Differential Privacy for Economists
Lecturers: Daniel Goroff, Dan Kifer, Frauke Kreuter, Ian Schmutte
Supported by the Alfred P. Sloan Foundation
July 17, 2020
The extent to which individual responses to household surveys are protected from discovery by outside parties depends on the summary information released by the collecting government or firm, and on the broader data environment. Rapid decline in the cost of computation, along with a rising number of publicly available data sets, often from private vendors, have increased the risk that a determined party could combine public and private data resources and identify the survey responses of small groups or even individual respondents. Differential privacy is a tool for assessing the trade-off between releasing more granular information based on survey responses and protecting the privacy of survey respondents. These lectures offer an introduction to differential privacy along with examples of its application in settings that range from the collection of data on a small group to the US census.