John Newman (Australian Bureau of Statistics)
Protecting aggregate business microdata at the Australian Bureau of Statistics
The ABS releases aggregate population data through the online tool TableBuilder. However, aggregate data relating to businesses in Australia has traditionally been released by the ABS in a time consuming manual manner. It involves the identification and suppression of estimates that, if released, would break ABS threshold and dominance rules. Given the range of tables released, the enforcement of these rules is complex, expensive and subject to error. The advantages of using TableBuilder for releasing business data centre on higher levels of confidentiality, with more timely delivery of estimates and significant cost savings. This paper will outline the methodology behind TableBuilder and report on the progress made at the ABS into the feasibility of releasing business data through TableBuilder. In addition, it will report on the development of a new framework for balancing risk and utility for releasing perturbed business estimates at the ABS. We compared the official published estimates from a number of business collections with perturbed estimates from an Excel-based simulator of TableBuilder. It was found that some collections were more suitable for TableBuilder than others. Reasons for this will be discussed.