This paper examines whether a socially embedded soft trait, systematically omitted from AI-automated credit decisions, can predict financial outcomes in fully automated lending environments, and through which economic channel. We develop theoretically and examine empirically a labor-market competitiveness mechanism in which facial trustworthiness can serve as a soft signal under information asymmetry, shaping income growth and repayment capacity. Leveraging a unique proprietary dataset from a nationwide FinTech platform in China with AI-based underwriting that explicitly excludes facial information, we link loan performance to employment histories and construct trustworthiness scores using computer-vision algorithms. We find that borrowers with more trustworthy facial appearances experience greater upward job mobility and significantly lower default risk, conditional on rich hard information. A one-standard-deviation increase in facial trustworthiness is associated with a 4.6 percent reduction in default relative to the sample mean. The effect is strongest among borrowers with limited formal credentials, consistent with facial trustworthiness substituting for hard-to-observe labor-market signals. Our findings identify an economically grounded channel through which biometric appearance predicts financial resilience and highlight limits of algorithmic models that omit socially embedded information.

Structural credit risk models are widely used in default prediction practice, but standard formulations do not explicitly account for external support that may prevent default. We incorporate external support into the Merton and, more importantly, the Leland--Toft (LT) model with endogenous default. In the Merton model, support reduces realized default mechanically through a joint-default channel. In the LT model, anticipated support additionally raises debt value and affects equity holders' default decision. We derive closed-form expressions for the support-adjusted default boundary and default probability, and show that, under an economically relevant condition, stronger support lowers the endogenous default boundary, generating an additional indirect reduction in default risk. Empirically, we construct a newly assembled dataset of corporate debt defaults among Chinese A-share listed firms that combines bank loan defaults with bond default events. Using a composite measure of external support, we find that incorporating support significantly improves default prediction in both models. For the LT model, both the mechanical and endogenous channels contribute to this improvement, while the specification incorporating both channels delivers the strongest predictive performance. Our results highlight the importance of institutional support for structural credit risk modeling in China.

Using the Automatic Earthquake Information Release System (AEIRS) that China launched in 2013 as a quasi-natural experiment, which disseminates real-time seismic information to the public, we explore the impact of frequent information release on the pricing in the credit markets and whether such pricing is efficient. The results show a significant increase in loan pricing for firms with higher peak ground acceleration (PGA) scores following the AEIRS. Such effect is more pronounced in firms in regions where seismic risk was previously unrecognized. The baseline results still stand with several robustness checks including parallel trend analysis, spatial regression discontinuity design, placebo tests, and controlling for loan demand effect, etc. Moreover, the release of earthquake information does not alter the actual frequency of seismic events, ex-post loan defaults, or corporate fundamentals. The increase in loan pricing is primarily driven by banks’ overreaction induced by the availability heuristic, which in turn hinders corporate capital investment and asset growth. Collectively our empirical evidence reveals the dark side of frequent information release.

This paper examines how the integration of artificial intelligence (AI) and big data affects banking operations, emphasizing the crucial role of big data in unlocking the full potential of AI. Leveraging a comprehensive dataset of over 4.5 million loans issued by a leading commercial bank in China and exploiting a policy mandate as an exogenous shock, we document significant improvements in credit rating accuracy and loan performance, particularly for SMEs. Specifically, the adoption of AI and big data reduces the rate of unclassified credit ratings by 40.1% and decreases loan default rates by 29.6%. Analyzing the bank's phased implementation, we find that integrating big data analytics substantially enhances the effectiveness of AI models. We further identify significant heterogeneity: improvements are especially pronounced for unsecured and short-term loans, borrowers with incomplete financial records, first-time borrowers, long-distance borrowers, and firms located in economically underdeveloped or linguistically diverse regions. Our findings underscore the powerful synergy between big data and AI, demonstrating their joint capability to alleviate information frictions and enhance credit allocation efficiency.

Whilst there is increasing research attention on predicting financial distress, the existing literature is subject to two specific limitations. The first is that a firm can experience a financial distress event (e.g., loan default, bankruptcy) more than once, yet most studies that model corporate financial distress prediction treat financial distress as occurring only once. This approach leads to an inefficient use of data with all subsequent events being ignored and subsequently a decrease in statistical power. Second, to account for the lack of independence between observations of repeated event data, the extant research utilising hazard analysis either has a separate analysis for successive distressed events or relies upon robust standard errors. In addition to a much smaller sample, a separate analysis yields the models that can be used to predict the survival of a distressed firm rather than the survival of a firm generally. The method of robust standard errors, while innocuous to one-time event data, ignores the possible downward bias in coefficient estimates for repeated event data. To address these two limitations, we treat financial distress as repeated events and apply more advanced methods (generalised estimating equations, random effects, fixed effects, and a hybrid approach) to account for the lack of independence between observations in discrete time hazard analysis. These different approaches are applied to a sample of listed companies in China over the 2007‒2021 period. We find that variables that are not statistically significant in models based on one-time events data become statistically significant in the models based on repeated events data, and that coefficient estimates are larger in their magnitude with more advanced methods than with the method of robust standard errors. We also find that among the advanced methods, a hybrid approach achieves substantially better out-of-sample prediction, particularly over a long-term horizon than other approaches. Our results remain robust in tests of robustness.

Building on the important study by Allen, Qian and Qian (2005) and Ayyagari, Demirgüc-Kunt and Maksimovic (2010), I examine whether third party guarantors play an effective role in assessing loan risk. Using a proprietary database of third party loan guarantees in China, I find strong evidence that guarantors and banks disagree on pricing loan risk, and that banks can better predict loan defaults than guarantors. I also find that the probability of loan default is affected by the capability of guarantor officers. My findings question the contribution of soft information in the improvement of credit scoring and support the view that informal finance should be limited. This paper also supports the implications of studies on human capital in financial intermediation.

We examine empirically whether internal corporate governance mechanisms play a role in reducing the probability of being sued by lending banks due to bank loan default and the market reaction to the announcement of bank loan litigation. Using bank loan litigation events in Chinese financial markets, our results show that companies with better internal corporate governance mechanisms are associated with a lower probability of being sued. We also find a significant negative market reaction to the announcement of a bank loan filing while insignificant market reaction to the announcement of bank loan litigation verdict. Moreover, we test whether internal corporate governance mechanisms can play a role in mitigating the effect of market reactions. Our findings indicate that there is no evidence of internal corporate governance in mitigating this effect. Our paper suggests that internal corporate governance mechanisms are important in preventing the trigger of external governance mechanisms (litigation) but do not play any role once external governance (litigation) takes over.