A new paper on US union wage effects finds that they’re much bigger at the bottom of the earnings distribution.
New estimates of union wage effects in the U.S
Maury Gittleman and Brooks Pierce
Economics Letters
We present new estimates for union wage effects, using a unique dataset with information on job skill requirements. We find the union wage premium to be smaller at higher skill levels, consistent with the view that unions work to decrease skill differentials.
And here’s their key (ok, only) table, showing that unions raise wages overall by 19 log points (about 20%), but by an even larger amount for low-paid workers.
Estimated union log wage differentials (standard errors in parentheses)
Sample Union wage premium All observations (105,597) .190 (.007) Quintiles  1 .172 (.014)  2 .218 (.013)  3 .206 (.013)  4 .155 (.008)  5 .124 (.018)
I remember a paper from a few years ago by Paul Miller in which he argued that the ‘union wage effect’ was merely a proxy for a ‘large firm effect’ on wages. Seemed to me to be a curious argument; I wonder what he’d make of this particular study.
Matt, they control for firm size, plus a bunch of other things.