The Human Cost of Trade

The United States emerged from World War II as an unrivaled economic and industrial superpower, with roughly 50 percent of the world's manufacturing capacity and 80 percent of its hard currency reserves. But if global competition, particularly from Japan in the second half of the 20th century, challenged US dominance, the emergence of China in the 21st century ended it: today China accounts for about 32 percent of global manufacturing; the US, just 15 percent. 

Image

MIT economist David Autor, PhD ’99

The MIT Professor David Autor, PhD ’99, in collaboration with the University of Zurich's David Dorn and Harvard's Gordon Hanson, call this economic earthquake the "China Shock." At first, and for many years, mainstream economists were mostly sanguine about China's impact on the global economy and about globalization in general. Workers would move from manufacturing to service sector jobs. They would retrain and reeducate themselves for higher-paying positions in new industries. In the meantime, their living standards would improve thanks to lower-cost goods from overseas. 

Today, Autor says that many economists were asking the wrong questions about the China Shock. Instead of looking at the aggregate costs and benefits of globalization, they needed to look at the places where the impact would be heavily concentrated. In a recent conversation with Harvard Griffin GSAS Communications, Autor draws from updated research with Dorn and Hanson to describe what happened to rust-belt communities where manufacturing jobs were lost, what was missing from the economic models of decades past, and how the China Shock continues to resonate in social unrest and politics. 

Immediately after World War II, the US economy accounted for something like half of the world’s gross domestic product. What happened between then and 2001, when China joined the World Trade Organization (WTO)?  

US manufacturing has been in decline as a share of employment since the early 1940s as other sectors have grown. But numerically, it was actually pretty stable. There were nearly 20 million US manufacturing workers in 1979. There were 18 million in 1999. And then it declined by about 4.5 million in the next seven years. So, the US was losing a lot of labor-intensive sectors. It lost ground in electronics, in automobiles, and so on.  

So, what happened when China joined the WTO?  

The China trade shock was different, both in the scale and the speed. The intensity was just unlike anything we've ever seen. And it was a combination of change in policy and currency manipulation by China, but that wasn't the biggest part of it. It was this incredible productive capacity for low-priced goods, and the US had no policies in place to ease the adjustment. 

The impact was very geographically concentrated in labor-intensive manufacturing in the South Atlantic and the deep South. On the scale of US employment, a few million jobs is not that big a deal, right? We have a country of more than 150 million workers. But the shock was so localized, just occurring in a handful of counties. And it didn’t eliminate a few jobs there; it took out the entire economic foundations of these places. No longer could you be making commodity furniture, or textiles, or cheap auto parts, or building materials. Those products were no longer viable. 

[The China Shock] was so localized, just occurring in a handful of counties. And it didn’t eliminate a few jobs there; it took out the entire economic foundations of these places.  

Let’s talk about the China side. How did the country go from crushing poverty and economic depredation during the time of Mao Zedong, to global manufacturing powerhouse? 

In the early 1980s, after Deng Xiaoping gained power following Mao’s death, the country started to reform and open to global trade. In the early 1990s, we start to see that having an effect. China allows foreign investment and technology, the price system. It creates these special economic zones. It allows free mobility of population into these zones. This was an enormous change. 

China’s productivity and capacity rose very, very rapidly throughout the 1990s. But starting from such a low base, even fast growth meant that it was initially just inching its way up the curve. And, of course, the US economy was very robustly growing especially because of the growth of the Internet. So China wasn’t really on most Americans’ people’s radar. 

Then, in 2000, China and the United States negotiated permanent normal trade relations. In 2001, China joined the WTO.  That did two things. One, it eliminated all uncertainty about the investment environment in China. Firms could now safely make investments, without fear that they were going to be reversed or shut down by policy. Second, it reduced not only China’s export barriers, but also barriers to imports from other countries. So, suddenly, it could import more cheaply the inputs to make things.  

It also caused a change in China's banking system. The country’s banks had to do a bunch of things to comply, and actually this did them a favor. The new regulations shut down state-owned enterprises and reformed the financial system. So, in many ways, China became a much fiercer competitor. 

And on the US side? 

The US just ran an enormous merchandise trade deficit, equal to a couple percentage points of GDP over that period. It exceeded 2 percentage points of GDP at one point. What that means in real terms is, all of a sudden, we're importing manufacturing goods that we would have been producing domestically. It wasn't just substitution from other countries; it was an enormous growth in imports, and that had to displace some manufacturing employment. I mean, if all of a sudden we start buying all our cars from another country, it means we're not making them here. And most of what we produce, we consume ourselves. This is true for all large countries, even for China. Most of what we produce, we don't produce for export. When you start importing a lot more things, generally what you're crowding out first is your own production. So, on some level, it had to be true that some of this was going to reduce US manufacturing employment. No one knew quantitatively how much.  

Right. Given all that, how did many mainstream economists view the China Shock? 

So, there's a story that says, okay, well, yeah, we move out of manufacturing, move into services, and yeah, it's a little bit of reduced demand for production workers, but they can do something else, and it's a big country, and it’s only a few million jobs, so not a big deal. 

And what was the truth? 

What actually turned out to be true is that we had a fall in employment in trade-impacted local labor markets and a rise in unemployment and non-employment. People didn't move fluidly across sectors. Wages fell, and lots of other social maladies occurred simultaneously, including more recent evidence of, for example, mortality effects. 

Why do you think economists misunderstood the China Shock so dramatically?  

Economists have understood for decades that trade has concentrated costs and diffuse benefits. The costs are people who suddenly lose careers or whose skill sets become devalued. The benefits are lower prices. 

In the case of the China Shock, the benefits were experienced by hundreds of millions of people—everyone who shopped at Walmart, for example. The costs were experienced by only a few million people, but their losses were in no sense commensurate with the gains they got from lower prices. You can’t be like, oh, sure, I have no job anymore, but I can get a big TV. 

Now, that doesn't mean that the aggregate benefits are negative, per se. My work with Dorn and Hanson is not a full welfare analysis. But the concentrated cost turned out to be much, much greater than people had anticipated. They had a notion that labor markets were much more fluid than they were, that people would adjust much more quickly, that the other jobs were closer substitutes, that as long as the unemployment rate was not high, that everyone would get re-employed. Economists didn't anticipate how scarring this would be for those places. 

[Economists] had a notion that labor markets were much more fluid than they were, that people would adjust much more quickly, that the other jobs were closer substitutes, that as long as the unemployment rate was not high, that everyone would get re-employed. [They] didn't anticipate how scarring this would be for those places. 

So how is your work different? 

What we did in our China Shock paper in 2016, which was really building on papers that we published in 2013 and 2013 in the American Economic Review and Quarterly Journal of Economics, was to take a geographic approach instead of looking at the national prices of goods and then the aggregate price of skill groups. That's not going to be precise. Because everything is visible in prices, but not in employment levels, and everything disperses nationally. And it turns out most of the effects of the China Shock were not in wages, but in employment. It's not that people took a wage cut, they just lost jobs. And it's not that it disperses nationally, it occurs in a very geographically concentrated way. 

Once you look at the level of the geography in which this occurred, as opposed to nationally, you see these very intense job losses, rise in transfer payments, and increase in other adverse outcomes. This is something that [Harvard Sociology Professor] William Julius Wilson wrote about decades earlier, how the decline of inner-city manufacturing damaged the Black community, and especially Black men. A lot of what we discovered really just recapitulated that for other areas. 

How did you update this research for your 2025 paper, “Places Versus People”?  

The 2013 paper was based on repeated cross-sections from the census of population. We could see how employment changed, but we weren’t following individuals longitudinally. It was also based on a 5 percent sample. 

The data we use in “Places versus People,” allows us to look at individual earnings records from the unemployment insurance system matched to tax records. This enables us to follow people longitudinally over decades to see where they started work, what happened to their earnings and employment, and whether they moved or changed sectors. It let us form a panel of people to look at where they were at the time of impact and what happened next, as well as a panel of places to see how they evolved. We can ask: Did employment fall because people moved out, or because they stayed and were no longer working? How much of the change is due to new people entering at lower wages versus high-wage people leaving? 

There is a direct mapping between what happens to the people and what happens to the places, but it has many moving parts—particularly, people moving in and out. What we show is that these places have now substantially reconstituted. They are growing relatively rapidly; they have many new young people, US-born Hispanics, and foreign-born college graduates moving in. While manufacturing has continued to decline steeply, these areas have more employment now in schools, medical services, big-box stores, and transportation. However, the wage structure is lower. 

You can view that as a type of economic dynamism, but the individuals who were in manufacturing at that point did not move out. Many of them "rode it down." Those who didn't lose work stayed in manufacturing with diminished prospects and lower wage growth. The world changed around them as their conditions eroded. 

If the US wants to hold onto what is left of its industrial capacity, it requires significant investment. Tariffs are only one small part of that. You cannot win a race by continually throwing sand at the feet of your competitors and hoping they trip; eventually, you need to bulk up and start running. 

You’re describing the concentration of these effects. Why didn’t it play out the way many economists thought it would? Why didn’t people move from the sectors where the impact was so concentrated? Why didn’t they move to places with other jobs or into growing sectors like services? 

I don’t think we should caricature all economists as believing the same thing. People had a variety of views and many were surprised, but not everyone thought the outcome was impossible. 

That said, there are several factors. First, labor economists have understood for a long time that losing career jobs is extremely damaging. People lose 10, 15, or 20 percent of their prior earnings for decades. Many don’t re-enter the labor force, and those who do often enter at lower wages. Involuntary job loss is very scarring. That was well-known to labor economists, if not to trade economists. 

Why was it so damaging? Manufacturing provided high-wage jobs to adults with lower levels of education. There wasn't an alternative job that was equally good. These were full-time, steady positions with specialized skills that were not applicable elsewhere. If you were working in a sawmill, those skills don't translate to becoming a Walmart cashier. 

Furthermore, there wasn't another "land of opportunity" for these people to go to. There wasn’t another sawmill in Pennsylvania waiting for them. Manufacturing is geographically concentrated—steel is made here, furniture is made there, textiles are over there. The world was not waiting for these people with open arms. There wasn’t an equally good alternative they could have moved to. 

More broadly, people generally do not make major occupational transitions in mid-adulthood. Most occupational change occurs at the margin of entry and exit: young people enter growing industries and older people retire from contracting ones. That is the natural rhythm of the labor market. It doesn't change rapidly because most changes occur across careers, not within them. 

Finally, how do we start to get out of this mess? Is it tariffs? Is it unions? Should we be trying to build up our industrial base? 

Gordon Hanson, and I wrote a New York Times op-ed about this last July. You cannot reverse the tide. If the US wants to hold onto what is left of its industrial capacity, it requires significant investment. Tariffs are only one small part of that. You cannot win a race by continually throwing sand at the feet of your competitors and hoping they trip; eventually, you need to bulk up and start running. 

For the US, this means targeted industrial policy, prioritizing certain sectors, and recognizing that we are now a small country compared to China in the scheme of world manufacturing. We need allies and coherent policies. It is not just about jobs anymore; it is extremely militarily and economically risky for the US to have so much of its industrial base under threat. This includes semiconductors. If China were to invade Taiwan, for example, it would be devastating because the vast majority of our advanced semiconductors come from TSMC in Taiwan. Our auto sector is no longer competitive in many areas. There are many fields of advanced technology where China is not only ahead of the United States, but is the sole provider. We are in a grave situation.