High-Frequency Spending Responses to Government Transfer Payments 
Findings from studies focusing on spending responses to various government transfer payments and tax changes have important implications for thinking about the efficacy of fiscal policy designed to stimulate the economy during recessions. The economic downturn associated with the COVID-19 pandemic was unique in that it originated from a public health shock—rather than an economic one—resulting in government-mandated restrictions on economic activity, changes in individual behavior, and substantial fiscal support. Restrictions on activity, especially early in the pandemic when the Coronavirus Aid, Relief, and Economic Security (CARES) Act fiscal stimulus payments were disbursed, likely affected individuals’ ability, desire, and need to spend the stimulus money. Studying the spending response to the CARES Act payments is therefore relevant for understanding the benefits of fiscal stimulus in response to a nontraditional economic downturn. This paper uses high-frequency, transaction-level payment-card data for a sample of lower-income, unbanked cardholders to examine consumers’ spending response to the stimulus payments. The data focus on a subset of the population that likely had a particular need for the stimulus monies. Unlike the data used by previous papers in the literature, the authors’ data allows them to study the MPC out of the stimulus payments while simultaneously controlling for spending out of non-stimulus income and tax refund payments.
Key Findings
- Among the consumers in the sample, the marginal propensity to consume (MPC) out of the CARES Act stimulus payment was 15 cents in the week the payments were received (that is, spending increased 15 cents per dollar of stimulus). The MPC then rose noticeably over time to reach about 66 cents cumulatively after 16 weeks.
- The MPC out of non-stimulus income was 20 cents initially and 46 cents cumulatively after 16 weeks.
- The cumulative MPC out of tax refunds was similar to that of non-stimulus income after 16 weeks.
- The finding that the spending response to the stimulus payments was at first somewhat slower than for other temporary income fluctuations but became larger over time is broadly consistent with the presence of constraints on economic activity around the time most of the CARES Act stimulus funds were disbursed.
- Spending responses to the stimulus payments were stronger for cardholders with particularly low 2019 income and for cardholders who received unemployment insurance benefits during the pandemic, despite the overall generosity of those benefits.
Implications
The pandemic disproportionately impacted low-income workers, especially early on, which may have resulted in consumers experiencing job and earnings losses at the same time as the stimulus payments were disbursed. Therefore, with the panel of low-income consumers used in this study, controlling for non-stimulus income is important for identifying the effect of the stimulus payments on MPC. Including the MPC estimates out of non-stimulus income and out of tax rebates also provides context for assessing the size of the spending response to the stimulus payments.
Furthermore, while the analysis in this paper focuses on only the first round of stimulus payments, those payments were disbursed during an economic downturn caused primarily by pandemic-related business shutdowns and changes in individual behavior. Thus, this paper’s findings provide insight into the efficacy of fiscal policy responding to the initial stages of a public health shock.
Abstract
This paper evaluates the marginal propensity to consume (MPC) out of the 2020 fiscal stimulus payments using high-frequency, transaction-level data for a sample of low-income cardholders, many of whom are unbanked. Consumers’ MPC out of non-stimulus income and their MPC out of tax refunds are estimated simultaneously. Spending responds less on impact to the stimulus payments than to non-stimulus income (15 cents versus 20 cents per dollar of income), but stimulus-payment spending quickly catches up and is noticeably higher than non-stimulus-income spending on a cumulative basis after 16 weeks (66 cents versus 46 cents). This finding is qualitatively quite robust, and there is relevant heterogeneity in the spending responses across cardholders that includes some pandemic-related effects.