We hypothesize two sources for sheepskin effects - signaling, and diplomas tied to jobs with downwards rigid wages. These theories have implications for diploma effects not only in the first, but also the second moments of the Mincerian earnings distribution that we are able to identify using a flexible econometric specification. Idiosyncrasies in Mexican labor market and educational institutions offer a natural experiment on which to train this methodology and test these theories. Correcting for heterogeneity in diplomas, we find no evidence of sheepskin effects, except on graduation from primary school. We find compelling evidence that returns to education (in both moments) are linked with labor market institutions and job-specific diplomas in the manner we hypothesize. Our econometric structure corrects for sample selectivity due to unemployment and allows us to observe behavior on the quantity axis of a labor market segmented by sheepskin effects. We also analyze the covariates of hours worked which helps to explain observed patterns in hourly earnings.