Method

How to Tell Whether a Pay Gap Is Really About the Job

A metro pay gap can come from two very different places. Separating them tells you whether you are looking at a cost-of-living story or a policy one.

When a salary comparison says that registered nurses earn 2.3 times more in San Francisco than in Orlando, there are two possible explanations, and they lead to completely different conclusions.

The first is that San Francisco pays everyone more. It does — the all-occupation median there is 1.63 times Orlando's. The second is that nurses specifically are paid disproportionately well in San Francisco, over and above that general premium. That would be a much more interesting finding, because it implies something is setting nurse pay other than the local cost of living.

The two explanations are separable with data that is already public, and the test is simple enough to do by hand.

The test

occupation premium ÷ metro general premium = residual

The occupation premium is the ratio between the highest and lowest metro median for one occupation. The metro general premium is the ratio between the all-occupation medians for those same two metros. Dividing one by the other removes the broad cost-of-living effect and leaves whatever is specific to the occupation.

A residual near 1.0 means the occupation is paid roughly in line with its local market, and the headline gap is a cost-of-living story. A residual well above 1.0 means something occupation-specific is setting pay: a union contract, a staffing mandate, a licensing rule, a state funding formula. That is a policy question, not a market one.

Across the eighteen occupations examined for this series, the residual sorts cleanly. Software developers, accountants and truck drivers sit near 1.0 — their pay tracks the local market. Bartenders, teachers, electricians and nurses sit well above it. The difference between those two groups is not demand. It is who sets the wage.

What the residual tells you

Reading the residual, across the occupations analysed on this site
Residual Reading Examples
0.89–0.95 Pay set in a broad national or regional market accountants and auditors 0.89×, heavy and tractor-trailer truck drivers 0.94×, software developers 0.95×
1.34–1.42 Some institutional effect on top of the local market police and sheriff's patrol officers 1.34×, registered nurses 1.42×
1.52–1.96 Pay is set primarily by policy, contract or mandate electricians 1.52×, elementary school teachers, except special education 1.73×, bartenders 1.96×

The analyses

What this method does not do

  • These are nominal wages. No cost-of-living adjustment is applied anywhere.
  • OEWS surveys employers, so self-employed and contract workers are excluded.
  • Wages only: no bonus, equity or employer benefit contributions are counted. For software and for police pensions, that omission is large enough to change the conclusion.
  • Every estimate carries a sampling margin. Where BLS suppressed a figure, this site does not publish a page at all.