Wednesday, February 16, 2011


Some people seem to be paying attention to this piece by Justin Weidner and John Williams at the San Francisco Fed. For example, someone at The Economist picked up on it. The FRBSF Economic Letters posts, of which the Weidner/Williams piece is one, are short, accessible summaries of research, but in this case there is not much research to back it up. There's really no news here.

I have written on sectoral reallocation issues in earlier posts, for example here and here. To summarize, there are some important labor market anomalies associated with the recent recession. The unemployment rate has risen much more, and employment has dropped much more than is typically observed in conjunction with a drop in real GDP of this magnitude. Further, in spite of the (somewhat slow) recovery in real GDP, employment is still in the cellar. As a result, there has been a large anomalous increase in productivity coming out of the recession.

My casual observations in previous posts, which should be interpreted as suggestive, and not serious research, tell me that there could be something interesting going on related to the the sectoral reallocation of labor and other factors of production. Another important anomaly is in labor market performance in Canada. The magnitude of the real GDP decline in Canada during the recession was similar to what it was in the US, but the unemployment rate (typically a couple of percentage points higher in Canada) is currently 7.8%. The sectoral reallocation process could help to explain the anomalies. To address this seriously, I think we have to go back to 2000 or earlier, and think about: (i) the secular reallocation of labor from manufacturing to services; (ii) secular labor reallocation across geographical regions; (iii) the effects of incentive problems in the mortgage market after 2000, which increased the share of employment in residential construction until 2006, and set the stage for the subsequent reduction in employment share in that industry; (iv) the role played by financial factors in the geographical and sectoral dispersion in employment.

To successfully disentangle what is going on, we need a model with sufficient disaggregated detail to capture the regional and sectoral allocation of labor, and search and matching among firms and workers. Activities of the economic agents in the model should be such that we can ask these agents the same questions that are asked in the BLS household survey, and have the answers make sense. Basic Mortensen-Pissarides won't do the trick, unless we try to somehow load all this sectoral detail into the matching function, which seems unsatisfactory.

Once such a model is up and running, what do we want to do with it? Here are some questions: What would have happened if the mortgage market had somehow been differently organized and regulated? How would the US economy have looked different, starting in 2000? Could there have been better policy responses to the financial crisis and the recession than what we had? Is the US unemployment insurance system adequate (actually, most research on optimal unemployment insurance already says it is far from perfect)? Does the financial crisis tell us something new about labor market policies? I'm hoping that people who know how to do these things are working on the problem. Please let me know if there is some work out there that I am missing.

What I am sure of is that gazing at some Beveridge curve scatter plots, pondering the "natural" unemployment rate, or trying to decompose unemployment into so-called "cyclical," "structural," "demand-deficient," "involuntary," "voluntary," "green," "yellow," "purple," or "chartreuse" components will not answer these questions.


  1. You say that your 'casual observations...tell me that there could be something interesting going on related to the the sectoral reallocation of labor and other factors of production.' David Autor and others have documented extensively that this has been going on for decades but what has it got to do with this recession? In which sectors has unemployment decreased and/or wages gone up as a result of sectoral reallocation during this recession? And if it is 'related to the the sectoral reallocation of labor and other factors of production' why has it not resulted in a recession until now as this sectoral reallocation has been going on for decades?

  2. 1. Look at employment in health services and education. That continued to grow through the recession.
    2. I'm not saying that secular sectoral reallocation caused the recession. I'm saying that it may have contributed to the dynamics we are seeing in the labor market. Regarding the financial crisis and the recession, a key sectoral aspect is what has happened since 2000 in the housing and mortgage markets relative to the rest of the economy, and the geographical sectoral aspects of that. The regions of the economy hardest-hit by the recession were those where there was most action in incentive-problem mortgages, and the biggest housing market crashes.

  3. What about the Samuelson-Jones specific factors model from international trade theory? Couldn't somebody fuse matching theory with the specific factors model to get a coherent model that might explain our structural unemployment?

    Also, 1st anonymous doesn't seem to be aware we've just gone through a massive misallocation of resources to housing, which is unprecedented in its scale over last 40 years.

    anon from tx

  4. Yes, in principle you could use some ideas from trade theory, as part of what is involved is inter-regional trade and issues of factor mobility.

  5. "there has been a large anomalous increase in productivity coming out of the recession"

    Steve, I don't follow. Looking at the productivity (output/hour basis, non-farm business sector) data since inception, I see that the rate of productivity growth increased in every recession, and the increase continued after the recession ended.

  6. anonymous:

    See this post:

  7. Thanks, I had read that post. I would suggest that there were a number of changes in the 80s/90s that might have affected productivity dynamics, e.g. the globalization of the labor market, the increased focus on shareholder value after the Milken/LBO boom, and the use of technology in the workplace.