In recent years, and during Covid-19 in particular, many countries have experimented with innovative approaches to fiscal stimulus. This column reports on an experiment in France in which individuals were given a debit card tied to a new bank account with an initial balance of €300, but with differing conditions. While certain groups, such as lower-income households, showed somewhat higher spending responses, the differences in spending caused by card design were much larger than differences across demographic groups, suggesting that the design of fiscal stimulus policies can strongly affect their impact on aggregate demand.
During economic downturns, policymakers often debate how to effectively stimulate consumer spending. In recent years, many countries have experimented with innovative approaches to fiscal stimulus, from traditional tax rebates to more targeted interventions like prepaid cards. For instance, during the Covid-19 pandemic, cities and countries as diverse as Milan, Seoul, and Hong Kong implemented stimulus programmes using prepaid cards or time-limited consumption vouchers. However, there is still little empirical evidence on how the design of such transfers affects their effectiveness.
Recent work analysing the 2008 US tax rebates suggests that traditional fiscal stimulus may have modest effects on consumption. Using robust statistical methods, researchers found that households spent only about 25% of these rebates in the first quarter (Borusyak et al. 2023, Orchard et al. 2023). These findings raise an important question: could alternative designs for fiscal stimulus lead to larger consumption responses?
An experiment to understand the role of stimulus design
To answer this question, we conducted a large-scale experiment in France where we provided money transfers to about 1,000 individuals that were randomly chosen from a pool of households that are roughly representative of the French population (Boehm et al. 2024). Selected recipients were randomly assigned one of three different transfer designs. The first group received a debit card that was tied to a new chequing account with an initial balance of €300, which they could spend in shops and on the internet however they wanted (with the only exception being that they could not withdraw cash). The cards were valid for five months, and recipients were informed that after these six months they would receive any unspent balance wired to their main chequing account. The second group received the same account and card, except that the card would expire after three weeks, and unspent balances were lost to the recipient. The third card had the same conditions as the first one, but remaining balances were lost at a rate of approximately 10% per week, mimicking a negative interest rate. Collaborating with a major French retail bank allowed us to track total consumption spending (using any means of payment) of the recipient households, as well as for an ex-ante identical control group. By comparing recipient households with control households, we estimated how these different stimulus designs affected total household spending.
Results
We summarise our results in five key facts about how households respond to a fiscal stimulus.
First, with a standard prepaid card, households increase consumption expenditure on average by 23% of the transfer in the first month – a response similar to recent estimates from US tax rebates. Even though many households ‘used up’ the balance on the stimulus card, they decreased their spending using other means of payment, so that the total increase in consumption expenditure is on average only €69 (=23% of 300).
Second, we find that the design of the transfer matters substantially: when the card expires after three weeks (as for the second group), households increase their consumption expenditure by 61% of the transfer amount in the first month. The average increase in consumption expenditure for the third group (which faced a negative interest rate on these account balances) lies in between those of the first and second group, at 35% of the transferred amount. Figure 1 shows the path of the consumption expenditure response by group.
Figure 1 Response of consumption expenditure, by group
Third, for all cards, the spending increase is concentrated in the first few weeks after receiving the transfer. After these first few weeks, weekly consumption expenditures are indistinguishable between recipient households and control households.
Fourth, while the consumption response varies to some extent with household characteristics like income, even households with substantial liquid wealth show large spending responses. This finding contrasts with standard economic models that predict that only liquidity-constrained households should respond strongly to transfers. Among the attributes of households or recipients that we observe, demographic characteristics like gender are the most important predictors of the consumption increase following our unanticipated transfer.
Fifth, there is substantial variation in spending responses across households that cannot be explained by observable characteristics. Many households have high marginal propensities to consume, regardless of their demographic or economic profile. The distribution of the marginal propensity to consume out of a transitory transfer (see Figure 2) is unimodal.
Figure 2 Distribution of four-week cumulative MPC estimates by card type
These findings have important implications for both economic theory and policy design. On the theory side, our results challenge standard macroeconomic models where money is fully fungible. The fact that spending responses differ dramatically based on whether the card expires suggests that households do not treat all money as equivalent. Indeed, we find that many households make purchases with their regular debit cards even when they have sufficient funds on their prepaid cards – behaviour that is difficult to reconcile with standard rational models of consumption behaviour.
Design-based stimulus transfers: Implications for macroeconomic policy
From a policy perspective, our results suggest that the design of fiscal stimulus policies can strongly affect its impact on aggregate demand. Our experiment was designed with scalability and policy-relevance in mind: the stimulus treatments we administered could be scaled to national levels and could be administered by fiscal or monetary authorities. Design-based stimulus policies also have the potential to be more effective than targeted transfers. While we find that certain groups (such as lower-income households) have somewhat higher spending responses, the differences in spending caused by card design are much larger than differences across demographic groups. Furthermore, tweaking the design of stimulus payments as opposed to targeting transfers may avoid equity concerns. As a result, policymakers may want to focus on how transfers are implemented rather than just who receives them.
Our findings are particularly relevant given recent estimates showing modest effects of traditional fiscal stimulus. For instance, studies of the 2020 US stimulus payments found relatively low spending responses (Parker et al. 2022, Gorodnichenko et al. 2020). Our results suggest that alternative designs – like transfers that expire if not spent quickly – could generate substantially larger effects on consumption. As countries continue to face economic challenges that may require stimulus measures, considering how transfers are structured – not just their size or targeting – may be crucial for maximising their impact.
Implementing direct stimulus transfers in the way we study would likely require significant legal and operational preparation. Central banks and fiscal authorities would need to develop or adapt payment systems that can handle time-limited transfers at scale. This could involve creating a framework for central bank digital currencies with programmable features, or partnering with private financial institutions to develop standardised prepaid card programmes. Legal frameworks may need updating to allow for the issuance of government-backed payment instruments with features like expiration dates or negative interest rates.
The infrastructure requirements are not trivial, but several recent examples show they are feasible. During the Covid-19 pandemic, cities including Seoul successfully implemented large-scale prepaid card programs (Woo et al. 2021), while Hong Kong distributed consumption vouchers to its entire adult population (Geng et al. 2022). These experiences provide valuable lessons for other jurisdictions. Given that developing and testing such systems takes time, policymakers would be well-advised to create the necessary infrastructure now, while the need for fiscal stimulus policies is low. Having these tools ready before the next economic downturn would significantly expand the policy options available for macroeconomic stabilisation.
Source : VOXeu
