Export performance of firms in Baltic countries i

Article history: Received 17.08.2015; last revision 22.12.2015; accepted 20.01.2016 doi: 10.24984/iel.2017.5.2.1 Export performance of firms in Baltic countries i Andrzej Cieślik ii, Jan Michałek iii and Anna Michałek iv Abstract: Following the new strand in the new trade theory literature that focuses on firm heterogeneity initiated by the Melitz (2003) model, in this paper we investigate the determinants of export activity of firms in three Baltic countries. The study covers Estonia, Latvia and Lithuania and is based on firm level data for the period starting in 2002 and ending in 2009. First, we start with estimating probit regressions for the pooled dataset that includes all Baltic countries, and then we disaggregate the sample into particular countries. Our estimation results obtained for the whole group of Baltic countries indicate that the probability of exporting increases with the higher share of university graduates in productive employment, larger spending on R&D activities, the use of foreign technology licenses, the foreign ownership, the higher productivity and the firm size. The results obtained for particular countries reveal some degree of heterogeneity among those countries. Keywords: Baltic countries, export activity, firm heterogeneity JEL Classification: F10, F12, L11 Introduction In the early 1990s, the Baltic countries faced transition from non-market to market economies, radically liberalized their multilateral and regional trade and integrated successfully into the European division of labor. Thus, those countries became small open economies and firms based in those countries gained the access to foreign markets but at the same time were exposed to i This research has been undertaken under the grant NCN (GR4353). The Authors would like to thank the National Science Centre for financial support. The opinions expressed by authors in the paper do not reflect neither positions of the Faculty of Economics at the University of Warsaw, nor of the European Central Bank. ii University of Warsaw, Department of Economics, 44/50 Długa St., PL-00241 Warszawa, Poland. Corresponding author. E-mail: cieslik@wne.uw.edu.pl iii University of Warsaw, Department of Economics, 44/50 Długa St., PL-00241 Warszawa, Poland iv European Central Bank, DG-Statistics/Monetary and Financial Statistics Division INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 26

increased foreign competition. The majority of previous studies for those countries evaluating trade effects were based on aggregate trade flows and gravity models derived either from the neoclassical or new trade theories. In the recent years, a new strand in the new trade theory literature that stresses the firm heterogeneity in terms of productivity and export performance has emerged. It turns out that only a small fraction of the most productive firms accounts for the majority of exports and most firms do not export concentrating their activities on domestic markets only. The majority of empirical studies based on firm-level data are conducted for particular developed and some developing countries while the empirical evidence for the post-transition economies of Central and Eastern Europe is much less abundant. In particular, the firm-level evidence on export performance for the Baltic economies is still rather scarce and mostly limited to country studies based on firm surveys. Therefore, the main goal of this paper is to complement previous micro-level empirical studies for particular countries with multi-country econometric evidence for the entire group of Baltic countries. In this paper, we study empirically the nexus between productivity and exporting, having controlled for other firm characteristics in three Baltic countries. Our study is based on the BEEPS firm level data for the period starting in 2002 and ending in 2009. First, we start with estimating probit regressions for the pooled dataset that includes the whole group of Baltic countries, and then we disaggregate the sample into particular countries. The structure of this paper is as follows. In Section 2 we review the relevant literature. In Section 3 we discuss the empirical methodology and the data. In Section 4 we discuss our empirical results. Section 5 summarizes and concludes. Literature review The literature on the evolution of trade patterns in the Baltic states is already quite extensive (see e.g. Streimikiene et al., 2016). Several strands in this literature can be identified. One of the strands focuses on the determinants of trade volumes. For example, the early study by Laaser and Schrader (2005) finds that due to the stepwise access to the EU Common Market the focus of their trade relations changed from former intra-soviet trade to intense trade integration with European partners. From the perspective of the neoclassical trade theory, exports of the Baltic countries were dominated by traditional labor-intensive goods. However, this study finds that significant heterogeneity in the patterns of specialization among the Baltic countries exists ranging from Estonia's exports with higher technological content to Lithuanian rawmaterial-intensive exports. www.ieletters.cz 27

The analysis of the general patterns of trade in the Baltic countries was continued in the number of subsequent studies. For example, Bernatonyte and Normantiene (2009) seek to define the nature and pattern of trade specialization in the Baltic States using the basic theories of trade specialization and methods of measurement of inter-industry and intraindustry trade specialization. They analyzed the inter-industry trade using the relative trade advantage index and the pattern of intra-industry trade specialization using the Grubel-Lloyd index. Their analysis showed the relevance of the new trade theory models as the share of intra-industry trade between the Baltic States and the EU had been growing rapidly. The pattern of intra-industry trade at a highly disaggregated level for the new EU members states (NMS), including the Baltic countries, has also been studied by Benkovskis and Rimgailaite (2011) on the basis of methodology proposed by Feenstra and Kee (1994) and extended by Hummels and Klenow (2005) and Broda and Weinstein (2006). To proxy for unobserved relative variety they assumed that the number of brands in each two-digit sector followed a Poisson distribution. According to their calculations, the smallest amounts of brands among NMSs was coming from the smallest countries such as Slovenia, Bulgaria, Romania and the Baltic States, while the bigger countries such as Poland and the Czech Republic exported a larger number of varieties (Strielkowski, 2012). Moreover, the number of exported varieties was increasing over time in all NMSs. Their calculations showed also that NMS exports were of lower quality compared with German exports. The Baltic States and Bulgaria appeared at the lower end of the range, while the highest export quality was observed in Hungary, Poland and the Czech Republic. However, export quality was not homogenous across different industries and countries. Moreover, it was found that all NMSs were able to increase average quality of their exports during the sample period. The highest increase was reported in Romania while the lowest in Latvia. The pattern of specialization and trade has also been studied separately for particular Baltic countries. For example, Saboniene (2009) studied Lithuanian export competitiveness in 2000-2007 using the modified indices of revealed comparative advantage (RCA). She also compared Lithuanian export competitiveness with two other Baltic states: Latvia and Estonia. The results of her analysis showed that Lithuanian export was largely dependent on the goods produced by traditional industries such as animal products, prepared foodstuffs, wood and wood articles, textile articles and furniture. Similarly, Latvia and Estonia retained strong positions in almost all traditional branches of industry. Moreover, her analysis showed that export of medium-high INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 28

technology and medium-low technology industries gradually increased, but the areas of high and low technology industries declined. Subsequently, Benkovskis (2012) analyzed the competitiveness of Latvia s exporters using disaggregated trade data from UN Comtrade. His paper extended the static methodology of Hummels and Klenow (2005) to the dynamic framework. His main contribution to existing empirical literature on trade in the Baltic states was the decomposition of changes in export market shares into intensive and extensive margins. v His study revealed that competitiveness of Latvian exporters, measured by the market share of their products in world trade, almost doubled between 1999 and 2010. This change was mainly driven by the intensive margin, as Latvia s exporters increased their presence on traditional markets. Moreover, the contribution of the extensive margin was also positive due to geographical expansion. Latvia exported also more new products but the intensity of this process was rather modest. Moreover, the share of traditional products exported by Latvia in world imports remained roughly unchanged and the share of traditional geographical destinations of Latvia s products decreased. He postulated further research, inter alia, based on microeconomic determinants of Latvia s competitiveness. This postulate was partly addressed by Putniņš (2013) who employed an international business approach to study the determinants of export competitiveness of Latvian firms. His study was based on a survey over 500 medium-sized Latvian companies that have an established operating record. His sample consisted of companies registered in Latvia that: (i) had annual turnover between EUR 500 thousand and EUR 50 million; (ii) were registered in 2007 or earlier; and (iii) were not from the real estate or financial sectors. He found that most medium-sized Latvian companies were exporters and on average exports constituted more than half of their total turnover. The direct exports of Latvian firms were concentrated on the neighboring Baltic countries while Scandinavia, Germany, Russia, and other EU countries also constituted a substantial proportion of their exports. However, he used mainly descriptive statistics to compare characteristics of exporters with nonexporters. He found that exporters were larger, younger, faster growing and payed higher wages compared to non-exporters. His findings regarding wages were consistent with the view that exporters have higher labor productivity or utilize more skilled labor. Especially, direct exporters tended to be more innovative, proactive and risk taking, and therefore had higher entrepreneurial v The extensive margin refers to the number of exporting firms, products exported and countries served while the intensive margin to the average value of exports per product and exporter across destinations. 29 www.ieletters.cz

orientation. Moreover, foreign-owned companies revealed a higher propensity to export, with 71% of foreign-owned companies exporting, compared to 56% of domestic-owned companies. The Latvian companies that discontinued exporting (58%) according to Putniņš (2013) analysis were unable to sustain international competition. This reinforced his main conclusion that productivity and competitiveness were considerable constraints to Latvian exports. His findings were generally in line with the findings of other empirical studies based on most recent strand in the new trade theory literature that focuses on the relationship between the level of labor productivity and exporting. The new strand in trade theory that was initiated by Melitz (2003) predicts that only most productive firms can become exporters. The large majority of empirical studies found support for the main prediction of his theoretical model, i.e. that more productive firms self-select into foreign markets. This has been demonstrated, for example, by Bernard and Wagner (1997) for German firms, Bernard and Jensen (1999) for US firms, Clerides et al. (1998) for Columbia, Mexico and Morocco, and Castellani (2002) for Italy. vi The relationship between productivity and the probability of exporting in the context of the whole group of Baltic states was studied by Cieślik et al. (2014a,b). In particular, Cieślik et al. (2014a) studied the firm-level determinants of export performance in three groups of countries: the Visegrad, Baltic and Caucasus countries. First, they estimated probit regressions for the pooled dataset that included all three groups of countries, and then they disaggregated the sample into particular country groups to study the differences and similarities between these groups. Their empirical results confirmed the importance of firm characteristics for export performance in these groups of countries. At the same time, they also found significant heterogeneity between these groups in terms of firm-level determinants of export performance. Cieślik et al. (2014b) extended the previous analysis to include also Eastern European and Central Asian countries in addition to the Visegrad, Baltic and Caucasus countries which mostly confirmed the previous findings. More recently, Cieślik et al. (2016) empirically examined the nexus between firm-level productivity and exporting for the Baltic states and the Central European countries. They hypothesized that this nexus should be more vi The survey of early empirical evidence on firm heterogeneity and exporting was provided by Tybout (2003). More extensive summaries of empirical evidence on the relationship between the productivity and export performance were provided by Wagner (2007, 2012). INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 30

pronounced in the Central European countries compared to the Baltic states and formally tested for differences between these two country groups. However, despite some differences in the estimated parameters between the Baltic states and the Central European countries, their formal tests indicated that the results obtained for these two groups of countries were not statistically different. In contrast, the results obtained separately for particular Baltic and Central European countries revealed some degree of heterogeneity among them with respect to the determinants of export performance. In particular, the productivity variable was significantly significant only in the Czech Republic, Hungary and Lithuania. Similarly, the use of foreign technology, foreign ownership, R&D expenditure and the share of university graduates in productive employment were statistically significant in only some countries. Firm size was the only variable that was statistically significant for all countries. Thus, the new strand in international trade theory provides a useful tool for the analysis of export performance. In this paper, we study empirically the nexus between firm-level productivity and exporting postulated by the Melitz (2003) model in the entire group of the Baltic countries as well as for particular countries. In this study, we focus on the determinants of firm decisions to export which is an equivalent of studying the extensive margin effects. vii The main contribution of this paper is to complement previous empirical studies for particular countries with formal econometric evidence for not only the whole group of Baltic states but also for individual countries within this group. In particular, we control for other firm characteristics that may affect export performance such as human capital, the level of internationalization, as well as firm age and firm size. Data description and empirical methodology 3.1. Data description Our analysis is based on "EBRD-World Bank Business Environment and Enterprise Performance Survey (BEEPS)" data collected by the World Bank and the European Bank for Reconstruction and Development in the postcommunist countries located in Europe and Central Asia (ECA) as well as Turkey. The main objective of the BEEPS survey was to obtain feedback from enterprises in the aforementioned countries on the state of the private sector. The survey examined the quality of the business environment as vii The alternative method of studying the changes in the extensive margin in the Latvia was presented by Benkovskis (2012). 31 www.ieletters.cz

determined by a wide range of interactions between firms and the state. The surveys covered manufacturing and services sectors and are representative of the variety of firms according to sector and location within each country. The data were collected for years 2002, 2005, and 2009. Our study focuses on three Baltic countries: Estonia, Latvia, and Lithuania. Along with the Central European countries the Baltic countries were the leaders in multilateral and regional trade liberalization in the early 1990s. Already on 13 September 1993 these three countries signed the agreement establishing the Baltic Free Trade Area (BAFTA) which came into force on 1 April 1994. On 1 January 1997, the BAFTA agreement was extended to cover also trade in agricultural produce. In addition, throughout the 1990s the Baltic countries signed a number of bilateral free trade agreements with the Central European countries. At the same time, the Baltic states started their integration with the European Union by submitting in 1995 the official request for the EU membership. The association agreement with the EU entered in force in 1998 and on 1 May 2004, all the Baltic states joined the European Union. Given the positive changes in the international institutional environment it can be expected that firms from the Baltic countries are also the leaders in export activity. Therefore, it is worth comparing the propensity to export of firms in the Baltic countries with other countries covered by the BEEPS. The export activity is defined as the situation when at least one percent of sales revenue comes from the sales made abroad. In Table 1 we present the export propensity of firms from the Baltic countries and the countries from the region treating Turkey as a benchmark a market economy which was from the region free of the communist past. Table 1 reveals a great degree of heterogeneity across the firms in the postcommunist countries including the Baltic countries. It can be noted that on average, firms residing in Turkey are the most export-oriented amongst the firms in the region. A high share of exporting firms is typical for the countries that emerged from the former Yugoslavia, which traditionally were more market-oriented and had more liberal trade regimes in the past compared to other communist countries. In contrast to the aforementioned countries, the share of exporters amongst all countries firms is the lowest for firms residing in the former Soviet Union, with the exception of the Baltic states. This is in line with the phenomenon often observed, that firms from bigger countries usually sell their products on the domestic markets. However, heterogeneity in export performance even among the Baltic countries cannot be explained only by referring to the country characteristics that are usually stressed by the traditional trade theory. Therefore, it is necessary to study also INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 32

the role of firm characteristics in determining the export performance in line with the recent trend in the new trade theory. Table 1: comparison of the propensity to export among the firms from Baltic countries and other countries in the region as well as Turkey Export (national sales less than or equal 99% of establishment's sales) Country Mean Freq. Turkey 0.5793 2 465 Slovenia 0.5533 685 Croatia 0.4199 1 148 Serbia 0.3733 900 Slovakia 0.3700 654 FYRMacedonia 0.3601 736 Estonia 0.3545 660 Lithuania 0.3544 680 Hungary 0.3516 1 149 Bosnia 0.3460 737 Czech Rep. 0.3454 857 Bulgaria 0.3184 1 853 Latvia 0.2857 651 Albania 0.2746 732 Poland 0.2734 2 008 Belarus 0.2583 848 Moldova 0.2356 887 Ukraine 0.2182 1 902 Romania 0.2135 1 382 Armenia 0.1899 895 Russia 0.1721 2 359 Kyrgyz Rep. 0.1708 609 Georgia 0.1689 746 Montenegro 0.1307 153 Uzbekistan 0.1260 921 Tajikistan 0.1197 735 Azerbaijan 0.1100 900 Kazakhstan 0.1009 1 378 Total 0.2867 29 630 Source: own calculations based on the BEEPS data. The probability of exporting of firms residing in the Baltic countries can related to the explanatory variables on firm and sector characteristics. These variables are based on survey questions regarding the individual characteristics of the firm, sector of activity, legal and economic status, characteristics of www.ieletters.cz 33

managers and the size of the firm, the infrastructure of services in the analyzed country, economic performance and key characteristics of the reviewed firms, as well as stakeholders, e.g. employers organizations, employees organizations, local government, central government, the ICT industry, SMEs, academics, etc. The sample used in our econometric analysis includes cross-section data for almost two thousand observations for firms located in the Baltic countries in all analyzed years. 3.2. Empirical methodology In order to investigate empirically the theoretical relationship between labor productivity and exporting in the context of the Baltic countries we employ the standard probit regression, having controlled for the additional firm characteristics. Building on the previous theoretical literature initiated by the Melitz (2003) model and earlier empirical studies discussed in the previous section, we develop the following empirical model to investigate the effects of various firm characteristics on export performance of firms in the Baltic countries. In particular, we assume that the export status of i-th firm denoted by Y i* can be related to the set of individual firm characteristics in the following way: * Yi X i i (1) where X i which is a vector containing explanatory variables that may affect export performance with the first term equal to unity for all i, is the vector of parameters on these variables that needs to be estimated and i is the error term that is assumed to be independent of X i and normally distributed with a zero mean. However, instead of observing the volume of exports, we observe only a binary variable Y i indicating the sign of Y i *. Thus, our dependent variable follows a binary distribution and takes the value 1 when the firm exports and 0 otherwise: 1 if Y i > 0 Y i = (2) 0 if Y i = 0 We can obtain the distribution of Y i given X i. Hence, the probability that a firm exports can be written as: P(Y i =1 X i) = Φ(X iθ) (3) where Φ( ) denotes the standard normal cumulative distribution function (cdf). INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 34

To be able to successfully implement the probit model, it is important to know how to interpret the vector of estimated parameters on the explanatory variables. Consider a specific explanatory variable x ij, which is an element of vector X i. The partial effect of x ij on the probability of exporting can be written as: P(Y i =1 X i)/ x ij = p(x i)/ x ij (4) When multiplied by Δx ij equation (4) gives the approximate change in P(Y i =1 X i) when x ij increases by Δx ij, holding all other variables constant. Estimation results In this section, we present two sets of our estimation results. First, we discuss the results obtained for the whole group of Baltic countries to provide comparability with the earlier empirical studies. Then, we show the results for the individual Baltic countries. 4.1 Pooled results for the whole Baltic group In columns (1)-(5) of Table 2 we report results that come from the specification that includes the measure of productivity, having controlled for standard factors mentioned in other studies. In column (1) we control for other factors related to human capital that may affect export activity. These factors include the level of the R&D spending (lr&d), the percentage of employees with university degrees (luniv) and the share of skilled workers in total full-time employment (lsk_prod). Our estimation results reveal that the parameter of the measure of productivity displays the positive sign is statistically significant at the 1 per cent level. This means the higher level of productivity is positively related to the probability of exporting. This result is in line with the main prediction of the Melitz (2003) model concerning the positive nexus between productivity and exporting. The estimated parameters on the human capital variables are not statistically significant with the exception of the share of employees with the tertiary degree (luniv) which is statistically significant at the 10 per cent level. The estimated signs of parameters on our explanatory variables are generally in line with the expectations and results of other studies, in particular the study by Putniņš (2013). This means that the higher level of productivity is positively related to the probability of exporting. Moreover, the share of workers with university degrees in total employment is positively related to the probability of 35 www.ieletters.cz

exporting. The role of labor productivity and human capital in export performance is also confirmed in the study by Putniņš (2013). Table 2: Baseline results for the whole Baltic group. VARIABLES (1) (2) (3) (4) (5) lprod 0.0813*** 0.0719*** 0.0491* 0.0355 0.2240** (0.019) (0.019) (0.029) (0.029) (0.113) lsk_prod 0.0218 0.0222 0.0603* 0.0583* 0.0644** (0.021) (0.022) (0.031) (0.031) (0.032) luniv 0.0040* 0.0060*** 0.0049 0.0059* 0.0048 (0.002) (0.002) (0.003) (0.003) (0.003) lr&d 0.0440 0.0680 0.1030 0.1700* 0.1820* (0.051) (0.052) (0.087) (0.096) (0.099) lage -0.0070** -0.0059-0.0049-0.0047 (0.003) (0.004) (0.005) (0.005) medium 0.5040*** 0.5330*** 0.4200** 0.4030** (0.130) (0.177) (0.182) (0.183) large 1.040*** 1.1450*** 1.0130** * 0.9550** * (0.150) (0.218) (0.223) (0.226) foreign_cap 0.0094*** 0.0092** * 0.0096** * (0.003) (0.003) (0.003) foreign_tech 0.7230** -0.0719-0.0821 (0.307) (0.383) (0.387) d_latvia 1.3540* (0.781) d_lithuania 0.9420** (0.437) Constant -1.395*** -1.637*** -1.619*** -0.482-3.577** (0.262) (0.281) (0.394) (0.501) (1.810) time dummies no no no yes yes Observations 581 581 351 351 351 Log likelihood -383.7-358.4-194.3-186.9-184.2 Pseudo R2 0.032 0.096 0.190 0.221 0.232 Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1. Source: Own results INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 36

In column (2) additionally we control for the age (large) and the size of the firm (medium and large). viii All these additional control variables are statistically significant although at different levels of statistical significance. In particular, the firm age variables is statistically significant at the 5 per cent level, while both firm size variables are statistically significant already at the 1 per cent level indicating the importance of economies of scale for exporting. Surprisingly, in contrast to studies for other EU countries the estimated parameter on the age variable displays a negative sign. However, this finding is in line with the recent study by Putniņš (2013) for Latvian firms who finds that younger firms are more dynamic and export oriented. This result may apply also to the exporting firms in other Baltic countries that were established relatively recently after the change of the economic system. The estimated parameters on the firm size variables display positive signs and are in line with our expectations and the results of other studies. This is the standard result in the empirical studies on firm heterogeneity. This result is in line with the results reported by Putniņš (2013) who found significant heterogeneity even among the medium sized firms. Moreover, it can be noted that the estimated parameters for larger firms is almost twice as high as the estimated parameter for the medium sized firms. It means the largest firms reveal the highest propensity to export. These results are in line with the results of other studies showing that there is a high concentration of exports among the largest firms. The inclusion of the additional control variables does not affect the statistical significance of the productivity variables but increase the statistical significance of the share of employees with the tertiary degree (luniv) which becomes statistically significant at the 1 per cent level. In column (3) we control for the internationalization of the firm by including two additional variables measuring the openness of the firm to the foreign technology licensed from a foreign-owned company (foreign_tech) and the foreign capital (foreign_cap). Both variables are statistically significant although at different levels of statistical significance. The foreign ownership of the firm (foreign_cap) is statistically significant at the 1 per cent level, while the use of foreign technology (foreign_tech) is statistically significant at the 5 per cent level. Both variables display the expected positive signs which means that the probability of exporting increases with the internationalization of the firm. The inclusion of these variables affects the level of statistical significance of the productivity variable and human capital variables. In particular, the viii The benchmark dummy variable for the size of the firm is the small firm (small). The estimated coefficient on this variable is not reported due to the full collinearity of this variable with two other firm size variables. 37 www.ieletters.cz

statistical of the productivity variable drops to the 10 per cent level. The share of employees with the tertiary degree (luniv) loses completely its previous statistical significance while the estimated parameter on the share of skilled workers in total full-time employment (lsk_prod) becomes statistically significant the 10 per cent level. Moreover, the estimated parameter on the age variable becomes statistically not significant. In column (4) we investigate the robustness of our results by controlling for individual time effects for particular years of the sample. The inclusion of time dummies affects the statistical significance of both the productivity variable and control variables. In particular, the productivity variable and the use of foreign technology (foreign_tech) lose completely their previous statistical significance. Moreover, two measures of human capital: the level of the R&D spending (lr&d) and the percentage of employees with university degrees (luniv) become statistically significant at the 10 per cent level. The statistical significance of the medium firm size variable drops to the 5 per cent level. In column (5) control for both time and country specific effects for Latvia and Lithuania while Estonia is treated as a point of reference. The estimated parameters on both dummy variables display positive signs and are statistically significant although at different levels of statistical significance. In particular, the estimated parameter for the Latvia dummy is statistically significant at the 10 per cent level, while the estimated parameter for the Lithuania dummy is statistically significant at the 5 per cent level. This suggests that firms from Latvia and Lithuania are more export oriented compared to firms from Estonia. Moreover, the magnitude of the estimated coefficient for Latvia is higher than for Lithuania. The inclusion of the country dummies affects the statistical significance of both the productivity variable and human capital variables. In particular, the productivity variable regains its statistical significance at the 5 per cent level. Moreover, the share of skilled workers in total full-time employment (lsk_prod) becomes statistically significant at the 5 per cent level while the percentage of employees with university degrees (luniv) loses its previous statistical significance. 4.2. Results for individual Baltic countries In Table 3 we show the estimation results obtained for the individual Baltic countries. In column (1) we display the estimation results for Estonia. These results are very different from the results obtained for the whole sample of the Baltic countries. In particular, the estimated coefficient on the productivity variable displays a positive sign but it is not statistically significant. Similarly, all the measures of human capital (R&D spending, skilled labour and university education) are INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 38

not statistically significant. Also, the age of the firm is not statistically significant. In contrast to the aforementioned variables, both firm size and firm internationalization variables display positive sign and are statistically significant although at different levels of statistical significance. Table 3: Results for individual Baltic countries. VARIABLES Estonia (1) Latvia (2) Lithuania (3) Lprod 0.1160-0.0227 0.5490*** (0.220) (0.225) (0.196) lsk_prod 0.0607 0.2860 0.0443 (0.061) (0.243) (0.047) Luniv -0.0031 0.0153** 0.0006 (0.007) (0.006) (0.005) lr&d 0.0190-0.4300* 0.3810** (0.170) (0.259) (0.153) Lage 0.0149-0.0095-0.0143* (0.010) (0.010) (0.008) Medium 0.9320*** 0.3220 0.3390 (0.345) (0.363) (0.302) Large 1.0910** 0.4470 1.5240*** (0.462) (0.490) (0.356) foreign_cap 0.0070* 0.0132** 0.0286*** (0.0042) (0.0061) (0.011) foreign_tech 1.4370** -0.3000 0.1660 (0.629) (0.789) (0.534) Constant -3.229-2.020-7.437*** (3.462) (2.183) (2.477) Observations 99 108 144 Log likelihood -50.12-53.58-68.46 Pseudo R2 0.269 0.204 0.313 Note: Standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1. Source: Own results This is line with empirical studies for other EU countries. In particular, the dummy variable for medium sized firms is statistically significant at the 1 per cent level, while the dummy variable for large firms is statistically significant at the 5 per cent level. However, the magnitudes of the estimated parameters on both variables are very similar. Moreover, the foreign ownership of the firm 39 www.ieletters.cz

(foreign_cap) is statistically significant at the 10 per cent level, while the use of foreign technology (foreign_tech) is statistically significant at the 5 per cent. In column (2) we report the estimation results for Latvia. Like in the case of Estonia the estimated coefficient on the productivity variable is not statistically significant. This means that the level of productivity does not seem to be important for the probability of exporting. Similarly, the estimated coefficient on the share of skilled workers in total full-time employment (lsk_prod) is not statistically significant. In contrast, two other measures of human capital are statistically significant although at different levels of statistical significance. The estimated coefficient on the percentage of employees with university degrees (luniv) is statistically significant at the 5 per cent and displays the expected positive sign. The estimated coefficient on the level of the R&D spending (lr&d) is statistically significant at the 10 per cent level but displays an unexpected negative sign. The estimated parameters on the age and the size of the firm are not statistically significant. The foreign ownership variable (foreign_cap) is statistically significant at the 5 per cent level, while the use of foreign technology (foreign_tech) is not statistically significant. These results demonstrating the role of human capital and foreign capital in exporting are in line with the results reported by Putniņš (2013). In column (3) we show estimation results for Lithuania. In contrast to the results obtained for Estonia and Latvia the level of productivity is statistically significant already at the 1 per cent level. These results confirms the main prediction of the Melitz (2003) model concerning the positive nexus between productivity and exporting. Among the human capital variables only the level of the R&D spending (lr&d) is statistically significant at the 5 per cent level and displays an expected positive sign. The estimated coefficient on the age variable is statistically significant at the 10 per cent level and displays a negative sign. Among the firm size variables only the dummy variable for the large firms is statistically significant at the 1 per cent level. This means that the exporting activity is highly concentrated among the largest firms. Like in the case of the other two Baltic countries the estimated parameter on the foreign ownership variable (foreign_cap) is statistically significant but at the 1 per cent level. Moreover, the magnitude of this parameter is the highest among all Baltic states. Finally, like in the case of Latvia the variable measuring the use of foreign technology (foreign_tech) is not statistically significant. Conclusions INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 40

In this paper, we investigated the determinants of export activity of firms in the Baltic countries. The study covered Estonia, Latvia, and Lithuania and was based on firm level data for the period starting in 2002 and ending in 2009. First, we performed a descriptive analysis comparing the export performance of the firms from particular Baltic countries with the export performance of firms from other countries in the region of Central and Eastern Europe. We demonstrated that firms from the Baltic countries were on average less export oriented compared to firms from countries of a similar size located in Central Europe such as Slovenia or Slovakia. Then, we estimated probit regressions for the pooled dataset that included all Baltic countries, and later we disaggregated the sample into particular countries. In our study, we focused on the determinants of firm decisions to export which was an equivalent of studying the extensive margin effects. This analysis complemented the previous study for Latvia made by Benkovskis (2012) and for all Baltic states by Cieślik et al. (2016). Our estimation results obtained for the whole sample indicated that the probability of exporting increases with the higher level of productivity, the measures of human capital, firm size, foreign ownership and the use of foreign technology. These results are generally in line with the results obtained by Putniņš (2013) for Latvian firms and the results for all Baltic states reported in Cieślik et al. (2016). The results obtained for particular Baltic countries revealed a great degree of heterogeneity among those countries clearly showing that the Baltic countries do not constitute a homogenous group. The key result on the positive exporting-productivity nexus was driven mainly by the results obtained for Lithuania, while in the case of Estonia and Latvia there seems to be no clear relationship between the level of productivity and the probability of exporting. This result is in line with the results reported in Cieślik et al. (2016). Therefore, the predictions of the Melitz (2003) model were only partly confirmed for the group of Baltic countries. Instead, the export performance of Estonian firms was based mainly on foreign knowledge and domestic human capital played no role while in the case of Latvia export performance was related to the domestic human capital proxied by the percentage of employees with university degrees. The size of the firm was statistically significant in the majority of countries, i.e. Estonia and Lithuania. The only explanatory variable that was statistically significant in all Baltic countries was the foreign ownership. This result is also supported by the study of Putniņš (2013) for Latvia. This means that more foreign direct investment is needed to improve export performance of firms in the Baltic countries. However, the estimation 41 www.ieletters.cz

results for individual Baltic countries must be treated with cautions due to a small sample of firms. References Benkovskis, K., (2012). Competitiveness of Latvia s Exporters. Baltic Journal of Economics, 12(2), 17-45. Benkovskis, K., Rimgailaite, R. (2011). The Quality and Variety of Exports from the New EU Member States: Evidence from Very Disaggregated Data. Economics of Transition, 19(4), 723 747 Bernatonyte, D., Normantiene, A. (2009). Estimation of Trade Specialization: the Case of the Baltic States. Inzinerine Ekonomika-Engineering Economics, 2(2), 7-17. Bernard, A, Eaton, J, Jensen, J., Kortum, S. (2003). Plants and Productivity in International Trade. American Economic Review, 93(4), 1268-90. Bernard A.B., Jensen J.B. (1999). Exceptional Exporter Performance: Cause, Effect, or Both? Journal of International Economics, 47(1), 1-26. Bernard A.B., Wagner, J. (1997). Exports and Success in German manufacturing. Weltwirtschaftliches Archiv, 133(1), 134-57. Broda, Ch., Weinstein, D., (2006). Globalization and the Gains from Variety. Quarterly Journal of Economics, 121(2), 541-585. Business (The) Environment and Enterprise Performance Survey (BEEPS) 2008-2009. A Report on methodology and observations, World Bank April 2010. Castellani, D. (2002). Export Behavior and Productivity Growth: Evidence from Italian Manufacturing Firms. Review of World Economics, 138(4), 605-628. Cieślik, A., Michałek, J.J. Michalek, A., Mycielski, J. (2016). Globalization, international trade, and human development: a case of Central and Eastern Europe. Czech Journal of Social Sciences, Business and Economics, 5(2), 6-15. Cieślik, A., Michałek J.J., Michałek A. (2014a). The Influence of Firm Characteristics and Export Performance in Central and Eastern Europe: Comparisons of Visegrad, Baltic and Caucasus States. Entrepreneurial Business and Economics Review, 2 (1), 7-18. Cieślik, A., Michałek, J.J., Michałek, T. (2014b). Firm Characteristics and Export Performance. In: N. Daszkiewicz and K. Wach (eds.). Firm-level Internationalisation and its Business Environment: Knowledge-based and Entrepreneurial Approach, Gdańsk University of Technology Publishing House, Gdańsk, 34-44. INTERNATIONAL ECONOMICS LETTERS University Service Publishing ISSN: 1805-7306 42

Clerides, S.K., Lach, S., Tybout, J.R. (1998). Is learning be Exporting Important? Micro-dynamic Evidence from Colombia, Mexico and Morocco. Quarterly Journal of Economics, 113, 903-48. European Bank for Reconstruction and Development EBRD (2010). Transition Report 2010. Recovery and Reform, European Firms in a Global Economy: EFIGE (2010). The Global Operations of European Firms. The second EFIGR Policy Report, Bruegel. Feenstra, R., Kee, H.L., (2004). On the Measurement of Product Variety in Trade. American Economic Review, 94(2), 145-149. Fernandes, A., Isgut, A. (2005). Learning-by-Doing, Learning-by-Exporting, and Productivity: Evidence from Columbia. World Bank Policy Research Working Paper 3544. Helpman, E., Melitz, M., Yeaple, S. (2004). Export versus FDI with Heterogeneous Firms. American Economic Review, 94(1), 300-316. Hummels, D., Klenow, P.J., (2005). The Variety and Quality of a Nation's Exports. American Economic Review, 95(3), 704-723. Laaser, C.L., Schrader K., (2005). Baltic Trade with Europe: Back to the Roots? Baltic Journal of Economics, 5(2), 15-37. Loecker, J. De (2007). Do Exports Generate Higher Productivity? Evidence from Slovenia. Journal of International Economics, 73, 69-98. Melitz, M. (2003). The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity. Econometrica, 71(6), 1695 1725. Putniņš, T.J. (2013). Exporting by Latvian companies: Vitality, drivers of success, and challenges. Baltic Journal of Economics, 13(2), 3-33. Saboniene, A. (2009). Lithuanian Export Competitiveness: Comparison with other Baltic States. Inzinerine Ekonomika-Engineering Economics, 2, 49-57. Strielkowski, W. (2012). Factors that determine success of small and medium enterprises. the role of internal and external factors. Journal of Applied Economic Sciences, 7(3), 334-350. Streimikiene, D., Strielkowski, W., Bilan, Y., & Mikalauskas, I. (2016). Energy dependency and sustainable regional development in the Baltic States-a review. Geographica Pannonica, 20(2), 79-87 Tybout, J.R. (2003). Plant and Firm Level Evidence on New Trade Theories. In: Choi E. Kwon, J. Harrigan (eds.) Handbook of International Trade, Oxford, Basil Blackwell. Wagner, J. (2007). Exports and Productivity: A Survey of the Evidence from Firm-Level Data. World Economy, 30(1), 60 82. Wagner, J. (2012). International Trade and Firm Performance: A Survey of Empirical Studies since 2006. Review of World Economy, 148, 235 267. WTO (2008). World Trade Report 2008. Trade in a Globalizing World. World trade Organization. 43 www.ieletters.cz