Trouble in the Tails? What We Know about Earnings Nonresponse Thirty Years After Lillard, Smith, and Welch

Abstract

Earnings nonresponse in household surveys is widespread, yet there is limited knowledge of how nonresponse biases earnings measures. We examine the consequences of nonresponse on earnings gaps and inequality using Current Population Survey individual records linked to administrative earnings data. The common assumption that earnings are missing at random is rejected. Nonresponse across the earnings distribution is U-shaped, highest in the left and right tails. Inequality measures differ between household and administrative data due in part to nonresponse. Nonresponse biases earnings differentials by race, gender, and education, particularly in the tails. Flexible copula-based models can account for nonrandom nonresponse.

Imputation Match Bias in Immigrant Wage Convergence

“…First, the share of the workforce who is an immigrant has increased dramatically since 1970. Second, earnings imputation rates have substantially increased over time (Bollinger et al 2015;Hirsch and Schumacher 2004). Third, as we show here, immigrants who have arrived more recently to the United States are more likely to have imputed earnings than those with longer duration in the United States.…”

“…Based on CPS data, Bollinger and Hirsch (2006) show that the magnitude of attenuation on the coefficient of a specific characteristic is approximately equal to the proportion of individuals with that characteristic whose wages are imputed. This line of research supports estimating earnings regressions by excluding observations with imputed earnings in order to mitigate imputation match bias (Bollinger at al. 2015).…”

The Impact of a Minimum‐Wage Increase on Temporary‐Contract Workers

“…As noted by Bollinger et al. (), non‐response in earnings surveys can be particularly prevalent in the tails of the earnings distribution, which is where our target group of minimum‐wage workers are likely to be located. In the presence of severe earnings non‐response, selection models can be used to test the robustness of results.…”

Nonresponse Bias in Inequality Measurement: Cross‐Country Analysis Using Luxembourg Income Study Surveys

“…Mistiaen and Ravallion (2003) add the count of item-nonresponse households to the type-A unit-nonresponse households, and use the methodology on them jointly. Londoño-Vélez, 2013), and 4-5 points in the United States (Hokayem, Raghunathan, and Rothbaum, 2016;Bollinger et al, 2019). The correction to a worldwide Gini has been estimated at 3-4 percentage points (Anand and Segal, 2015).…”

“…To evaluate this more fully, supplementaryTable A4reports a similar but quadratic model specification:g i = 1 + 2 loge y i + 3 (log e y i ) 2 .The problem of bottom-income nonresponse and the need to impute for it has been raised previously bySafari et al (2018),Bollinger et al (2019),Morelli and Muñoz (2019), and Hlasny, Ceriani, and Verme (forthcoming).…”