This is the third in a series of blog posts on the topic of measuring labor market slack. In this post, we assess the relationships between short- and long-term unemployment and wages by comparing the differences in states’ experiences over the business cycle. While all states felt the impact of the Great Recession, some fared better than others. Consequently, it is possible to use differences in the composition and shifts of short- and long-term unemployment to determine whether short-term unemployment exerts a greater influence on wage determination. The results suggest that there is little difference in how long-term and short-term unemployment affect wages, and as a consequence, the long-term unemployed shouldn’t be dismissed when evaluating labor market slack.
Most models of the labor market tell us that the effect of the unemployment rate on wage inflation is primarily through the wages of new hires, and eventually through the wages of continuing workers as their wages reset over time. The mechanism is straightforward: the presence of a larger stock of unemployed workers relative to the number of available jobs makes it more difficult for an unemployed worker to find a job. This effect lowers the reservation wage for accepting a job offer compared with what it might be in a labor market where it is easy to find a job.
With this framework in mind, we formulate two alternative hypotheses about the extent of wage pressure from the long-term unemployed. If the long-term unemployed are at the margins of the labor market and behave essentially as nonparticipants, we would expect higher levels of long-term unemployment to exert little to no downward pressure on wages. Both job seekers and employers recognize the long-term unemployed as nonparticipants and effectively ignore them during the hiring and wage-setting process. Alternatively, if employers and job seekers still consider the long-term unemployed to be searching for jobs, the long-term unemployed would continue to exert downward pressure on the wages of new hires.
To evaluate our hypotheses, we use data from the Quarterly Workforce Indicators (QWI) published by the U.S. Census Bureau. The QWI comprise consolidated data from individual state unemployment insurance databases covering the vast majority of wage and salaried workers. The pooled state-level data allow us to measure the average monthly earnings of workers in the private sector while distinguishing between the newly hired and workers who continue with the same job. One caveat is that we don’t observe the number of hours worked, so the earnings measures may vary because of changes in both hours and wages. We pool cross sections of the QWI from the first quarter of 1994 through the first quarter of 2013 and merge that data with state-level measures of short- and long-term unemployment, with long-term being defined as more than 26 weeks. The state-level measures of short- and long-term unemployment are computed from the Current Population Survey (CPS) micro data. The CPS, produced by the Census Bureau and the Bureau of Labor Statistics, is the U.S. government’s monthly survey used to estimate unemployment and labor force participation.
We estimate the following regression model
as the average monthly earnings of new hires within quarter t in state s and industry k. We decompose a state’s overall quarterly unemployment rate,
, as the sum of the short-term
unemployment. We find that both the short- and long-term unemployed appear relevant for the wages of new hires, with both measures exerting roughly equivalent effects on earnings. In a variety of specifications, estimates indicate that, if anything, the long-term unemployment rate puts more downward pressure on earnings. In all specifications, we can’t reject the hypothesis that βL = βS.
One concern is that there may be state-level shocks that are driving changes in wages and that are correlated with higher long-term unemployment. To address this concern, we leverage the idea that we expect state-level shocks to affect the earnings both of new hires and of continuing employees. With this in mind, we use the behavior of the average earnings of continuing employees as a control group within state, quarter, and sector. We estimate a model
where the left-hand side is the difference in log average monthly earnings between new and continuing employees within the state, sector, and quarter.
The regression results have a smaller estimate of β on unemployment across various specifications. We expect the estimated coefficients to be smaller since the earnings of continuing workers also respond to high levels of unemployment; they just do so more gradually because only some continuing workers will have their wages reset. Again, the estimated coefficients on the short- and long-term unemployment aren’t statistically different.
In our second blog post, we showed that the labor market outcomes of the long-term unemployed were closer to the outcomes of the nonparticipants who wanted a job than to nonparticipants who did not. Using a broader measure of unemployment that includes nonparticipants who want a job, we estimate the same model for the average monthly earnings of new hires. Across most specifications we find that, similar to the relationship with the long-term unemployment rate, states with higher shares of nonparticipants who want a job also have more downward pressure on earnings.
Overall, there is little evidence in the cross-state data that the long-term unemployed exert less pressure on wages. This finding, as well as the differences between the labor market outcomes of long-term unemployed workers and nonparticipants, suggests that the long-term unemployed should not be dismissed when considering labor market slack. The results also mean that the unusually large gap between short-term and long-term unemployment rates that has existed since the Great Recession doesn’t reduce the ability of the overall unemployment rate to track the state of the economy’s labor market.
The views expressed in this post are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
Robert C. Dent is a research analyst in the Federal Reserve Bank of New York’s Research and Statistics Group.
Samuel Kapon is a senior research analyst in the Bank’s Research and Statistics Group.
Fatih Karahan is an economist in the Bank’s Research and Statistics Group.
Benjamin Pugsley is an economist in the Bank’s Research and Statistics Group.
Ayşegül Sahin is an assistant vice president in the Bank’s Research and Statistics Group.