Trade and Employment. An Overview

Transcription

1 Trade and Employment An Overview

2

3 TRADE AND EMPLOYMENT: AN OVERVIEW 3 Contents Executive Summary 5 Introduction 7 Theoretical Considerations on Trade and Employment 8 Trade and Jobs 10 Trade and GDP Growth 16 Trade and Productivity 19 Trade, Job Quality and Distributional Effects 21 Conclusion 25 References 26

4 4 TRADE AND EMPLOYMENT: AN OVERVIEW

5 TRADE AND EMPLOYMENT: AN OVERVIEW 5 Executive Summary The bulk of economic research on the impacts of trade has for a long time neglected aggregate effects on jobs. Mainstream theories of international trade generally assume perfectly flexible labor markets and full employment and hence fail to account for net aggregate employment changes. In contrast, both traditional and modern theories of international trade have much more to say about the reallocation of jobs following trade liberalization. Traditional theories of international trade predict specialization of production according to comparative advantage and reallocation of workers from contracting to expanding sectors following trade liberalization. By changing the relative prices of goods, trade may also affect real wages, always creating winners and losers. Modern trade theories break with the tradition of modeling sectoral gains and losses and focus on firms within industries, predicting that only the most productive firms will expand, while less productive, import-competing firms will be driven out of markets following trade liberalization. As such, labor reallocation may as well occur across firms within the same sector. While conferring an important role of trade for employment, empirical studies often struggle to attribute employment outcomes to trade policies in the long run. Empirical research focusing on the net employment effects of international trade suggest that a substantial number of jobs depend on trade (e.g. one in seven EU jobs depends on EU exports). However, there is no consistent evidence for trade policy induced long term employment changes. Similarly, there is little evidence for persistent trade policy induced sectoral reallocation of workers, even though trade and global integration may contribute significantly to fostering structural change. Research focusing on short-run effects, however, have repeatedly identified labor churning and unemployment as more or less persistent characteristics of trade liberalization. Some evidence suggests that trade policies may induce temporary surges in unemployment as job reallocation takes place (notably within sectors), reflecting the existence of adjustment costs to trade reform. The duration and severity of such transitory unemployment varies greatly across countries, but appears to be concentrated mainly in industries with the highest degree of importcompetition. The latter may also display longer spells of slightly higher post-liberalization unemployment rates. The link between trade flows and domestic employment outcomes becomes stronger when looking at value-added data instead of gross trade data, suggesting that expanding domestic value addition is key to creating employment. Official trade statistics have historically been recorded in gross terms, registering values that do not distinguish between domestic and foreign value addition embedded in the price of a given good. Emerging evidence using value-added data suggests that domestic value addition embodied in exports and imports is a key variable affecting domestic employment outcomes and increasingly so. While the exact implications for trade policy are yet unclear and warrant more research, a simple corollary to these findings would be that trade policies should seek to carve out larger niches in global value chains, creating a supportive framework not only for exports, but also for imports of intermediary products and services. Employment and GDP growth often go hand in hand and trade protectionism can hamper both. The literature on trade and GDP growth offers additional albeit indirect insights on employment outcomes, as there is a close empirical relationship between economic growth and employment across countries and over time. While there are both theoretical and empirical justifications for growth enhancing effects of trade barriers for industrial policy purposes, there are at least three key reasons against erecting these: 1) Trade barriers are prone to political capture and have been used in detrimental ways historically; 2) Trade barriers may constitute beggarthy-neighbor policies that may prove detrimental for global welfare and are hence tightly regulated by international trade law; 3) The emergence and intensification of global value chains has magnified the cost of trade protection and will likely continue to do so in the future. Increased trade flows and trade liberalization are often associated with average productivity gains and higher average incomes that in line with modern trade theory arise at the industry level, but often extend to the aggregate. Ample empirical evidence suggests that trade-induced competitive pressures, access to cheaper and higher quality inputs, foreign direct investment and knowledge spillovers, as well as adjustment to foreign quality valuation are some of the main channels prompting firms to ʻupgrade productivity, through technology adoption or increased hiring of high-skill workers.

6 6 TRADE AND EMPLOYMENT: AN OVERVIEW However, a number of studies document increased wage inequality as a result of trade liberalization. A large strand of literature focusing on the industry level tends to conclude that trade can contribute to greater wage inequality, through the differential wage impact according to skill level. Higher skills are rewarded with increasing skill premia. Nevertheless, less skilled workers seem to gain from trade as well, albeit less proportionally than higher skilled workers. Other studies focusing on the occupation level suggest that import competition can have a significant wage-depressing effect that industry-level studies are unable to pick up. Yet another source of country-wide wage inequality can be due to between-firm differences: Firms that engage in international trade or are foreign owned pay higher wages than domestically-owned and domestically operating firms. At the same time, trade appears to have a positive effect on broader aspects of job quality, notably in developing countries. Fears that international trade exerts competitive pressures that may induce a race-to-the-bottom effect in aspects of job quality such as health and safety at the workplace, freedom of association or job security and benefits is not substantiated by data. To the contrary, the evidence rather suggests that open economies have significantly better working conditions than more closed economies. Exposure to international trade and foreign direct investment tends to prompt firms to improve working conditions even beyond the core labor standards as defined by the International Labor Organization. The impact of trade on the informal economy is unclear. Evidence on the existence or the direction of a link between trade and the informal economy is not systematic, owing to significant data constraints. Trade has a positive effect on female labor participation. A number of studies find evidence for a positive role of international trade in promoting female labor participation, which, however, may also increase their exposure to trade shocks.

7 TRADE AND EMPLOYMENT: AN OVERVIEW 7 Introduction The bulk of economic research evaluating the impact of trade on countries economies has for a long time avoided focusing on employment. This can be in part explained by the lack and/or low quality of data on labor market activities. A more important reason, however, is that the main theories of international trade have until recently used the simplifying assumption of full employment in modeling the effects of trade on a given economy. Levels of employment have been included as exogenous elements of the theory at hand, meaning that these are determined by outside factors such as population growth. As such, these theories have been limited to making predictions over trade policy induced changes in e.g. output, productivity or returns to factors of production and are by construction unable to offer any insights on net aggregate employment changes. They do, however, have implications for worker reallocation across sectors and their returns (e.g. wages). For the purpose of this study, it is hence helpful to look at employment in the broader context of the fundamental identity that links employment to output and labor productivity as follows: output = employment productivity It is important to note that this relationship holds by definition, as productivity is defined as output divided by employment. The identity does not rely on any underlying causal mechanism, but can nevertheless be useful in structuring our analysis around the three key variables present in the equation. For example, any given growth of output can be achieved with high productivity growth and low employment growth, suggesting a low overall employment intensity of growth. Conversely, a high employment intensity of output growth implies low productivity growth, but high employment growth. The structure of this paper will follow these lines, examining the effect of trade on each of these variables in turn, while discussing potential interrelations. We start by sketching a brief overview of the predictions of relevant trade theories, drawing on Jansen & Lee (2007). We then turn to the empirical evidence, reviewing first the literature on the link between trade and jobs, as this relates most directly to the focus of this study. Subsequently, we extend the analysis to the effect of trade on economic growth and productivity respectively, as these may impact employment outcomes indirectly. The last section zooms in on the big picture and explores the interrelations between trade, job quality and distributional effects. or: output growth employment growth + productivity growth

8 8 TRADE AND EMPLOYMENT: AN OVERVIEW Theoretical Considerations on Trade and Employment Traditional models of international trade stress the importance of country specific production capacities and endowments of factors of production like capital, land and labor as determinants of trade patterns. Accordingly, countries are endowed with certain production capacities that confer them with a comparative advantage in the production of certain goods. Opening up to trade hence results in a specialization in an identifiable set of exporting sectors to the detriment of shrinking import-competing sectors. The respective expansion and contraction of output in affected sectors translates into a concomitant reshuffling of labor across sectors, while aggregate labor remains fixed. This is the so-called full employment assumption, which does not allow those models to make predictions concerning net changes in employment. Moreover, the movement of labor is generally modeled to be frictionless, assuming that no adjustment costs are borne by labor in the process of changing jobs, resulting in immediate re-adjustment from contracting to expanding sectors. With regards to wages, however, traditional models do make quite important predictions. As such, the move towards free trade always entails winners and losers. In particular, as trade liberalization effects changes in relative prices, it will cause the real return of a country s relatively abundant factor to rise, while the real return of the country s relatively scarce factor will fall. In other words, if an industrialized country with a high proportion of high-skilled labor enters into a trade agreement with a developing country that has a relatively high proportion of low-skilled workers, real wages of low-skilled workers in the industrialized country are expected to decrease, while they would increase for low-skilled workers in the developing country. These traditional and mainstream models of international trade have seen much variation over the years, with a few extensions that explicitly model unemployment. For example, search frictions have been added that model the time it takes laid off employees and new employers to find their match, leading to temporary unemployment.

9 TRADE AND EMPLOYMENT: AN OVERVIEW 9 Nevertheless, the basic mechanism has remained the same, namely the change in demand for certain workers over others and across sectors. Modern trade theories break with the tradition of thinking in terms of sectoral gains and losses from trade. Departing from the empirical observation that most international trade actually occurs within industries and is carried out by the most productive firms, it lays the focus of analysis on firms and their differences in productivity. As such, only the most productive firms can pay the fixed costs associated with exporting and are able to compete on international markets, regardless of the economic sectors they operate in. Reciprocal trade liberalization entails a lowering of these fixed costs of exporting, while increasing international competition for all firms, resulting in an expansion of the most productive firms and cessation of activity of the least productive ones (which were not exposed to import competition before liberalization). As such, labor reallocation occurs within and not across sectors, as workers move from previously active firms into those benefiting from trade in that same sector. Finally, the literature on offshoring adds another layer of complexity to making predictions on the job effects of trade, by arguing that the extent to which jobs are affected from trade largely depends on the tradability of the services or tasks associated with the job in question. Such tradability goes beyond the distinction between high- vs. low-skilled jobs and is immaterial to economic sectors, as e.g. routine tasks performed by low-skilled workers are in principle as tradable as high-skilled tasks that do not require face-to-face interaction with customers, such as segments in research and development (R&D). These recent developments in the theory of international trade have revolutionized how analysts perceive international trade and complicated the tasks of empirical researchers looking for theoretical guidance to inform their evidence-based analyses. The following sections aim to summarize where the literature stands today, taking account of the challenges researchers face in coming up with their findings. Employments effects and theories of trade Traditional models of international trade assume full employment and are thus incapable of explaining tradeinduced changes in employment levels. They, however, can be useful to predict changes in wages. Trade liberalization is usually expected to increase the real return of a country s relatively abundant factor and reduce the real return of the country s relatively scarce factor. This hypothesis is largely behind the opposition to trade agreements by workers representatives in industrial countries which would, accordingly, expect a reduction of real wages for unskilled labor (usually the relatively scarce factor in industrial countries), in spite of potential long term economic gains for the country as a whole. To which extent this hypothesis holds has been extensively debated but it underscores that there are winners and losers of trade liberalization and trade integration. Newer trade theories break with the tradition of thinking in terms of sectoral gains and losses from trade and focus on productivity differences at firm level. Empirical observations suggest that only the most productive enterprises are able to compete in international markets, regardless of the economic sectors they operate in. The large shares of trade taking place within sectors and industries would indicate that workers are largely reallocating across firms within the same sector, with workers moving to those firms able to benefit the most from trade and away from existing firms unable to keep pace with international competition. In a world of increasingly fragmented production processes, theories based on differences across sectors seem ill-suited to adequately explain the effects of trade on employment. When global value chains dominate production patterns the tradability of the services and tasks associated with a job may be much more relevant to predict how the job is affected by trade than the level of skills it requires or the sector in which the job is located. While advances in trade theories in the last decades have contributed to better understand the relationship between trade and employment, economist are still far from having a comprehensive overview, not to speak about a consensus, of the various channels, mechanisms and ways in which trade can impact employment.

10 10 TRADE AND EMPLOYMENT: AN OVERVIEW Trade and Jobs Given the initial lack of theories that explicitly account for unemployment in trade models, the literature that examines the direct link between trade and employment is quite recent and still rather inconclusive. Moreover, methodological constraints often make generalizations of specific findings difficult. On the one hand, most empirical investigations focus on the manufacturing sector and thus neglect the impact of trade and trade policy in other sectors. As such, the large majority of developing country employment, which is in agriculture and services, is left out of the analysis. On the other hand, a substantial part of this literature focuses on industrialized countries. Not only do these fit better with theoretical models that assume full employment, but they also generally have better data quality and availability. Studies focusing on reported employment and unemployment rates in developing countries are inevitably dealing with systematic measurement errors, as notably developing country economies are characterized by a large degree of informality. Jobs in the informal sector are not accounted for in official statistics, even though they can represent large shares of total employment. Sinha (2011) estimates that, on average, 80 per cent of workers in low-income countries, 40 per cent of those in middle-income countries and 15 per cent of workers in high-income countries are employed in the informal economy. Given these limitations, the respective findings need to be taken with a grain of salt. The long term Over the long run, the potential employment creation from greater trade integration can be significant. Arto et al. (2015) estimate that one in seven EU jobs (31 million) depends on EU exports. However, the explicit role of trade policy in fostering such integration is less clear. For example, access to world markets can boost domestic industries and create new employment opportunities, but better access to supplies of inputs from abroad can be a precondition for such expansion. In general, reviews of the empirical literature such as undertaken by Hoekman and Winters (2005) and Newfarmer and Sztajerowska (2012) conclude that the direct effects of trade reform on aggregate employment are muted. Using a rich dataset including many developing countries and spanning 26 years, McMillan and Verduzco (2011) are unable to detect any significant correlation between tariff changes and changes in industrial employment. Substantial variation in country-specific experiences exist that make the identification of a common effect impossible. The authors cite the example of China and India, who have both effected a reduction in tariffs of about 40 percentage points during the observed period of time. While industrial employment in China has boomed in that same period, industrial employment in India has risen only modestly. Such divergent employment outcomes in countries with similar trade policy changes suggest that other factors appear to be more important for aggregate employment and warrant further research. The evidence on sectoral employment is similarly inconclusive. The bulk of the empirical literature does not find evidence for the prediction of traditional trade theories that trade liberalization induces significant reallocation of workers across sectors Hoekman and Porto (2010). This is of course different from longer-term structural change phenomena, implying that there is a strong long-term association between higher incomes and the reallocation of the share of labor from agriculture to manufacturing and eventually services, to which trade can contribute significantly. Yet, some analysts stress the fact that trade and globalization can contribute to fostering structural change, but not necessarily into the right direction (Rodrik et al., 2014). Trade can support such positive structural change only if it contributes to expanding sectors that have denser linkages with the broader economy, allowing countries to move up the industrial structure with greater ease (Hidalgo et al., 2007). Moreover, Rodrik et al., (2014) point out that developing countries display larger inter- and intra-sectoral productivity gaps, as well as more important and persistent unemployment rates than developed economies. Hence trade in these countries will have a different impact on labor than elsewhere. The authors continue to stress that while most studies document how trade entails productivity upgrades within firms and hence is associated with average industry productivity increases as low productivity firms exit, they rarely investigate what happens to displaced workers. While not investigating the effects of trade and globalization per se, the authors note that some developing countries have benefited from greater global integration while others have not. They note that in poorly performing developing countries import competition has caused many industries to contract and

11 TRADE AND EMPLOYMENT: AN OVERVIEW 11 release labor to less productive activities, such as agriculture and informality. This adverse effect of openness is associated with countries that 1) have a comparative advantage in natural resources, 2) maintain overvalued and uncompetitive exchange rates, and 3) have inflexible labor markets. Such considerations lead some observers to conclude that comparative advantage driven trade may not necessarily translate into the expected theoretical benefits, as it traps countries into producing what they are currently relatively best at, preventing them from upgrading their future industrial base. In this vein, Singh (2011) points out that many Asian countries employed trade-related industrial policies that helped them successfully move into industries in which they clearly did not have a comparative advantage at that time. For example, Japan and then Korea started producing steel (a relatively capitalintensive industry) when their per capita income was only 2.5 per cent that of the United States. The not so long term A general finding of the trade literature is the existence of adjustment costs to trade liberalization, which entail significant losses to some affected groups. Adjustment costs can be broadly evaluated as the value of output that is forgone in the transition to new long-run production patterns, because of the time it takes factors to reallocate from their pre- to their post-liberalization occupations (Francois et al., 2011). The losses incurred to individuals can result from prolonged unemployment, lower pay in new occupations, retraining costs, as well as other personal consequences resulting from job loss. Traditional trade theory would predict adjustment costs to be largest in sectors adversely affected by trade liberalization, notably those in which a country is at a comparative disadvantage relative to the rest of the world. New trade theories, conversely, predict that adjustment costs can also be large within industries, as low-productivity firms are driven out of markets and/or shrink. The magnitude of adjustment Trade, structural transformation and the quality of jobs Trade can support positive structural change only if it contributes to expanding sectors that have denser linkages with the broader economy and allow countries to move up the industrial structure. In poorly performing developing countries, import competition has caused many industries to contract and release labor to less productive activities, such as agriculture and informality. This adverse effect of openness is associated with countries that 1) have a comparative advantage in natural resources, 2) maintain overvalued and uncompetitive exchange rates, and 3) have inflexible labor markets. Comparative advantage driven trade may risk trapping countries into producing what they are currently relatively best at, preventing them from upgrading their future industrial base. When the positive employment effects from trade are limited to sectors or tasks with shallow economic linkages and few possibilities to expand the country s skills and knowledge base, the surge in jobs is likely to be short-lived. Thus, for structural transformation the quality of jobs is as at least as important as the number of jobs generated by trade integration. Quality not only in terms of labor and social standards but above all in terms of the potential associated to those jobs for skills upgrading and to generate knowledge spillovers. The success of the exportled strategies of various Asian countries in the 20th century was less the result of expanding export volumes than the result of their ability to steadily diversify and upgrade their export-base through the accumulation of skills and knowledge by their labor force. Furthermore, the recent availability of data disaggregating the domestic and foreign value components of exports suggests that trade policies which encourage the progressive development of industries with high domestic labor content both in exports and imports may contribute to increasing countrywide employment levels over long periods of time. For instance, three emerging economies Brazil, China and India, managed to create more domestic jobs through trade than most other countries for the period. Policies oriented to foster employment linked to GVCs in activities with high potential for skills and technological upgrading may have played an important role.

12 12 TRADE AND EMPLOYMENT: AN OVERVIEW costs in turn depends on a large number of factors that are external to the mere trade liberalization shock. Examples are the restrictiveness of a country s administrative regulations or externally imposed labor market contract requirements that may prevent firms in expanding sectors to adjust quickly. Access to credit is important, as long as new investments are needed to seize novel trade-induced opportunities. Workers skill sets are another determining factor in that they may allow workers to find new placements more rapidly. Similarly, firms in declining sectors may be faced with stringent labor laws, or persistent government support for industrial policy purposes that offer disincentives to perform the required adjustment smoothly and swiftly, unnecessarily prolonging the inevitable and making adjustment more costly. Hence, there is certain trade-off between the need to protect workers from adverse shocks on the one hand and the risk of hampering structural transformation processes on the other. Striking a balance between these two considerations is key and may provide for formidable political challenges in practice. Regarding the actual magnitude of adjustment costs, most studies find that these are only a small fraction of total benefits from trade, reinforcing the perception that trade liberalization generally benefits countries. Bradford et al. (2005) point out that the US federal government spends less than $2 billion annually on explicit trade adjustment assistance less than 1 percent of prospective annual benefits from complete free trade. Of course adjustment costs are not negligible for those individuals affected, but among economists the predominant view is that the permanent gains from liberalization are so enormous that countries can easily afford the modest sums necessary to alleviate the temporary pains of adjustment Bradford et al. (2005). Many of the factors that raise adjustment costs are much more widespread in developing countries, suggesting that the magnitude of adjustment costs may be higher there than in industrialized countries, but there is no robust empirical backing for such a conjecture. A number of studies reviewed in Newfarmer and Sztajerowska (2012), Francois et al. (2011) and Hoekman and Porto (2010) suggest that there is surprisingly little evidence for transitional aggregate unemployment surges following trade liberalization, but the evidence is mixed. Dutt, et al. (2009), for example, find robust cross- country evidence that liberalization periods are followed by immediate rises in unemployment in the short run, but a reversal of that rise and an eventual decline in unemployment in the long run. Moreover, large-scale job displacement may take place when trade policy changes are implemented during periods of broader economic stress, such as recessions or major structural change. The predictions of new new trade theory, namely increased labor churning within industries and sometimes even within firms is a more regular feature of the empirical evidence (Hoekman and Porto (2010). Francois et al. (2011) and Newfarmer and Sztajerowska (2012) review a number of studies that suggest that unemployment, as a small share of the total labor force, may be a more or less persistent post-liberalization characteristic notably within affected industries. Trefler (2004) analyzes the impact of tariff cuts within the US-Canada Free Trade Agreement of 1989 on Canadian employment and find evidence for substantial job dislocation within affected sectors. Accordingly, the agreement was associated with 12 percent employment losses in the most impacted, import-competing industries and 5 percent for manufacturing as a whole, translating into jobs that were lost. Meanwhile, the evidence cited in this paper also suggests that these job losses were of temporary nature, as the overall share of manufacturing jobs in Canada remained constant at 62 percent in 1988 and Similarly, studying the impact of offshoring on employment in the US manufacturing sector, Ebenstein et al. (2014) find evidence for substantial crowding out of US employment through employment creation abroad. A one percentage point increase in import competition is associated with a 0.6 percentage point decrease in manufacturing employment in the United States, which disproportionally affects workers with a high school diploma or less. Notably for developing countries, a few studies suggest that labor that is shed in the context of adjustment to trade liberalization can exceed the rate of new hiring within firms. Menezes-Filho & Muendler (2011) obtain a rich dataset on Brazil, which allows them to follow workers through a liberalization period and to observe the paths of their employment histories. They show that the liberalization episode is associated with a lower likelihood of a worker moving from one formal-sector job to another, suggesting labor market congestion. Nevertheless, a reduction of tariffs in intermediate goods is associated with employment gains. A study by McMillan et al. (2003) suggests that overall efficiency gains can in some cases be entirely canceled out by resulting unemployment. Studying a trade liberalization

13 TRADE AND EMPLOYMENT: AN OVERVIEW 13 event in Mozambique, they find that the resulting closing of plants led to large numbers of unemployed workers, of whom 90% were still unemployed more than ten years later. This timeframe largely exceeds any reasonable definition of temporary adjustment costs and may reinforce the conjecture that adjustment costs can be of higher magnitude in developing countries. The authors conclude that in order to predict liberalization outcomes, it is key to understand the initial conditions a country is in, making sense of the linkages that exist between the liberalized sector and other sectors of the economy. Another strand of the literature looks at whether tradedisplaced workers differ fundamentally from workers displaced by non-trade shocks such as during economic recessions, broader structural adjustment programs, or technological change. Francois et al. (2011) review a few of those studies and conclude that both groups appear to be similar in terms of education and work experience, even though trade displaced unemployed workers appear to be slightly older, have more tenure and slightly higher earnings in the former job. This point is important to bear in mind from a policymaker s point of view, as it may argue in favor of non-discrimination in designing assistance programs. In sum, the nexus between trade and employment is unclear. Industry-level studies may offer clues about how trade policy affects employment in certain industries, but usually cannot say much about net job creation effects in the aggregate. For policy makers eager to design trade policy so as to offer domestic employment opportunities, the issue has become even more complicated due to the emergence of global value chains (GVCs). Concentrating export promotion efforts in a certain sector does not necessarily translate in a commensurate rise in domestic employment opportunities. While a lot more research needs to be done in this regard, recent evidence suggests that trade has in most countries generated more foreign than domestic jobs (Jiang, 2013). Traditionally, researchers have simply focused on the domestic labor content of exports, as well as the foreign labor contained in imports. However, such a simplistic analysis masks the fact that the global slicing of production chains implies that exports and imports contain respectively foreign and domestic labor as well (through imported and exported intermediates), as well as third-country labor. Making use of this information provided by the World Input-Output Database, Jiang finds that notably the three emerging economies Brazil, China and India have achieved to create more domestic jobs through trade than most other countries for the period Hence, trade policies that encourage the development of industries with high domestic labor content both in exports and imports may contribute to increasing countrywide employment levels. These findings on domestic labor notwithstanding, the income derived from domestic value added embodied in foreign final demand (that is, exports of domestic value added ), has increased by 106% between 1995 and 2009 in real terms (OECD et al, 2013). Again, the three emerging economies of Brazil, China and India have reaped spectacular benefits from inclusion in GVCs; their resulting income increased 3-fold, 5-fold and 6-fold respectively. These income gains have also translated into increased domestic employment for most countries, suggesting that trade and high domestic value added content in particular is becoming ever more important for the creation of employment opportunities at home (see figure 1). Figure 1: Jobs sustained by foreign final demand, as a % of total employment 30% 25% 20% 15% 10% 5% 0% AUS BRA CAN CHN DEU FRA GBR IND ITA JPN KOR MEX RUS TUR USA Source: OECD et al. (2013)

14 14 TRADE AND EMPLOYMENT: AN OVERVIEW Increased participation in GVCs hence makes the link between gross trade and employment weaker, as the domestic value added content of trade is the key variable determining domestic employment creation. In fact, UNCTAD (2013a) confirms that domestic employment is more responsive to changes in value added exports than in gross exports. Similarly, the employment elasticity (the percentage change in employment in reaction to a one percent change in exports) is different across sectors, indicating differential sensitivity of domestic employment to trade across sectors. In particular, a one percent increase in value added exports in manufacturing is associated with a 0.42 percent increase in manufacturing employment, whereas for services this elasticity is only 0.12 (see figure 2). The relatively low elasticity for services stems from the fact that the share of services trade in services output is lower than in manufacturing. UNCTAD (2013) point out that if one were to correct for this asymmetry, the responsiveness of employment in the services sector to services exports may be even higher than in manufacturing. However, it is important to note the interconnection between broad economic sectors. Associating manufacturing exports with manufacturing jobs alone does not do justice to the interwoven intricacies of a modern economy. For example, Arto et al. (2015) estimates that most jobs associated with exports of the EU are actually to be found in the service sector. For the whole EU, almost 60 percent of jobs that depend on EU exports (31 million) are actually located in the services sector. The trend towards using internationally comparable input-output data for trade analysis hence promises to yield new and highly policy-relevant findings in the near future that may run counter to long-held tenets about international trade. Its role for domestic economic outcomes is likely to gain importance, calling for policy to play an active role in shaping a supporting environment, rather than obstructing it. Figure 2: Employment and exports correlation for manufacturing and services Manufacturing Services e( logemp X ) 0-1 e( logemp X ) e( logvaexp X ) e( logvaexp X ) coef = , se = , t = coef = , se = , t = 8.68 Source: UNCTAD (2013)

15 TRADE AND EMPLOYMENT: AN OVERVIEW 15 Offshoring The rise of GVCs has been accompanied by the phenomenon of offshoring, which comprises the physical relocation of activities overseas through the establishment of overseas affiliates, and offshore outsourcing, where inputs are sourced from foreign third-party suppliers. Existing studies do sometimes provide evidence that offshoring can increase productivity, but the effects are heterogeneous and depend on the country under study, the type of firms (ownership structure, exporting status) and the type of offshoring activity (services vs. material). In a pioneering study, Amiti and Wei (2009) combine input-output information with trade data to measure service and material imports in the US and estimate a production function at the industry level. They find that service offshoring accounts for 10% of labor productivity growth, whereas the effect of material offshoring is less substantial and not robust to different specifications of the estimation equation. The pioneering theoretical model explaining the mechanics of offshoring by Feenstra and Hanson (1995) predicts that material offshoring increases the relative demand for skilled labor in both countries involved. This is because offshoring usually occurs from rich to poorer countries and involves the relocation of activities that are relatively less skilled in rich countries, but relatively high skilled in comparison to other tasks performed in poor countries. The implication for resulting wage inequality is that it is supposed to rise in both countries, a prediction that has since received widespread empirical support. Feenstra and Hanson (1997) find such evidence for increased wage inequality in Mexico as the receiving country, while Feenstra and Hanson (1999) document this effect for the US, as the home country. Nevertheless, the latter study also finds that real wages of production workers in the US were probably not affected, while real wages of non-production workers have increased as a result of offshoring. The employment effects of offshoring appear to be less clear. Setupathy (2013) finds that operating profits and domestic wages actually rise at US firms likely to take advantage of new offshoring opportunities in Mexico, while there is no indication of greater job loss than elsewhere. Ebenstein et al. (2014) find that US offshoring to high-wage countries is positively correlated with US manufacturing employment, whereas offshoring to low-wage countries is associated with declines in US employment. And while wages for those who remain in manufacturing are generally positively affected, the authors find that workers driven out of the manufacturing sector incur wage losses, which are magnified for those workers who in addition have to change occupations. Hummels et al. (2011) also focus on displaced workers from offshoring and find that those workers generally experience greater and more persistent wage and earnings losses than workers displaced for other reasons, with the burden of adjustment falling disproportionally on low-skilled workers. Services offshoring is a relatively new phenomenon and has been made possible through spectacular developments in information technology (IT) and transport. Unsurprisingly, very little is known to date on how services offshoring interacts with employment outcomes. In a groundbreaking study Jensen and Kletzer (2005) estimate that around 40% of US jobs are potentially affected by services trade. Workers in tradable service activities (like business services) are, on average, substantially more educated than those in non-tradable activities (like retail trade) and other sectors, like manufacturing. The findings on the effects of services offshoring on labor in high-income countries are generally mixed with some studies echoing the findings on material offshoring (see Newfarmer and Sztajerowska (2012)So far, however, no study has found significantly negative effects of services offshoring on high-income countries, as some commentators might have feared (Blinder 2006). For developing countries, services trade is widely seen as an opportunity. Ghani (2010) notes that as the services produced and traded across the world expand with globalization, the possibilities for all countries to develop based on their comparative advantage expand. A study by the International Labor Organization on offshoring and working conditions in Argentina, Brazil, India and the Philippines concludes that overall, jobs in the BPO [business process outsourcing] industry in developing countries are of reasonably good quality by local standards in terms of their working and employment conditions (regarding wages, hours of work and non-wage benefits, for example) Gosheh and Messenger (2010). Gonzales et al. (2012) point out that a number of impediments to trade in services persist notably in developing countries, which may prevent them from taking better advantage from integration into global services trade.

16 16 TRADE AND EMPLOYMENT: AN OVERVIEW Trade and GDP Growth While highly controversial, the widely used metric of choice in evaluating countries economic growth and potential for development more generally is the growth rate of gross domestic product (GDP). The use of GDP as a relevant metric in assessing labor market outcomes may be justified to the extent to which there seems to be a close empirical negative relationship between economic growth and unemployment across countries and over time (Ball et al., 2012) This empirical relationship has been first articulated by Arthur Okun in the early 1960s and has been included in a list of core ideas that is widely accepted by the economics profession (Blinder, 1997). Assuming the validity of Okun s law for the time being and letting trade influence employment through the channel of broader economic growth, it makes sense to briefly review the rather extensive literature on trade, trade policies and GDP growth, drawing on Schwarzer (2013) and Newfarmer and Sztajerowska (2012). Traditional theories of international trade make a very clear case for free trade, stipulating that any barriers to trade will only hamper efficient allocation of economic activity and hence reduce welfare. At least since the seminal work of Krugman (1991), the neoclassical view of international trade has changed. Krugman introduced the concept of increasing returns to scale essentially meaning that economic activity becomes more efficient with scale into international trade theory, giving a formal underpinning for the infant industry argument, which states that certain industries can become viable only after reaching a certain size. Hence, such infant industries need to be protected in their early stages and trade policy is one means to do so. As long as there is no retaliation, trade restrictions such as tariffs and quotas can theoretically increase output from the restriction-imposing country if the protected infant industry offers scope for learning and agglomeration spillovers (Melitz, 2005). It is important to bear in mind, however, that such improvements in welfare will accrue only to the tariff imposing country and will generally reduce overall welfare globally, which maintains the argument in favor of free trade on a global basis. Clemens and Williamson (2001) and O Rourke (2000) find a strong positive correlation between import tariffs and economic growth across countries during the late 19th century, hypothesizing that protection allowed countries to accelerate the growth of what were then emerging sectors (industry), benefiting from synergies such as knowledge spillovers among firms, labor pooling, or input-output linkages where upstream firms provide inputs for downstream firms, taking advantage of low transportation costs. However, this type of association could not be upheld for more recent times. Reviewing nearly 200 empirical studies on the link between openness to trade and growth, Harrison and Rodríguez-Clare (2009) conclude that studies using trade volumes as a measure of openness (measured as the sum of imports and exports over GDP) generally find a positive relationship between changes in openness and growth, making a strong case for overall trade-friendly policies. Studies that use tariffs as a measure of openness generally find inconclusive or negative effects of average tariffs on growth. Newfarmer and Sztajerowska (2012) assert that no study finds that trade restrictions positively affect long-run growth, a point conceded by even the most sophisticated critics (p.10). However, looking at tariff averages might not be very revealing altogether. Lehmann and O Rourke (2008) look at the pattern of protection during the 19th century and refine earlier findings, noting that agricultural tariffs were negatively correlated with growth, whereas the reverse was true for industrial tariffs. More recently, Nunn and Trefler (2010) found strong evidence that high correlation of tariff structure across industries with those industries respective required skill levels had long-term growth-enhancing effects. That is, if a country s tariff lines are designed such that higher tariffs coincide with goods that necessitate greater skill in their production, long-term GDP growth rates will be higher. They find that at least 25% of the correlation between long-term growth and the skill bias in tariffs corresponds to a causal effect. Decomposing their analysis into high-skill and low-skill sectors, their analysis reveals the differential impact of tariffs on each sector: while tariffs in high-skill sectors are associated with sector growth, tariffs in low-skill sectors are associated with negative sectoral growth, somewhat echoing Lehmann and O Rourke s (2008) findings with 19th century data. Overall this analysis suggests that careful sectoral targeting (skill-intensive sectors) can have significant growthenhancing effects beyond the targeted sector.

17 TRADE AND EMPLOYMENT: AN OVERVIEW 17 Similarly, Estevadeordal and Taylor (2008) find that tariffs on intermediate and capital goods affect growth more negatively than tariffs on consumption goods. This finding is consistent with mainstream economic theory, stressing that firms access to cheaper imported inputs is vital for keeping production costs low, and increasing their competitiveness on international markets. Blonigen (2013) made a similar finding more recently, analyzing industrial policies from 1975 to 2000 in the steel sectors of 21 countries. This study finds strong evidence that import protection policies and export subsidies have had significantly negative effects on the export performance of downstream sectors. Both are tools that will raise domestic prices for steel, raising production costs for domestic users in other sectors. Hence, the fact that studies looking at average tariffs find mostly weak or even negative associations with overall long term growth implies that countries have used tariffs in very inefficient, even detrimental ways. Indeed, existing evidence suggests that protection is usually motivated by revenue generation, to protect special interests or for other political reasons, rather than for sound economic development purposes (Gawande, Krishna & Olarreaga, 2009). The highest tariff protections, as well as other targeted forms of support such as capital subsidies, have frequently been granted to declining sectors, even in countries that are often considered as industrial policy success stories, including Japan (Beason & Weinstein, 1996) and Korea (Lee, 1996). Besides the demonstrated empirical regularity that openness is associated with economic growth, there is still substantial disagreement on which way causality runs. Countries whose incomes are high for reasons not related to trade, may well be characterized by a high trade share. We have seen how the evidence on the nexus between trade policies as a measure of openness and growth is similarly inconclusive. Moreover, trade policy changes do not necessarily translate into concomitant changes in trade flows, as evidenced by an often disappointing supply response to trade policy reform in many developing countries. Nevertheless, a number of economists tend to converge towards a positive net effect of openness to trade (see e.g. Winters and Martuscelli (2014) for an overview). A noteworthy paper by Brückner & Lederman (2012) employs an innovative methodology allowing them to conclude that trade openness has spurred economic

18 18 TRADE AND EMPLOYMENT: AN OVERVIEW growth in Sub-Saharan Africa, reinforcing the view that developing countries also benefit from greater openness. In sum, while there appears to be scope for using trade protectionism as an instrument for long-term economic growth in the tariff-imposing country, governments generally seem to have had great difficulty applying it optimally. Furthermore, governments today are bound by a much tighter set of trade rules as embodied in the World Trade Organization (WTO), significantly restricting the scope of their discretion in setting up trade barriers and making the possibility of retaliation very likely in practice. These developments and the detrimental effect of trade protection for global welfare currently rather make the case for trade-friendly policies. Finally, and perhaps most importantly, trade has undergone significant structural changes over the last decades. GVCs have become a dominant feature of world trade throughout the globe, implying increased international fragmentation of production. The implications for trade policy are not yet clear and research on the topic is currently in full swing. Preliminary results suggest that countries increasingly rely on foreign inputs for their own firms exports, which may in turn be further processed abroad. Moreover, services play a much greater role in international trade than implied by gross trade statistics. While trade in services, as conventionally measured, makes up roughly 20% of global trade flows, this number more than doubles when measuring flows in terms of trade in value added. For the link between trade policies and growth this implies yet another layer of complexity, as trade policies become more interdependent and have more immediate and pervasive effects. The cost of protection increases since, for example, tariffs are cumulative when intermediate inputs cross borders multiple times and the final product is taxed as well. The reason is that nominal duties are calculated on the basis of gross value (see figure 3). Figure 3: Tariffs on the gross value and the domestic value-added of exports, 2009 Manufacturing Agriculture 35% 35% 30% 30% 25% 25% 20% 20% 15% 15% 10% 10% 5% 5% 0% 0% United States Chile UK Japan Germany EU France Netherlands Vietnam China Chile Japan Vietnam UK Netherlands EU Germany France United States China Tariffs on the domestic VA of exports Tariffs on gross exports Source: OECD et al. (2013)

19 TRADE AND EMPLOYMENT: AN OVERVIEW 19 Trade and Productivity Productivity is a crucial element of competitiveness and economic viability, as it reflects efficient production and low production costs and hence the ability to sustainably provide employment opportunities. At the same time, it is an important element in the more imminent tradegrowth-employment nexus described in the introduction. The basic relationship linking GDP and employment growth hinges critically on the growth rate of productivity. Productivity growth can be the result of a higher growth rate of output compared to employment, even if the latter is still positive. However, productivity growth can also be the result of employment losses, resulting in a labor-saving effect. In economic terms, such productivity growth still indicates greater competitiveness and economic viability, albeit at the expense of labor in this case. The evidence on the prevailing developments in employment that are at the heart of productivity growth is mixed and appears to depend on the level and geographical focus of analysis. For example, Dew-Becker and Gordon (2005) find a robust negative correlation between growth in labor productivity and growth in employment per capita across Europe. On the other hand, looking at historical data of the US manufacturing sector and using econometric techniques to identify causality, Nordhaus (2005) finds that rapid productivity growth has actually led to increased rather than decreased employment. The job intensity of growth is certainly a major topic for policymakers and should therefore take a central role in policy agendas. However, looking at job creation alone does not do justice to the wider employment topic. A perhaps more important reason that warrants the use of the productivity metric to (indirectly) assess the effect of trade on employment is that productivity represents an indicator for the quality of jobs, as output and income over employment represent two sides of the same coin. Against this background, this chapter will briefly review the main findings on trade and productivity, before the next chapter will look more closely at more direct measures of job quality and distributional impacts of trade. The available evidence on economy-wide productivity effects of trade paints a similar picture as the one on economic growth. Rising trade ratios appear to be broadly correlated with economy-wide increases in productivity in the long run (Newfarmer and Sztajerowska, 2012). Theoretically, productivity plays a crucial role in modern trade theory, as described above. The predictions of modern trade theory find ample empirical backing. The main channels through which trade liberalization affects firms is through competitive pressures, which allows only the most efficient firms to export and drives less efficient firms out of the market. Overall, a concomitant rise in affected average industry productivity can be observed with trade liberalization, sorting out less productive firms and encouraging entry and survival of high-productivity firms, leading to average productivity growth primarily within, but even across industries (see, for example, Pavcnik, 2000, or Tybout, 2003). The number of firms that engage in international trade is typically very small compared with a country s total number of firms. This finding dates back to Bernard, Jensen and Lawrence (1995), who examined U.S. manufacturing firms, but it has subsequently been confirmed for other countries in a number of other studies (see, for example, Eaton, Kortum & Kramarz, 2008, or Mayer & Ottaviano, 2008, for an overview). Moreover, total export value is typically concentrated in a few large, established firms that employ substantially more people, are more productive and pay higher wages than non-exporters, a finding that is valid for both developed and developing countries (von Uexkull, 2012) Competitive pressures are not the only channel through which trade liberalization appears to impact productivity, indicating that potentially employment reducing cost-cutting measures are not the only way to expand output through international trade. Reducing tariffs on input goods can drive learning about foreign technology, expansion in the variety of intermediate inputs available to production, and availability of higher-quality intermediate inputs (Bernard, Jensen, Redding & Schott, 2011). Indeed, lowering tariffs on input goods appears to be associated with large productivity gains in firms (see, for example, Goldberg, Khandelwal, Pavcnik & Topalova, 2010, or Amiti & Konings, 2007). These benefits support the notion that trade can be a vehicle for international Research & Development (R&D) spillovers, in that countries importing intermediate and capital goods can benefit from R&D done in the exporting country (Harrison & Rodríguez- Clare, 2009). Exporters in both developing (Alvarez & López, 2005) and developed (Bernard, Jensen, Redding & Schott, 2007) countries are typically more capital- and skill-intensive

20 20 TRADE AND EMPLOYMENT: AN OVERVIEW than non-exporters. This finding is difficult to square with old trade theory, which stresses comparative advantage as a determinant of trade flows and predicts skill differences among countries, rather than within them. From a policy perspective, it is of course important to know in which way causality runs, does skill and capital intensity foster trade or the other way round? A number of studies offers evidence of trade-induced quality and technology upgrading, often proxied by an increased share of high-skill, or non-production workers, or spending on technology in datasets. Bustos (2007) investigates the reaction of Argentinian firms following trade and capital account liberalization in Argentina and finds a strong relationship between these events and increases in technology spending and skill upgrading in both exporting and foreign-owned firms. Technology upgrading and increases in skilled labor occurred faster for new exporters post-liberalization than for non-exporting firms. Van Biesebroeck (2003) finds similar effects on sub-saharan African manufacturing firms. Brambilla et al. (2012) offer evidence for differentiated skill upgrading in Argentinian firms according to export destination. They argue that consumers in higher income export destination markets display a higher valuation for product quality, which should induce exporters to upgrade technology more intensively, as well as hire a larger share of high-skilled workers than domestic firms and those exporting to countries of similar income levels. Their findings largely confirm their theory, except that exporters to destinations of similar income levels do not differ substantially from firms operating solely domestically. They explain the latter finding by similarities in quality valuation between the Argentinian market and similar low-and middle-income countries that these firms are exporting to. Their work suggests that productivity gains resulting from technology and skill upgrading should be muted in some regional integration schemes, where countries do not vary much in terms of quality valuation. (FDI) (see Schwarzer, 2013). Here, most researchers have found that firms with foreign equity participation display higher output, higher output per worker, or higher levels of total factor productivity (Harrison & Rodríguez-Clare, 2009). The caveat of most such research, however, is to identify a causal effect. While it is possible that FDI increases productivity, it is also likely that sectors or firms with most promising productivity growth prospects attract most FDI. Using econometric techniques that rule out the possibility that foreign firms tend to acquire the most productive domestic enterprises, studies have been able to conclude that, indeed, foreign equity infusion does confer productivity benefits to domestic firms (see, for example, Matthias & Javorcik, 2009). However, there is little evidence for the existence of horizontal spillovers, the extent to which foreign ownership benefits indigenous firms in the same industry. Developing countries often seek to attract FDI in order to facilitate technology transfer to domestic firms operating in the same sector, hoping to encourage the development of indigenous firms. Foreign investors usually lack incentives to actively engage in technology transfer with their own domestic competitors. On the other hand, there is ample empirical backing for the conjecture that FDI might be beneficial to domestic firms through vertical spillovers, that is, positive externalities stemming from the relationship of foreign enterprises with domestic suppliers and customers (Harrison & Rodríguez-Clare, 2009). Evidence on forward spillovers the supply of inputs embodying new technologies or processes is scant. However, the existence of backward spillover effects productivity gains in foreign firms domestic suppliers has been confirmed in various studies for a number of countries (see, for example, Gorodnichenko, Svejnar & Terrell, 2007, and Javorcik, 2004). A few studies also confirm differential productivity growth between exporters and non-exporters, notably in developing countries (see Van Biesebroeck, 2003, and De Loecker, 2007). From a development perspective, the most important conclusion of such work is probably that learning from exporting is most likely in technologically backward countries and among less productive firms (Harrison & Rodríguez-Clare, 2009). Similar evidence emerges from the literature on Foreign Direct Investment

21 TRADE AND EMPLOYMENT: AN OVERVIEW 21 Trade, Job Quality and Distributional Effects Wages and Inequality Basic economics suggests that labor s real compensation should grow in line with its productivity. This relationship generally holds in the long-term, but is far from perfect. In fact, labor compensation in the US lagged productivity gains since 2002 and the share of employees compensation in total output declined since 2000 (Newfarmer and Sztajerowska, 2012). As mentioned before, traditional theories of international trade have much more to say about wages than about aggregate employment. The workhorse prediction is given by the Stolper-Samuelson Theorem, which states that a trade liberalization induced rise of the price of a particular good will lead to a rise in the return to the factor that is used most intensively in the production of that good. For developing countries, this usually meant to imply that the returns to unskilled workers are expected to increase, leading to poverty reduction and increased wage equality within these countries. Global developments notably in the manufacturing sector appear to be a case in point for this: While the manufacturing sector in high income countries has shed labor over recent decades, employment in developing countries mainly in East Asia has expanded significantly, from 54 million in 1980 to over 108 million in 2005 (McMillan and Verduzco, 2011). At the same time, these global manufacturing hubs have witnessed rising wages over recent years, potentially paving the way for other developing countries to take on a greater role in manufacturing (Manyika et al.,2012). Fukase (2013) examines the implications of a bilateral Free Trade Agreement between Vietnam and the USA and finds some support for the predictions of the Stolper-Samuelson Theorem. His study documents how provinces that were more exposed to the increase in export opportunities created by the FTA experienced a larger wage growth for unskilled workers, which was driven by increased demand for those. Nevertheless, they caution that their study did not take other factors into account, which may have played a significant role in wage development. In fact, they observed a steady rise in skill premium in the aggregate sample of individuals (over all regions and skill types), which the Stolper Samuelson type effect merely mitigated, but did not dominate. Most studies, in fact, find little support for Stolper- Samuelson-like predictions and the evidence rather points towards the opposite: Skill premia and wage inequality have risen following trade liberalization in many developing countries, including Mexico, Colombia, Brazil and Argentina (Winters and Martuscelli, 2014). Moreover, we have discussed above how there is little or sometimes even conflicting evidence on labor reallocation from contracting to expanding sectors. Unsurprisingly, some commentators hence declare that Stolper-Samuelson is dead (Davis and Mishra, 2007). Again, modern trade theory offers an alternative look on how trade might have distributional impacts within countries. The observation that a large part of the effects of trade liberalization happens within sectors has led to a growing focus on intrasectoral changes in the wage distribution. The mainstream literature has focused on how changes in the relative returns to workers with different characteristics (e.g. skill level) and in the composition of labor demand have contributed to wage variation across firms within sectors. This is motivated by the fact that the returns to factors vary with worker characteristics and trade liberalization may increase demand for workers with certain characteristics. These characteristics can be either observable (e.g. education), for which some datasets provide relevant information, but may also be unobservable (e.g. innate ability or competency), which firms can to some extent screen for, but which are impossible to be revealed in datasets. 1 Brambilla et al. (2012) find robust evidence for skill upgrading and concomitant higher payrolls induced by exporting to high-income countries, but this result does not hold for exporters to countries with similar market characteristics as the home country (Argentina). This effect is, however, only partly explained by the additional hiring of high-skilled workers, but remains valid when controlling for the number of high skilled workers. Hence, within firm wage inequality does rise, but lower skilled production workers seem to also benefit from higher wages (of course the dichotomy high skilled vs. low skilled has certain nuances to it and it cannot be observed from the data whether those workers on the lower side of the scale do benefit or not from these average wage raises). The authors argument is that exporting to high-income 1 See the review in Winters and Matuscelli (2014) for a brief overview of such studies

22 22 TRADE AND EMPLOYMENT: AN OVERVIEW countries increases pressures for skill upgrading through higher quality valuation in those markets. A similar result has been found by Verhoogen (2008), who observed a similar phenomenon in Mexican plants following the devaluation of the Mexican peso, a proxy for trade liberalization, as the lower value of the currency promotes exports. His results suggest that the export share of higher productivity plants has increased with the devaluation and wages have increased as well as the share of white collar to blue-collar workers. Thus, inequality within Mexican industries has increased through expanding trade. Nevertheless, he also finds that both wages to skilled and non-skilled workers have increased in absolute terms, suggesting that both types of workers have gained from greater trade. Wage differences and employment levels according to trading destination are also a feature of Dutch importers and exporters examined in a study by Fortanier et al. (2011). Accordingly, while firms participating in international trade are generally larger in terms of employment and pay higher wages than strictly domestically operating firms, there are differences in terms of where firms import from and export to. For example, the authors find that importing from developing countries is associated with lower wages, suggesting that cheap imports could indeed substitute for domestic labor to some extent. However, they find the same effect for imports from the US, which does not support this hypothesis. Apart from export destinations, the composition of trade seems to be relevant as well for firm characteristics, notably whether the good in question is an input or an output good. Using detailed Indonesian manufacturing census data, Amiti and Davis (2012) find strong evidence that a fall in tariffs on output goods lowers wages at importcompeting firms but boosts wages at exporting firms. Likewise, a reduction of tariffs on inputs leads to higher wages paid by importing firms relative to their locally sourcing counterparts. In the likely event that domestically operating firms use unskilled labor more intensively, this finding may reinforce the view so far expressed that trade can contribute to rising inequality. In general, the results so far suggest that trade can contribute to greater wage inequality, through the differential wage impact according to skill level. Nevertheless, less skilled workers seem to gain from trade as well, albeit less proportionally than higher skilled workers. Notwithstanding the large body of literature, Ebenstein et al. (2014) suggest that the industry level might be the wrong level of analysis. They hypothesize that workers are likely to be mobile across industries, but more likely to remain in their initial occupation as they switch jobs, in which case the industry level of analysis would miss an important point in detecting potential wage effects. Constructing occupation-specific measures of offshoring, import competition and export activity, they show that a 1 percentage point increase in occupation-specific import competition in the US is associated with a 0.25 percentage point decrease in real wages, an effect that industryspecific studies are unable to detect. Some occupations in the US have faced substantially more import competition than others, an example of the former being shoe manufacturers and teachers for the latter. They further show that workers driven by import competition to leave the manufacturing sector, often end up in other sectors where they face substantial wage declines (for example 6 to 22 percentage points when moving from manufacturing into the service sector). Another strand of the literature focuses on between-firm differences in wages as a mechanism for trade to affect wage inequality. Drawing on data from Brazil, (Helpman et al., 2012) find significant trade-induced inequality impacts that cannot be explained by changes in relative wages between different sectors and types of workers. As such, an important channel of the effect of trade on inequality has been overlooked in earlier research. The authors estimations imply a sizable effect of trade participation on inequality, with exporters on average paying higher wages than non-exporters even after controlling for employment and labor composition. However, the relationship between trade openness and wage inequality is non-monotonic, meaning that it first exacerbates inequality and then reduces it. These results are consistent with a large body of literature that confirms that foreign owned firms pay higher wages, even when controlling for labor and industry characteristics. Harrison and Rodriguez-Clare (2009) conclude that at any rate, foreign owned firms do not pay wages below those prevailing in the host country. Trade and Job Quality We have hitherto focused on the implications of trade for the quantity of jobs (employment) and wages, but the broader implications on the quality of jobs are also of interest in order to improve living conditions of the indi-

23 TRADE AND EMPLOYMENT: AN OVERVIEW 23 vidual. Trade-induced competitive pressures may lead to increased international competition in prices, which may translate into cost cutting and lowering of labor standards, the so called Race to the bottom effect. Core labor standards are defined by the ILO as the elimination of all forms of forced or compulsory labor, effective abolition of child labor, equality in opportunity and treatment, and freedom of association and right to collective bargaining. Yet, quality of jobs can be defined in an even broader sense, taking into account e.g. working hours, health and safety at the workplace, and job security and benefits. Labor Standards While several examples of trade related flagrant disregard of workers safety repeatedly grab the headlines of international news (e.g. child labor, extreme working conditions, disregard for union rights, etc.), a number of studies find that there is no indication of a systematic link between trade and low working standards. Some of those studies reviewed in Newfarmer and Sztajerowska (2012) suggest that in general countries with lower core labor standards have not enjoyed better export performance. Flanagan (2006) reviews a larger sample of countries over a period of 30 years and concludes that open economies, in fact, have significantly better working conditions than more closed economies, including fewer accidents at work and greater freedom of association. Moreover, low labor standards deter, rather than encourage foreign direct investments by multinational companies. Contrarily to often voiced public concerns, it appears that foreignowned and export oriented firms may even improve working conditions, even beyond the core labor standards as defined by the ILO. One mechanism through which trade and labor standards may be positively linked is the nexus between per capita income and development. Concomitant shifts in sectoral production patterns from agriculture towards manufacturing and services reinforce this trend, as agriculture is usually associated with the worst working conditions. Resulting urbanization may be a pull-factor for the highly visible phenomenon of urban poverty in developing countries, but studies confirm that, in general, urban jobs tend to deliver higher wages and better working conditions (Robertson et al., 2009). Another mechanism through which the trend towards trade-induced improvement of working conditions occurs is that foreign and export oriented firms are more susceptible to pressure from labor groups, leading them to exhibit greater compliance with minimum wages and labor standards (Harrison and Rodriguez-Clare (2009). This is due to increasing activity of reputation-sensitive international buyers and pressure from advocacy groups, which appear to have a discernible impact. Harrison and Scorse (2008) find that foreign and export oriented firms in Indonesia were much more likely than domestic enterprises to adhere to minimum wages as a consequence of the anti-sweatshop campaigns there. They also find that the employment costs of the anti-sweatshop campaigns were minimal, as garment and footwear subcontractors were able to reduce profits to pay the additional wage costs without reducing the number of workers. Informality Most of the available evidence centers on workers and firms in the formal sector. Informality is a much more important phenomenon in developing countries, where up to 80% of employment is informal (Sinha, 2011). The informal sector is characterized by generally worse working conditions, as any form of legal protection, which is normally benefiting workers in the formal sector, does not cover informal economic activities. Trade may impact entry and/or exit of firms and/or workers into or out of the informal sector. The impact of trade on the size of the informal sector is unclear, with some studies finding a positive link, and others a negative one (Sinha, 2011). Trade liberalization in Brazil has been associated with a higher probability of displaced workers from import competing firms exiting into the informal sector (Menezes-Filho and Muendler, 2011). However, there seems to be no systematic evidence for this effect. Average wages in the informal sector are generally lower than in the formal sector and firms participating in international trade generally operate in the formal sector. Hence, in conjuncture with the previous findings, one would expect trade to increase inequality across formal and informal sectors, even though one cannot account for potential benefits informal suppliers of formal firms may incur through the latter s trade. Nevertheless, Sinha (2011) notes that the shares of workers in informality have been persistent over time, while trade has increased dramatically over recent decades. She concludes that country-specific trade composition, supply capacity, details of the trade liberalization scenario and other specific circumstances such as labor market condi-

24 24 TRADE AND EMPLOYMENT: AN OVERVIEW tions, determine the direction of the impact of trade on informal employment. Trade policies appear to be less determinant in informal employment outcomes than income per capita, labor market policies and administrative barriers to formalization. Gender Through altering employment patterns, trade may have implications for the gender composition of the labor force. Greater participation of women in the labor force offers ways to break out of traditional gender stereotypes that negatively affect women s social status and income. Moreover, evidence suggests that women tend to spend a larger proportion of their income on their family than men, which has important implications for fighting poverty and inequality. Intersectoral reallocation changes may occur if trade liberalization favors female-intensive sectors, in which case traditional trade theory would predict higher female employment and wages. Aguayo-Tellez et al. (2010) find tariff reductions in Mexico in the context of NAFTA have increased women s share of the wage bill by 5.3 percentage points resulting from both increased hiring and higher wages for women. Moreover, the authors find evidence of household bargaining power shifting in favor of women. Household expenditures shifted from goods associated with male preference, such as men s clothing and tobacco and alcohol, to those associated with female preference such as women s clothing and education. Berik (2011) describes how trade expansion has brought a substantial increase in employment for women workers in developing countries notably in labor-intensive exportoriented manufacturing industries since the mid-1970s. Given their limited access to education and hence inferior skill levels women may have benefitted disproportionally from these new work opportunities, as these industries are generally associated with lower skill requirements. Moreover, relatively high concentration of female workers in sectors producing tradable goods and lack of opportunity in the domestic economy in the developing world has made them more vulnerable to external shocks. ILO (2012) reports that the recent financial crises and the resulting contraction in export-oriented manufacturing industries and tourism have led to greater job losses among women in many developing countries. Gaddis and Pieters (2012) examine the effects of trade liberalization in Brazil and find that Brazilian States with greater exposure to trade liberalization experienced faster increases in female labor force participation and employment. However, the increase in the share of the female labor force is mainly driven by female labor absorption in the service sector. Another channel that may promote greater employment of women instead of men is that trade liberalization stimulates technological upgrading in firms, which may result in greater demand for women in less physically intensive activities. Juhn et al. (2013) hypothesize that new technologies replace physically demanding tasks with computerized production, making women relatively more productive in low-skill jobs. The authors confirm this hypothesis with Mexican data around the time of NAFTA trade liberalization and find that indeed new export entrants upgraded their technology and increased the ratio of female to male blue-collar workers as well as their relative wages. The same trend can be observed in service offshoring activities. Within IT and IT enabled services, an important services subsector in terms of its potential for services trade, the situation appears to be more favorable for women than in other sectors, even though research is far from conclusive on that question. A World Bank analysis by Sudan et al. (2010) finds that women account for about 65% of the total professional and technical workers in the IT services and IT enabled services in the Philippines. In India, women make up 30 percent of the workforce a much higher rate of female participation than in the service sector in general. In Ireland, 70 percent of call center employees are women. According to this study, women account for a greater number of high-paying jobs than in most other sectors of the economy in each of these countries. Finally, gender discrimination may simply be reduced through trade liberalization s competitiveness effects. Evidence on this transmission channel, however, is mixed. For example, Ederington et al. (2010) find that firms facing greater tariff reduction in Colombia have reacted by hiring more women, increasing the share of female employment. On the other hand, Menon and Rogers (2009) find that the wage gap in the Indian manufacturing has actually widened as a result of trade-induced competitive pressures. They argue that pressures to cut costs have had to be borne by women, as they have less bargain power than men.

25 TRADE AND EMPLOYMENT: AN OVERVIEW 25 Conclusion International trade is neither the culprit of massive unemployment as critics of globalization like to proclaim, nor the engine of job creation that its supporters like to think. Trade is a double edged sword that can create better employment opportunities for some, while denying those to others. While trade policy can play an important role in steering toward the desired employment outcome, a multitude of other factors can interfere and make this task a challenging one for policy makers. The desired response of trade flows to policy changes is not straightforward and even less so is the link with domestic employment outcomes. Nevertheless, on balance, research supports the case for overall trade friendly policies that if carefully designed may translate into more and better employment opportunities than policies on the other side of the spectrum. International trade is becoming an ever more present reality and sustains an increasing number of jobs worldwide, as progress in transport, and information and communication technology continues unabated. As the global economy gets increasingly interconnected, the nature of international trade is changing. The last decade has witnessed a revolution in the analysis of international trade, both theoretically, driven by the emergence of modern trade theories, but also empirically, as internationally comparable input-output data have revealed how trade in intermediates is an ever more prominent feature of international trade. As such, trade policy is likely to gain greater importance as a tool to steer economic development. At the same time, as long held and rather simplistic tenets of the traditional trade literature are being put into question, new evidence based directions for policy makers need to emerge.