3 The Pains from Trade

IN EARLY MARCH 2018, President Trump signed new tariffs on steel and aluminum, surrounded by steelworkers in their hard hats. Shortly after, the IGM Booth panel, which we talked about in the introduction, asked its roster of experts, all senior economics professors at top economics departments, Republicans and Democrats, whether “imposing new US tariffs on steel and aluminum will improve Americans’ welfare.” Sixty-five percent “strongly” disagreed with the statement. All the others merely “disagreed.” No one agreed. No one was even unsure.1 When asked the additional question of whether “adding new or higher import duties on products such as air conditioners, cars, and cookies (to encourage producers to make them in the US) would be a good idea,” once again all of them agreed it would not be.2 Paul Krugman, the standard-bearer of liberal economics, likes trade but so does Greg Mankiw, a Harvard professor who headed the Council of Economic Advisors under President George W. Bush and a frequent critic of Krugman’s views.

In contrast, in the United States the general public opinion about trade is mixed at best, and more often than not these days, negative. On the steel and aluminum tariffs, opinions were split. In a survey conducted during the fall of 2018 where we asked a representative sample of Americans exactly the same question as in the IGM Booth panel, only 37 percent of people either disagreed or strongly disagreed with Trump’s proposal to increase tariffs. Thirty-three percent agreed.3 But, more generally, the sentiment seems to be, both on the right and on the left, that the United States is too open to goods from other countries. Fifty-four percent of our respondents agreed that using higher tariffs to encourage producers to produce in the US would be a good idea. Only 25 percent disagreed.

Economists mostly talk about the gains of trade. The idea that free trade is beneficial is one of the oldest propositions in modern economics. As the English stockbroker and member of Parliament David Ricardo explained two centuries ago, since trade allows each country to specialize in what it does best, total income ought to go up everywhere when there is trade, and as a result the gains to winners from trade must exceed the losses to losers. The last two hundred years have given us a chance to refine this theory, but it is a rare economist who fails to be compelled by its essential logic. Indeed, it is so rooted in our culture that we sometimes forget the case for free trade is by no means self-evident.

For one, the general public is certainly not convinced. They are not blind to the gains of trade, but they also see the pains. They do see the advantages of being able to buy cheap abroad, but worry that, at least for the direct victims of cheaper imports, the gains are swamped by the costs. In our survey, 42 percent of respondents thought low-skilled workers are hurt when the United States trades with China (21 percent thought they are helped), and only 30 percent thought everyone is helped by the fall in prices (27 percent said they thought everyone was hurt).4

So is the public simply ignorant, or might it have intuited something the economists have missed?

Stanislas Ulam was a Polish mathematician and physicist, one of the co-inventors of modern thermonuclear weapons. He had a low opinion of economics, perhaps because he underestimated economists’ capacity to blow up the world, albeit in their own way. Ulam challenged Paul Samuelson, our late colleague and one of the great names in twentieth-century economics, to “name me one proposition in all of the social sciences which is both true and non-trivial.”5 Samuelson came back with the idea of comparative advantage, the central idea in trade theory. “That this idea is logically true need not be argued before a mathematician; that it is not trivial is attested by the thousands of important and intelligent men who have never been able to grasp the doctrine for themselves or to believe it after it was explained to them.”6

Comparative advantage is the idea that countries should do what they are relatively best at doing. To understand how powerful the concept it, it is useful to contrast it to absolute advantage. Absolute advantage is simple. Grapes don’t grow in Scotland, and France does not have the peaty soil ideal for making scotch. Therefore, it makes sense that France should export wine to Scotland, and Scotland should export whisky to France. Where it gets confusing is when one country, like China today, looks like it’s pretty much better at producing everything than most other countries. Wouldn’t China simply swamp all markets with its products, leaving other countries with nothing to show for themselves?

David Ricardo argued in 1817 that even if China (or in his era, Portugal) was more productive at everything, it could not possibly sell everything, because then the buyer country would sell nothing and would have no money to buy anything from China or anywhere else.7 This implied that not all industries in nineteenth-century England would shrink if there was free trade. It was then evident that if any industries in England were to shrink because of international trade, it should be the least productive ones.

Based on this argument, Ricardo concluded that even if Portugal was more productive than England at producing both wine and cloth, once trade between them opened up, they would nonetheless end up specializing in the product for which they had a comparative advantage (meaning where their productivity was high relative to their productivity in the other sector: wine for Portugal, cloth for England). And the fact that both countries make the goods they are relatively good at making and buy the rest (instead of wasting resources producing a product ineptly) must add to the gross national product (GNP), the total value of goods people in each country can consume.

Ricardo’s insight underlines why there is no way to think of trade without thinking about all the markets together. China could win in any single market and yet there is no way for it to win in every market.

Of course, the fact that GNP goes up (both in England and in Portugal) does not mean there are no losers. In fact, one of Paul Samuelson’s most famous papers purports to tell us exactly who the losers are. Ricardo’s entire discussion assumed production required only labor, and all workers were identical, so when the economy became richer everyone benefitted. Once there is capital as well as labor, things are not that simple. In a paper published in 1941, when he was just twenty-five, Samuelson set out the ideas that remain the basis of how we are taught to think about international trade.8 The logic, once you understand it, as is often the case with the best insights, is compellingly simple.

Some goods require relatively more labor than others to produce and relatively less capital; think of handmade carpets versus robot-made cars. If two countries have access to the same technologies of production for both goods, it should be obvious the country relatively abundant in labor will have comparative advantage in producing the labor-intensive product.

We would therefore expect a labor-rich country to specialize in labor-intensive products and move out of capital-intensive ones. This should raise the demand for labor compared to when there was no trade (or more restricted trade), and therefore wages. And, conversely, in a relatively capital-abundant country, we should expect instead that the price of capital goes up (and wages go down) when it trades with a more labor-abundant partner.

Since labor-abundant countries tend to be poor, and laborers are usually poorer than their employers, this implies freeing trade should help the poor in the poorer countries, and inequality should fall. The opposite would be true in rich countries. So opening trade between the United States and China should hurt US workers’ wages (and benefit Chinese workers).

That does not mean the workers in the United States must necessarily end up worse off. This is because, as Samuelson showed in a later paper, the fact that free trade raises GNP means there is more to go around for everybody, and therefore even workers in the United States can be made better off if society taxes the winners from free trade and distributes that money to the losers.9 The problem is that this is a big “if,” which leaves workers at the mercy of the political process.

The Stolper-Samuelson theorem (as this result is now widely known in economics, after Samuelson and his co-author, Stolper) is beautiful, at least as much as any theoretical result in economics is beautiful. But is it true? The theory has two clear and encouraging implications, and one that is less encouraging. Opening up to trade should increase GNP in all countries, and in poor countries inequality should go down; however, in rich countries, inequality can go up (at least before any redistribution the government might undertake). The slight problem is that the evidence more often than not refuses to cooperate.

China and India are often portrayed as the poster children for trade-fueled growth in GNP. China opened up its markets to trade in 1978, after thirty years of communism. For most of those thirty years, China barely acknowledged the world market. Forty years later, it is the world’s exporting powerhouse, about to seize the position of the world’s biggest economy from the United States.

India’s story is less dramatic, but perhaps a better example. For about forty years, until 1991, its government controlled what it called the “commanding heights of the economy.” Imports required licenses that were at best grudgingly granted and in addition required the importer to pay import duties that could quadruple the price of the imports.

Among the things essentially impossible to import were cars. Foreign visitors to India would write about the “cute” Ambassador, a barely updated replica of the 1956 model of the Morris Oxford, a British sedan of no particular distinction, that was still the most popular car on Indian roads. Seat belts and crumple zones were entirely unknown. Abhijit can still remember his one ride in a 1936 Mercedes-Benz (this must have been in 1975 or thereabouts), and the sense of exhilaration from being in a car with a genuinely powerful engine.

Nineteen ninety-one was the year after Saddam Hussein’s invasion of Kuwait that eventually led to the First Gulf War. This resulted in the interruption of oil flows out of Iraq and the Gulf, and sent oil prices through the ceiling. It delivered a huge shock to India’s oil import bill. Coming at the same time as the war-driven exodus of Indian émigrés from the Middle East, who therefore ceased sending money to their loved ones at home, the country experienced a massive foreign exchange shortage.

India was forced to seek help from the International Monetary Fund (IMF), an opportunity the IMF was waiting for. China, the USSR, Eastern Europe, Mexico, and Brazil, among others, had begun to take serious steps toward letting markets decide who should produce what. India at the time was the last of the big holdouts, an economy that continued to adhere to the anti-market ideology fashionable in the 1940s and 1950s.

The deal the IMF offered would change all that. India could have the funds it needed, but only if it opened its economy to trade. The government had no choice. The import and export licensing regime was abolished, and import duties came down very quickly from an average of nearly 90 percent to something closer to 35 percent, in part because many of the leading figures in the economic ministries had long desired a chance to do something like this, and they were not going to let the opportunity pass.11

There were, unsurprisingly, many who predicted this would lead to disaster. Indian industry, raised behind high tariff walls, was too inefficient to compete with the rest of the world’s powerhouses. The Indian consumer, starved of imports, would go on a binge and bankrupt the economy. And so on.

Remarkably, the dog hardly barked. After a sharp drop in 1991, by 1992 GDP growth was back at its 1985–1990 trend of about 5.9 percent per year.12 The economy did not collapse, nor did it dramatically take off. Overall, during the period 1992–2004, growth inched up to 6 percent and then jumped to 7.5 percent in the mid-2000s, where it has remained, more or less, ever since.

So should India be counted as a shining example of the wisdom of trade theory, or something closer to the opposite? On the one hand, that growth weathered the transition smoothly, echoing the predictions of trade optimists. On the other hand, that growth took more than a decade to accelerate after 1991 seems disappointing.13

This particular debate has no real resolution. There is only one India with its one history. How would anyone know whether pre-1991 growth would have continued had there been no crisis and the trade barriers not been brought down in 1991? To complicate matters, trade was being liberalized gradually starting in the 1980s; 1991 just sped that up (a lot). Was the big bang necessary? We will never know unless we are allowed to rewind history and let it go down the other path.

Unsurprisingly, however, economists find it very hard to let go of this sort of question. The issue is less about India per se. There is no way around the fact that there was a large shift in Indian growth, at some point in the 1980s or 1990s, associated with the move from socialism (of sorts) to capitalism. The growth rate before the mid-1980s was around 4 percent. Now it is closer to 8 percent.15 Such changes are rare and what is especially rare is that the change seems to have been sustained.

At the same time, inequality increased dramatically.16 Something very similar, if perhaps even more dramatic, happened in China in 1979, in Korea in the early 1960s, and in Vietnam in the 1990s. It is clear that the kind of extreme state control these economies operated under before liberalization was very effective at keeping inequality down, but at a high cost in terms of growth.

Where there is much more disagreement, and therefore more scope for learning, is about the best way to run an economy once a nation gives up extreme government control. How important is it to get rid of the remaining tariff protections India holds on to, which are significant barriers to trade, but nothing like what there was before? Will that further speed up growth? What will happen to inequality? Will the Trump tariffs derail growth entirely in the United States? And will they actually help the people he is purportedly trying to protect?

To answer such questions economists often compare countries. The basic idea is simple: some countries (like India) liberalized trade in 1991 but others more or less like them did not. Which groups grew faster in the years immediately after 1991, in absolute terms or perhaps relative to their pre-1991 growth rates? Those who liberalized, those who had always been open, or those who stayed closed all along?

There is a voluminous literature on this question, perhaps not surprisingly given the importance of free trade among economists and its popularity in the business press. The answers run the gamut from very positive assessments of the effect of trade on GDP to much more skeptical positions, though it must be said that there is little or no evidence for strongly negative effects.

The skepticism comes from three distinct sources. First, reverse causality. The fact that India liberalized trade, whereas another similar country did not, might reflect that India was ready for the transition, and would have grown faster than its comparator even without the change in trade policy. In other words, was it growth (or the potential for growth) that caused trade liberalization, and not the other way around?

Second, omitted causal factors. Liberalization in India was part of a much bigger set of changes. Among them was the fact that the government essentially stopped trying to tell business owners what they should produce and where. There was also a more nebulous but perhaps equally important shift in the attitude of the bureaucracy and the political system toward the business sector: the idea that business was a legitimate pursuit of honest people, something that could even be “cool.” It is essentially impossible to separate the effects of all these changes from that of trade liberalization.

Third, it is hard to know what in the data constitutes trade liberalization. When the tariffs are 350 percent, there are no imports, so cutting them quite a bit might change very little. How do we distinguish relevant policy changes from irrelevant posturing? Moreover, such sky-high taxes invited defiance; people found creative ways to get around them. In response, the governments would often set up arcane rules to trap violators. A lot of these things changed when the country liberalized, but different bits changed at different speeds in different countries. How do we decide which country liberalized more, given that different countries chose different reforms?

All these are issues that make cross-country comparisons particularly fraught. The reason why different researchers get different answers about the effect of trade policy on growth has a lot to do with the different choices they make on each of these issues—how to measure changes in trade policy and which of the many possible sources of confusion about causality one is willing to tolerate.

For this reason, it is very hard to have a lot of faith in the results. There are always going to be a million ways to do cross-country comparisons, depending on exactly which brave assumption one is willing to swallow.

The same constraints get in the way of being able to test the other prediction of the Stolper-Samuelson theory. Does inequality fall in poorer countries when they open up to trade? There are relatively few cross-country studies on this subject, reflecting a pattern we will see again and again. Trade economists have tended to stay away from thinking about how the pie is shared, despite (or perhaps because?) Samuelson’s early warning that, in rich countries at least, trade could come at the expense of the workers.

There are exceptions, but not ones that inspire confidence. A recent research report by two members of the IMF’s staff finds that countries that are close to many other countries, and as a result trade more, tend to be both richer and more equal. They ignore the inconvenient fact that Europe is where there are many small countries that trade a lot with each other, and those countries tend to be both richer and more equal, but probably not primarily because they trade a lot.17

One other reason to be skeptical of this rather optimistic conclusion is that it flies in the face of what we know from a number of individual developing countries. In the last three decades, many low- to middle-income countries have opened up to trade. Strikingly, what happened to their income distribution in the following years has almost always gone in the opposite direction of what the basic Stolper-Samuelson logic would suggest. The wages of the low-skilled workers, who are abundant in these countries (and should therefore have been helped), fell behind relative to those of their higher-skilled or better-educated counterparts.

Between 1985 and 2000, Mexico, Colombia, Brazil, India, Argentina, and Chile all opened up to trade by unilaterally cutting their tariffs across the board. Over the same time period, inequality increased in all those countries, and the timing of these increases seems to connect them to the trade liberalization episodes. For example, between 1985 and 1987, Mexico massively reduced both the coverage of its import quota regime and the average duty on imports. Between 1987 and 1990, blue-collar workers lost 15 percent of their wages, while their white-collar counterparts gained in the same proportion. Other measures of inequality followed suit.18

The same pattern, liberalization followed by an increase in the earnings of skilled workers relative to the unskilled, as well as other measures of inequality, was found in Colombia, Brazil, Argentina, and India. Finally, inequality exploded in China as it gradually opened up starting in the 1980s and eventually joined the World Trade Organization (WTO) in 2001. According to the World Inequality Database team, in 1978 the bottom 50 percent and the top 10 percent of the population both took home the same share of Chinese income (27 percent). The two shares starting diverging in 1978, with the poorest 50 percent taking less and less and the richest 10 percent taking more and more. By 2015, the top 10 percent received 41 percent of Chinese income, while the bottom 50 percent received 15 percent.19

Of course, correlation is not causation. Perhaps globalization per se did not cause the increase in inequality. Trade liberalizations almost never take place in a vacuum; in all these countries, trade reforms were part of a broader reform package. For example, the most drastic trade policy liberalization in Colombia in 1990 and 1991 coincided with changes in labor market regulation meant to substantially increase labor market flexibility. Mexico’s 1985 trade reform took place amid privatization, labor market reform, and deregulation.

As we mentioned, India’s 1991 trade reform was accompanied by the removal of the industrial licensing regime, capital market reforms, and a general shift of power and influence to the private sector. China’s trade liberalization was of course the capstone of the massive economic reform undertaken by Deng Xiaoping, which legitimized private enterprise in an economy where it had been almost forbidden for thirty years.

It is also true that Mexico and other Latin American countries opened up exactly at the time when China was also opening up, and therefore they all faced competition from a more labor-abundant economy. Perhaps that was what hurt the workers in these economies.

Showing anything definitive about trade by just comparing countries is difficult, because both growth and inequality could depend on so many different factors, trade being just one of those ingredients, or indeed an effect rather than a cause. There have, however, been some fascinating within-country studies that do throw a shadow over the Stolper-Samuelson theorem.

Looking at different regions within countries clearly reduces the number of potential things going on at the same time that might obscure the effects of trade; there is usually a single policy regime, a shared history, and common politics, making the comparisons more convincing. The problem is that the central predictions of trade theory, by their very nature, encompass every market and region in the economy, and not just the ones where imports come in or exports take off.

In the Stolper-Samuelson view of the world, there is one unique wage for every worker with the same skills. A worker’s wage does not depend on his sector or region, but only on what he brings to the table. This is because the steelworker in Pennsylvania who loses his job because of foreign competition should move immediately to wherever he can find a job, to Montana or to Missouri, to plating fish or making fisher-plates. After brief transitions, all workers with the same skills will earn the same.

If this were true, then the only legitimate object of comparison for learning about the impact of trade would be the entire economy. We would not learn anything by comparing workers in Pennsylvania with workers in Missouri or Montana because they would all have the same wage.

Rather paradoxically, therefore, if one believes the assumptions of the theory, it is almost impossible to test it, since the only impact one observes is what happens at the country level, and we just demonstrated the many pitfalls of cross-country comparisons and country case studies.

However, as we saw with migration, labor markets tend to be sticky. People do not move even when labor market conditions would suggest they ought to, and as a result wages are not automatically equalized across the economy. There are in effect many economies inside the same country and it is possible to learn a lot by comparing them, as long as the changes in trade policy affecting these subeconomies are not all the same.

One young economist, Petia Topalova, who was a PhD student at MIT at the time, decided to take this idea seriously, and to start from the premise that people may be stuck, both in a place and in a line of trade. In an important paper, she studied what happened in India after the massive trade liberalization of 1991.20 It turned out that even though we think of “India liberalizing,” there were very different changes in trade policy that affected different parts of the country. This is because, even though eventually all the tariffs were brought down to more or less the same level, since some industries were much more protected than others to start with, there were much bigger reductions in tariffs for some industries. Moreover, India has over six hundred districts that differ enormously in the kinds of businesses they are home to. Some are mainly agricultural; others have steel plants or textile factories. Since different industries fared differently, the liberalization led to very different reductions in tariffs in different districts. Topalova constructed, for each Indian district, a measure of how much it was affected by liberalization. For example, if one district mainly produced steel and other industrial manufacturing products, whose tariff dropped from almost 100 percent to about 40 percent, she would say this district was strongly affected by liberalization. If another district just grew cereals and oilseeds, whose tariff essentially did not change, it was almost unaffected.

Using this measure of exposure, she looked at what happened before and after 1991. The national poverty rate dropped rapidly in the 1990s and 2000s, from about 35 percent in 1991 to 15 percent in 2012.21 But, against this rosy backdrop, greater exposure to trade liberalization clearly slowed poverty reduction. Contrary to what the Stolper-Samuelson theory would tell us, the more exposed a particular district was to trade, the slower poverty reduction was in that district. In a subsequent study, Topalova found that the incidence of child labor dropped less in districts more exposed to trade than in the rest of the country.22

The reaction to her findings in the economics profession was surprisingly brutal. Topalova ran into a barrage of very unfriendly comments suggesting she had the wrong answer, even if her methods were correct. How could trade actually increase poverty? The theory tells us trade is good for the poor in poor countries, so her data had to be wrong. Blackballed by the academic elite, Topalova finally took a job at the IMF, which, somewhat paradoxically given the IMF had pushed for the massive liberalization in the first place, was more open-minded about her research than the academic community.

Topalova’s paper was also rejected by the top economic academic journals despite the fact that it eventually inspired a literature dedicated to the debate. There are now many papers applying Topalova’s approach in other contexts and, incidentally, finding the same results in Colombia, Brazil, and, as we will see below, eventually the United States.23 It was only several years later that she got some measure of vindication from academic economists when her findings won the Best Paper Award from the journal in which the paper had been published.

Topalova had always insisted she had no intention of claiming anyone had been hurt by the trade liberalization. Since she was comparing regions within the same country, all she could say was that some areas (those most affected by trade) were less successful in reducing poverty than others. This is entirely consistent with the possibility, which her paper is careful to underline, that the tide of liberalization had lifted all boats, just some more than others. And her work does not imply that inequality increased in India as a whole, just that it went up more in the more trade affected districts. In fact, because the places most touched by liberalization tended to be somewhat richer to start with, the fact that they did not fare particularly well after liberalization, paradoxically, reduced countrywide inequality. In other papers, Topalova and her colleagues demonstrated some clearly positive economy-wide consequences of the Indian trade liberalization. For example, Indian firms, challenged to find new markets, started introducing new products they could now sell abroad. Moreover, the fact that they could import cheaper and better inputs, indeed ones they could not even find in India before, meant they could make new products for the domestic and international markets.24 This increased their productivity and, along with other reforms undertaken by the government in the early 1990s (and some luck with worldwide growth), contributed to the rapid growth of the Indian economy since the 1990s.

Nevertheless, it is easy to see why trade economists felt threatened by Topalova’s paper. The benefits of trade in traditional theory come from the reallocation of resources. The very fact that Topalova finds any difference between more exposed and less exposed districts tells us resources (workers, but also capital) do not move easily, as we noted earlier. If they did, wages everywhere would have been more or less the same. And she is not the only one to find this; a number of other studies also found very little evidence of resource reallocation.25 But once we give up on the idea that people and money will chase opportunities, how do we hold on to our faith that trade is good?

If workers are slow to move across district boundaries, it is plausible they are also slow to move from one kind of job to another. This is entirely consistent with what we know about labor markets. In India, Topalova found the negative effect of trade liberalization on poverty was exacerbated in states where strict labor laws made it very difficult to fire workers and shrink unprofitable firms, allowing profitable ones to take their place.26

There is also a body of solid evidence showing that, at least in developing countries, land does not easily change hands. Capital also tends to be sticky.27 Bankers are slow to cut credit to firms that are not doing well, but also to lend to those firms that are doing well, for the interesting reason that many credit officers, the people who make lending decisions, are terrified of being held responsible for loans that go bad. The easiest way to avoid this is to make no decision; just rubber-stamp whatever decision has been made in the past, by someone else, and let yet another person deal with the loans in the future. The one exception, unfortunately, is when loans are about to fail—then bankers actually give the ailing firms new loans to pay back their old ones, in the hope of postponing the default and perhaps benefitting from a reversal of fortune. This is the phenomenon, in banking parlance, of “evergreening” loans, one of the main reasons why so many banks with seemingly impeccable balance sheets suddenly wake up to a looming disaster. Sticky lending means existing firms that should have been put out of their misery continue to hang on. At the same time, it also means new businesses have a hard time raising capital, especially in the middle of the uncertainty that comes with, say, a trade liberalization, because the loan officers shy away from taking on new risks.

Given these various forms of stickiness, it is plausible that when bad news arrives in the form of greater competition from outside, instead of embracing it and moving resources to their best possible use, there is a tendency to hunker down and hope the problem will go away on its own. Workers are laid off, retiring workers are not replaced, and wages start to drift down. Business owners take a big hit on their profits, loans get renegotiated, all in order to preserve as much as possible of the status quo ex ante. There is no improvement in efficiency, just a fall in the earnings of everyone associated with the industries that lose their protection.

This might seem extreme, but Topalova finds something like this in the Indian data. For one, there was very little migration out of the districts affected by liberalization.28 Even within a region, resources were slow to move among industries.

More strikingly still, this was true within firms. Many firms in India produce more than one product, so one would expect firms to close down product lines competing with cheaper imports and reorient production toward products facing less of a disadvantage. There is nothing to stop this even where labor laws make it hard to fire people, but Topalova’s research found very little “creative destruction.” Firms never seem to discontinue a product line that has become obsolete. Perhaps it is because the managers find the transition process costly: workers need to be retrained, new machines need to be purchased and installed.29

These internal barriers notwithstanding, resources did eventually move (at least in some countries) and exports are a big part of the remarkable success stories of East Asia in particular. Despite what you hear from President Trump and others, it was not because rich countries were naively welcoming. Rich countries heavily regulate imports, which have to meet strict safety, labor quality, and environmental standards.

It has been argued that regulations often serve to keep imports out. California avocado producers successfully lobbied for a complete federal ban on Mexico Hass avocados from 1914 until 1997. This was on the grounds of keeping Mexican pests out, despite the facts that Mexico is territorially adjacent and that pests do not require visas to cross the border. In 1997, the federal ban was lifted but remained in effect in California until 2007. More recently, researchers found that during the 2008 crisis in the United States, the Food and Drug Administration suddenly became more likely to refuse, on food-safety grounds, shipments of imported foods coming from developing countries; for exporters from developing countries, the cost associated with shipments being refused quadrupled during the period! Obviously, the quality of shipments from Mexico could not have changed because of the subprime crisis in the United States, but because demand for avocados went down, it became all the more valuable to keep them out to protect local growers.30 Domestic pressures for protection mount during bad times and safety regulations are often used as an excuse to protect the domestic producers.

That said, some of these standards also reflect genuine consumer preferences for safety (e.g., some Chinese toys have been found to contain lead), the protection of the environment (e.g., pesticide use in agricultural products), or the condition of workers (e.g., child labor). Indeed, the success of the Fairtrade branding shows that many consumers are willing to pay more to intermediaries who can assure them that a product meets some environmental and ethical standards. And, partly inspired by this, many well-known brand names these days impose quality standards over and above any regulatory requirements, making it even harder for new exporting countries to enter.

There is something else quite specific about developing countries trying to be the next China that adds to all these challenges.

The World Trade Organization established an Aid for Trade initiative in 2006, and as of mid-2017, over $300 billion had been disbursed for various programs to help developing countries trade.31 Behind all such initiatives and funding is the belief that trade is a route out of poverty for these countries. A project from Aid to Artisans (ATA), a US-based NGO helping producers of handmade products in developing countries to access international markets, allowed researchers to put that assumption to the test.32

In October 2009, ATA received funding to implement a new program in Egypt. The program followed a standard procedure. First, ATA looked for a suitable product that appealed to high-income markets and was produced in the country relatively cheaply. The research team helped ATA identify the ideal product: carpets. Handmade rugs are an important source of employment in Egypt, and there is demand for them in the US.

Second, ATA had to find a location. They chose Fowa, a town located two hours southeast of Alexandria that is home to hundreds of small firms producing a specific type of rug. A typical firm in Fowa is a one-man (never woman!) operation; the owner operates a single loom out of his home or a shed.

Third, ATA always works through a local intermediary firm with on-the-ground knowledge, which receives the order and finds small-scale producers to manufacture the products. The hope is that ATA will work in the country for some years but then pull out, leaving the intermediary strong enough to keep the project going and growing. A big appeal of the town of Fowa, from this point of view, was the presence of a natural intermediary, Hamis Carpets. Hamis was already marketing many of the carpets produced in the town, although for the most part they were not exported.

Hamis Carpets and ATA then set out to decide what kind of carpets to make, find the buyers, and generate orders. That took a lot of effort. ATA brought the CEO of Hamis to the United States for a training course, hired an Italian consultant to design rug samples, and showcased Hamis’s products in every gift fair and to every importer they knew. Despite all this, it was only after one and a half years of searching for customers that Hamis Carpets secured its first significant export order, from a German buyer.

From this point on, business picked up. Between 2012 and 2014, orders arrived rapidly, and five years after the project had started total orders exceeded $150,000. A US NGO with good contacts and financing, a fearless team of very committed and talented young researchers, and a solid firm with a good domestic reputation took five years to get a decent amount of orders, enough to give sufficient work to occupy thirty-five small firms. Without the external push from ATA, it probably would not have been possible for the local intermediary to make this work.

Why was it so difficult? A large part of the problem seems to be that from the point of view of a foreign buyer (often a large retailer or online store with a brand name), buying from a small carpet manufacturer in Egypt is a gamble. For them quality is critical. Customers expect it; they want flawless carpets. So is timing. If the carpets are not ready for the launch of the new spring collection, the sellers take a big hit. Finally, there is no way to pass the entire risk back to the manufacturer. While it is possible to refuse to pay the manufacturers if quality is low or there is a delay, what the retailer can claw back by returning carpets or refusing to pay is peanuts relative to the reputational loss (think of irate buyers’ web posts about the low quality of products from Wayfair) or the cost of missing the spring collection deadline. In principle, firms can also agree on penal damages (the manufacturer agrees to pay a certain large amount of money for every day of delay, say), but good luck collecting from a small-town Egyptian firm that could vanish overnight. Nor is it feasible for the retailer to check every single carpet to avoid any reputational risk; it would cost way too much in staff time.

Another possibility might be to offer the products so cheaply that consumers were willing to accept the risk of some defects, knowing they could always send the carpet back. Why stake reputation on delivering a product as close to perfection as possible? Why not lower expectations along with prices?

It turns out this does not always work, because in many cases the price cannot go low enough for consumers to waste their time with a product they don’t trust. We once purchased a DVD player in Paris. When it came, we realized the flap through which one puts in the DVD was stuck. After about an hour spent trying to make it work, and another hour looking for technical help on the manufacturer’s site, we went online to chat with a nice Amazon employee who offered a full refund. To get the refund we had to drop the DVD player off at a grocery store near us.

The first time Abhijit went to the grocery store, the shop owner refused to take the player because they had too many Amazon shipments. The second time, the owner made him wait twenty-five minutes before taking the package, because he was getting another consignment of packages at the same time that he needed to log. In the meantime, we bought another DVD player from a different retailer (we were in a rush since we wanted it for our daughter’s birthday). Unfortunately, when it came we realized it would not work with the television in our apartment. We attempted to return it through the product’s website, but since the purchase had not yet been logged as completed, it was not possible until a few days later. At the time of writing, the second DVD player sits, nicely repacked but unreturned, on the table in our entryway. Meanwhile, we gave up on buying a DVD player. Esther’s father lent us one.

Why this long story about our misadventures with a DVD player? It drives the point home that for the ultimate consumer, time is money, as is reliability, and it’s money we will never recover. It is not like Amazon will pay Abhijit his hourly wage for his two trips to the grocery store or the two hours spent trying to fix the machine.

Or think about the pretty T-shirt you bought cheap on some website, which infected the entire wash with its brilliant blue color. Who will compensate you for the $100 blouse that now has blue stains across the front? Or for the time it took you to find that blouse by rummaging through every consignment store in the Village?

This is why Amazon goes to great trouble to maintain its reputation for excellent service. In some cases, for example, they protect the customer’s time by not requiring they return the defective product. For the same reason, Amazon then wants to deal with a producer it can totally trust, ideally a company they have dealt with before, or at least one with a reputation for good products and good service. For both customer and retailer, time is money.

The structure of global inequality is such that the kind of customers in the West who would buy a handmade carpet or a hand-printed T-shirt (labor-intensive products for whose manufacture poor countries have a comparative advantage) are often so much richer than the makers that any savings from a new entrant offering cheaper prices will be insufficient to compensate the customer for their lost time or the ruin of a favorite blouse.

Take the example of an Egyptian manufacturer trying to compete with China on T-shirts. Average monthly wages in China are $915, while those in Egypt are about $183.33 Assuming a work-week of forty hours, the hourly wage in China is about $5 an hour, while in Egypt it is $1. So the saving in labor cost to hand print a T-shirt that takes an hour to make (a very, very nice T-shirt) in Egypt rather than in China is at most $4. In fact, it is probably much less since T-shirt makers tend to pay a lot less than the average wage. As buyers, many of us would happily pay the extra $4 for the peace of mind its quality assures. Amazon knows that. Why would it pay to experiment with the unknown guy in Egypt when it has a known and reliable supplier in China?

In the case of the Egyptian carpets, an intermediary (in fact, two: ATA and Hamis Carpets) was needed because it was impossible for each individual carpet weaver to build a reputation. They were just too small. Hamis at least had the volume needed to establish a track record of identifying good producers and monitoring their work effectively, and thereby establishing a reputation for quality. It was also in a position to teach them to improve their quality: the exporting firms improved quality very quickly and were soon much better technically than similar firms that had lost the lottery for being included in the study. But since no one outside Egypt knew Hamis, it is no surprise that hardly anyone initially wanted to deal with it or give it a chance to build a reputation.

Making matters worse, when Hamis finally got the chance to export, it had the reverse problem to deal with. A foreign buyer might also be tempted to misbehave: to not pay for an order or change their mind on what they wanted. Hamis had to be the trusted intermediary on both sides. For example, one buyer had asked the carpets be given an antique look by bathing them in tea and sprinkling them with acid. Unfortunately, when they received the carpets, they hated the result and blamed the manufacturer.

In such cases, Hamis was caught between a rock and a hard place. It could try to push back against the buyer, but there was never going to be adequate documentation of all the back-and-forth before the order was filled (“Yes, there was an email, but remember what we said on the phone”). So Hamis would be put into a he said–she said situation where, being a new player and from Egypt to boot, it was unlikely matters would turn out well. On the other hand, the manufacturers in Egypt felt they had done what they were asked to do and would be very upset if they did not get paid. They could not afford not to be. In the end, Hamis often had to absorb the losses.

We first encountered the pain of establishing a reputation in the nascent Indian software industry in the late 1990s. Software in India initially developed around the southern city of Bangalore, then a sleepy town known for its pleasant climate (and now a sprawling metropolis with impossible traffic). Indian firms specialized in customized products for specific clients. If a company wanted a new accounting software, they could get a standard one customized for them, or they could get one built from scratch by an Indian firm.

India had several clear advantages in this sector: a supply of graduates from engineering colleges well known for their excellence, good internet access, English as a first language, and a different time zone, which allowed software engineers to work on different shifts from their American clients. The infrastructure needs were minimal: an office, a small team, a few computers. In Bangalore, this was made even easier by the establishment, as early as 1978, of Electronic City, an industrial park reserved for firms in what would later be called the infotech sector, which came with an assured supply of electricity and reliable communication lines.

All this made it relatively easy for anybody with the right diploma and a willingness to work hard to hang up their shingle and establish themselves as a software firm. But surviving in the industry was not easy.

In the winter of 1997–1998, we asked the CEOs of over a hundred Indian software firms about their experiences with their most recent two projects. For CEOs of young firms, life was unglamorous and hard. A client would specify what they wanted, the firm would try their best to build it, but the client would often claim it was not exactly what they had requested. The CEOs almost always felt the client had changed their mind, but the client typically took the view that the firm had not understood the requirements. In any case, for the most part disagreeing was futile, since the deal with young firms almost always involved a contract where they got paid a fixed amount irrespective of the amount of work done, and only when the buyer was satisfied.

We suspect the choice of this type of contract reflected the buyer’s sense that it was taking a risk by contracting with an unknown supplier in faraway India. Consistent with this interpretation, as firms matured and presumably became better known, we saw a switch from fixed-price contracts to cost-plus contracts, where the buyer paid for whatever time and materials it cost the seller to produce the software.34 Our story also explains why the relatively few cases where a young firm got a cost-plus contract tended to be when the firm had already done a project for the client and therefore had established a reputation.

One of the young CEOs we met was exhausted. He felt he was working night and day on uninteresting projects (and their endless adjustments) just to stay afloat. He had recently taken up a Y2K project, which meant hunting through thousands of lines of code to eliminate dates written in the form “1/1/99” rather than in the form “1/1/1999.” There were dire warnings of the disasters that would ensue if computers started thinking the year was 2099. Companies were rushing to fix their databases.

The work was predictable—there was relatively little risk of a disastrous cost overrun—but mind numbing. The CEO was considering shutting down and joining a bigger firm. The life of slogging through mindless projects, haggling with clients who did not know what they wanted, and constantly wondering whether he could pay his rent was not what he had signed up for when he launched his dream of software entrepreneurship.

Young firms lacking a reputation need to start with deep pockets. Although people often refer to Infosys, started in 1981 by seven engineers with $250 borrowed from the first CEO’s wife and now the third-largest software company in India, it is probably not a coincidence that India’s two biggest software firms today are Wipro, owned by a family that had a successful cooking-oil business before branching out into software, and Tata Consultancy Services (TCS), part of the large industrial Tata Group that produces everything from salt to steel. Of course, it took more than money. In these two cases there was also someone with vision and talent. But clearly money helped.

Having a name also helps. It is no accident that Gucci, originally a high-end leather goods producer, now sells everything from car seats to perfume, and that Ferrari, which started with sports cars, now sells eyeglasses and laptops. Buyers of Gucci perfumes or Ferrari laptops probably don’t expect particularly innovative products from those brand names. They are going, rather, for the assurance Gucci and Ferrari value their good names too much to sell low-quality products, and perhaps the bragging rights that come with buying something clearly expensive.

The value of a brand name is that it wards off competition. That the buyers are so much richer than the producers makes it very important for the seller or the intermediary to focus on quality rather than price. What makes, for any potential new entrant, the challenge of undercutting the incumbent even harder is that the price paid to the supplier tends to be a small part of what a good-quality product is worth to the buyer. Indeed, branding and distribution costs are often much larger than manufacturing costs. For many items, the cost of production is no more than 10–15 percent of the retail cost. This means a more efficient producer can do very little to affect the final price of the product in proportional terms. Cutting his cost of production by 50 percent would only reduce the overall cost of putting the product in the hands of the buyer by at most 7.5 percent.

That could still be a significant amount of money, but as a large literature has demonstrated, proportional changes are what buyers seem to care about. In a classic experiment, one group was asked whether they would drive twenty minutes to save $5 on a $15 calculator and another group whether they would do the same for a $125 calculator. Twenty minutes is twenty minutes, and $5 is $5, but the answers were very different: “68 percent of the respondents were willing to make an extra trip to save $5 on a $15 calculator; only 29 percent were willing to exert the same effort when the price of the calculator was $125.” The point is that $5 is a third of $15 but only 4 percent of $125, which is why they switch in one case but not the other. Consumers are unlikely to switch sellers to save 7.5 percent.35

What this means is China’s prices can increase quite a bit without anyone really noticing. Moreover, there is no reason for these prices to significantly increase anytime soon. China is a big country with a lot of very poor people willing to take jobs at current wages, so costs will remain low. Countries like Vietnam and Bangladesh that aspire to be the next China, the supplier of every kind of cheap manufacture to the world, might spend a long time waiting in the wings. And it is a bit frightening to imagine just how long that could be for Liberia, Haiti, and the Democratic Republic of the Congo, which would like to inherit the same mantle one day, once Bangladesh and Vietnam are too rich to want it.

The outsized role of reputation means international trade is not just about good prices, good ideas, low tariffs, and cheap transportation. It is very difficult for a new player to enter and take over a market, because they start without reputation. This along with the stickiness of labor means the easy flow of people and moneys that free trade is meant to leverage, and which the Stolper-Samuelson thesis is based on, does not work nearly as well on the ground.

To make matters worse for, say, a new country trying to get into the fray, it is not only your own name that counts. Japanese cars are known to be well built, Italian cars are famous for being stylish, German cars are great to drive. A new Japanese entrant, like Mitsubishi when it first entered the US market in 1982, probably benefitted significantly from the success of older Japanese brands. Conversely, buyers are unlikely to want to try out a car produced in Bangladesh or Burundi, even if it is supposedly made to the most exacting standards, the price is low, and the reviews are good. God knows, they will wonder, what might go wrong in a few years. And they may well be right. It is possible that it would take many years of experience producing for the domestic market to know how to make a good car. That is how Toyota, Nissan, and Honda got started.

However, suspicion of newcomers can also turn into a self-fulfilling prophecy. If almost no one buys the car, the company will collapse and customer service will cease. Or if everybody expects the Egyptian rugs to fade, then they will sell for very little money and therefore it would not pay for entrepreneurs in Egypt to invest in producing higher quality rugs. It’s a vicious cycle.36

The curse of low expectations can be very hard to overcome. Even if a firm chooses to deliver the highest-quality products, sufficiently pessimistic buyers will assume it is just a matter of time before the quality goes down. This is where it can be very useful to have the right connections: someone who knows you and will vouch for you.

It is no accident that ethnic Indians and Chinese who lived and worked in Western countries played an important role in their native countries’ transition when they returned home. They used their reputation earned and business cards collected to assure buyers (often firms where they had already worked) that things would be okay.

The presence of some success stories can set off a virtuous cycle. Buyers tend to flock to firms that have had one successful breakthrough, reassured by the fact that others have continued to do business with them. Most young sellers who get an order, recognizing this is their one chance to break the vicious cycle of low expectations, will try their best to deliver when given a chance.

For example, in the rose export market in Kenya,37 local producers work with intermediaries to export their roses to Europe. Neither the buyer nor the seller in this industry can rely solely on formal contracts to enforce good behavior. Roses are very perishable, so upon receiving a shipment a buyer could always claim the roses were not of an acceptable quality and refuse to pay. But, on the other hand, the seller could also claim the buyer somehow spoiled the roses to avoid paying. This means that establishing a reputation for reliability is important. During a period of political unrest in Kenya after the disputed presidential election of 2007, when workers were scarce and transportation was dangerous, new producers who were yet to establish a reputation went to great lengths to continue delivering to their buyers. Some even hired armed guards to protect their roses during delivery. The buyers stayed happy and the Kenya rose market survived the unrest.

Of course, even such desperate measures may not always save your skin. The overall reputation of the industry matters, and it may take only a few bad eggs to ruin the reputation of an otherwise high-quality industry. Governments, recognizing this, have tried to find ways to penalize individual producers who cheat on quality. In 2017, the Chinese government decided these penalties needed to be upped. China Daily quoted Huang Guoliang, director of the administration’s quality supervision department: “Current law generally imposes administrative penalties on violators of product quality law, which are too lenient… A system under which violators of the law would suffer devastating consequences would act as a deterrent [italics added].”38

The best-case scenario in this world of fragile and interconnected reputations is often an “industrial cluster,” a concentration of firms in the same industry in one location, all benefitting from the reputation associated with the cluster.

There have been knitwear factories in Tirupur in India since 1925, and throughout the 1960s and 1970s, the industry grew, producing mainly the white cotton tank tops Indian men wear under their shirts. In 1978, an Italian garment importer, a Mr. Verona, was desperately looking for a large shipment of white T-shirts. The association of garment exporters in Mumbai directed him to Tirupur. Happy with his first lot, he came back for more. In 1981, the first major European chain, C&A, followed him to Tirupur. Its exports were still only $1.5 million until 1985. Then they grew exponentially. By 1990, Tirupur’s export volumes had passed $142 million.39 Exports peaked at $1.3 billion in 2016, though the industry is now facing severe pressure from China, Vietnam, and other recent entrants to the market.40

China has scores of very large specialized manufacturing clusters (“socks city,” “sweater city,” “footwear capital,” etc.). For example, the Zhili cluster in Huzhou has more than ten thousand enterprises producing children’s wear, employing 300,000 workers. In 2012, it was responsible for 40 percent of the GDP of its region. The United States has clusters too, some better known than others. Boston has a biotech cluster. Carlsbad, near Los Angeles, specializes in golf equipment, and Michigan has clocks.41

The organization of the garment industry in Tirupur reveals the value of a name. The whole industry is organized around jobbers, subcontractors who take care of one or more stages of the production process, or even do all the stages for part of a shipment. The jobbers are the invisible people. Buyers deal instead with a smaller number of known names who secure orders and then distribute them among the jobbers. The advantage of this model of production is that it allows production at a very large scale, even if no one has the wherewithal to invest in a single immense factory. Everyone invests what they can and leave it to the intermediaries to put the pieces together. This is another reason why the industry needs to be clustered.

A similar system operates in many large exporting clusters throughout the developing world, where the reputation of some secures the employment of many others. Intermediaries, just like Hamis Carpets in Egypt or the sellers in Tirupur, mediate the relationship with foreign buyers. They have a lot to lose if there is a problem with quality from any of the jobbers and therefore take care of quality control. And while there can be a lot of teething pain, as we saw in the case of Hamis, the eventual rewards are probably quite decent.

Interestingly, this system may be changing. A substantial part of the business model of two of the world’s most successful companies, Amazon and Alibaba, is to insert themselves in place of these intermediaries by allowing individual producers to build their own reputations on their sites, for a price of course, thereby not requiring certification from the intermediary. This is why after you receive a package ordered through Amazon Marketplace, you get repeated entreaties for feedback from Amazon sellers. It is in pursuit of these ratings that they are selling you the socks or the toy for an absurdly low price. Their hope is that one day they will have ratings both numerous and high enough that they can name their price. Of course, it will take some time for these new marketplaces to cement their reputations as guarantors of quality (and they may yet fail). Until they succeed, it is essentially impossible for an isolated producer in the third world to start competing on the international market, however good its product is and however low its prices are.

The Italian maverick Marxist, Antonio Gramsci, once wrote: “The old is dying and the new cannot be born; in this interregnum all manner of morbid symptoms appear.”42 He could have well been writing about the post-liberalization world. As we saw, there are many very good reasons why resources tend to be sticky, especially in developing countries, and breaking into export markets is hard. One consequence of this fact is that trade liberalization anywhere may not be as much of a slam dunk as is often implied by economists. Wages may go down instead of up, even in labor-abundant developing countries where workers should benefit from trade, because everything that labor needs to be productive—capital, land, managers, entrepreneurs, and other workers—is slow to shift from the old job to the new one.

If machines, money, and workers continue to be used in the old sectors, there will be many fewer resources moving to the potential exporting sectors. In India, the effect of the 1991 liberalization was not a massive and sudden change in import and export volumes. Between 1990 and 1992, the openness ratio (the sum of all the imports and exports, as a percentage of the GDP) only increased a little bit, from 15.7 percent to 18.6 percent. But eventually both imports and exports went up, and India today is actually more open than China or the United States.43

Resources eventually moved and new products started being produced. And since existing producers benefitted from being able to import what they needed more easily, what they produced was of better quality and more saleable outside. The software industry, for example, benefited from the ability to import smoothly the hardware they needed, and software exports boomed. Indian firms were quick to switch to imports when they became cheap. Moreover, they also eventually introduced new product lines (for domestic and international use) to take advantage of those cheaper imports. But it took time.44

There is some evidence for the view (held by many policy makers) that the best way to speed up this process is to adopt “export promotion policies,” that help exporters export more. All the East Asian success stories of the postwar era—Japan, Korea, Taiwan, and most recently China—have used one strategy or the other to help exporters speed up their expansion. Most observers believe China, for example, systematically undervalued its exchange rate throughout the 2000s (until about 2010) by selling renminbi and buying foreign currencies to keep its products artificially cheap against the competing products sold in dollars.

In 2010, Paul Krugman called China’s policy the “most distortionary exchange rate policy any major nation ever followed.” It was not cheap: China already owned $2.4 trillion in reserves and it added $30 billion to it per month.45 Given how good the Chinese were at exporting and just how frugal Chinese consumers are, China has a natural tendency to sell more than it buys, and this ought to have pushed the exchange rate up and choked off export growth. The policy prevented this from happening.

Was the promotion of exports good economics? It is possible that it did help the exporters by raising their profits in renminbis (if you sell your shoes for the same number of dollars, the lower the exchange rate, the more local currency you get for them). This made it easier for them to afford to keep the dollar price of their exports low, which encouraged foreigners to buy Chinese, and thereby helped build the reputation of Chinese products. It also helped the exporters accumulate more capital and hire more new workers.

On the other hand, it was at the expense of Chinese consumers who paid for those overvalued imports (this is the flip side of having a weak currency). It is not easy to say what would have happened if the policy had not been adopted. First, the Chinese government also adopted a range of other policies that also favored exporters. China continued to remain competitive when it stopped manipulating its currency after 2010. Second, even if exporters had expanded more slowly, the domestic market might have grown faster and absorbed the surplus. China even today only exports about 20 percent of its GDP; the rest goes to local production.

Even if export promotion did work for China—and it could have—the same strategy is unlikely to work for too many other countries, at least in the near future. The problem in part is China itself. Its success and its enormous size make it harder for others to succeed. The sheer fragility of the process of acquiring a reputation, the critical importance of the right connections, and all the breaks needed to succeed also make us question whether trying to break into international trade is the way forward for the average poor country.

J. D. Vance’s 2016 book Hillbilly Elegy is a lament on behalf of America’s left-behind people, though reading it, one senses the author’s deep ambivalence about how much to blame the victims.46 Part of the economic hollowing out of the parts of Appalachia the book is set in occurred due to trade with China. The fact that poor people got hurt is what we would expect from the Stolper-Samuelson theorem: in rich countries it is the workers who suffer. What is surprising is how geographically concentrated the suffering ends up being. The left-behind people live in left-behind places.

The approach taken by Petia Topalova to examine the impact of trade liberalization on India’s districts was replicated in the United States by David Autor, David Dorn, and Gordon Hanson.47 China’s exports are heavily concentrated in manufacturing, and within manufacturing they are concentrated in specific classes of products. For example, within the apparel sector, sales of some goods in the US, such as women’s nonathletic footwear or waterproof outerwear, are completely dominated by China, but for other goods, such as coated fabrics, almost nothing comes from China.

Between 1991 and 2013, the United States was hit by the “China shock.” China’s share of world manufacturing exports grew from 2.3 percent in 1991 to 18.8 percent in 2013. To examine its labor market impacts, Autor, Dorn, and Hanson constructed an index reflecting the exposure of each US commuting zone to the China shock. (A commuting zone is a cluster of counties constituting a labor market, in the sense that it is possible to commute between them for a job.) The index is built on the idea that if Chinese exports to countries other than the US of a specific commodity are particularly high, implying China is generally successful in that industry, the commuting zones in the US producing that particular commodity will be hurt more than those producing another commodity. For example, since China’s growth in female nonathletic footwear was particularly rapid after China’s accession to the WTO, a commuting zone producing lots of footwear in 1990 would be more affected by the China shock than a commuting zone producing mostly coated fabrics, where China was not so present. So the China shock index measures the vulnerability of a region’s industrial mix to China’s strength by weighing each product type by China’s import to the EU.

US commuting zones fared very differently depending on what they happened to produce. Those zones more affected by the China shock experienced substantially larger reductions in manufacturing employment. More strikingly, there was no reallocation of labor to new kinds of jobs. The total number of jobs lost was often larger than merely the number of jobs lost in the industries that were hit, and rarely less. This is presumably a consequence of the clustering effect we talked about. Those who lost their jobs tightened their belts, further reducing the economic activity in the area. Nonmanufacturing employment did not pick up the slack. If it had, we would have seen an increase in nonmanufacturing employment in the most affected regions. In fact, for lower-skilled workers, the increase in nonmanufacturing employment in affected commuting zones was lower than in other regions. Wages also declined in these areas compared to the rest of the country (and this was a period of stagnant wage growth overall), especially for low-wage workers.

Despite the fact that there were neighboring commuting zones essentially unaffected by the shock (and zones that actually benefitted, say, by importing certain components from China), workers did not move. The working-age population did not decline in the adversely affected commuting zones. They had no work.

This experience is not unique to the United States. Spain, Norway, and Germany all suffered similarly from the impact of the China shock.48 In each case the sticky economy became a sticky trap.

The problem was exacerbated by the clustering of industries. As we already saw, there are many good reasons for industries to cluster, but one potentially negative consequence is that a trade shock may hit with particular violence, potentially affecting all the firms concentrated in the region. In one single year, between October 2016 and October 2017, exports in Tirupur, the Indian T-shirt cluster, went down 41 percent.49

This can set off a downward spiral. Laid-off workers spend less in local businesses, such as shops and restaurants. The value of their houses declines, sometimes catastrophically, since to a large extent the value of my house depends on how nicely your house is maintained. When most of a neighborhood starts to go down, everyone goes down together. Households with larger declines in housing wealth experience a tightening of their credit limit and their ability to refinance, which further reduces their consumption.50 This hits the shops and the restaurants, and some of them end up closing. The disappearance of these amenities, the dearth of nice neighborhoods, and the catastrophic decline in the local tax base that makes it harder to provide water, schools, lights, and roads can eventually make an area so unattractive that it becomes impossible to revive. No new firm will want to move there to take the place of those that have died.

This logic applies just as much to the manufacturing clusters in the United States as it does to those in India or China. Tennessee, for example, had a large concentration of clusters producing goods directly competing with China, from furniture to textiles. The closure of these firms has produced a series of ghost towns. Bruceton, Tennessee, which was profiled in the Atlantic, had been home to the factory of the Henry I. Siegel Company (H.I.S.). At its peak, H.I.S. made jeans and suits in three giant plants, employing seventeen hundred people. It started winding down in the 1990s. In 2000, it laid off its last fifty-five workers. Afterward, according to the Atlantic article,

The neighboring town of McKenzie lost its pajama factory and a shoe company in the 1990s. It is still trying to convince new businesses to come. Whenever the town hears a new factory wants to move, city employees call the decision maker and try to sell the town to them. They have had some interest, but no taker yet. The Atlantic article goes on:

This is not a reason to try to prevent clustering, since the gains from clustering are potentially very large, but a warning to be willing to step in and deal with what happens when the cluster unravels.

Even though they clearly overestimated the extent to which the market would take care of those directly affected by trade, trade theorists have always known some people would get hurt. Their response has always been that since many people do benefit, we should be willing and able to compensate those who are negatively affected.

Autor, Dorn, and Hanson looked at the extent to which the government stepped in to help the regions ill-affected by trade with China. They found that while they received somewhat more money from public programs, it was much too little to fully compensate for the lost incomes. For example, comparing the residents of the most affected commuting zones to those of the least affected, incomes per adult went down by $549 more in the former, whereas government welfare payments went up by only approximately $58 per adult.52

Furthermore, the composition of these transfers may have contributed to worsen the situations of the workers who lost their jobs. In principle, the primary program to help newly unemployed workers who have lost their jobs due to trade is the Trade Adjustment Assistance (TAA) program. Under the TAA, a qualifying worker can extend unemployment insurance for up to three years as long as they receive training to work in other sectors. They may also get financial help to relocate, to search for jobs, or to get health care.

TAA is a longstanding program, in place since 1974, and yet it provided a minuscule share of the already small transfers toward the affected counties. Of the $58 in additional transfers that went to the more affected regions, only twenty-three extra cents came from TAA. A very big part of what did grow was disability insurance; out of every ten workers who lost their jobs due to trade, one went on disability insurance.

The huge increase in disability insurance is alarming. It is unlikely that trade had a direct effect on the physical health of these workers, especially since the most physically demanding jobs were those that typically disappeared. Some workers were undoubtedly depressed; for others, disability insurance became a strategy they had to adopt to survive. Either way, unfortunately, going to disability is usually a one-way street out of employment. For example, research on a veterans’ program that newly recognized diabetes as a reason to claim disability for those exposed to Agent Orange showed that for every hundred veterans who entered the disability program as a result of the policy change, eighteen dropped out of the labor force for good.53 In the United States, those who join the disability rolls rarely leave them,54 partly because being classified as disabled hurts their employment prospects. Having to adopt disability after a trade shock to pay the bills is likely to push some people who could have otherwise found a new job out of the labor force entirely.

For workers who need to resort to disability benefits to survive, being classified as disabled adds insult to injury. When they go on disability, workers who have spent their lives in a physically demanding job lose not only their occupation, but their claim to dignity. So not only did the United States not come close to compensating the workers who lost out, but what little help people could get through the existing social protection apparatus seemed designed to make them feel denigrated.

Partisan politics has played a role in this disaster. When someone who has lost their job needed healthcare, a recourse was supposed to be Obamacare. Unfortunately, many Republican states like Kansas, Mississippi, Missouri, and Nebraska decided to make a show of resisting the federal government by denying their citizens this option. This pushed some people to apply for disability status in order to get healthcare. Indeed, after the adoption of the Affordable Care Act (a.k.a. Obamacare), disability claims increased by 1 percent in states that refused to expand Medicaid, while they decreased by 3 percent in expansion states.55

But the causes run deeper. US politicians are wary of subsidizing specific sectors (since others would feel slighted and would lobby for their own protection), which is probably partly the reason why TAA has remained such a small program. Economists have also traditionally been unwilling to embrace place-based policies (“help people, not places” as the slogan goes). Enrico Moretti, one of the few economists who has actually studied such policies, actively dislikes them. For him, channeling public funds into regions doing poorly is throwing good money after bad. Blighted towns are meant to shrink while others take their place. It is the way of history. What public policy needs to do is to help people move to the places of the future.56

This analysis seems to give too little weight to the facts on the ground. As we know, the same reasons that make clusters develop also mean they fall apart quickly. Theoretically, the obvious response to this wholesale unwinding ought to be for a lot of people to leave, but as we saw already, they don’t. At least not nearly fast enough. Instead, when their county was hit by the China shock, fewer people got married, fewer had children, and of the children who were born, more were born out of wedlock. Young men—and, in particular, young white men—were less likely to graduate from college.57 “Deaths of despair” from drug and alcohol poisoning and suicides skyrocketed.58 These are all symptoms of a deep hopelessness once associated with African American communities in inner cities of the United States but are now replicated in white suburbs and industrial towns up and down the Eastern Seaboard and the eastern Midwest. A lot of this damage is irreversible, at least in the short run. The school dropouts, the drug and alcohol addicts, and the children growing up without a father or a mother have lost a part of their futures. Permanently.

Donald Trump decided the solution to the negative effect of trade was tariffs. He welcomed a trade war. It started in the first few months of 2018, with new tariffs on aluminum and steel. Trump then talked about $50 billion in tariffs on Chinese goods, and then when China retaliated, suggested another $100 billion.

The stock market tumbled on the announcement, but the basic instinct that we should close our economy and, in particular, defend it against China is shared by many Americans on both sides of the aisle.

Meanwhile, economists were jumping up and down. They evoked the specter of the “worst ever tariff,” the Smoot-Hawley Tariff Act, which precipitated a global trade war in 1930 by imposing tariffs on twenty thousand goods imported into the United States. The Smoot-Hawley bill coincided with the onset of the Great Depression, and although it may or may not have caused it, it certainly gave sweeping tariffs a bad rep.

The idea that more trade is good (on balance) is deeply engrained in anybody who went to graduate school in economics. In May 1930, over a thousand economists had written a letter encouraging President Hoover to veto the Smoot-Hawley bill. And yet there is something else economists do know but tend to keep closely to themselves: the aggregate gains from trade, for a large economy like the United States, are actually quantitatively quite small. The truth is, if the US were to go back to complete autarky, not trading with anybody, it would be poorer. But not that much poorer.

Arnaud Costinot and his longtime collaborator Andrés Rodríguez-Clare managed to make themselves infamous in the community of trade economists for making that point. In March 2018, they released a timely new article, “The US Gains from Trade,” with the following prescient first paragraph:

This article builds on a line of research they developed over several years, both together and with others, and on decades of research in trade. The key idea is that the gains from trade depend primarily on two things: how much we import and the extent to which these imports are influenced by tariff, transportation costs, and the other costs of trading internationally. If we import nothing, clearly it does not matter if we erect a wall and stop importing. Second, even if we import a lot, if we stop doing so when import prices increase even a little bit, because it becomes a little more expensive to bring the goods here, it must mean we have many available substitutes at home, so the value of imports is not that high.

Building on this idea, we can compute the gains from trade. If the United States only imported bananas and produced apples, it would be fairly easy. We could look at the share of bananas in consumption, and the extent to which consumers were willing to switch between apples and bananas as the prices of bananas and apples changed. (These are what economists call cross-price elasticities.) In fact, the United States imports products in about eighty-five hundred categories, so to do this calculation properly, we’d need to know the cross-price elasticity between every product and the price of every other product around the world—apples and bananas, Japanese cars and US soybeans, Costa Rican coffee and Chinese undershirts—making this approach unfeasible.

But in fact we don’t actually need to look at products one by one. We can get reasonably close to the truth by assuming all imports are a single undifferentiated good that is either directly consumed (imports represent 8 percent of US consumption) or used as input for US production (another 3.4 percent of consumption).60

To get the final gains from trade, we need to know just how sensitive our imports are to trade costs. If they are very sensitive, it means it is easy to replace what we import with things we produce locally, and it is not very valuable to trade with other countries. If, on the other hand, the value remains unchanged even as the costs change, it means we really like what we buy abroad, and trade increases welfare a lot. There is some guessing involved here, since we are in fact talking about a good that does not exist, a composite of thousands of widely differing products. The authors therefore present the results for a range of situations, going from a scenario where traded goods can very easily be substituted with domestic goods (leading to gains of trade of 1 percent of GDP) to one where it is very difficult to substitute them (leading to an estimate of 4 percent of GDP).

Costinot and Rodríguez-Clare’s preferred estimate is that the gains from trade are about 2.5 percent of GDP. This is really not a lot. The US economy grew 2.3 percent in 2017,61 so one year of decent growth could pay for sending the US economy into complete autarky, in perpetuity! Did they get something wrong in their calculations? One can argue with many of the details, but the order of magnitude has to be right. Simply put, despite its openness to trade, the US import share (8 percent) is one of the lowest in the world.62 So the gains from international trade to the United States cannot be that large. Belgium, a small open economy, has an import share of above 30 percent, so there trade matters much more.

This is not so surprising. The US economy is very large and very diverse, and therefore capable of producing much of what is consumed there. Moreover, a lot of consumption is of services (everything from banking to house cleaning) not typically traded internationally (yet). Even the consumption of manufactured goods involves a significant share of locally produced services. When we buy an iPhone assembled in China, we also pay for US design and local advertising and marketing. The phone is sold in shiny Apple stores built by local firms and manned by local tech lovers.

We should not be carried away by the US example, however. Large economies like the United States and China have the skills and the capital to produce most things at a very high level of efficiency somewhere in the country. Moreover, their internal markets are large enough to absorb production from many factories in many sectors operating at the appropriate scale. They would lose relatively little by not trading.

International trade is much more important for smaller and poorer countries, like those in Africa, Southeast Asia, or southeastern Europe. Skills there are scarce and so is capital, and the domestic demand for steel or cars is unlikely to be big enough, given that incomes are low and populations are small, to sustain production at scale. Unfortunately, it is precisely those countries that face the biggest barriers to becoming players in the international market.

But for larger developing countries like India, China, Nigeria, or Indonesia, the bigger problem is often internal integration. Many developing countries suffer from a lack of internal connectivity. Nearly a billion people worldwide live more than a mile from a paved road (one-third of them are in India), and nowhere near a train line.63 Internal politics sometimes add to that. China has excellent roads, but Chinese provinces have found ways to discourage domestic firms from importing goods from the rest of the country.64 And until the recent introduction of unified taxes on goods and services in India, each state had the power to set its own tax rates, and often used them to favor local producers.

But perhaps the very idea of comparative advantage is overrated, and even small countries can live in autarky. Or to push the logic even further, perhaps every community can learn to produce what it needs.

This idea has a long and somewhat infamous pedigree. During the Great Leap Forward in China, Chairman Mao argued, among other things, that industrialization could be willed to happen in every village, and that steel could be produced in backyard steel furnaces. The project failed miserably, but not before peasants melted down their pots and pans and plowshares to comply with the chairman’s wishes, and busied themselves producing steel while fields remained fallow and crops rotted on the ground. Many China observers think this might have contributed to the Great Chinese Famine of 1958–1960, when upward of thirty million people died.

The idea of self-sufficient village communities was also the centerpiece of Gandhi’s economic philosophy. His vision of a society clothed in homespun and living mainly off the land had a durable effect on Indian economic policy in the post-independence era. Until the WTO forced India to do away with the policy in 2002, 799 goods, from pickles to fountain pens, dyes, and many items of clothing, were reserved for tiny firms that could be set up in villages.

The problem of course is that small is not beautiful. A minimum scale is required to allow firms to employ specialized workers or to use high-productivity machines. In the early 1980s, Abhijit’s mother, Nirmala Banerjee, an economist with quite left-wing views, surveyed small firms in and around Kolkata, and was astounded by just how unproductive they were.66 Later evidence confirmed her insight. In India, small firms are much less productive than larger ones.67

But firms can only be large if the market is large. As Adam Smith wrote in 1776: “The division of labour is limited by the extent of the market.”68 This is why trade is valuable. Isolated communities cannot have productive firms.

Indeed, national integration via railroad has had transformative impacts in many economies. In India, between 1853 and 1930, the British colonial administration oversaw the building of nearly forty-two thousand miles of railroad in India. Before the railways, commodities were transported by bullocks on dirt roads, and could travel at most twenty miles per day. Railroads could transport these same commodities almost four hundred miles in a day, at a much lower cost, and with less risk of spoilage. Inland regions all but cut off from the rest of the country got connected.69 The railroad network dramatically reduced trade costs. The transportation cost per mile traveled was nearly two and a half times higher for roads than for railroads. And places brought together by railways started to trade more and became richer; the value of agricultural production increased 16 percent faster in districts that got a train line, relative to those that did not.

The United States was another large country integrated through a vast network of railroads at about the same time. Although the role of railroads in the development of the US economy has been controversial, recent research suggests agricultural land value would have been 64 percent lower in the absence of railroad construction.70 These land prices embody all the gains farmers expected from better connections with other counties. And the gains came in large part from the ability to specialize in what each region was good at. Between 1890 and 1997, agriculture became more and more locally specialized. Farmers increasingly chose the crop that each field (due to its climate, soil, etc.) was ideally suited for, which led to large gains in overall agricultural productivity and income.71

Poor internal integration also makes economies sticky, eliminating the gains from international trade for the common men and women, or even turning them into losses. Bad roads discourage people from taking new jobs in cities. In India, the unpaved roads connecting villages to main roads have been shown to be a deterrent for rural dwellers to get nonagricultural jobs outside their villages.72 Bumpy rides add so much to the final price of goods that consumers in remote villages enjoy almost no benefits from international trade. In Nigeria and Ethiopia, by the time imported goods arrive at those villages, if they make it at all, they are unaffordable.73 Poor transportation, both for inputs and for the final products, erode the cost advantages of a cheap labor force. Internal connections must improve for international integration to be beneficial.

The examples and analyses in this chapter come from cutting-edge research conducted by the most respected departments of economics, yet the main conclusions may seem to put us at odds with decades of conventional wisdom. While every economics undergraduate learns there are large aggregate gains from trade, and that everybody can be made better off as long as we can redistribute those gains, the three main lessons from this chapter are decisively less rosy.

First, the gains of international trade are fairly small for a large economy like that of the United States. Second, while the gains are potentially much larger for smaller and poorer countries, there is no magic bullet. Just as we saw in the chapter on migration that opening a border widely would not be enough to get everyone to move, removing trade barriers is not enough to ensure new countries can join the party. Declaring trade is free is not the magic bullet for development (or even for trade). Third, the redistribution of gains from trade has proven extremely tricky, and people negatively affected by trade have suffered, and are still suffering, a great deal.

Taken together, the exchange of goods, people, ideas, and cultures made the world much richer. Those lucky enough to be in the right place at the right time, with the right skills or the right ideas, grew wealthy, sometimes fabulously so, benefitting from the opportunity to leverage their special gifts on a global scale. For the rest, the experience has been mixed. Jobs were lost and not replaced. Rising incomes have paid for more new jobs—as chefs and chauffeurs, gardeners, and nannies—but trade has also created a more volatile world where jobs suddenly vanish only to turn up a thousand miles away. The gains and the pains ended up being very unequally distributed and they are, very clearly, starting to bite back at us; along with migration they define our political discourse.

So do protectionist tariffs help? No. Reintroducing tariffs now will not help most Americans. The reason is simple: one of our main argument so far has been that we need to worry about transitions. Many of those displaced by the China shock never really recovered because the sticky economy meant they could not move sectors or regions to get back on their feet, and the resources could not move to them.

But shutting off trade with China now will clearly create a new set of displacements and many of those new losers will be in counties we have not yet heard anything about, simply because they are doing just fine. Indeed, among the 128 products on which China announced tariffs on March 22 and April 2, 2018, the majority were agricultural: a.p.p. (apples, pears, and pork), rather than apps. US exports in agriculture have risen steadily over the last few decades (from $56 billion in 1995 to $140 billion in 2017). Today a fifth of US agricultural production is exported. And the biggest export destination is East Asia. China alone buys 16 percent of US agricultural exports.74

The first-order effect of a trade war with China is therefore likely to be a loss of jobs in agriculture and in the industries supporting it. The US Department of Agriculture estimates that in 2016, agricultural exports were responsible for over a million jobs in the United States, almost three-quarters of which were in the nonfarm sector.75 The five states with the largest share of agricultural employment are California, Iowa, Louisiana, Alabama, and Florida.76 For precisely the same reasons people who lost their manufacturing jobs in Pennsylvania were not able to get other jobs near home, these agricultural jobs will not be replaced by manufacturing jobs in the region. And we know from everything we have seen in this chapter and the previous one that just as manufacturing workers did not move when their jobs were lost, farm workers would probably not move. Alabama and Louisiana are two of the ten poorest states in the United States,77 and a trade war would throw them under the bus.

For the United States, a trade war would not be the end of the world as we know it. But while it may save some jobs in steel, it would likely cause significant new damage to others. The US economy will be fine. Hundreds of thousands of people will not.

Since the main problem with trade is that it creates many more losers than the Stopler-Samuelson theory suggests, it seems any solution should involve either limiting the number of losers by helping them move or change jobs, or finding a way to compensate them better.

One side benefit of the negative effect of trade being so concentrated is that we actually know where to look for the victims. Why not target some help directly to the workers in industries that lost out to the China shock? Indeed, this was the idea behind the Trade Adjustment Assistance program. The TAA pays for training (up to $10,000 a year) and the trained workers get up to three years in unemployment benefits, precisely to give them some time to land on their feet. The only problem, as we saw, is that the program remained tiny.

Sadly, this was not because TAA was ineffective as a concept; it was just severely underfunded. To qualify for the program, a worker must petition the Department of Labor. A caseworker is then allocated the worker’s file and tasked with determining whether in this case the job in the worker’s former firm disappeared because of competition from imports, the offshoring of jobs, or ripple effects from the trade-induced distress of other companies that either bought from or sold to that firm.

A complex judgment goes into this decision, and some caseworkers are much more willing than others to rule in favor of the worker and allocate them aid. One study makes the case that the assignment of a petition to a particular caseworker, and therefore the eventual judgment, is more or less random.78 Using a database of 300,000 petitions, it compares workers assigned to more or less lenient caseworkers. Workers assigned to more lenient caseworkers are more likely to receive the TAA and therefore more likely to be trained, move sectors, and earn more money. Overall, workers awarded TAA initially had to forego $10,000 in earnings (since they could not work while they got the training), and the government spent some money for the training, but over the next ten years the retrained worker earned $50,000 more than the untrained worker. It took ten years for the salary levels of the retrained and untrained workers to converge. This was thus a worthwhile investment for them, although not one they could undertake without the government’s support, since getting a bank loan for this purpose would have been very difficult.

So why was an effective program like the TAA underfunded and underused? Partly because neither policy makers nor the public knew it worked until that study came out, quite recently. This probably reflects the lack of interest in these kinds of policies among trade economists. Economists also don’t like programs that rely so much on a judgment call; they worry about potential abuse. At a political level, spending large sums of money on trade adjustment would have made it more explicit that trade adjustment costs are in fact large, and this may not have been palatable.

One obvious path is therefore to expand a program like the TAA, making it both more generous to individuals and more easily awarded. For example, the revamped TAA could be modeled on the GI bill, paying enough for someone who is a “veteran” of a trade shock to get a new start with their education. The GI Bill provides up to thirty-six months of education benefits, pays for full tuition at public schools, and up to $1,994 toward tuition for a full-time student (and a prorated rate for part-time programs), as well as a stipend for housing.79 The new TAA could be something like that, combined with extended unemployment insurance for the duration a person is in school. And since we know there are strong local market effects from trade disruptions, the TAA could be more generous in regions known to have been particularly affected by trade shocks, to avoid sending the affected labor markets into a downward spiral.

More generally, much of the hardship caused by trade is related to the immobility of both people and resources. The free movement of goods across borders is not matched by movement within countries. All the solutions we discussed at the end of chapter 2 to encourage internal migration, and the seamless integration of movers (subsidies, housing, insurance, help with childcare, etc.) would help in adjusting to trade shocks.

But it is also clear that mobility, TAA induced or not, is not the ideal solution for all workers. Some may not want to, or not be able to, be retrained; others may not want to change their job, particularly if this involves moving. This may be especially true for older workers. For them retraining would be difficult, and they might be less likely than younger workers to find a new job afterward. Indeed, a study found that after mass layoffs, older workers find it very difficult to find another job. Two and four years after losing their job, men and women swept in a mass layoff at age fifty-five were at least twenty percentage points more likely to be unemployed than those lucky enough to escape job loss at fifty-five.80 This kind of job loss has a permanent effect on younger workers as well, but the impact is nowhere near as large.81

Older workers who get fired also tend to be those who spent a long career working at a particular job. For them, the work they do provides a sense of pride and identity and defines the place they have in their communities. It is difficult to compensate them with an invitation to be trained to do something entirely different.

Why not then offer to subsidize firms adversely affected by trade (particularly those located in the most affected regions) as long as they keep employing older workers? Larry Summers (the head of the National Economic Council from 2009 to 2012) and Edward Glaeser have recently argued for a payroll-tax reduction in some specific areas.82 A tax reduction may, however, be insufficient to convince a firm to keep its employees if it has become uncompetitive. By being more specific about the sector and the areas, and by restricting the program to already employed workers between the ages of fifty-five and sixty-two (when they can claim social security and retire), it would be possible to spend much more money on each person, possibly compensating the firm for more than the cost of a full-time worker if that is what it takes. That won’t save every firm, but it might preserve a significant amount of employment where it matters the most, prevent communities from falling apart, and be part of the necessarily long transition to a new path. The right way to pay for this is to use general tax revenue. To the extent we are all benefitting from trade, we should collectively pay for the cost. It makes no sense to ask agricultural workers to lose their jobs just so steelworkers can keep theirs, which is what tariffs accomplish.

Of course, the proposal is not without practical difficulties. Affected firms would need to be identified, and there would certainly be lobbying and attempts to circumvent the rules. The proposal may be seen as a form of trade protection and run afoul of WTO rules. But these issues could be solved. The principle of identifying firms that have been subject to trade shocks is already accepted by the TAA program, which has developed a mechanism to adjudicate claims. To avoid casting it as trade protection, the provision could be extended to jobs lost due to technological disruptions.

The overarching takeaway is that we need to address the pain that goes with the need to change, to move, to lose one’s understanding of what is a good life and a good job. Economists and policy makers were blindsided by the hostile reaction to free trade, even though they have long known that as a class workers were likely to suffer from trade in rich countries and benefit from it in poor countries. The reason is they have taken it for granted that workers would be able to move jobs or places, or both, and if they were not able to do this, it was somehow their failing. This belief has colored social policy, and set up the conflict between the “losers” and the rest that we are experiencing today.