Employer size or skill-group size effect on wages?

It turns out that the employer-size effect on individual wages dwindles away once one control for the number of workers of the same skill-group (educational type) as the observed individual within the establishment. The skill-group size effect on wages is substantial. The main results, a dwindling employer size effect and a significant group size effect, remain after controlling for both individual and establishment specific heterogeneity. This observation rejects most of the proposed explanations for the employer-size effect, while it lends considerable support for the notion that there are frictions in the labor market and that each establishment faces an upward sloping supply curve for each type of labor.