Policy Research Working Paper 10737 Linking Export Activities to Productivity and Wage Rate Growth Luis Aguilar Luna Deborah Winkler Macroeconomics, Trade and Investment Global Practice March 2024 Policy Research Working Paper 10737 Abstract This paper examines the relationship between trade and job activities as well as spillovers to non-export activities. Coun- quality, using productivity and wage rate data for export tries’ specialization in global value chains and sectors also and non-export activities in a sample of 60 countries across matters for the relationship between exports and job qual- all income levels and 45 sectors spanning the whole econ- ity, with manufacturing, agriculture, and business services omy over 1995–2019. First, the analysis finds that workers showing stronger associations. The link between exports involved in export activities are more productive and better and the wage rate is smaller than for productivity. Finally, paid than those in non-export activities. While the produc- productivity and wage rate growth decompositions suggest tivity premium for export activities is confirmed in low- and that growth within rather than between activities was the middle-income countries, there is no wage rate premium. driving force. Within export activities, productivity and Second, this study finds a positive relationship between wage increases were dominated by within-sector growth, exports and labor productivity at the country-sector level, although labor movement toward more productive sectors which can be attributed to productivity gains within export also matters in low- and middle-income countries. 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 dwinkler2@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 Linking Export Activities to Productivity and Wage Rate Growth Luis Aguilar Luna1 and Deborah Winkler2 Keywords: Economic development, international trade, export activity, labor productivity, wage rate JEL codes: F14, F16, F66 1Consultant, Trade and Regional Integration, World Bank. 2Corresponding author, Senior Economist, Trade and Regional Integration, World Bank. eMail: dwinkler2@worldbank.org. This paper was developed for the project “Leveraging trade for more and better job opportunities in developing countries” and also serves as input to the project “Jobs and growth in developing countries” (both reports are forthcoming). The authors thank Sébastien Dessus, Aart Kraay, Maryla Maliszewska, Gaurav Nayyar, Martín Rama, and Shu Yu for very helpful suggestions and discussions. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank Group and its affiliated organizations, its Executive Directors or the governments they represent. 1. Introduction At the firm level, evidence suggests a positive association between a firm’s trading status and productivity. Exporting firms are on average more productive, larger and more skill-intensive than domestic firms (e.g., Wagner 2007, 2012). While the positive link between exporting and productivity (and to a lesser extent employment) has been established at the firm level in low- and middle-income countries (LMICs), positive productivity spillovers also arise from the availability of better inputs and services, as well as engagement in global value chains (GVCs) (Alfaro-Urena, Manelici, and Vasquez 2022, Amiti and Konings 2007, Arnold, Javorcik, and Mattoo 2011). Additionally, foreign direct investment contributes to positive spillover effects (Fernandes and Paunov 2012, Havranek and Irsova 2011, and Javorcik 2004). Participation in GVCs tends to magnify the productivity premium of traditional one-way trade (World Bank 2020). Demand for worker skills tends to be higher among GVC-participating firms relative to non- GVC firms because of higher global quality standards imposed by lead firms in GVCs (World Economic Forum 2015; Criscuolo and Timmis 2017). Higher labor productivity and average wage rates capture aspects of job quality. Can a productivity and wage rate premium of exporting activities be confirmed at the aggregate level of countries and sectors? In this paper, we rely on the OECD Trade in Employment and Trade in Value Added databases for direct estimates of value added, labor compensation, and employment which can be observed separately for export activities 3 and the rest of the economy. The data are available for 45 goods and services sectors (based on ISIC Rev. 4 sectors) spanning the whole economy in 60 countries across all income levels for the period 1995-2020. Appendices 1 and 2 describe the underlying data and computations in more detail. Appendices 3 and 4 give an overview of the country sample (including income classifications) and sector coverage. This paper first examines if workers tied to export activities are more productive and are paid higher wage rates than workers in non-export activities – at both the aggregate and sector levels within countries. To assess this question, we compute mean differences between export and non-export activities at both the country and county-sector levels. We find evidence for a productivity premium at the aggregate level and for the 26 high-income countries (HICs) in the sample, while a wage premium cannot be confirmed in the remaining 34 low- and middle-income countries (LMICs). Our findings, however, reject the existence of a productivity and wage rate premium for export activities within country-sectors which may be explained by positive productivity spillovers from export to non-export activities within sectors. In a second step, this study therefore explores the relationship between increases in exports and job quality within country-sectors more explicitly, also allowing for spillovers from export to non-export activities. We regress labor productivity and wage rate on exports at the country-sector level. Spillovers from export activities to non-export activities within sectors, e.g., due to competition and learning from trading firms, can be an additional source of productivity gains within country-sectors (e.g., Crespo and Fontoura 2007, Alvarez and Lopez 2008). To allow for possible spillover effects, we regress productivity and wage rates for non-export activities on exports in a country-sector. Our results suggest a positive association between growth in exports and job quality for export activities as well as positive spillovers from export to non-export activities. Besides the productivity increases within export activities and spillovers to non-export activities, economy-wide productivity gains from exports can also stem from the reallocation of workers from less productive to more productive activities – both within and across sectors. Activities in this paper refer to 3 This paper focuses on direct export activities within sectors, i.e., not accounting for indirect effects in domestic supplying sectors through input-output linkages. 2 two basic groups: export and non-export activities within a sector (one can also think of activities of exporting and non-exporting firms in the aggregate). To understand what is driving aggregate labor productivity in a country requires decomposing productivity growth into productivity gains within activities and sectors, and productivity gains between activities and sectors. Productivity gains within a sector can, for instance, occur if workers move from less productive non-export to more productive export activities. Workers can also move from less productive non-export or export activities to more productive export activities between sectors. 4 In a third step, we decompose the growth of countries’ aggregate productivity and wage rates to understand the extent to which their growth is also driven by movements of workers towards more productive and better-paid export activities or sectors. One challenge of this exercise stems from the fact that our dataset contains both a sector dimension (45 sectors) and an activity dimension (export and non- export) within sectors. In this paper, we show within and between decompositions (i) by activities, (ii) by sectors (for export and non-export activities separately), and (iii) by sector-activity pairs. 5 Each of these decompositions allows to shed some light on the role of factor reallocation towards more productive activities or sectors. Our paper is structured as follows. Section 2 shows broad trends of average labor productivity and wage rates for export activities and the rest of the economy – overall and for HICs and LMICs separately. It also assesses econometrically at the country level if the difference in observed job quality between export activities and the rest of the economy is statistically significant. Section 3 explores whether higher exports are associated with increased productivity and wage rates at the country-sector level, including through spillovers from export to non-export activities. Section 4 proposes three types of decompositions of labor productivity and wage rate growth to understand the extent to which their growth is driven by reallocation of workers between activities, sectors, and sector-activity pairs. Section 5 concludes. 2. Do export activities hold a productivity and wage rate premium over the rest of the economy? 2.1 Trends in average labor productivity and wage rates at the country level Overall, export activities show a strong productivity premium over the rest of the economy. 6 The average labor productivity of direct export activities is larger than for the rest of the economy in a sample of 60 countries (Figure 1, upper left panel). Value added per worker (in 2015 constant USD) of export activities rose from 49.6 thousand in 1995 to 65.7 thousand by 2011 but stagnated over the following decade. Splitting the sample into two income groups suggests that export activities in 34 LMICs (as per their 1995 income classification) became less productive after 2012, dropping from over 36 thousand to 31 thousand USD per worker by 2020 (bottom left panel). By contrast, average labor productivity of export activities in 26 HICs increased until 2018 and remained flat afterwards (middle left panel). Interestingly, labor productivity in the rest of the economy expanded at a constant pace until 2019 but was hit during Covid-19 in the full sample, and in HICs and LMICs separately. 4 In this context, it is also important to note that employment gains from exports at the firm level do not necessarily materialize at the aggregate level of sectors or countries if exporting firms pull away workers from domestic firms or other sectors. 5 There are 90 sector-activity pairs (45 sectors x 2 activities). 6 In this section, we use the ‘rest of the economy’ rather than ‘non-export activities’ as these activities can include domestic input production for export activities. 3 Figure 1: Labor productivity and wage rate, total, export activities vs. rest of economy, 1995-2020 Average labor productivity (thousand USD) Average wage rate (thousand USD) Labor productivity (in thousand USD), total Average wage rate (in thousand USD), total 70 28 65 60 26 55 24 50 45 22 40 35 20 30 18 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Export activities Rest of the economy Export activities Rest of the economy Labor productivity (in thousand USD), HICs Average wage rate (in thousand USD), HICs 120 50 110 45 100 90 40 80 70 35 60 30 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Export activities Rest of the economy Export activities Rest of the economy Labor productivity (in thousand USD), LMICs Average wage rate (in thousand USD), LMICs 40 11 35 10 30 9 25 8 20 7 15 6 10 5 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Export activities Rest of the economy Export activities Rest of the economy Source: WB staff computations. Data: OECD TiE (employment, labor compensation), OECD TiVA (value added) and WDI (deflators). Note: Labor productivity is defined as value added per worker and the average wage rate is labor compensation per worker (in thousand USD) in constant 2015 USD$ terms. WBG income classifications as of 1995. 4 Similarly, the average labor income per worker tied to export activities is larger than that for the rest of the economy in the full country sample. It increased from 19.6 thousand USD in 1995 to over 27 thousand USD by 2020 (in 2015 constant USD) for export activities, while it grew from 18.6 thousand USD to almost 25 thousand USD in the rest of the economy (Figure 1, upper right panel). This finding is driven by the HICs in the sample where the premium increased over time (middle right panel). While the average premium was 2.6 thousand USD in 1995, it reached 6 thousand USD by 2020. By contrast, LMICs do not show a wage premium for export activities. It rather appears that export activities paid slightly lower wage rates than activities in the rest of the economy over this period (bottom right panel). 2.2 Productivity and wage premium in export activities at the country level In this section, we test whether the labor productivity premium of export activities is statistically significant. Table 1 confirms that the average labor productivity of export activities is significantly higher than that for the rest of the economy in a sample of 26 HICs over the period 1995-2019. We excluded 2020 from the analysis due to Covid-19. For export activities, average labor productivity reached 97.9 thousand USD (in 2015 constant terms) compared to 77.2 thousand USD for the rest of the economy. Similarly, the difference is statistically significant in the sample of 34 LMICs, with export activities showing an average labor productivity of 31.7 thousand USD compared to 19.6 thousand USD in the rest of the economy. Table 1: Difference in average labor productivity, export activities vs. rest of economy, 1995-2019, t-tests High-income countries (as of 1995) Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 650 97.89292 1.986919 50.6567 93.99135 101.7945 Labor productivity, rest of the economy 650 77.18721 0.850787 21.69091 75.51658 78.85784 diff 650 20.70571 1.462526 37.28725 17.83386 23.57756 mean(diff)=mean(lp_expdir-lp_rest) t=14.1575 H0:mean(diff)=0 df=649 Ha:mean(diff)>0 Pr(T>t)=0.0000 Low- and middle-income countries (as of 1995) Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 850 31.66738 2.254097 65.71767 27.24312 36.09164 Labor productivity, rest of the economy 850 19.62278 0.399976 11.66121 18.83773 20.40784 diff 850 12.04459 2.116946 61.71904 7.889534 16.19966 mean(diff)=mean(lp_expdir-lp_rest) t=5.6896 H0:mean(diff)=0 df=849 Ha:mean(diff)>0 Pr(T>t)=0.0000 Source: WB staff computations. Data: OECD TiE (employment), OECD TiVA (value added) and WDI (deflators). Note: lp_expdir = labor productivity in export activities, lp_rest = labor productivity in rest of the economy. Labor productivity is defined as value added per worker (in thousand USD) in constant 2015 USD$ terms. WBG income classifications as of 1995. For additional information, see Appendices 1 to 4. Figure 2 shows the differences for 1995, 2005 and 2019 separately. In HICs, the productivity premium is especially pronounced in commodity-exporting countries like Norway and Saudi Arabia (and Australia in 2019), but also in services-exporting countries like Ireland, Luxembourg, and Switzerland (and Singapore in 2019). In LMICs, export activities show a lower average labor productivity in several instances, such as Brazil or Bulgaria. Interestingly, there is no evidence for a productivity premium of export activities in 1995, while it becomes statistically significant in 2005 and 2019. The underlying t-tests for each year are shown in Appendices 5 and 6. The productivity premium of export activities is particularly high in Kazakhstan and the Russian Federation but also Mexico in 2019. 5 Figure 2: Difference in average labor productivity, export activities vs. rest of economy, 1995, 2005 and 2019 High-income countries, 1995 Low- and middle-income countries, 1995 High-income countries, 2005 Low- and middle-income countries, 2005 High-income countries, 2019 Low- and middle-income countries, 2019 Source: WB staff computations. Data: OECD TiE (employment), OECD TiVA (value added) and WDI (deflators). Note: Labor productivity is defined as value added per worker (in thousand USD) in constant 2015 USD$ terms. WBG income classifications as of the year shown in the graphs. Excludes SAU in 1995 due to extremely high values in export activities. For additional information, see Appendices 1 to 4. 6 By contrast, LMICs do not show a wage premium for export activities which is only apparent in HICs. Table 2 shows that the average wage rate of export activities is significantly higher than for the rest of the economy in the sample of 26 HICs over the period 1995-2019. The average wage rate for export activities was around 44.7 thousand USD (in 2015 constant terms) compared to 40.2 thousand USD for the rest of the economy. In the sample of 34 LMICs, export activities and the rest of the economy show no wage difference, with an average wage rate of between 8.2 thousand USD and 8.5 thousand USD in both groups. Table 2: Difference in average wage rate, export activities vs. rest of the economy, 1995-2019, t-tests High-income countries (as of 1995) Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 650 44.6951 0.6382756 16.2729 43.44177 45.94844 Wage rate, rest of the economy 650 40.19281 0.5141029 13.1071 39.1833 41.20231 diff 650 4.502295 0.2002548 5.105516 4.10907 4.895521 mean(diff)=mean(w_expdir-w_rest) t=22.4828 H0:mean(diff)=0 df=649 Ha:mean(diff)>0 Pr(T>t)=0.0000 Low- and middle-income countries (as of 1995) Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 850 8.189872 0.2052051 5.982707 7.787103 8.592641 Wage rate, rest of the economy 850 8.494331 0.1920752 5.599907 8.117333 8.871329 diff 850 -0.3044592 0.0695322 2.027194 0.4409343 0.1679841 mean(diff)=mean(w_expdir-w_rest) t=-4.3787 H0:mean(diff)=0 df=849 Ha:mean(diff)>0 Pr(T>t)= 1.0000 Source: WB staff computations. Data: OECD TiE (employment), OECD TiVA (value added) and WDI (deflators). Note: w_expdir = wage rate in export activities, w_rest = wage rate in rest of the economy. The wage rate is defined as the labor compensation per worker (in thousand USD) in constant 2015 USD$ terms. WBG income classifications as of 1995. For additional information, see Appendices 1 to 4. Figure 3 shows the wage differences for 1995, 2005 and 2019 separately. In HICs, the wage premium is particularly strong in services-oriented exporting countries like Ireland, Luxembourg, Singapore, and Switzerland – which also stood out in terms of their productivity premia. In LMICs, we find a high variation across the country sample but no evidence for a wage premium of export activities. Malaysia for instance showed a wage premium for the rest of the economy in all three years, while Argentina paid substantially higher wages in non-export activities in 1995. Similarly, export activities were paid a much lower wage rate in Brazil, Bulgaria, and Costa Rica in 2019. The underlying t-tests for each year are shown in Appendices 7 and 8. Overall, the findings imply a decoupling of wages from productivity in LMICs. Several factors may explain the absence of a wage premium despite the productivity premium. First, research suggest that participation in GVCs can be linked to declining labor shares driven by more capital-intensive production which reduces the relative demand for less-skilled workers (e.g., Reshef and Santoni 2023). New technologies are related to higher-quality standards and high-skilled labor, raising the barriers for low- skilled labor in LMICs to participate in GVCs (Rodrik 2018). Second, the fact that there is a productivity but not a wage premium could be explained by uncompetitive labor markets that do not optimally allocate workers to the most productive firms at market-clearing wage levels. This could, for example, be due to higher search costs and frictions in low- income countries, especially in non-tradable activities, so that the average wage rates do not necessarily reflect the marginal value product of workers (Donovan and Schoellman 2023). 7 Figure 3: Difference in average wage rate, export activities vs. rest of the economy, 1995, 2005 and 2019 High-income countries, 1995 Low- and middle-income countries, 1995 High-income countries, 2005 Low- and middle-income countries, 2005 High-income countries, 2019 Low- and middle-income countries, 2019 Source: WB staff computations. Data: OECD TiE (employment), OECD TiVA (value added) and WDI (deflators). Note: The wage rate is defined as the labor compensation per worker (in thousand USD) in constant 2015 USD$ terms. WBG income classifications as of the year shown in the graphs. For additional information, see Appendices 1 to 4. 8 Such labor market frictions have been linked to labor market power of firms which appears to be higher in LMICs (Brooks et al. 2021). Frictions could also relate to informality which is more prevalent in LMICs (La Porta and Shleifer 2014). While informal work serves as a buffer when formal job opportunities are rare, it can also increase firms’ labor market power when wage employment becomes more appealing, putting downward pressure on wages (Amodio et al. 2022). Another possible factor may be a discrepancy between the compensation of general skills (based on experience and schooling) relative to firm-specific skills (based on tenure and training), as observed for Kenya, although the latter are associated with larger productivity gains (van Biesebroeck 2011).7 3. What is the relationship between exports and job quality within country- sectors? 3.1 Productivity and wage premium in export activities at the country-sector level While the previous section established a labor productivity premium at the aggregate level of countries for HICs and LMICs, and a wage premium for HICs, this section examines whether such premia can also be detected at the level of sectors within countries. We therefore specify the following labor productivity regression model: = + _ + + (1) where LP designates labor productivity (value added per worker) in logs, c a country, s a sector, and a the type of economic activity (direct export activity or rest of the sector). EX_dummy is a dummy taking the value of 1 if a worker is linked to an export activity a within sector s, and 0 if the worker is linked to non-export activities in the sector. We rely on a vector of unobserved country and sector fixed effects. Similarly, we specify the following wage rate regression model: = + _ + + (2) where W designates the wage rate (labor compensation per worker) in logs. Table 3 shows the mean differences, as specified in equation (1) (columns 1 to 3) and equation (2) (columns 4 to 6). Surprisingly, our results for 2019 do not find a significant difference in average productivity and wage rates between export and non-export activities within country-sectors in our sample of 60 countries, nor if we focus on HICs and LMICs separately. The absence of a productivity and wage rate premium for export activities within country-sectors could be driven by internationally uncompetitive sectors in the economy. Focusing on sectors only that have a revealed comparative advantage, as defined by the Balassa (1965) index does not change our results. There seems to be no productivity and wage rate premium for workers in export activities (results available upon request). 7While not the focus of this paper, it is important to highlight that workers can also benefit from trade liberalization through the consumer channel. Research for the United States finds, for instance, that increased trade with China benefitted consumers through substantial price declines (Jaravel and Sager 2019). In India, on the other hand, trade liberalization primarily benefitted producers by providing them access to cheaper inputs and a larger variety of inputs, lowering their production costs which resulted in reduced marginal costs and higher markups. While these cost savings were not fully transferred to consumers in the form of lower prices, consumers could nonetheless have benefitted from access to higher quality products and longer-term dynamic gains (de Loecker et al. 2016). 9 Table 3: Within country-sector productivity and wage differences between export activities and the rest of the sector, 2019 (1) (2) (3) (4) (5) (6) All HICs LMICs All HICs LMICs VARIABLES LP LP LP W W W export dummy -0.007 0.010 -0.018 -0.013 0.003 -0.024 (0.681) (0.611) (0.447) (0.373) (0.856) (0.302) Constant 3.893*** 4.642*** 3.334*** 2.988*** 3.852*** 2.347*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 4,787 2,044 2,743 4,817 2,051 2,766 R-squared 0.776 0.749 0.658 0.786 0.693 0.611 Country YES YES YES YES YES YES Sector FE YES YES YES YES YES YES Source: WB staff computations. Note: *** p<0.01, ** p<0.05, * p<0.1. pval in parentheses. What could explain the absence of a productivity and wage rate premium for export activities within sectors – even in sectors of revealed comparative advantage? First, the learning-from-exporting theory suggests that firms increase their productivity and/or quality as they engage in exporting. In the case of Egyptian rug production, for example, this has led to a situation where new exporters reduced their output per hour worked, as global buyers were demanding higher-quality rugs which take a longer time to produce. As long as exporters find it profitable to produce rugs despite falling productivity levels (because their global buyers pay more), they will continue to engage in export activities (Atkin, Khandelwal and Osman 2017). 8 Second, demonstration and competition effects can result in learning spillovers from exporting to domestic firms within sectors (e.g., Crespo and Fontoura 2007, Alvarez and Lopez 2008) which could give domestic firms a productivity boost. It may be further augmented by the aspiration of domestic producers to become exporters. Section 3.2 therefore will explore whether increases in exports in a country-sector are associated with higher labor productivity and wage rates, also accounting for spillover effects from export to non-export activities. 3.2 Allowing for spillovers to non-export activities While the previous section established a labor productivity premium at the aggregate level of countries both for HICs and LMICs, and a wage premium for HICs, this section examines the relationship between exports and job quality within country-sectors. We therefore specify the following labor productivity regression model applying a fixed effects estimator: = + + + + (3) where LP designates labor productivity (value added per worker) in logs, c a country, s a sector, and t the time. EX is the value of gross exports in logs. We rely on a vector of unobserved country-sector, country-time and sector-time fixed effects. To allow for spillovers from export to non-export activities, we additionally assess the relationship between exports and labor productivity for export activities x and non-export activities n in sector s 8 Controlling for rug specifications and quality, however, shows a higher productivity for exporting firms. 10 separately. We therefore replace the dependent variable in equation (3) with activity-specific productivity measures as follows: , = + + + + (3a) , = + + + + (3b) Similarly, we specify the following wage rate regression models: = + + + + (4) , = + + + + (4a) , = + + + + (4b) where W designates the wage rate (labor compensation per worker) in logs. The findings suggest that export-labor productivity and export-wage rate elasticities in a country- sector are positive in the full sample of 60 countries (Figure 4). A 10 percent increase in exports over the period 1995-2019 was associated with a 0.95 percent increase in average labor productivity and a 0.6 percent increase in the average wage rate (in constant 2015 terms). These relationships are stronger for LMICs compared to HICs. The full regression results are shown in Appendix 9. Figure 4: Export-labor productivity and export-wage rate elasticities, 1995-2019, overall and by income level Export-labor productivity elasticities Export-wage rate elasticities 0.16 0.16 0.14 0.14 0.12 0.12 0.10 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00 All HICs LMICs All HICs LMICs All HICs LMICs All HICs LMICs All HICs LMICs All HICs LMICs LP in sector LP of export LP of non-export W in sector W of export W of non-export activities activities activities activities Source: WB staff computations. Note: All results are statistically significant at the 1% level. Blue bars refer to equations (3) and (4), orange bars to equations (3a) and (4a), and yellow bars to equation (3b) and (4b). See Appendix 3 for country coverage, Appendix 4 for sector coverage and Appendix 9 for the full regression results. Not correcting for the possible endogeneity of exports could lead to overestimating their productivity impact, as more better performing firms are also more likely to participate in global markets. To mitigate this potential bias, the analysis is conducted separately for exporting and non-exporting activities, using sector-level exports as the indicator of trade. Should the influence be attributed to the tendency of more productive firms to enter global markets, one would expect to see a negative correlation between sectoral exports and labor productivity among non-exporting activities. 11 Focusing on the relationship between exports and job quality for export activities within country- sectors only (Figure 4, orange bars), elasticities increase for export activities compared to the average sector (blue bars). Specifically, a 10 percent growth in exports in LMICs (HICs, resp.) is associated with productivity increases of 1.49 (1.27, resp.) percent for export activities. However, exports are also associated with productivity increases in non-export activities within sectors, suggesting positive productivity spillovers (yellow bars). In LMICs, raising exports by 10 percent is correlated with productivity gains of 1.09 percent in non-export activities (left panel). The general patterns are similar for wage rates, although elasticities tend to be generally lower, as discussed in section 2.2. 3.3 Differences by type of global value chain participation and broad sectors This section focuses on differences in the relative magnitude of elasticities by GVC taxonomy groups and broad sectors. Splitting the country sample by the GVC country typology of the World Development Report 2020 (World Bank 2020) suggests major productivity boots from exports for countries specialized in limited manufacturing compared to those specialized in commodities. While a 10 percent increase in exports is only associated with productivity gains of 0.16 percent for the latter group, the gains rise to 1.35 percent for the former group. The positive relationship is slightly smaller for countries in advanced manufacturing and services and further declines for countries specialized in innovative activities (Figure 5, left panel). While the gains are stronger when focusing on export activities only (Figure 5, orange bars), the results again point to strong productivity spillovers to non-export activities (yellow bars). The relationship between exports and wage rates across GVC typology groups is similar but elasticities are smaller (right panel). Interestingly, there is no positive association for commodity exporters overall, which is driven by the lack of wage spillovers from export to non-export activities. The full regression results are shown in Appendix 10. Figure 5: Export-labor productivity and export-wage rate elasticities, 1995-2019, by GVC taxonomy group Export-labor productivity elasticities Export-wage rate elasticities 0.20 0.20 0.18 0.18 0.16 0.16 0.14 0.14 0.12 0.12 0.10 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00 commodity limited commodity limited commodity limited advanced innovation advanced innovation advanced innovation commodity limited commodity limited commodity limited advanced innovation advanced innovation advanced innovation LP in sector LP in export LP in non-export W in sector W in export W in non-export activities activities activities activities Source: WB staff computations. Note: All results are statistically significant at the 1% level. Missing bars are statistically insignificant. Blue bars refer to equations (3) and (4), orange bars to equations (3a) and (4a), and yellow bars to equation (3b) and (4b). See Appendix 3 for country coverage and taxonomy group, Appendix 4 for sector coverage, and Appendix 10 for the full regression results. Commodity = commodity exporters, limited = limited manufacturing, advanced = advanced manufacturing and services, and innovation = innovative activities (see World Bank 2020). 12 Splitting the sample into broad economic sectors shows that export growth is associated with the strongest labor productivity boosts in manufacturing and agriculture, followed by business services (Figure 6, upper panel). Within manufacturing, low-tech and medium-to-high tech manufacturing seem to benefit more strongly from exports compared to medium-tech manufacturing sectors which are more resource-intensive (such as rubber and plastics, minerals, and metals). 9 In addition, we also find high productivity spillovers from exports to non-export activities in these sectors. By contrast, no positive link between exports and productivity can be established for mining, while the relationship is relatively small for other services. The results for wage rates are similar in nature but elasticities are again smaller. Appendices 11 and 12 show the full regression results. Another source for productivity gains is labor reallocation towards more productive activities and sectors within countries which will be explored in more detail in section 4. Figure 6: Export-labor productivity and export-wage rate elasticities, 1995-2019, by broad sector Export-labor productivity elasticities, by sector Export-wage rate elasticities, by sector 0.20 0.20 0.18 0.18 0.16 0.16 0.14 0.14 0.12 0.12 0.10 0.10 0.08 0.08 0.06 0.06 0.04 0.04 0.02 0.02 0.00 0.00 LP in sector LP in export activities W in sector W in export activities LP in non-export activities W in non-export activities Source: WB staff computations. Note: All results shown are statistically significant at the 1% level. Missing bars are statistically insignificant. Blue bars refer to equations (3) and (4), orange bars to equations (3a) and (4a), and yellow bars to equation (3b) and (4b). See Appendix 3 for country coverage and taxonomy group, Appendix 4 for sector coverage and Appendices 11 and 12 for full regression results. agric = agriculture, mfg_medhig = medium-to-high-tech manufacturing, mfg_med = medium-tech manufacturing, mfg_low = low-tech manufacturing, bserv = business services, otherserv = other services. 4. What drives aggregate labor productivity and wage rate gains? 4.1 Growth decomposition method In this section, we decompose the growth of countries’ aggregate productivity and wage rates to understand the extent to which their growth is driven by movements of workers towards more productive and better-paid export activities or sectors. Define the aggregate and per worker values in country c at time t as ≡ ∑ , ≡ ∑=1 , =1 ≡ / , and = , where Y is value added, L employment and y labor productivity. 9 Sectors are classified using the UNIDO classification by technology intensity: https://stat.unido.org/content/learning- center/classification-of-manufacturing-sectors-by-technological-intensity-%28isic-revision- 4%29;jsessionid=B99E902A3918AB9F3DF9859923DFC4F4 13 Also define the activity a shares in the total as = and = , where and designate the employment and value-added shares, respectively. With this notation, the Bennet (1920) decomposition of changes in productivity is: − 0 0 − 0 ( � − � ) � = � � � �+ � ( − 0 ) 0 0 0 � 0 =1 =1 � ≡ ( + 0 )/2, where � ≡ ( + 0 )/2. � ≡ ( + 0 )/2, and = Average labor productivity (wage rate) in country c at time t. The decompositions are based on 1995 and 2019 as start and end years with 0 = 1995, 1 = 2019. 10 One challenge arises from the fact that our dataset contains both a sector dimension (45 sectors) and an activity dimension (export and non-export activities) within sectors. Figure 7 illustrates the possible sources of productivity growth at the sector-activity level. From the perspective of export activities in sector 1, there can be productivity gains within the sector-activity (top left panel highlighted in yellow). Three types of labor movements can further contribute to productivity gains, as depicted by the white arrows: First, workers can move within sector 1 from non-export to export activities (between activities, i.e., from orange to yellow). Second, labor engaged in export activities can reallocate from other sectors to sector 1 (between sectors, i.e., from green to yellow). Third, there can also be labor movement from non-export activities outside of sector 1 (between sector-activities, i.e., from blue to yellow). Figure 7: Possible movement of labor towards export activities in sector 1 (yellow section) Activities Sectors Export Non-export within sector-activity gains labor movement within sector 1 to export 1 activity (between activities) 2 3 labor movement within export activity to labor movement to export activity in sector 1 .. sector 1 (between sectors) (between sector-activities) S Source: WB staff illustration. In the following, we differentiate between three types of growth decompositions: 1. Growth decomposition by activities, where the economy consists of two activities, namely export and non-export activities. we ran year-on-year decompositions over the period 1995-2019, as well as decompositions for 1995-2000, 2000- 10 Alternatively, 2005, 2005-2010, 2010-2015, and 2015-2019. A key observation is that the start and end years matter considerably. Results are available upon request. 14 This decomposition shows the growth contributions (i) within export activities, (ii) within non-export activities, (iii) due to labor movement to export activities (between), and (iv) due to labor movement to non-export activities (between). Movement of labor across sectors is possible but cannot be detected separately. 2. Growth decomposition by sectors, where the economy consists of 45 sectors. This decomposition shows the growth contributions (i) within sectors, and (ii) between sectors due to labor movements to more productive sectors. While this decomposition can be performed for export and non-export activities separately, movement of labor towards export (and non-export) activities cannot be detected separately. 3. Growth decomposition by sector-activity pairs, where the economy consists of 90 pairs (45 sectors X 2 activities). This decomposition shows the growth contributions (i) within sector-activity pairs (yellow panel in Figure 7), and (ii) between sector-activity pairs due to labor movements to more productive sector-activity pairs (orange, green and blue panels in Figure 7). Movement of labor towards export (and non-export) activities within and across sectors cannot be detected separately. 4.2 Contributions to productivity and wage rate growth within and between activities Figure 8 shows the decomposition of productivity and wage rate growth within and between activities for the full sample of 60 countries, and for HICs and LMICs separately (using their 1995 income classifications). 11 Overall, almost three quarters of the productivity growth was driven by growth within non-export activities, while growth within export activities contributed around a quarter. Growth within activities can also include movement of workers from less productive to more productive sectors. The results also suggest that productivity growth within export activities was slightly higher for HICs (almost 28 percent) than for LMICs (22 percent). Appendices 13 to 14 show the decompositions for each country separately. Productivity growth within export activities among HICs was highest for Luxembourg, Singapore, and Ireland, while Saudia Arabia and Greece stood out among the group of LMICs. However, the growth contributions due to labor movements towards more productive or better-paid activities are negligeable, possibly reflecting labor market frictions such as discrepancies or imperfect information between the required capabilities and available worker skills or geographical barriers (Donovan and Schoellman 2023). Interestingly, the between contributions in LMICs are slightly negative. In other words, workers have moved from more productive non-export activities to less-productive export activities which contributed negatively to overall productivity growth (-1.5 percent). Notable exceptions include, e.g., Mexico and Viet Nam where reallocation towards more productive export activities contributed 4.8 and 7.2 percent, respectively (Appendix 14). Another explanation might be that export activities are geographically more concentrated within countries, creating larger barriers to labor mobility (Donovan and Schoellman 2023). In many countries, exporting firms are clustered around core regions with good infrastructure such as special economic zones and close to ports or large foreign markets. Labor movement from non-export to export activities could thus require workers to overcome larger geographical barriers than movement within activities. 12 11Growth decompositions here and in the following are normalized to 100 percent. Growth rates are available upon request. 12In additional robustness checks, we changed the definition of export activities to also include indirect exports of domestic sectors supplying inputs to export activities. However, using this alternative definition does not increase the between effects. 15 Table 4 shows the decomposition of productivity and wage rate growth, respectively, by five broad sectors (agriculture, mining, manufacturing, business services and other services). In manufacturing, growth within export activities contribute 44 and 41 percent to productivity and wage rate growth, respectively, over the period 1995-2019, and more (less) in HICs (LMICs). Growth within export activities contribute more than a fifth to productivity and wage rate growth in agriculture, with a higher share in HICs compared to LMICs. Growth within export activities in business services contributes a third to productivity growth of HICs, and over a fifth to growth in LMICs. Figure 8: Decomposition of productivity and wage rate growth, 1995-2019, within and between two activities Productivity growth Growth of wage rate 81.0 79.3 78.1 74.8 74.4 90.0 90.0 69.1 80.0 80.0 70.0 70.0 60.0 60.0 50.0 50.0 28.8 25.3 40.0 40.0 23.9 22.5 21.3 19.2 30.0 30.0 20.0 20.0 1.1 0.9 1.2 0.3 0.5 0.4 0.2 0.0 10.0 10.0 0.0 0.0 -10.0 -10.0 -0.3 -0.4 -0.2 -1.5 within within rest between between rest within exports within rest between between rest exports exports exports All HICs LMICs All HICs LMICs Source: WB staff computations. Note: within exports = growth within export activities; within rest = growth within rest of the economy; between exports = growth due to labor reallocation to export activities; between rest = growth due to labor reallocation to rest of the economy. See Appendix 3 for country coverage and Appendices 13 and 14 for the country decompositions. Averages for LMICs and All exclude Saudi Arabia for productivity decompositions and Argentina for wage rate decompositions. Table 4: Decomposition of productivity and wage rate growth, 1995-2019, within and between export activities and the rest of the economy, by broad sector Labor productivity Wage rate within exports within between between within within between between Sample Broad sector rest exports rest exports rest exports rest All Agriculture 23.0 77.0 -0.1 0.1 22.1 77.5 0.1 0.3 All Mining 26.3 68.4 2.0 3.3 27.2 71.1 1.4 0.3 All Manufacturing 43.9 55.5 0.2 0.4 41.2 57.8 0.6 0.4 All Business Services 26.2 73.7 0.1 0.0 28.6 71.4 -0.4 0.4 All Other Services 1.3 94.6 3.8 0.2 2.6 99.1 -1.6 -0.1 HICs Agriculture 27.0 72.2 0.0 0.8 24.6 73.8 0.6 1.1 HICs Mining 25.3 65.7 4.5 4.5 33.4 61.5 2.6 2.5 HICs Manufacturing 49.0 49.9 0.4 0.6 45.6 53.3 0.6 0.5 HICs Business Services 32.9 67.6 -0.5 0.0 38.7 62.5 -2.1 0.9 HICs Other Services -1.8 92.7 8.4 0.6 2.7 98.7 -1.3 0.0 LMICs Agriculture 20.2 80.5 -0.2 -0.4 20.3 80.2 -0.2 -0.2 LMICs Mining 27.1 70.5 0.1 2.3 22.2 78.7 0.5 -1.5 LMICs Manufacturing 39.4 60.5 0.0 0.2 37.4 61.7 0.7 0.3 LMICs Business Services 21.3 78.2 0.5 0.0 21.8 77.3 0.7 0.1 LMICs Other Services 3.7 96.1 0.3 0.0 2.6 99.4 -1.8 -0.2 Source: WB staff computations. Note: within exports = growth within export activities; within rest = growth within rest of the economy; between exports = growth due to labor reallocation to export activities; between rest = growth due to labor reallocation to rest of the economy. See Appendix 3 for country coverage. Dark blue = highest values, dark red = lowest values. 16 4.3 Contributions to productivity and wage rate growth within and between sectors This section decomposes labor productivity and wage growth within and between sectors. In the full sample of 60 countries, labor productivity growth was determined by growth within the 45 sectors (Figure 9). Focusing on export activities only, labor productivity within sectors increased by 87 percent over the period 1995-2019, while movements of workers to more productive export activities across sectors contributed 13 percent. In HICs, the within contribution (98 percent) dominates the between contribution (2 percent). In LMICs, within sector productivity growth contributed 80 percent, while growth due to worker reallocation to export activities in more productive sectors contributed 20 percent (left panel). The findings from the previous section and this decomposition seem to suggest that the productivity growth contribution from export activities mainly stems from growth within such activities; within export activities they come mainly from within the same sector. For non-export activities in LMICs, the reallocation effect across sectors was larger than for export activities (34 percent) while growth of the within-component was relatively smaller (66 percent). In HICs, the growth stemming from reallocation towards more productive sectors only contributed 8 percent, which is larger than for export activities, but substantially lower than the within-growth contribution of 92 percent (right panel). Figure 9: Decomposition of productivity growth, 1995-2019, within and between 45 sectors Export activities Non-export activities 98 100 87 100 92 80 77 80 80 66 60 60 40 34 40 20 23 20 13 20 8 2 0 0 within sectors between sectors within sectors between sectors All HICs LMICs All HICs LMICs Source: WB staff computations. Note: See Appendix 3 for country coverage and Appendix 4 for sector coverage. The labor productivity growth decomposition average excludes Norway and South Africa due to extremely high values for export activities, and Greece, Italy, Luxembourg, and Saudi Arabia due to extremely high values for the rest of the economy (see Appendices 15 and 16). Similarly, growth of the average wage rate was dominated by growth within the 45 sectors (Figure 10). Focusing on export activities, the wage rate growth within sectors contributed 90 percent over the period 1995-2019, while movements of workers to better-paid export activities across sectors contributed 10 percent. The within-sector contributions were higher for HICs than for LMICs, whereas the between- sector component was higher for LMICs (left panel). For non-export activities, the within-sector growth component was slightly lower than those for export activities. It was 84 percent overall but lower again for LMICs (77 percent) where labor reallocation towards better-paid sectors contributed 23 percent. Appendices 15 and 16 show the labor productivity and wage rate growth decompositions within and between sectors for all 60 countries in the sample individually. 17 Figure 10: Decomposition of wage rate growth, 1995-2019, within and between 45 sectors Export activities Non-export activities 96 93 100 90 86 100 84 77 80 80 60 60 40 40 23 14 16 20 10 20 7 4 0 0 within sectors between sectors within sectors between sectors All HICs LMICs All HICs LMICs Source: WB staff computations. Note: See Appendix 3 for country coverage and Appendix 4 for sector coverage. The wage rate growth decomposition average excludes Brazil and Ukraine due to extremely high values for export activities, as well as Italy and South Africa due to extremely high values for the rest of the economy (see Appendices 15 and 16). 4.4 Contributions to productivity and wage rate growth within and between sector- activity pairs Figure 11 shows the final decomposition of productivity and wage rate growth based on 90 sector- activity pairs (45 sectors x two activities). In HICs, growth within sector-activity pairs contributed over 89 percent to productivity growth, while movement of workers to higher productivity sector-activity pairs – either between activities within the same sector or between sectors within the same activity or from other sectors and activities – contributed less than 11 percent. In LMICs, growth within sector-activity pairs contributed almost 70 percent to labor productivity growth, while movement of workers to more productive sector-activity pairs contributed 30 percent. As was the case before, the between component in LMICs is stronger than in HICs. Based on the previous findings, we can assume that the between effect is driven by labor movement from other sectors within export activities rather than from reallocation from non-export activities within the same sector or from another sector. Figure 11: Decomposition of productivity and wage rate growth, 1995-2019, within and between sector- activity pairs Productivity growth Wage rate growth 97.3 100 89.4 100 77.7 82.2 80 69.4 80 71.1 60 60 40 30.6 40 28.9 22.3 17.8 20 10.6 20 2.7 0 0 within sector-activity pairs between sector-activity pairs within sector-activity pairs between sector-activity pairs All HICs LMICs All HICs LMICs Source: WB staff computations. Note: See Appendix 3 for country coverage and Appendix 4 for sector coverage. The labor productivity growth decomposition average excludes Italy, Luxembourg, Greece, and Saudi Arabia due to extremely high values (see Appendices 17 and 18). 18 In HICs, growth within sector-activity pairs contributed over 97 percent to wage rate growth, while labor reallocation to higher paid jobs between sector-activity pairs contributed less than 3 percent. In LMICs, the within contribution to wage rate growth was slightly higher than for productivity (71 percent) while the between contribution reached almost 29 percent (Figure 11, right panel). Appendices 17 and 18 show the labor productivity and wage rate growth decompositions within and between sector-activity pairs for all 60 countries in the sample individually. 5. Conclusions This paper contributes to the broader research question whether trade helps increase the quality of jobs. It first assessed if productivity and wage rate premia can be confirmed at the aggregate country level. Using a sample of 26 HICs and 34 LMICs covering export and non-export activities over the period 1995- 2019, our analysis finds evidence at the country level that workers linked to export activities are more productive and better paid in HICs. In LMICs, the productivity premium is also confirmed, while the findings reject the existence of a wage rate premium for export activities. This difference could be explained by various factors, such as the higher capital intensity of firms engaged in GVCs putting downward pressure on less-skilled workers, as well as less competitive labor markets that do not optimally allocate workers to the most productive firms at market-clearing wage levels in LMICs. We also examined if the observed patterns hold at the country-sector level but surprisingly cannot confirm a productivity or wage rate premium for export activities, even in sectors of revealed comparative advantage. One possible explanation may be that exporters tend to focus more strongly on higher quality exports to meet stricter global standards which can lower their overall productivity levels. An alternative explanation may be the existence of demonstration and competition effects resulting in learning spillovers from export to non-export activities within sectors. Our results find a positive association between exports and job quality at the country-sector level; it can be attributed to productivity and wage rate gains within export activities and also positive spillovers to non-export activities within sectors. While higher elasticities can be expected for export activities, the gains from productivity spillovers to non-export activities are not substantially smaller. In LMICs, raising exports by 10 percent is correlated with productivity gains of 1.49 percent for export activities and 1.09 percent for non-export activities. The correlation between exports and productivity is smaller for HICs. For both types of countries, we find smaller correlations between exports and the average wage rate, although the magnitude of the estimated coefficients is generally smaller. Countries’ specialization in GVCs and broad sectors also matters for the relationship between exports and job quality, with manufacturing, agriculture, and business services showing stronger links. Moving from commodities to limited manufacturing promises a huge boost in productivity gains from exports. The positive relationship is slightly smaller for countries in advanced manufacturing and services and further declines for countries specialized in innovative activities. In line with these results, elasticities are largest in low- and medium-high-tech manufacturing sectors and agriculture, followed by business services. A productivity growth decomposition by activities suggests that growth within rather than between activities was the driving force (although less pronounced for export activities). We decomposed aggregate labor productivity and wage rate growth over the period 1995-2019 to examine the extent to which growth is driven by movements of workers towards more productive activities, sectors, and sector- activity pairs. In LMICs, growth within export activities contributed almost 23 percent to overall productivity growth while growth within the rest of the economy accounted for 79 percent. By contrast, the growth contribution due to workers’ reallocation from non-export to export activities was slightly 19 negative in LMICs, i.e., workers moved from more productive non-export activities to less productive export activities, with Mexico and Viet Nam being notable exceptions. Two possible explanations for the low movement of workers towards more productive export activities include larger skills gaps and larger geographical barriers that need to be overcome. Possible discrepancies in required capabilities or worker skills between export and non-export activities may prevent workers from moving towards more productive export activities. Another explanation might be that export activities are geographically more concentrated within countries (e.g., closer to markets, ports, special economic zones), creating larger barriers to labor mobility compared to movements within the same activities. Within export activities, productivity increases were driven by within-sector growth although labor movement towards more productive sectors also matters in LMICs. Decomposing productivity growth of export activities by sectors suggests that labor movement to more productive sectors accounts for 20 percent in LMICs, whereas growth within sectors contributes 80 percent. Similarly, the decomposition by sector-activity pairs for LMICs suggests that growth within sector-activities accounts for 69 percent of overall productivity growth, whereas labor reallocation across sector-activities accounts for 30 percent. Future research may want to simultaneously decompose labor productivity growth along the sector and activity dimensions to directly account for all possible sources of productivity growth. Finally, labor reallocation towards more productive sectors contributes more strongly to the productivity growth of export activities in LMICs. Previous research suggests that the between-sector contribution increases with countries’ income levels: while it is marginal in HICs, factor reallocation between sectors plays a stronger role in emerging economies and has contributed roughly half of the productivity growth in low-income countries since 2000 (Dieppe and Matsuoka 2021). In our country sample, the productivity growth contribution due to labor movement towards more productive sectors is a fifth for export activities and roughly a third for non-export activities in LMICs. Given that our data sample for LMICs covers more middle-income countries, our results are aligned with previous findings. 20 References Alonso Alfaro-Ureña, Isabela Manelici, and Jose P. Vasquez. 2022. “The Effects of Joining Multinational Supply Chains: New Evidence from Firm-to-Firm Linkages”. The Quarterly Journal of Economics 137 (3): 1495-1552. Alvarez, Roberto and Ricardo A. López. 2008. “Is Exporting a Source of Productivity Spillovers?”, Review of World Economics 144(4): 72-749. Amiti, Mary, and Donald R. Davis. 2012. “Trade, Firms, and Wages: Theory and Evidence.” The Review of Economic Studies 79(1): 1-36. Amiti, Mary, and Jozef Konings. 2007. “Trade Liberalization, Intermediate Inputs, and Productivity: Evidence from Indonesia.” American Economic Review 97(5): 1611-1638. Amodio, F., N. de Roux, P. Medina, and M. Morlacco. 2022. “Labor market power, self-employment, and Development.” CEPR Discussion Papers Number 17543). Centre for Economic Policy Research Arnold, Jens M., Beata Javorcik, and Aaditya Mattoo. 2011. “The Productivity Effects of Services Liberalization: Evidence from the Czech Republic”. Journal of International Economics 85 (1): 136-46. Atkin, D., A. K. Khandelwal, and A. Osman. 2017. “Exporting and Firm Performance: Evidence from a Randomized Experiment.” The Quarterly Journal of Economics 132: 551-615. Bennet, T. L. 1920. “The Measurement of Changes in the Cost of Living.” Journal of the Royal Statistical Society 83(3): 455-462. Brooks, W. J., J. P. Kaboski, Y. A. Li, and W. Qian. 2021. “Exploitation of labor? Classical monopsony power and labor’s share.” Journal of Development Economics, 150(102627). Crespo, N., and M. Fontoura. 2007. “Determinant Factors of FDI Spillovers—What Do We Really Know?” World Development 35 (3): 410-25. Criscuolo, C. and J. Timmis. 2017. “The Relationship Between Global Value Chains and Productivity.” International Productivity Monitor 32: 61-83. Das, D. K., A. A. Erumban, S. Aggarwal, and P. C. Das. 2017. “Measuring Productivity at the Industry Level.” The India KLEMS data base manual 2017 (Version 4). De Loecker, J. K., P. K. Goldberg, A. K. Khandelwal, and N. Pavcnik. 2016. “Prices, Markups, and Trade Reform.” Econometrica 84(2): 445–510. Dieppe, A. and H. Matsuoka. 2021. “Sectoral Decomposition of Convergence in Labor Productivity: A Re- examination from a New Dataset.” World Bank Policy Research Working Paper No. 9767, Washington, DC: World Bank Group. Donovan, K. and T. Schoellman. 2023. “The role of labor market frictions in structural transformation.” Oxford Development Studies 51. Fernandes, Ana M., and Caroline Paunov. 2012. “Foreign direct investment in services and manufacturing productivity: Evidence for Chile.” Journal of Development Economics 97(2): 305-321. Havranek, Tomas, and Zuzana Irsova. 2011. “Estimating vertical spillovers from FDI: Why results vary and what the true effect is.” Journal of International Economics 85(2): 234-244. Horvát, P., C. Webb and N. Yamano. 2020. “Measuring employment in global value chains”, OECD Science, Technology and Industry Working Papers, 2020/1, Paris: OECD. Jaravel, X. and E. Sager. 2019. “What are the price effects of trade? Evidence from the US and implications for quantitative trade models.” CEP Discussion Papers dp1642, Centre for Economic Performance, LSE. Javorcik, Beata. 2004. “Does foreign direct investment increase the productivity of domestic firms? In search of spillovers through backward linkages.” American Economic Review 94(3): 605-627. La Porta, Rafael and Andrei Shleifer. 2014. “Informality and Development.” Journal of Economic Perspectives 28(3): 109-26. Reshef, A. and G. Santoni. 2023. “Are Your Labor Shares Set in Beijing? The View through the Lens of Global Value Chains.” European Economic Review 155(104459). 21 Rodrik, D. 2018. “New Technologies, Global Value Chains, and Developing Economies.” NBER Working Paper 25164, National Bureau of Economic Research, Cambridge, MA. Van Biesebroeck, Johannes. 2011. “Wages Equal Productivity. Fact or Fiction? Evidence from Sub Saharan Africa.” World Development 39(8): 1333-1346. 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 Economics 148: 235-67. World Economic Forum. 2015. What Companies Want from the World Trading System. Geneva: World Economic Forum. World Bank. 2020. World Development Report 2020: Trading for Development in the Age of Global Value Chains, Washington, DC: World Bank Group. 22 Appendices Appendix 1: Datasets for econometric analysis The dataset covers 60 countries, drawing on the OECD Trade in Employment (TiE) and OECD Trade in Value Added (TiVA) databases to obtain value added, labor compensation and employment measures for 45 sectors over the period 1995-2020 (see Appendix 3 and 4 for the country and sector coverage). Total value added, labor compensation and employment are available for each sector in a country. In addition, the datasets publish these measures for direct export activities, i.e., domestic value added in exports, compensation in exports, and employment in exports. By taking their difference (total minus export-linked) one obtains measures for value added, labor compensation and employment for the rest of the economy (or sector). These different measures allow to compute the labor productivity (value added per worker) and the average wage rate (labor compensation per worker) for export activities and the rest of the economy. We deflated the value added and labor compensation data, using value-added deflators for four broad sectors (agriculture, non- manufacturing industry, manufacturing, and services) with 2015 as the base year from WDI. The countries are listed in Appendix 3. Domestic employment by industry statistics are drawn from various sources such as OECD’s Annual National Accounts and Structural Analysis (STAN) databases, official national statistics and, in a very few cases, from research projects such as, for example, India KLEMS (Das et al., 2017). Employment is defined as persons engaged in production activity within the National Accounts boundary of the resident institutional unit (domestic concept) and includes both employees and self-employed. If there are no estimates of employment by industry in official National Accounts statistics, then Labor Force Survey (LFS) statistics are exploited. National Accounts are preferred to LFS as a source for employment by industry since LFS are usually based on residential households and thus exclude non-resident workers while including resident workers commuting abroad (national concept). Appendix 2: Measures of employment and labor compensation in exports Domestic employment in gross exports, y, can be computed as follows: = ( − )− e is a vector of exports with N designating the number of goods and services industries, AD the N x N domestic coefficient matrix denoting the amount of domestic products used in the production of goods or services, I the N x N identity matrix with ones on the diagonal and zeros elsewhere, such that( − )− is the Leontief inverse. The latter ensures that all output related to exports is considered, not only that from the export sector, but also of other domestic sectors that contribute through the delivery of intermediate inputs. c is a 1 x N vector with domestic employment over output coefficients in industry n. Employment in exports for a country can be further decomposed into three elements: • Direct: employment in industry n used in the production of goods and services exported by industry n. • Indirect: employment in other, upstream, domestic industries (different from industry n) that is embodied in the exports of industry n. • Re-imported: employment by any industry, used to produce exports of intermediate goods or services subsequently embodied in imports used in the production of exports by industry n. Our analysis relies on precomputed measures from OECD’s Trade in Employment database (2023 release), covering the period 1995-2020. Labor compensation in exports is computed similarly. For more details, see Horvát, Webb and Yamano (2020) “Measuring employment in global value chains”. 23 Appendix 3: Country sample and World Bank income classifications, 1995, 2005 and 2019 Country iso3 income1995 income2005 income2019 1 Argentina ARG UM UM UM 2 Australia AUS H H H 3 Austria AUT H H H 4 Belgium BEL H H H 5 Bulgaria BGR LM UM UM 6 Brazil BRA UM UM UM 7 Canada CAN H H H 8 Switzerland CHE H H H 9 Chile CHL UM UM H 10 China CHN L LM UM 11 Colombia COL LM LM UM 12 Costa Rica CRI LM UM UM 13 Cyprus CYP H H H 14 Czechia CZE UM H H 15 Germany DEU H H H 16 Denmark DNK H H H 17 Egypt, Arab Rep. EGY LM LM LM 18 Spain ESP H H H 19 Estonia EST LM H H 20 Finland FIN H H H 21 France FRA H H H 22 United Kingdom GBR H H H 23 Greece GRC UM H H 24 Croatia HRV UM UM H 25 Hungary HUN UM H H 26 Indonesia IDN LM LM LM 27 India IND L LM LM 28 Ireland IRL H H H 29 Iceland ISL H H H 30 Israel ISR H H H 31 Italy ITA H H H 32 Japan JPN H H H 33 Kazakhstan KAZ LM LM UM 34 Korea, Rep. KOR H H H 35 Lithuania LTU LM UM H 36 Luxembourg LUX H H H 37 Latvia LVA LM UM H 38 Mexico MEX UM UM UM 39 Malta MLT UM H H 40 Malaysia MYS UM UM UM 41 Netherlands NLD H H H 42 Norway NOR H H H 43 New Zealand NZL H H H 44 Peru PER LM LM UM 45 Philippines PHL LM LM LM 46 Poland POL LM UM H 47 Portugal PRT H H H 48 Romania ROU LM UM UM 49 Russian Federation RUS LM UM UM 50 Saudi Arabia SAU UM H H 51 Singapore SGP H H H 52 Slovak Republic SVK LM H H 53 Slovenia SVN H H 54 Sweden SWE H H H 55 Thailand THA LM LM UM 24 56 Türkiye TUR LM UM UM 57 Ukraine UKR LM LM LM 58 United States USA H H H 59 Viet Nam VNM L L LM 60 South Africa ZAF UM UM UM Source: World Bank. Note: H = high-income, UM = upper-middle income, LM = lower-middle income, L = low-income. Appendix 4: Sector coverage Code OECD TiVA sectors D01T02 Agriculture, hunting, forestry D03 Fishing and aquaculture D05T06 Mining and quarrying, energy producing products D07T08 Mining and quarrying, non-energy producing products D09 Mining support service activities D10T12 Food products, beverages and tobacco D13T15 Textiles, textile products, leather and footwear D16 Wood and products of wood and cork D17T18 Paper products and printing D19 Coke and refined petroleum products D20 Chemical and chemical products D21 Pharmaceuticals, medicinal chemical and botanical products D22 Rubber and plastics products D23 Other non-metallic mineral products D24 Basic metals D25 Fabricated metal products D26 Computer, electronic and optical equipment D27 Electrical equipment D28 Machinery and equipment, nec D29 Motor vehicles, trailers and semi-trailers D30 Other transport equipment D31T33 Manufacturing nec; repair and installation of machinery and equipment D35 Electricity, gas, steam and air conditioning supply D36T39 Water supply; sewerage, waste management and remediation activities D41T43 Construction D45T47 Wholesale and retail trade; repair of motor vehicles D49 Land transport and transport via pipelines D50 Water transport D51 Air transport D52 Warehousing and support activities for transportation D53 Postal and courier activities D55T56 Accommodation and food service activities D58T60 Publishing, audiovisual and broadcasting activities D61 Telecommunications D62T63 IT and other information services D64T66 Financial and insurance activities D68 Real estate activities D69T75 Professional, scientific and technical activities D77T82 Administrative and support services D84 Public administration and defence; compulsory social security D85 Education D86T88 Human health and social work activities D90T93 Arts, entertainment and recreation D94T96 Other service activities Activities of households as employers; undifferentiated goods- and services- D97T98 producing activities of households for own use 25 Appendix 5: Difference in average labor productivity, export activities vs. rest of the economy, high-income countries, 1995, 2005, 2019, t-tests High-income countries, 1995 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 27 58.32376 4.753334 24.69905 48.55314 68.09437 Labor productivity, rest of the economy 27 51.73463 3.07069 15.95577 45.42274 58.04652 diff 27 6.589127 2.653666 13.78885 1.134439 12.04382 mean(diff)=mean(lp_expdir-lp_rest) t=2.483 H0:mean(diff)=0 26 Ha:mean(diff)>0 Pr(T>t)=0.0099 High-income countries, 2005 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 31 91.39272 11.34166 63.14772 68.22995 114.5555 Labor productivity, rest of the economy 31 65.02 3.798424 0.14873 57.26258 72.77742 diff 31 26.37272 10.7709 59.96983 4.375607 48.36983 mean(diff)=mean(lp_expdir-lp_rest) t=2.4485 H0:mean(diff)=0 df=30 Ha:mean(diff)>0 Pr(T>t)=0.0102 High-income countries, 2019 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 41 92.71778 9.569983 61.27779 73.37613 112.0594 Labor productivity, rest of the economy 41 69.98319 4.690187 30.03185 60.50397 79.46241 diff 41 22.7346 7.093486 45.42047 8.398126 37.07107 mean(diff)=mean(lp_expdir-lp_rest) t=3.2050 H0:mean(diff)=0 df=40 Ha:mean(diff)>0 Pr(T>t)=0.0013 Note: lp_expdir = labor productivity in export activities, lp_rest = labor productivity in rest of the economy. 26 Appendix 6: Difference in average labor productivity, export activities vs. rest of the economy, low- and middle-income countries, 1995, 2005, 2019, t-tests Low- and middle-income countries, 1995 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 34 14.88049 6.327741 36.89675 2.006607 27.75438 Labor productivity, rest of the economy 34 9.226941 1.243653 7.25168 6.69671 11.75717 diff 34 5.653551 5.958854 34.74579 6.469829 17.77693 mean(diff)=mean(lp_expdir-lp_rest) t=-0.9488 H0:mean(diff)=0 df=33 Ha:mean(diff)>0 Pr(T>t)=0.178 Low- and middle-income countries, 2005 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 30 14.06927 1.512436 8.283951 10.97599 17.16254 Labor productivity, rest of the economy 30 12.66139 1.562665 8.559067 9.465381 15.8574 diff 30 1.407878 0.566511 3.102907 0.249233 2.566522 mean(diff)=mean(lp_expdir-lp_rest) t=2.4852 H0:mean(diff)=0 df=29 Ha:mean(diff)>0 Pr(T>t)=0.0095 Low- and middle-income countries, 2019 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Labor productivity, export activities 20 19.08225 2.198925 9.83389 14.47985 23.68466 Labor productivity, rest of the economy 20 16.48909 1.684595 7.533739 12.9632 20.01499 diff 20 2.593161 1.392772 6.228668 0.321946 5.508267 mean(diff)=mean(lp_expdir-lp_rest) t=1.8619 H0:mean(diff)=0 df=19 Ha:mean(diff)>0 Pr(T>t)=0.0391 Note: lp_expdir = labor productivity in export activities, lp_rest = labor productivity in rest of the economy. 27 Appendix 7: Difference in average wage rate, export activities vs. rest of the economy, high-income countries, 1995, 2005, 2019, t-tests High-income countries, 1995 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 27 28.92765 2.252792 11.70585 24.29697 33.55833 Wage rate, rest of the economy 27 26.82784 1.761325 9.152116 23.20739 30.4483 diff 27 2.099809 0.612173 3.180944 0.841469 3.358149 mean(diff)=mean(w_expdir-w_rest) t=3.4301 H0:mean(diff)=0 df=26 Ha:mean(diff)>0 Pr(T>t)=0.0010 High-income countries, 2005 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 31 2.752607 15.32587 30.87586 42.119 2.752607 Wage rate, rest of the economy 31 2.279023 12.68906 28.7526 38.06137 2.279023 diff 31 0.808698 4.50264 1.43886 4.742024 0.808698 mean(diff)=mean(w_expdir-w_rest) t=3.8215 H0:mean(diff)=0 df=30 Ha:mean(diff)>0 Pr(T>t)=0.0003 High-income countries, 2019 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 41 39.68023 3.28187 21.01422 33.04732 46.31313 Wage rate, rest of the economy 41 35.94488 2.734896 17.51188 30.41744 41.47231 diff 41 3.735351 0.7933593 5.079978 2.131912 5.33879 mean(diff)=mean(w_expdir-w_rest) t=4.7083 H0:mean(diff)=0 df=40 Ha:mean(diff)>0 Pr(T>t)=0.0000 Note: w_expdir = wage rate in export activities, w_rest = wage rate in rest of the economy. 28 Appendix 8: Difference in average wage rate, export activities vs. rest of the economy, low- and middle- income countries, 1995, 2005, 2019, t-tests Low- and middle-income countries, 1995 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 34 3.622869 0.563495 3.285711 2.47643 4.769308 Wage rate, rest of the economy 34 3.93123 0.584063 3.405645 2.742945 5.119516 diff 34 -0.30836 0.236129 1.376857 0.788769 0.172047 t=- mean(diff)=mean(w_expdir-w_rest) 1.3059 H0:mean(diff)=0 df=33 Ha:mean(diff)>0 Pr(T>t)=0.8997 Low- and middle-income countries, 2005 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 30 4.985667 0.6426531 3.519956 3.671294 6.30004 Wage rate, rest of the economy 30 5.356375 0.7274037 3.984154 3.868667 6.844082 Diff 30 -0.3707078 0.1558118 0.853416 0.6893787 0.0520369 t=- mean(diff)=mean(w_expdir-w_rest) 2.3792 H0:mean(diff)=0 df=29 Ha:mean(diff)>0 Pr(T>t)=0.9879 Low- and middle-income countries, 2019 Variable Obs Mean Std. err. Std. dev. 95% conf. interval Wage rate, export activities 20 6.032277 0.6547997 2.928353 4.661765 7.402788 Wage rate, rest of the economy 20 6.809661 0.8011414 3.582813 5.132853 8.486469 diff 20 -0.777384 0.4084061 1.826448 1.632188 0.0774198 t=- mean(diff)=mean(w_expdir-w_rest) 1.9035 H0:mean(diff)=0 df=19 Ha:mean(diff)>0 Pr(T>t)=0.9639 Note: w_expdir = wage rate in export activities, w_rest = wage rate in rest of the economy. 29 Appendix 9: Export-labor productivity and export-wage rate elasticities in country-sectors, 1995-2019, overall and by income level (1) (2) (3) (4) (5) (6) (7) (8) (9) Labor productivity in Labor productivity of export Labor productivity of non-export sector activities activities All HICs LMICs All HICs LMICs All HICs LMICs exports 0.095*** 0.081*** 0.109*** 0.136*** 0.127*** 0.149*** 0.093*** 0.074*** 0.109*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant 3.103*** 3.932*** 2.465*** 2.783*** 3.557*** 2.174*** 3.116*** 3.978*** 2.464*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 61,456 26,574 34,882 60,010 25,929 34,081 61,281 26,472 34,809 R-squared 0.963 0.958 0.945 0.964 0.958 0.948 0.962 0.959 0.946 Country-sector YES YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES YES Note: *** p<0.01, ** p<0.05, * p<0.1. pval in parentheses. (1) (2) (3) (4) (5) (6) (7) (8) (9) Wage rate in sector Wage rate of export activities Wage rate of non-export activities All HICs LMICs All HICs LMICs All HICs LMICs exports 0.060*** 0.044*** 0.071*** 0.090*** 0.077*** 0.101*** 0.059*** 0.040*** 0.072*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant 2.428*** 3.413*** 1.698*** 2.199*** 3.152*** 1.479*** 2.429*** 3.440*** 1.691*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 61,287 26,412 34,875 59,856 25,791 34,065 61,115 26,312 34,803 R-squared 0.963 0.946 0.933 0.965 0.946 0.935 0.963 0.946 0.932 Country-sector YES YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES YES Note: *** p<0.01, ** p<0.05, * p<0.1. pval in parentheses. 30 Appendix 10: Export-labor productivity and export-wage rate elasticities in country-sectors, 1995-2019, by GVC taxonomy group (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Labor productivity in sector Labor productivity in export activities Labor productivity in non-export activities commodity limited advanced innovation commodity limited advanced innovation commodity limited advanced innovation exports 0.016*** 0.135*** 0.113*** 0.062*** 0.045*** 0.198*** 0.132*** 0.101*** 0.017*** 0.128*** 0.111*** 0.060*** (0.009) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.008) (0.000) (0.000) (0.000) Constant 3.607*** 2.352*** 2.676*** 4.090*** 3.390*** 1.924*** 2.501*** 3.749*** 3.604*** 2.383*** 2.687*** 4.105*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 11,014 12,775 17,758 18,720 10,624 12,414 17,418 18,358 11,013 12,728 17,697 18,654 R-squared 0.972 0.957 0.952 0.965 0.974 0.959 0.954 0.965 0.972 0.957 0.953 0.964 Country-sector FE YES YES YES YES YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES YES YES YES YES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Wage rate in sector Wage rate in export activities Wage rate in non-export activities commodity limited advanced innovation commodity limited advanced innovation commodity limited advanced innovation exports -0.000 0.078*** 0.079*** 0.026*** 0.024*** 0.125*** 0.092*** 0.055*** 0.000 0.073*** 0.078*** 0.027*** (0.994) (0.000) (0.000) (0.000) (0.006) (0.000) (0.000) (0.000) (0.952) (0.000) (0.000) (0.000) Constant 2.603*** 1.799*** 1.937*** 3.588*** 2.441*** 1.483*** 1.821*** 3.349*** 2.600*** 1.821*** 1.942*** 3.575*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 11,013 12,771 17,756 18,564 10,623 12,411 17,407 18,224 11,012 12,725 17,695 18,501 R-squared 0.965 0.954 0.954 0.954 0.966 0.954 0.956 0.954 0.965 0.952 0.952 0.954 Country-sector FE YES YES YES YES YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES YES YES YES YES Note: *** p<0.01, ** p<0.05, * p<0.1. pval in parentheses. Commodity = commodity exporters, limited = limited manufacturing, advanced = advanced manufacturing and services, and innovation = innovative activities (see World Bank 2020). 31 Appendix 11: Export-labor productivity elasticities in country-sectors, 1995-2019, by broad sector (1) (2) (3) (4) (5) (6) (7) (8) Labor productivity in sector VARIABLES agric mining manufacturing mfg_medhigh mfg_med mfg_low bserv othserv exports 0.141*** 0.019 0.143*** 0.143*** 0.108*** 0.162*** 0.122*** 0.020*** (0.000) (0.183) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant 1.940*** 4.589*** 2.692*** 2.864*** 2.759*** 2.477*** 2.933*** 3.520*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 2,954 3,449 23,903 9,820 5,664 8,419 16,189 14,835 R-squared 0.981 0.964 0.967 0.960 0.973 0.977 0.959 0.983 Country-sector FE YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES (1) (2) (3) (4) (5) (6) (7) (8) Labor productivity in export activities VARIABLES agric mining manufacturing mfg_medhigh mfg_med mfg_low bserv othserv exports 0.188*** 0.123*** 0.166*** 0.178*** 0.117*** 0.184*** 0.140*** 0.051*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant 1.581*** 3.842*** 2.491*** 2.576*** 2.688*** 2.276*** 2.800*** 3.309*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 2,732 2,859 23,599 9,634 5,636 8,329 16,162 14,323 R-squared 0.987 0.973 0.967 0.961 0.975 0.976 0.960 0.982 Country-sector FE YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES (1) (2) (3) (4) (5) (6) (7) (8) Labor productivity in non-export activities VARIABLES agric mining manufacturing mfg_medhigh mfg_med mfg_low bserv othserv exports 0.136*** 0.015 0.140*** 0.135*** 0.114*** 0.160*** 0.122*** 0.020*** (0.000) (0.291) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant 1.962*** 4.589*** 2.716*** 2.923*** 2.715*** 2.496*** 2.924*** 3.520*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 2,938 3,386 23,896 9,815 5,664 8,417 16,091 14,835 R-squared 0.980 0.964 0.966 0.959 0.972 0.977 0.960 0.983 Country-sector FE YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES Note: *** p<0.01, ** p<0.05, * p<0.1. pval in parentheses. agric = agriculture, mfg_medhig = medium-to-high-tech manufacturing, mfg_med = medium-tech manufacturing, mfg_low = low-tech manufacturing, bserv = business services, otherserv = other services 32 Appendix 12: Export-wage rate elasticities in country-sectors, 1995-2019, by broad sector (1) (2) (3) (4) (5) (6) (7) (8) Wage rate in sector VARIABLES agric mining manufacturing mfg_medhigh mfg_med mfg_low bserv othserv exports 0.113*** -0.003 0.079*** 0.058*** 0.087*** 0.121*** 0.076*** 0.006 (0.000) (0.805) (0.000) (0.000) (0.000) (0.000) (0.000) (0.162) Constant 0.754*** 3.469*** 2.294*** 2.635*** 2.113*** 1.853*** 2.423*** 2.691*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 2,954 3,369 23,897 9,820 5,664 8,413 16,106 14,835 R-squared 0.988 0.961 0.969 0.968 0.980 0.971 0.968 0.973 Country-sector FE YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES (1) (2) (3) (4) (5) (6) (7) (8) Wage rate in export activities VARIABLES agric mining manufacturing mfg_medhigh mfg_med mfg_low bserv othserv exports 0.166*** 0.090*** 0.088*** 0.092*** 0.081*** 0.095*** 0.091*** 0.027*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Constant 0.387*** 2.808*** 2.216*** 2.387*** 2.156*** 2.018*** 2.312*** 2.545*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 2,716 2,806 23,593 9,634 5,636 8,323 16,079 14,323 R-squared 0.989 0.970 0.969 0.968 0.981 0.971 0.969 0.972 Country-sector FE YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES (1) (2) (3) (4) (5) (6) (7) (8) Wage rate in non-export activities VARIABLES agric mining manufacturing mfg_medhigh mfg_med mfg_low bserv othserv exports 0.104*** -0.003 0.077*** 0.053*** 0.094*** 0.117*** 0.078*** 0.006 (0.000) (0.835) (0.000) (0.000) (0.000) (0.000) (0.000) (0.157) Constant 0.805*** 3.440*** 2.315*** 2.678*** 2.063*** 1.888*** 2.405*** 2.690*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Observations 2,938 3,308 23,890 9,815 5,664 8,411 16,009 14,835 R-squared 0.988 0.961 0.967 0.965 0.980 0.971 0.967 0.973 Country-sector FE YES YES YES YES YES YES YES YES Country-year FE YES YES YES YES YES YES YES YES Sector-year FE YES YES YES YES YES YES YES YES Note: *** p<0.01, ** p<0.05, * p<0.1. pval in parentheses. 33 Appendix 13: Decomposition of labor productivity and wage rate growth, 1995-2019, within and between two activities, by HICs Labor productivity growth Wage rate growth within within between between within within between between iso3 export rest export rest export rest export rest AUS 31.4 70.3 -1.6 -0.2 16.1 83.7 0.2 0.0 AUT 29.3 69.6 0.8 0.3 29.8 67.4 2.1 0.6 BEL 31.2 67.8 0.8 0.2 16.4 83.0 0.5 0.2 CAN 18.9 82.2 -1.0 -0.2 16.6 84.2 -0.6 -0.1 CHE 38.4 54.5 5.6 1.5 33.4 60.4 4.9 1.3 CYP 66.3 39.0 -4.1 -1.2 59.7 41.2 -0.9 0.0 DEU 23.0 71.7 4.5 0.8 22.0 68.8 7.8 1.4 DNK 30.5 67.5 1.6 0.4 24.0 75.4 0.5 0.1 ESP 15.9 85.1 -0.9 -0.1 17.7 83.8 -1.3 -0.2 FIN 13.4 87.0 -0.4 -0.1 18.7 81.7 -0.3 -0.1 FRA 10.8 88.9 0.2 0.0 13.3 86.2 0.4 0.1 GBR 17.6 82.0 0.3 0.1 17.6 81.7 0.7 0.1 IRL 61.7 23.2 8.1 7.0 53.7 38.9 3.7 3.7 ISL 14.4 85.6 0.0 0.0 18.8 80.8 0.3 0.1 ISR 21.4 77.8 0.7 0.1 22.3 76.8 0.8 0.1 ITA 23.3 76.8 -0.1 0.0 46.5 51.8 1.5 0.3 JPN 6.3 90.2 3.2 0.2 4.8 87.5 7.2 0.5 KOR 25.6 75.2 -0.7 -0.1 16.8 83.4 -0.2 0.0 LUX 79.3 -15.1 19.8 16.0 28.7 62.3 5.0 3.9 NLD 20.6 77.6 1.4 0.4 24.5 74.6 0.7 0.2 NOR 36.2 73.3 -8.4 -1.1 8.8 92.3 -1.0 -0.1 NZL 19.5 79.1 1.1 0.2 16.7 82.4 0.7 0.1 PRT 17.1 85.5 -2.1 -0.5 16.0 86.5 -2.0 -0.5 SGP 69.1 29.8 0.8 0.4 54.2 45.1 0.5 0.3 SWE 20.3 79.5 0.2 0.0 18.6 81.3 0.1 0.0 USA 8.1 91.9 0.0 0.0 6.9 93.1 0.0 0.0 TOTAL 28.8 69.1 1.1 0.9 23.9 74.4 1.2 0.5 Note: HICs as of 1995. Dark blue = highest values, dark red = lowest values. 34 Appendix 14: Decomposition of labor productivity and wage rate growth, 1995-2019, within and between two activities, by LMICs Labor productivity growth Wage rate growth within within between between within within between between iso3 export rest export rest export rest export rest ARG -10.9 111.3 -0.3 -0.1 -211.4 270.2 37.9 3.3 BGR 25.3 90.3 -10.6 -5.0 16.0 90.7 -4.2 -2.5 BRA 8.5 94.4 -2.7 -0.2 1.9 102.0 -3.5 -0.3 CHL 20.2 80.7 -0.8 -0.1 7.7 91.6 0.6 0.1 CHN 6.0 94.0 0.0 0.0 5.9 94.1 0.0 0.0 COL 31.8 65.6 2.0 0.6 23.7 67.7 7.4 1.2 CRI 21.2 77.9 0.6 0.2 19.4 78.4 1.6 0.6 CZE 24.8 75.7 -0.3 -0.1 25.7 74.4 -0.1 0.0 EGY 5.5 94.9 -0.3 0.0 -3.4 103.8 -0.4 0.0 EST 31.0 68.8 0.1 0.1 31.9 68.1 0.0 0.0 GRC 94.9 56.9 -45.6 -6.3 22.9 80.4 -3.0 -0.4 HRV 22.0 79.6 -1.4 -0.3 18.7 82.2 -0.7 -0.2 HUN 29.9 69.2 0.6 0.4 25.0 73.3 1.3 0.4 IDN 11.1 91.5 -2.3 -0.3 9.1 91.2 -0.2 0.0 IND 10.3 89.2 0.5 0.0 8.2 91.7 0.1 0.0 KAZ 29.8 79.9 -8.9 -0.9 22.5 81.2 -3.5 -0.3 LTU 28.3 70.3 0.9 0.4 24.6 75.0 0.2 0.1 LVA 18.7 81.2 0.1 0.0 18.4 81.4 0.2 0.0 MEX 14.2 80.2 4.8 0.8 4.3 98.0 -1.9 -0.4 MLT 55.6 47.2 -1.8 -1.0 50.9 49.8 -0.5 -0.2 MYS 30.2 68.4 1.0 0.4 29.2 63.0 5.5 2.3 PER 12.9 84.4 2.4 0.3 10.7 90.8 -1.3 -0.2 PHL 28.5 71.6 -0.2 0.1 30.4 69.7 -0.2 0.2 POL 18.1 81.2 0.5 0.1 17.2 82.7 0.1 0.0 ROU 13.1 88.6 -1.3 -0.3 10.5 90.3 -0.6 -0.2 RUS 22.0 78.0 0.0 0.0 11.3 88.7 0.0 0.0 SAU 203.8 10.7 -108.5 -6.0 40.2 71.2 -10.9 -0.5 SVK 26.5 73.9 -0.3 -0.1 27.6 72.2 0.1 0.0 SVN 33.6 66.7 -0.3 0.0 28.5 71.2 0.2 0.1 THA 20.9 77.0 1.7 0.5 18.1 82.2 -0.2 -0.1 TUR 17.8 82.0 0.1 0.0 16.2 83.9 -0.1 0.0 UKR 3.1 92.6 3.8 0.4 -13.8 111.9 2.2 -0.2 VNM 15.2 76.0 7.2 1.6 16.2 77.6 5.1 1.1 ZAF 23.3 76.7 0.0 0.0 58.4 41.0 0.5 0.1 TOTAL 22.5 79.3 -1.5 -0.3 19.2 81.0 -0.2 0.0 Note: LMICs as of 1995. Dark blue = highest values, dark red = lowest values. TOTAL excludes Saudi Arabia (SAU) for labor productivity decompositions and Argentina (ARG) for wage rate decompositions. 35 Appendix 15: Decomposition of productivity and wage rate growth, 1995-2019, within and between 45 sectors, export activities vs. rest of economy, by HICs Labor productivity growth Wage rate growth exports rest of economy exports rest of economy iso3 within between within between within between within between sectors sectors sectors sectors sectors sectors sectors sectors AUS 30.2 69.8 75.1 24.9 38.7 61.3 72.9 27.1 AUT 95.2 4.8 102.3 -2.3 92.1 7.9 93.8 6.2 BEL 101.0 -1.0 108.7 -8.7 135.4 -35.4 113.3 -13.3 CHE 115.1 -15.1 85.3 14.7 114.4 -14.4 102.7 -2.7 CYP 71.2 28.8 79.9 20.1 55.0 45.0 139.6 -39.6 DEU 110.9 -10.9 114.4 -14.4 116.1 -16.1 140.2 -40.2 DNK 90.3 9.7 83.6 16.4 94.8 5.2 100.7 -0.7 ESP 103.4 -3.4 37.2 62.8 96.9 3.1 91.3 8.7 FIN 107.9 -7.9 109.8 -9.8 90.8 9.2 83.4 16.6 FRA 112.7 -12.7 106.3 -6.3 97.4 2.6 97.1 2.9 GBR 111.0 -11.0 80.2 19.8 107.1 -7.1 112.4 -12.4 IRL 141.8 -41.8 139.5 -39.5 190.6 -90.6 6.7 93.3 ISL 138.8 -38.8 86.3 13.7 131.5 -31.5 104.0 -4.0 ISR 35.7 64.3 75.9 24.1 16.3 83.7 104.5 -4.5 ITA 54.0 46.0 104.4 -4.4 0.0 0.0 JPN 134.8 -34.8 103.3 -3.3 179.8 -79.8 123.3 -23.3 KOR 86.2 13.8 90.0 10.0 83.6 16.4 84.6 15.4 LUX 91.0 9.0 34.7 65.3 13.7 86.3 NLD 150.9 -50.9 138.2 -38.2 115.8 -15.8 126.8 -26.8 NOR 54.1 45.9 63.6 36.4 87.8 12.2 NZL 87.2 12.8 63.1 36.9 95.4 4.6 84.8 15.2 PRT 72.1 27.9 56.9 43.1 67.7 32.3 51.3 48.7 SGP 113.4 -13.4 124.0 -24.0 73.7 26.3 111.9 -11.9 SWE 101.9 -1.9 104.0 -4.0 96.8 3.2 96.4 3.6 TOTAL 98.1 1.9 91.7 8.3 95.5 4.5 93.2 6.8 Note: HICs as of 1995. The labor productivity growth decomposition average excludes Norway due to extremely high values for export activities, and Italy and Luxembourg due to extremely high values for the rest of the economy. The wage rate growth decomposition average excludes Italy due to extremely high values for the rest of the economy. Dark blue = highest values, dark red = lowest values. 36 Appendix 16: Decomposition of productivity and wage rate growth, 1995-2019, within and between 45 sectors, export activities vs. rest of economy, by LMICs Labor productivity growth Wage rate growth exports rest of economy exports rest of economy iso3 within between within between within between within between sectors sectors sectors sectors sectors sectors sectors sectors ARG 48.3 51.7 93.4 6.6 58.6 41.4 20.1 79.9 BGR 137.5 -37.5 -34.5 134.5 133.9 -33.9 72.7 27.3 BRA 122.5 -22.5 62.3 37.7 0.0 0.0 59.0 41.0 CHL 66.9 33.1 101.6 -1.6 62.4 37.6 84.7 15.3 CHN 91.6 8.4 82.9 17.1 93.6 6.4 89.7 10.3 COL 53.5 46.5 146.4 -46.4 43.8 56.2 112.9 -12.9 CRI 69.3 30.7 56.7 43.3 73.3 26.7 89.9 10.1 CZE 91.0 9.0 84.9 15.1 94.9 5.1 97.4 2.6 EGY 69.2 30.8 26.2 73.8 120.2 -20.2 79.8 20.2 EST 86.2 13.8 103.9 -3.9 90.2 9.8 97.3 2.7 GRC 10.3 89.7 64.6 35.4 89.7 10.3 HRV 91.6 8.4 70.1 29.9 90.0 10.0 79.5 20.5 HUN 100.0 0.0 92.2 7.8 83.7 16.3 86.9 13.1 IDN -19.2 119.2 50.2 49.8 45.9 54.1 63.9 36.1 IND 73.5 26.5 89.4 10.6 71.8 28.2 76.1 23.9 KAZ 70.2 29.8 75.2 24.8 61.1 38.9 68.3 31.7 LTU 96.6 3.4 90.6 9.4 97.2 2.8 90.7 9.3 LVA 99.8 0.2 92.7 7.3 98.7 1.3 90.7 9.3 MEX 76.9 23.1 142.5 -42.5 51.7 48.3 119.3 -19.3 MYS 98.9 1.1 -12.3 112.3 105.0 -5.0 135.4 -35.4 PER -0.8 100.8 63.3 36.7 39.3 60.7 59.4 40.6 PHL 8.6 91.4 39.8 60.2 19.9 80.1 62.6 37.4 POL 79.3 20.7 76.2 23.8 88.0 12.0 79.9 20.1 ROU 104.4 -4.4 81.2 18.8 108.7 -8.7 71.2 28.8 RUS 107.0 -7.0 99.8 0.2 103.1 -3.1 92.6 7.4 SAU 143.4 -43.4 160.2 -60.2 45.6 54.4 SVK 99.7 0.3 92.7 7.3 98.5 1.5 98.9 1.1 SVN 92.4 7.6 48.7 51.3 97.5 2.5 68.0 32.0 THA 44.8 55.2 48.6 51.4 61.7 38.3 59.2 40.8 TUR 77.9 22.1 50.0 50.0 86.7 13.3 64.0 36.0 UKR 168.1 -68.1 52.0 48.0 28.9 71.1 VNM 89.7 10.3 10.2 89.8 86.5 13.5 18.7 81.3 ZAF -40.4 140.4 164.8 -64.8 TOTAL 79.7 20.3 65.7 34.3 85.7 14.3 76.7 23.3 Note: LMICs as of 1995. The labor productivity growth decomposition average excludes South Africa due to extremely high values for export activities, as well as Greece and Saudi Arabia due to extremely high values for the rest of the economy. The wage rate growth decomposition average excludes Brazil and Ukraine due to extremely high values for export activities, as well as South Africa due to extremely high values for the rest of the economy. Dark blue = highest values, dark red = lowest values. 37 Appendix 17: Decomposition of labor productivity and wage rate growth, 1995-2019, within and between sector-activity pairs, by HICs Labor productivity growth Wage rate growth iso3 within sector-activity between sector-activity within sector-activity between sector-activity AUS 63.1 36.9 67.2 32.8 AUT 99.2 0.8 91.3 8.7 BEL 105.0 -5.0 118.8 -18.8 CHE 92.9 7.1 101.2 -1.2 CYP 77.2 22.8 93.3 6.7 DEU 108.7 -8.7 122.8 -22.8 DNK 84.6 15.4 98.7 1.3 ESP 55.2 44.8 95.5 4.5 FIN 109.4 -9.4 86.4 13.6 FRA 107.6 -7.6 97.1 2.9 GBR 87.5 12.5 110.6 -10.6 IRL 106.3 -6.3 104.5 -4.5 ISL 94.4 5.6 109.1 -9.1 ISR 66.1 33.9 82.4 17.6 ITA 188.4 -88.4 JPN 104.4 -4.4 123.6 -23.6 KOR 89.5 10.5 84.5 15.5 LUX 21.3 78.7 NLD 140.7 -40.7 122.9 -22.9 NOR 27.8 72.2 86.8 13.2 NZL 66.8 33.2 85.9 14.1 PRT 62.5 37.5 56.8 43.2 SGP 115.1 -15.1 90.3 9.7 SWE 103.3 -3.3 96.4 3.6 TOTAL 89.4 10.6 97.3 2.7 Note: HICs as of 1995. The labor productivity growth decomposition average excludes Italy and Luxembourg due to extremely high values. Dark blue = highest values, dark red = lowest values. 38 Appendix 18: Decomposition of labor productivity and wage rate growth, 1995-2019, within and between sector-activity pairs, by LMICs Labor productivity growth Wage rate growth within sector- between sector- within sector- between sector- iso3 activity activity activity activity ARG 123.0 -23.0 30.6 69.4 BGR -13.8 113.8 86.4 13.6 BRA 70.2 29.8 66.8 33.2 CHL 96.5 3.5 82.6 17.4 CHN 83.5 16.5 89.9 10.1 COL 106.7 -6.7 85.3 14.7 CRI 58.9 41.1 83.1 16.9 CZE 87.8 12.2 97.0 3.0 EGY 28.0 72.0 76.2 23.8 EST 97.8 2.2 94.9 5.1 GRC 87.5 12.5 HRV 76.4 23.6 82.4 17.6 HUN 93.3 6.7 85.3 14.7 IDN 45.0 55.0 62.5 37.5 IND 87.4 12.6 75.8 24.2 KAZ 74.0 26.0 66.4 33.6 LTU 91.3 8.7 91.9 8.1 LVA 94.1 5.9 92.0 8.0 MEX 130.3 -30.3 118.3 -18.3 MYS 24.3 75.7 110.3 -10.3 PER 53.0 47.0 58.0 42.0 PHL 32.0 68.0 49.1 50.9 POL 76.8 23.2 81.6 18.4 ROU 85.1 14.9 75.7 24.3 RUS 100.3 -0.3 94.1 5.9 SAU 8.3 91.7 SVK 95.5 4.5 99.0 1.0 SVN 65.9 34.1 78.1 21.9 THA 46.7 53.3 59.7 40.3 TUR 54.7 45.3 67.7 32.3 UKR 69.4 30.6 50.8 49.2 VNM 19.8 80.2 28.1 71.9 ZAF -1.8 101.8 -67.9 167.9 TOTAL 69.4 30.6 71.1 28.9 Note: LMICs as of 1995. The labor productivity growth decomposition average excludes Greece and Saudi Arabia due to extremely high values. Dark blue = highest values, dark red = lowest values. 39