Why did we think wages are rigid for all those years?

Thelarge spike at zero in the distribution of year-to-year nominal wage changes in household surveys is often seen as evidence of nominal wage rigidity. But measurement error—especially from workers rounding their reported wages—can exaggerate this spike. Using U.S. Current Population Survey data, we adjust for potential rounding behavior and find that the zero-change spike falls from 15–20 percent to 7–12 percent, aligning closely with recent estimates from administrative data.

That is from See-Yu Chan, Stephan Hobler, and Thijs van Rens.  Note that Hobler, from LSE, is on the job market this year (with a different job market paper).