We examine whether lenders incorporate behavioral signals into consumer credit pricing. Exploiting regulatory credit card reforms that altered the information content of a sophisticated credit card behavior, we show that borrowers exhibiting less sophisticated credit card usage subsequently experienced significantly higher default and bankruptcy rates across multiple consumer lending products. Despite this predictive power, lenders did not price this signal ex ante. Instead, contract terms adjusted only after repayment deterioration became observable through traditional credit metrics. These findings reveal a timing gap between observing predictive behavioral information and incorporating it into pricing decisions.
In many countries, small businesses rely on loans backed by personal guarantees from company managers, making managers personally liable for loan repayment if the business fails. Such guarantees may discourage managerial risk-taking by increasing the personal costs of failure. Exploiting a Japanese policy reform aimed at reducing the use of personal guarantees by encouraging monitoring-based loans, this paper examines which firms opt out of personal guarantees and choose monitoring-based loans, and how these loan arrangements are associated with firm risk-taking and performance. We find that: (1) monitoring-based loans carry higher interest rates; (2) firms with stronger credit ratings and lower managerial ownership are more likely to choose monitoring-based loans; and (3) after accounting for positive borrower selection, firms using monitoring-based loans exhibit greater risk-taking, reflected in a higher likelihood of default but stronger growth conditional on survival. Without controlling for selection through managerial ownership and credit ratings, firms borrowing through monitoring-based loans exhibit both lower default probabilities and even higher growth, indicating substantial positive self-selection into opting out of personal guarantees. Overall, the results suggest that allowing borrowers to choose between loans with and without personal guarantees facilitates positive selection into monitoring-based lending and may promote performance-enhancing risk-taking among SMEs.
This paper documents debt misreporting in unaudited financial statements and examines its sources and implications for bank lending. Using a unique combination of datasets, we measure debt misreporting as the difference between firms’ bank debt reported on the balance sheet and the corresponding information recorded in the public credit registry. We find that debt misreporting is not randomly distributed and that firms with greater financing needs are more likely to underreport, rather than overreport, their debt. We further show that underreporting is more prevalent when firms approach new lenders, consistent with firms using it in an attempt to obtain credit. In terms of credit supply, these firms receive less credit, particularly from banks with greater prior exposure to misreporting and from banks that verify borrower information through the credit registry. In addition, underreporting firms exhibit subsequent signs of financial weakness, including delays in supplier payments and a higher probability of non-performing loans, charge-offs, and insolvency. Overall, our findings suggest that a substantial part of debt underreporting is strategic and that banks take it into account when making credit decisions.
We study tradeoffs that lenders face when borrowers breach covenants and quantify the implications for the lender's revealed preference for continuing lending relationships. When a borrower breaches a pre-set covenant threshold, the lender faces a choice between enforcement-which generates fees and behavioral concessions that reduce default risk-and forbearance, which preserves the relationship but forgoes these benefits. Using a fuzzy regression discontinuity design and simple tradeoff model, we estimate that the lender's revealed preference for continuing the marginal lending relationship at the enforcement threshold is 11.6% of loan amount and find heterogeneity in this estimate that is consistent with theories of information-based relationship value.
