Association for Public Policy Analysis & Management

As the largest program in the federal nutrition safety net, the Supplemental Nutrition Assistance Program (SNAP) aims to support food spending among low-income households. Household-level SNAP benefit allotments are benchmarked to a needs standard that reflects the estimated amount of resources required to purchase a minimal cost grocery basket. This needs standard does not account for geographic cost of living variation, making the needs standard a target of calls for program innovation. Although there is an extensive literature examining the effect of SNAP on food spending, we know less about this relationship in the context of geographic cost of living variation.

This paper contributes to the broader literature about the effect of SNAP on food spending, while also examining the influence of geographic variation in housing prices. We focus on housing prices for two reasons. First, housing is an important driver of area-level cost of living. Second, housing takes up the largest share of low-income households’ budgets. SNAP may play a distinct role in the food budgets of households in relatively high housing cost areas compared to those in relatively low housing cost areas.

We use nationally representative data from the National Household Food Acquisition and Purchase Survey (FoodAPS). Our analytic sample includes 4,115 households with 1) 6 or fewer household members, 2) non-missing food-at-home (FAH) spending, and 3) plausible shelter expenses. We use geographic data on place of residence to link FoodAPS with Regional Price Parities (RPP) for rents. RPPs allow us to examine variation in housing prices across metropolitan and non-metropolitan areas within states.

Our empirical approach considers interactions between 1) cash income and SNAP benefits, 2) SNAP participation status, and 3) housing prices to understand the effect of SNAP on FAH spending. We allow the effect of housing prices to be non-linear so that the effect of SNAP may vary for households living in relatively low or high housing cost areas. Our preliminary descriptive analysis, which does not yet consider RPPs, assesses regional and urban-rural differences in FAH spending for SNAP households (Table 1).

Table 1. Mean FAH spending for SNAP households

		Northeast	Midwest	South	West
FAH spending relative to maximum SNAP benefit	Urban	1.20 (0.15)	0.84 * (0.06)	0.86 * (0.05)	0.99 (0.07)
	Rural	0.92 (0.02)	1.44 (0.26)	1.06 (0.11)	1.53 (0.59)

* : significantly different, with reference to Northeast urban tracts

Bold: significantly different, with reference to urban tracts in same region

Notes. Estimates are weighted, with jackknife standard errors in parentheses. The analytic sample includes 1,273 SNAP households.

SNAP households living in Northeastern urban areas, where housing prices are likely high, spend 0.34 and 0.36 percentage points more on FAH compared to SNAP households in Southern or Midwestern urban areas, respectively. Northeastern rural SNAP households have similar levels of FAH spending compared to rural SNAP households in other regions. Further analysis will determine whether these differences persist after accounting for household resources. This research may inform debates around program innovation, helping SNAP better support food spending needs of low-income households across the U.S.