Policy Research Working Paper 9432 Exporting and Female Labor Market Outcomes in Georgia Claire H. Hollweg Anne Ong Lopez Macroeconomics, Trade and Investment Global Practice October 2020 Policy Research Working Paper 9432 Abstract Using firm-level data for Georgia, the paper estimates the of this elasticity. The data are from the National Statistics quasi-elasticity of employment and wages with respect Office of Georgia Statistics Survey of Enterprises merged to the share of exports in total sales, to explore whether with customs data for 2006–17. The instrumental variables changes in the structure of sales (exporting versus selling regression results show that the act of exporting improves to the domestic market) matter for labor market outcomes. female employment but reduces overall average wages and The methodology uses exogenous fluctuations in exchange female wages. Increasing exports to the European Union as rates combined with firms’ initial exposure to various mar- well as high-income countries drives this positive result for kets as instrumental variables to identify a causal effect. The female employment, whereas exporting to upper-middle-in- results differentiate employment levels and average wages come countries is found to have a negative relationship with by gender and consider whether export destination or the female employment. competiveness of economies matters for the magnitude This paper is a product of the Macroeconomics, Trade and Investment Global Practice. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at chollweg@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Exporting and Female Labor Market Outcomes in Georgia Claire H. Hollweg Anne Ong Lopez1 Keywords: Trade and labor markets, Wages and employment JEL classification codes: J20, J30, F14, F16 1 Claire H. Hollweg is Senior Economist in the Macroeconomic, Trade and Investment Global Practice of the World Bank and Anne Ong Lopez is Economist in the Strategy, Policy and Budget Department of the Asian Infrastructure Investment Bank. The corresponding author can be contacted at chollweg@worldbank.org. We thank Cristina Constantinescu Massimiliano Cali for comments and Elizabeth Ruppert-Bulmer for inputs to an earlier draft. The findings, interpretations, and conclusions expressed in this paper are those of the authors and do not represent the views of the World Bank, its Executive Directors, or the countries they represent. 1. Introduction and motivation It is well established in the literature that exporters are larger and more productive than non- exporting firms. Recent literature has also shown that exporters achieve better labor market outcomes than non-exporters. Firms that have a larger share of exports in total sales are found to have higher levels of employment and pay higher wages (Bambrilla, Lederman and Porto 2012). Cebeci, Lederman and Rojas (2013) attempt to identify a causal effect of changes in the structure of sales on labor market outcomes using exogenous fluctuations in exchange rates combined with firms’ initial exposure to various markets as instrumental variables. Under this approach, increases in exports as a share of total sales in Turkish firms are associated with increases in employment levels within firms, but a higher export share generates no observed impact on wages. In this analysis, we explore whether changes in firms’ sales structure – exporting versus selling to the domestic market, as well as by export destination – matters for labor outcomes. We use firm- level data for Georgia to estimate the quasi-elasticity of employment and wages with respect to the share of exports in total sales. Following the methodology of Cebeci, Lederman and Rojas (2013) allows us to test for a causal effect of firms’ growth in exports relative to domestic sales (measured by the change in a firm’s share of exports in total sales 2) on their labor market outcomes, namely employment level (total employment, as well as female employment) and average wages (average for total firm employment, and for female employment). 3 In addition, we assess whether exporting to different destination markets such as the European Union or high-income countries has different impacts on labor-market outcomes. The remainder of the paper proceeds as follows. Section 2 presents the methodology and Section 3 describes the firm-level data source used. Section 4 discusses stylized facts about exporters in Georgia. Section 5 concludes with the results of the analysis. 2. Methodology 2 The share, rather than the level, of exports is used to capture the structure of firms’ sales (as only using levels would capture size effects). 3 Whereas an OLS regression can identify statistically significant correlations between changes in the structure of firms’ sales and employment and wage growth, we employ an instrumental variables approach to identify causal relationships. We use exogenous fluctuations in exchange rates combined with firms’ initial exposure to various markets as instrumental variables (see Cebeci, Lederman and Rojas 2013). 2 To identify the effect of firms’ sales structure on labor outcomes, we estimate the effect of within- firm changes in the share of sales that are exported on the level of employment (total, skilled, unskilled, female) and average wages (total, female) of the firm, the latter being a proxy for the quality of workers employed in each firm. The analysis follows a methodology similar to Bambrilla, Lederman and Porto (2012) and Cebeci, Lederman and Rojas (2013). The relationship between exporting and labor-market outcomes is specified as: ,, = + + + 1 ,, + 2 ,,, + 3 ,, + ,, where ,, stands for the log of firm ’s employment level or average wage in industry in year ; ,, is the share of firm-level exports in total sales; ,,, is the share of firm-level exports to destination market in total sales; ,,, is total firm-level sales; are firm-, industry-, and year- specific effects to control for other factors that can affect firm size and sales; 1 is a first parameter to be estimated, which is an export-sales elasticity of employment and wages; 2 is a second parameter to be estimated, which tells whether exporting more to the specific destination is better for employment and wages than exporting on average; and the error term ,, is assumed to be white noise. Controlling for total sales avoids potential biases in the estimation of 1 due to exogenous changes in export conditions. For example, a positive external shock that increases exports could possibly also increase domestic sales if firms invest part of the money from increased exports to stimulate domestic sales as well. Such an effect would over-estimate the quasi-elasticity of labor market outcomes vis-à-vis the share of exports in total sales. Alternatively, the positive external shock could decrease domestic sales if the supply response is inelastic and the firm decides to serve only the export market at the expense of the domestic market, in which case the bias would go in the opposite direction. OLS estimates of the elasticities 1 and 2 may be biased due to reverse causality, as changes in firm size and wages may affect the structure of firms’ sales (given that exporters are larger and more productive than non-exporters). Following Brambilla, Lederman and Porto (2012) and Cebeci, Lederman and Rojas (2013), our methodology will use exogenous fluctuations in exchange rates interacted with firms’ initial exposure to external markets as instrumental variables 3 to identify causal effects of export growth within a firm (as a share of total sales) on a firm’s labor market outcomes. The first-stage regression specification is: ,, = + + 1 ,,=0 ∗ + 2 ,, + ,, ,,, = + + 1 ,,=0, ∗ , + 2 ,, + ,, where ,,=0 ∗ is the initial share of exports in total sales of each firm 4 times the real effective (trade weighted) exchange rate of the country and ,,=0, ∗ , is the initial share of firm-level exports to destination market in each firm’s total sales times the nominal effective (trade weighted) exchange rate between the origin country and the destination. A valid instrument is one that is correlated with the endogenous regressor and uncorrelated with the error term. The choice of instruments is valid given that (i) a change in a countries’ effective exchange rate affects export competitiveness of a firm, (ii) the initial level of firms’ exposure to exports is pre- determined, and (iii) the assumption that a country’s trade-weighted exchange rate is not influenced by any single firm. 5 While the effective exchange rate is common to all firms within a country at a point in time, the interaction with a firms’ initial export or destination-market share allows the instrument to vary over time and across firms (so it can explain variation in changes in export shares over time and across firms). 6 We test the model using firm-level survey data and customs data for Georgia (described below), which provides a sufficiently large and rich data set. 3. Variable definitions and data sources We utilize firm-level data to estimate the effects of within-firm changes in sales structure on firm- level employment and average wages. The data sets include firm-level employment (total and 4 The first year a firm appears in the sample is when it first exported, such that the initial share of exports is not zero for any firm. 5 We alternatively control for firm-specific effects by estimating the model in first-differences, which wipes out the influence of time-invariant firm-specific characteristics. Industry and year dummy variables are included to control for industry- and year-specific effects. Because the specification in levels has firm-level fixed effects, which control for time-invariant firm-specific factors, first differencing the model is not necessary to remove the effect of such factors. However, estimating the specification in changes arguably allows for a stronger instrument, given that changes in the effective exchange rate as opposed to the level are a better exogenous correlate to export competitiveness. 6 Changes in firms’ sales may also be endogenous to the specification, for example if firm size is correlated with unobservable firm characteristics that change over time. Notwithstanding, because we are less interested in estimating this coefficient, and because not including it potentially biases the coefficient we are interested in estimating, we maintain the specification. 4 female workers), firm-level average wage (total and female workers), the share of exports in firm- level total sales, and the share of exports to specific countries in firm-level total exports. We use the National Statistics Office of Georgia (GEOSTAT) Statistics Survey of Enterprises for 2006-2017, which is a panel with the same firm identified across years. The data are then merged with customs-level transactions data using unique firm IDs. The GEOSTAT survey covers more than 61,000 firms and 132,000 firm-year observations during this period, with data on employment (total and female), wages (total and female), and sales. Customs data record covered more than 81,000 entries, with information on export value and destination for each transaction. Data on effective exchange rates are collected from the International Monetary Fund’s International Financial Statistics. Export share: Share of establishment’s sales that are exported. Firm-level export shares for Georgia are collected from the GEOSTAT Statistics Survey of Enterprises for 2006-2017, calculated as total sales by each type of activity without VAT and excise divided by total exports (converted from USD to GEL using the corresponding year’s official exchange rate reported by the World Bank World Development Indicators). Note that because a panel dimension is necessary to conduct the analysis of changes in export shares and changes in labor market outcomes, the selection of firms was limited to those observed in the data across multiple years. A firm was included in the sample if it was observed in any two sample years, had exported in one of those years, and had non-zero exports in the first year or a previous year. 7 Market share: Share of establishment’s exports that are exported to a particular country market. Firm-level market shares for Georgia are collected from the GEOSTAT Statistics Survey of Enterprises for 2006-2017, calculated as exports to a particular market (e.g., the European Union, the Russian Federation, non-EU non-Russia ECA, high income, upper-middle income, lower- middle income, and low income) divided by total exports. Exports were converted from USD to GEL using the corresponding year’s official exchange rate reported by the World Bank World Development Indicators. 7 The first year a firm appears in the sample is when it first exported, such that the initial share of exports is not zero for any firm. 5 Sales: Firms’ total turnover. Sales are in real 2005 national currency units using the consumer price index reported by the World Bank World Development Indicators. Firm-level sales for Georgia are calculated from the GEOSTAT Statistics Survey of Enterprises for 2006-2017. Employment: Number of persons employed. Firm-level employment for Georgia is collected from the GEOSTAT Statistics Survey of Enterprises for 2006-2017, measured as the average number of persons employed (employees, employed shareholders and employed family members in case of family owned enterprise) in the enterprise during the year. Female employment: Number of female persons employed. Firm-level female employment for Georgia is collected from the GEOSTAT Statistics Survey of Enterprises for 2006-2017, measured as the average number of female persons employed (employees, employed shareholders and employed family members in case of family owned enterprise) in the enterprise during the year. Wages: Total annual cost of labor or remuneration (including wage, salary, premium, bonus, social payments, etc. of both permanent and temporary employees) which was accrued or paid in kind to employees (including income tax) during the year divided by employment. Wages are in real 2005 national currency units using the consumer price index reported by the World Bank World Development Indicators. Firm-level average wages for Georgia are calculated from the GEOSTAT Statistics Survey of Enterprises for 2006-2017. Female wages: Remuneration (including wage, salary, premium, bonus, social payments, etc.) which was accrued or paid in kind to female employees during the year divided by female employment. Female wages are in real 2005 national currency units using the consumer price index reported by the World Bank World Development Indicators. Firm-level average female wages for Georgia are calculated from the GEOSTAT Statistics Survey of Enterprises for 2006-2017. Nominal effective exchange rate: A measure of the value of a currency against a trade-weighted average of several foreign currencies. The nominal effective exchange rate for Georgia for each destination market was calculated from the World Bank World Development Indicators using trade weights calculated from UNCOMTRADE for 2006-2012. 8 8 We experiment with an alternative instrument that is also interacted with the country’s deposit dollarization rate, but the results do not change. 6 Real effective exchange rate: The nominal effective exchange rate divided by a price deflator or index of costs, where 2007=100. The real effective exchange rate was collected from the IMF’s International Financial Statistics. 4. Summary statistics In Georgia, 8 percent of firms in the sample were exporters in any year, and 2 percent of total sales were represented by exports. The data however suggest that of firms that export, a significant and increasing share of their sales tend to come from exports. On average for Georgian exporting firms, 13 percent of sales came from exports. For firms that export to the EU or high-income countries, exports tend to represent a larger share of sales (see Table 1). Table 1: Share of exports and exporting firms All firms Exporters EU exporters HIC exporters Exported in at least one year 8.4 -- -- -- Share of exports in total sales 1.9 12.9 19.5 18.4 Number of firms 61,849 5,090 1,447 1,713 Number of observations 132,028 18,296 7,138 8,151 Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. Note: HIC = high income country. Exporting firms tend to be bigger and more productive than firms that sell exclusively to the domestic market (Bernard and Jensen 1995, Bernard and Jensen 1999, Bernard, Jensen, Redding and Schott 2007, Clerides, Lach and Tybout 1998). This holds true in our Georgia sample with respect to firm size in terms of employees, where exporters are four times as large as non-exporters. For example, the average number of employees in Georgian exporting firms was 83 as compared to 20 in non-exporting firms. The difference in turnover is even larger, with exporters seven times as large as non-exporters. Exports also pay more, with average wages (total compensation per worker) nearly twice as large for exporters than non-exporters (see Table 2). This difference may reflect a higher fixed cost of exporting in Georgia, which only larger or more productive firms can overcome. On average, Georgian exporters hire more female workers than non-exporters, but the share of female employees is smaller in exporting firms, with 35 percent of the workforce being women in exporting firms compared to 51 percent in non-exporting firms (see Table 2). Exporters also pay 7 female employees relatively higher wages than non-exporters, but female workers earn lower wages than male workers in both exporting and non-exporting firms. However, the ratio of female wages to total wages is lower in exporting firms than non-exporting firms. These differences are statistically significant (see the Appendix). Table 2: Size and wages of exporting vs. non-exporting firms All firms Non-exporters Exporters Average number of employed 28.7 19.7 82.8 Average number of female employed 15.4 11.9 32.1 Female employment share 0.48 0.51 0.35 Average wages 4,647 4,076 7,530 Average female wages 3,759 3,264 5,974 Female-to-male wage ratio 0.95 0.97 0.88 Average turnover 1,825,241 963,991 6,924,577 Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. Notes: Exporters are firms that exported in at least 1 year of the sample. Labor market outcomes in terms of employment and wages exist between firms that export to different destinations. Firms that export to the EU or high-income countries tend to be larger than other exporters, in terms of sales as well as number of employees (total and female). They also pay higher wages (total and female). However, the share of female workers in total workers as well as the female-to-male wage ratio is lower for firms that export to the EU or high-income countries than other exporters (see Table 3). These differences are statistically significant (see the Appendix). Table 3: Size and wages of exporting firms by destination Non-EU EU exporter Non-HIC HIC exporter exporter exporter Average number of employed 22.9 127.1 22.6 119.0 Average number of female employed 12.5 52.0 12.3 49.4 Female employment share 0.49 0.38 0.49 0.38 Average wages 4,376 8,419 4,344 8,304 Average female wages 3,525 6,561 3,500 6,472 Female-to-male wage ratio 0.96 0.86 0.96 0.86 Average turnover 1,325,639 10,187,536 1,299,063 9,472,563 Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. 8 Notes: HIC = high income country. Exporters are firms that exported in at least 1 year of the sample. EU / HIC exporters are firms that exported to the EU / high income country in at least 1 year of the sample. This holds across sectors of Georgia’s economy that export. The female labor share is higher in manufacturing than in primary sectors for both exporting firms and non-exporting firms, but even in these sectors on average the female labor share is lower for exporters than non-exporters. 9 The aggregate result of the wage rate holds across broad sectors of the economy, with the exception of agriculture, hunting, forestry and fishing where the female-male wage gap is lower for exporters than non-exporters. 5. Regression results and conclusions Using the regression methodology described above, we test the hypothesis that firms that sell to global markets generate greater employment and pay higher wages than firms that sell only to the domestic market, and that exporting is causally linked to employment and wage growth. We report the results for OLS regressions – which identify non-causal statistical correlations between changes in firms’ export shares and employment and wage growth – as well as the instrumental variables approach to identify causal relationships. The OLS regression results are summarized in Table 4 and the IV regression results are summarized in Table 5. 10 Tables 4 and 5 identify some significant correlations between exporting and labor market outcomes (when all export destinations are considered together). Although we do not find evidence that firms that increase their share of exports in total sales also increase their employment level, there is a negative correlation with wages that is also statistically significant for the instrumental variable results. The low value of the point estimates indicates a relatively small effect, however; a 10 percentage point increase in the share of sales that are exported leads to a decrease in wages by 1.8 percent. The instrumental variable results suggest however there is a significant and positive correlation with exporting and female employment, but a significant and negative correlation with female 9 The female employment share for manufacturing non-exporters if 43 percent compared to 35 percent for exporters. For mining and quarrying it is 18 percent for exporters compared to 14 percent for non-exporters, for fishing it is 25 percent compared to 22 percent, and for agriculture, hunting and forestry it is 35 percent compared to 32 percent. 10 Full regression results are available from the authors upon request. First-stage estimation results, results in first differences, and results with an alternative instrument are also available upon request. 9 wages. A 10 percentage point increase in the share of sales that are exported causes female employment levels to increase by 1.2 percent and female wage levels to decrease by 1.5 percent. The insignificant OLS yet significant IV results indicate that reverse causality is likely a concern, such that the average correlation is not significant. Exploring further the positive relationship with female employment but negative relationship with female wages for Georgian exporters is limited due to data constraints. Nevertheless, the results are in line with the global literature. In a cross-country data set, World Bank (2019, 2020) shows that exporters tend to hire more women than non-exporters, but these women are more likely to be in labor-intensive production jobs or occupations that require lower skills and pay less. At the same time, these firms are significantly less likely to be majority female owned and are also significantly less likely to have a top female manager. Women’s placement stems, in part, from disadvantages in endowments—such as assets, access to credit, education, skills training, and social capital, among others. These arise because of gender-biased regulations or discriminatory social norms (World Bank 2019, 2020). Segregation of female employment across sectors may be another explanation for higher female employment but lower female wages on average for Georgian exporters if women are more likely to work in sectors that pay lower wages. As identified in World Bank (2018), women tend to be employed in lower-productivity sectors such as agriculture in Georgia, which is confirmed with the data used for this paper’s analysis, as in Figure 1. Because the regressions control for industry fixed effects at the 1-digit industry level, they would capture this type of segregation of women across broad industries. Nevertheless, if this segregation also happens within more closely defined industries that also have pay gaps for women, such as garment production within the manufacturing industry, which also has higher female labor shares, this could be one possible explanation of the results. The inefficient allocation of female employment across sectors therefore may similarly prevent female workers from benefiting as much as they could from gains in employment driven by exporters’ performance. 11 11 We also explored whether the results change when restricting the sample to 2007-2013 when Russia had imposed an embargo on trade with Georgia, but the overall results hold. 10 Figure 1: Female employment share, exporters vs. non-exporters 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 Agriculture, Fishing Mining and Manufacturing hunting and quarrying forestry Exporters Non-exporters Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. It is worth discussing the validity of the instrumental variables. The F-statistics for the first stages (a test of the exclusion restrictions) are all above the desired critical values, with corresponding p- values for the null hypothesis of irrelevant instruments close to zero. However, the estimated coefficient on the instrumental variable is positive and highly statistically significant in all of the specifications, which implies that real appreciations of the exporting country’s currency are associated with increases in firms’ export over sales ratios. We cannot fully test the validity of the exclusion restrictions by conducting over-identification tests because the models are just identified. Differentiating by export destination, the analysis finds evidence that exporting to ECA and middle-income countries has significant correlations with employment. We consider the following export destinations in separate regressions: EU, Russia, non-EU non-Russia ECA, high-income, upper-middle income, lower-middle income and low-income markets. Although the OLS results suggest a significant correlation with the share of exports to ECA and middle-income countries, the IV results are not significant. Instead, there is a positive and statistically significant relationship with exporting to lower middle-income countries and employment levels. A 10 percentage point increase in the share of sales that are exported causes female employment levels to increase by 0.5 percent. 11 Considering female employment, there is a positive and significant correlation with the share of firms’ exports to Russia, ECA, and lower middle-income countries and the levels of female employment. We find a negative correlation with the share of firms’ exports to EU countries. These become insignificant with the IV results, suggesting reverse causality is a concern. Although we do not find any evidence that the relationship between exporting and female wages varies by export destination (OLS and IV), we do find evidence that firms that increase their export share to ECA countries as well as lower-middle-income countries pay lower wages (IV). In summary, the instrumental variables regression results show that the act of exporting improves female employment but reduces overall average wages and female wages. Increasing exports to the EU as well as high income countries drives this positive result for female employment, whereas exporting to upper middle income countries is found to have a negative relationship with female employment. 12 Table 4: Georgia’s elasticity of labor market outcomes with respect to exporting, OLS Destination market Coefficient Employment Female employment Wages Female wages World Export share 0.00074 -0.00083 -0.00136** 0.00037 [0.00068] [0.00072] [0.00058] [0.00068] EU Export share 0.00078 -0.00077 -0.00132** 0.00040 [0.00068] [0.00072] [0.00058] [0.00068] EU market share -0.00028+ -0.00044* -0.00021 -0.00014 [0.00021] [0.00026] [0.00021] [0.00027] Russia Export share 0.00067 -0.00090 -0.00142** 0.00031 [0.00068] [0.00072] [0.00057] [0.00068] Russia market share 0.00066+ 0.00094* 0.00077+ 0.00095 [0.00047] [0.00052] [0.00051] [0.00091] ECA Export share 0.00064 -0.00091 -0.00136** 0.00040 [0.00068] [0.00072] [0.00058] [0.00069] ECA market share 0.00034** 0.00029* 0.00001 -0.00008 [0.00015] [0.00016] [0.00014] [0.00019] HIC Export share 0.00075 -0.00079 -0.00129** 0.00041 [0.00068] [0.00072] [0.00057] [0.00068] HI market share -0.00009 -0.00021 -0.00032+ -0.00016 [0.00020] [0.00022] [0.00020] [0.00025] UMIC Export share 0.00059 -0.00092 -0.00144** 0.00032 [0.00068] [0.00073] [0.00058] [0.00069] UMI market share 0.00034** 0.00022+ 0.00021* 0.00013 [0.00015] [0.00016] [0.00013] [0.00020] LMIC Export share 0.00071 -0.00088 -0.00132** 0.00039 [0.00068] [0.00072] [0.00058] [0.00069] 13 LMI market share 0.00044 0.00102** -0.00089** -0.00029 [0.00045] [0.00051] [0.00043] [0.00081] LIC Export share 0.00074 -0.00083 -0.00136** 0.00038 [0.00068] [0.00072] [0.00058] [0.00068] LI market share -0.00057 -0.00088 0.00080+ 0.00098 [0.00128] [0.00142] [0.00056] [0.00104] Number of firm-year observations 8,876 7,900 8,443 7,848 Number of firms 3,385 2,786 3,031 2,752 Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. Notes: Robust standard errors clustered at firm level in brackets. Includes turnover, firm fixed effects, and year and industry effects. *** p<0.01, ** p<0.05, * p<0.10, + p<0.20. 14 Table 5: Georgia’s elasticity of labor market outcomes with respect to exporting, IV Destination market Coefficient Employment Female employment Wages Female wages World Export share 0.01213+ 0.01492* -0.01832*** -0.01516** [0.00750] [0.00820] [0.00697] [0.00731] EU Export share 0.01203+ 0.01453* -0.01793*** -0.01495** [0.00746] [0.00843] [0.00695] [0.00732] EU market share 0.00053 0.00361+ -0.00216 -0.00219 [0.00223] [0.00265] [0.00235] [0.00263] Russia Export share 0.12181 0.26543 -0.02507 -0.09927 [0.30288] [1.39088] [0.04939] [1.52254] Russia market share 0.07970 0.19164 -0.00572 -0.07098 [0.21238] [1.04916] [0.04006] [1.27907] ECA Export share 0.01338* 0.01648** -0.01639** -0.01430** [0.00733] [0.00791] [0.00637] [0.00677] ECA market share -0.00251 -0.00214 -0.00387** -0.00125 [0.00217] [0.00247] [0.00197] [0.00226] HIC Export share 0.01119+ 0.01402* -0.01702*** -0.01523** [0.00708] [0.00823] [0.00660] [0.00715] HI market share 0.00292 0.00496 -0.00434 0.00042 [0.00431] [0.00465] [0.00459] [0.00457] UMIC Export share 0.00711 0.00550 -0.01911** -0.01838** [0.00834] [0.00826] [0.00839] [0.00888] UMI market share -0.00291+ -0.00669** -0.00055 -0.00244 [0.00211] [0.00262] [0.00207] [0.00242] LMIC Export share 0.00835 0.01259+ -0.01389** -0.01417* [0.00695] [0.00772] [0.00650] [0.00727] 15 LMI market share 0.00529** 0.00345 -0.00685*** -0.00152 [0.00249] [0.00306] [0.00253] [0.00278] LIC Export share 0.00835 0.01259+ -0.01389** -0.01417* [0.00695] [0.00772] [0.00650] [0.00727] LI market share 0.00529** 0.00345 -0.00685*** -0.00152 [0.00249] [0.00306] [0.00253] [0.00278] Number of firm-year observations 8,064 7,471 7,936 7,442 Number of firms 2,720 2,462 2,648 2,450 Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. Notes: Robust standard errors clustered at firm level in brackets. Includes turnover, firm fixed effects, and year and industry effects. *** p<0.01, ** p<0.05, * p<0.10, + p<0.20. 16 References Brambilla, I., D. Ledermna, and G. Porto (2012). “Exports, Export Destinations, and Skills.” American Economic Review, 102(7): 3406-38. Bernard, A., and B. Jensen (1995). “Exporters, Jobs and Wages in US Manufacturing, 1976-1987.” Brookings Papers on Economic Activity: Microeconomics, 67-112. Bernard, A., and B. Jensen (1999). “Exceptional Export Performance: Cause, Effect, or Both?” Journal of International Economics, 47(1): 1-25. Bernard, A., B. Jensen, S. Redding and P. Schott (2007). “Firms in International Trade.” Journal of Economic Perspectives, 21(3): 105-30. Calderón, C., and V. Poggio (2011). “Trade and Economic Growth: Evidence on the role of complementarities for CAFTA-DR countries.” In J. H. Lopez and R. Shankar (Eds.), Getting the Most out of Free Trade Agreements in Central America. Washington, DC: The International Bank for Reconstruction and Development / The World Bank, pp. 83- 122. Cebeci, T., D. Lederman, and D. Rojas (2013). “The Structure of Exports across Destinations and Labor-Market Outcomes: An Empirical Case Study of Turkey.” Unpublished. Clerides, S., S. Lach and J. Tybout (1998). “Is Learning by Exporting Important? Micro-dynamic evidence from Colombia, Mexico, and Morocco.” The Quarterly Journal of Economics, 113(3): 903-47. Word Bank (2018). Georgia at Work: Assessing the Jobs Landscape. Washington, DC: World Bank. World Bank (2019). World Development Report 2020: Trading for Development in the Time of Global Value Chains. Washington, DC: World Bank. World Bank (2020). Women and Trade: The Role of Trade in Promoting Women’s Equality. Washington, DC: World Bank. 17 Appendix Table A1: Difference between exporting vs. non-exporting firms Exporter EU exporter HIC exporter Average number of employed (log) 1.4904*** .6878*** .6473*** (.0107) (.0225) (.0222) Average number of female employed (log) 1.0341*** .6266*** .5870*** (.0104) (.0220) (.0219) Female employment share -.1245*** .0271*** .0355*** (.0023) (.0036) (.0036) Average wages (log) .9177*** .2975*** .2755*** (.0089) (.0138) (.0136) Average female wages (log) .8062*** .2322*** .2103*** (.0089) (.0142) (.0141) Female-to-male wage ratio -.0572*** -.0316 -.0488** (.0122) (.0194) (.0192) Average turnover (log) 2.5000*** .9636*** .9103*** (.0171) (.0292) (.0288) Source: GEOSTAT Statistics Survey of Enterprises for 2006-2017. Notes: HIC = high income country. Exporters are firms that exported in at least 1 year of the sample. EU / HIC exporters are firms that exported to the EU / high income country in at least 1 year of the sample. Column 2: difference in means in log workers, log wages, log turnover between exporters and non-exporters, controlling for year and 1- digit industry. Column 3: difference in means between firms that export to at least one EU destination and non-EU exporters, controlling for year and 1-digit industry. Column 4: difference in means between firms that export to at least one HIC destination and non-HIC exporters, controlling for year and 1-digit industry. ***Significant at the 1% level, **Significant at the 5% level, *Significant at the 10% level. 18