We report the results of a field experiment in which treated employers could not observe the compensation history of job applicants, whereas control employers could. We find no evidence that treated employers relied more heavily on alternative signals of productivity. Instead, they responded by...