On behalf of the German Federal Ministry of Education and Research, the Federal Statistical Office conducts the project „Disclosure of Official Business Survey Data”. The project’s aim is the development of ways to practically anonymize microdata surveyed in corporations and businesses. Within certain legal limits Statistical Offices may transmit such data to research institutions. The project presents a contribution to fundamental research. It serves the improvement of the empirical economic sciences’ working base.
Within this project the ZEW and the Federal Statistical Office cooperate in assessing the protection effect of anonymization procedures allowing for external sources of information.
In the context of this cooperation, the ZEW and the Federal Statistical Office have been developing methodological approaches to file matching (record linkage) and the derivation of re-identification probabilities from the latter. The attribution of characteristics holders (paired cases) by the means of continuous and discrete parameter values from the different files has been on the top of the agenda. On the basis of the methods developed the ZEW has developed an executable record linkage program including documentation.
The ZEW provides excerpts of its company databases (“ZEW-Gründungspanel”, “Mannheimer Unternehmenspanel” etc.) usable only for matching tests carried out within the project’s framework. The Federal Statistical Office staff runs matching tests with these databases and project data from the cost structure survey in manufacturing industries at the Federal Statistical Office. In a joint project report by both partners the Federal Statistical Office staff has compiled the results of these matching tests including a detailed report.
ZEW and Federal Statistical Office staff conduct research into the appropriateness of resampling anonymizing procedures. For this purpose, for Federal Statistical Office project data series, ZEW staff at the Federal Statistical Office generates simulated observations directly from the microdata series. The anonymized data series created by the resampling procedures are tested as to their properties to reproduce univariate distributions as well as to their effects on regression methods. The generated anonymized sample files are examined towards their protection effect by ZEW staff in cooperation with the Federal Statistical Office. Furthermore, in cooperation with the Institute of Applied Economic Research (IAW) in Tübingen, they are examined towards their potential for analysis. Both examinations are documented in detail.
01.06.2003 - 31.01.2004
Institut für Angewandte Wirtschaftsforschung, Tübingen, DE