BEYOND THE MINIMUM: FIELD EXPERIMENT AND MACHINE LEARNING ON CREDIT CARD PAYMENT WARNINGS

The minimum payment warning — a mandatory credit card disclosure in many countries — has been described as a perverse nudge because it may reduce repayments through anchoring bias. This paper introduces a novel “statement balance warning” highlighting the consequences of repaying less than in full. In a large-scale randomised field experiment with 179,706 borrowers, email payment reminders were randomly assigned to four conditions: a minimum payment warning, a statement balance warning, both, or no warning. The statement balance warning increased full repayment by 0.6–0.7 percentage points and reduced revolving interest charges by 6.8–9.0 percent. Causal random forests reveal significant heterogeneous effects: the top 40 percent of cardholders receiving the statement balance warning increased full repayment by up to 2.4 percentage points, particularly those who deliberately vary monthly payments and face a narrower minimum-statement balance gap. The results suggest that cardholders treat warnings as targets rather than anchors, prioritising the more attainable goal when faced with two salient amounts — providing new evidence on dual-level targets in financial decision-making. An online experiment confirms these findings and shows that the warning improves financial understanding. Based on this evidence, Chile has approved the statement balance warning for inclusion in its new national credit card regulation.