We Expected States With More Tradespeople to Pay Them Less. The Data Says No
This is a methods piece. It reports a null result from our own data rather than a finding.
The theory
It is a reasonable thing to believe. If a state has three times as many roofers per resident as its neighbour, competition should push the price of a roofing hour down.
We have employment counts and wages for ten trades across 51 states, so this is testable.
The test
| Trade | Correlation (employment vs pay) |
|---|---|
| Carpenters | +0.32 |
| Floor Layers | +0.24 |
| Painters, Construction and Maintenance | +0.15 |
| Drywall and Ceiling Tile Installers | +0.12 |
| Tile and Stone Setters | +0.09 |
| Roofers | +0.06 |
| Plumbers, Pipefitters and Steamfitters | +0.03 |
| Construction Laborers | -0.02 |
| Electricians | -0.04 |
| Cement Masons and Concrete Finishers | -0.07 |
A correlation of zero means the two numbers move independently. A positive number means more tradespeople go with higher pay, which is the opposite of the theory.
Carpenters are the strongest at +0.32, and that is almost certainly a size effect: big states have both more carpenters and higher wages, and the two are being read as a relationship.
Why the theory fails
Geography is doing the work. Florida and Texas have the most roofers and below-average pay. Illinois has a twentieth as many and the highest pay. Both patterns are explained by state size and local price levels, not by a supply curve.
Wages are sticky in both directions. A market with too many roofers does not see pay fall quickly, because the workers leave or move to other trades first.
The buyer does not see the market. You get three quotes, not thirty. The price you pay reflects those three contractors' backlogs, not the state aggregate.
What this means for the site
We do not use employment counts to adjust any calculator on this site. The adjustment is the state wage median, and nothing else.
It is worth saying plainly: a plausible theory that the data does not support should be reported as unsupported. That is the difference between an analysis and an argument.
Frequently asked
So what does drive prices between states?
Wage levels and local price levels, in that order. Employment counts do not add explanatory power in this dataset.
Could the correlation be real but hidden?
Possibly. A per-capita measure would be better than a raw count, and we do not have state population in this dataset. That is a limitation, stated rather than papered over.
Updated October 2026