The Baltic states and Europe: common factors of economic activity

Transcription

1 Baltic Journal of Economics ISSN: X (Print) (Online) Journal homepage: The Baltic states and Europe: common factors of economic activity Ludmila Fadejeva & Aleksejs Melihovs To cite this article: Ludmila Fadejeva & Aleksejs Melihovs (2008) The Baltic states and Europe: common factors of economic activity, Baltic Journal of Economics, 8:1, 75-96, DOI: / X To link to this article: Copyright 2008 Taylor and Francis Group LLC Published online: 03 Jun Submit your article to this journal Article views: 325 Citing articles: 3 View citing articles Full Terms & Conditions of access and use can be found at

2 The Baltic states and Europe: common factors of economic activity 75 The Baltic states and Europe: common factors of economic activity Ludmila Fadejeva, Aleksejs Melihovs 1,2 Abstract Participation in the European Economic Area (EEA) and the intention to join the monetary union increased the motivation of the new Member States to achieve a high level synchronisation of economic activity with the euro area. This paper aims to characterise fluctuations of economic activity that are common for the Baltic States, Central and Eastern European (CEE) countries, euro area countries, and Russia. For analysis of real standardised GDP growth, the dynamic factor model is employed. The results of the study show that the Baltic economies are similar in economic development and share a common factor. After 2000, GDP growth between the Baltic States and the main euro area countries indicates growing synchronisation of economic development between these country groups. JEL code: business cycle synchronisation, dynamic factor model, dynamic correlation. JEL classification: E32, F20, C10 1 Introduction One of the major EU objectives is to establish not only a common economic area but also a common monetary area. To enter the euro area, the new EU Member States should meet the Maastricht criteria, which include successful participation in the European Exchange Rate Mechanism II (ERMII) for two consecutive years. The majority of the new EU Member States (except Slovenia, which has already adopted the euro, Poland, the Czech Republic, and Hungary) have already joined the ERMII. Participation in the European Economic Area (EEA) and the intention to join the monetary union have increased the motivation of the new Member States to achieve high level synchronisation of economic activity with the euro area. In the case of synchronised economic 1 Division of Monetary Research and Forecast, Monetary Policy Department, Bank of Latvia, Riga, Latvia. s: Ludmila.Fadejeva@bank.lv, Aleksejs.Melihovs@bank.lv 2 The views expressed in this publication are those of the authors and do not necessarily represent the official views of the Bank of Latvia. The authors assume responsibility for any errors or omissions. The authors wish to thank Zoja Rasmusa, Uldis Rutkaste, Viktors Ajevskis (Bank of Latvia) and an anonymous referee for their valuable comments and recommendations.

3 76 Baltic Journal of Economics 8(1) (2008) fluctuations, the costs of possible counter-cyclical monetary policy are minimised 3, which is in line with the theory of optimum currency area (OCA) 4. The ability to assess the magnitude of cross-country co-movements in the economic activity of the Baltic States and to obtain a common development pattern with other countries of the European Union (EU), particularly the euro area, is of great importance. This paper aims to characterise the fluctuations of economic activity that are common for the Baltic States, Central and Eastern European (CEE) countries represented by the Czech Republic, Poland, Slovakia, and Slovenia, euro area countries represented by France, Germany, and Italy, as well as the Commonwealth of Independent States (CIS) represented by Russia. Real standardised GDP quarterly growth is chosen as an indicator of economic development of countries. The main questions to be answered within the current research are as follows: Do the Baltic States share a common factor in real GDP growth? Are the Baltic States significantly different from other Central and Eastern European countries in respect to co-movements in economic activity? To what extent does real GDP growth in the euro area, Russia, and CEE countries explain developments in real GDP growth in the Baltic States? In order to answer these questions, the dynamic common factor model is employed. This assesses the impact of various common factors on real standardised GDP growth of individual countries. The paper is structured as follows. Section 2 lays out the methodology employed. Section 3 provides some details of data. Section 4 presents the research results. Section 5 sums up. 2 Analysis of business cycle synchronisation The definition of business cycles has changed over time. In early studies, a business cycle, the so-called classical cycle, was defined as sequences of expansions and contractions in series representing levels of economic development. This approach is typically associated with the National Bureau of Economic Research (NBER) 5. Later, due to very high rates of real economic growth after WW II and a slowdown rather than absolute declines in overall economic activity, a view that an economic time series should be decomposed into the sum of trend and cyclical components, commonly referred as a growth cycle 6, was developed. According to the OECD definition, a growth cycle is a more accurate 3 See, for example, Darvas and Szapary (2005). 4 The theoretical foundations of currency unions have been developed in the literature on OCA pioneered by Mundell (1961) and continued by McKinnon (1963) and Kenen (1969). 5 See Monch and Uhlig (2004) for an example of applied NBER developed methodology and the Bry-Boshan procedure. 6 The term "growth cycle" was introduced by OECD in 1960.

4 The Baltic states and Europe: common factors of economic activity 77 definition of cycles of economic activity where contractions (expansions) include not only absolute declines (increases), which is in line with the NBER approach, but also slowdowns (accelerations). The main questions in respect to the OECD approach are how the trend and cyclical component should be identified and estimated. In order to solve these issues, a range of parametric and non-parametric measures have been developed 7. Since the end of the 1980s, business cycles have been viewed in a wider international context, taking into account the economic interactions of different countries. Special attention was paid to the definition of the common business cycle given by Burns and Mitchell in , which states that the common business cycle can be seen as the co-movement of aggregate economic activity of countries, defined by the similar timing of expansions and contractions in economic developments. These theoretical concepts were empirically proved by Stock and Watson (1988, 1991, 1998) who used a dynamic factor model to capture co-movements by obtaining a single common factor from a set of many macroeconomic series, and Hamilton (1994) who developed a nonlinear model with a discrete regime switching between periods of expansion and contraction to assess the dynamics of real GNP 9. A number of different methods have been applied to study the degree of synchronisation between economic variables. Three main groups are mentioned in the literature. First, the correlation approach; second, VARs, particularly structural VARs; and third, the factor model approach. The survey paper by Firdmuc and Korhonen (2006) reports 35 studies (the first two approaches are mainly used) on business cycle correlation between the euro area and all CEE countries. The results of most studies indicate a high level of synchronisation between economic indicators of euro area and CEE countries, such as Poland, Slovenia, the Czech Republic, and Slovakia, with almost no synchronisation between the Baltic States and the euro zone. The results for the Baltic States could well be explained by the specific shock to the Baltic economies due to the Russian crisis of 1998 and the length of data series (mostly up to 2002). Longer data series allow us to eliminate the effect of the Russian crisis and to come up with new results of synchronisation between GDP growth series for the euro zone and Baltic States after The current paper is based on analysis of co-movement of economic indicators and follows the dynamic factor model specification of Monfort et al. (2004) and Moneta and Ruffer (2006). According to the literature on synchronisation of business cycles available to the authors, the existence of common factors between the euro zone and the Baltic States, eliminating the negative shock caused by the Russian crisis of 1998, have not been evaluated using the above approach. In the specification of the dynamic factor model it is assumed that the n-dimensional stochastic process y i,t (i = 1,...,n) depends linearly on m unobservable factors z k (k = 1,..., m ), which t in turn follow the first order autoregressive process. The linear state-space model can be written in a matrix notation as follows: 7 See Mitchell and Mouratidis (2002), Zarnowitz and Ozyildirim (2002) for an overview of parametric and nonparametric measures of the growth cycle. 8 The full definition can be found in Burns, A.F. and Mitchell, W.C. (1964), Measuring Business Cycles. New York: National Bureau of Economic Research. 9 See Diebold and Rudebush (1994) for explicit representation of the two concepts.

5 78 Baltic Journal of Economics 8(1) (2008) Y t = AY t-1 + BZ t + ε t Z t = DZ t-1 + η t (1) where A and D are diagonal matrixes, ε t and η t are independent Gaussian white noise vectors. The variance-covariance matrix V[ε] of disturbances ε t is assumed to be diagonal: where δ 2 represents variances of the error term. The B matrix measures the instantaneous i impact (factor loadings or sensitivities) of common factors on each series y i. The advantage of this specification is that it is fairly flexible and allows for distinguishing between the factors common for all y i and factors common for a group of y i (specific common factors). In the case of specific factors, a system of equations [1] can be decomposed into the following system of equations: Y t = AY t-1 + BZ t + CW t + ε t Z t = DZ t-1 + η t z (2) W t = D w W t-1 + η t w where Z t is the specific factor for a group series and W t is the common factor affecting all series 10. The Kalman filter approach is used to estimate the parameters of the model Data The source of quarterly seasonal and working day adjusted European data for the period from 1996 to 2007 is the EUROSTAT database, the source of Russian GDP data is the Federal State Statistics Service of Russia. GDP data for Russia have been seasonally adjusted using the Census X12 method. The data set is log-differenced and standardised to remove the scale effect of different economies and to ensure comparability of time series fluctuations. This provides a comparison of estimated coefficients between countries within the frame of a single model without implementing additional weights. Quarterly real standardised GDP data for European countries and Russia are used to evaluate synchronisation of economic activities. Countries are subdivided into three main groups, according to the interest of the current paper: the Baltic States (Latvia, Lithuania, and Estonia), CEE countries (the Czech Republic, Poland, Slovakia, and Slovenia), euro area countries (France, Germany, and Italy) and the CIS (Russia). 10 See Monfort (2004) for more details. 11 For detailed methodology on the Kalman filter see chapter 13 of Hamilton (1994).

6 The Baltic states and Europe: common factors of economic activity 79 4 Results Since the Baltic and most of the CEE countries have joined ERM II and synchronisation of economic activity of countries within the euro area is gaining importance, special attention in the current paper is given to analysis of common fluctuations of real economic activity in the euro area, the Baltic and CEE countries. In order to obtain a deeper understanding of the impact of Russia's economic activity on the economic activity of the Baltic States and, in particular, of the 1998 crisis, Russia is also included in the analysis 12. Additional analysis of synchronisation between Scandinavian (Sweden, Norway, Finland, and Denmark) and euro area countries proved that these countries share a common factor. The individual common factor of Scandinavian countries in addition to a common factor for Scandinavia and the euro area is negligible during the period analysed. Therefore, the authors found no evidence that the economic development of Scandinavian countries could have a different influence on the Baltic States compared to euro area countries. For this reason, the current publication does not provide a detailed analysis of a common factor between Scandinavia and the Baltic States, which to a large extent is similar to the common factor for the euro area and the Baltic region. 4.1 Common dynamic factor models of real standardised GDP growth in the Baltic States This section presents the results of four common dynamic models for the Baltic States Common factor for real standardised GDP growth in the Baltic States To estimate a common factor for real standardised GDP growth in the Baltic States, a onefactor model is employed. Parameter estimates for a model of one common factor appear in Table 2. Lagged dependent variables are not significant, while the impact coefficients of the common factor are all significant and similar in magnitude among countries (see Table A1 in the Appendix). Table 1 Correlation between real standardised GDP growth series and common factor LV EE LT corr(z,y i ) Analysis of static correlation between the common factor for the Baltic States and real standardised GDP growth (see Table 1) indicates that on average the economic activity of Lithuania is less synchronised with the common factor and hence also with fluctuations of the economy in Latvia and Estonia. 12 In the current paper the term "Russian crisis" is used to denote the financial crisis in Russia in 1998, which occurred as a result of unsustainable public debt dynamics and the correction needed to an overvalued real exchange rate. Detailed analysis of the crisis can be found in Pinto et. al. (2005).

7 80 Baltic Journal of Economics 8(1) (2008) , 14 Table 2 Parameter estimates (model with one common factor) LV EE LT a i,i (0.245) (0.128) (0.166) b i 0.594**** 0.479**** 0.358*** (0.171) (0.130) (0.173) σ i 0.574**** 0.766**** 0.849**** (0.168) (0.109) (0.111) d 0.751**** (0.148) Y t = AY t-1 + BZ t + ε t 14 Z t = dz t-1 + η t Note: coefficient is significant at 1% (***), 5% (**), 10% (*) The results of correlation analysis and estimated b coefficients of common factor impact suggest that the estimated factor to a large extent represents economic activity in Latvia. Real standardised GDP growth of Estonia and Lithuania shows that additional specific fluctuations in the economic development of these countries are more pronounced. As Chart 1 shows, there are two specific medium-term deviations from the zero level in common factor for real GDP dynamics in the Baltic countries: the first corresponds to the negative exogenous shock in 1998 (the Russian crisis) and the second captures the positive exogenous shock in 2004 (accession to the EU) and the following period of consistent and rapid economic growth up to the end of Chart 1 Common factor for real standardised GDP growth of Latvia, Lithuania, and Estonia; +/- 2RMSE To assess the degree of synchronisation among countries and to investigate if any dynamics exist, the correlation analysis over a four-year moving window is estimated (see Chart A2 in the Appendix). The moving correlation between the common factor obtained and real standardised GDP growth for Estonia and Latvia is quite stable. The moving correlation for Lithuania during is stable as well; however, later the tendency changes. Correla- 13 Standard error in parenthesis. 14 Hereinafter, the variance of error term (ε t ) is set at 1 (due to data standardisation and for identification purposes).

8 The Baltic states and Europe: common factors of economic activity 81 tion gradually vanishes and remains close to zero after The period of stable correlation can be explained by a similar response to the Russian crisis and further recovery. The low overall correlation for Lithuania and the downward slope of moving correlation after 2002 may be explained by the country's specific economic structure compared with other Baltic States. One possible reason for this may be the large share of the oil processing industry in the manufacturing sector and differences in credit market development. For example, due to low inflation, the real interest rate in Lithuania remained positive up to 2005, which is not the case for Latvia and Estonia where the respective rate had already turned negative in 2004 (see Table A3). Compared with other Baltic States, such dynamics of the real interest rate in Lithuania during the period under review stimulated private consumption to a lesser extent, bringing about moderation of growth in domestic demand. Therefore Lithuania is the only Baltic country which does not show signs of its economy overheating in the period under review after 2004, which indicates moderate GDP growth level compared with neighbouring countries Common factors for real standardised GDP growth in the Baltic States and CEE countries In order to check if a common factor exists in the development of real standardised GDP growth for the CEE and Baltic countries, the previous model is augmented by an additional common factor for the Baltic States and CEE, while retaining the common factor for the Baltic States. The number of factors to be included in the model is checked by information criteria (see Table A4); the specification with the smallest value is chosen. Table 3 shows parameter estimates for the model. Table 3 Parameter estimates (model with two common factors) 15 LV EE LT CZ PL SK SI a i,i **** **** **** (0.359) (0.187) (0.304) (0.074) (0.062) (0.170) (0.148) c i (0.279) (0.235) (0.243) (0.031) (0.220) (0.266) (0.174) b i 0.511*** 0.346*** (0.249) (0.168) (0.247) σ i 0.567**** 0.829**** 0.830**** 0.238**** 0.443**** 1.014**** 0.756**** (0.223) (0.178) (0.149) (0.043) (0.153) (0.108) (0.149) d w 0.730**** (0.268) d z 0.774**** (0.228) Y t = AY t-1 + CW t + BZ t + ε t W t = d w W t-1 + η w t Z t = d z Z t-1 + η z t - 1st common factor (for Baltic and CEE countries) - 2nd common factor (for Baltic countries only) Note: coefficient is significant at 1% (****), 5% (***), 10% (**), 15% (*) 15 Standard error in parenthesis.

9 82 Baltic Journal of Economics 8(1) (2008) The impact of the common factor for the CEE and Baltic countries is not significant for all countries, implying that no common factor exists for the CEE and Baltic countries (see Table 3). CEE and Baltic countries do not share a common factor that explains fluctuations in real standardised GDP growth Common factors for real standardised GDP growth in the Baltic States and the main euro area countries Similar to analysis of the common factor for the CEE countries and Baltic States, the common factor in development of real standardised GDP growth for the main euro area countries and Baltic States is estimated. Parameter estimates of the two-factor model, one of them common for the Baltic States and euro area, and the other only for the Baltic States, appear in Table 4. The inclusion of two common factors is proved by the Akaike information criteria (see Table A4). Table 4 Parameter estimates (model with two common factors) 16 LV EE LT IT FR DE a i,i ** (0.221) (0.155) (0.209) (0.197) (0.171) (0.138) c i 0.459*** 0.396*** **** 0.462**** 0.378**** (0.190) (0.197) (0.166) (0.189) (0.141) (0.142) b i 0.427*** 0.316*** 0.307*** (0.180) (0.159) (0.141) σ i 0.524**** 0.789**** 0.773**** 0.396* 0.773**** 0.832**** (0.164) (0.123) (0.109) (0.273) (0.110) (0.132) d w 0.625**** (0.153) d z 0.838**** (0.138) Y t = AY t-1 + CW t + BZ t + ε t W t = d w W t-1 + η w t - 1st common factor (for the Baltic States and the main euro area countries) Z t = d z Z t-1 + η z t - 2nd common factor (for the Baltic States only) Note: coefficient is significant at 1% (****), 5% (***), 10% (**), 15% (*). The common factor for the euro area and Baltic States as well as that for the Baltic States only has a statistically equal effect on the dynamics of real standardised GDP growth in Latvia and Estonia (see Table A5). Compared with the one-factor model, coefficient b i for the common factor of the Baltic States became smaller due to inclusion of an additional factor, which picked up some explanatory power from the common factor for the Baltic States. Together with statistically significant coefficients of b i and c i this indicates subdivision of fluctuations of real standardised GDP growth between the factors and proves that the Baltic and euro area countries share a common development pattern. Subdivision of fluctuations caused a 16 Standard error in parenthesis.

10 The Baltic states and Europe: common factors of economic activity 83 decrease of static correlation coefficients for the common factor for the Baltic States (see Table 2 and Table 5). Table 5 Correlation between real standardised growth series and common factors LV EE LT IT FR DE corr(w,y i ) corr(z,y i ) Implementation of the second common factor decreased the volatility of the common factor for the Baltic States; thus the effect of economic fluctuations common to the euro area and Baltic States is removed from this factor (see Chart 2). 1st common factor (LV, EE, LT, IT, FR, DE) 2nd common factor (LV, EE, LT) Chart 2 Common factors for real standardised GDP growth in the two-factor model; +/- 2RMSE In order to analyse the contribution of each common factor in relation to economic growth of the respective country, real standardised GDP growth has been subdivided according to the importance of common factors and the individual factor, which is defined as a part of economic activity growth not explained by common factors. The contribution of estimated common factors to the dynamics of real standardised quarterly GDP growth in the Baltic States appears in Chart A6. The common factor for the Baltic States explains the slowdown of economic activity due to the common response of the Baltic Region to the Russian crisis of The positive development of the euro area, supplemented by successful reorientation of Baltic exports from the Russian to the European market (see Table A7), speeded up recovery of the Baltic States' economic activity in The contribution of the common factor for the Baltic States after 2001 is mostly positive, showing an overall positive economic development in the Baltic Region. A somewhat negative impact of the decline in economic activity of the euro area was present in After accession to the EU, the common factor for the Baltic States represents a pronounced positive effect on the economy. The correlation over a four-year moving window between common factors and the series of real standardised GDP growth of countries appears in Chart A8. The results show that after 2002 real standardised GDP growth in Latvia became more correlated with the common fac-

11 84 Baltic Journal of Economics 8(1) (2008) tor for real standardised GDP growth in the euro area and Baltic States. Estonian real standardised GDP growth is stable during this period. After 2004, the share of exports from the Baltic States to the new EU countries increased significantly (see Table A7) due to strong growth of domestic demand in the EU10, which induced an increase in imports of goods and services (see Table A9 and Table A10). Together with moderate economic growth in the EU12, this determined a change in the structure of exports from the Baltic States to the EU (see Table A7). The share of foreign trade to the EU12 decreased, mainly due to decrease of export share. The decline observed in the share of exports to the euro area after 2004 is explained by the recorded decrease in the value of four-year moving correlation between the first factor (common for the Baltic States and euro area) and real GDP growth series after 2004 (see Chart A8). Additional analysis of synchronisation between the CEE and euro area countries suggested that these countries also share the common factor; however, the common factor for CEE countries only is not significant. The four-year moving correlation between the common factor for the euro zone and CEE and real standardised GDP growth for CEE countries has on average been increasing since This finding, together with the results of the current section, proves that the CEE and Baltic countries are both moving towards higher synchronisation with the euro area. This could be seen as a positive sign in the process of integration into European monetary union for new European countries. Taking into account the above, it could be foreseen that synchronisation between Baltic and CEE states might improve Common factors for real standardised GDP growth in the Baltic States, main euro area countries, and Russia To evaluate the impact of Russian economic activity and, in particular, the crisis of 1998 on development of the Baltic economies, the two-factor model used in Section was augmented by a third factor common for Russia and the Baltic States. The estimated results of the three-factor model appear in Table 6. The coefficient values for each of the factors are statistically equal for neighbouring countries, suggesting that the factors obtained have a similar impact on economic activity in Latvia, Lithuania, and Estonia (see Table A11). The results of the third common factor show that the coefficients of the common factor for the Baltic States and Russia are not statistically significant (see Table A4 and Table 6). However, graphic representation of the extracted common factor (see Chart 3) shows well-defined and statistically significant periods in development of the Russian economy, important for the explanation of economic fluctuations in the Baltic States, captured by the third common factor, namely the Russian crisis of 1998.

12 The Baltic states and Europe: common factors of economic activity 85 Table 6 Parameter estimates (model with three common factors) 17 LV EE LT IT FR DE RU a i,i (0.254) (0.277) (0.363) (0.267) (0.217) (0.194) (1.025) c i 0.503*** 0.402* **** 0.537**** 0.415*** - (0.236) (0.260) (0.273) (0.243) (0.171) (0.183) - b 1i 0.414*** 0.254* 0.261* (0.205) (0.170) (0.163) b 2i (0.465) (0.506) (0.631) (1.344) σ i 0.489**** 0.809**** 0.815**** 0.452** 0.751**** 0.853**** (0.188) (0.187) (0.220) (0.282) (0.129) (0.189) (1.387) d w 0.988**** (0.075) z d **** (0.162) z d (0.835) Y t = AY t-1 + CW t + B 1 Z 1 + B 2 Z 2 + ε t t W t = d w W t-1 + η w t - 1st common factor( for the Baltic and main euro area countries) Z 1,t = d 1z Z 1,t-1 + η 1, - 2nd common factor( for Baltic countries only) zt Z 2,t = d 2z Z 2,t-1 + η 2, - 3rd common factor (for Baltic countries and Russia) zt Note: coefficient is significant at 1% (****), 5% (***), 10% (**), 15% (*) 1st common factor (LV, EE, LT, IT, FR, DE) 2nd common factor (LV, EE, LT) 3rd common factor (LV, EE, LT, RU) Chart 3 Common factors for real standardised GDP growth in the three-factor model; +/- 2RMSE 17 Standard error in parenthesis.

13 86 Baltic Journal of Economics 8(1) (2008) Comparison of common factors for real standardised GDP growth in the Baltic States obtained from the two- and three-factor models with and without a specific factor for Russia (see Chart A12) shows that inclusion of a common factor for the Baltic States and Russia influenced the dynamics of the common factor for the Baltic States during The decline in economic activity in the Baltic States has a less pronounced trough in the second half of 1998 due to exclusion of the Russian crisis pattern represented by a sharp drop in GDP growth (see Chart 3, the third common factor). The common factor for the Baltic States and Russia is mainly formed by the dynamics of Russian real standardised GDP growth. The average correlation of Latvia and Estonia with this factor is rather high mainly due to a similar response to the Russian crisis of 1998 (see Table 7). Table 7 Correlation between series of real standardised GDP growth and common factors LV EE LT IT FR DE RU corr(w,y i ) corr(z 1,Y i ) corr(z 2,Y i ) The graphic representation of common factor contributions to the dynamics of real standardised GDP growth for Latvia appears in Chart A13. The decrease in Russian economic activity in the first half of 1998 resulting from negative tendencies in the world financial markets and the financial crisis in Russia in the second half of 1998 is removed from the common factor for the Baltic States. The response of the Lithuanian economy to the Russian crisis was weaker and the following recovery slower compared with neighbouring countries. In the present study, the structure of the dynamic factor model is defined as a first-order auto-regressive process, so that the effect of economic growth in Russia on GDP growth in Lithuania should be evaluated with caution (see Chart A13). Correlation between real standardised GDP growth for Latvia, Lithuania, Estonia and the factor common for the Baltic States and Russia indicates that until 2003 the correlation for the Baltic States is stable and rather high due to a similar response to the Russian crisis as discussed earlier. After 2002, when the impact of the crisis is excluded from the analysis, the moving correlation between real standardised GDP growth for Latvia and Estonia and the third factor common to the Baltic States and Russia becomes less stable and, on average, smaller (see Chart A14). 5 Conclusion This paper identifies a number of facts about observed common factors in real standardised GDP growth of the Baltic, CEE and main euro area countries, as well as Russia. Dynamic factor models are used to extract joint fluctuations (common factors) in growth of real standardised GDP series for different countries and to assess the contribution of those factors to total GDP growth in each country.

14 The Baltic states and Europe: common factors of economic activity 87 The main results of four specifications of the dynamic common factor model are as follows: - The Baltic States share a common pattern in GDP growth, represented by an obtained common factor for these countries. Lithuania, due to its specific economic structure, represents a lower degree of synchronisation between the common factor and real GDP growth in comparison to the other Baltic States. - The existence of a common factor in the economic activity of the Baltic States and the main euro area countries is proved by the results of the two- and three- common factor models. The results of four-year moving correlation between the common factor and series of real standardised GDP growth of countries show that after 2000 real standardised GDP growth in the Baltic States became more synchronised with GDP growth of the main euro area countries. - Results of the two- and three-factor models show that the role of a common factor for the Baltic States remains significant and pronounced under different specifications. The existence of a common factor for the Baltic States could be interpreted as a region-specific development path, associated with pronounced restructuring and convergence processes of the economies. - The Baltic and CEE countries do not share a common factor in development of real GDP growth during the period analysed. Additional analysis of the CEE and euro area countries suggested that these countries share a common factor, with increasing synchronisation between regions since This finding shows that the CEE and Baltic countries are both moving towards higher synchronisation with the euro area, and with time the synchronisation between the CEE and Baltic States might improve. - Inclusion of Russian standardised GDP growth in the analysis helps to extract a factor common for the Baltic States and Russia, indicating the period of the Russian crisis when development of the Russian economy significantly influenced the dynamics of GDP growth in the Baltic countries. Results of four-year moving correlation indicate that after 2002 synchronisation between the common factor for the Baltic States and Russia and GDP growth of the Baltic States decreased and became less stable. Bibliography Agresti, A.-M., Mojon, B. (2001) "Some Stylised Facts on the Euro Area Business Cycle", ECB Working Paper, No. 95. Avouyi-Dovi, S., Kierzenkowski, R., and Lubochinsky, C. (2006) "Are Business and Credit Cycles Converging or Diverging? A Comparison of Poland, Hungary, the Czech Republic and the Euro Area.", Banque de France, Working Paper NER-E#144. Camacho, M., Perez-Quiros, G., Saiz (2006) L. "Are European business cycles close enough to be just one?" Journal of Economic Dynamics and Control, Volume 30, pp Croux, C., Forni, M., and Reichlin, L. (1999) "A Measure of Co-movement for Economic Variables: Theory and Empirics", CEPR Discussion Paper No Darvas, Z., Szapary, G. (2005) "Business Cycle Synchronisation in the Enlarged EU", CEPR Discussion paper No.5179.

15 88 Baltic Journal of Economics 8(1) (2008) Diebold, F.X., Rudebusch, G.D. (1994) "Measuring Business Cycles: A Modern Perspective", NBER Working Paper No Fidrmuc, J., Korhonen, I. (2006) "Meta-Analysis of the Business Cycle Correlation Between the Euro Area and CEECs", CESifo Working Paper No Hamilton, J.D. (1989) "A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle", Econometrica, Vol. 57, No. 2., pp Hamilton, J.D. (1994). "Time Series Analysis", New Jersey: Princeton University Press. Kenen, P. (1969) The Theory of Optimums Currency Areas: An Eclectic view in Mundell, R.A, Swoboda,K.A Monetary Problems of the International Economy(University of Chicago Press, 1969, ( Exchange Rates and the Monetary System by Kenen, P.B, Edward Elgar Publishing Limited. 1994). Massmann, M., Mitchell, J. (2003) "Reconsidering the Evidence: Are Eurozone Business Cycles Converging?", National Institute of Economic and Social Research, NIESR Discussion Paper, No McKinnon, R. I. (1963), Optimum Currency Areas, The American Economic Review, Vol.53, No Mitchell, J., Mouratidis, K. (2002) "Is there a Common Euro-Zone Business Cycle?", Eurostat: Eurostat Colloquium, Modern Tools For Business Cycle Analysis. Monch, E., Uhlig, H. (2004) "Towards a Monthly Business Cycle Chronology for the Euro Area", CEPR Discussion paper No Moneta, F., Ruffer, R. (2006) "Business Cycle Synchronisation in East Asia", ECB Working Paper Series No Monfort, A., Renne, J.P., Ruffer, R., and Vitale G. (2004) "Is Economic Activity in the G7 Synchronised? Common Shocks Versus Spillover Effect", CEPR Discussion Paper No Mundell, R.A. (1961) "A theory of optimum currency areas", The American Economic Review 51 (4), Pinto, B., Gurvich, E., and Ulatov, S. (2005) "Lessons from the Russian Crisis of 1998 and Recovery" in Managing Volatility and Crisises: A Practitioner's Guide, ed. J.Aizenman and B.Pinto, Cambridge University Press, Cambridge, pp Stock, H.J., Watson, M.W. (1988) "A Probability Model of the Coincident Economic Indicators", NBER Working Paper No Stock, H.J., Watson, M.W. (1991) A probability model of coincident economic indicators, in K. Lahiri and G.H. Moore, eds., Leading Economic Indicators: New Approaches and Forecasting Records, Cambridge University Press, Cambridge, pp Stock, H.J., Watson, M.W. (1998) Diffusion indexes, NBER Working Paper No Tripier, F. (2002) "The Dynamic Correlation between Growth and Unemployment", Economics Bulletin, Vol.5, No. 4 pp 1-9. Zarnowitz, V., Ozyildirim, A. (2002) "Time Series Decomposition and Measurment of Business Cycles, Trends and Growth Cycles", NBER Working Paper No. W8736.

16 The Baltic states and Europe: common factors of economic activity 89 Appendix Table A1 Wald statistics (one-factor model; Baltic States) Chi-square value df Probability b LV = b LT b LV = b EE b EE = b LT Chart A2 Correlation between series of real standardised GDP growth and common factor for the Baltic States computed over a four-year moving window (Latvia, Lithuania, Estonia) Table A3 Real three-month interest rate* (Estonia, Latvia, Lithuania; ) Estonia Latvia Lithuania * 3-month interest rate index versus HICP year-on-year index Source: Eurostat database Table A4 Information criteria for different model specifications Number and specification of factors Akaike info criterion Common factor for CEE 1 factor (Balt+CEE) and Baltic States 2 factors (Balt+CEE; Balt) Common factor for euro area 1 factor (Balt+EA) and Baltic States 2 factors (Balt+EA; Balt) Common factor for euro area, 1 factor (Balt+EA+RU) Baltic States and Russia 2 factor (Balt+EA; RU) factors (Balt+EA; Balt; RU) factors (Balt+RU; Balt+EA; Balt) Notes: Balt Baltic States, CEE - Central and Eastern European countries, EA euro zone countries, RU - Russia

17 90 Baltic Journal of Economics 8(1) (2008) Table A5 Wald statistics (two-factor model; main euro area countries and Baltic States) Chi-square value df Probability c LV = b LT c EE = b EE c LV = C EE Chart A6 Contribution of common and individual factor to the dynamics of real standardised GDP growth (two-factor model; Latvia, Estonia, Lithuania)

18 The Baltic states and Europe: common factors of economic activity 91 f1 1st common factor (LV, EE, LT, IT, FR, DE) f2 2nd common factor (LV, EE, LT) f* individual factor Chart A6 (continued) Table A7 Structure of exports of Baltic countries (% of total) Baltic States EU EU15* RU EU10** Other Estonia EU EU15* RU EU10** Other Latvia EU EU15* RU EU10** Other Lithuania EU EU15* RU EU10** Other Export to Baltic States Scandinavia Estonia from Latvia Lithuania * EU15* represents a part of exports to the EU15 not included in EU12 ** Exports to nine new EU Member States (excluding the respondent country) Source: EUROSTAT, Comext database (SITS)

19 92 Baltic Journal of Economics 8(1) (2008) st common factor (LV, EE, LT, IT, FR, DE) 2nd common factor (LV, EE, LT) Chart A8 Correlation between series of real standardised GDP growth and common factors computed over four-year moving window

20 The Baltic states and Europe: common factors of economic activity 93 Table A9 Structure of imports of Baltic countries (% of total) Baltic States EU EU15* RU EU10** Other Estonia EU EU15* RU EU10** Other Latvia EU EU15* RU EU10** Other Lithuania EU EU15* RU EU10** Other Export to Baltic States Scandinavia Estonia from Latvia Lithuania * EU15* represents a part of imports from EU15 not included in EU12 ** Imports to 9 new EU Member States (excluding the respondent country) Source: EUROSTAT, Comext database (SITS) Table A10 Real annual growth in imports of goods and services ( ) Czech Republic Estonia Latvia Lithuania Hungary Poland Slovenia Slovakia Source: EUROSTAT database Table A11 Wald statistics (three factor model; main euro area countries, Baltic States, Russia)

21 94 Baltic Journal of Economics 8(1) (2008) Chi-square value df Probability c LV = c LT = c EE b 1,LV = b 1,EE = b 1, LT b 2,LV = b 2,EE = b 2, LT Chart A12 Comparison of common factors for real standardised GDP growth in the Baltic States obtained from two- and three-factor models (with and without a specific factor for Russia; Baltic States, main euro area countries, Russia) Chart A13 Contribution of common and individual factors to dynamics of real standardised GDP growth (three-factor model)

22 The Baltic states and Europe: common factors of economic activity 95 f1 1st common factor (LV, EE, LT, IT, FR, DE) f2 2nd common factor (LV, EE, LT) f3 3rd common factor (LV, EE, LT, RU) f* individual factor Chart A13 (continued) 1st common factor (LV, EE, LT, IT, FR, DE) Chart A14 Correlation between series of real standardised GDP growth and common factors computed over four-year moving window

23 96 Baltic Journal of Economics 8(1) (2008) nd common factor (LV, EE, LT) 3rd common factor (LV, EE, LT, RU) Chart A14 (continued)