Robert Feenstra is one of the most authoritative scholars in the field of International Trade. Throughout the years, he has produced innovative theories, has collected data that in many cases have been made publicly available, and has conducted rigorous and original empirical research. One of the fields where he has made seminal contributions is the analysis of offshoring, in joint work with Gordon Hanson (Feenstra and Hanson, 1996, 1999).Footnote ¹ He has introduced a measure of offshoring based on input–output tables that has been widely used in subsequent studies, and showed how such a phenomenon might have contributed to labor market outcomes for the US since the 1980s.

Offshoring in the Global Economy, a collection of the Ohlin lectures given by Feenstra in September 2008, is a very welcome contribution to the offshoring literature. It carefully reviews the workhouse models, with a constant focus on their capacity to explain stylized facts about the US labor market and economy. It provides new, and challenging, macroeconomic implications that should be kept in consideration in the policy debate about this phenomenon. It shows new channels through which offshoring can have relevant productivity effects.

The book is divided into three chapters. The first discusses the microeconomic structure of the various models in the literature, with a focus on the labor market consequences of offshoring (in terms of wages and employment of different types of workers and their productivity). The second presents three applications on the macroeconomic effects of offshoring, namely volatility, exchange-rate pass-through, and terms of trade. The concluding chapter presents some ideas for further research on the mechanisms through which international trade can affect productivity.

The first chapter starts with an exposition of some stylized facts on labor markets in the 1980s. In the US and other countries, both the employment and wages of skilled workers increased relative to unskilled workers. A simple supply and demand framework suggests that this could only happen due to an outward shift in the relative demand for skilled workers. While most economists have explained this with skill-biased technological change (henceforth, SBTC), Feenstra belongs to the group of economists who believe that trade has something to do with this. His point is that a slight modification of the Heckscher–Ohlin model with a continuum of goods, with activities replacing goods, is sufficient to yield results that are observationally equivalent to SBTC: an increase in the employment and wage premium of skilled labor.Footnote ² Since the two explanations are observationally equivalent, only econometric analysis can assess their relevance. From the analysis of Feenstra and Hanson (1999), it emerges that roughly one quarter of the increase in relative wage is explained by material offshoring, and about 30% by the use of high-tech capital, a measure of SBTC.Footnote ³ Unfortunately, the author does not mention any study performing a similar exercise for the employment premium.

The evidence for the 1990s is that of an increase in the relative wage of skilled workers, coupled with a decrease in their relative employment. These trends cannot be explained by an outward shift in the relative demand for skilled labor. The author notes that the only possible explanations are: labor demand shifts that occurred within the category of skilled workers – namely, an increase in the relative demand for workers towards the high end of the skill distribution and away from workers with medium skills – or service offshoring. Once again, the author discusses the relative contributions of skill-biased technical change (in this more sophisticated version that looks at the spectrum of skilled workers rather than at the dichotomy skilled–unskilled) and offshoring in the review of the empirical literature. He mentions the study by Sitchinava (2008), who shows that the increased use of computers explains 95% of the rise in the relative wage of skilled labor, while service offshoring contributes to 15% of that increase and material offshoring to 7%. These figures imply that services offshoring might have a larger impact than the materials type on the dynamics of relative wages in the 1990s. However, the explanation advanced by labor economists, focusing on technological factors that may have little to do with trade – let alone with offshoring – seems to be much more empirically relevant.Footnote ⁴

Although the author chooses to focus on the less empirically relevant explanation for the patterns of relative wages and employment, his exposition of the theoretical model of offshoring that is consistent with these stylized facts is clear and elegant. He starts with an excellent overview of the ‘trade in tasks’ model of Grossman and Rossi-Hansberg (2008) (henceforth, GRH), pointing out the crucial assumption that home firms bring their own technology abroad when they offshore a task, and the crucial result that offshoring of low-skilled intensive tasks acts exactly like a low-skill-biased technical innovation that requires less of this input for unit of output. GRH refer to this as ‘productivity effect’, which raises the relative wage of low-skilled labor. This result is however contrary to the empirical evidence. In the large-country extension, GRH allow for output changes due to offshoring: if full employment of low-skilled labor is to be preserved, the increase in the output of the low-skill intensive sector must be larger than the reduction in the output of the other sector (a Rybczynski effect). This leads to an increase of its relative supply on world markets and to a subsequent reduction in its relative price, which (by the Stolper–Samuelson channel) reduces the relative wage of low-skilled workers. GRH define this ‘price effect’.Footnote ⁵ A very useful result that Feenstra is able to show is that the price effect dominates the productivity effect (thus leading to the reduction in the relative wage of skilled labor shown by the data) if the elasticity of substitution in production is less than one. This is a testable implication in a cross-section of sectors, because both the elasticity parameter and factor returns are industry-specific in the data. The second important point made by Feenstra is that even the simple small-country version of the GRH model can replicate the stylized fact of an increase in relative wage, and decrease of relative employment of skilled labor, simply by assuming that offshoring is done for high-skilled tasks rather than low-skilled ones. In this case, offshoring has a productivity effect on skilled labor that raises its relative wage, but at the same time it reduces its employment share because of a reduction in the demand for domestic skilled workers.

The first chapter concludes with some factor content calculations for the US. The question is: can changes in factor content of trade explain the observed patterns of relative wages? Feenstra argues that it can, provided that factor intensities are imputed from highly disaggregated trade flows to correct for the bias due to the aggregation problem.Footnote ⁶ He claims that trade (recorded at the 10-digit level) has an impact on the effective factor endowment that, since 1994, is large enough to explain changes in relative factor prices. For instance, 1994 trade patterns predict a net import of 10 million production (low-skilled) workers, which increases the ratio of production to non-production (high-skilled) workers from 2.3 to 3.9. This is supposedly related to offshoring. The link between offshoring and factor content is however not made explicit by the author. It is not clear how much offshoring, relative to, say, policy factors, contributed to this increase in the effective endowment ratio. Moreover, calculations for 1997 and 2000 were done using the same factor requirements of 1994 (only updating trade data). The numbers reported by the author on the effective endowment ratio for these two years cannot therefore take into account any type of factor-biased technological change that might have occurred after 1994.

The second chapter shifts focus from the microeconomic structure to the macroeconomic implications of offshoring. Feenstra presents three implications: the effect on business cycle volatility, the impact on prices and exchange rates, and the impact on terms of trade.

The first implication is that, if country i offshores to country j, it exports its business cycle fluctuations to j. For instance, if demand is booming, employment volatility is amplified in j, while the impact of the demand shock is offset in i by shifting production abroad. This result is confirmed empirically: Mexican maquiladora industries are more volatile than similar industries beyond the border (be it in the US as a whole, in California, or in Texas). Two other interesting things are worth mentioning here: first, the extensive margin of employment adjustment in maquiladora industries is large (plant exit and entry accounting for one third of the adjustment);Footnote ⁷ second, the focus on US demand shocks, rather than on productivity shocks, is typical of the business-cycle models. Bergin, Feenstra, and Hanson (2009b) show that demand shocks, as opposed to productivity ones, are the main source of extra volatility in the Mexican offshoring sector.

The second macroeconomic implication of offshoring explored by Feenstra is the impact on the pass-through of exchange rates on import prices. The argument goes as follows. If the US imports a lot from countries with fixed exchange rates (say, China), a depreciation of the US dollar will raise the dollar price of goods imported by countries with a flexible exchange rate (say, Mexico) by less than one-to-one. This is because competition from China limits price increases from other countries, dampening the pass-through of the dollar exchange rate to US import prices. Feenstra supports this idea with a theoretical model with flexible mark-ups and with empirical estimations showing incomplete pass-through of the exchange rate to import prices in the presence of competition from China and other countries with fixed exchange rates. Interestingly, in the theoretical model the dampening effect on the pass-through is caused by the extensive margin of Chinese competition (number of Chinese firms exporting to the US), rather than the intensive margin. Unfortunately, the author does not have data on the number of Chinese firms serving the US market or on their market share, so in the regressions he has to use the intensive margin (overall Chinese import share). If such data were available, it would be interesting in future research to quantify the effects of the two margins of Chinese competition.

The third, and in my view most interesting, macroeconomic implication concerns the effects on the terms of trade. Feenstra makes two related points. First, if import competition lowers import prices, this should be reflected in the price deflator used to construct real GDP measures. Second, if import price deflators do not reflect tariff cuts and increased import variety, terms of trade and the GDP price deflator are measured with a downward bias. Once such adjustment is made, it turns out that one-fifth of the increase in US real GDP growth since 1995 reflects improved terms of trade, while it is incorrectly counted as productivity growth. The reader might wonder: Why does it matter? Has offshoring anything to do with all this? Concerning the first question, Feenstra argues that this matters not only for statistical purposes, but especially because it implies a more benign view of import competition than the one commonly held in the policy debate and in the popular press. I could not agree more. As for the second question, the answer is positive. Reviewing the evidence from the Information Technology Agreement (ITA), Feenstra et al. (2005) show that tariff reductions have a magnified effect on price reductions, not only because of the multilateral nature of tariff cuts, but also because of the fragmentation of production (offshoring).

In the final chapter, Feenstra argues that the evidence on the skill premium since 2000 is mixed (rather erratic movements in relative wage and employment of skilled workers). This calls for new, more articulated explanations, based for instance on the observation that the most significant changes have occurred during downturns in the US economy.

Next, Feenstra proposes some research ideas on how offshoring can affect productivity. The first is based in the idea of supermodularity of production, which means that different tasks within a firm are complementary. Offshoring allows firms to narrow the skill distribution of their workers, for instance if they systematically offshore low-skilled labor. In this case, the marginal product of high-skilled workers would increase because of complementarity in effort with coworkers with similar skills.

The second research idea puts together supermodularity of production and the endogenous choice of effort by workers. Feenstra presents a theoretical model in which trade endogenously increases product variety and, consequently, the level of effort exerted by workers and their productivity.Footnote ⁸ The idea that trade is associated with productivity gains is not new: it is the new source of gains from trade identified by Melitz (2003). What is new, and very interesting, here is the underlying mechanism. While in Melitz (2003) trade is associated with selection and share-shifting effects towards more productive firms, in the model sketched by Feenstra it is associated with an increase in workers' effort. In this respect, the only criticism that I would make is that the model is not specific to offshoring. In fact, trade liberalization is equivalent to an increase in country size.Footnote ⁹

As a concluding remark, I believe that the book under review will be extremely useful for academics interested in understanding the offshoring literature and/or looking for research ideas. But more importantly, it can serve as an authoritative guide on the topic for policy-makers, whose decisions contribute to shape the extent and deepness of trade integration and, consequently, the potential impact of offshoring on the economy.