Title：A Bayesian Look at American Academic Wages
Speaker： Professor Michel Lubrano
Time：9:30AM, April 14th, 2016
Location: The first conference hall in School of Economics, 8th floor of the northern building.
This paper tests modern theories of wage formation in the specific context of the academic market. We have access for that to the wage data base of Michigan State University (and numerous other variables) in 2006. We found evidences for validating the tournament theory. We adjusted a hybrid mixture to the wage distribution for modelling both average wages and high wages. But contrary to what one can expect, we could not validate the superstar theory for explaining professor wages. In fact we observe a wage compression for the higher wages, even when we study wage dynamics using 2012 data. However, we found some superstar phenomenon when recruiting Assistant professors, and mainly when recruiting sport coaches who can get some of the highest wages. Our econometric model is a hybrid mixture of a lognormal and a Pareto density, the latter being compatible with superstars.
We provide Bayesian inference for that mixture and derive various inequality characteristics. Separating in probability the Pareto sample from the lognormal one, we can relate wage formation to the dynamics of promotion after six years and explain the rate of exit. The question of heterogeneity and dynamics of inequalities is of central interest.