Trump is an ass, but that doesn’t mean that his attacks on the biased media are always wrong. The New York Times has recently come out with a howler so egregious that even the overly nice Tyler Cowen had to call them out.
The headline in the Times read “Amid ‘Trump Effect’ Fear, 40% of Colleges See Dip in Foreign Applications.” The article linked to a survey. If you actually follow the hyperlink the opening line of the study reads “39% of responding institutions reported a decline in international applications, 35% reported an increase, and 26% reported no change in applicant numbers.” A slightly different vibe, no?
Cowen, who too charitably called the article a “blooper,” has called for it to be retracted. Yet, you can still find it on the website.
This reminds me of the time the paper characterised a Stanford University study of Michigan’s charter schools with “half the charters perform only as well, or worse than, Detroit’s traditional public schools.” Dig into the report, as the Cato Institute did, and the actual figures are (depending on whether you are testing reading or math skills) 1% or 7% of the charter schools performed significantly worse, 52% or 44% performed basically the same, and 47% or 49% performed better. So, why didn’t the NYT report that nine out of ten Detroit charter schools performed as well or better than their public school peers? Or why not be just neutral, as a true “newspaper of record” should be, and report that roughly half the charter schools performed as well as public schools, roughly half performed significantly better, and a very small percentage performed worse?
Anybody can make mistakes. But mistakes are usually randomly distributed. When the mistakes predominately fall on one side of the distribution, statisticians use a different term. That term is bias.
My wonkier readers will recognize this name as that of an MIT economist who co-authored a widely cited paper entitled “The China Syndrome: Local Labor Market Effects of Import Competition in the United States.”
The benefits of free trade are broadly accepted among economists – in fact, there is probably no economic doctrine that garners wider support. This belief is a logical extension of the manifest benefits from the division of labor and exchange, which clearly improve welfare domestically and have the same effect when extended to the international stage.
However, economists have never claimed that free trade is good for everyone. The profession has always acknowledged that there can be winners and losers, but the gains to the winners have been thought to vastly outweigh the losses (as my very able co-author pointed out here). Economists have therefore tended to downplay this concern.
Autor’s study has shined a light on the losers, something that has taken on greater relevance following the election of Donald Trump (although there is plenty of evidence that most of Trump’s supporters have not been hit hard by globalization). Autor found that import competition could have very substantial local effects, such as on the furniture manufacturing business in South Carolina, and that these are lingering. The study found that “rising imports cause higher unemployment, lower labor force participation, and reduced wages in local labor markets that house import-competing manufacturing industries.”
Overall, Autor found that Chinese competition cost America about 1 million manufacturing jobs between 2000 and 2007, a number that is now a commonplace in political discourse. But this figure is in dispute. A recent editorial in the Wall Street Journal cites two new studies that find much less deleterious results. As pointed out in my recent blog piece on the limitations of statistical analysis in economics, don’t expect an early and definitive conclusion to this debate.
However, even if we take Autor’s claims at face value, an Econtalk podcast with him shows the limitations of his analysis and the questions it leaves unanswered. Here are some of them:
The study made no attempt to identify employment gains elsewhere in the economy arising from the spending power that was liberated through the lower purchase prices of imported goods. This point is straight out of Economics in One Lesson, the classic libertarian book by Henry Hazlitt that reminds us that in economics we must always look beyond the simplistic, first-order effects. In a certain sense, Autor’s study is a more sophisticated variant on the “broken window fallacy.”
The study made no attempt to explain why the unemployed manufacturing workers had, to use Autor’s phrase, “no good outside option.” In other words, why did losing a manufacturing job mean falling off an employment cliff? I have openly wondered about this before and Autor’s study, and his talk on the podcast (“you could argue that they were doing much better in manufacturing than another sector…” – but why?), doesn’t help with the question.
There was also no discussion of the impediments to labor mobility. The rise of Chinese competition can explain a hit to local labor markets, but it cannot explain the lingering effects of this hit. This can only be explained by a lack of labor mobility. Why?
After listening to Autor, I was struck by how the authors brought to bear a great deal of statistical firepower to answer fundamentally limited questions. I was also struck by how the common discussion of this research has left out so much of the detail necessary to understand the results.
In general, I was left with the impression that, even if we grant the conclusions of the research, the “China shock,” in which a huge and very industrious country emerged suddenly onto the international trade scene from behind a wall of communism, was a one-time event that should not dictate future trade policy. We are, once again, significantly at risk of driving the car while looking in the rearview mirror.
Weybridge, United Kingdom
I Wish That I Had Said That…
“All process arguments are insincere” by the journalist Michael Barone, coining a rule of politics that is useful to remember with all the debate around Senate procedures and precedents surrounding the Gorsuch nomination
“Susan Rice operationalized the NSC during the last administration. I was put on to ensure that it was de-operationalized. General McMaster has returned the NSC to its proper function” by Steve Bannon, explaining why he will no longer be a principal on the National Security Council and earning a big “WTF does that mean?” from me
“I think we’ve had one of the most successful 13 weeks in the history of the presidency” by Donald Trump, speaking 11 weeks into his term
 As an aside, it appears that the hyperlink has created a new form of deception. Links give the appearance of support for a claim, but if you follow them, frequently the underlying articles do nothing of the sort and sometimes even show the exact opposite. Deceptive writers rely on the laziness of readers.
 In the statistical sense of this term, I suspect.