Crowdsourcing peer information to change spending behavior

GAICIL Insights

Keeping up with the Joneses is usually treated as vanity.¹ This paper suggests it is often a search for information. Crowdsourcing peer information to change spending behavior, by Francesco D'Acunto, Alberto G. Rossi and Michael Weber, isolates the information channel of peer effects in a setting that excludes common shocks and social pressure. The authors pair a spending panel with crowdsourced information about anonymous peers, delivered through a FinTech app.

The findings are clear. Consumers converge to their peers' spending, and more so when peer signals are more informative. Convergence is asymmetric: within 12 months of receiving information, overspenders close 17% and underspenders 5% of their gap relative to peers. The authors exploit quasi-random assignment to peer groups in an instrumental-variable strategy and replicate the results in a randomized controlled trial on a population not selected for interest in peer information, supporting external validity. The results fit information-based theories of overconsumption.

The effects build persistently, and users exposed to more informative signals respond more strongly. People use peer spending as information about what is appropriate, adjusting when they learn they are out of line. Nobody is watching these anonymous peers, and nobody needs to be: the data alone move behavior. Many overspenders simply lack a reference point.

The authors place the study among the very first of robo-advising tools targeting consumption and saving choices, perhaps the most important economic decisions consumers make over their lives. As artificial intelligence enters personal finance, tools that deliver tailored information about spending and saving will become common. This paper shows that such information changes behavior measurably. The robo-advisor of the future may spend more time on the grocery bill than on the stock portfolio.

The asymmetry is a gift to policy. If raising the average saving rate were the objective, peer information might succeed, because users spending more than their peers react more strongly. Algorithmic tools that deliver peer comparisons could nudge households toward greater saving at low cost and at scale. The tool mainly corrects excess.

The authors flag hard open questions. Whether reacting to peer information is optimal, and whether peer spending carries relevant information for a given consumer, remain to be established. Peers may differ in circumstances, and converging to their spending may not always be wise. Heterogeneous and redistributional effects deserve isolation in future work. Designers of AI financial tools should take these questions seriously, ensuring comparisons are meaningful and recommendations serve users. A benchmark is only as good as the peer group behind it.

Design is where AI earns its keep. AI makes it possible to tailor comparisons to finely matched peer groups, to time information for maximum relevance and to combine peer data with other guidance. Each design choice could affect outcomes, and rigorous experimentation of the kind presented here identifies the best designs. The authors call for research on optimal design and on the channels transmitting the effects.

Crowdsourcing itself is the innovation. By aggregating anonymous information from many users, FinTech platforms provide benchmarks previously unavailable to individuals, turning dispersed data into useful signals. That is a distinctive capability of digital platforms. Protecting user privacy while delivering such benchmarks is essential to maintaining trust; a benchmark that leaks identities would destroy the very anonymity that makes it work.

Translated into policy, information-based tools can influence household finances without mandates. Encouraging tools that help consumers understand spending relative to relevant peers could support saving and financial resilience. Regulators can focus on transparency and data protection while letting such tools develop.

Read as a whole, information shapes consumption, and algorithms deliver information at scale. Consumers respond to peer benchmarks, especially on learning they overspend. As AI-driven financial tools become ubiquitous, their design will influence spending and saving across the economy. This study offers early, rigorous evidence of that influence and a roadmap for harnessing it.

One question to put to savings policy: a well-built comparison chart might move household saving. People who learn they are outliers tend to come back toward the middle on their own.