An official website of the United States government
Imputing In-Kind Benefits in Survey Data (PDF)
Nikolas Mittag - CERGE-EI - nikolasmittag@posteo.de
This presentation reviews methods to impute in-kind benefits in survey data that are misreported or missing entirely. I focus on what we learned from imputing benefits for poverty measurement to improve imputation of in-kind benefits when measuring consumption. In the first part of the presentation, I outline three broad strategies to impute benefits: based on predicted eligibility status, using information on reported receipt and from prediction equations derived from other data sources such as linked administrative data or surveys that ask for information on the missing programs. I illustrate how each strategy works and underline their merits and problems using the evaluation of imputing SNAP benefits in Mittag (2019). In the second part of the presentation, I discuss what I consider key lessons from the paper and the experience with SNAP for the case of imputing in-kind benefits, such as WIC, NSLP and LIHEAP, to measure consumption.
Last Modified Date: November 10, 2021