Online Program

Disclosure Limitation for the Quarterly Census of Employment and Wages

Keywords: Longitudinal Data; fully synthetic; confidential data; multiscale data; multiple imputation; bootstrap.

The U.S. Bureau of Labor Statistics Quarterly Census of Employment and Wages (QCEW) releases data to the public aggregated across different scales both longitudinally and by economic sector. In order to protect confidential data, some cells are suppressed in the publicly released data. One problem with this disclosure limitation method is that many of the missing observations can be accurately imputed, using dynamic linear models that respect the constraints imposed by the multiscale nature of the data. An empirical investigation demonstrated that many of the suppressed cells can be imputed within 1\% accuracy, thus causing potential concerns for data confidentiality. We describe the difficulties that the constraints imposed by the structure of the data present in developing a suitable disclosure limitation method for the QCEW. We then propose a new disclosure limitation method based on multiple fully synthetic data sets obtained from repeated samples of the original micro-data.