arXiv.itsnick.co

an independent field guide to 23 million papers from arXiv & OpenAlex — not affiliated with arXiv.org

Social Sciences · Sociology and Political Science

JEL Code: C61

James J. Heckman, Paul A. Lafontaine, James J. Heckman, Paul A. Lafontaine

See on the map

Abstract

Currently, over 30 % of all wage earners refuse to answer one or more income related questions in the Current Population Survey (CPS) Monthly data. The CPS attempts to correct for any re-sulting non-response bias by allocating wages for non-respondents through a hot deck imputation procedure. The hot deck method matches wage non-respondents to wage respondents according to certain demographic and labor force characteristics. This paper shows that this imputation pro-cedure can lead to serious bias in estimating GED returns using CPS data. Although all estimates of returns to education are affected, we focus on estimating the returns to General Educational De-velopment (GED) certiÞcation. The imputation bias arises from the limited number of educational categories used to impute wages and is independent of any potential non-response bias which may also be present. After correcting for this bias in various ways, we show that the CPS data yields estimated returns to GED certiÞcation similar to those previously found in Cameron and Heckmans (1993) study using the National Longitudinal Survey of Youth (NLSY) data which do not suffer from imputation bias. We further show that all positive wage returns to GED certiÞcation in the CPS appear to be from an unobserved ability bias. This holds true for both the native and foreign born, as well as older versus more recent birth cohorts.

More like this