| The vast empirical literature on wage differentials reflects extreme interest in the subject from economists, social scientists, and policymakers alike. A review of the literature reveals two noticeable shortcomings. First, nearly all past studies have assumed a linear relationship between earnings and worker characteristics without consideration as to the potential specification error this imposes on their estimates. Second, researchers have generally focused on simple mean effects rather than estimating the effects of characteristics and wage rates on earnings distributions as a whole. Using data from the American Community Survey, this dissertation addresses both issues with a simple nonparametric approach, and applies the method to measure wage differentials between federal and private sector workers, as well as wage differences between males and females. |
