CARD combines planning, controlled generation, and metric-guided revision to produce credit card Reddit threads whose aggregate statistics match real threads better than OASIS and SynthPAI.
Weighing Anchor on Credit Card Debt
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
We find it is common for consumers who are not in financial distress to make credit card payments at or close to the minimum. This pattern is difficult to reconcile with economic factors but can be explained by minimum payment information presented to consumers acting as an anchor that weighs payments down. Building on Stewart (2009), we conduct a hypothetical credit card payment experiment to test an intervention to de-anchor payment choices. This intervention effectively stops consumers selecting payments at the contractual minimum. It also increases their average payments, as well as shifting the distribution of payments. By de-anchoring choices from the minimum, consumers increasingly choose the full payment amount - which potentially seems to act as a target payment for consumers. We innovate by linking the experimental responses to survey responses on financial distress and to actual credit card payment behaviours. We find that the intervention largely increases payments made by less financially-distressed consumers. We are also able to evaluate the potential external validity of our experiment and find that hypothetical responses are closely related to consumers' actual credit card payments.
citation-role summary
citation-polarity summary
fields
cs.AI 1years
2026 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
CARD: Controlled Agentic Reddit Discussions for Credit Card Simulation
CARD combines planning, controlled generation, and metric-guided revision to produce credit card Reddit threads whose aggregate statistics match real threads better than OASIS and SynthPAI.