The idea that labor markets are increasingly internationally interconnected (i.e. global) has serious purchase on the public imagination. It also has implications for policy. To a government whose primary objective is to compete for jobs, increasingly interconnected labor markets strengthen arguments for devaluing the national currency, for reducing workers’ ability to collectively bargain for higher wages, and for teaching schoolchildren to turn a jack rather than analyze history. Growing interconnections also reduce the net benefits of fiscally stimulating your home economy in the face of a recession, unless you can convince your trading partners to do so too. Anybody familiar with economic policy discussions since the 1990s should not need convincing that these arguments are frequently made. Because these policy shifts all carry large costs, one has to ask whether the idea of an increasingly global labor market is empirically sound.
 
But what does one really mean when claiming that national labor markets have become increasingly interconnected? And what do (or can) we know about whether this has happened? Liming Chen, Jesus Felipe, Andrew Kam and I tackle these questions in a recent working paper. Our conceptual conclusion is that commentators pushing the global labor market meme are implicitly invoking at least three different notions of international labor market connectivity, and that each of these notions matters for deciding whether to pull a different policy lever. Moreover, our empirical findings with respect to the three notions of interconnection, which we measure in 40-68 countries, are completely inconsistent with the idea that labor markets globalized across the board.¹  Many countries globalized in one dimension but deglobalized in one or two others, with the dimensions registering increases and decreases differing across countries. And in many cases, it is impossible to know whether they have globalized in a particular dimension or not.

Three policy debates, and three dimensions of a “global” labor market

1. Export Induced Labor Demand (EILD)
A global labor market is often thought to boost the case for international coordination of stimulus policies when recessions hit. As a country becomes increasingly reliant on exports to create demand for its labor, its incentives to go into debt to stimulate its economy diminish—far simpler to let its trading partners do it for them. To assess the relevance of this argument, we ask whether exports generate a growing fraction of the demand for each country’s labor.  We label this share Export-Induced Labor Demand (EILD). 
 
Figure 1 plots EILD in 2011 and 1995 for 40 countries. If a country lies above the diagonal line in Figure 1, EILD and its employment-related incentives to free-ride grew, and vice versa. Clearly, there is not strong evidence, outside Europe and South Korea, of high or fast growing EILD. The other non-European countries with the most systemically important labor forces (China, India, The United States, Brazil, Japan, Canada, and Russia) are notable for the very small and roughly stagnant level of EILD. For example, only 8% of US employment exists to meet demand for US exports, a fact that may help to explain why pro-trade arguments have failed to register on the hustings. This suggests that—outside Europe and South Korea—concerns over growing incentives to free-ride on other countries’ stimulus policies are somewhat exaggerated.

Figure 2 splits the change in each country’s EILD share over time (the blue dot) into two components. First, as one would expect, most industries became more export-intensive as trade ramped up—an effect captured as the “within industry” shift, which usually increased EILD. Second, as output per worker increased much faster in more export intensive industries, labor shifted to less export-oriented sectors, typically reducing EILD—the “between-industry” shift. It is the balance of these two forces that determines whether an economy became more reliant on exports for labor demand. This might help to explain why exaggerated claims about the importance of stimulus coordination have become lore—industry experts may be extrapolating from what they see within their industries, while forgetting that employment in the most export-dependent industries has been declining.

2. Employment in Tradable Industries
Governments must also consider how “globalized” their national labor markets are when confronting a balance of payments crisis. One option that countries running low on foreign reserves consider is a currency devaluation. Devaluations tend to make imports unattractive but exports attractive. This, in turn, reduces the trade deficit, slows reserve depletion, and creates jobs in those sectors of the economy that produce products that are either exported or compete against imports. However, devaluations also increase the price of tradable products, which tend to be set in dollars in world markets, relative to the earnings of workers who sell non-tradable services—whose prices are set in local currency in the local labor market. Thus workers in non-tradable sectors are hurt by devaluations, often badly. If the share of workers who produce internationally tradable goods and services—the “tradable employment” share—rises, then, other things equal, devaluation becomes a relatively attractive means of achieving balance.
 
Tradable employment is a totally different measure from IELD, and is also a lot harder to peg empirically. What is tradable at any point in time depends on two factors: whether it is physically possible to produce it far from consumers (i.e. whether the product is intrinsically tradable), and, if so, whether governments would allow it to be transported across borders (i.e. whether it is also de facto tradable). All goods, except a few with unusual physical properties (e.g. cement and ice), have always been intrinsically tradable, but trade liberalization made them gradually more de facto tradable. Technological change has rendered several services more tradable over time as well.  There are no good measures of how tradable different goods and services are over time.

To get around this, our paper puts bounds on tradable employment shares over time under extreme assumptions about what is and isn’t intrinsically tradable,² and then exploit the fact that all de-facto tradable products are intrinsically tradable. We conclude that the fraction of workers that produce intrinsically tradable products declined precipitously in most developing countries in recent decades, as workers were squeezed out of agriculture and into non-tradable services. This trend was not as acute in advanced economies, but we nevertheless find zero countries (out of 68) in which we can show that employment producing intrinsically tradable products increased. We also argue that in the future, the share of workers that produces de facto tradables is likely to lie in the range of 15-40%—a far cry from a totally globalized labor market.
 
We therefore conclude that in this dimension, particularly in developing countries, employment has de-globalized. What people do for a living has shifted in a way that makes currency devaluations less, not more attractive, relative to other ways of dealing with payments imbalances.
 
3. Trade-linked employment
We developed our final measure, trade-linked employment, to check whether labor market globalization does indeed call for pro-competitive reforms to human resource policies (labor protections and education). Trade-linked employment includes not only workers producing tradables (e.g. the workers who produce toys for export, or oil that could have been imported), but also workers who produce the non-tradable services required to produce tradables (e.g. truck drivers who serve the toy and oil sectors). When these workers are well paid or lack skills needed to produce cheap toys and oil, the toy and oil industries—and by extension, the economy—become less competitive. Thus, if globalization increases the fraction of employment that is trade linked, this strengthens arguments for repressing wages and for turning schools into vocational training centers.

Trade linked employment is the most difficult of our three notions to peg empirically. Indeed, the assumptions needed to do so pile up so quickly, that the best we could do was to report on a very wide range of levels for each country and year within which we feel comfortable the true estimate must lie. These bounds are so large that for 30 out of our 40 countries we were unable to reach any qualitative conclusion as to whether the true number had increased or decreased. For the other 10 countries, we have a reasonable suspicion that it decreased. While inconclusive results are dissatisfying, they tell us something very important. If, after careful examination with the best data we could find, we cannot determine that trade-linked employment has increased, then there is little reason to think that governments know that this has happened either. There does not appear to be a globalization-related case for making labor market institutions more flexible in all industries, or for wholesale education reforms to promote production skills over (say) civic sense or cultural education.

Conclusion

As empirically conservative researchers, we therefore urge policy-makers to proceed cautiously. Think carefully about policy proposals that are forged in response to the ostensibly globalized labor market; understand what measure of interconnection informs your policy decisions; measure it carefully in your labor force; and then proceed in a fashion that takes the limits of what you know seriously. Make adjustments to policies affecting those sectors of the economy that have clearly globalized, but be careful about extending those policies to sectors that have not. Work with a scalpel, not a broom.
 
Uncritically accepting the global labor market meme often results in policies that have the potential to increase economic inequality. Yet our results imply that some governments have greater latitude to combat inequality than the “globalization-everywhere” discourse suggests. We recommend they use it.