The Ins and Outs of Unemployment in the Long Run : A New Estimate for the Natural Rate?

In this paper, we present a simple, reduced-form model of comovements in real activity and worker flows and use it to uncover the trend changes in these flows, which determine the trend in the unemployment rate. We argue that this trend rate has several key features that are reminiscent of a “natural rate.” We show that the natural rate, measured this way, has been relatively stable in the last decade, even after the most recent recession. This was due to two opposing trend changes: On the one hand, the trend in the job-finding rate, after being relatively stable for decades, declined by a significant margin after 2000, pushing trend unemployment up. But the trend in the separation rate has somewhat offset that effect, with a continued secular decline since the early 1980s. We also show that, contrary to the business-cycle movements of the unemployment rate, most of the low-frequency variation in the rate can be accounted for by changes in the trend of the separation rate, not the job-finding rate. The notable exception is during the last decade, when the trend changes in the flows that caused the opposing effects on the trend unemployment rate also implied a slower rate of worker reallocation in the U.S. economy. This slow rate of worker reallocation implies a much slower decline in the unemployment rate over the near term than would have been possible with a high degree of churning, which was a feature of U.S. labor markets before the past decade. We also show that the estimated trend for the unemployment rate is very robust to labor force movements and to the use of different filters, as long as one utilizes the information on the underlying worker flow rates