Well I finally went on a long run and so got to listen to the 72 minute long podcast of Russ Roberts talking with MIT economist David Autor.  Disappointing. Mostly the blame falls on Russ Roberts, and he basically admits it himself during the podcast and on the website. He spends the first 45 minutes engaging Autor in an undergraduate-level discussion of intermediate trade theory.  Talk about comparative advantage gone wasted: Autor is an empirical economist, after all.

Finally about 45 minutes into the podcast you get some facts.  Surprisingly few, though.  U.S. employment in manufacturing has been declining for 70 years, since the 1950s.  China’s share of manufacturing in global trade and in manufacturing value added increased a lot, from 1990 to 2010.  Autor and co-authors estimate three numbers it seems.

1. The magnitude of the initial displacement in manufacturing employment.  Autor and co-authors find that 25% of decline in manufacturing employment over period 1990-2007 and 45% of the decline in manufacturing over the period 2000-07 was due to China productivity-competition effect (i.e. import competition from Chinese firms).  Autor then says, “In net that would be a reduction of one and a half million manufacturing jobs.”
2. The increase in transfer payments attributable to the displacement.  Autor says, “Every thousand dollar increase of import penetration leads to approximately about a $58 increase in transfer payments.” The podcast finally gets to a discussion of the length of time of this displacement at 1:01.  Autor cites a paper using Social Security data.
3. The length of time for “adjustment” to the displacement.  On average top third of earners “do fine” changing employers.  Bottom tercile “lose employment and earnings in way that are not able to recover from.”  “Over the long-term… on average the effects are not that large at the individual level… among the people who initially have lower earnings levels the effects are very pronounced.”

As I listened, I wondered about the magnitude of the 1-2 million jobs lost in manufacturing.  Roberts unfortunately does not clarify and neither does Autor: is this an average displacement per year, or the total displacement over 20 years of China’s manufacturing imports?  The context (and other references to the work) suggests it is the total net cumulated job loss over 20 years.  So 200,000 a year, say.  That just doesn’t sound like a big number in an economy like the United States, where job openings at any given time are on the order of magnitude of 4 million.  Another way of saying that is that about 5 million people move jobs every month.  Net job creation in the U.S. economy is about 200,000 every month (since 2010).  Autor’s estimates sound like about 5-10% of all displacement per year is due to China.  A lot of that would have happened anyway, as automation and import competition from other countries would have led to job displacement.  So I think maybe we are talking about 2-5% of all displacement is due to one of the biggest economic changes of the past 30 years.  Again, that seems fairly small, to me.  “Small” meaning that it seems like what I would have guessed… is anyone surprised by this number?  200,000 a year means 200 factories a year close (employing 1,000 each), due to trade with China.  Sure, viral videos of factory closing are heart-breaking.  But my local hair and nails parlor closed when the boss retired, and her two assistants were displaced.  Heartbreaking as it is, that closing, like the closing of 200 factories in an economy of 150 million workers, does not suggest the need for a major shift in public policy.  A modest shift.  But not a “Oh my we’ve been wrong for 20 years” shift.  And definitely not a reason to elect Donald Trump with his promises of 35% tariffs on Mexican manufacturing exports.

On the transfer payments.  Say annually there is increase in import penetration of $100 billion from China.  So the increase in transfer payments is on the order of magnitude of $6 billion per year.  Say we cumulate this over 20 years.  Realistically, of course, the initial transfer payments decline after a few years as people do get jobs, or start collecting social security, etc. So maybe we are talking about maybe $50 billion in annual transfer payments.  Presumably this is the peak, since Chinese manufacturing imports have started to slow in recent years.  Given the Federal, state and local government transfer budgets, this is about 2% of all government transfer payments.  So the biggest economic “shock” to the economy, Autor is saying, has a total negative effect of about 2% per year on transfer payments (and probably has already peaked).  My take again is that doubling transfer payments to displaced workers is well within the budget possibilities (especially when combined with dropping a few United Technologies weapons programs, which of course would displace a different group of manufacturing workers).

Coming back to the magnitudes again. Suppose the job loss indeed is 1.5 million. Suppose it is the bottom tercile that is significantly affected for a long period of time.  Suppose $20,000 per year for these workers is a reasonable amount to cushion the shock.  Then we are talking about 500,000 x 20,000 =  $10 billion per year in trade adjustment assistance.  For obvious reasons, this is a huge budget issue, with TAA being opposed by Republicans (they would rather have tariffs than adjustment assistance?).  The current budget for TAA seems to be about $1 billion.  My point is that $10 billion just isn’t that much.  (One stealth bomber apparently costs about $1 billion.)

Remember, this discussion of the magnitudes of the displacement and the transfer payments is not counting any of the positive effects (see below), in the United States or in China.  (To be fair to Autor, he does mention and I agree that job displacement has some pretty negative psychic effects. That is why TAA is a good thing, if well-implemented.)

All in all, to me Autor’s (and coauthors) findings sound like a pretty convincing case that the displacement costs are actually fairly small.  Should we not be pursuing even greater trade liberalization?  Autor by the way apparently is in favor of TPP.  So really the argument is about what should be very reasonable reform and increase in TAA, but instead has become this way-out-of-proportion political football.

So that was my take.  A very good podcast for students in an intermediate trade class.

Here are a few other points.

• Autor says he is perfectly willing to concede the empirical work is not doing a good job measuring the indirect trade creation that may accompany the rise of trade in China, because that measurement is probably very hard.  He is also clear that he and co-authors do not measure the consumer benefits from the imports.  Finally, he is frank that they are not measuring the benefits to workers in China of the huge increase in value-added.  All these omissions, Autor notes in a throwaway line, mean that the case for continued trade liberalization is probably very strong… he is definitely not, in the podcast or paper, apparently, making a case for protectionism as policy.)
• Autor notes that China and trade are significant, but that technology has played a very big part, “technology is the most important,” he says, “technology is the more important.”
• Roberts does not ask for, and Autor does not offer (in the podcast), a how the job displacement might be concentrated by geographic, ethnic, gender, and age.  They never discuss the demographic adjustment needed when estimating the welfare implications of the displacement.  Was this displacement concentrated among the older work force (say above 50?).  If so, from a public policy perspective perhaps it should be addressed more by transfer payments than by relocation incentives? Autor suggests as much saying disability and early retirement and exit from labor market have been a big part of the adjustment mechanism.
• On the slow adjustment to displacement, Autor proffers the very interesting possibility that men are much more reluctant to enter the “personal services economy.”   Of course, it sounds like blaming the victim.  And a perusal of cooking shows suggests that men are very happy to be cooks.  And as Autor mentions, the building trades (plumbing, solar roof installing) remains vibrant service sector jobs.
• It is convenient for Autor, and journalists have probably put him in this fairly shady territory, where he tells an “intellectual history” story about how economists ignored these displacement costs… that nobody measured displacement before, and nobody ever cared about adjustment before, because in the Bretton Woods era there was little displacement due to trade.  As an undergraduate in the 1980s who took undergraduate trade, we certainly talked about adjustment costs.  I remember reading the Japan, Inc. comic book in late 1980s!  OK so that is after Bretton Woods.  So Autor is telescoping economics intellectual history to a 25 year period (1945-70)?  Even that isn’t true.  A quick search on JSTOR for articles on trade and unemployment finds a few nuggets. (Incidentally, some practically replicate word for work the undergraduate-level discussion that Autor and Roberts had for the first 45 minutes of the podcast.  Maybe there is some room for job displacement of economists who are still saying the same thing after 40 years?).
  • The Trade Adjustment Bills: Their Purpose and Efficacy, Bruce E. Clubb and Otto R. Reischer, Columbia Law Review, Vol. 61, No. 3 (Mar., 1961), pp. 490-503
  • The Political Economy of Steel Import Quotas,Ferdinand L. Molz, Journal of Economic Issues, Vol. 4, No. 2/3 (Jun. – Sep., 1970), pp. 60-76
  • Are National Full-Employment Policies Consistent with Freer Trade? Clark Lee Allen, Nebraska Journal of Economics and Business,Vol. 8, No. 1 (Winter, 1968/1969), pp. 3-15
  • Coal and Steel Community Policies for Averting Unemployment,Monthly Labor Review, Vol. 86, No. 10 (October 1963), pp. 1139-1144.
  • Full Employment, Trade Expansion, and Adjustment Assistance, Tracy W. Murray, Michael R. Egmand, Southern Economic Journal, Vol. 36, No. 4 (Apr., 1970), pp. 404-424