Alternative Disclosure Limitation Methodologies for Small Establishments in the Quarterly Census of Employment and Wages Program

• Activity Number: 138
• Type: Contributed
• Date/Time: Monday, August 5, 2013 : 8:30 AM to 10:20 AM
• Sponsor: Survey Research Methods Section
• Author(s): Spencer Jobe*+ and Michael Buso and Shail Butani and David Hiles and Randall Powers and Daniell Toth
• Companies: Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics and Bureau of Labor Statistics
• Keywords: data smearing ; noise perturbation ; synthetic models ; small establishments ; averaging by industry

Abstract

The U.S. Bureau of Labor Statistics' Quarterly Census of Employment and Wages program is a virtual census (97%) of employees on nonfarm payrolls, published in fine geographic and industry detail. To preserve respondent confidentiality, a large proportion (60%) of the tabulated cells is suppressed, which significantly reduces the utility of the data. In this work, we compare a number of alternative disclosure limitation methods that replace sensitive data with altered or synthetic values. Among the methods included in the comparison are "data smearing," which replaces each establishment's data with an average from a sample of data from similar establishments in the area; random noise perturbation; averaging by industry; as well as a number of other methods for perturbing or replacing sensitive data. The methods are applied to establishments of all sizes, but, the comparison focuses on the utility and protection of data from small establishments (fewer than 10 employees) that comprise about 80 percent of the population.