Is the U.S. Labor Market Becoming More Sclerotic? And Does It Matter?
The U.S. economy continued to exhibit signs of a painfully slow recovery in the last month. In particular, labor markets seem to be front and center in setting the sluggish pace. One long-term trend that was confirmed in the latest recession is that there is less churning in the labor market. My colleague Murat Tasci keeps track of long-term trends in worker movements into and out of jobs, and his findings show that job-finding rates have been declining since the 1950s, while separation rates have exhibited the same tendency since the early 1980s.
Reduced rates of worker turnover are usually associated with the more regulated labor markets of some European countries, and it is because of their slower turnover that those labor markets have been characterized as sclerotic. Now that it is clear that the U.S. labor market is becoming more sclerotic, too, the question is whether it matters, in the sense of whether it is detrimental to the growth of the U.S. economy. So far, the answer appears to be no. There are at least three ways in which reduced worker flows might be harmful to the economy, and none appears to be showing up in the U.S.
The first is that reduced turnover may contribute to higher unemployment. While the unemployment rate has certainly gone up in the last four years, its trend was clearly negative through the early 1980s and up until the mid-2000s. In fact, low job-finding rates can be compatible with either high or low unemployment rates. It all depends on whether the separation rates are high or low.
Second, lower turnover could reflect the fact that searching for jobs and workers has gotten costlier or harder, resulting in poorer worker-job matches, and therefore lower productivity. Firms and workers are involved in a bilateral sorting and matching process, and high rates of churning can be an indication that the process is working, while low rates would indicate that either firms or workers (or both) are settling for matches that are substantially suboptimal. What is optimal, of course, depends not only on the quality of the match, but also on the costs of searching. If it is the case that search costs have increased in the U.S., then one should see poorer-quality matches, as firms and workers settle earlier in the searching process. But when we look at labor productivity as a proxy for match quality, we get no such picture. Labor productivity has been growing at a fairly constant rate. The same is true, by the way, when we look at multifactor productivity measures, which take into account not only how productive labor is but also how productive capital is at the same time.
A different story, one that seems to better match the data, is that firms and workers have become increasingly more efficient at sorting and matching, and therefore they are able to generate similar quality matches while searching less. This efficiency should come as no surprise when one considers that information dissemination in job searching (for both firms and workers) is not even remotely comparable to what it used to be 30, or even 20 years ago, when the internet was not around.
Finally, the third way reduced turnover might harm the economy is by increasing unemployment durations. The idea is that diminished worker flows may give rise to longer unemployment spells, which in turn would imply larger skill losses and lower productivity once long-unemployed workers get a job. Again, note that decreasing worker flows do not necessarily imply longer median unemployment spells. Unemployment duration only started to increase in any meaningful way in the mid-2000s, while worker flows have been decreasing since the mid-1980s.
One final word regarding the effects of long unemployment spells. While their negative effects are clear, and that is what economists focus on most of the time, let me play devil’s advocate and point to some theoretically possible positive effects. I say theoretical because I know of no research that empirically supports my hunches, nor am I arguing that these effects dominate the negative ones. But if long unemployment spells fall disproportionately on those whose skills are becoming irrelevant or obsolete, a long period without a job might work to provide workers with the right incentives to retool. If I am an unemployed typewriter repairman who is going through a spell of unemployment that I estimate will last a week or two, I have less of an incentive to go back to school and become a Java programmer, say, than I would if I estimated the unemployment spell would last a whole year. To understand the importance of this effect, we will need to know more about who the long-term unemployed are.