Knowledge that Transforms

Machine learning (ML) could strengthen banks’ resilience through improved credit risk screening and ultimately benefit financial stability. Yet, ML adoption in banking remains limited, with simpler linear models still predominating. We argue that the emergence of highly flexible ML models has created a new challenge for forecasting tasks: ‘model multiplicity’—where equally accurate ML models at the aggregate level produce divergent individual-level predictions (‘predictive multiplicity’) or differ in their decision surfaces (‘procedural multiplicity’). These issues raise fundamental questions: Why should an individual or firm be subject to an adverse credit risk model outcome when there is an equally accurate model that treats them more favorably? Using the world’s largest loss database of corporate defaults, we examine these two phenomena in recovery rate ( RR ) modeling and propose heterogeneous ML ensembles as a natural solution. By combining predictions and decision surfaces from multiple well-performing ML models, ensembles mitigate risks associated with predictive multiplicity by ensuring that borrowers are not subject to the fluctuations of a single model, and reduce procedural multiplicity by providing a robust measure of features that ultimately improve out-of-sample RR predictions. By addressing the ‘multiplicity of good models’ problem, our study emphasizes the importance of model stability and provides new insights for the future development of ML models.

We use the staggered deregulation of interstate banking in the U.S. to show that CEOs who have gained career experience at multiple banks are more likely to pursue acquisitions when competition intensifies. Acquisitions completed by these CEOs perform better than those led by CEOs whose career experience is confined to a single institution. Analyzing the sources of performance gains, we find that CEOs with greater across-bank experience are more effective at identifying and integrating dissimilar targets. Our findings cannot be explained by other formative CEO experiences or a CEO general ability. The results highlight the importance of externally versus internally acquired experience in explaining how managers respond to a competitive shock.

Using an extensive matched hedge fund-prime broker panel dataset for the period 2001–2021, we document a strong positive relationship between hedge fund leverage and prime broker’s stock price crash risk after controlling for other crash risk drivers. Our results are not only statistically, but also economically significant, showing that a one-standard-deviation increase in hedge fund leverage is associated on average with an increase of around 5% of a standard deviation in the negative skewness or the down-to-up-volatility of bank stock returns. Moreover, they remain robust when accounting for endogeneity and conducting many robustness checks. We also document that some investment strategies, such as one focusing on fixed income, appear to decrease the slope of the risk metrics of prime brokers, and ultimately leading to lower stock price crash risk.