(Wonkish. Part One of Two. This part covers theory, part two some data.)
Can the number of people quitting their jobs tell us anything useful about slack in the labor market? No. At most, it tells us to focus even more on the stuff we already knew to be watching.
Evan Soltas has argued that the rate at which people quit their job tells us all we need to know about the unemployment rate, in particular how likely it is that the long-term unemployed and people who have left the labor force could be brought back into the labor force. There were several responses from Dean Baker, Jared Bernstein, and Cardiff Garcia.
As a reminder, there are currently 2.5 million Americans who want a job but aren’t looking, and as such are counted as “marginally attached to the labor force” instead of unemployed. There’s a major debate about how much of the collapse in the labor force participation rate is due to de facto unemployment, or people who will come back into the labor force if it gets better (with much of the research pointing to about half).
We should distinguish between what quit rates can tell us about the employed and what they can tell us about the status of the unemployed. What’s the theory of quits in which they tell us what the “true” unemployment rate is? Soltas: “Think about the decision to quit. It's a function of your confidence that you'll find a better job quickly -- which embodies some unobserved but holistic measure of labor-market tightness.” If quits go up, then the long-term unemployed, in this view, “no longer [have] the power to restrain wage growth or discourage the employed from quitting and switching jobs.”
If I understand this correctly, this assumes that quits are a measure of people first becoming unemployed and then looking for a new job (trying to “find a better job quickly”).[i] But I don’t think this is right. A very large percentage, perhaps half, of quits are people moving job-to-job rather than becoming unemployed.[ii]
Under Soltas' view, quits make it more difficult for the existing unemployed to find jobs by increasing unemployment. But doesn’t a quit create a job opening? As such, doesn’t it create an opportunity for an unemployed person, rather than a limitation? Also, even if the employed became unemployed, would that mean that the long-term unemployed couldn’t find work, or just that they are at the end of a very long line?
Let’s try to develop a more firmly grounded theory of quits. What can quits tell us about unemployment as a matter of economic logic? Well, first off, every person who quits his or her job creates a job opening. Someone has to do the work that person was doing. So we should think of “vacancy chains” -- a term made famous by George Akerlof, Andrew Rose, and Janet Yellen (!) back in 1988 -- and their characteristics.
Suppose Amy works at Acorp, and she leaves her job to take a new one. Acorp now has a job opening. They hire Bill away from Bcorp. Now Bcorp has a job opening. So they hire Charlotte from Ccorp. Now Ccorp has a job opening. They hire Dan from Dcorp. Now Dcorp has a job opening.
Note: Amy, Bill, Charlotte and Dan are all counted as “quits,” though they wouldn’t count as having been unemployed.
So Dcorp hangs a sign that says “help wanted.” They can do one of three mutually exclusive things. They can (1) hire an employed person, which will create another quit, which will create another vacancy, which Erin from Ecorp will fill. Thus the vacancy chain is extended one more step. They’ll likely have to pay the employed person a higher wage than that person currently makes to change jobs (or at least in aggregate this will happen). Or they can (2) hire an unemployed person, which will not create another job opening, but instead just close the vacancy chain.[iii] Or, if they can’t fill the job with an employed person or an unemployed person, they (3) leave the job opening unfilled.
So let’s rephrase that. The only thing interesting about the quits rate in this context is what it is implying about the length and conclusion of vacancy chains. If the length of the vacancy chain is infinitely increasing, then wage growth must be skyrocketing. And if the unemployed aren’t capable of taking jobs and closing the vacancy chain, then the number of job openings must rise relative to the unemployment rate.
Which means the interesting things the quits rate tells us about unemployment are entirely captured in wage growth or the number of job openings relative to unemployment (i.e. the Beveridge Curve).
So where are we? Wage growth is still weak, and nowhere near what it was in the late 1990s. Meanwhile, the Beveridge Curve shifted at the height of the crash, but has been remarkably consistent since. The shifting of the Beveridge Curve has been thoroughly debated, with the complicating issues of circularity and endogenous firm search at the forefront. What stands out for me is that the Beveridge Curve hasn’t deteriorated in the past several years, which implies that the increasing duration and prominence of the long-term unemployed in the labor market isn’t showing up in increased job openings.
The quits rate is important. It hints at the lived experience of working throughout the Great Recession and points to the stagnating wages workers face. But from the point of view of the status of the unemployed, it doesn’t tell us the “true” unemployment rate. Rather, it is important because it flags the things we should be watching -- things that are slowly improving, as long as the Fed doesn’t put a stop to it.
[i] Soltas also has a statistical relationship between the two based over the past two business cycles, but if we can’t think of a good theory of what quits tell us about unemployment I’m not sure what weight we should put on this relationship.
[ii] The Census will start releasing job-to-job transition data in late 2014. Hurry Census! We need this data to be publicly digestible.
[iii] This is why Akerlof, Rose, and Yellen believe quits to be counter-cyclical to unemployment, as opposed to the pro-cyclicality predicted by search models. There are more unemployed hanging around to close chains earlier in a recession, reducing the number of quits necessary to close a vacancy chain.