U.S. Earnings Mobility: Comparing Survey-Based and Administrative-Based Estimates

Earnings mobility has been studied both at the macro level (how much of a certain kind of mobility is there in the economy?) and at the micro level (what are the correlates of change in income or position?). Many empirical mobility studies provide estimates of the amount of mobility in a country over time and the correlates of individual mobility within the income distribution. While measurement error is recognized as potentially important at both these levels, very little is known about the degree to which earnings mobility estimates are affected by measurement error. In this paper, we use a new dataset that contains individually reported total annual labor earnings from the Survey of Income and Program Participation (SIPP) linked to employer-reported total annual labor earnings from the Social Security Administration’s Detailed Earnings Record (DER; these are taken directly from Box 1 on theW-2 form and are not capped by FICA) to compare micro and macro earnings mobility estimates for the U.S. during the 1990s using the two different earnings measures. We ask how much difference it makes to mobility estimates to use administrative-based earnings rather than survey-based earnings, and we obtain two major findings. Qualitatively, we find that the results are similar but not identical when administrative-based earnings are used rather than survey-based earnings. Quantitatively, we find that magnitudes are often very different when administrative-based earnings are used rather than survey-based earnings. The administrative-based results are neither systematically larger nor systematically smaller than the survey-based ones.