I really appreciate the analysis you share. Thank you.
I wonder, how do you think about the fact that there's cyclicality in U's ability to capture jobseeking? As you know, people become newly employed from either unemployment (a UE transition) or from not in the labor force (an NE transition). These days, only 1 in 4 newly employed people are coming from U and 3 in 4 are coming from N. However, this relationship isn't stable. The UE share of (UE+NE)
rises during recessions and falls during expansions, as can be seen here:
So vacancies per unemployed (V/U) would seem to overstate tightness when the labor market is tight. Including a correction, factor to U -- something like U*(NE/(UE+NE)) -- might give a different picture of how far we are from optimal tightness?
Actually our welfare analysis does not assume a matching function. It does not assume either that only unemployed workers find jobs. It only assumes that a Beveridge curve links unemployment to vacancies (which you can see in the data), and that the size of the labor force is acyclical (which you can also verify in the data). But the framework does not rule out job-to-job transitions or transitions from outside the labor force. These transitions do not feature explicitly because they are not directly relevant to the welfare analysis. In that, we very much follow the sufficient-statistic methodology (https://youtu.be/NSh1amPVlEg). We specify the minimum amount of structure to be able to conduct the welfare analysis. Any structural model that shares this minimum structure will be covered by the analysis.
In our results, the v/u ratio is not meant to capture the job-finding rate of jobseekers. It is here to summarize whether v > u or v < u, which in turn indicates whether the market is too tight or too slack. This characterization would still hold in models with job-to-job transitions or transitions from outside the labor force, as long as they satisfy our basic assumptions (Beveridge curve, acyclical labor force).
However, you are absolutely right that to obtain a model of the labor market that is more realistic than the standard DMP model, it is important to add the pool of workers out of the labor force to the pools of workers employed and unemployed. And indeed this has been an active area of research in the past decade (see for instance https://www.aeaweb.org/articles?id=10.1257/aer.20121662).
In such a model, unlike in the standard model, the v/u ratio might not adequately capture the state of the labor market. I have never seen the correction that you propose in these models, but it would be interesting to see how it performs compared to the standard v/u ratio.
Finally, I was trying to understand why the share of UE in NE + UE is much larger in bad times. This fact might have a simple explanation. Given that the pool of unemployed workers is much larger in bad times than in good times, while the pool of the workers out of the labor force is roughly acyclical, the size of the pool of unemployed workers relative to the total pool of jobseekers will be high in bad times and low in good times. If the job-finding rates of all jobseekers move in tandem, then we expect the share of UE in NE + UE to be much larger in bad times.
I really appreciate the analysis you share. Thank you.
I wonder, how do you think about the fact that there's cyclicality in U's ability to capture jobseeking? As you know, people become newly employed from either unemployment (a UE transition) or from not in the labor force (an NE transition). These days, only 1 in 4 newly employed people are coming from U and 3 in 4 are coming from N. However, this relationship isn't stable. The UE share of (UE+NE)
rises during recessions and falls during expansions, as can be seen here:
https://fred.stlouisfed.org/graph/?g=16Wed
So vacancies per unemployed (V/U) would seem to overstate tightness when the labor market is tight. Including a correction, factor to U -- something like U*(NE/(UE+NE)) -- might give a different picture of how far we are from optimal tightness?
Actually our welfare analysis does not assume a matching function. It does not assume either that only unemployed workers find jobs. It only assumes that a Beveridge curve links unemployment to vacancies (which you can see in the data), and that the size of the labor force is acyclical (which you can also verify in the data). But the framework does not rule out job-to-job transitions or transitions from outside the labor force. These transitions do not feature explicitly because they are not directly relevant to the welfare analysis. In that, we very much follow the sufficient-statistic methodology (https://youtu.be/NSh1amPVlEg). We specify the minimum amount of structure to be able to conduct the welfare analysis. Any structural model that shares this minimum structure will be covered by the analysis.
In our results, the v/u ratio is not meant to capture the job-finding rate of jobseekers. It is here to summarize whether v > u or v < u, which in turn indicates whether the market is too tight or too slack. This characterization would still hold in models with job-to-job transitions or transitions from outside the labor force, as long as they satisfy our basic assumptions (Beveridge curve, acyclical labor force).
However, you are absolutely right that to obtain a model of the labor market that is more realistic than the standard DMP model, it is important to add the pool of workers out of the labor force to the pools of workers employed and unemployed. And indeed this has been an active area of research in the past decade (see for instance https://www.aeaweb.org/articles?id=10.1257/aer.20121662).
In such a model, unlike in the standard model, the v/u ratio might not adequately capture the state of the labor market. I have never seen the correction that you propose in these models, but it would be interesting to see how it performs compared to the standard v/u ratio.
Finally, I was trying to understand why the share of UE in NE + UE is much larger in bad times. This fact might have a simple explanation. Given that the pool of unemployed workers is much larger in bad times than in good times, while the pool of the workers out of the labor force is roughly acyclical, the size of the pool of unemployed workers relative to the total pool of jobseekers will be high in bad times and low in good times. If the job-finding rates of all jobseekers move in tandem, then we expect the share of UE in NE + UE to be much larger in bad times.