The Western Balkans on the road to the EU: An assessment of their integration with the euro zone using the Generalized Purchasing Power Parity approach.
Dimitrios Sideris
Bank of Greece
ABSTRACT
The present paper has two aims. The first aim is to test whether the five Western Balkan countries form an optimum currency area (OCA) with the eurozone, in an attempt to assess their integration with the European Union (EU). The second aim is to examine whether the decision of the countries to seek to join the EU created any forces fostering their convergence, evidence which would be in line with the theory on the endogeneity of the OCA criteria. Our findings indicate that the decision of the countries to seek to join the EU at some time in the future and the measures they took towards this aim, did promote integration of the five South and Eastern European (SEE) countries and that, at present, they are quite well aligned with the euro zone.
Keywords: EU enlargement; OCA; real exchange rates; cointegration; GPPP.
JEL classification: C32; F33; F36; F42.
Correspondence:Dimitrios Sideris. Economic Research Department. Bank of Greece. 21, El. Venizelou Ave.,10250 Athens.Greece.e-mail:
Acknowledgements: Technical assistance by Zacharias Bragoudakis is gratefully acknowledged. All remaining errors are entirely my responsibility.
1. Introduction
The main objective of the present study is to assess the degree of integration of the Western Balkan countries with the European Union (EU). We focus on these South Eastern European (SEE) countries, because they are not EU members and they share a number of common characteristics. The countries of interest areAlbania, Croatia, the FormerYugoslavianRepublic of Macedonia (FYROM), Moldova andSerbia.[1] The period of interest is the post-1990 period, which thus covers the transition phases of the Western Balkans from centrally planned economies to market economies. During this period all five countries pursued monetary and exchange rate policies and implemented legal and institutional reforms with the aim to foster macroeconomic stability and integration with the EU. The aim of the present paper is to investigate whether these measures have influenced the process of the fiveeconomies towards convergence with the EU. The five countriesshare common features: they are economies in transition; they are members of the Regional Cooperation Council for South Eastern Europe (SEE), which encourages the mutual cooperation and European and Euro-Atlantic integration of SEE countries (it succeeded the Stability Pact for South Eastern Europe in 2008); they are potential EU accession countries.[2]
In the present paper, we use two methods to evaluate convergence of the countries with the EU (and the euro area): (1) We first test whether there exists the Purchasing Power Parity (PPP) hypothesis between each country under consideration and the Euro zone. Validity of PPP would imply high degree of trade and goods markets’ integration between each of the five SEE countries and the EU. According to our knowledge, no previous study has tested this hypothesis for these countries. (2) We then test whether the five SEE countries form an Optimum Currency Area (OCA) with the euro zone;such an investigation has not been performed before in the literature, either. We test this hypothesis by applying the approach of the Generalized Purchasing Power Parity (GPPP). GPPPproposes testing whether the real exchange rates of a group of economies with respect to a base currency form a cointegrating vector or not. The theory is based on the following idea: it could be that the real exchange rates of a number of economies are not themselves stationary, as a result of the non stationarity of the fundamental economic variables; nevertheless, if the fundamentals are sufficiently integrated as in a currency area, the real exchange rates will share common trends and therefore, will form a cointegrating relationship.
In the study, we use monthly data starting at the beginning of the transition phase of the five economies at the early 1990s and ending in March 2009. In order to make use of all available observations, we approximate the eurozone by Germany given that Germany has been the reference country for all European countries during the pre-euro years of the European Monetary System and its central bank pursued an anti-inflationary monetary policy similar to that pursued by the European Central Bank. Besides, Germany still weights for roughly one-third of the euro area GDP.[3]The period thus covers a number of major political and economicevents that affected the performance of the economies individually (e.g. the fall of the Milocevic regime in Serbia in 2001) and/or jointly (e.g. the introduction of the euro in 1999, the 2007-2009 global financial crisis).
An additional issue of interest is whether the decision of the five SEE countries to seek to join the EU at some time in the future, and the policies they pursued towards this target, have facilitated their route towards the formation of an OCA with the euro area members. Evidence in favor of this hypothesis would be in line with the endogenous OCA theory.[4] The endorsement of the Stabilisation and Association Process (SAP) in November 2000 can be considered as a structural change that affected the economic policies of the five SEE countriestowards the euro area members. The SAP was followed by specific Stabilization and Association Agreements (SAA) which explicitly included provisions for future EU membership of the Balkan country involved. [SAAs are similar in principle to the Europe Agreements signed with the Central and Eastern European countries in the 1990s and the Association Agreement with Turkey.] To analyze the effects of such agreements we test for GPPP for the period after November 2000. In other words, we test whether the decision of the Western Balkan countries to seek to becomeEuropean Union members had any impact on the behavior of the real exchange rates and the fundamentals of the economies -as is suggested by the endogenous OCA theory.
The empirical work entails univariate stationarity analysis of the individual real exchange rate series; stationarity of a real exchange rate would mean that PPP between the country and the euro zone is valid, which in turn would imply a high degree of economic integration. Then, and in the event that the real exchange rates turn out to be non stationary, we test whether they form a cointegrating vector, in a multivariate setting. The model specification used for cointegration allows for different long-run relations and short-run dynamics. As evidenced in the relevant literature, if the short-run dynamics are different from the long-run relations, the specification of the short-run dynamics turns out to be crucial for the estimation of the equilibrium relationships.
The rest of the paper is organized as follows. Section II presents briefly a review of the GPPP theory and its relevance to the OCA theory. Section III presents the empirical results. The final section summarizes and concludes.
2. The economic background
The GPPP theory is based on the following idea: It could be that the fundamental economic variables determining real exchange rates of a group of economies are non-stationary, and consequently the real exchange rates of the economies are non-stationary. However, the fundamental variables can still be sufficiently integrated; in such an event, the real rates will share common trends and form a cointegrating relationship (Enders and Hurn, 1994). If this holds true, the economies constitute an optimal currency area in the sense of Mundell (1961) who argues that two economies constitute a currency area if they present similar real disturbances. Following this rationale, the existence of an equilibrium path for a linear combination of real exchange rates rules out the presence of real asymmetries and implies long-run sustainability of a monetary area.
The theory also suggests that, when economic interdependence in a group of economies is high, an economy’s bilateral real exchange rate is influenced by the exchange rates of the other economies in the group and the fundamentals of the other economies. The theory thus questions the validity of the standard bilateral tests for the validity of the Purchasing Power Parity (PPP) hypothesis as they ignore the influence that outside countries may have on bilateral exchange rates.[5]
Following the notation of Enders and Hurn (1994), GPPP can be described as follows: given an n-country world, an m-country (m≤n) currency area exists such that a long-run equilibrium relationship exists between the m - 1 bilateral exchange rates, of the form:
r21t = a + b31t r31t + b41t r41t + b51t r51t + ….. + bm1t rm1t + et (1)
where ri1t is the log of the bilateral real exchange rate in period t between country 1 and country i; a is the intercept term; bi1s are the parameters of the cointegrating vector, which represent the degree of comovement of the real exchange rates; and et is a stationary stochastic disturbance term.
It is clear that if all bi1s are equal to zero, then the traditional PPP -between countries 1 and 2- is valid. GPPP holds when the combination of the non-stationary bilateral real exchange rates is shown to be itself stationary. The bi1 parameters reflect the economic interdependencies within the group of economies. Enders and Hurn (1994) show that the estimated bi1s are closely linked to the aggregate demand functions of a goods market-clearing relationship. They also indicate that the more similar the aggregate demand functions in each country of the group, the lower the bi1s in magnitude.
The GPPP method has been used in a large number of papers, in order to test whether a group of countries form an OCA or not and, consequently, whether a group of economies, considered as a whole, is suitable for monetary integration.[6]
3. The empirical evidence
3.1 The data set
The econometric work entails initially univariate analysis of each real exchange rate series. In other words, we first test for stationarity of the series, applying a number of unit root tests. Then, and in the event that the real exchange rates turn out to be non stationary, we test whether there holds a GPPP relationship among them, using a cointegration technique.
In the study we use monthly observations for the domestic currencies of the five countries against the German mark. The price variables are measured by the consumer price index (CPI), given that CPIs are the indices published for all involved countries and are broadly similar as far as coverage is concerned. The sample period varies in the different economies, depending on the period when the reforms started and the availability of the data. Reforms started in the early 1990s in all five countries but data observations are available for the period after 1996 for Moldavia. All data are taken from the International Financial Statistics electronic database. To investigate any possible effects coming from the decision of the countries to try to join the EU, analysis is performed: (i) for the whole period 1996.1-2009.3, (ii) for the period following the SAP agreement 2000.12 – 2009.3.
3.2 Univariate analysis - Testing for PPP
The real exchange rate series are denoted as ri, where the subscript i takes the values a, cr, m, mo and s, which stand for Albania, Croatia, the FYROM, Moldova and Serbia, respectively. We test for the order of integration of the real exchange rate series using a set of standard Augmented Dickey –Fuller tests. Stationarity of a bilateral real exchange rate implies that PPP holds between the two economies, evidence which, in turn, indicates that the two economies are well integrated.
The results of the ADF unit root tests are presented in Table 1. Overall, thetests provide evidence for a unit root in all real exchange rate series. The results thus imply that there is no evidence for PPP linking any of the economies under consideration with the euro zone as approximated by Germany.
3.3 Testing for GPPP using cointegration analysis
The cointegration rank
Based on the results of the unit root tests, we then investigate whether the five real exchange rates with respect to the mark cointegrate in a GPPP relationship. We test for cointegration using the Johansen methodology on Vector Auto-Rregressive (VAR) models (Johansen, 1995). The analysis is performed for the two different periods. To this end, we estimate two unrestricted VARs for the vector = (ra, rcr, rm, rmo, rs) using multivariate least squares. The VAR systems are estimated assuming a constant in the deterministic variable set. The number of included lags in the VARs is determined on the basis of the Akaike information criterion and is set equal to eight.
The cointegration results for the full sample are reported in Table 2, Panel A. Both the maximum eigenvalue test and the trace test developed by Johansen (1995) do not reject the hypothesis of no cointegration. In other words, the results indicate that these countries did not form an OCA with the eurozone,when the whole period is considered.However, the results for the post–SAP agreement period, presented in Table 2, panel B, provide evidence for one cointegrating vector: The real exchange rates are closely integrated and form a GPPP relation during this period.
According to the results, the five countries have been operating as an optimal currency area in the recent period, but not before. This change in the findings possibly reflects the impact of the trade increase between the five economies and the euro zone, as a result of the new trade agreements and the change in the exchange rate and monetary policies of most of the five economies,which targeted alignment with the euro area countries. They may also be a result of the new and dominant role of the euro, -introduced in 1999.1- in the European markets.It may also reflect the higher level of coordination in their economic policies in recent years. [In fact, the monetary policy institutions, the goals and institutional settings of the central banks of the economies have converged to some degree in the recent period.]
The estimated cointegrating relationship.- The loadings
Table 3 presents the estimated cointegrated vector, which describes the GPPP relationship between the five real rates for the period 2000.12-2009.3. The normalized vector reflects the interrelationship among these real exchange rates. The estimated coefficients can be interpreted as long-run elasticities. All coefficients are lower than unity, implying a small size affect. For example, the estimated coefficients show that a 1% rise (fall) in the Croatian currency/mark real exchange rate will induce a 0.18 % fall (rise) in the Serbian dinar / mark real exchange rate. According to Ender and Hurn (1994), if the real exchange rates are only influenced by real output processes of the various nations, the normalized vector coefficients will be smaller the more similar are a country’s aggregate demand parameters.
The Johansen maximum likelihood approach also estimates the adjustment coefficients of each variable in the VARs. The adjustment coefficients indicate the speed at which the variables adjust towards their long-run equilibrium. The speed of adjustment shows how quickly any deviation from GPPP tends to correct itself. According to Johansen (1995), if a certain variable adjustment coefficient is insignificantly different from zero, then the variable is known to be weakly exogenous, as the dynamics of this variable are not influenced by the long-run equilibrium relationship.
Table 3also presents the speed of adjustment coefficients. The largest coefficient is found in the case of real exchange of the Czech koruna and the Serbian dinar against the mark. The coefficient 0,874 for the Serbian dinar implies that the dinar/mark real exchange rate adjusts at the rate of 87,4 percent per month toward the long-run equilibrium.
4. Conclusions
The present study aims to analyze the degree of convergence of the five Western Balkan economies with the eurozone. The work examines whether these countries form an OCA with the eurozone by using the GPPP theory. The paper also investigates whether the decision of the five countries to seek to join the EU had any impact on fostering their integration with the euro area. We argue that this decision of the five economies to join the monetary union and the policy steps made towards convergence with it, have already promoted their integration.
In the empirical work, cointegration analysis is employed to test the GPPP hypothesis –whether the real exchange rates converge in the long run– after an initial assessment of the stationarity of each real exchange rate series. The cointegration analysis examines the joint behavior of the rates, in two different periods: the full period and the period after the endorsement of the SAP agreement. The results provide evidence in favor of an OCA with the euro area only for the period following the SAP agreement. The results indicate that the group of the five economies has enjoyed a reduction in their real exchange rate instability in the recent period. This could be due to increased trade integration of the five economies with the EU caused by the introduction of the euro and the swift of the economic policies of most of the fiveSEE countries towards integration with the EU. They also indicate that a significant increase in policy convergence has been achieved.