A Test for Selection in Matched Administrative Earnings Data

Abstract

We test whether individuals in the Health and Retirement Study who consented to have administrative earnings data matched to survey responses represent a non-random sample. For both men and women, there is a general pattern of negative selection across three measures of pre-entry labor-market behavior: labor-force participation, self-employment, and earnings. However, for some outcomes the estimates are not precise enough to draw firm conclusions. The strongest results are that men who consented were 4.7 percentage points less likely to be self-employed than those who did not, and women who consented earned 13 percent less than those who did not.

Keywords: sample selection, administrative data, survey data

JEL Classification: J26, H2