COMPETITION from Chinese imports may have cost some Americans jobs, but economists have done pretty well out of it. Since 2013 David Autor, David Dorn and Gordon Hanson have published nine separate studies digging into the costs of trade. They have found that, of the fall in manufacturing jobs between 1990 and 2007, one-quarter could be attributed to a surge in imports from China. Other sectors failed to soak up the extra workers. Their research also suggested that the China shock has cut the supply of marriageable men and opened the door of the White House to Donald Trump.

In recent weeks a dispute has erupted over their results. Jonathan Rothwell, an economist at Gallup, a pollster, alleged “serious flaws” in one paper, prompting a fierce eight-page response from the authors, and an acrimonious public tiff.

The row centres on how the effect of the China shock is measured. The trio wanted to isolate the effects of extra Chinese supply, rather than of something happening in America, so they checked that imports of particular Chinese products were surging in other rich countries, too. They then compared places in America more exposed to these Chinese imports—typically those with lots of labour-intensive manufacturing—with less exposed ones.

Mr Rothwell’s critique does not attempt to debunk their research completely. But he asks whether combining changes in the 1990s and the 2000s makes sense. When he splits this period up, he confirms the finding that Chinese imports had large effects on American manufacturing employment. But several other effects of Chinese imports become smaller or no longer statistically significant. For example, the effect of Chinese imports on the size of the labour force falls to a quarter of its 1990s size in the 2000s. This is hardly conclusive—slashing sample sizes inevitably reduces the power of a test.

Mr Rothwell has not disproved anything. But he has provided an opportunity to think through the assumptions of the original research by Messrs Autor, Dorn and Hanson. Their attempt to isolate the effects of China would not have been entirely successful, for instance, if other countries were experiencing non-China-related shocks similar to those hitting America.

More broadly, it is impossible to know what would have happened had Chinese imports not surged. Monetary policy might have been different. And what a company such as Apple would have done without low-cost Chinese assembly workers is unknowable. Moreover, adding up individual effects over the whole economy could miss important interactions.

Mr Rothwell’s strongest criticism is not of that China-shock literature at all, so much as of the way it was received. Some have taken evidence of disruption as proof that tariffs would be a good idea, or that trade with China has hurt America. But, as Mr Autor says himself, “our research does not tell you the net societal costs and benefits of trade.” It does not estimate the benefits to exporters as China opened up (though this was smaller than the rise in imports) or to American shoppers able to buy cheaper stuff.

Other research is emerging that attempts to answer those questions. One paper, by Kyle Handley and Nuno Limão, found that the extra trading certainty associated with China’s accession to the WTO lowered American manufacturing sales and employment by more than 1%, but also lowered American prices and raised consumers’ incomes by the equivalent of a 13-percentage-point cut in tariffs. It also helped poor Chinese workers get richer, which isn’t to be sniffed at either.