How much is international trade costing Americans jobs? It turns out, not much.
A major objection by protectionists is that international trade with other countries, especially low-wage countries, will undermine American workers. As a consequence, there will be a “giant sucking sound” (to use Ross Perot’s phrasing) as jobs are pulled from the United States to these countries.
From a purely theoretical perspective, there is little reason to think this will happen. Wages are determined by the marginal productivity of workers. Workers across countries are not identical, which is why firms do not necessarily pursue low wage workers. Workers in other countries earn less than Americans do not simply because they’ll accept a lower standard of living (and hence, lower wages) but because they are less productive on the margin than American workers are. The underlying reason is the same one that explains why the New England Patriots opted not to draft me (a 37-year-old who hasn’t played any football, ever) and pay me a few thousand dollars, but instead draft a 23-year-old UNC quarterback, whom they pay millions of dollars.
Indeed, according to mainstream trade theory, there will only be such a significant offshoring of jobs that wages are expected to decline under very precise circumstances. Specifically: when there are identical levels of productivity, identical technology, and one country produces labor-intensive goods while the other produces capital-intensive goods. (Under these conditions, we’re more likely to hear a “giant sucking sound” of jobs going to Canada rather than to low-wage workers in Mexico or China.) To the extent that trade does affect jobs, we should expect minimal effects on aggregate employment levels as workers are reallocated, some jobs are lost while others are created.
Standard theory gives us good reason to doubt protectionist claims. But the protectionists present an alternative theory. We can test the empirical claims of the protectionists to see which theory better explains the empirical evidence.
First, consider job security. One measure of job security, developed by University of Chicago economist Steven J. Davis, is initial weekly jobless claims as a percentage of the total employed population. Jobless claims are claims made by those who separated from their jobs unwillingly and through no fault of their own—in other words, they were not fired for incompetence but rather a downturn in business. Under this measure, a high percentage suggests that workers are less secure in their jobs, while a lower percentage suggests that workers are more secure in their jobs. The graph below shows this percentage from 1967–2019 (I stopped the graph pre-COVID because the data got weird during the pandemic, but then returned to normal once lockdowns were lifted).
Two things are obvious:
First, from the 1980s to present (with recessions excluded) the average percentage has been trending downward—suggesting that American workers are generally becoming more secure in their jobs. Even in the 2008 recession, the percentage was lower than during recessions in the 1970s and 80s.
Second, major international trade deals, such as NAFTA (1994) and China joining the WTO (2000) did not affect job security levels in the aggregate. The general downward trend that starts in the 1980s is uninterrupted (except for recessions). If these trade deals really led to the devastation of American workers, as asserted by protectionists, we should expect to see some indication of that devastation in the data. But we do not. The increases we do see are tied to recessions, not trade policies.
“But wait!” I hear some protectionists cry. “Of course it wouldn’t show up in the aggregate. People get laid off from manufacturing jobs but become more secure in their minimum wage jobs. That’s what we really object to!” Fair enough, but we see the same trend in manufacturing.
This second graph shows data for American manufacturing jobs. It shows layoffs and discharges in manufacturing as a percentage of total employment in manufacturing. The time frame shown on this graph is shorter because the Bureau of Labor Statistics only started collecting this data in 2000.
Again, recessions aside, we see that manufacturing workers become more secure in their jobs over the nearly two decades covered. In fact, this chart works especially hard against the protectionists’ claims. Manufacturing job security actually became stronger post-China Shock than pre-China Shock. It seems to have moved little after NAFTA.
This, too, may fail to respond to protectionists’ concerns. They may reply, “Again, you’re looking at too broad a scope. It is the regional effect that matters. Look at the devastation of the Rust Belt! There’s all the evidence you need of the horrors of international trade!”
Regional impact is a legitimate concern. As Douglas Irwin documents in his excellent 2017 book Clashing Over Commerce, U.S. trade policy and concerns have long been regional in nature. But alas, the Rust Belt does not support the protectionist story either. A 2023 paper by Simeon Alder, David Lagakos, and Lee Ohanian in the Journal of Political Economy finds “Labor conflict accounts for half of the decline in the region’s share of manufacturing employment. Foreign competition plays a smaller role, and its effects are concentrated after most of the region’s decline had already occurred.” In other words, they find that labor market factors, not international trade, are the primary cause for the Rust Belt’s struggles.
To refute every argument against international trade would require a book. Adam Smith’s Wealth of Nations is a good starting point; he refuted protectionist arguments of his day, and protectionist arguments have not changed since the 1700s. See also Douglas Irwin’s Free Trade Under Fire. The evidence indicates that international trade has not been the “gutting” factor that many protectionists claim. More can be explained by the theory that macroeconomic factors primarily determine the number of jobs a country holds, not international trade. This is not to say that international trade has no effect. Some jobs were lost, others were created by international trade. But the net effect on jobs has been close to zero.
Readers will notice I do not address the protectionist argument that trade has “deindustralized” the United States. That omission is intentional. The reason is that there are two meanings of “deindustrialization.” Depending on which definition one uses, the claim is correct or incorrect. I call the two meanings absolute deindustrialization and relative deindustrialization.
Absolute deindustrialization is when manufacturing output is falling in absolute terms. This is what most politicians and political activists against free trade tend to mean when they talk about deindustrialization. Under this meaning, trade hasn’t deindustrialized the United States. According to the U.S. Federal Reserve’s monthly report on Industrial Production, U.S. manufacturing output is below its 2007 record high, but remains much higher than before China joined the WTO or NAFTA. Manufacturing has plateaued, but it is not declining.
Relative industrialization is when manufacturing output relative to other nations begins to fall. Trade economist Richard Baldwin uses deindustrialization in this sense (see, for example, this blog post). In the relative sense, deindustrialization is occurring due to trade, but the reason is that the United States is moving to a more service-based economy. We continue to produce manufactured goods (and high-valued goods, at that. In the first quarter of 2026, U.S. manufacturing produced and sold goods at a level consistent with a seasonally-adjusted annual rate of $7.5 trillion nominal dollars), but we produced a relatively larger share of tech, medical, finance, education, and the like.
America’s comparative advantage is in knowledge-based goods, rather than manufacturing. Consequently, our economy is one that takes advantage of our relatively high level of skills and knowledge. Indeed, even our manufacturing output is in goods that require high levels of skill to produce. One last implication is that re-shoring manufacturing will be costly, since a skills-mismatch exists.
