Economists have persistently found a positive relationship between the employer size (measured in number of employees) and the wage received by employees. This employer size effect is robust to many different controls of the wage equation such as education, industry, capital-labor ratio, occupation, etc. Theoretical models such as those from Kremer (1993), Bulow and Summers (1986), Weiss and Landau (1984), Strand (1987) and Burdett and Mortensen (1998) provide explanations for the employer size effect, which we test using a wage equation with fixed effects that control for unobserved heterogeneity at three levels: worker, employer and match quality, following Woodcock (2008). To our knowledge, former papers in the empirical literature related to the employer size effect have not controlled for these heterogeneities simultaneously as we do here, so their results might be biased. We show that more than half of the employer size effect is explained by correlation between the employer size and the three heterogeneities. However, based on our results, we cannot conclude on the empirical validity of the theoretical explanations provided by those authors.