Earnings Dynamics and Measurement Error in Matched Survey and Administrative Data

ABSTRACT

This article analyzes earnings dynamics and measurement error using a matched longitudinal sample of individuals’ survey and administrative earnings. In line with previous literature, the reported differences are characterized by both persistent and transitory factors. Estimating a model consistent with past results, survey errors are mean-reverting when administrative reports are assumed correct, but not when this assumption is relaxed. Although most reported earnings variation is true, we conclude that measurement errors dominate observed changes, and that transitory earnings contribute little to overall earnings inequality. The results imply the reliability of matched administrative data should be treated with caution.

Keywords:

• Earnings dynamics
• Measurement error
• Panel data
• Validation study

ACKNOWLEDGMENTS

This research was supported in part by the Royal Society of New Zealand Marsden Fund Grant MEP1301. The authors thank seminar participants at the Asian and Australasian Society of Labour Economists (AASLE) Canberra conference, Hitotsubashi University, University of Technology Sydney, University of Tokyo, Victoria University of Wellington, two referees, and Dave Maré for excellent comments and discussions, and Corey Allan for research assistance in putting together the data extract for the analysis.

DISCLAIMER

Access to the anonymized data used in this study was provided by Statistics New Zealand in accordance with security and confidentiality provisions of the Statistics Act 1975, and secrecy provisions of the Tax Administration Act 1994. The results in this article are the work of the authors, not Statistics NZ, and have been confidentialized to protect individuals and businesses from identification. See Hyslop and Townsend (2016) for the full disclaimer.