Reconciling Trends in Male Earnings Volatility: Evidence from the SIPP Survey and Administrative Data

Michael D. Carr, Robert A. Moffitt, Emily E. Wiemers

Research output: Contribution to journalArticlepeer-review

3 Scopus citations


As part of a set of papers using the same methods and sample selection criteria to estimate trends in male earnings volatility across survey and administrative datasets, we conduct a new investigation of male earnings volatility using data from the Survey of Income and Program Participation (SIPP) survey and SIPP-linked administrative earnings data (SIPP GSF). We find that the level of volatility is higher in the administrative earnings histories in the SIPP GSF than in the SIPP survey but that the trends are similar. Between 1984 and 2012, volatility in the SIPP survey declines slightly while volatility in the SIPP GSF increases slightly. Including imputations due to unit nonresponse in the SIPP survey data increases both the level and upward trend in volatility and poses a challenge for estimating a consistent series in the SIPP survey data. Because the density of low earnings differs considerably across datasets, and volatility may vary across the earnings distribution, we also estimate trends in volatility where we hold the earnings distribution fixed across the two data sources. Differences in the underlying earnings distribution explain much of the difference in the level of and trends in volatility between the SIPP survey and SIPP GSF.

Original languageEnglish (US)
Pages (from-to)26-32
Number of pages7
JournalJournal of Business and Economic Statistics
Issue number1
StatePublished - 2022


  • Administrative data
  • SIPP
  • Survey Data
  • Volatility

ASJC Scopus subject areas

  • Statistics and Probability
  • Social Sciences (miscellaneous)
  • Economics and Econometrics
  • Statistics, Probability and Uncertainty


Dive into the research topics of 'Reconciling Trends in Male Earnings Volatility: Evidence from the SIPP Survey and Administrative Data'. Together they form a unique fingerprint.

Cite this