Behavioral Household Finance
Paper Session
Sunday, Jan. 3, 2027 8:00 AM - 10:00 AM (EST)
- Chair: Arna Olafsson, Copenhagen Business School
Targeting Higher Credit Card Payments
Abstract
In a field experiment, I test two nudges in a credit card payment app: shrouding the credit card minimum payment amount and an option to pay 50% of the statement balance. After six statements, neither nudge significantly increases credit card payments. This finding contrasts with prior evidence from highly-replicated online experiments that finds large increases in hypothetical payments from shrouding the minimum payment. My results demonstrate that the minimum payment does not act as an anchor that lowers payments. Instead, credit cardholders frequently lack liquid cash and target paying at least the minimum payment, which acts as a reference point.The Financial Psychology of Shrinking Families: How Demographic Decline Shapes Investor Be
Abstract
Fertility rates are declining worldwide, producing a growing share of only children. We exploit China's One-Child Policy in a regression discontinuity design to study how family structure shapes investor attitudes and their consequences for financial innovation adoption. Using nationally representative data covering over 30,000 household-year observations, we first establish that only children are significantly more risk-averse, less trusting, less socially engaged, and less competitive than individuals with siblings, extending a landmark psychology finding to the field. We then show that these attitude differences reduce participation in FinTech (internet lending, digital wealth management, and online payments) but not traditional finance, with economically large effects that survive controls for wealth, income, and internet access. The effect is concentrated among less educated individuals and in regions with weaker investor protection. Only children also react more negatively to FinTech platform scandals, confirming trust as the dominant channel. As the financial system becomes increasingly digital, demographic decline may impose a psychological constraint on financial modernization.Eliciting Stopping Times
Abstract
"We study how individuals plan dynamic risk-taking by eliciting stopping times—complete contingent plans of when to continue or stop taking risk. We introduce an experimental method to elicit stopping times and apply it to a setting in which subjects can repeatedly take a binary, fair and symmetric risk up to five times. By focusing on the simplest possible single risk, we isolate dynamic aspects of risk-taking. The method yields the first incentivized data on unconstrained stopping times and enables a direct comparison between precommitted risk-taking plans and sequential risk-taking actions without commitment.The elicited stopping times display known salient properties of repeated risk-taking, suggesting that our method captures economically meaningful behavior. Most subjects start taking the repeated risk and do so using continue-when-winning and stop-when-losing strategies. While such behavior is inconsistent with risk-averse expected utility (EU) preferences, it is expected under non-EU preferences (e.g., Barberis 2012; Ebert and Strack 2015) and has been documented in previous studies (Strack and Viefers 2021; Dertwinkel-Kalt and Frey 2024; Heimer et al. 2025).
We then document three new facts about stopping times. First, stopping times exhibit substantial heterogeneity, but an unsupervised machine-learning algorithm identifies a small number of economically interpretable strategy clusters, including stop-loss, take-profit, buy-and-hold, take-the-risk-once, and never-start strategies. Second, we propose a structural estimation of prospect theory parameters from stopping-time data. The estimated parameters map well to the strategy clusters, implying that prospect theory is a useful starting point for modeling stopping behavior. Third, path-dependence and randomization are both used often. About 55% of subjects use path-dependence at least once, while about 80% use randomization at least once.
In the second part of the paper, we use our experimental method to obtain new insights on several economic applications. First, we provide evidence on subjects’ mental models when planning repeated risk-taking by analyzing the order in which they construct stopping times. About 60% of subjects plan using forward induction, whereas only about 27% use the normatively prescribed backward induction.
Second, we show that trailing stop-loss strategies are a better description of observed stopping times than threshold stop-loss strategies and are used about 1.5 times as often. Restricting subjects to threshold strategies, however, does not significantly affect aggregate measures of stopping times.
Third, comparing each subject’s risk-taking plan with her sequential risk-taking actions, we document the standard shift: subjects plan to stop after losses and to continue after gains, but systematically deviate from these plans by continuing longer after losses and stopping earlier after gains (cf. Disposition Effect). At the same time, we find that subjects follow their plan for 89% of their risk-taking actions, implying that the standard shift arises from a small number of systematic deviations rather than from an overall inconsistency.
Fourth, we show that restricting plans to threshold strategies mechanically overstates the standard shift by about 50%.
Fifth, we study the sources of dynamic inconsistency using additional between-subject studies and provide a behavioral decomposition of the standard shift. We find that 38% of the shift is attributable to a lack of commitment, 34% to memory constraints, 15% to planning itself, and 13% to mechanical inconsistencies induced by planning constraints. Nudges in the form of plan-consistent defaults have no measurable effect.
By eliciting unconstrained stopping times, we hope to inspire experimental and empirical work on applied research questions. Many, if not most, risks that people face can be taken repeatedly and stopping problems abound. Examples include the sale of an asset, job search, real options, or the decision to conclude data collection. The experimental method introduced in this paper can be adapted for such purposes. We also hope to inform the fundamental and applied theoretical stopping literature. Given this large literature (summarized, for example, in the textbooks of Shiryaev 2007 or Björk, Khapko, and Murgoci 2021), it may be surprising that unconstrained stopping times have not been elicited yet."
Discussant(s)
Francesco D'Acunto
,
Georgetown University
Ben Keys
,
University of Pennsylvania
Da Ke
,
University of South Carolina-Columbia
Michael Ungeheuer
,
Aalto University
JEL Classifications
- G4 - Behavioral Finance