What makes a job better? Survey evidence from job changers

To maximize the precision of these estimates and to make these results relevant for the universe of job changes, the main analyses include all job changers—voluntary and involuntary. The process of moving jobs is very different, however, for voluntary quits and/or promotions compared with involuntary layoffs. To assess the extent that the results are different for these two types of changes, this section presents separate results for individuals who were and were not laid off in the past year.

First, we show that job changers who were laid off were less likely to say that their new job was better overall or along any other characteristic, with the exception of the physical demands of the job. (See chart 2.) As might be expected, a larger share of those laid off said their new job was worse along all characteristics asked about. Even among those who were laid off, 53 percent said that their new job was better overall, which may reflect the relatively strong job market during the sample period. Additionally, the sample only includes individuals who were laid off but found subsequent employment, which is a selected sample of laid off individuals.

We find that error rates are perhaps slightly lower among those laid off. As shown in table 8, interestingly, the error rates for having a worse characteristic are much lower for those laid off. This implies that having a worse or the same job characteristic is more likely to yield a job with worse or the same quality. This is consistent with those who are laid off getting worse jobs, on average, compared with those who are not laid off, where those who are not laid off may be accepting jobs with one characteristic that is worse or the same, even though the job is better overall.

When we examine the correlations between changes in job characteristics by layoff status in table 9, we find that individuals who were laid off generally experienced stronger correlations in the changes of their job characteristics compared with those who were not laid off.

There are at least two possible reasons why there could be a stronger relationship between characteristics for individuals who were laid off. One explanation has to do with how the decision is structured. A worker who was not laid off will likely have the option of continuing in their current job and will be much less likely to choose a job that is worse off in all characteristics, barring some other forcing event like a strong geographic preference or the anticipation of a layoff. Workers who are laid off, however, could be more likely to accept a job that has worse characteristics, since they are unlikely to have the option of returning to their last job. Another explanation involves differences in the group of workers who are laid off relative to those who were not, since these correlations do not control for characteristics that influence layoff status. For example, the group of workers who were laid off tended to be less educated than the group of workers who changed jobs for other reasons, and table 5 shows that workers with less education have stronger correlations between job characteristics.

Finally, when we redo the regression analysis by layoff status, we find relatively similar results with broadly similar coefficients, with one exception. (See table 10.)

As shown in table 10, better pay and benefits conditional on other characteristic changes is more predictive of a better overall job for those who were laid off. They are 27 percentage points more likely to say the job is better overall compared with those with the same pay and benefits. Those who were not laid off are only 17 percentage points more likely to say their job is better overall if the pay and benefits are better.