SHEEPSKIN EFFECTS BY COHORT - IMPLICATIONS OF JOB MATCHING IN A SIGNALING MODEL

In the presence of job matching, the returns to education signals are shown to decline in value as additional work experience allows more di rect observation of productivity. This is tested by estimating sheepsk in effects across five age cohorts of non-minority males in 1991. The effects are large and significant in early cohorts and virtually nonex istent in later cohorts. This pattern is partially confirmed with esti mations within cohorts showing sheepskin returns declining from 1979 t o 1991. The pattern within cohorts suggests that the 1991 pattern is n ot merely the result of vintage effects but caution is expressed in dr awing conclusions.