By David Wiczer, Economist
When workers move from one job to another, economists generally assume it was an optimal choice. However, nearly half of all workers switching employers directly from one job to another experience earnings losses. It would seem that a vast number of workers are making choices that are difficult to justify, at least from an earnings perspective.
As part of my ongoing work with Carlos Carrillo-Tudela and Ludo Visschers, I have been studying the distribution of earnings changes when workers switch from one job to another. We are using data from the Survey of Income and Program Participation (SIPP), which is a short, high-frequency panel dataset that follows workers for about four years and beginning with samples in 1996, 2001, 2004 and 2008.1
We see that the median change in earnings associated with a job-to-job switch is only 2.6 percent.2 This is to say, for the average worker changing employers, earnings in the first full month of employment with the new employer are only 2.6 percent higher than earnings in the last full month with the old one. During the two recessions in this period, the median earnings change associated with a job-to-job change was zero percent.
Overall, 48 percent of workers changing jobs had higher wages with their previous employer than the next. In recessions, this number becomes 50 percent. There are also some very large losses: 20 percent lost more than half of their initial earnings.
Economists have long recognized the importance of job changes to earnings dynamics in part because they have also long understood the reciprocal: Earnings are generally stable among those staying in the same job. Robert Topel and Michael Ward attributed one-third of all early career earnings growth to job changes.3
But this is a statement about the average wage growth, and it would be consistent for many workers to have less or negative earnings growth along with a job change. Their data are also of relatively low frequency, so some of this growth may accrue after the initial month's match. More recently, figures similar to ours have been documented in the U.K., where again about half of job-to-job changes do not result in immediate earnings gains.4
Why do we see so many people take earnings losses in seemingly voluntary transitions? Perhaps the simplest explanation is that earnings are not the only motivation for taking a job. Many times, one job is simply more pleasant than another, and a worker is willing to take a pay cut to enjoy these amenities. This explanation, however, is quite difficult to test for given that the amenities are not often perfectly observable or quantifiable.
Another potential theory harkens to Topel and Ward's evidence that job changes are associated with long-term earnings growth. Workers may be willing to accept a temporary earnings loss for the promise that they will rise further in the future. There are good theoretical reasons why we should expect this.5 If a job is offering learning opportunities or training that will increase earnings in the future, it should be able to pay its new hires less because some of their compensation comes in the form of this training.
Finally, we may be mismeasuring “voluntary” job-to-job changes. Many times, workers may leave employers and take earnings cuts because their original employers gave warnings and negative signals. Or even more concrete mismeasurement: the worker may have been unemployed for such a short time, it was entirely in between the CPS' monthly surveys.
All of these explanations may be at work, but they have a quantitative challenge to explain the motive for nearly half of all job-to-job switches.
1 For our purposes, the SIPP is particularly well suited because workers report their earnings every month. Thus, we can observe the earnings just before and just after a job-to-job change. The other commonly used dataset to study job-to-job changes is the Bureau of Labor Statistics' Current Population Survey. While it has a long-enough panel to observe the employer transitions, earnings data are too limited to see earnings at both the original and destination employer. The SIPP theoretically has both monthly earnings and hourly wage rates. However, because most workers are not paid by the hour, the wage rate is generally imputed based on an imprecise measure of hours. Hence, we use monthly earnings, the more reliable measure.
2 Median figures mean half of the people made larger gains and half made smaller gains.
3 Topel, Robert H.; and Ward, Michael P. “Job Mobility and the Careers of Young Men,” The Quarterly Journal of Economics, 1992, Vol. 107, Issue 2, pp. 439-79.
4 Carrillo-Tudela, Carlos; Hobijn, Bart; She, Powen; and Visschers, Ludo. “The Extent and Cyclicality of Career Changes: Evidence for the U.K.,” Working Paper 2014-21, Federal Reserve Bank of San Francisco, August 2014.
5 For example, see Rosen, Sherwin, “Learning and Experience in the Labor Market,” Journal of Human Resources, 1972, Vol. 7, Issue 3, pp. 326-42.