Empirica DOI 10.1007/s10663-014-9275-x ORIGINAL PAPER The transmission of foreign shocks to South Eastern European economies Goran Petrevski • Jane Bogoev • Dragan Tevdovski Ó Springer Science+Business Media New York 2014 Abstract This paper investigates the transmission of foreign shocks to economic activity and macroeconomic policies in three South Eastern European (SEE) economies: Croatia, Macedonia and Bulgaria. Specifically, we provide empirical evidence on the influence of several policy and non-policy shocks (euro-zone output gap, money market rate and inflation) on economic activity as well as monetary and fiscal policies in the three countries. The main motivation behind our empirical investigation is the fact that all of these economies are small open economies with rigid exchange rate regimes and different degree of integration within the European Union (EU). As for the methodological issues, we employ recursive vector autoregressions to identify the exogenous shocks in the euro-area. Generally, the estimated results imply that euro-zone economic activity has significant and relatively strong influence on these economies where foreign output shocks are transmitted relatively quickly. The results also suggest that the effects of foreign shocks are of larger magnitude in the countries that are more integrated with the EU. An additional finding is that positive foreign interest rate shocks trigger a contractionary response of domestic monetary policy notwithstanding the fact that domestic money While writing the paper Jane Bogoev was with the National Bank of the Republic of Macedonia. G. Petrevski (&)
D. Tevdovski Faculty of Economics, Saints Cyril and Methodius University, Goce Delchev Blvd. 9B, 1000 Skopje, Macedonia e-mail: goran@eccf.ukim.edu.mk D. Tevdovski e-mail: dragan@eccf.ukim.edu.mk J. Bogoev Kuzman Josifovski Pitu 1, 1000 Skopje, Macedonia e-mail: janebogoev@yahoo.com J. Bogoev 1818 H street, Washington, DC 20433, USA 123 Empirica market rates are not linked with euro-zone interest rates. Finally, euro-zone inflation is instantly transmitted to domestic inflation. We can explain these effects by several factors, such as: the fixed exchange rates, the relatively high trade integration of SEE economies within the euro-zone as well as the dependence of SEE banks on foreign financing. Keywords Monetary policy
Vector autoregression
Exogenous shocks JEL Classification C3
E52
E58
E61 1 Introduction Since the very beginning of the transformation process, former communist countries have declared their aspiration towards joining the EU. In this respect, they have engaged in massive institutional and economic reforms, which have facilitated them to converge gradually to the ‘‘old’’ EU members. As a result, first Bulgaria and Romania, and later on Croatia succeeded in achieving the goal of EU accession, while Macedonia has been given the candidate status. Looking forward, SEE countries are faced with another challenge related to the EU accession process, namely, that of adopting the single currency. In these regards, it is important to examine how euro-area shocks are transmitted to SEE countries, i.e. what is the response of domestic macroeconomic variables to these shocks. Therefore, the transmission of EU macroeconomic and policy shocks represents a relevant research issue for policymakers in the former transition economies from SEE, such as: Croatia, Macedonia, and Bulgaria. Our paper aims to examine the following issues: What is the reaction of fiscal and monetary policy in SEE countries to a shock in economic activity within the EU? What is the reaction of fiscal and monetary policy in SEE countries to a shock in EU inflation rate? Do monetary policies in SEE countries follow the monetary policy implemented by the European Central Bank (ECB)? In addition, we seek to find out whether there are any significant differences in the impact of foreign shocks on domestic macroeconomic variables depending on the level of EU integration. In principle, the answers to the above questions seem obvious due to the small open economy features of SEE countries as well as their fixed exchange rate regimes. Under these circumstances, foreign supply and demand shocks are expected to be transmitted quickly and fully to these economies as predicted by standard textbook open economy macroeconomics models. At the same, it is known that fixed exchange rate regimes (especially under perfect capital mobility) impose severe limitations to the ability of policy makers to offset exogenous shocks. Yet, there are several factors that might cause the effects of foreign shocks as well as the response of domestic macroeconomic policies to differ from one country to another or even to diverge from those suggested by textbook models. For instance, a country might be exposed predominantly by idiosyncratic domestic shocks, which could be asymmetric and thus, blur the response of 123 Empirica domestic macroeconomic variables to foreign shocks (for empirical evidence on CEE economies, see Fidrmuc and Korhonen 2003; Benczur et al. 2004). Also, macroeconomic variables in individual countries might show different response to a common foreign shock due to various factors, such as: the differences in economic structure, high import dependence, export diversification, output and consumption composition, production factors mobility, labor market flexibility, and credibility of domestic policies. All of the aforementioned factors might be one of the reasons for asymmetric shocks or SEE policy makers might pursue macroeconomic policies that diverge from those in the euro-zone (Jiménez-Rodriguez et al. 2010; Velickovski 2013 provide empirical evidence for CEE and SEE countries, respectively). In addition, segmented financial systems might weaken the transmission of foreign interest rate shocks to domestic money market and bank lending rates, which might explain why the interest rate pass-through could be slow and incomplete (Petrevski and Bogoev 2012 provide empirical evidence for SEE countries). Under imperfect capital mobility, the ‘‘impossible trinity’’ might not hold, implying that countries with fixed exchange rates might retain some autonomy in conducting monetary policy. For example, the implementation of various noninterest rate tools by some SEE central banks (Croatia is the most notable example) might provide some maneuver room in the conduct of monetary policy. Therefore, the questions raised above are worth of being explored. In these regards, the paper provides empirical evidence on the effects of several euro-zone macroeconomic and policy shocks (output gap, inflation, and interest rates) on several macroeconomic and policy variables in the SEE countries (output, inflation, interest rates and budget surpluses) based on the impulse response functions estimated with recursive VARs. In order to assess how exogenous factors affect macroeconomic performances and policy variables in SEE economies we impose the so-called block exogeneity restrictions regarding the interrelationship between the variables included in the VAR. The main findings from our study can be summarized as follows. First, notwithstanding the countercyclical fiscal policy, euro-zone output shocks have strong effects on economic activity and inflation, reflecting both the trade and capital channel. Second, positive foreign inflation shocks lead to higher domestic inflation, which in turn triggers a contractionary behavior of monetary policy makers, aimed at curbing the adverse effects on inflation expectations. Third, an increase in foreign interest rate leads to a decline in domestic output, despite the fact that it is not transmitted into domestic money market rates. Fourth, we find some cross-country differences in the response of macroeconomic variables. For instance, the effects of output shocks on output and inflation are larger in Bulgaria and Croatia, and Euribor shocks affect domestic output only on these two countries. Therefore, it seems that the level of integration with the EU matters for the transmission of foreign shocks. As can be seen, some of our findings support standard textbook predictions that foreign output and inflation shocks have strong effects on domestic output and inflation in small open economies. One can explain these findings by the combined effects of rigid exchange rate regimes, high trade openness, capital flows as well as the dominant presence of foreign banks in the banking systems of SEE economies. 123 Empirica For instance, high trade openness makes these countries vulnerable to fluctuations in foreign demand so that euro-zone business cycles are quickly transmitted through the trade channel. However, it is not only trade openness that matters in the transmission of foreign shocks, but financial integration plays important role, too. Specifically, due to the dominant role of foreign banks in SEE banking systems, domestic banks are highly dependent on foreign financing. Hence, when their parent banks are faced with an increase (decline) in lending activity associated with economic expansion (contraction) in the euro-zone, this fuels (reduces) lending activity of SEE banks, exacerbating the transmission of foreign output shock. Moreover, these finding have strong implications for the effectiveness of macroeconomic policies in SEE countries in the sense that both inflation and domestic output fluctuations are mostly driven by foreign shocks. Therefore, under fixed exchange rate regimes, domestic macroeconomic policies are largely ineffective tools for macroeconomic management and policy makers need to accommodate most of the external shocks. In these regards, it is interesting to notice that, despite the countercyclical reaction of fiscal authorities, they are not able to offset the effects of foreign output shocks on domestic output and inflation. This finding implies that in small open economies with fixed exchange rates fiscal policy may not be effective instrument as suggested by the Mundell–Fleming model. On the other hand, we find some deviations in the response of domestic macroeconomic variables to foreign shocks from those predicted by standard small open economy models. For instance, our results suggest that domestic interest rates are not linked with foreign interest rates, thus, suggesting that the ‘‘impossible trilemma’’ might not work, i.e. there may be some limited room for autonomous monetary policy even in the presence of rigid exchange rate regimes. The rest of the paper is organized as follows. Section 2 provides a brief overview of the empirical literature on the transmission of foreign shocks and the response of domestic policies. The data description and the estimation methods are presented in Sects. 3 and 4, respectively. The findings of the empirical study are presented in the Sect. 5. 2 An overview of the empirical literature on the transmission of foreign shocks The effects of foreign real and monetary shocks and the response of domestic economic policies have been discussed extensively in the open economy macroeconomics literature (For instance, see Agénor and Montiel 1996; Hossain and Chowdhury 1996; Jha 1994; Karakitsos 1992; Krugman 1988; Obstfeld and Rogoff 1996; Prachowny 1984; Shone 1989). The empirical literature focusing on the effects of foreign shocks on former transition economies has emerged only recently, reflecting the accession process towards the EU and the prospective membership in the euro area. In these regards, one strand of the empirical literature deals with the issue of business cycle synchronization between the new member states and the euro-zone (for instance, see Frenkel and Nickel 2005; Fidrmuc and Korhonen 2003; Korhonen 2003; 123 Empirica Fidrmuc and Korhonen 2006; Benčı́k 2011). These studies usually rely on the Blanchard and Quah (1989) methodology by employing a small-scale Structural VAR (SVAR) model that decomposes supply and demand shocks. The main findings of these studies are that only the most advanced CEE economies show high degree of business cycle synchronization with the euro-zone, while for the rest of these countries the shocks are idiosyncratic. As for SEE economies, the empirical evidence suggests that business cycle synchronization vis-à-vis the euroarea is low, the convergence of supply and demand shocks is slow or absent and the synchronization with the euro-zone business cycle depends on the degree of integration with the EU (Velickovski 2010, 2013; Petrovska 2012). Another strand of the empirical literature focuses on the transmission mechanism of foreign shocks. Hence, it can be regarded as complementary to the previous one in the sense that it is not only shock asymmetry that matters, but the response of domestic macroeconomic variable to foreign shocks is important, too. Here, some papers investigate only the effects of specific foreign shocks, such as fiscal policy shocks (Crespo-Cuaresma et al. 2011) or interest rate shocks (see Velickovski 2010, 2013; Minea and Rault 2011; Petrevski and Bogoev 2012 for papers covering SEE economies). Others take a broader perspective and deal with various supply, demand and policy shocks (for instance, see Benczur et al. 2004; Eickmeier and Breitung 2006; Horváth and Rusnák 2009; Jiménez-Rodriguez et al. 2010; Krznar and Kunovac 2010; Unevska-Andonova and Petkovska 2012). The usual approach within this strand of literature is to employ some variant of small-scale SVAR models, relying on the impulse response functions and/or variance decomposition in order to trace the dynamic response of domestic macroeconomic variables as well as the relative importance of foreign shocks. Eickmeier and Breitung (2006) represent a notable exception since they work within a large-dimensional structural factor framework. The main findings of these studies may be summarized as follows: foreign price shocks have strong effects on domestic inflation; foreign shocks have strong effects on domestic output; while the results on the link between foreign and domestic interest rates are weak. Our paper adds to the second strand of the empirical literature by exploring the effects of various foreign shocks on domestic macroeconomic variables in SEE countries, including the response of fiscal policy, which is something that is often omitted in most of the papers. Having in mind that most of the SEE economies have rigid exchange rate regimes (pegs, currency boards and official dollarization), which impose serious constraints on the use of monetary policy as a stabilization tool it is important to trace the response of fiscal policy to foreign shocks. Further on, from a methodological point of view, we impose more appropriate restrictions in our empirical methodology. Specifically, alongside contemporaneous restrictions as usually implemented in VAR-based papers, we impose the so-called block exogeneity restrictions regarding the interrelationship between the variables included in the VAR, which imply that not only current but also all past values of domestic variables do not affect euro-zone variables. In this way, we are able to conduct our empirical investigation following assumptions that are both economically appealing and closer to the definition of exogenous shocks. 123 Empirica 3 Data description For the empirical investigation we use quarterly data from the first quarter of 1999 to the fourth quarter of 2011. More precisely, for Bulgaria the data set starts from the first quarter of 1999 and we do not use previous data because of the highly unstable macroeconomic environment prevailing in late 1990s as well as the change in the policy regime (the introduction of currency board in 1997). For Macedonia, the sample starts in the first quarter of 2000 due to the change in the main monetary policy instrument that occurred in the beginning of 2000. For Croatia, the sample starts from the second quarter of 2000 for two reasons: first, we wish to avoid the effects of the banking crisis from 1998 to 1999, and second, the money market rate data is available only from the second quarter of 2000. The variables used in the empirical research include: primary cyclically adjusted government balance (as a ratio of GDP), money market interest rate (for Croatia and Macedonia), M0-to-GDP ratio for Bulgaria, quarterly annualized inflation rate and output gap. In addition, the output gap in EMU is included as an indicator of foreign economic activity; euro-zone money market rate (the 3-month Euribor) is included as a foreign reference interest rate and euro-zone inflation as a foreign inflation. We have done a seasonal adjustment by using the ‘‘CENSUS X-12’’ method of some of the data series, such as: real GDP, the consumer price index (CPI) and the M0 aggregate used for Bulgaria. Inflation rate is based on the CPI data. The output gap is calculated as a percentage difference between the actual and potential GDP. In estimating potential GDP andoutput gap we use one of the most commonly used statistical methods in the empirical literature, i.e. Hodrick–Prescott (HP) filter with the default lambda of 1,600 (k = 1,600). In addition, we have conducted the following unit root tests in order to check the stationarity of the data series: Augmented Dickey–Fuller (ADF) with various lag length selection criteria (Akaike, Scwartz and Hannan-Quin), Phillips–Perron (PP), and Kwiatkowski–Phillips–Schmidt–Shin (KPSS) tests. The unit root tests reveal that money market rates in Croatia and Macedonia, and the M0-to-GDP ratio in Bulgaria are all non-stationary, while the rest of the series are stationary.1 After performing the same unit-root tests on the first differences of these variables we obtain stationarity. Consequently, we conclude that the money market rates for Croatia and Macedonia, and the M0-to-GDP ratio are I(1) variables. Therefore, we proceed by working with the first differences of these variables. Primary cyclically-adjusted government balance is employed as an indicator of fiscal policy stance. For Bulgaria, we use general government balance as a broader measure of fiscal policy stance, whereas for Croatia and Macedonia we take the central government balance because quarterly data for the general budget balance are not available for such a long time period. In that respect, we argue that using the central government budget balance is a relatively good proxy for general 1 The results from the unit root tests are available from the authors upon request. 123 Empirica government policy stance.2Primary government balance is calculated as difference between revenues and primary expenditures, i.e. interest payments are subtracted from total expenditures. The rationale for that, according to Mackiewicz (2008), is that interest payments represent an exogenous category. Ultimately, in designing the current fiscal policy and the size of expenditures, fiscal authorities cannot influence the size of interest payments and they take them as an exogenous factor, which is determined by the past fiscal policy decisions related to public borrowing (Angelovska-Bezoska et al. 2011). For consistency, the data related to fiscal revenues and expenditures, throughout the whole sample period are adjusted according the Governmental Financial Statistics (GFS) 2001 methodology set by the International Monetary Fund (IMF). The cyclical adjustment of the budget balances was done according to the aggregated approach advocated by IMF, OECD, European Commission and the World Bank Group.3 Domestic money market interest rates are used as indicators of monetary policy in Croatia and Macedonia. In spite of the fixed exchange rate regime, we believe that there is a room for autonomous monetary policy in these two countries due to the following reasons: first, the interest rate parity holds only under perfect capital mobility where domestic and foreign assets are perfect substitutes. Obviously, these assumptions are too strong for Croatia and Macedonia; second, both Macedonian and Croatian central banks have relied on a series of non-interest rate policy tools, thus, being able to affect domestic money market rates. Certainly, the above mentioned interpretation cannot be valid for Bulgaria where, due to the features of the currency board and the full capital account liberalization, the central bank is not capable of conducting active monetary policy through the conventional tools like the interest rates (Minea and Rault 2011). Yet, due to the excess coverage with foreign reserves the Bulgarian central bank may have very limited space for maneuver by relying on some other tools, such as the reserve requirement. In that respect, we use the M0-to-GDP ratio as some kind of a monetary policy indicator (though an imperfect one, admittedly). M0 is composed of currency in circulation plus banks’ reserves (required reserves and excess reserves). We decided to use this indicator because the Bulgarian National Bank may have some influence on banks’ reserves through the reserve requirement. Nevertheless, it is true that banks’ reserves contain endogenous component, even when the required reserve ratio remains unchanged they may vary according with the changes in deposit volume. We are aware of the weakness of this monetary policy indicator, but we take it as a second best alternative. Consequently, in interpreting the results of the analysis based on this indicator, some of the conclusions stated should be taken with caution. 2 This argument is based on estimating correlation coefficients between central and general government balances for a broader sample of 31 countries that are member states of the European Union (EU) including the aggregated data for EU and the Euro-zone (EMU). The correlation coefficients based on annual data set indicated that there is a highly positive association (estimated higher than 0.9), in almost all countries in the sample with the exceptions of Latvia, Poland and Sweden where the level of correlation was estimated a bit lower around 0.8. All these correlation coefficients are statistically significant at 1 % level. 3 Cyclical adjustment of the budget revenues and expenditures was done according to the output gap movements with imposed elasticity coefficients for budget revenues of 1 and for budget expenditures of 0. 123 Empirica 4 Estimation method The estimation method applied in this research is based on recursive vector autoregression (VAR) models, which have become the main econometric tool for assessment of the effects of monetary and fiscal policy shocks (Enders 2010; Lutkepohl and Kratzig 2004; Stock and Watson 2001). The rationale for using recursive VAR models is as follows. First, these models are seen as most appropriate choices when the model consists of endogenous variables and the possible two way causation among the variables. Second, they enable us to estimate impulse response functions that indicate the interrelations and the transmission mechanism of the imposed shocks in each equation to the rest of the variables in the model. Third, these methods allow for including various restrictions about the contemporaneous impact of the variables in the model. Within the VAR methodology, several popular approaches have emerged recently, such as the factor-augmented VAR (FAVAR) and the global VAR (GVAR). Though they offer substantial advantages over recursive VARs we have decided not to employ them for the following reasons: our study does not contain a large number of variables for which it is useful to be summarized by a small set of factors using the principal component analysis within the FAVAR approach. Also, as pointed by Belvisio and Milani (2006), one drawback of FAVAR is the fact that the factors can be difficult to interpret economically, which is important when VAR models are used for drawing policy implications. On the other hand, the GVAR approach has proven to be very useful in the analysis of the interactions of global macroeconomic data networks where both the cross-section and the time dimensions are large (Chudic and Pesaran 2014). This is beyond our research topic, since we are interested in modeling the effects of foreign shocks in three individual SEE countries, where there is no feedback from macroeconomic variables in SEE economies to the ones from the euro-zone. The dependent variables in our VAR model are: yf, if, pf, yd, Fd, id and pd. The variables containing the superscript f are the foreign variables, while the variables with the superscript d are domestic variables. The variables: yf, if and pf represent the output gap, money market rate (the 3-month Euribor) and inflation rate in the euro-zone, respectively. The variables yd, Fd, id and pd indicate the output gap, fiscal policy variable, the money market rate and inflation in the domestic economy, respectively. We estimate the VARs separately for the three economies. The structure of the recursive VAR implies the following restrictions: (a) foreign variables (output gap, money market rate and inflation in the euro-zone) have contemporaneous impact on each of the variables in the three sample countries while the opposite relationship is precluded; (b) economic activity (output gap) in the eurozone and in the three analysed countries contemporaneously influence the policy variables (fiscal and monetary policy), while the policy variables do not have a contemporaneous impact on economic activity because they affect the ‘real’ sector with a certain time lag (see Blanchard and Quah 1989). The last restriction implies that economic activity variable and policy variables in both the euro-zone and domestic economies of the three countries have contemporaneous impact on inflation while inflation does not have contemporaneous feedback on these variables. 123 Empirica We build our model on the assumption that euro-area shocks affect exogenously the sample economies because all the three economies are relatively small as measured by their GDP and GDP per capita as well as their share in the euro-zone trade. Due to the small open economy assumption, we assume that SEE economies cannot influence economic developments and economic policy in the euro-zone. In addition, one may list several possible explanations as to why foreign shocks are expected to have strong effects on SEE economies: first, they are quite open in the sense that exports plus imports combine more than 100 % of their respective GDP; second, their foreign trade is highly integrated with the euro-area since more than 40 % of their trade is withthe euro-zone member countries; third, their banking sectors are also closely connected with the euro-zone banking system because majority of the foreign-owned bankcapital originates from the EMU economies. As a result, the banks in these countries are dependent on foreign financing from the EMU financial markets (Bogoev 2011); fourth, these economies import many products for final consumption from the EMU countries that may have direct impact on their inflation, while the percentage share of their exports in the total imports of the EMU economies is almost negligible. Following the approach of Cushman and Zha (1997), we impose the blockexogeneity restriction in the model such that in the basline unrestricted VAR specification, the lags of foreign variables are included in the equations of domestic variables, while the lags of domestic variables are excluded from the equations offoreign variables. These restrictions imply that the first block of foreign variables (euro-zone output gap, money market rate and inflation) is exogenous to the model whereas the lags of domestic variables do not enter in their equations (they are restricted to zero). After explaining the estimation methods used and the restrictions included in the VAR models, we now briefly explain our estimation strategy: 1. 2. 3. 4. We first specify unrestricted VAR model in order to determine the optimum number of lags of the variables. Here, we select the most parsimonious model due to the relatively limited number of observations compared to the number of variables included. The selection of the lag length is done on various lag length selection criteria, such as: Akaike (AIC), Schwarts (SIC), Hannan-Quinn (H-Q), sequential modified likelihood ratio test statistic (LR) and Final prediction error (FPE). After specifying the maximum number of lags, in the cases where more than one lag is suggested, then, due to the limited number of observations relative to the number of variables used, we do a subset model selection by dropping those lags of the variables of the unrestricted VAR that may improve the criterion value. In doing this we employ the so-called ‘‘top-down’’ procedure in selecting the number of lags in each individual equation in the VAR (for more details see Lutkepohl et al. 2006). We estimate the unrestricted VAR model by the feasible generalized least squares (FGLS) estimator; In order to explore whether the unrestricted VAR model is correctly specified and stable, we also conduct residual-based diagnostic tests, such as: 123 Empirica 5. 6. Portmanteau and Breusch–Godfrey LM tests for autocorrelation, Jarque–Bera Normality test, and Autoregressive conditional heteroskedasticity test (ARCHLM). If the selected model by steps 1 and 2 satisfies the residual-based diagnostic tests, we proceed further with them. Otherwise, we re-specify the unrestricted VAR by reducing or increasing the number of lags until the residual-based diagnostic tests provide satisfactory results. We tests for the stability of the estimated coefficients of the unrestricted VAR by employing several structural break tests for unknown breakpoint: the cumulative sum of the recursive residuals (CUSUM) and the squared cumulative sum of the recursive residuals (CUSUM SQ), suggested by Brown et al. (1975). If we find no structural breakpoint then we proceed by estimating the recursive and SVARs by employing maximum likelihood (ML) estimator with scoring algorithm (Amisano and Giannini 1997). 5 Model selection and discussion of the results In this section we present the model selection of the unrestricted VAR and then we continue by explaining the estimated results (mainly the impulse response functions—IRFs) for each country separately. 5.1 Model specification As already explained in Sect. 4, we have selected the unrestricted VAR model for each of the three sample economies according to several lag length selection criteria. We also checked whether the specified lag length satisfies the residual diagnostic tests, which suggested non-rejection of the null hypotheses of no serial correlation, normal distribution of the residuals and homoskedastic error terms at least at 5 % level of significance.4 Having determined the number of lags used in the unrestricted VAR, we proceed with estimating the unrestricted VAR and conducting the structural stability tests: CUSUM and CUSUMSQ. The structural stability test results for Croatia and Macedonia indicated that there has not been any structural break during the sample period for any of the variables included. For the case of Bulgaria the test results also indicated to the same conclusion for almost all of the variables used, with the exception of the money market rate for which mixed results are obtained. Namely, the CUSUM test suggests no structural break while the CUSUMSQ test indicates one structural break in the beginning of 2008. Because the results are mixed between the two methods and there is no a priori reason why we would expect a structural break of the money market rate in the period suggested, we proceed by estimating the recursive VARs as there is not structural break. 4 The results from the lag length selection criteria, residual-based diagnostic tests and stability tests are available from the authors upon request. 123 Empirica 5.2 Discussion of the estimated IRFsfrom recusrsive VARs In this section we interpret the cumulative impulse responses from recursive VARs (see ‘‘Appendix’’). In assessing the IRFs we calculate the 95 % confidence bands of Efron (Efron and Tibshirani 1993) and Hall (1992), estimated with bootstrap method of 100 replications. In what follows, we discuss the impulse responses generated by shocks in foreign variables (the output gap, Euribor, and inflation in the euro-zone, respectively). In contrast to many empirical studies, which do not address the statistical significance of the obtained impulse responses and discuss the policy implications even in the case of insignificant impulse responses, we only pay attention to those that are statistically significant by the two 95 % confidence bands mentioned above. The interpretation of the results is done according to the shocks originating from each of the exogenous variables, respectively. 5.2.1 Foreign output shock As presented in Appendix 1, a positive shock to the euro-area output gap leads to a rise in economic activity in the three SEE economies with a magnitude between 1.6 and 2 % points. Our results confirm that there is some degree of synchronization of business cycles between the euro-area and SEE economies, reflecting the importance of trade channel in the transmission of foreign shocks. This finding is quite expected given the high trade openness of SEE economies as well as the EU as the main market for their exports. As a result, an economic expansion in the eurozone leads to an increase in the import demand and thus, higher exports and output in SEE economies. The response of economic activity to foreign output shocks is larger and more persistent in Croatia and Bulgaria as compared to Macedonia, which has not yet started the negotiation process for joining the EU. The heterogeneous magnitude of the effects of foreign output shocks in the three SEE economies cannot solely be explained by the trade channel, since all of them have high trade openness. In other words, our results underline the importance of the capital channel, too, in the transmission of foreign shocks. In these regards, economic expansion in the euro-zone may be associated with higher exports of capital (foreign direct investments for example), which is mainly attracted by those countries that have closer political and economic links with the EU. Therefore, it seems that the degree of business cycle synchronization may indeed depend on the level of integration with the EU as usually found in the empirical literature (Jiménez-Rodriguez et al. 2010; Velickovski 2010, 2013; Petrovska 2012). At the same time, the euro-areaoutput shock leads to a temporary increase in domestic inflation in the three SEE economies by fuelling demand-side pressures. More precisely, as shown above, the euro-zone expansion spurs economic activity, leading to higher domestic demand in the three SEE economies, which puts an upward pressure on domestic inflation. The response of domestic inflation to the euro-zone output shock is very strong, ranging from 3 to 5 % points. Once again, we find that the effects of foreign output shocks are larger in Bulgaria and Croatia as compared to Macedonia, which may be related by the closer integration of these economies with the EU, probably through the capital channel. 123 Empirica Further on, we find that a positive output shock in the euro-area leads to an improved fiscal policy stance in Croatia and Macedonia. This might reflect the attempts of policymakers to use the favourable external environment in order to improve their fiscal position. In fact, as rigid exchange rate regimes impose limitations on the ability to pursue countercyclical monetary policy, it may be desirable for fiscal authorities to offset the spillover effects from the EU-wide expansion (positive aggregate demand shocks). In addition, the countercyclical fiscal policy increases the credit rating of the countries and the confidence in the fiscal policy makers for maintaining fiscal sustainability, which enables them to borrow from Euro-zone markets when needed. This result is in line with the models that advocate for the use of countercyclical discretionary fiscal policy at individual country level as a response to asymmetric EU shocks (Beetsma and Jensen 2002; Bryson 1994; Galı́ and Monacelli 2008). 5.2.2 Foreign interest-rate shock A positive shock in the foreign interest rate (a rise in the 3-month Euribor) exerts negative effects on domestic economic activity in Bulgaria and Croatia, but not in Macedonia (Appendix 2). In the former countries, a positive shock in the Euribor affects negatively economic activity with a magnitude from 1.5 to 2 % points. These findings can be explained by both the trade and the capital channel. As for the former, a rise in the Euribor contracts aggregate demand and export demand in the euro-zone, causing a fall of the output in SEE economies. In addition, rising Euribor leads to higher costs for the private sector (banks and companies) from external financing (through loans from financial institutions, intercompany loans etc.). More precisely, when foreign interest rate increases, companies’ costs of financing investment activities increase, too. This, in turn, lowers the rate of return on new investment, leading to a decline in investment expenditures with negative effects on domestic economic activity. Similarly, due to high foreign ownership in the SEE banking sector, a foreign interest rate shock increases the costs of domestic banks from borrowing abroad, which may lead to a contraction in bank lending with adverse effects on domestic output (for more details see Bogoev 2011). In assessing foreign interest rate shocks, we are especially interested in the response of domestic monetary policy variables. Here, we find that a foreign interest rate shock triggers an increase in domestic money market rates only in Macedonia with a magnitude of 0.8 % points. In Croatia, the initial response of domestic market rate is quite oscillating, though a positive Euribor shock leads to a temporary decline in domestic money market rate. These results for Croatia are quite puzzling since they contradict conventional wisdom that under rigid exchange rate regimes domestic interest rates should be closely tied to foreign interest rates, thus, imposing severe constraints on the ability for conducting autonomous monetary policy. Therefore, this result can be interpreted as evidence that the Croatian central bank, which has implemented a series of non-interest rate measures throughout the sample period (see Kraft 2003; Lang and Krznar 2004), may have preserved some autonomy in the conduct of monetary policy. Moreover, we find that a foreign interest rate shock does not have significant effects on monetary conditions in 123 Empirica Bulgaria (M0-to GDP ratio), too, despite the operation of the currency board arrangement. This finding is in line with the results obtained by Minea and Rault (2011), who find that Bulgarian interest rates and money supply are not linked with ECB’s interest rates in the short-run. This may suggest that the M0 aggregate was influenced mainly by foreign currency flows to Bulgaria, driven by the economic and political developments in the country and its EU accession process, but not as a result of the changes in the Euribor rate. 5.2.3 Foreign inflation shock As shown in Appendix 3, a shock inforeign inflation has a strong but temporary effect on domestic inflation in SEE economies with a magnitude of around 5 % points in Bulgaria, and around 2 % points in Croatia and in Macedonia. This finding implies that domestic inflation in the SEE economies is closely linked with eurozone inflation, which is quite expected for small economies with rigid exchange rate regimes and heavy import dependence from the euro-zone. Specifically, higher foreign inflation is instantly transmitted to domestic inflation via two channels: directly, via tradable goods, whose prices expressed in domestic currency are linked to prices on the international markets); and indirectly, via intermediary goods, which cause second-round inflationary effects. In addition, Appendix 3 reveals that a foreign inflation shock triggers a reaction by monetary policy in SEE countries, while it does not affect fiscal policy behavior and domestic economic activity. In these regards, a positive shock in foreign inflation leads to an increase in domestic money market rates in Macedonia and a decline in the M0-to-GDP ratio in Bulgaria. In contrast, the money market rate in Croatia declines in response to higher euro-area inflation, although this reaction is barely statistically significant. As shown above, the response of monetary policy in Macedonia to a shock of foreign inflation is consistent with the link between domestic and foreign money market rates. Specifically, when euro-zone inflation intensifies, the ECB reacts by increasing its key policy rate, which affects the euroarea money market rates. In turn, the rise in the Euribor is transmitted on the Macedonian money market rate with a certain delay. Moreover, rising foreign inflation may also induce a reaction of domestic monetary policy in order to prevent the second-round effects and thus, to curb inflation expectations. In Bulgaria, the decline in the M0-to-GDP ratio can be explained within the standard money demand theory, i.e. higher foreign inflation increases the risk of rising domestic inflation so that economic agents reduce their cash balances. The decline in the money market rate in Croatia in a response to higher euro-area inflation is consistent with the response of Croatian money market rate to a shock in the Euribor. These two results can be seen as evidence that Croatian money market rate does not follow Euribor, which might be a consequence of the massive use of various non-interest rate measures by the Croatian central bank. Finally, the lack of any response of fiscal policy in SEE economies to foreign inflation shock reflects the different preferences of policy makers, i.e. while central banks are mainly concerned with inflation shocks, fiscal authorities respond mainly to real economic activity, i.e. foreign output shocks. 123 Empirica Based on the analysis of impulse response functions obtained from the recursive VAR model we can summarize the following main findings for the effects of foreign shocks in the three SEE economies: first, despite the countercyclical fiscal policy, euro-zone output shocks have strong effects on economic activity and inflation, reflecting both the trade and capital channel. Second, positive foreign inflation shocks lead to higher domestic inflation in line with the small open economy features of these countries. Also, foreign inflation shocks trigger contractionary behavior of domestic monetary policy, aimed at curbing the adverse effects on inflation expectations. Third, an increase in foreign interest rate leads to a decline in domestic output through indirect channels, despite the fact that it is not transmitted into domestic money market rates. Fourth, besides these common results, which are predicted by economic theory, we find some cross-country differences in the response of macroeconomic variables. For instance, the effects of output shocks on output and inflation are larger in Bulgaria and Croatia, and Euribor shocks affect domestic output only on these two countries. Therefore, it seems that the level of integration with the EU matters for the transmission of foreign shocks. As can be seen, some of our findings provide support to standard textbook predictions that foreign output and inflation shocks have strong effects on domestic output and inflation. These effects are quite expected in small open economies under fixed exchange rate regimes and currency boards. Also, these findings imply that policy makers in these economies are faced with serious constraints in the conduct of economic policy, i.e. they often need to accommodate foreign macroeconomic shocks. In these regards, it is interesting to notice that, despite the countercyclical reaction of fiscal authorities, they are not able to offset the effects of foreign output shocks on domestic output and inflation. This finding suggests that in small open economies with fixed exchange rates fiscal policy may not be effective instrument as suggested by the Mundell–Fleming model. On the other hand, we find some deviations in the response of domestic macroeconomic variables to foreign shocks from those predicted by standard small open economy models. For instance, our results suggest that domestic interest rates are not linked with foreign interest rates, suggesting that the ‘‘impossible trilemma’’ may not work, i.e. there may be some limited room for autonomous monetary policy even in the presence of rigid exchange rate regimes. The limited autonomous monetary policy is conducted through non-market measures like reserve requirement, direct credit limits and other regulatory measures. 5.3 Comparison with CEE economies This section aims to compare and contrast the estimated results for the three SEE economies discussed in the previous subsection with the findings for CEE countries, such as: the Baltic States, the Czech Republic, Hungary, Poland, Slovakia and Slovenia. Both groups of countries, the CEE and SEE economies share similar characteristics in terms of their size and openness, i.e. they are small open economies, which are expected to be under strong influence of foreign shocks. On the other hand, CEE countries have been members of the EU for a longer period than the two SEE countries and thus, they have higher level of economic and 123 Empirica financial integration with the EU. Therefore, it may be interesting to see whether these two groups differ in their response to euro-area shocks. A common finding in the empirical literature on the transmission of foreign shocks to CEE economies is that euro-area output shocks are quickly transmitted to CEE countries, but the response depends on the type of monetary policy regimes in individual countries. For example, stronger adjustment of domestic output to a euroarea output shock is usually found in the Baltic States, Slovenia, Slovakia and Hungary, whereas the response of domestic output is weaker in Czech Republic and Poland (Eickmeier and Breitung 2006; Aslanidis 2010; Jiménez-Rodriguez et al. 2010; Balabanova and Brüggemann 2012). One possible explanation for the heterogeneity among the CEE countries refers to the type of monetary policy regime: the countries with inflation targeting are able to use their nominal exchange rate as a temporary buffer against external shocks (Eickmeier and Breitung 2006). Compared with the results for the SEE economies (Sect. 5.2.1), it is worth noting that, in general, the response of domestic output to a euro-area output shock in the Czech Republic, Poland and Hungary is lower and less persistent than in the three SEE economies. Similar conclusions hold for the transmission of euro-zone inflation shocks on domestic macroeconomic variables. Again, larger response of domestic inflation and interest rates can be noticed in the Baltic States, Slovenia and Slovakia as compared to Poland and Czech Republic (Eickmeier and Breitung 2006; Horváth and Rusnák 2009; Balabanova and Brüggemann 2012; Benkovskis et al. 2011). As before, the adjustment of domestic inflation is lower in inflation targeting countries (the Czech Republic and Poland) due to the reaction of their central banks, which raise policy rates in order to offset the impact of higher imported prices from the euro-zone. An additional explanation for the stronger transmission of euro-zone inflation to the Baltic States, Slovenia and Slovakia, apart from the rigid exchange rate regimes, may also be the small size of their economies, i.e. these countries are smaller compared to Poland and the Czech Republic, which are less dependent on euro-area imports (Benkovskis et al. 2011). The transmission of euro-zone interest rate shocks is quite strong and synchronized in all the CEE countries with a slightly weaker magnitude in the Czech Republic and Poland (Benkovskis et al. 2011). This implies that changes in the ECB’s monetary policy stance are quickly and strongly transmitted to CEE countries, which does not comply with or results for SEE economies as presented in Sect. 5.2.2. The difference in the reaction of the monetary policies between SEE and CEE economies may be explained by the lower degree of integration of financial markets in SEE economies with the euro-area as compared to CEE economies. The greater financial integration of CEE economies with the EMU can be noticed not only in the banking sector but also in capital flows (FDI and portfolio investment). Further on, the implementation of non-conventional monetary policy instruments by SEE central banks may be another factor contributing to their heterogeneous response compared to CEE countries. In brief, this section suggests that the transmission of foreign shocks to economic activity, prices and interest rates to some extent depends on the type of the monetary policy regime. In these regards, the transmission of foreign shocks is stronger in the 123 Empirica countries with rigid exchange rates because they cannot use nominal exchange rates as a temporary buffer against external disturbances. As for the transmission of foreign interest rate shocks, besides the exchange rate regime, it also depends on the level of financial integration with the EMU as well as the implementation of nonconventional monetary policy measures by central banks. 6 Conclusions In this paper we examine the effects of foreign shocks on SEE economies with fixed exchange rate regimes, such as: Bulgaria, Croatia and Macedonia, based on the impulse response functions estimated with recursive VARs. Specifically, we have conducted empirical investigation in the response of several macroeconomic and policy variables (output, inflation, interest rates and budget surpluses) in SEE countries to various euro-zone shocks (output, interest rates and inflation). In order to assess how exogenous factors affect macroeconomic performances and policy variables in SEE economies we impose the so-called block exogeneity restrictions regarding the interrelationship between the variables included in the VAR. The main findings from our study can be summarized as follows: first, notwithstanding the countercyclical fiscal policy, euro-zone output shocks have strong effects on economic activity and inflation, reflecting both the trade and capital channel. Second, positive foreign inflation shocks lead to higher domestic inflation, which in turn triggers a contractionary behavior of monetary policy makers, aimed at curbing the adverse effects on inflation expectations. Third, an increase in foreign interest rate leads to a decline in domestic output through indirect channels, despite the fact that it is not transmitted into domestic money market rates. Fourth, we find some cross-country differences in the response of macroeconomic variables. For instance, the effects of output shocks on output and inflation are larger in Bulgaria and Croatia, and Euribor shocks affect domestic output only on these two countries. Therefore, it seems that the level of integration with the EU matters for the transmission of foreign shocks. As can be seen, most of our findings imply that policy makers in SEE economies are faced with serious constraints in the conduct of economic policy, i.e. they often need to accommodate foreign macroeconomic shocks. However, we find some deviations in the response of domestic macroeconomic variables to foreign shocks from those predicted by standard small open economy models. For instance, our results suggest that domestic interest rates are not linked with foreign interest rates, thus, implying that the ‘‘impossible trilemma’’ may not work, i.e. there may be some limited room for autonomous monetary policy even in the presence of rigid exchange rate regimes through unconventional measures. Acknowledgments This research was supported by a grant from the CERGE-EI Foundation under a programme of the Global Development Network. The authors are grateful to the Journal’s editor and the two anonymous referees, as well as Sergey Slobodyan, Karsten, Staehr, Magdalena Petrovska, and participants at workshops and conferences in Skopje and Banja Luka for their helpful comments and suggestions. All opinions expressed are those of the authors and have not been endorsed by CERGE-EI, GDN, the National Bank of the Republic of Macedonia or the World Bank Group. 123 Empirica Appendices: IRFs of recursive VAR with 95 % confidence intervals of Efron and Hall, respectively Appendix 1: Impulses generated from the output gap in the euro-zone 123 Empirica Appendix 2: Impulses generated from the Euribor 123 Empirica Appendix 3: Impulses generated from the euro-zone inflation References Agénor PR, Montiel PJ (1996) Development macroeconomics. Princeton University Press, Princeton Amisano G, Giannini C (1997) Topics in structural VAR econometrics, 2nd edn. Springer, Heidelberg Angelovska-Bezoska A, Bogoev J, Mitreska A, Kadievska-Vojnovic M (2011) Investigating the cyclical behavior of fiscal policy in the Republic of Macedonia during the period of transition. Croat Econ Surv 13(1):57–104 Aslanidis N (2010) Business cycle synchronization between the CEEC and the Euro-area: evidence from threshold seemingly unrelated regressions. Manch Sch 78:538–555 Balabanova Z, Brüggemann R (2012) External information and monetary policy transmission in new EU member states: results from FAVAR models. Working paper series, no. 05. Department of Economics, University of Konstanz Beetsma RMWJ, Jensen H (2002) Monetary and fiscal policy interactions in a micro-funded model of a monetary union. Working paper no. 166. European Central Bank 123 Empirica Belvisio F, Milani F (2006) Structural factor-augmented VARs (SFAVARs) and the effects of monetary policy. B E J Macroecon 6(3):1–46 Benčı́k M (2011) Business cycle synchronisation between the V4 countries and the euro area. Working paper 1/2011. National Bank of Slovakia Benczur P, Koren M, Ratfai A (2004) The transmission of Eurozone shocks to CEECs. Central European University (manuscript), Budapest Benkovskis K, Bessonovs A, Feldkircher M, Worz J (2011) The transmission of Euro area monetary shocks to Czech Republic, Poland and Hungary: evidence from a FAVAR model. Focus Eur Econ Integr 3:8–36 Blanchard O, Quah D (1989) The dynamic effects of aggregate demand and aggregate supply disturbances. Am Econ Rev 79(4):655–673 Bogoev J (2011) What drives bank lending in domestic and foreign currency loans in a small open transition economy with fixed exchange rate? The case of Macedonia. Comp Econ Stud 53(2):307–331 Brown RL, Durbin J, Evans JM (1975) Techniques for testing the constancy of regression relationships over time. J R Stat Soc 37(2):149–192 Bryson JH (1994) Macroeconomic stabilization trough monetary and fiscal policy coordination: implication for European Monetary Union. Open Econ Rev 5(4):307–326 Chudic A, Pesaran MH (2014) Theory and practice of GVAR modeling. Working paper no. 180. Federal Reserve Bank of Dallas, Globalization and Monetary Policy Institute Crespo-Cuaresma J, Eller M, Mehrota A (2011) The economic transmission of fiscal policy shocks from Western to Eastern Europe. BOFIT discussion papers no. 12 Cushman DO, Zha T (1997) Identifying monetary policy in a small open economy under flexible exchange rates. J Monet Econ 39(3):433–448 Efron B, Tibshirani RJ (1993) An introduction to the bootstrap. Chapman & Hall, New York Eickmeier S, Breitung J (2006) How synchronized are new EU member states with the euro area? Evidence from a structural factor model. J Comp Econ 34(3):538–563 Enders W (2010) Applied time series econometrics, 3rd edn. Wiley, Hoboken Fidrmuc J, Korhonen I (2003) Similarity of supply and demand shocks between the euro area and the CEECs. Econ Syst 27(3):313–334 Fidrmuc J, Korhonen I (2006) Meta-analysis of the business cycle correlation between the Euro area and the CEECs. J Comp Econ 34(3):518–537 Frenkel M, Nickel C (2005) How symmetric are the shocks and the shock adjustment dynamics between the Euro area and Central and Eastern European countries? J Common Mark Stud 43(1):53–74 Galı́ J, Monacelli T (2008) Optimal monetary and fiscal policy in a currency union. J Int Econ 76:116–132 Hall P (1992) The bootstrap and edgeworth expansion. Springer, New York Horváth R, Rusnák M (2009) How important are foreign shocks in a small open economy? The case of Slovakia. Glob Econ J 9(1). doi:10.2202/1524-5861.1420 Hossain A, Chowdhury A (1996) Monetary and financial policies in developing countries. Routledge, London Jha R (1994) Macroeconomics for developing countries. Routledge, London Jiménez-Rodriguez R, Morales-Zumaquero A, Égert B (2010) The effect of foreign shocks in Central and Eastern Europe. J Policy Model 32(4):461–477 Karakitsos E (1992) Macrosystems—the dynamics of economic policy. Blackwell, Oxford Korhonen I (2003) Some empirical tests on the integration of economic activity between the euro area and the accession countries. Econ Transit 11(1):177–196 Kraft E (2003) Monetary policy under dollarisation: the case of Croatia. Comp Econ Stud 45(3):256–277 Krugman P (1988) External shocks and domestic policy responses. In: Dornbusch R, Helmers FLCH (eds) The open economy—tools for policymakers in developing countries. Oxford University Press for The World Bank, Oxford Krznar I, Kunovac D (2010) Impact of external shocks on domestic inflation and GDP. Working paper series, W-26. Croatian National Bank Lang M, Krznar I (2004) Transmission mechanism of monetary policy in Croatia. In: Paper presented at the tenth Dubrovnik economic conference Lutkepohl H, Kratzig M (2004) Applied time series econometrics. Cambridge University Press, Cambridge 123 Empirica Lutkepohl H, Kratzig M, Boreiko D (2006) VAR analysis in JmulTi. http://www.jmulti.de/. (Accessed 15 July 2012) Mackiewicz M (2008) Determinates of cyclicality of fiscal surpluses in the OECD countries. Munich personal RePEc archive working paper no. 16304 Minea A, Rault C (2011) External monetary shocks and monetary integration: evidence from the Bulgarian currency board. Econ Model 28:2271–2281 Obstfeld M, Rogoff K (1996) Foundations of international macroeconomics. MIT Press, Cambridge Petrevski G, Bogoev J (2012) Interest rate pass-through in South East Europe: an empirical analysis. Econ Syst 36(4):571–593 Petrovska M (2012) Business cycle regimes in a group of South East European economies. Evidence from a threshold SUR approach. Int J Econ Policy Emerg Econ 5(3):272–292 Prachowny MFJ (1984) Macroeconomic analysis for small open economies. Clarendon Press, Oxford Shone R (1989) Open economy macroeconomics. Harvester-Wheatsheaf, New York Stock JH, Watson MW (2001) Vector autoregressions. J Econ Perspect 15(4):101–115 Unevska-Andonova D, Petkovska M (2012) The transmission of external shocks to the Macedonian economic activity. Working paper. National Bank of the Republic of Macedonia Veličkovski I (2010) Potential costs of Euro adoption for transition countries: a case study for Macedonia. VDM, Frankfurt Velickovski I (2013) Assessing independent monetary policy in small, open and euroized countries: evidence from Western Balkan. Empir Econ 45(1):137–156 123