This post is not about dating at work, though dating is a good way of thinking about a phenomenon that used to puzzle economists: why are some jobs vacant at the same time that some workers are unemployed? The explanation uncovered by research is that, just like in dating, before committing to a relationship each party wants to verify that the counter-party is a good match, that they possess the qualities that are important to them. For example, rather than hire the first person that responds to its ad, a firm will typically wait until it has a large-enough pool of applicants. Then, it will screen this pool by reviewing resumes, conducting interviews, and so on. Similarly, a job-seeker will typically not take the first job available. Instead, they will apply for several jobs and wait until they have found out more about the salary, location, requirements and potential for advancement of each job before deciding which offer, if any, to take. Because this process is lengthy, in the meantime vacant jobs remain unfilled and job-seekers remain unemployed.
One may argue that while this may be true in the short run, given enough time eventually all vacancies will be filled so any unemployment that persists past that point must be due to a lack of jobs. This would be true if workers that found a job kept it for the rest of their working life. The reality is, however, that as in love so in employment, relationships do not always survive the test of time. Currently, in a typical month about 5.5 million unemployed workers will be matched with a vacant job. However, during the same month approximately an equal number of workers will become separated from their job. Eventually, these or other newly vacant jobs and these or other newly jobless workers will start looking again for a match. Because this match-and-break-up process takes place every month, vacant jobs and unemployed workers will always coexist, though it is important to remember that the vacancies and job-seekers of today are not necessarily the same as those of previous months.
Observing how the job-search process works helps us understand why unemployment goes up during recessions and down during expansions. The data suggest that this phenomenon is due to fluctuations in the number of job openings. During good times, firms post more vacancies so it takes less time for a job-seeker to find the "right" job. Hence, unemployment declines. During a recession, firms reduce their hiring so it takes longer for a job-seeker to find a job. This inverse relationship between the vacancy rate, the percentage of jobs that are vacant, and the unemployment rate, the percentage of workers that are looking for employment, can be plotted to produce the so-called Beveridge curve. Below, I graph the curve corresponding to the time-period from December 2000 (when the Bureau of Labor Statistics started conducting the Jobs Openings and Labor Turnover Survey) until June 2003, which is when the unemployment rate peaked following the 2001 recession. The vacancy rate is on the vertical axis and the unemployment rate on the horizontal. As can be seen, the curve is downward sloping capturing the inverse relationship between the vacancy rate and the unemployment rate. The left end-point, corresponding to December of 2000, is affiliated with a high vacancy rate and a low unemployment rate of 3.9%. The right end-point, corresponding to June 2003, is affiliated with a low vacancy rate and a high unemployment rate of 6.3%.
What happens if we extend the time-period to show the recovery? The Beveridge curve below plots the data from December 2000 to November 2007. The left end-point still corresponds to December of 2000. We see that from 2003 until the end of 2007, vacancies rise and unemployment declines. Graphically, the economy starts moving up and to the left along the same curve.
In 2008, the economy was hit by a recession so severe that it has been named, The Great Recession. How does the Beveridge Curve look if we extend the time-period to capture its effect? Below, the left end-point still corresponds to December of 2000 but the right end-point corresponds to October 2009, which is when the unemployment rate peaked. Starting in December of 2007, the economy starts sliding down along the same curve, as the vacancy rate begins to fall and the unemployment rate rises. Compared to the 2001 recession, the rate of vacancies experiences a bigger drop resulting in a bigger increase in the unemployment rate. The economy passes the right end-point of the 2001 recession and stops at 10%.
Based on the behavior of the economy up to this point, what should we expect to see happening if we extend the time-period to capture the recovery from the Great Recession? A good guess would be that the economy should move up and to the left along the same curve, as it did when it recovered from the 2001 recession. However, if we plot the data all the way until May of 2018, we get this instead. Like before, as firms open more vacancies it becomes easier for job-seekers to find a job so the unemployment rate declines. Unlike before, instead of moving along the same curve, the economy starts moving along a new Beveridge curve that lies to the northeast of the old.
What does the shift of the curve signify? It means that since the end of 2009, there have been more unemployed workers for each vacancy, or alternatively there have been more vacant jobs for each unemployed worker. This implies, in turn, that it takes more time for a given number of unemployed workers to fill a given number of vacancies. In other words, there seems to be greater mismatch between those looking for a job and the jobs available. Although the unemployment rate, currently at 4%, is low by historical standards, it is still higher than what we would have expected based on our previous experience given that there are so many vacant jobs. One possible explanation for the mismatch is that the vacant jobs are located in different geographic areas than the bulk of unemployed workers, and that these workers are reluctant to move. Another explanation is that the existing vacant jobs require skills that are different from the skills that most jobless workers possess. Examining the validity of each explanation is beyond the scope of this post. The shift does suggest, however, that to understand current labor market outcomes we should stop viewing the labor market as homogeneous and instead focus on how the characteristics of the job-seekers vary relative to those of the available jobs.
One may argue that while this may be true in the short run, given enough time eventually all vacancies will be filled so any unemployment that persists past that point must be due to a lack of jobs. This would be true if workers that found a job kept it for the rest of their working life. The reality is, however, that as in love so in employment, relationships do not always survive the test of time. Currently, in a typical month about 5.5 million unemployed workers will be matched with a vacant job. However, during the same month approximately an equal number of workers will become separated from their job. Eventually, these or other newly vacant jobs and these or other newly jobless workers will start looking again for a match. Because this match-and-break-up process takes place every month, vacant jobs and unemployed workers will always coexist, though it is important to remember that the vacancies and job-seekers of today are not necessarily the same as those of previous months.
Observing how the job-search process works helps us understand why unemployment goes up during recessions and down during expansions. The data suggest that this phenomenon is due to fluctuations in the number of job openings. During good times, firms post more vacancies so it takes less time for a job-seeker to find the "right" job. Hence, unemployment declines. During a recession, firms reduce their hiring so it takes longer for a job-seeker to find a job. This inverse relationship between the vacancy rate, the percentage of jobs that are vacant, and the unemployment rate, the percentage of workers that are looking for employment, can be plotted to produce the so-called Beveridge curve. Below, I graph the curve corresponding to the time-period from December 2000 (when the Bureau of Labor Statistics started conducting the Jobs Openings and Labor Turnover Survey) until June 2003, which is when the unemployment rate peaked following the 2001 recession. The vacancy rate is on the vertical axis and the unemployment rate on the horizontal. As can be seen, the curve is downward sloping capturing the inverse relationship between the vacancy rate and the unemployment rate. The left end-point, corresponding to December of 2000, is affiliated with a high vacancy rate and a low unemployment rate of 3.9%. The right end-point, corresponding to June 2003, is affiliated with a low vacancy rate and a high unemployment rate of 6.3%.
What happens if we extend the time-period to show the recovery? The Beveridge curve below plots the data from December 2000 to November 2007. The left end-point still corresponds to December of 2000. We see that from 2003 until the end of 2007, vacancies rise and unemployment declines. Graphically, the economy starts moving up and to the left along the same curve.
In 2008, the economy was hit by a recession so severe that it has been named, The Great Recession. How does the Beveridge Curve look if we extend the time-period to capture its effect? Below, the left end-point still corresponds to December of 2000 but the right end-point corresponds to October 2009, which is when the unemployment rate peaked. Starting in December of 2007, the economy starts sliding down along the same curve, as the vacancy rate begins to fall and the unemployment rate rises. Compared to the 2001 recession, the rate of vacancies experiences a bigger drop resulting in a bigger increase in the unemployment rate. The economy passes the right end-point of the 2001 recession and stops at 10%.
Based on the behavior of the economy up to this point, what should we expect to see happening if we extend the time-period to capture the recovery from the Great Recession? A good guess would be that the economy should move up and to the left along the same curve, as it did when it recovered from the 2001 recession. However, if we plot the data all the way until May of 2018, we get this instead. Like before, as firms open more vacancies it becomes easier for job-seekers to find a job so the unemployment rate declines. Unlike before, instead of moving along the same curve, the economy starts moving along a new Beveridge curve that lies to the northeast of the old.
What does the shift of the curve signify? It means that since the end of 2009, there have been more unemployed workers for each vacancy, or alternatively there have been more vacant jobs for each unemployed worker. This implies, in turn, that it takes more time for a given number of unemployed workers to fill a given number of vacancies. In other words, there seems to be greater mismatch between those looking for a job and the jobs available. Although the unemployment rate, currently at 4%, is low by historical standards, it is still higher than what we would have expected based on our previous experience given that there are so many vacant jobs. One possible explanation for the mismatch is that the vacant jobs are located in different geographic areas than the bulk of unemployed workers, and that these workers are reluctant to move. Another explanation is that the existing vacant jobs require skills that are different from the skills that most jobless workers possess. Examining the validity of each explanation is beyond the scope of this post. The shift does suggest, however, that to understand current labor market outcomes we should stop viewing the labor market as homogeneous and instead focus on how the characteristics of the job-seekers vary relative to those of the available jobs.
