برونسپاری، دستمزد، و استخدام : تئوری و شواهد

This paper investigates the wage and employment effects of offshoring. I use firm-level data and two events in Mexico as a natural experiment to identify the effects of a fall in the marginal cost of offshoring to Mexico. I find that domestic wages actually rise at U.S. firms likely to take advantage of this new offshoring opportunity. At the same time, domestic wages fall at U.S. firms unlikely to take advantage of this opportunity. Furthermore, I find no evidence of greater domestic job loss at the former compared to the latter firms. These findings are consistent with productivity effects from offshoring. To explain the mechanism, I develop a theoretical framework that combines heterogeneous firms with imperfect labor markets and rent-sharing. Firms likely to take advantage of new offshoring opportunities increase their productivity and profitability at the expense of their competitors. Through rent-sharing, this channel leads to higher domestic wages at the former firms relative to the latter. Further, there is no empirical evidence of greater domestic job loss at the firms likely to expand their offshoring compared to their competitors that are unlikely to increase their offshoring.

Offshoring1 has been a source of controversy, with opponents charging that it hurts domestic workers. This view contends that firms that offshore reduce wages and shed jobs. However, economists have long challenged these notions. One notable instance occurred when Gregory Mankiw, while serving as chairman of the Council of Economic Advisors, caused an uproar by commenting that offshoring is only “the latest manifestation of the gains from trade that economists have talked about at least since Adam Smith.“2 But as trade generally increases welfare in the aggregate, I also produce “winners” and “losers.” If offshoring is the latest manifestation of gains from trade, then who are the winners and losers? Briefly summarizing the results, this paper finds that profitability and average domestic wages actually rise at firms likely to take advantage of a new offshoring opportunity relative to the profitability and average domestic wages at firms unlikely to take advantage of the new offshoring opportunity. Second, while there may be employment losses at the former firms, these losses are no greater than employment losses at the latter firms. These results are consistent with evidence of productivity effects from offshoring. To empirically test the productivity effects from offshoring, I first put forth a model that provides clear, testable predictions. The model incorporates aspects of Grossman and Rossi-Hansberg (2008) in a context of heterogeneous firms and imperfect labor markets. There are three key features to my model. First, imperfect labor markets are characterized by search costs and bargaining, leading to wages as a function of firm-specific rents. Second, firms are heterogeneous in productivity, which leads to differences in scale. There exist fixed costs of offshoring while the benefits from offshoring increase in the scale of the firm; as a result only the most productive firms are able to offshore (“MNCs”). Among the MNCs, the most productive are able to offshore to multiple destinations (type I MNCs), while moderately productive firms are able to offshore to a single destination abroad (type II MNCs). Meanwhile, the least productive firms source solely from the domestic market (“purely domestic firms”). The third feature of the model, following Melitz and Ottaviano (2008), is that markups are endogenous. Offshoring reduces marginal costs of production for a firm (productivity effect), enabling it to not only expand output but also increase its markups and profitability, which are then shared in the form of higher wages. This mechanism, which I call the productivity plus rent-sharing (PRS) effect, varies by firm and is stronger at more productive firms. In the model, an exogenous shock provides a new offshoring opportunity. Endogenously, it is profitable for type I MNCs to take advantage of this opportunity while it is not profitable for type II MNCs and purely domestic firms. The productivity effect from this additional offshoring leads to higher profitability and domestic wages at type I MNCs as discussed above. In addition, the productivity effect from offshoring allows type I MNCs to expand production at the expense of type II MNCs and purely domestic firms (“business-stealing effect”). The business-stealing effect leads to a fall in the relative competitiveness of type II MNCs and purely domestic firms, thereby leading to lower markups, lower profitability, and consequently lower wages at these firms. This mechanism implies that wage dispersion across firms in the sector increases, with most of the wage dispersion occurring at the upper tail of the wage distribution and little or none occurring at the lower tail, which is consistent with recent empirical findings by Autor et al. (2008). Finally, the business-stealing effect generates a reallocation of market share and employment from purely domestic and type II MNCs towards type I MNCs. Hence, at type I MNCs, the net effect on domestic employment is ambiguous, with the direct job loss due to new offshoring potentially offset by expansion from the productivity effect. The net effect at type II MNCs is also ambiguous: they contract and shed jobs domestically but they also are forced to inshore some of their activities because of lower profitability. Purely domestic firms simply contract and shed jobs domestically. In this paper, I think of the productivity effect/efficiency gain from offshoring simply as lowering the marginal costs of production for the new offshoring firm. However, it is possible to think of additional and potentially more powerful productivity benefits. For example, if a firm is able to offshore its non-core tasks and focus on its core tasks, there could be productivity gains from specialization. Also, the firm could reinvest the cost savings from offshoring in research and innovation, providing long-term productivity boosts. Examining these other productivity channels would be interesting avenues for future research. The contributions of this paper are several-fold. First, I extend previous work on the productivity effects of offshoring3 to develop firm-level predictions on domestic wages and employment. Then, using firm-level data and a natural experiment, I am able to carefully identify the causal link between offshoring and wages and test and some of the theoretical predictions. Merely demonstrating that offshoring firms pay higher wages is not sufficient as offshoring and wages are endogenous. High wages may cause a firm to offshore, or a third factor, such as productivity, could cause a firm to pay high wages and offshore. Therefore, by employing firm-level data and an exogenous shock, I can provide evidence of a causal relationship between offshoring and domestic wages and employment. In particular, for the identification strategy, I take advantage of two episodes in Mexico as exogenous shocks to the marginal cost of offshoring to Mexico for US firms.4 First, the Foreign Investment Law (FIL) of 1993 relaxed restrictions and reduced both pecuniary and non-pecuniary costs of foreign ownership of Mexican firms.5 Second, the peso depreciation at the end of 1994 significantly lowered real wages of Mexican workers in dollar terms, thereby making Mexico a more attractive platform for offshoring. I have separated firms into treatment and control groups. The treatment group, which is the empirical analog to type I MNCs in the model, includes US firms with an offshoring presence in Mexico as of 1993. These firms, having already paid the fixed costs of entry, would be in position to take advantage of a fall in the marginal cost of offshoring. The control group, which is the empirical analog to type II MNCs in the model, includes US firms that were offshoring to other Latin American countries, excepting Mexico, as of 1993. These firms would be less likely to respond to a fall in the marginal cost of offshoring since they would still have to pay the fixed entry cost. Comparing the two, I find that offshoring, operating profits per domestic worker (profitability), and average domestic wages increased more for treatment than control firms during the period 1993–1997. Further, these differential changes are statistically significant when compared with the time period 1997–2001 for all the same outcome variables. In addition, this empirical analysis finds no evidence of greater job loss at treatment firms relative to control firms. One potential issue is that a labor composition effect (see Feenstra and Hanson, 2003) could also be consistent with the aforementioned findings. If type I MNCs offshore the lowest-paying jobs, then average domestic wages would increase mechanically with offshoring. I develop two different proxies for firm-level labor composition and using these different measures as controls, I still find that domestic wages increased more at treatment firms compared to control firms, evidence of the PRS effect. Using back-of-the-envelope calculations, I estimate that 20–30% of the wage differential between treatment and control firms can be explained by the PRS channel, with some or all the remainder due to the labor composition effect or alternative channels. Section 2 delves further into related literature. 3 and 4 develop the theoretical framework, and Section 5 examines the comparative statics, given an exogenous fall in the marginal cost of offshoring. Section 6 describes the data and presents some descriptive statistics. Section 7 provides background on the episodes in Mexico, describes the empirical methodology, and presents the main results. Section 8 addresses alternative hypotheses. Finally, Section 9 describes my conclusions.

This paper investigates the idea of offshoring as technology, with a focus on trying to understand the productivity effects of offshoring on firm level domestic wages and employment. This paper finds evidence that US MNCs that were more likely to take advantage of a new offshoring opportunity experienced a rise in relative profitability and average domestic wages during the shock period, 1993–1997. This differential change was greater during the 1993–1997 period than during other periods without similar shocks to offshoring costs. In addition, the differential change was greater for firms in industries with ostensibly higher levels of rent-sharing. I interpret this as evidence of a causal link between offshoring and greater profitability and increased domestic wages at the parent firm. Meanwhile, the empirical analysis finds no evidence of greater employment loss due to offshoring between treatment and control firms. The evidence described above is consistent with offshoring's impact on wages and employment through a productivity and rent-sharing mechanism. Among heterogeneous firms, only certain firms are able to take advantage of offshoring opportunities. Offshoring enables these firms to experience productivity gains, which raises their profitability and domestic wages. The productivity gain allows these firms to become more competitive and steal business from firms unlikely to enter into or expand their offshoring. This productivity-fueled expansion helps offset the direct loss of domestic jobs to offshoring while the business-stealing effect leads non-offshoring firms to contract, which lowers their profitability, domestic wages, and employment. The findings in this paper challenge the concept that offshoring most negatively affects domestic wages and employment at firms that offshore. Rather, this research suggests that offshoring should be viewed as a technology, enhancing productivity and competitiveness, and offers new evidence on winners and losers from offshoring. Workers at MNCs who are not easily substitutable could benefit the most from offshoring while workers at purely domestic firms, interestingly, could be hurt due to their firms' loss in competitiveness. What, then, about the workers at MNCs whose jobs have been offshored? It would be interesting to know if these workers were re-hired at expanding MNCs at higher wages or if they had to take an outside job at a far lower wage. This avenue of empirical work would be useful for developing better targeted policies aimed at helping workers who are hurt by offshoring. Further, this paper assumes that workers are homogeneous. However, worker heterogeneity likely plays an important role. For example, offshoring could have heterogeneous effects on the bargaining power of workers depending on the type of tasks that they perform. Extending the theory and empirics to consider these additional heterogeneous effects on workers would be an interesting direction for future research.