All Times ET
Federal administrative tax data are invaluable for research, but because of privacy concerns, access to these data is typically limited to select agencies and a few individuals. An alternative to sharing microlevel data are validation servers, which allow individuals to query statistics without accessing the confidential data. This talk will present results from a feasibility study of using differentially private (DP) methods to implement such a server. This work was done as part of a project to expand access to administrative tax data. The talk will cover the background for the study, the methods studied, and the results using real administrative tax data obtained from the Internal Revenue Service Statistics of Income (SOI) Division. The methods studied include DP algorithms for releasing tabular statistics, means, quantiles, and regression estimates. Our findings show that a validation server would be feasible for simple statistics but would struggle to produce accurate regression estimates and confidence intervals. I will outline challenges and offer recommendations for future work on validation servers.