Ben Casselman's NYT piece on economist Tim Kane's run for a congressional seat in Ohio called my attention to an economists' poll that I had missed. The poll posed the following question to a group of elite economists:

"An important reason why many workers in Michigan and Ohio have lost jobs in recent years is because US presidential administrations over the past 30 years have not been tough enough in trade negotiations."

Of the whole group, 64 percent either strongly disagreed or disagreed. Only 5 percent agreed. (The rest were uncertain or didn't answer.)

This outcome is striking because standard trade models absolutely predict that some people will be losers from trade. The basic story is that workers in an sector where the U.S. has a comparative disadvantage will end up with lower pay as a result of removing trade barriers.

To make it concrete, suppose the auto industry is protected by a 20 percent tariff barrier and we make the tariff zero. The expected result is that we would have fewer workers employed in the auto industry. The workers who lose jobs in the industry will, in general, get lower pay in their new jobs, as will the workers who remain employed in the industry.

The argument for free trade is that the gains in aggregate will be larger than what these workers lose. In principle, the winners can redistribute to the losers and make everyone better off, but there is no dispute that the workers in the auto industry are made directly worse off by the removal of the tariff taken by itself.

What is striking is that there is considerable research by many economists, most notably a group led by MIT economist David Autor, showing that workers in Michigan and Ohio were badly hurt by the trade opening to China in the last decade. In effect, these economists were asked whether they thought autoworkers were hurt by the elimination of a tariff on imported autos after they had been shown evidence of large-scale job loss and wage declines in the sector.

The vast majority still said "no."