The level of globalization in the world economy has been steadily increasing. Is it really a common phenomenon, in which all economic sectors and companies in a country participate? Is the opening-up of economies, seen from a microeconomic perspective, characterized by similar mechanisms in different countries? Do firms from MENA countries, where trade liberalization has taken place relatively recently, behave as expected, according to existing trade theories? Do they behave like firms in OECD countries (including Turkey and Israel)? To this end, in this paper, we would like to verify several theoretical hypotheses, based on empirical testing, using microeconomic techniques to analyse firm-level data.
The main objective of this paper is to verify whether or not the transition of firms in the selected MENA countries in order to meet the requirements of globalized market economies has been completed. In particular, we analyse and compare the export behaviour of firms from the selected MENA countries with that of firms from CEE, Israel and Turkey. We want to treat firms from CEE, Israel and Turkey as the benchmark, since there are many similarities, in terms of transition, between the countries selected (MENA and CEE countries, Turkey and Israel). Moreover, recent reports (e.g., IMF 2014) demonstrate that the transition process in CEE countries has already been completed successfully. The export performance of CEE firms in various regions has already been analysed, compared to the behaviour of firms in the EU-15 countries, by academics who are part of this project (e.g., Cieślik, Michałek and Michałek 2014; Cieślik et al. 2015).
In particular, the main goal of the present study is the empirical verification of the main hypotheses resulting from the Melitz (2003) model, which represents a positive relationship between the productivity of companies and their involvement in export activities using firm-level data for the selected MENA countries. In the course of the study, we will endogenize the productivity of companies, among others, by referring to their spending on R&D, human capital, new technologies and their size. We will also control for ownership characteristics and the significance of corruption. This analysis will enable the identification of key factors influencing the export competitiveness of individual firms, as well as shed some light on the entire economies of MENA countries. Finally, we will try to determine to what extent the determinants of export behaviour of companies in this region are similar to specific firms operating in the group of more developed CEE countries, as well as Turkey and Israel.
This will allow for the empirical verification of key relationships, as described in the theoretical model of Melitz (2003), for companies in MENA countries. The aggregated analysis will cover the following eight MENA countries: Egypt, Israel, Jordan, Lebanon, Morocco and Tunisia, Turkey, and the West Bank and Gaza. We will study the behaviour of firms in selected larger MENA countries, for which the number of observations was sufficient for the econometric analysis. Special attention will be devoted to Egypt, for which we have the largest number of observations. We will also separately analyse the behaviour of firms in Israel and Turkey, which are the most developed countries among the analysed countries.
In the paper, we will undertake the following research questions, resulting directly from the theory:
Does the selection mechanism operate as described in the Melitz (2003) model, i.e., is it true that only the most productive companies are able to export their products and that less productive firms only sell them in the domestic market? To what extent is productivity determined by innovation activities (including expenditure on R&D), and to what extent by other factors? Do economies of scale, as measured by the number of employees, have a significantly positive impact on the exports of the analysed companies? What is the impact of the quality of human capital (skilled workers, university graduates) available to companies on their export competitiveness? Does the internationalization of firms facilitate export performance (relationships with licensors, parent companies, foreign investment)? Is the set of factors relevant to the export development for firms of MENA countries different from the factors affecting the competitiveness of companies from Turkey, Israel, CEE and other EU countries? Do firms in the MENA region have different propensities towards exports? What should be the role of government in raising the export competitiveness and internationalization of domestic enterprises?
The empirical study will enable us to obtain comparable results on the potential role of various firm characteristics in the selected MENA countries and for particular countries, as well as formulate conclusions concerning policy recommendations aimed at increasing the export competitiveness of economies and the creation of an appropriate institutional environment. The originality of our recommendations comes from the fact that no direct export promotion is needed, provided that the appropriate structural policies are adopted. The recommendations resulting from the analysis can be treated as an indirect way of promoting exports.
This paper is organized as follows. In Section 1, we survey the relevant literature on the determinants of export performance. In Section 2, we describe the research methodology and the data set. In Section 3, we report our empirical results. Finally, the last section summarizes and concludes the paper with policy guidelines and directions for future research.
The empirical literature on firm heterogeneity commenced with the work of Bernard, Jensen and Lawrence (1995), for the US, and Clerides
The majority of empirical studies support the theoretical prediction of the Melitz model, i.e., that more productive firms self-select themselves into foreign markets. A survey on early empirical evidence on the relationship between firm productivity and exporting was provided by Tybout (2003), while extensive summaries of more recent empirical evidence for this relationship in particular countries were offered by Wagner (2007; 2012). According to the first survey by Wagner (2007), a large number of studies using data from different countries shows that exporters and importers are more productive than non-exporters and non-importers. In particular, his review provides clear-cut evidence in favour of the self-selection hypothesis, while arguing that future exporters tend to be more productive than future non-exporters in the years before they enter the export market and often have higher
This picture was largely confirmed in a more recent survey by Wagner (2012); i.e., his review provides extensive evidence in favour of the self-selection hypothesis. It has also been suggested that the empirical results, which result from the learning-by-exporting hypothesis, may not be robust with respect to the specific methodologies and data sets. In particular, the learning-by-exporting hypothesis was confirmed for some countries in early studies, such as those by Isgut (2001) for Colombia, Blalock and Gertler (2004) for Indonesia, and Alvarez and Lopez (2005) for Chile. However, more recent firm-level evidence fails to support this hypothesis. In particular, a lack of evidence for the learning-by-exporting hypothesis has been reported by Arnold and Hussinger (2005) for Germany, Damijan and Kostevc (2006) for Slovenia, Pisu (2008) for Belgium, and Smeets and Warzynski (2010) for Denmark.
More recent studies focus on the role of product mix in exporting. Examples include Bernard, Redding and Schott (2011), Eckel and Neary (2010) and Mayer, Melitz and Ottaviano (2014). Bernard, Redding and Schott (2011), argue that product switching for US firms is correlated with both firm and firm-product attributes, and that product adding and dropping induce large changes in firm scope. Eckel and Neary (2010) examine how globalization affects the scale and scope of multi-product firms. Their model identifies a new source of gains from trade: productivity increases as firms concentrate on their core competence. Finally, Mayer, Melitz and Ottaviano (2014) theoretically and empirically show that, for French firms, tougher competition in an export market induces a firm to skew its export sales toward its best performing products.
Empirical evidence for the relationship between productivity and exporting, based on multi-country firm-level data sets, is still rather scarce. In the literature, the number of empirical studies devoted to the verification of hypotheses derived from the Melitz model for the broader group of MENA and CEE countries is limited. According to Wagner (2012, 261): “Any attempt to extract information on the size of the effects – the economic relevance, not the statistical significance – is hindered by the absence of a reasonably high degree of comparability across the studies. This lack of comparability is due to differences in the unit of analysis (establishment vs. enterprise), the sampling frame (all firms vs. firms with a number of employees above a certain threshold only), the specification of the empirical models estimated and the econometric methods applied.”
One of the first multi-country studies was the European Firms in the Global Economy (EFIGE) report (2010), which was the outcome of an international research project based on comparable firm-level data from several EU countries. The results of this project confirmed the importance of firms’ productivity in exporting. In this report, it was demonstrated that firms’ export performance in seven EU countries was dependent on labour productivity, as well as other firm characteristics. The study showed that, in all countries, exporting firms were, on average, more productive and bigger than non-exporters. Moreover, the study showed that the probability of exporting increased with firm age, the share of university graduates in total employment, R&D spending, and foreign ownership.
The analysis of differences in firm productivity in the selected MENA countries has been studied in the recent FEMISE report (2015). There is also a recent study of export performance of firms in the MENA region (Fakih and Ghazalian 2014). The authors analysed the significance of some firm characteristics, such as size, age, share of skilled workers and type of ownership, in relation to their export performance.
In our study, we focus on testing the hypothesis concerning the impact of productivity differences on the probability of exporting in MENA countries. In addition, we will try to take into account other firm characteristics that may affect export performance, such as R&D spending, foreign technology licences and foreign ownership of firms.
In contrast to the majority of previous studies conducted for selected countries, which were carried out using national data from national statistical offices, our study will be based on a comparable database collected as part of a joint project between the EBRD and World Bank BEEPS (2013), which includes the selected MENA countries (Algeria, Egypt, Jordan, Lebanon, Morocco and Tunisia), all countries of CEE, as well as Israel and Turkey. The BEEPS data offer information on various characteristics of firms. This allows us to analyse the role of the variables described by the Melitz (2003) model, as well as other characteristics of firms describing their innovation, the use of various forms of human capital and the use of new technologies that may affect their productivity and thus export activity.
The results of our empirical study should allow us to formulate policy recommendations, aimed at increasing the export competitiveness of firms and whole economies, and the creation of an appropriate institutional environment promoting international trade. In particular, it should highlight the most important firm-level characteristics affecting exports in the analysed countries and which policies can help to stimulate those exports.
In this paper, we empirically investigate the determinants of export activity of firms in selected MENA countries. The analysis is conducted in the light of a new strand in the theory of international trade, originating mainly from the theoretical Melitz (2003) model. The empirical study is based on micro-econometric modelling. First, to estimate the relationship between the characteristics of firms and the probability of exports, we use a probit regression. In the case of probit regressions, our dependent variable, which describes the export activities of the company in a given year, is a binary variable taking the value of 1 in the case of positive exports for the company in a given year, and 0 in the absence of exports for the company in a given year.
In our model, the probability of exports in the
The key explanatory variable stressed by the Melitz (2003) model is labour productivity, which is expressed as the total amount of annual sales per full-time employee (
Description of variables used in the empirical study The role of product mix for exporters is analysed by Mayer, Melitz and Ottaviano (2014). This variable is only used in the robustness test in the Appendix. Source: BEEPS data set.Variable Description Dependent binary variable, which takes the value of 1 if the establishment is exporting (directly or indirectly) at least 1% of its sales, and 0 otherwise Logarithm of productivity expressed as the total amount of annual sales per full-time employee Number of years since the start of operation Binary variable, which takes value of 1 if the establishment was spending money on R&D in the last three years, and 0 otherwise Logarithm for the percentage of employees at the end of the fiscal year with a university degree Logarithm for the number of permanent, full-time employees in this firm at the end of the last fiscal year Binary variable, which takes the value of 1 if its shares are owned by private foreign individuals, companies or organizations, and 0 otherwise Binary variable, which takes the value of 1 if the establishment uses technology licensed from a foreign-owned company, and 0 otherwise multi 100 minus the share of the main product in total sales. This variable measures whether the firm is producing many (multiple) products (0 means that the main product represents 100% of supply) The percentage of a firm owned by the state/government Corruption: its perception as an obstacle to current operations of the establishment (0 stands for no obstacles and 4 for a very serious obstacle)
Summary statistics for variables used in the empirical study for the whole sample of 22,449 observations Source: BEEPS data set.Variable Observations Mean Std. dev. Min. Max. 22,220 16.261 13.591 0 190 22,260 0.109 0.312 0 1 21,271 32.908 30.903 0 100 22,274 76.287 338.188 1 21,000 22,449 0.085 0.279 0 1 22,220 0.131 0.337 0 1 22,181 0.766 7.115 0 99 21,588 1.456 1.508 0 4
We use the probit model to study the relationship between labour productivity and exporting, having controlled for other firm characteristics. Building on the previous theoretical literature, we develop an empirical model to investigate the effects of various firm characteristics on their export performance. Our variable follows:
where
Instead of observing the volume of exports, we only observe a binary variable indicated by the sign of
The probability that a firm is involved in exports as a function of firm, industry and country characteristics can be written as:
Our study is based on the BEEPS V data collected by the EBRD and the World Bank in post-communist countries located mainly in Europe and Central Asia (ECA) and MENA countries. 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. The survey questions concerned the identification of firms, sectors of activity, legal and economic status, characteristics of managers, firms’ size, the infrastructure of services in the analysed countries, the economic performance and key characteristics of reviewed firms, and stakeholders.
The sample includes data covering the period 2011–2014. Almost 60% of surveys of firms in every country were conducted in 2013. The numbers of observations (surveys) per year were as follows: 2,884 in 2011, 1,833 in 2012, 13,435 in 2013 and 4,287 in 2014. The total number of observations was 22,449. The only exception was Albania. The details concerning the sampling methodology are explained in the Sampling Manual available at For example, Stiebale (2011) argues that there is a potential problem of endogeneity between the measure of export performance and financial variables. However, in our estimation equations, we do not include the measures of financial constraints.
The variables expressed in local currencies (e.g., sales) were converted into a common currency (US dollars). The BEEPS surveys covered both manufacturing and services sectors and are representative of the variety of firms according to sector and location within each country. The number of firms operating in the service sector was relatively small compared to the manufacturing sector. Therefore, it was not possible to perform estimations separately for the manufacturing and service sectors. Moreover, particular industries within each sector can differ with respect to their capital intensity and export performance. Therefore, to control for heterogeneity across industries in our estimations, we used industry-specific effects in addition to individual firm characteristics.
In all countries where a reliable sample frame was available, the sample was selected using stratified random sampling. Therefore, we used the standard probit procedure for the pooled cross-sectional data set without controlling for individual firm effects. However, we controlled for country-specific and sector-specific effects. The list of countries in our sample is presented in Tab. A in the Appendix. In the majority of cases, the data include about 250-350 observations per country. The largest samples of firms are available for Russia (4,220), Egypt (2,897), Turkey (1,334) and Ukraine (1,002). The summary statistics for variables used in our empirical study are presented in Tab. 2.
In this section, we present three sets of our estimation results. First, in Tab. 3, we report the results obtained for the joint sample of both CEE and MENA countries included in the BEEPS V database, then for the whole group of eight MENA countries and compare them with the estimation results obtained for the whole group of CEE countries. Then, in Tab. 4, we report the estimation results obtained for various subgroups of CEE countries, in particular, in terms of the split of the group into two subgroups: EU members and non-EU members. We also present the estimation for six MENA countries (excluding Turkey and Israel). Finally, in Tab. 5, we report the estimation results obtained separately for the individual MENA countries.
Results of probit estimations for the entire sample of countries, MENA countries and all CEE countries p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.05 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.05 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.05 p<0.05 p<0.05 p<0.1 p<0.1 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 p<0.01 Robust standard errors in parentheses,All countries MENA-8 All CEE VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) -0.012 0.0414 0.070 0.049 0.044 0.053 -0.025 0.039 0.091 (0.004) (0.009) (0.020) (0.007) (0.013) (0.037) (0.006) (0.012) (0.021) 0.007 0.002 0.002 0.002 0.002 0.003 0.008 0.001 0.000 (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) 0.599 0.492 0.439 0.631 0.502 0.509 0.620 0.508 0.415 (0.033) (0.035) (0.057) (0.058) (0.060) (0.147) (0.042) (0.045) (0.049) -0.027 0.021 0.031 -0.006 0.024 0.032 -0.044 0.017 0.030 (0.005) (0.005) (0.006) (0.007) (0.008) (0.009) (0.006) (0.007) (0.009) 0.253 0.286 0.265 0.308 0.333 0.306 0.203 0.255 0.242 (0.009) (0.010) (0.025) (0.015) (0.016) (0.040) (0.013) (0.014) (0.025) 0.596 0.495 0.469 0.452 0.487 0.476 0.707 0.509 0.466 (0.040) (0.043) (0.065) (0.070) (0.074) (0.150) (0.051) (0.054) (0.057) 0.330 0.267 0.250 0.431 0.245 0.244 0.316 0.285 0.258 (0.033) (0.035) (0.062) (0.060) (0.063) (0.161) (0.041) (0.043) (0.057) -0.007 -0.005 -0.004 -0.006 -0.005 -0.004 -0.005 -0.004 -0.003 (0.002) (0.002) (0.002) (0.004) (0.004) (0.003) (0.002) (0.002) (0.002) -0.008 0.012 0.018 -0.043 0.019 0.016 -0.005 0.010 0.021 (0.008) (0.009) (0.011) (0.012) (0.014) (0.010) (0.011) (0.012) (0.016) Constant -1.582 -2.270 -2.474 -2.217 -2.451 -2.419 -1.333 -2.483 -2.593 (0.066) (0.133) (0.279) (0.108) (0.185) (0.561) (0.088) (0.203) (0.238) Country effects No Yes Yes No Yes Yes No Yes Yes Sectoral effects No No Yes No No Yes No No Yes Observations 16,113 16,113 16,113 5,453 5,453 5,453 10,237 10,237 10,237 Log likelihood -7846 -7078 -6730 -2925 -2682 -2617 -4605 -4223 -3924 Pseudo R2 0.126 0.211 0.250 0.151 0.221 0.240 0.115 0.189 0.246
Results of probit estimations for CEE members of the EU, CEE non-members of the EU and for six MENA countries (Turkey and Israel excluded) p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.05, p<0.05, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.05, p<0.05, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.05, p<0.05, p<0.05, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, p<0.01, Robust standard errors in parentheses,CEE EU members CEE non-EU members Six MENA countries VARIABLES (1) (2) (3) (4) (5) (6) (7) (8) (9) lprod 0.009 0.027 0.088 -0.004 0.045 0.089 0.074 0.055 0.074 (0.013) (0.021) (0.027) (0.007) (0.015) (0.028) (0.008) (0.017) (0.044) age -0.002 -0.003 -0.004 0.009 0.003 0.002 0.001 0.000 0.001 (0.002) (0.003) (0.004) (0.001) (0.002) (0.001) (0.001) (0.001) (0.002) R_D 0.607 0.561 0.444 0.588 0.484 0.406 0.601 0.456 0.450 (0.078) (0.080) (0.113) (0.051) (0.054) (0.060) (0.068) (0.071) (0.070) luni 0.011 0.020 0.034 -0.052 0.017 0.030 0.002 0.012 0.026 (0.010) (0.010) (0.010) (0.008) (0.010) (0.014) (0.009) (0.010) (0.013) lsize 0.237 0.248 0.234 0.204 0.258 0.247 0.304 0.342 0.321 (0.024) (0.025) (0.027) (0.015) (0.016) (0.034) (0.017) (0.018) (0.043) fo 0.579 0.550 0.512 0.640 0.482 0.445 0.435 0.378 0.347 (0.087) (0.088) (0.074) (0.064) (0.068) (0.079) (0.076) (0.079) (0.128) folicences 0.173 0.180 0.116 0.359 0.325 0.303 0.363 0.380 0.411 (0.076) (0.078) (0.081) (0.049) (0.052) (0.073) (0.078) (0.079) (0.159) share_gov -0.016 -0.014 -0.016 -0.003 -0.002 -0.002 -0.007 -0.007 -0.006 (0.007) (0.007) (0.007) (0.002) (0.002) (0.002) (0.004) (0.004) (0.002) corruption -0.054 -0.022 -0.012 0.020 0.022 0.032 -0.028 0.010 0.013 (0.020) (0.022) (0.016) (0.013) (0.014) (0.019) (0.015) (0.016) (0.013) Constant -1.354 -1.851 -2.160 -1.753 -2.610 -2.648 -2.625 -2.533 -2.695 (0.169) (0.277) (0.307) (0.110) (0.241) (0.325) (0.124) (0.222) (0.727) Country effects No Yes Yes No Yes Yes No Yes Yes Sectoral effects No No Yes No No Yes No No Yes Observations 2,472 2,472 2,472 7,765 7,765 7,765 4,277 4,277 4,277 Log likelihood -1404 -1371 -1161 -3083 -2845 -2721 -2171 -2033 -1974 Pseudo R2 0.107 0.128 0.261 0.116 0.184 0.220 0.153 0.206 0.230
Results of probit estimations for individual MENA countriesEgypt Turkey Israel Tunisia Morocco VARIABLES 1 2 3 4 5 6 7 8 9 10 0.144*** 0.158*** 0.021 0.0212 0.140 0.139 -0.089* -0.009 -0.090 -0.059 (0.026) (0.027) (0.025) (0.0254) (0.088) (0.091) (0.049) (0.055) (0.055) (0.058) 0.003 0.006** 0.013*** 0.0104** 0.013*** 0.014*** -0.009** -0.008** -0.003 -0.000 (0.002) (0.002) (0.004) (0.00454) (0.004) (0.005) (0.004) (0.004) (0.005) (0.005) 0.512*** 0.539*** 0.037 0.0205 1.793*** 1.807*** 0.699*** 0.653*** 0.023 0.135 (0.131) (0.138) (0.144) (0.145) (0.241) (0.251) (0.161) (0.163) (0.221) (0.233) 0.020 0.042** 0.025 0.0265 0.051* 0.051* -0.008 -0.003 -0.021 0.001 (0.018) (0.020) (0.020) (0.0202) (0.027) (0.028) (0.031) (0.032) (0.036) (0.039) 0.389*** 0.392*** 0.369*** 0.342*** 0.161** 0.165** 0.311*** 0.261*** 0.344*** 0.315*** (0.027) (0.028) (0.047) (0.0499) (0.063) (0.066) (0.048) (0.050) (0.067) (0.071) 0.254** 0.212* Omitted 0.891** 0.920** 0.752*** 0.746*** 0.300 0.300 (0.120) (0.123) (0.390) (0.400) (0.181) (0.185) (0.230) (0.237) 0.680*** 0.713*** 0.077 0.0163 -0.325 -0.376 0.145 0.215 0.117 0.209 (0.126) (0.130) (0.116) (0.118) (0.322) (0.327) (0.245) (0.252) (0.217) (0.231) -0.009* -0.009* Omitted Omitted Omitted -0.003 -0.002 0.043 0.053 (0.005) (0.005) (0.009) (0.009) (0.070) (0.069) 0.043* 0.043* 0.078* 0.0490 0.050 0.063 -0.010 0.001 -0.015 0.028 (0.025) (0.026) (0.041) (0.0428) (0.113) (0.119) (0.042) (0.042) (0.052) (0.057) Constant -4.335*** -4.697*** -1.708*** -2.095*** -3.704*** -3.698*** -0.068 -0.902 -0.636 -1.501* (0.333) (0.353) (0.351) (0.453) (1.222) (1.285) (0.576) (0.638) (0.755) (0.808) Sectoral effects No Yes No Yes No Yes No Yes No Yes Observations 2,177 2,177 719 719 410 410 576 576 319 319 Log likelihood -805.1 -770.3 -425.6 -413.0 -172.2 -164.7 -330.4 -318.1 -167.5 -151.7 Pseudo R2 0.232 0.266 0.142 0.167 0.300 0.331 0.172 0.203 0.113 0.197
Our estimation results for the entire sample of countries, including both CEE and MENA countries, are presented in Columns (1–3). In Column (1), we present the results obtained without controlling for individual country and sector-specific effects; in Column (2), we control for country-specific effects; and, in Column (3), we control for both country- and sector-specific effects. We treat these estimations as the benchmark results for all other comparisons in this section. Tab. 4 and 5 apply the same format of presentations for other country aggregates.
The estimation in Column (1) reveals that the estimated parameter for the measure of productivity displays an unexpected negative sign and is statistically significant at the 1% level. However, this surprising result disappears when we control for country- and sector-specific effects (Columns (2) and (3), respectively). The change in the sign of the estimated parameter for the measure of productivity probably reflects large differences in labour productivity among the analysed countries in the full sample.
The positive sign of the estimated parameter for the productivity variable reported in Columns (2) and (3), which is statistically significant at the 1% level, means that a 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 the probability of exporting.
Another slightly surprising result appears in the case of the variable See, e.g., Cieślik et al. (2015). In some estimations the sign of the age variable displays negative signs. For further discussion on the role of state control in the past, see the comments on the variable
The majority of our control variables are statistically significant at the 1% level. The estimated signs of parameters for standard explanatory variables are in line with the expectations and results of other studies discussed in the literature review section. The estimated parameters for the human capital variable ( The sign of the
The firm size variable also displays an expected positive sign, at a 1% level of statistical significance, indicating the importance of economies of scale for exporting. The variables measuring foreign ownership (
Finally, the additional variables describing the role of state ownership and the level of corruption reveal some amount of pre-transition legacy. In particular, the variable
In Columns (4–6), we estimate the same model for the aggregate of eight MENA countries (Egypt, Jordan, Lebanon, Morocco, Tunisia, the West Bank and Gaza, Israel and Turkey). The estimation results for MENA countries reveal certain similarities to the results for all countries in terms of signs and the statistical significance of estimated parameters, but only for some control variables. In particular, the estimated parameters for the variables
However, there are also significant differences. In particular, the estimated parameter for the key variable in the Melitz model (2003),
Another important difference exists in the case of the
The other significant difference appears in the case of the
The estimations of the model for all CEE countries are shown in Columns (7–9) of Tab. 4. In the majority of cases, the values of parameters and their statistical significance is very similar to the results obtained for all countries listed in Columns (1–3). In particular, the values of the parameters for variables
There are only differences in the case of additional variables describing the engagement of state ownership and the level of corruption. Both variables, i.e.,
The estimation results for the 11 CEE countries, which are currently members of the EU, are presented in Columns (1–3) of Tab. 4. The same scheme as in Tab. 5 is used, i.e., in Column (1), we do not control for country- and sector-specific effects, in Column (2), we control for country-specific effects and, in Column (3), we control for both country- and sector-specific effects. The results for these countries are very similar in terms of statistical significance, signs and values for estimated parameters to those obtained for all countries.
In particular, the parameter for the variable
A minor difference only appears in the case of two variables. The variable
The results for the other CEE countries, which are non-members of the EU, are presented in Columns (4–6) of Tab. 4. Many of those countries are former republics of the Soviet Union. The results for these countries are very similar in terms of statistical significance, signs and values of estimated parameters to those obtained for all countries. Minor differences appear in the following cases.
First, the variable In principle, we should interpret this as implying that a more corrupted environment facilitates exports. For a more detailed interpretation of the role of corruption in the economic activities of firms in CEE, see Cieślik and Goczek (2015).
The results for the six MENA countries are presented in Columns (7–9) of Tab. 5. These estimations were performed for six out of the eight countries (as in Tab. 4), i.e., with the exception of Israel and Turkey, which are the most advanced of MENA countries in the process of economic development.
The results for these six countries are somewhat similar in terms of statistical significance, signs and values of estimated parameters to those previously obtained for all MENA countries. The major differences can be summarized as follows. First, the sign of estimator for the
The robustness of the estimates obtained for aggregate country groups has been tested using an alternative estimation method, i.e., the fractional logit model. The results of these estimations are presented in Tab. A2 in the Appendix. These results are somewhat different in comparison to those obtained by the probit model in the benchmark results presented in Tab. 4. In particular, the variables
We also tried to analyse the role of product concentration and multi-product firms in exporting. The results of these estimations, including the additional variable
In the subsequent part of this section, we present the results of the estimations for individual MENA countries (see Tab. 5). For each country, we first present estimation results without controlling for sectoral effects, followed by the results obtained when controlling for sectoral effects. We only performed the estimations for countries for which the number of observations was larger than 300. However, a large number of observations (exceeding 700 firms) was only available for Egypt and Turkey. Therefore, the results for other countries should be treated with caution. We will treat the results for all eight MENA countries as the benchmark for comparison.
The estimation results for Egypt are presented in Columns (1) and (2) of Tab. 5. The number of observations is relatively large (2,177), while the results are similar to those obtained for the group of eight MENA countries. The parameters for
The estimation results for Turkey are presented in Columns (3) and (4) of Tab. 5. The number of observations is relatively large (719), but the results are different to those obtained for the group of eight MENA countries. Only two parameters for the variables The
The estimation results for Israel are presented in Columns (5) and (6) of Tab. 5. The number of observations is quite small (410), but the results are somewhat similar to those obtained for the group of eight MENA countries. A major difference is that the productivity variable ( The role of domestic research (
The estimation results for Tunisia are presented in Columns (7) and (8) of Tab. 5. The number of observations is not high (576) and the results are slightly different from those obtained for the group of eight MENA countries. Three control variables are statistically significant at the 1% or 5% levels when we control for sectoral effects, as well as reveal the expected positive signs (
The estimation results for Morocco are presented in Columns (9) and (10) of Tab. 5. The number of observations is small (319) and the results are very different from those obtained for the group of eight MENA countries. Only one parameter
In conclusion, we can state that the determinants of export performance are heterogeneous among firms from individual MENA countries. Only one variable describing the size of company employment (
In this paper, we attempted to address some important questions by relying on a comparison of different country experiences, which carry relevant implications for policy recommendations. In other words, we tried to analyse to what extent the determinants of export behaviour of companies in the MENA region are similar to those of companies operating in CEE countries, which have been discussed in previous studies on these countries. We draw our conclusions on the basis of probit estimations, given our awareness that fractional logit estimations do not fully support these recommendations.
Our research confirmed that the productivity of labour, in accordance with the Melitz model (2003), affects firms’ propensity to export in MENA and CEE countries if we control for country- and sector-specific effects, but this key variable of the model does not seem to work in the case of the MENA when taken as a whole. Moreover, we confirmed that other variables, such as i) the size of the company, ii) the use of human capital and iii) the level of firms’ internationalization measured by foreign ownership and the use of foreign licences, contributed to an increase in the propensity to export among firms in the analysed CEE countries, although less so in MENA countries.
Clearly, there are some important differences between these two groups of countries as well. Labour productivity is a statistically significant variable in CEE countries but not in MENA countries. The use of foreign technology (foreign licences) is also statistically significant in the case of CEE countries, but not in all eight MENA countries, if we control for country- and sector-specific effects. The age of the company is not significant in CEE countries, but is significant in a majority of MENA countries (usually, older companies are more export-oriented). Finally, the measures reflecting a non-market economy legacy, i.e., the engagement of state ownership and corruption, are statistically significant in some groups of CEE countries, but very rarely in the case of individual MENA countries.
What are the preliminary policy implications of the above? Oddly, if the aim of MENA governments is to improve export performance, fighting corruption does not seem to be particularly helpful. It seems that a policy of privatizing firms, such as the one practised after 1989 in CEE, is not going to help much in improving export performance (maybe with the exception of Egypt). The fact that firms’ age is a significant variable in the case of the MENA means that, over time, export performance should improve as a result of accumulated experience.
Given that many firm-level determinants of exports are sometimes dissimilar in CEE and MENA countries, it also follows that the export competitiveness of the analysed MENA countries can be improved by the development of modern education systems and the facilitation of the accumulation of human capital. Financial support for R&D and innovation should have a positive impact on export performance. MENA countries should also seek to attract export-oriented foreign direct investments. On the other hand, transfers of technology via licences does not seem to work as well in CEE countries (with the exception of Egypt).
The specificities of MENA countries, with respect to product concentration, should also be taken into consideration. Multi-product firms in MENA countries are more likely to export, while this characteristic is statistically not significant for exports in the case of CEE firms. This could mean that firms from the MENA do not concentrate their exports on products that correspond to where they are most efficient. It would be desirable to carry out an in-depth investigation into this phenomenon.
To sum it up, according to our estimations, it appears that corruption and state ownership do not result in serious barriers to exports at present for either group of countries. However, the situation is differentiated among individual MENA countries, as discussed in the context of individual country results presented in Tab. 6. More in-depth studies, based on broader databases, and research on the additional determinants of exports are needed in the future. It would also be desirable to study the direction of causality between exporting and productivity using firm-level panel data, which are not available from BEEPS V.
More attention should also be given to the role of innovations, given that, thus far, we have focused on the input side of innovation. Therefore, in future studies, it would be useful to study the output side of innovation as well. In particular, it would be desirable to separately investigate the role of various types of innovation, such as product or process innovations.