Bulletin of Economic Research

1. INTRODUCTION

Following the global crisis of 2008–09, many governments in developed countries put in place fiscal stimuli measures aimed at promoting the recovery from a protracted housing bust (Castro, 2010; Agnello and Schuknecht, 2011; Agnello and Sousa, 2013a; Agnello et al., 2015). The subsequent concerns about long‐term sustainability and spikes in government bond yields ultimately translated into the sovereign debt crisis. In this context, fiscal consolidation programmes were designed and restrictive fiscal packages were implemented.

In Europe, the conduct of fiscal policy has been guided by the assumption that fiscal consolidation is a necessary condition for sustainable growth over the medium to long‐term. In other countries – namely, the United Kingdom and the United States –, governments have recognized that there is a (negative) trade‐off between fiscal austerity and economic growth, at least, in the short‐term. For instance, Agnello et al. (2013) show that government spending cuts can bring public debt back to a sustainable path, but fiscal discipline also plays a key role as it helps shorten the duration of fiscal consolidations. Mallick and Granville (2005) also find that poverty reduction can only be temporarily achieved via fiscal adjustments.

Countries under financial stress and struggling to restore their public finances via consolidation measures appear to have no alternative but to pay the price of creating a recessionary environment (at least, in the short term). On the other hand, countries that managed to avoid fiscal austerity seem to be weathering better in terms of economic performance. This raises a number of questions on the impact of fiscal adjustments on business cycle synchronization. Specifically, do fiscal consolidation programmes lead to decoupling? What are the effects of fiscal stimulus episodes on the synchronization of business cycles? Are fiscal adjustments implemented by a single country more likely to result in de‐synchronized business cycles (i.e., to reduce the correlation of the cyclical component of economic activity across countries) than programmes that are simultaneously adopted by different countries?

Experience says that economies can be subject to shocks that are asymmetric: for instance, the macroeconomic impact of large depreciations is different from small depreciations; and large wage adjustments can affect real economic activity more substantially than small movements. Moreover, it is also well‐known that the impact of common shocks can be asymmetric: for example, positive commodity price shocks can favour large net commodity exporters at the expense of large net commodity importers.

Under these conditions, if asymmetries are partially offset by fiscal adjustments, then fiscal consolidation/stimuli measures can lead to a higher degree of co‐movement of business cycles across countries. In contrast, if fiscal policy is pro‐cyclical (Gavin and Perotti, 1997; Lane, 2003) and changes in the institutional or the political background make sizeable fiscal adjustments unavoidable in a given country, then consolidation/stimulus programmes can be associated with a fall in the correlation of the cyclical component of real output between any pair of countries.

Thus, the relationship between fiscal adjustments and business cycle synchronization cannot be determined ex‐ante. In the current work, we explore empirically this link with a suitable econometric framework.

We start by considering the identification of fiscal consolidation programmes of Devries et al. (2011), who rely on a narrative approach. Using quarterly data for a panel of industrialized countries, we show that fiscal consolidations implemented by a single country (i.e., unilateral fiscal consolidation programmes) do not significantly impact on business cycle synchronization or can even reduce it. In contrast, fiscal consolidation measures that are adopted by both countries (i.e., synchronized fiscal consolidation programmes) make their business cycles behave more closely.

Then, we use the statistical approach of Alesina and Ardagna (2010) to identify fiscal stimulus episodes. We find similar evidence of a positive and significant impact on business cycle synchronization from the bilateral implementation of fiscal stimuli. However, unilateral fiscal stimulus programmes do not tend to exert such effect.

Further, the results provide some evidence of decoupling when an inflation targeting regime is adopted unilaterally, but business cycle synchronization improves when both countries move towards this monetary policy regime. In addition, there is an increase in business cycle synchronization when both countries fix their exchange rates and when countries become members of a monetary union. Finally, we uncover a positive effect of bilateral trade on business cycle synchronization.

The main policy implication of our work is that a synchronization of fiscal consolidation measures coupled with an inflation targeting regime can induce stronger co‐movement of business cycles across countries. The reason is that when monetary policy is conducted with the goal of achieving price stability over the medium to long‐term, central banks respond to foreign output shocks by stabilizing prices and allowing domestic output to adjust. Thus, business cycle synchronization tends to rise. If governments also react to foreign output shocks by implementing fiscal consolidation measures that overlap in time across countries, this will further increase the co‐movement of their business cycles.

The paper is organised as follows. Section II. reviews the related literature. Section III. describes the econometric methodology and presents the data. Section IV. provides the empirical results. Section V. concludes.

II. LITERATURE REVIEW

Our analysis is related to various other studies on the determinants of business cycle synchronization. In particular, Artis and Zhang (1997, 1999) and Artis (2004) detect an ‘European business cycle’ since the early eighties. Devereux and Engel (2001, 2003) find that floating exchange rates protect the domestic economy from foreign monetary shocks (the so‐called ‘insulation’ effect). Baxter and Kouparitsas (2005) conclude that the degree of bilateral trade between a pair of countries has a robust effect on business cycle synchronization. Kose et al. (2003, 2012) and Imbs (2006) emphasize that countries with closer financial ties tend to have more synchronized business cycles. Darvas and Szapáry (2000) do not find a significant impact of the exchange rate regime on capital flows across countries. Darvas et al. (2005) highlight that the business cycles of countries with similar government budget positions move more closely. Thus, fiscal divergence i.e., deviations from a fiscally responsible behaviour, creates idiosyncratic shocks and is associated with lower business cycles synchronization.

Darvas and Szapáry (2008) examine business cycle synchronization in the new EU members of Central and Eastern Europe and the euro zone countries. They show that, despite the dramatic improvement in the correlation of the cyclical components of GDP, industrial production and exports among the new EU members, the degree of synchronization of private consumption with the euro zone countries remained low. Artis (2008) shows that as the process of international trade deepens, regional business cycle affiliations are superseded by wider business cycle clubs. Araújo and Oliveira Martins (2009a, 2009b) show that the deep fall in economic activity during the so‐called ‘Great Recession’ largely reflected the ‘Great Synchronization’ of trade flow declines across countries.

More recently, Flood and Rose (2010) argue that business cycles in countries targeting inflation are only slightly synchronized with foreign ones. Artis et al. (2011) study the behaviour of the international business cycle across advanced and emerging market economies and find evidence of a secular increase in international business cycle synchronization within a group of European and a group of English‐speaking economies.

We contribute to this literature by assessing the impact of fiscal adjustments on business cycle synchronization, an issue that, to the best of our knowledge, has not been investigated thoroughly yet. Arghyrou (2000) finds that the accession to the European Monetary Union (EMU) has led to important changes in the trade patterns, even though it did not seem to have a significant effect on prices (Arghyrou, 2007). Fatás and Mihov (2003) show that fiscal policy discretion leads to macroeconomic instability. Arghyrou and Luintel (2007) develop a new empirical approach and find that the Maastricht criteria had a positive effect on the government's intertemporal budget constraint of a series of EMU countries. Along the same lines, Afonso et al. (2011) provide a new methodology to investigate long‐term fiscal developments, which accounts for the persistence, the cyclicality and the discretion of fiscal adjustments. Castro (2011) highlights the importance of fiscal rules for the economic growth of the European Union. Afonso and Jalles (2012) use different analytical approaches and a policy‐action framework to assess the success of fiscal consolidations, and emphasize the importance of the cyclically adjusted primary balance and the duration of the programmes. Agnello and Sousa (2013b) show that fiscal prudence is key for economic prosperity and Agnello and Sousa (2014) find that political instability is associated with more discretion in fiscal policy.

III. EMPIRICAL METHODOLOGY

In order to explore the empirical relationship between business cycle synchronization and fiscal adjustments, we estimate the following model using a pooled Ordinary‐Least Squares (OLS) estimator:

where is the sample correlation coefficient (Pearson's coefficient) between real output in country i and real output in j over 5‐year non‐overlapping windows.

Since we are interested in the business cycle deviations from the trend, we consider two alternative techniques for de‐trending real output (GDP), namely: (i) the Baxter‐King band‐pass filter; and (ii) the fourth difference, i.e., the annual growth rate implied by the quarterly data. Data on the seasonally adjusted GDP are obtained from the International Financial Statistics (IFS) and the World Economic Outlook (WEO) of the International Monetary Fund (IMF) and the Organization for Economic Co‐Operation and Development (OECD).

Flood and Rose (2010) show that inflation targeting has a positive (albeit small) impact on business cycle synchronization. Thus, we test whether the advent of inflation targeting (IT) provides some sort of insulation from external shocks. In this context, IT(1) and IT(2) are dummy variables that take the value of one if one or both countries are inflation targeters and zero, otherwise. Information about inflation targeting regimes is based on the identification put forward by Mishkin (2008).

Mundell (1968) shows that under a fixed exchange rate regime, monetary shocks increase the synchronization of business cycles across countries, but the effect of real shocks cannot be determined a priori. In the case of a flexible exchange rate, real shocks have a positive impact on business cycle synchronization, but monetary shocks have an ambiguous effect. To assess the impact of fixed exchange rate regimes on business cycle synchronization, we add Fix(1) and Fix(2) to the set of explanatory variables. These are dummy variables that take the value of one if one or both countries have a fixed exchange rate regime and zero, otherwise.

Similarly, we check whether membership of a currency union affects the degree of synchronization of business cycles. Thus, we consider MU(1) and MU(2) among the set of regressors. These are dummy variables that take the value of one if one or both countries are members of a monetary union.

Both fixed exchange rate regimes and monetary unions are measured in accordance with the work of Reinhart and Rogoff (2004). It should be noted that Fix(2) and MU(2) are equal to one if both countries have a fixed exchange rate vis‐à‐vis each other and if both countries are in the same currency union, respectively. This implies that, for countries that are in a currency union but do not share a common currency, MU(2) takes the value of zero. Flood and Rose (2010) show that countries fixing exchange rates against each other have more synchronized business cycles and the same applies in the case of monetary unions, albeit to a smaller extent.

In addition, we include Cons(1) and Cons(2) in the set of explanatory variables. These are also dummy variables that take the value of one if one (unilateral adjustment) or both countries (synchronized adjustment) have implemented a fiscal consolidation programme for m consecutive years within the 5‐year non‐overlapping window. We expect synchronized fiscal consolidations to have a positive impact on the co‐movement of business cycles across countries, as ‘coordinated’/’simultaneous’ fiscal adjustments tend to generate effects on economic growth of the same direction. Thus, they are likely to increase business cycle synchronization. In contrast, the impact of unilateral fiscal consolidation episodes is undetermined.

Finally, is a vector of control variables, which includes: 1) the degree of bilateral trade between countries i and j; 2) the log distance between countries i and j; 3) the Chicago Board Options Exchange (CBOE) Volatility Index (VIX); and 4) the U.S. term spread.

According to Baxter and Kouparitsas (2005), the degree of bilateral trade between a pair of countries is an important determinant of business cycle synchronization, with effects that are above and beyond those of the impact of the monetary regime. Thus, we consider their measure of bilateral trade among the set of control variables. In particular, we compute the sum of all four bilateral trade flows (i.e., exports from country i to country j, imports into country j from country i, exports from country j to country i, and imports into country i from country j), divided by the multilateral sums (i.e., country i’s exports, country j’s exports, country i’s imports and country j’s imports). Data on exports and exports is sourced from the IMF's Direction of Trade dataset.

Along the same lines, we follow Baxter and Kouparitsas (2005) and Flood and Rose (2010) and include the natural logarithm of the distance between a pair of countries in the vector of control variables. Countries‧ locations in latitude and longitude are sourced from the CIA's World Factbook location. We expect distance to be negatively correlated with the degree of business cycle synchronization.

The VIX is typically considered a proxy for risk aversion and uncertainty and a good way to track the dynamics of the integration of countries in the world economy due to, for instance, the co‐movement implied by the global financial cycle (Rey, 2015). An increase in risk aversion and uncertainty should be associated with growth decoupling across countries.

As for the U.S. term spread, it captures expectations about future growth, as well as the recent unconventional monetary policies that have been put in place in recent times. We expect a tightening of (nonstandard) monetary policies to have a negative effect on business cycle synchronization.

We also account for the effects of fiscal stimuli on business cycle synchronization. To that end, the dummy variables Cons(1) and Cons(2) are replaced with the dummy variables Stim(1) and Stim(2), which take the value of one if one or both countries have adopted fiscal stimuli packages for m consecutive years within the 5‐year non‐overlapping window.

Fiscal consolidation (stimulus) episodes are identified using data for government spending and government revenue from the OECD's Economic Outlook. As in the case of fiscal consolidations, synchronized fiscal stimuli should lead to more co‐movement of business cycles, but the effect of unilateral fiscal stimuli episodes should be ambiguous.

Therefore, we run the following panel‐data regression using a pooled Ordinary‐Least Squares (OLS) estimator:

We identify fiscal consolidation episodes using a narrative approach based on the work of Devries et al. (2011). As emphasized by these authors, the standard statistical approach, which builds on variation in the cyclically‐adjusted primary budget balance (CAPB), suffers from three major limitations. First, the procedure to adjust the budget balance for the effect of the business cycle is subject to uncertainty. Second, the measurement error of the CAPB may be correlated with economic developments. Finally, the statistical approach omits periods of fiscal consolidation that were followed by negative shocks and offsetting discretionary fiscal policies.

Thus, Devries et al. (2011) analyse policy measures that are motivated by deficit reduction by examining accounts and records of what governments were planning to do at the time of publications (such as the IMF Recent Economic Developments reports, the IMF Staff Reports, or the OECD Economic Surveys). In this context, the narrative approach eliminates the endogeneity of the response of fiscal policy to the economy.

Yet, one disadvantage of the database compiled by Devries et al. (2011) is that it does not distinguish between episodes of fiscal consolidation and of fiscal stimulus respectively. For this reason, we identify fiscal stimulus programmes using a statistical approach based on the work of Alesina and Ardagna (2010), and thus correct the primary surplus for year‐to‐year changes in the unemployment rate. More specifically, a fiscal stimulus episode corresponds to a year in which the cyclically‐adjusted primary balance deteriorates by at least 1.5 percent of GDP.

Table 1 provides a summary of the descriptive statistics of the variables. The correlation between the two dependent variables is large (0.7221) and significant at the 1 percent level. This implies that our measure of business cycle synchronization is not sensitive to the de‐trending process. In the case of the dummy variables, their means are typically larger for ‘unilateral’ events than for ‘bilateral’ events that apply to each pair of countries. Among other control variables, the dispersion of distance and VIX is particularly high.

In Table 2, we provide the list of episodes of fiscal consolidation identified using the narrative approach of Devries et al. (2011) and the list of episodes of fiscal stimulus identified using the statistical approach of Alesina and Ardagna (2010). Given that the two types of episodes are identified using different approaches, there is some degree of overlapping between fiscal consolidations and fiscal stimuli even though it is very small. Additionally, it can be seen that there is also some synchronization between the two types of episodes across countries.

IV. EMPIRICAL RESULTS