The Employer Size Wage Effect PDF Download

The strong empirical regularity of higher wages in large establishments seems to be ubiquitous. It was first brought to light by Moore (1911) and later confirmed by, among others, Brown and Medoff (1989), it prevails across countries and over time, across studies that span a broad range of industries and occupations. Yet, in the literature, the particular reason why the size of a firm should be a determinant of a worker's wage rate is still controversial. This research examines the size-wage relationship using the Institut fuer Arbeitsmarkt und Berufsforschung (IAB) employment sample over the period 1980-1995. Firstly, we summarise previous theoretical research and empirical evidence on the size-wage relationship.Secondly, the different theoretical arguments are tested empirically. The attempt to account for the size premium using observable worker and plant characteristics through level estimation meets with limited success. Size could reflect various unobserved determinants and it is generally argued that, the greater the possibility of controlling for these latent factors, the less likely that a significant size effect will appear; to control for constant unmeasured worker and firm characteristics, we apply fixed effect and first difference models. In the longitudinal analysis, the size wage differential is still sizeable and is quite similar to level estimates. We find that the wage change closely resembles the differential estimated in the level regression and pre-displacement plant size plays also a role in post-displacement wage determination. The empirical findings demonstrate that positive effects of firm size on wages persist after controlling for observed and unobserved worker and firm characteristics. They suggest that although each of the explanations of size-wage differentials can account for some of the observed variation in worker wages, none can fully unravel the observed employer size-wage differentials. In the end, the unexplained premium remains large and significant.