11/2/2010. The Katz-Murphy (1992) formulation. As relative supply increases, relative wage decreases. Katz-Murphy (1992) estimate

Transcription

1 The Katz-Murphy (1992) formulation As relative supply increases, relative wage decreases Katz-Murphy (1992) estimate KM model fits well until 1993 Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

2 Katz-Murphy (1992) estimate Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008): Relative wages differ by cohort >20 yrs experience College wage /High School wage < 10 yrs experience < 10 yrs experience College Grads /High School Grads >20 yrs experience Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

3 Card and Lemieux (2001) strategy allowing more flexibility in the CES Define education cohorts with substitution among cohorts MP Chain Rule Revised estimating equation Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

4 How credible is the minimum wage explanation? How credible is the minimum wage explanation? Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

5 How credible is the minimum wage explanation? Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008): How credible is the minimum wage explanation? Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

6 Lemieux(2006): Points to the poor fit of the Katz- Murphy model after 1987 as requiring a different explanation for widening inequality in the 1990s+ Unobserved skills Prices of the unobserved skills Measurement error in skills Presumption that ε it increases due to p it, but it could be Var(e it ) or Var(v it ) 90 th percentile log wage residuals, different variances ε 90 ε' 90 6

7 The need for a new explanation 1) Surge in college supply lowered returns to skill in the 1970s, but residual inequality continued to rise College and unobserved skill must be different inputs 2) College/High School premium stops rising in the 1990s, even though the relative supply change is similar to the 1980s. If trend growth is slowing why is residual inequality still rising? 3) If price of unobserved skill is rising, why is black-whote or male-female gap not rising? Solution: Composition of labor market more experienced There is no puzzle Assume initially that Define for each education-experience group j Conditional variance of unobserved skill for group j Variance decomposition using formulation in handout Group j share of the total population 7

8 Problem : cannot observe e it Conditional variance of jth group wages is Imposing that Change in variance from period s to period t Within group variance Between group variance If change is within group, it is due to prices If change is between groups, it is due to composition 8

9 Inequality rises when MW falls 9

10 Within group error variance not much going on after 1987 Controlling for changes in composition: Data from 1973, 2003 Estimate (P 03 x): probability of being in 2003 data using individual attributes Higher educated, experienced people downweighted Vertical distance due to composition, horizontal path due to prices 10

11 What about measurement error? March CPS: asks Earnings and Hours for the past year Wage = (Earnings/Hours) May CPS: asks Hourly wage directly Ditto Outgoing Rotation Group Outgoing Rotation Group: CPS respondents spend 4 months in, 8 months out and then 4 months in Salaried workers, similar treatment in March, May Hourly workers should have more accurate information in May MEN small big Variance in log hourly wages: Classical measurement error in hourly wages, not in salaried positions WOMEN 11

12 MEN Between Group Variance Classical measurement error does not affect means: between group variation should not be affected WOMEN MEN Residual Wage Variation WOMEN 12

13 MEN Changes in the proportion of workers in the CPS paid hourly Fraction paid hourly is increasing, raising measured residual inequality by percentage points over the sample period WOMEN Autor et al--- 1) What if we consider the upper and lower tail of the error distribution 2) What if we hold prices fixed and let composition evolve according to Vertical distance due to change in prices and Vertical distance due to change in prices and horizontal movement along path is due to changes in labor market composition. 13

14 Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008): Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

15 Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008): Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

16 Facts: Occupational labor markets Is there a polarization of the labor market? Rising 90/50 inequality Falling 50/10 inequality What would explain shrinking demand for the median worker with rising demand at the upper and lower tails? Dictionary of Occupation Titles defines education level and task intensity by occupation Task intensity of 50 means the occupation is at the median intensity of the task in 1960 Mean years of education by occupation Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):

17 ΔE = α + βδw + ε β = 3.0 in β = 2.96 in Autor, David H., Lawrence Katz and Melissa S. Kearney. Trends in U.S. Wage Inequality: Revising the Revisionists. The Review of Economics and Statistics 90(2) (May 2008):