Online Program

Rethinking Sensitivity and Disclosure Risk: QCEW Establishment Data
*Michael Buso, Bureau of Labor Statistics 
Shail Butani, Bureau of Labor Statistics 
David Hiles, U.S. Bureau of Labor Statistics 

Keywords: Sensitivity, disclosure risk, cell suppression, noise perturbation, synthetic data

The U.S. Bureau of Labor Statistics (BLS) and NORC have jointly developed a new disclosure limitation methodology for the Quarterly Census of Employment and Wages (QCEW), building on the micro-level noise perturbation ideas of Evans, Zayatz, and Slanta (EZS). Although the new method is expected to reduce the number of cell suppressions by 20 percent and offer improved protection, it would still suppress approximately one-half of all cells.

To further improve data utility, we propose the consideration of a new definition of sensitivity and protection for establishment-level data. The new paradigm would feature the following: 1) Perturbed values need no additional protection. 2) Perturbed data may equal original values on a probability basis. 3) Persistent zero-values do not need protection. 4) Trend data for wages would receive protection for outliers only. 5) Employment data present little additional disclosure risk and thus need no protection. These suggestions, however, require extensive evaluation on part of all stakeholders.