Offshoring is the movement of jobs to other countries. While this may raise overall welfare, it means lower employment opportunities and possibly lower wages domestically. Offshoring is the major cause for the decline in self-reported work life quality in many countries. For example, in the US there are six workers that view offshoring negatively for every one worker that sees it as a positive influence (Pew Center 2006). Such a negative impression makes public policy backlash likely. No less important for the policy debate is offshoring’s prevalence. The implications for public policy will be quite different depending on whether offshoring will affect 3 million jobs or 30 million in countries such as the US (Blinder 2007).
Offshoring takes lower wages … and what, exactly?
Alas, there is no agreement on what makes a job offshorable. Are jobs that are routine and therefore can be done by computers most susceptible to offshoring (Levy and Murnane 2004)? Perhaps, but computers tend to replace low-skilled workers anyway. Also high-skilled activities such as tax return preparation might be offshorable if they can be easily delivered over the internet (Blinder 2007). Moreover, while the classic examples of offshoring to low-wage countries (call centres in India) are well known, what about the sizable offshoring to richer countries (Dell Computers in Ireland)? All told, estimates of the importance of offshoring in terms of the number of affected jobs vary by an order of magnitude, for the US, e.g., from 2% to perhaps 40% of the workforce.1 And the range of forecasts is even larger.
Limit #1: Geographic distance
While it is clear that lower factor costs are the main draw for firms to offshore, the limits to offshoring are not as well understood. Our research stresses the two principal factors.
A good way to examine offshoring is by considering FDI, where a domestic company owns a plant in another country.2 Figure 1 presents the size of US FDI into other countries.
The figure shows a clear negative relationship between the sales of US multinational affiliates and the trade costs (physical and policy barriers) from the US. In addition to the usual gravity finding – larger markets attract more sales – Figure 1 indicates that offshoring is helped by geographic proximity.3 Moreover, this pattern holds also at the level of individual companies. That is, a typical US multinational company sells more in Mexico than it does in Ireland, and its sales in Ireland are larger than those in New Zealand (conditional on host country market size). Why is there more offshoring to nearby countries?
Transferring information is costly
Offshoring frequently requires incurring technology transfer costs because there is a phase of communication involving problem-solving between multinational parent and affiliate to get the technology to ‘work’. Arrow (1969) is an early formulation of this idea. The following illustrates why transfer costs explain that affiliate sales are declining as trade costs are increasing.
Firms’ transfer of technology: disembodied or embodied?
A multinational has two options to serve a foreign country.
- It can complete a task at home and export it as an intermediate good. In this case the firm’s technology is embodied in the intermediate good, which incurs trade costs.
- The firm can transfer the required technology (disembodied) to its affiliate, at cost, and the task is completed offshore.
As distance to a foreign market increases, raising trade costs, the multinational firm will want to transfer more tasks to its affiliate. But with either transfer, costs are higher than before, implying that the affiliate is smaller and offshoring is lower – a pattern as in Figure 1 emerges. Information on individual US multinational firms strongly supports this idea.
Offshoring requires complementary inputs from home
What is the key difference between our framework, giving a negative relationship between trade costs and affiliate sales, and the traditional exports-versus-FDI analysis resulting in a positive correlation? The latter is included in Figure 1 as well in the form of the (smoothed) proximity-concentration prediction of Brainard (1997). In contrast, we obtain a negative correlation because technology cannot be transferred to affiliates at zero cost. More generally, geography imposes a limit to offshoring because tasks that a firm can better complete at home are often complementary to those that it can offshore.
Limit #2: Technological complexity
Offshoring and trade often result from firms spreading their production networks across countries. There are many individual intermediate goods that each may be either produced domestically or in various countries abroad, after which they are combined at the point of final consumption. Figure 2 shows a typical sourcing strategy in our theory. Here, we have a US multinational firm (headquarters in the US) that is selling a final product in Canada, and the intermediates are sourced from India, Canada, and the United States.
Offshoring costs have both country- and task-based characteristics
There are two reasons why the technological complexity of intermediates imposes a limit to the extent of offshoring.
- The offshoring host economy has a limited absorptive capacity to manage complex technology.
- More technologically complex tasks are harder to offshore because they require more problem-solving communication between the multinational parent and affiliate’s workers.
A country’s absorptive capacity for offshored production might be low due to a lack of infrastructure, as well as skilled labour. At the same time, labour endowments may in fact matter less than the ease with which the information to complete a task as such can be transferred (the correlation of skill requirements and offshorability is weak). In Figure 2, therefore, we focus directly on tasks, which are ordered left to right from easiest- to hardest-to-transfer.
Relatively low factor costs in India are the main reason why the US multinational will complete the least-complex tasks there (Figure 2). For task complexity in the middle range, Canada will be the optimal location of production. Technology transfer costs to Canada are likely lower than to India (more web servers per capita, lower travel costs for face-to-face interaction). Completing tasks in Canada also saves on intermediate trade costs since Canada is the location of final consumption. Finally, tasks for which the required information is most difficult to transfer will be completed at the headquarters in the US and then exported. This means that the low-wage offshoring advantage of poorer countries can be reduced or even overturned by lower costs of technology transfer.
First of all, most of the R&D of multinational firms is done at the headquarter (parent) company, which supports the idea that the most complex tasks are completed in the headquarters (see Figure 2). Moreover, data on individual US multinational firms confirms that offshoring in highly complex industries (measured by headquarter R&D intensity) is much more difficult than for the typical industry (about 33% lower). Another implication of this framework, related to Figure 2, is that the tasks embodied in headquarter exports are more and more technologically complex for destinations further and further away from the headquarter country. Detailed information on the skill- and occupation-content of US exports strongly supports this result.
The bottom line
Current and future trends in offshoring will be best understood by considering both what makes offshoring attractive (lower factor costs) and what are the major limits to it (technology transfer costs as well as trade costs). Which jobs in which industries will be most heavily affected depends not only on whether and how fast factor prices are converging, but perhaps more importantly on the race between technology transfers and trade costs.
1. See Blinder (2007) who discusses several influential estimates.
2. Another form of offshoring occurs when a domestic company buys from or sells to a foreign independent company (sometimes called outsourcing). Our main argument presented below applies to this case as well.
3. Affiliate sales in Figure 1 are divided by domestic sales to ensure data confidentiality. This also adjusts for productivity differences across firms.
Arrow, Kenneth (1969), “Classificatory Notes on the Production and Transmission of Technological Knowledge”, American Economic Review 59(2): 29-35.
Blinder, Alan (2007), “How Many US Jobs Might be Offshorable?”, Princeton University, CEPS Working Paper No. 142, March.
Brainard, S. Lael (1997), “An Empirical Assessment of the Proximity-Concentration Trade-off Between Multinational Sales and Trade”, American Economic Review 87(4): 520-544.
Keller, Wolfgang, and Stephen Yeaple (2009) “Global Production and Trade in the Knowledge Economy”, CEPR Discussion Paper 7175.
Levy, Frank, and Richard Murnane (2004), “How Computerized Work and Globalisation Shape Human Skill Demands”, working paper, MIT.
Pew Center (2006), “Public Says American Work Life Is Worsening, But Most Workers Remain Satisfied with Their Jobs”, Pew Research Center, Labour Day 2006.