Reference no: EM131762961

Question: This problem set considers some of the empirical issues involved in implementing a comparable worth policy (a policy that assigns equal wages to jobs that are judged to be "comparable"). Economists typically do not think of the "worth" of a job independently from market forces, but the use of formal job evaluation methods has existed for some time. Hay Associates, a national compensation consulting company, has conducted job evaluations for a number of states. The company relates the wages for a given job to a number of "compensable" factors including skills, responsibility, effort, and working conditions. The trained job evaluators assign a specified number of points ("Hay Points") to each of the four job characteristics for each job. The scores are then added to get a total point score for the job. This problem set employs data from 116 job titles used by the state of New York for its employees in 1982. Workers with these job titles represent 1/3 of the state's employees. These data are found in https://www.wiley.com/college/ashenfelter.

a. Define male-dominated jobs as those with fewer than 30% women and female-dominated jobs with greater than 70% women (this is the same approach Hay follows). Compute the mean percentage of women in both male- and female-dominated jobs.

b. What is the mean wage in female-dominated and male-dominated jobs? Does the differential necessarily indicate discrimination? What is the mean score for the five compensable factors (skills, effort, responsibility, working conditions, and total points) for male-dominated and female-dominated jobs and how do they compare?

c. Run a regression of wages on skills, effort, responsibility, and working conditions. Do the coefficients have the anticipated signs? Are they significantly different from zero? What fraction of the variance in wages do they explain?

d. Run a regression of wages on points. What is the relationship between this regression and the regression in part (c)? What restriction on the regression in part (c) would yield this specification? How would you test this restriction? Conduct this test and report your results.

e. Now add another variable to the regression you ran in part (c). Define the variable such that it equals 1 if the job is female-dominated and zero otherwise. Interpret the coefficients from this regression. What is the importance of the coefficient on the new variable added to the regression? What does it seem to suggest?