Erik Hurst, Mark Aguiar, Conny Arvis, Adrien Auclert, Michael Boskin, Matilde Bombardini, Michael Bordo, Michael Boskin, Elizabeth Caucutt, James Coughlin, John Cochrane, Sebastian di Tella, Marcus Ferrara, Jer Garcia, Tom Gilligan, Bob Hall, Rick Hanushek, Caroline Hoxby, Anjini Koshar, Pablo Kurlat, Lance Lochner, Frederic Martenet, Michael Nordeen, Stavros Panageas, Elena Pastorino, Paul Peterson, Monika Piazzesi, Josh Rauh, Martin Schneider, Michael Schoonover, Amit Seru, Gopi Shah Goda, George Shultz, Richard Sousa, Victoria Vanasco, John Taylor, Ramin Toloui, Chris Tonetti, Francesco Trebbi, Sean Xu.
Erik Hurst, V. Duane Rath Professor of Economics at the University of Chicago, and Deputy Director of the Becker Friedman Institute, talked about “The Transformation of Manufacturing and the Decline in U.S. Employment.”
The macro economy has seen two large shifts — the transformation of manufacturing, and the decline in employment particularly among low skill workers. Erik documented both facts, and argued they are linked.
We have had declines in the past such as the 1980s which did not lead to employment declines. Why now?
He concluded with some thoughts about the mechanisms.
First, Erik documented annual hours worked for men 21-55. This declined sharply in the 2008 recession, and has not recovered. Currently we are 180 hours per year below the last peaks. Most of the adjustment are people not working (the “extensive margin”).
Broken down by education levels, much of the decline in work is among the less educated. Even though labor markets have supposedly recovered, the decline in work since 2008 is as large as the total decline in the 1982 recession.
Most striking, the fraction of individuals who do not work at all in a year is up sharply. 20% of men with less than a high school education did not work at all in the past year.
We discussed the obvious questions — what do they do all day, and how do they live. Time use surveys don’t show a big increase in illegal activity.
Caroline Hoxby pointed out that over the long term, people are getting more educated, so people with < 12 years of education are a smaller and less skilled tail of the labor force these days. But she conceded that the surge in non-work in 2008 is too sharp to be just that effect.
The numbers are similar across white, black and urban, rural.
There are similar trends among women, despite the historic increase of women’s labor force participation. The last decade saw the first actual decline in employment among women with lower education levels.
Despite low unemployment (people looking for work), actual work especially in low education groups has declined a lot. Much of the decline in unemployment is people leaving the labor force.
Turning to manufacturing, manufacturing employment also showed a big decline from 2000 to 2010. We are down about 7 million manufacturing jobs since 2000.
There are two stories for decline 1) import competition 2) increasing capital — machines and robots.
Manufacturing output has risen, a fact in favor of the machines story. The ratio of capital services to hours has also grown, and capital intensity has grown. Manufacturing output has grown even though employment has not. We’re doing more with fewer people.
The people who do work in manufacturing are shifting to higher education — they fix and run machines.
Josh Rauh noted big declines in employment and increases in output could be shifts in what we are making, not how we are making the same things.
Bob Hall pointed out that lots of manufacturing employment is clerical. The GM “standard job” is tracking down missing invoices. Perhaps the decline in manufacturing employment is really office automation, not factory floor automation.
Now, are manufacturing and non-employment linked? A plot of men by education level, showing the ratio of those who work in manufacturing + work in construction + don’t work at all / all men is remarkably flat, and high. It accounts for half of all low-education men. It looks a lot like there are shifts between these categories — and the manufacturing decline is entirely the rise in construction and non-work.
The heart of the paper is to explore regional variation. Places where manufacturing used to be are places where employment is low now.
A map of manufacturing share of employment shows the rust belt. The change in manufacturing is strongly related to the initial level. Places with higher shares of manufacturing employment lost more. It looks like the loss of manufacturing employment is pretty evenly spread out.
Hurst wants to associate changes in manufacturing employment and changes in total employment. To that end, he used an instrument to account for the fact that maybe people just stop working. The instrument is the predicted change in manufacturing from the initial level and the national trend.
The result is that places where the employment share of manufacturing went down 10 percentage points show a 3.8 percentage point lower employment share. The result is larger for low skill workers.
While it’s dangerous to add up cross sections, if one does this in a naive way, a big chunk of the aggregate decline is consistent with this cross-region estimates.
Is it trade? Hurst considered a measure of the China trade shock. The employment declines were biggest in industries with trade shocks. But effects on overall employment are about the same. It doesn’t matter where the manufacturing decline comes from.
Hurst compared the current situation with the 1980s. There was also a decline in manufacturing. But in that case places that lost manufacturing jobs did not see smaller shares of people working overall. Those that lost manufacturing jobs either moved or got new jobs.
So Why now?
Hurst investigated several stories. First, Mobility. Americans are moving less than they used to. A good discussion followed on why they might be moving less. Hurst measured that the population response to manufacturing contraction is much smaller now. People don’t move out as much.
Second, is there a persistent skill mismatch? Is the demand for unskilled people just going down everywhere so there is no point in moving? Caroline Hoxby explained a few times that it’s not obvious job opportunities are any better in other places. We debated whether low skill people have flexible skills or not. It would seem that factory worker can do construction. But in Hurst’s view, it’s harder for a gruff factory worker to be a greeter at Walmart. Perhaps some people really are missing central people skills needed to get service jobs.
Third, Transfers and disability? Are people going on disability or otherwise choosing government programs over work? To what extent are government programs including health care stopping them from moving.
Finally, Hurst investigates the interplay between labor markets and opioid use. This (like social programs) can lead to hysteresis — when people lose jobs, they get addicted, then won’t get new jobs. He showed a correlation between places with big manufacturing declines, and increases in opioid prescriptions and overdose data. Manufacturing decline predicts opioid use, prescriptions, and deaths.
- John Cochrane