M-RCBG Associate Working Paper No. 116
Fungibility of In-Kind Transfers: Evidence from the Supplemental Nutrition Assistance Program (SNAP)
Honorable Mention, 2019 Dunlop Thesis Prize
Ryan Andrew Davis
I compare the marginal propensity to spend out of in-kind transfers and cash income. Using six years of scanner data on 832 grocery stores and 1,253 households, I exploit two quasi-experiments to estimate these parameters. First, I exploit the 2009 American Recovery and Reinvestment Act's nationwide expansion of the Supplemental Nutrition Assistance Program's (SNAP) benefit levels. Then, I exploit plausibly exogenous variation in cash-on-hand coming from Earned Income Tax Credit (EITC) refunds. I find that households spend significantly more on food following an increase in SNAP benefits than they do after an increase in cash. This is also the case for hygiene and cleaning products, a prominent SNAP-restricted category that is also sold in grocery stores. Spending on beer and cigarettes, however, is indistinguishable between SNAP benefits and cash-on-hand, and the marginal increase is close to zero. Thus, policymakers should not be overly concerned about SNAP subsidizing alcohol and tobacco. In addition, I find that the partial fungibility of an in-kind transfer not only distorts spending on the good or service directly associated with it, but also on other restricted products that might be grouped in the same mental accounting bucket.