From More Openings to Better Jobs: Reading the Evidence

The Federal Reserve study’s central empirical question is whether a labor market with more openings changes what workers do and what they receive from a job change. The evidence is easiest to understand in two stages: first, does greater tightness increase search and movement; second, do the resulting moves improve job quality? Reading the estimates in that order prevents a common mistake—treating a rise in vacancies as if it automatically meant that every new job was better.

The paper reports estimates for a 10 percent increase in state-level JOLTS openings per thousand residents. That phrase needs care. The estimate is not a prediction that each worker’s openings will rise by exactly 10 percent, nor does it mean a ten-percentage-point increase in the probability of an outcome. It is a convenient way to express the coefficient on the logarithm of openings per capita. Technically, a 10 percent change corresponds to a log change of about 0.095, which the authors approximate as 0.1 for presentation.

The results are also estimates from a particular design using post-COVID local and temporal variation, geography fixed effects, and individual controls. They are evidence about a relationship the authors interpret as plausibly causal, not a universal constant that applies to every worker, occupation, or future labor market.

More openings expand search and movement

The first set of outcomes covers prime-age respondents regardless of whether they were employed at the survey date. A 10 percent increase in job openings per thousand state residents is associated with a 0.5 percentage-point increase in the probability of applying for a new job. The probability of starting a new job rises by 0.7 percentage points. The probability of quitting rises by 0.8 points, and the probability of changing jobs rises by 0.8 points.

Those percentage-point changes become easier to interpret when compared with the average incidence of each event. The paper reports proportional increases of about 1.6 percent for applying, 3.9 percent for starting a job, 7.0 percent for quitting, and 6.7 percent for changing jobs. The same 10 percent increase in openings is associated with a 0.3-point, or roughly 4.3 percent, reduction in the probability of being laid off. Tighter markets also predict a greater likelihood of asking for and receiving a raise, although the central point is the direction of the bargaining response rather than one single raise estimate.

The distinction between percentage points and percentages is not bookkeeping. Suppose an outcome has a baseline probability of 10 percent. An increase of 0.8 percentage points moves it to 10.8 percent, which is an 8 percent proportional increase. If the baseline were 40 percent, the same 0.8-point change would be only a 2 percent proportional increase. The paper’s reported proportional effects therefore depend on the mean of each outcome. A reader who sees “0.8” and silently turns it into “8 percent” can badly misread the scale.

The pattern is consistent with the offer-arrival mechanism. More openings are associated with more applications, then more starts and job changes. The effects are not identical because each step has additional conditions. A person can apply without receiving an offer.