Using longitudinal wage information in linked data sets

Abstract

"Using longitudinal wage information in linked datasets such as ALWA-ADIAB is not straight-forward. First, due to the upper earnings limit for social security contributions, parts of the wages have to be imputed. Second, although information from the ALWA survey is linked to administrative data from the Federal Employment Agency (BA) on the personal/individual level, matching single employment spells may not always be possible. I propose a polynomial time function to smoothen wage information over time and close gaps between spells in the administrative data. The information can then be matched to the respondents in ALWA and adjusted to meet the episode structure given in the survey. Moreover, survey information on contract hours allows for the calculation of hourly wages. The result is the supplement of ALWA with longitudinal wage information for all main employment episodes in West Germany, starting after January 1975, and for episodes in East Germany starting after January 1993. Although the estimation of wages obtained through the proposed method is connected to some assumptions and thus to a certain degree of uncertainty it produces more complete wage information than is provided through simple matching on a monthly basis. Moreover, a comparison with the net income that respondents reported for their interview date shows that predicted daily and hourly wages are highly reliable." (Author's abstract, IAB-Doku) ((en))

Cite article

Reichelt, M. (2015): Using longitudinal wage information in linked data sets. The example of ALWA-ADIAB. (FDZ-Methodenreport 01/2015 (en)), Nürnberg, 14 p.