The Within-Month Pattern of Medical Care Utilization among SNAP Households

This paper investigates the implications of the Supplemental Nutrition Assistance Program (SNAP) benefit cycle, whereby the benefits are paid once per month, but often do not last for the entire month. Prior literature has found that monthly government cash transfers affect health and health care (Evans and Moore 2011; Gross and Tobacman 2014). However, much less is known about how in-kind transfers such as SNAP payments affect medical care utilization.

Conceptually, there are two reasons why medical care behavior may be impacted by SNAP payments. First, households may reduce their purchase of non-food goods, such as those for medical care, to free up cash income for food purchases when SNAP benefits are exhausted. Second, SNAP payments may impact health care consumption indirectly, via changes in health. This could occur if households use the payments to increase certain types of consumption which affect the need for hospital care. Such a mechanism would be consistent with the finding that SNAP payments change the pattern of alcohol-related accidents (Cotti, Gordanier, and Ozturk 2016).

We use data from the 1996-2013 Medical Expenditure Panel Survey (MEPS) to investigate whether the SNAP benefit cycle alters consumption patterns for medical care, and specifically, whether benefit-receiving households reduce their medical care utilization towards the end of the benefit month. To do so, we rely on exogenous variation in the timing of SNAP payments. We merge state-level SNAP disbursement schedule to the MEPS to calculate the probability that each calendar week is the last week of the benefit month for SNAP recipients. We then regress different measures of healthcare utilization on the state-level probability variable. We include state and time fixed effects in our models, so identification comes from changes in disbursement schedules within states over time. In order to investigate the mechanisms involved, we estimate our models separately for outpatient visits, emergency room (ER) visits, and inpatient admissions. We further decompose medical visits by the characteristics of the patients and their medical conditions.

Preliminary results suggest that visits to both outpatient providers and to the ER decline at the end of the benefit month among SNAP recipients. While the reduction in the use of ER is concentrated among the conditions that are urgent, the reduction in visits to outpatient providers is uniform across different visit categories. However, the likelihood of an inpatient admission does not vary significantly across the SNAP benefit month. We find similar results when we estimate our models on the sub-sample of SNAP recipients who are also Medicaid or SCHIP beneficiaries. Overall, our results suggest that both mechanisms may play a role in our finding that SNAP households reduce their use of medical care at the end of the SNAP benefit month.