IN MARCH 2000, two months before a crucial vote in America’s Congress on whether to make normal trading relations with China permanent, Bill Clinton gave a press conference. In the first year of his presidency, 1993, he had made a bold case for the North Atlantic Free Trade Agreement (NAFTA) with Canada and Mexico, claiming it would create 200,000 jobs in America. Now, in the final year of his second term, he was even more bullish about a trade pact with China, which would allow that country to join the WTO. It would require China quickly to cut its average import tariff from 24% to 9%, to abolish import quotas and licences and to open up some industries to American investment. America, for its part, would not have to do anything. “This is a hundred-to-nothing deal for America when it comes to the economic consequences,” said Mr Clinton.

Sixteen years on the mood is rather different. Job losses in manufacturing states such as Michigan, Ohio and Pennsylvania have made trade a key issue in America’s presidential election. Donald Trump has risen to prominence in part by promising to impose steep tariffs on imports from China and Mexico, claiming America’s trade deficit with both countries (see chart) shows it is “losing”.

Hillary Clinton is no longer supporting the TPP trade deal she had earlier favoured. The demise of furniture-makers and textile firms, unable to compete with low-cost imports, belies the predictions made by her husband. Bernie Sanders, Mrs Clinton’s opponent in the Democratic Party primaries, said trade deals had been “a disaster for American workers”. A YouTube clip earlier this year showing the graceless manner in which bosses of Carrier, a maker of air-conditioners, told its workforce that it was moving production to Mexico seemed to confirm every fear about the exodus of jobs and the heartlessness of capitalism.

What is behind the change in mood? The years after the NAFTA agreement came into force, in 1994, were actually rather good ones for America’s economy, including manufacturing. But China’s accession to the WTO caused a big shock. The country’s size, and the speed at which it conquered rich-world markets for low-cost manufacturing, makes it unique. By 2013 it had captured one-fifth of all manufacturing exports worldwide, compared with a share of only 2% in 1991.

This coincided with a fresh decline in factory jobs in America. Between 1999 and 2011 America lost almost 6m manufacturing jobs in net terms. That may not be as dramatic as it sounds, since America is a large and dynamic place where around 5m jobs come and go every month. Still, when David Autor of the Massachusetts Institute of Technology (MIT), David Dorn of the University of Zurich and Gordon Hanson of the University of California, San Diego, looked into the job losses more closely, they found something worrying. At least one-fifth of the drop in factory jobs during that period was the direct result of competition from China.

Moreover, the American workers who had lost those jobs neither found new ones close by nor searched for work farther afield. They either swelled the ranks of the unemployed or, more often, left the workforce. That contradicts the widespread belief that America’s jobs market is fluid and flexible.

When men lose a factory job, they often stay put. Those who managed to find new jobs were paid less than before and were working in industries that were vulnerable to competition from imports. In subsequent research, the authors found that lost factory jobs also had a depressing effect on aggregate demand (and thus non-manufacturing jobs) in the affected areas. In total, up to 2.4m jobs may have been lost, directly and indirectly, as a consequence of imports from China.

In other rich countries, regions or industries with heavy exposure to Chinese imports also suffered material losses in factory jobs. A study of Spain’s jobs market by Vicente Donoso, of the Complutense University of Madrid, and others found that provinces with the greatest exposure to Chinese imports saw the largest falls in the share of manufacturing employment between 1999 and 2007, but this was compensated for by an increase in non-factory jobs.

Research in Norway, though, found that the main effect was to raise unemployment. João Paulo Pessoa of the London School of Economics found that British workers in industries exposed to high levels of import competition from China spent more time out of work than those in other industries. A wide-ranging study of the effect on Germany of more trade with China and eastern Europe in the two decades after 1988 concluded that industries competing with imports suffered job losses, but these were outweighed by job gains in regions focused on export industries. Those gains were due almost entirely to trade with eastern Europe, not China.

China’s accession to the WTO was supposed to be a great bonus for America. So why was its impact on trade and jobs so unexpectedly large? One reason was that China got a very significant advantage out of the pact. A paper by Justin Pierce, of the Federal Reserve, and Peter Schott, of Yale School of Management, argues that joining the WTO removed the risk for China of a steep increase in America’s tariffs, making it less perilous for its companies to invest in new factories. The authors found that industries where the threat of tariff increases was most reduced suffered the greatest job losses in America. But the lopsided nature of trade between China and the rich world also played a part. After China joined the WTO, its current-account surplus widened from an average of around 2% of GDP in the 1990s to about 5% in the following decade. In other words, China saved more. That helps explain the modest offsetting gains in exports in the regions affected by Chinese imports.
Done workin’
It is important to note that America’s growing inability to bounce back from losing manufacturing jobs predates the rise of China as an exporting power. A report published in June by the Council of Economic Advisers (CEA) charts the long-term decline in prime-aged men in America’s workforce.

It shows that in the mid-1960s almost all men aged between 25 and 54 were either in work or looking for a job, but that in the past half-century the participation rate for this group has dropped below 90%.

In every recession the rate falls more sharply, and when the economy picks up again it fails to make up all the lost ground.

But there are big differences between the participation rates of different groups of men. In 1964 male high-school graduates were about as likely to be in the workforce as college-educated men, but now only 83% of those with a high-school degree or less are in the workforce, against 94% of those who finished college (see chart). This mirrors a growing divergence in wages. In the mid-1960s the pay of less educated men averaged 80% of college-educated ones, but by 2014 that proportion had fallen to 60%.

It is unlikely that men are dropping out of work voluntarily. More than a third of inactive men live in poverty; less than a quarter have a working spouse. So the most obvious explanation is a fall in demand for less-skilled men. That in turn is partly linked to a long-term decline in manufacturing, whose share of the jobs market peaked in the days when almost all prime-age men worked. The CEA study found that states with a higher-than-average share of jobs in construction, mining and (to a lesser degree) manufacturing tend to have more prime-age men in the workforce. It does not help that men who lose their jobs are increasingly rooted in unemployment black spots. The propensity of people to move in search of work has dropped sharply since the early 1990s, for reasons that are not yet fully understood.

A steady drop in the share of prime-age men in the workforce going back half a century cannot be pinned on America signing free-trade agreements or China’s emergence as an exporter of manufactures, both of which happened fairly recently. Factory jobs peaked in the 1970s, but manufacturing output has continued to increase. Indeed, America’s share of world manufacturing output, on a value-added basis, has been fairly stable at a bit under a fifth for the past four decades.

Thanks to advances in technology, fewer workers are needed to produce the same quantity of goods.

But since trade with lower-cost countries and technological change have similar effects on labour-intensive production in the rich world, it is hard to disentangle their effects.

Still, some rich countries, such as Germany, Britain and Canada, have done rather better than America at keeping prime-age men in work, though others, including France, Italy and Spain, have done even worse. That is partly a matter of policy. Members of the OECD, a club of mostly rich countries, set aside an average of 0.6% of GDP a year for “active labour-market policies”—job centres, retraining schemes and employment subsidies—to ease the transition to new types of work. America spends just 0.1% of GDP. By neglecting those whose jobs have been swallowed by technology or imports, America’s policymakers have fuelled some of the anger about freer trade.

Have trade deals really been a disaster for American workers? Trade with China seems to have had an unusually large effect. Since 1985, America has signed 15 free-trade agreements (FTAs) covering 20 countries. Exports to these countries account for nearly half of all the goods America sells abroad, even though FTA countries make up just a tenth of GDP outside America. In the five years after a new trade pact comes into force, America’s exports to new FTA partners typically grow around three times as fast as its overall exports, at least keeping pace with imports. In 2012, exports to the 20 countries covered by FTAs grew twice as fast as the average. In America, exporting firms pay a wage premium of between 13% and 18%, compared with non-exporters. This is hardly a disaster.

America has run a trade deficit every year since 1976. On the other side of the global ledger are countries that consistently run big trade surpluses. These days the record is held not by China but by Germany, which last year had a current-account surplus of 8% of GDP (see chart). But this does not mean that America is “losing” at trade, as Mr Trump suggests, and China and Germany are winning.

The purpose of exports is to pay for imports, either now or later. A trade surplus is not a virility symbol. In some cases, it is a sign of a strong national preference for saving (though other countries might describe it as a symptom of weak domestic demand). Countries rarely have balanced trade, where the value of exports and imports is exactly the same. It might seem plausible that restricting trade to eliminate deficits will create jobs, channelling existing demand towards goods made at home.

But the reality is more complicated. In most rich countries, particularly America, the trade deficit widens when GDP growth is strong, and shrinks during recessions. The factors that drive demand for imports are the same as those that drive overall demand, and thus jobs. To balance trade, Americans would have to invest less or save more. Neither would create jobs.

It would help a sluggish world economy if surplus countries, like China and Germany, were to spend more on imports. But for America to aim to balance trade with any one country would be pointless.

In any case, a finished product exported from China to America, say, will include components made in third countries, and probably only a small fraction of the value will have been added in China itself. Four-fifths of all trade takes place along supply chains within, or organised by, multinational firms. Slapping a tariff on imports of intermediate goods from, say, Mexico would raise the price of America’s exports, which would probably be bad for its trade balance.

Around 40% of the value of Mexico’s exports of final goods to America, for instance, was added in America itself.

Sober advocates of free trade know that over time the gains from it come from greater efficiency, not from more jobs, the number of which is largely determined by demography and the strength of aggregate demand. It is easier to spot the link between freer trade and factory closures than the more dispersed benefits trade brings to workers across other industries.

Exporting firms in all countries and across a variety of industries are more productive, grow faster and pay higher wages than non-exporting firms. But a lot of the gains from trade come from the direct benefit of cheaper imports and their indirect effect on productivity.

The cost of protectionism
A study by Pablo Fajgelbaum of the University of California, Los Angeles, and Amit Khandelwal, of Columbia University, suggests that in an average country, people on high incomes would lose 28% of their purchasing power if borders were closed to trade. But the poorest 10% of consumers would lose 63% of their spending power, because they buy relatively more imported goods. The authors find a bias of trade in favour of poorer people in all 40 countries in their study, which included 13 developing countries. An in-depth study of European industry by Nicholas Bloom, of Stanford University, Mirko Draca of Warwick University and John Van Reenen of the LSE found that import competition from China led to a decline in jobs and made life harder for low-tech firms in affected industries. But it also forced surviving firms to become more innovative: R&D spending, patent creation and the use of information technology all increased, as did total factor productivity.

Taken together, these are large and permanent benefits. What is clear from the studies of Mr Autor and others is that the one-off integration of China had bigger and more lasting effects than expected. Too little attention has been paid in America to those whose jobs are displaced by new technology or imports. That has given an opening to protectionists, who are peddling a solution that will hurt the poor most. A similar sort of populism is rearing its head in Europe in response to migration.