Advances in survival analysis methods and applications to the banking industry

The structure of the commercial bank industry in the United States changed considerably over the last four decades. The number of institutions insured by the Federal Deposit Insurance Corporation (FDIC) shrunk by almost two-thirds due to failures and mergers & acquisitions occurred during the Savings and Loan and the subprime mortgage crises. Since bankruptcies may have severe consequences on the whole financial sector and the real economy, it is important for regulators to identify which factors lead banks into financial distress and to estimate their default risk.

In this thesis, we present methodological advances in survival analysis, i.e. the statistical analysis of time-to-event data. First, we consider a location-scale regression model and we propose goodness-of-fit tests for the conditional mean and variance. Then, we focus on a more complex semi-parametric cure model, which allows researchers to handle situations where a portion of the population under investigation is likely to be immune to the event of interest. We extend it to time-varying covariates and we provide a variable selection technique based on its penalized-likelihood. To this effect we developed the penPHcure R package, which latest release is available on CRAN. Finally, we propose the use of Generalized Extreme Value regression to model the incidence distributions in a competing risks framework.

In support of further research, we believe that these methods could be applied beyond the field of management science. For example, in medicine, to address the current Coronavirus outbreak, spawning novel scientific collaborations to help mitigate the worldwide blow from COVID-19.