutional quality measure are positive, but they slightly reduce compared to the estimation output of Table 2 and 3. Furthermore, the coefficients of both institutional variables remain only significant at the ten percent level. This hints on a substitutional effect of fiscal rules and institutional quality on the government budget.

(19)

19 and three constitutive terms, results are sometimes hard to interpret and cannot easily be displayed in a simple figure. For reasons of parsimony, I have constructed a 0-1 dummy variable (𝐹𝑅𝑖,𝑡) indicating

whether fiscal rules were weak (𝐹𝑅𝑖,𝑡= 0) or strong (𝐹𝑅𝑖,𝑡 = 1).13

Table 5 presents the estimation output for eq. (3) and Figure 4 shows the marginal effect plots. The results for the planning and implementation phases are very similar. When fiscal rules are weak, real-time fiscal policy is procyclical if institutional quality is low. When the quality of government administration increases, the procyclical stance of fiscal policy reduces, becomes acyclical after passing a minimum threshold, and even becomes countercyclical when quality is high. In countries with strong fiscal rules, real-time fiscal policy is acyclical when institutional quality is low. As institutional quality increases, fiscal policy becomes countercyclical.

The graphs for the ex-post fiscal outcomes show that fiscal policy is procyclical when fiscal rules are weak and institutional quality is low. That is, if the economy is operating one percentage point above (below) potential, this leads to a deterioration (improvement) of the government budget between 0.5% – 0.7% of GDP. Similar to earlier results, when governments become more effective, the cyclical stance of fiscal policy improves and becomes acyclical after the institutional quality measure passes a certain threshold (approximately 1.0). In countries with strong fiscal rules, the output gap has a marginally low impact on the government budget. A positive (negative) output gap of one percentage point leads to a deterioration (improvement) of the government budget of approximately 0.1% of GDP. Different to previous results, the quality of government administration does not really seem to reduce the procyclical stance of fiscal policy, but in spite of this, fiscal policy becomes acyclical when the institutional quality passes a minimum threshold (approximately 1.5). The marginal effect of the output gap seems to be insignificant for countries with low institutional quality as well, but the wider confidence intervals at the left-hand side of the graph can be explained by the fact that there are only a few observations with a low value for the institutional quality measure in case fiscal rules are strong.

(20)

20

[Insert Table 5 about here] [Insert Figure 4 about here]

Overall, the results suggest that both institutional factors have a significant impact on the fiscal policy stance in its various stages, but they provide no support for the hypothesis that the institutional factors are complementary in their effect on the reaction of fiscal policy to the business cycle. During real-time fiscal policy, the conditional effect of government efficiency does not seem to change for weak or strong fiscal rules. Furthermore, the results for ex-post fiscal outcomes indicate that institutional quality plays only a marginal role when fiscal rules are strong. This suggests that fiscal rules and the quality of government administration are more likely to be institutional substitutes in terms of improving the cyclical reaction of actual fiscal policies, what seems to be in contrast with previous studies (e.g., Bergman and Hutchison, 2015; Bova et al., 2018).

5. Robustness

5.1 Potential bias

So far, all the results are estimated using panel fixed effects, but with the inclusion of the lagged dependent variable the dynamic panel models contains a potential bias (Nickell, 1981). To control for this so-called ‘Nickell-bias’, a generalized methods of moments (GMM) estimator can be used. The widely used Arelano-Bond (1991) GMM and system GMM (Blundell and Bond, 1998) estimators require mean-stationarity of the variables. This assumption is unlikely to hold, considering that I use a panel consisting of macroeconomic variables which generally do not revert to a mean.14_{Therefore, I use}

the GMM estimator of Ahn and Schmidt (1995), which does not require mean stationarity of the variables.

14_{GMM is well suited for micro panels (large N, small T), but performs less with macro data (large N, large T). A violation of}

(21)

21 Tables A.3 and A.4 in the appendix present the results when eq. (2) is estimated using a GMM approach. The corresponding marginal plot figures (Figure A.1 and Figure A.2) can be found in the appendix. The results are similar to those reported in Tables 2 and 3, i.e. fiscal rules and institutional quality improve the cyclical fiscal stance. When the institutional variables are considered simultaneously, the results remain similar to a great extent. Table 6 shows the results for eq. (3) using the GMM approach and the corresponding marginal effects graph can be found in the appendix (Figure A.3). Compared to the FE results presented in Table 5 and Figure 4, the GMM results suggest on a more procyclical stance during the implementation of the budget, but the effect of fiscal rules and government quality does not change. Also similar to the results of Table 5, institutional quality does not really seem to have an additional stabilizing effect anymore when fiscal rules are strong. In contrast to earlier results, however, the marginal effect of the output gap remains significant for high levels of institutional quality. The economic significance of this effect, however, is rather low. When the economy is operating one percentage point above (below) potential, this seem to result in a deterioration (improvement) of the government budget close to 0.1% of GDP. Overall, the results do not differ much when the model is estimated using a GMM approach. Therefore, the results do not appear to suffer from the Nickell bias, such that panel fixed effects estimation of the model seems to be allowed.

5.2 Endogeneity of fiscal rules

(22)

22 use instruments that influence the adoption of fiscal rules based on the findings of Altunbas and Thornton (2017). I consider a dummy variable whether a country has joined the EMU, as adopting the supranational fiscal rules laid down in the SGP is a necessity for joining the monetary union. Furthermore, I consider an index for the development of financial markets and institutions, and a variable that captures the openness to trade as instruments. Deep credit markets and international trade make the government more vulnerable to domestic and foreign market developments, respectively, and might increase the probability that a country will adopt a fiscal rule. See Table A.2 in the appendix for a description of the instrumental variables and its data sources.

Table 7 show the results for eq. (2) using the IV approach. Figure A.4 in the appendix shows the marginal effects plot. The IV results are similar to the panel fixed effects results of Table 2. Fiscal rules do not seem to have a significant impact on policymakers in the planning phase, but fiscal rules do seem to reduce the procyclical stance in later stages of fiscal policy. Furthermore, when the strength of fiscal rules passes a minimum threshold, fiscal policy becomes acyclical. The under-identification tests (Kleibergen-Paap LM test of whether the equation is identified) reject the null hypothesis that the equation is under identified, which suggests that the instruments are ‘relevant’. The over-identification tests (Hansen test) cannot reject the null hypothesis that the instruments are ‘valid’ instruments, indicating that the instruments are uncorrelated with the error term. Only in column (2) – results for the planning stage with interaction effects – the tests whether the instruments are relevant and valid do not hold. In all cases, however, the endogeneity test does not reject the null hypothesis that the specified endogenous variable (i.e. the fiscal rules index) can actually be treated as exogenous, suggesting that panel fixes effects estimation is allowed.

5.3 Different types of fiscal rules

(23)

23 expenditure and balanced budget rules reduce the procyclicality bias of fiscal policy more effectively (Nerlich and Reuter, 2015; Combes et al., 2017; Bergman and Hutchison, 2018). To capture whether the different types of fiscal rules have a non-uniform effect on fiscal policy in its various stages, I construct indexes for each type of fiscal rule – i.e. expenditure rules, revenue rules, deficit rules, and debt rules – similar to the aggregated fiscal rules index.

Tables 8, 9 and 10 present the results for eq. (2) using the disaggregated indexes for respectively the planning phase, implementation phase, and the eventual fiscal outcomes. The corresponding marginal effect figures can be found in the appendix (Figure A.5-A.8). In general, debt rules and to a lesser extent balanced budget rules seem to significantly improve the government budget, whilst the general effect of expenditure and revenue rules on the government budget balance is negligible in all stages of policymaking. Similar to the results for the aggregated fiscal rules index, neither of the different types of fiscal rules seem to influence the cyclical reaction of fiscal policy in the planning stage (i.e. fiscal policy plans are acyclical irrespective of fiscal rules). In the implementation phase of fiscal policy, expenditure and debt rules (and to a lesser extent balanced budget rules) seem to significantly improve the cyclical fiscal stance.

[Insert Tables 8-10 about here]

(24)

24 is very strong (i.e. when the debt rules index takes on a value of 4.25 or higher), whereas balanced budget and revenue rules only seem to alleviate procyclicality.

These results, however, have to be interpreted with some caution. The conditional effect of the disaggregated indexes is sometimes measured with a lot of uncertainty, which could partly be explained by the fact that for some of the indexes the bulk of the observations is located in a small interval. For instance, only a few EU countries did have a revenue rule in place during the 1999-2015 period. In contrast, because of the SGP all EU member countries had considerably stringent balanced budget and debt rules over the sample period. The skewness of the disaggregated indexes therefore partly explains the wider confidence intervals for the conditional effect of the output gap at, respectively, higher and lower values of the disaggregated indexes. Especially in the examination of revenue rules the lack of observations forms a problem. Hence, it is far too early to conclude that revenue rules only marginally affect fiscal policy stabilization, as they might be an effective tool to constrain the use of windfall revenues for (procyclical) government expenditures.

6. Conclusions

EU countries experienced a severe economic downturn during the global financial crisis and the subsequent European sovereign debt crisis. Several countries showed that they did not save enough for rainy days, whilst fiscal policy as economic stabilizer is of the utmost importance for member countries of the EU. To shed more light upon this matter, this thesis studies the cyclical reaction of fiscal policy in 27 EU member states using a real-time data approach. The results indicate that ex-post fiscal policy is on average procyclical in EU countries. However, planned fiscal policy shows an acyclical stance, whereas the results for the real-time implementation of the budget give only a weak indication of procyclicality.

(25)

25 expenditure rules – improve the cyclical stance. Furthermore, strong institutional variables result into acyclical fiscal outcomes once passing a minimum threshold and even induce countercyclicality in fiscal policy plans and their implementation. The results considering real-time fiscal policy do not show a differentiated effect of government administration when fiscal rules are either weak or strong. Moreover, instead of being complementary, these factors seem to be institutional substitutes for improving the cyclical stance in the actual fiscal policies.

Whilst governments did not improve their institutional quality over the sample period, the reinforcement of national and supranational fiscal governance is reassuring for the stabilization function of national fiscal policies in the EU. Nevertheless, the results also indicate that fiscal rules are yet not capable of inducing countercyclical fiscal policy. A greater role for fiscal rules in the preparation of budgetary plans seems desirable, as neither of all types of fiscal rules appear to influence the cyclical stance in the planning phase of fiscal policy. Furthermore, the literature is far from conclusive to answer the question of why fiscal policy is often procyclical and more attention should be paid on the determinants of fiscal procyclicality.

(26)

26 through focusing on the numerical constraints laid down in fiscal rules. Eventually, this could provide more insight to whether the global trend of optimizing the design of fiscal frameworks to enforce compliance is justified or whether more attention should go to setting the optimal fiscal targets.

References

Ahn, S.C. and P. Schmidt, (1995). Efficient estimation of models for dynamic panel data. Journal of Econometrics 68: 5-27.

Alesina, A., F.R. Campante and G. Tabellini, (2008). Why is fiscal policy often procyclical? Journal of the European Economic Association 6: 1006-1036.

Altunsbas, Y. and J. Thornton, (2017). Why do countries adopt fiscal rules? The Manchester School 85: 65-87. 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: 277-297.

Armingeon, K., V. Wenger, F. Wiedemeier, C. Isler, L. Knöpfel, D. Weisstanner and S. Engler, (2018). Comparative Political Data Set 1960-2016. Bern: Institute of Political Science, University of Berne.

Asatryan, Z., C. Castellón and T. Stratmann, (2018). Balanced budget rules and fiscal outcomes: Evidence from historical constitutions. Journal of Public Economics 167, 105–119.

Avellan, L. and G. Vuletin, (2015). Fiscal procyclicality and output forecast errors. Journal of International Money and Finance 55: 193-204.

Azzimonti, M., M. Battaglini and S. Coate, (2016). The costs and benefits of balanced budget rules: Lessons from a political economy model of fiscal policy. Journal of Public Economics 136, 45–61.

Beetsma, R., X. Debrun, X. Fang, Y. Kim, V. Lledó, S. Mbaye and X. Zang, (2019). Independent fiscal councils: Recent trends and performance. European Journal of Political Economy 57: 53-69.

Beetsma, R. and M. Giuliodori, (2010). Fiscal adjustment to cyclical developments in the OECD: an empirical analysis based on real-time data. Oxford Economic Papers 62: 419-441.

Beetmsa, R., M. Giuliodori and P. Wierts, (2009). Planning to cheat: EU fiscal policy in real time. Economic Policy 31: 753-804.

Bénétrix, A.S. and P.R. Lane, (2013). Fiscal cyclicality and EMU. Journal of International Money and Finance 34: 164-176.

Bergman, M. and M. Hutchison, (2015). Economic stabilization in the post-crisis world: Are fiscal rules the answer? Journal of International Money and Finance 52: 82-101.

Bergman, M. and M. Hutchison, (2018). Fiscal procyclicality in developing economies: the role of fiscal rules, institutions and economic conditions. Mimeo, University of Copenhagen / University of California, Santa Cruz. Blundell, R.W and S.R. Bond, (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics 87: 115-143.

Bova, E., N. Carcenac and M. Guerguil, (2014). Fiscal rules and the procyclicality of fiscal policy in the developing world. IMF working paper 14/122.

Bova, E., P. Medas and T. Poghosyan, (2018). Macroeconomic stability in resource-rich countries: the role of fiscal policy. Journal of Banking and Financial Economics 1(9): 103-122.

(27)

27

Bun, M. and V. Sarafidis, (2013). Dynamic panel data models. UvA-Econometrics Working Papers 13/01, University of Amsterdam.

Calderón, C., R. Duncan and K. Schmidt-Hebbel, (2016). Do good institutions promote countercyclical macroeconomic policies? Oxford Bulletin of Economics and Statistics 78: 650-670.

Caselli, F. and J. Reynaud, (2019). Do fiscal rules cause better fiscal balances? A new instrumental variable strategy. IMF Working Paper 19/49.

Cimadomo, J., (2016). Real-time data and fiscal policy analysis: a survey of the literature. Journal of Economic Surveys 30: 302-326.

Combes, J., A. Minea and M. Sow, (2017). Is ﬁscal policy always counter- (pro-) cyclical? The role of public debt and ﬁscal rules. Economic Modelling 35: 138-146.

Croushore, D., (2011). Frontiers of real-time data analysis. Journal of Economic Literature 49: 72-100.

Debrun, X., L. Moulin, A. Turrini, J. Ayuso-i-Casals and M.S. Kuma, (2008). Tied to the mast? The role of national fiscal rules in the European Union. Economic Policy 23: 298-362.

Debrun, X. and L. Jonung, (2019). Under threat: rules-based fiscal policy and how to preserve it. European Journal of Political Economy 57: 142-157.

Eyraud, L., V. Gaspar and T. Poghosyan, (2017). Fiscal politics in the euro area. In Gaspar, V. et al, Fiscal Politics. Washington, DC: International Monetary Fund.

Fatás, A. and I. Mihov, (2011). Fiscal policy as a stabilization tool. CEPR Discussion Papers No. 8749. Feldstein, M.S., (2009). Rethinking the role of fiscal policy. NBER Working Paper No. 14684.

Frankel, J.A. and J. Schreger, (2013). Over-optimistic official forecasts in the Eurozone and fiscal rules. Review of World Economics 149: 247-272.

Frankel, J.A. and J. Schreger, (2016). Bias in official fiscal forecasts: can private forecasts help? NBER Working Paper No. 22349.

Frankel, J.A., C.A. Végh and G. Vuletin, (2013). On graduation from fiscal procyclicality. Journal of Development Economics 100: 32-47.

Galí, J. and R. Perotti, (2003). Fiscal policy and monetary integration in Europe. Economic Policy 18: 533-572. Gavin, M. and R. Perotti, (1997). Fiscal policy in Latin America. In Bernanke, B. and J. Rotemberg, NBER Macroeconomics Annual 1997, Cambridge, MA: MIT Press.

Gilbert, N. and J. de Jong, (2017). Do European fiscal rules induce a bias in fiscal forecasts? Evidence from the Stability and Growth Pact. Public Choice 170: 1-32.

Golinelli, R. and S. Momigliano, (2009). The cyclical reaction of fiscal policies in the euro area: the role of modelling choices and data vintages. Fiscal Studies 30: 39-72.

Gootjes, B., J. de Haan and R.M. Jong-A-Pin, (2019). Do fiscal rules constrain political budget cycles? DNB Working Paper No. 634.

Guerguil, M., P. Mandon and R. Tapsoba, (2017). Flexible fiscal rules and countercyclical fiscal policy. Journal of Macroeconomics 52: 189-220.

Heinemann, F., M. Moessinger and M. Yeter, (2018). Do fiscal rules constrain fiscal policy? A meta-regression analysis. European Journal of Political Economy 51: 69-92.

Holm-Hadulla, F., S. Hauptmeier and P. Rother, (2012). The impact of numerical expenditure rules on budgetary discipline over the cycle. Applied Economics 44: 3287-3296.

Ilzezki, E. and C.A. Végh, (2008). Procyclical fiscal policy in developing countries: truth or fiction? NBER Working Paper No. 14191.

(28)

28

Jong, J. de and N. Gilbert, (2018). Fiscal discipline in EMU? Testing the effectiveness of the Excessive Deficit Procedure. DNB Working Paper No. 607.

Kaminsky, G.L., C.M. Reinhart and C.A. Végh, (2004). When it rains, it pours: procyclical capital flows and macroeconomic policies. In Gertler, M. and K. Rogoﬀ (eds.) NBER Macroeconomic Annual 2004, Cambridge, MA: MIT Press.

Kaufmann, D., A. Kraay and M. Mastruzzi, (2011). The Worldwide Governance Indicators: methodology and analytical issues. Hague Journal of the Rule of Law 3: 220-246.

Milesi-Ferretti, G.M, (2004). Good, bad or ugly? On the effects of fiscal rules with creative accounting. Journal of Public Economics 88: 377-394.

Musgrave, R., (1959). The theory of public finance. New York, McGraw Hill.

Nerlich, C. and W.H. Reuter, (2015). Fiscal rules, fiscal space and procyclical fiscal policy. ECB working paper No. 1872.

Nickell, S., (1981). Biases in dynamic models with fixed effects. Econometrica 49: 1417-1426.

Reuter, W.H., (2015). National numerical fiscal rules: Not complied with, but still effective? European Journal of Political Economy 39: 67-81.

Reuter, W.H., (2019). When and why do countries break their national fiscal rules? European Journal of Political Economy 57: 125-141.

Schaechter, A., T. Kinda, N. Budina and A. Weber, (2012). Fiscal Rules in Response to the Crisis—Toward the “Next-Generation” Rules. A New Dataset. IMF Working Paper 12/187.

Talvi, E. and C.A. Végh. Tax base variability and procyclical fiscal policy in developing countries. Journal of Development Economics 78: 156-190.

Tujula, M. and G. Wolswijk, (2004). What Determines Fiscal Balances? An Empirical Investigation in Determinants of Changes in OECD Budget Balances. ECB Working Paper No. 422.

Tornell, A. and P.R. Lane, (1998). Are windfalls a curse? A nonrepresentative agent model of the current account. Journal of International Economics 44: 83-112.

Tornell, A. and P.R. Lane, (1999). The Voracity Effect. American Economic Review 89: 22-46.

Woo, J., (2009). Why do more polarized countries run more procyclical fiscal policy? The Review of Economics and Statistics 91: 850-870.

(29)

29

Tables

Table 1 Fixed Effects (FE) results baseline specification eq. (1)

(1) (2) (3)

Dependent variable: ∆𝐶𝐴𝑃𝐵𝑖,𝑡𝑡−1+𝑠 Planning; t|t-1 Implementation; t|t+1 Outcomes; t|ex post

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.218** -0.475*** -0.510*** (0.0813) (0.0612) (0.0497) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 0.00827 -0.125* -0.192*** (0.0387) (0.0677) (0.0428) 𝐷𝐸𝐵𝑇𝑖,𝑡−1𝑡−1+𝑠 0.0150*** 0.0321*** 0.0315*** (0.00489) (0.00747) (0.00556) 𝐼𝑁𝐹𝐿𝐴𝑇𝐼𝑂𝑁𝑖,𝑡𝑡−1+𝑠 0.146** 0.103 0.0258 (0.0645) (0.122) (0.0488) 𝐸𝐿𝐸𝐶𝑖,𝑡 -0.204 0.0328 -0.244 (0.186) (0.214) (0.260) 𝐶𝐻𝐸𝐶𝐾𝑆𝑖,𝑡 0.159** -0.140 -0.141 (0.0755) (0.184) (0.168) 𝐼𝐷𝐸𝑂𝐿𝑂𝐺𝑌𝑖,𝑡 0.00162 -0.00172 -0.00230 (0.00105) (0.00264) (0.00294) 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 -1.299*** 0.390 -0.619 (0.416) (1.079) (0.777) Observations 325 348 401 R-squared 0.287 0.370 0.393 Number of countries 27 27 27

Note: Fixed effects (FE) estimation of eq. (1) with robust standard errors shown in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Column (1) provides the results for the planning phase of fiscal policy; column (2) for the implementation phase; column (3) for the ex-post fiscal outcomes. Results for the time dummies are not displayed for reasons of parsimony, but are available upon request.

Table 2 Fixed Effects (FE) results eq. (2) with fiscal rules

(1) (2) (3) (4) (5) (6)

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.222** -0.222** -0.482*** -0.480*** -0.526*** -0.524*** (0.0826) (0.0830) (0.0611) (0.0557) (0.0478) (0.0447) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 0.00859 -0.0746 -0.126* -0.563* -0.200*** -0.417*** (0.0393) (0.135) (0.0708) (0.280) (0.0452) (0.126) 𝐹𝑅𝐼𝑖,𝑡 0.368 0.403* 0.750 0.845* 0.707** 0.730** (0.225) (0.225) (0.507) (0.489) (0.319) (0.324) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝐼𝑖,𝑡 0.0330 0.178 0.103* (0.0482) (0.112) (0.0503) Observations 325 325 348 348 401 401 R-squared 0.295 0.297 0.376 0.386 0.405 0.413 Number of countries 27 27 27 27 27 27

See notes Table 1. Results for the time dummies and control variables of vector 𝑋𝑖are not displayed for reasons of parsimony, but are available

(30)

30

Table 3 Fixed Effects (FE) results eq. (2) with institutional quality

(1) (2) (3) (4) (5) (6)

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.227*** -0.260*** -0.491*** -0.515*** -0.532*** -0.551*** (0.0815) (0.0801) (0.0640) (0.0612) (0.0543) (0.0528) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 -0.00835 -0.236** -0.167** -0.333** -0.227*** -0.347*** (0.0405) (0.0888) (0.0655) (0.142) (0.0416) (0.0608) 𝐼𝑄𝑖,𝑡 0.601* 0.945*** 2.196** 2.286** 1.919** 1.693** (0.352) (0.339) (0.874) (0.890) (0.860) (0.814) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐼𝑄𝑖,𝑡 0.221*** 0.204 0.155** (0.0613) (0.129) (0.0682) Observations 325 325 348 348 401 401 R-squared 0.293 0.351 0.382 0.393 0.404 0.416 Number of countries 27 27 27 27 27 27

upon request.

Table 4 Fixed Effects (FE) results eq. (3) with fiscal rules and institutional quality

(1) (2) (3)

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.233*** -0.499*** -0.541*** (0.0826) (0.0637) (0.0513) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 -0.00944 -0.169** -0.226*** (0.0408) (0.0688) (0.0437) 𝐹𝑅𝐼𝑖,𝑡 0.391* 0.798 0.580* (0.219) (0.494) (0.334) 𝐼𝑄𝑖,𝑡 0.653* 2.267** 1.530* (0.343) (0.863) (0.822) Observations 325 348 401 R-squared 0.302 0.388 0.412 Number of countries 27 27 27

(31)

31

Table 5 Fixed Effects (FE) results eq. (3) with fiscal rules and institutional quality – three-way

interactions

(1) (2) (3)

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.272*** -0.519*** -0.551*** (0.0845) (0.0568) (0.0434) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 -0.234** -0.630*** -0.503*** (0.0895) (0.205) (0.162) 𝐹𝑅𝑖,𝑡 0.474 1.247 1.094* (0.351) (0.797) (0.587) 𝐼𝑄𝑖,𝑡 1.115*** 2.170** 1.575** (0.372) (0.793) (0.672) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝑖,𝑡 0.0197 0.571*** 0.397*** (0.119) (0.188) (0.133) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐼𝑄𝑖,𝑡 0.244** 0.437 0.294 (0.0981) (0.294) (0.216) 𝐹𝑅𝑖,𝑡∗ 𝐼𝑄𝑖,𝑡 -0.347 -0.444 -0.274 (0.270) (0.511) (0.553) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝑖,𝑡∗ 𝐼𝑄𝑖,𝑡 -0.0486 -0.437 -0.296 (0.116) (0.279) (0.212) Observations 325 348 401 R-squared 0.358 0.417 0.434 Number of countries 27 27 27

upon request.

Table 6 Generalized Methods of Moments (GMM) results eq. (3) with fiscal rules and

institutional quality – three-way interactions

(1) (2) (3)

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.274*** -0.399*** -0.460*** (0.0747) (0.122) (0.0737) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 -0.227*** -0.584*** -0.480*** (0.0870) (0.191) (0.141) 𝐹𝑅𝑖,𝑡 0.343 0.443 1.042** (0.451) (0.767) (0.510) 𝐼𝑄𝑖,𝑡 1.302** 3.581** 1.290* (0.603) (1.441) (0.662) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝑖,𝑡 -0.00856 0.451** 0.418*** (0.115) (0.188) (0.130) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐼𝑄𝑖,𝑡 0.249*** 0.398 0.283 (0.0948) (0.281) (0.188) 𝐹𝑅𝑖,𝑡∗ 𝐼𝑄𝑖,𝑡 -0.313 -0.380 -0.272 (0.265) (0.361) (0.456) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝑖,𝑡∗ 𝐼𝑄𝑖,𝑡 -0.0319 -0.286 -0.314 (0.113) (0.272) (0.192) Observations 325 348 401 Number of countries 27 27 27

See notes Table 1. Generalized Methods of Moments (GMM) estimation of eq. (3) with robust standard errors shown in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Results for the time dummies and control variables of vector 𝑋𝑖are not displayed for reasons of parsimony, but

are available upon request.

(32)

32

Table 7 Instrumental Variable (IV) results eq. (2) with fiscal rules

(1) (2) (3) (4) (5) (6)

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.240*** -0.246*** -0.482*** -0.490*** -0.536*** -0.526*** (0.0749) (0.0749) (0.0842) (0.0818) (0.0746) (0.0755) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 -0.0106 -0.0931 -0.126 -2.221** -0.205*** -0.820*** (0.0381) (0.499) (0.0815) (1.094) (0.0474) (0.301) 𝐹𝑅𝐼𝑖,𝑡 0.423 0.828 0.844 2.923 1.127** 1.023 (0.492) (0.550) (1.154) (1.927) (0.571) (0.654) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝐼𝑖,𝑡 0.0321 0.854* 0.294** (0.196) (0.442) (0.143) Kleibergen-Paap stat. (p) 0.000 0.246 0.000 0.039 0.000 0.000 Hansen-J stat. (p) 0.530 0.003 0.635 0.323 0.722 0.469 Endogeneity test (p) 0.852 0.368 0.912 0.678 0.561 0.915 Observations 325 325 348 348 401 401 R-squared 0.267 0.252 0.376 0.209 0.401 0.383 Number of countries 27 27 27 27 27 27

See notes Table 1. Instrumental Variable (IV) estimation of eq. (2) with robust standard errors shown in parentheses: *** p<0.01, ** p<0.05, * p<0.1. Robust standard errors shown in parentheses: *** p<0.01, ** p<0.05, * p<0.1; In columns (1), (3), and (5), the fiscal rules index is the endogenous regressors. In columns (2), (4), and (6) the endogenous regressors are both the constitutive term of the fiscal rules index as the interaction with the output gap variable; The Kleibergen-Paap LM statistic shows the result for an under-identification test whether the instruments are relevant. Under the null hypothesis the equation is under-identified; The Hansen-J statistic shows the result for the overidentification test of the restrictions. Under the null hypothesis the instruments are valid instruments; The endogeneity test checks the endogenous regressors. Under the null hypothesis the specified endogenous regressors can actually be treated as exogenous. Results for the time dummies and control variables of vector 𝑋𝑖are not displayed for reasons of parsimony, but are available upon request.

Table 8 Fixed Effects (FE) results eq. (2) with differentiated fiscal rules – Planning phase

(1) (2) (3) (4) (5) (6) (7) (8)

Dependent variable: ∆𝐶𝐴𝑃𝐵𝑖,𝑡𝑡−1+𝑠 Expenditure rules Expenditure rules Revenue rules Revenue rules Budget rules Budget rules Debt rules Debt rules

𝐶𝐴𝑃𝐵𝑖,𝑡−1𝑡−1+𝑠 -0.219** -0.217** -0.218** -0.219** -0.219** -0.220** -0.221*** -0.221** (0.0823) (0.0832) (0.0821) (0.0827) (0.0814) (0.0821) (0.0795) (0.0800) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠 0.00886 -0.0144 0.00827 0.00174 0.00697 -0.263 0.00397 -0.0664 (0.0389) (0.0617) (0.0388) (0.0429) (0.0400) (0.307) (0.0408) (0.245) 𝐹𝑅𝐼𝑖,𝑡 0.0700 0.0877 -0.00206 0.0230 0.523 0.599* 0.833** 0.855** (0.0800) (0.0798) (0.132) (0.138) (0.330) (0.317) (0.374) (0.357) 𝐺𝐴𝑃𝑖,𝑡𝑡−1+𝑠∗ 𝐹𝑅𝐼𝑖,𝑡 0.0106 0.0143 0.0752 0.0212 (0.0229) (0.0195) (0.0828) (0.0734) Observations 325 325 325 325 325 325 325 325 R-squared 0.289 0.290 0.287 0.288 0.295 0.298 0.306 0.307 Number of countries 27 27 27 27 27 27 27 27