Revisiting the Economics of Privacy: Population Statistics and Confidentiality Protection as Public Goods
Abowd, John; Schmutte, Ian M.
We consider the problem of the public release of statistical information about a population–explicitly accounting for the public-good properties of both data accuracy and privacy loss. We first consider the implications of adding the public-good component to recently published models of private data publication under differential privacy guarantees using a Vickery-Clark-Groves mechanism and a Lindahl mechanism. We show that data quality will be inefficiently under-supplied. Next, we develop a standard social planner’s problem using the technology set implied by (ε, δ)-differential privacy with (α, β)-accuracy for the Private Multiplicative Weights query release mechanism to study the properties of optimal provision of data accuracy and privacy loss when both are public goods. Using the production possibilities frontier implied by this technology, explicitly parameterized interdependent preferences, and the social welfare function, we display properties of the solution to the social planner’s problem. Our results directly quantify the optimal choice of data accuracy and privacy loss as functions of the technology and preference parameters. Some of these properties can be quantified using population statistics on marginal preferences and correlations between income, data accuracy preferences, and privacy loss preferences that are available from survey data. Our results show that government data custodians should publish more accurate statistics with weaker privacy guarantees than would occur with purely private data publishing. Our statistical results using the General Social Survey and the Cornell National Social Survey indicate that the welfare losses from under-providing data accuracy while over-providing privacy protection can be substantial.
A complete archive of the data and programs used in this paper is available via http://doi.org/10.5281/zenodo.345385.Any opinions and conclusions expressed herein are those of the authors and do not necessarily represent the views of the Census Bureau, NSF, or the Sloan Foundation. We also thank the Isaac Newton Institute for Mathematical Sciences, Cambridge, for support and hospitality during the Programme on Data Linkage and Anonymisation, supported by EPSRC grant no. EP/K032208/1. Abowd also acknowledges the Center for Labor Economics at UC Berkeley, where he was a visiting scholar when this work was initiated. We are grateful for helpful comments from Larry Blume, David Card, Michael Castro, Cynthia Dwork, John Eltinge, Stephen Fienberg, Mark Kutzbach, Ron Jarmin, Dan Kifer, Ashwin Machanavajjhala, Frank McSherry, Gerome Miklau, Kobbi Nissim, Mallesh Pai, Jerry Reiter, Eric Slud, Adam Smith, Bruce Spencer, Sara Sullivan, Lars Vilhuber and Nellie Zhao along with seminar and conference participants at the U.S. Census Bureau, Cornell University, CREST, George Mason University, Georgetown University, University of Washington Evans School of Public Policy, and the Society of Labor Economists. We thank Jennifer Childs and Casey Eggleston for providing data from the Federal Statistical System Public Opinion Survey conducted by the Census Bureau’s Center for Survey Methodology. William Sexton provided excellent research assistance. No confidential data were used in this paper.
Abowd and Schmutte acknowledge the support of Alfred P. Sloan Foundation Grant G-2015-13903 and NSF Grant SES-1131848. Abowd acknowledges direct support from the U.S. Census Bureau (before and during his appointment as Associate Director) and from NSF Grants BCS- 0941226, TC-1012593. Some of the research for this paper was conducted using the resources of the Social Science Gateway, which was partially supported by NSF grant SES-0922005.
Demand for public statistics; Technology for statistical agencies; Optimal data accuracy; Optimal confidentiality protection
A complete archive of the data and programs used in this paper is available via http://doi.org/10.5281/zenodo.345385.