JOB CHARACTERISTICS, GRADING SYSTEMS AND COMPARABLE WORTH (DISCRIMINATION)
COVO, MARIO MINO
Doctor of Philosophy thesis
Utilizing a unique and comprehensive data set on the personal and job characteristics of workers from three firms, we test the equalizing wage differential theory and study the determinants of the persistent male-female earnings gap. Access to information on the point-grading systems used by the three firms to set their wages, permits us to expose the methodology behind the systems which are being considered by Comparable Worth advocates as potential substitutes for the market in the determination of wages. Our data on grades, and on directly observed job characteristics that are homogeneous across firms, reduce the biases of the traditional wage hedonic estimators and enhances their efficiency. Our results show systematic support for the theory of equalizing wage differentials. The inclusion of grade information increases the flexibility of traditional models by complementing human capital theory with important administrative constraints that usually accompany internal labor markets. Our model reduces the unexplained portion of the observed earnings gap between males and females and highlights the difficulties associated with using inter-firm data from national samples. Using an arbitrary set of job characteristics we explain approximately 80 percent of the grades used in the firms under study. By contructing a variable to simulate the job analyst's estimate of the sex of the worker, we show that analysts can, by choosing the appropriate weights, set wages and manipulate grades to perpetuate the earnings gap. A high correlation between market wages and grades is found, which confirms that it is impossible to escape market realities using grading systems in their present form. Therefore, if, as Comparable Worth advocates desire, grading system are to be used to set wages for all workers across the country in a way that market distortions are not reflected by the grades, the existing systems are not adequate. We show that if analysts were forced to use the same weights across firms the flexibility of their system to adapt to local market conditions would be severely hindered.