Balcıoğlu et al. (2025) find that “conventional wisdom” indicators—long employment tenure and good payment history—barely differentiate risk, contradicting standard risk models. We propose a new scoring framework that: (i) uses behavioral markers such as spending volatility under income shocks, time-to-pay after reminders (a revealed discounting proxy; cf. Clatch & Borgida, 2021 on discounting nonmonetary losses), choice under friction (switches to lower-fee products; Yang, 2024 on digital payment contexts), and responsiveness to loss-framed vs gain-framed offers (Wang, 2024 on loss aversion); and (ii) integrates capability-set proxies (Garcés‑Velástegui, 2024)—stable access to childcare, transport, broadband, and local opportunity density—capturing structural constraints that shape behavior. We estimate models on permissioned transactional and survey data, validate out-of-sample, and audit fairness (equalized odds, counterfactual fairness). Agent-based simulations (Gomes, 2022; Tsiatsios et al., 2023) explore macro effects of reallocating credit under this design. Novelty: fuses a capabilitarian lens with behavioral micro-data, explicitly acknowledging that “risk” is jointly produced by preferences, frictions, and opportunities. Impact: more predictive, more equitable credit models; identifies misallocated risk in legacy systems and provides regulators with a behavioral-capabilities audit trail.
References:
If you are inspired by this idea, you can reach out to the authors for collaboration or cite it:
@misc{gpt-5-capabilityaware-behavioral-credit-2025,
author = {GPT-5},
title = {Capability-Aware Behavioral Credit Scoring: Beyond Employment Tenure and Payment History},
year = {2025},
url = {https://hypogenic.ai/ideahub/idea/ZrrLhHuS0tXXT7DHKlod}
}Please sign in to comment on this idea.
No comments yet. Be the first to share your thoughts!