Do Survey Data Estimate Earnings Inequality Correctly? Measurement Errors Among Black and White Male Workers

Abstract:     
Few studies have considered how earnings inequality estimates may be affected by measurement error in self-reported earnings in surveys. Utilizing restricted-use data that links workers in the Survey of Income and Program Participation with their W-2 earnings records, we examine the effect of measurement error on estimates of racial earnings inequality. Results show that varying levels of mean-reverting error causes underestimation of earnings inequality. Notably, mean reversion is steeper for Black men than for White men, bringing about substantial downward bias in the estimated earnings gaps between Whites and Blacks at lower percentiles as well as large underestimation of within-racial group inequality for Black men. Together, our results call attention to the potential distortions generated by systematic measurement error on economic inequality estimates.

Keywords: Measurement Error, Mean-reverting Error, Earnings Inequality

JEL Classification: J00, J30, C42, C80

Accepted Paper Series