Abstract
dc:descriptionDuring the last twenty years, the large number of bank failures channeled a lot of interest to credit rating methodologies and techniques for the early identification of a potentially failing bank. The financial crisis initiated in 2007, raised serious concerns on the regulatory policies in effect and led to significant supervisory and regulatory reforms to stabilize and increase the resilience of the banking system. Stress-testing became a popular tool among regulators and researchers in the effort to assess the resilience of banks under alternative scenarios. In this dissertation, a novel forecasting model of bank failures based on machine-learning is developed and presented. The proposed methodology defines a linear decision boundary that separates the solvent banks from the failed ones. This setup generates a novel alternative stress-testing tool for the assessment of solvency and provides significant insights on how to prevent a bank from failure. Moreover, based on this model, I investigate the effect of the regulatory changes that happened after the financial crisis on the resilience of the banking system as a whole. In doing so, I consider all 5826 banks of the U.S. banking system for 19 years. I use only quantitative and publicly available information from banks' financial statements from 2000 to 2018 aiming to develop a model for a) forecasting bank credit ratings b) forecasting bank failures and c) assessing the resilience of U.S. banking system as a whole before and after the recent financial crisis. The results indicate that bank credit ratings largely rely on historical data thus responding and updating very late and very sluggishly, at times long after the reaction of the market. The proposed forecasting model for bank failures exhibits a 99.22% overall forecasting accuracy being very parsimonious by using only two explanatory variables. I provide evidence that in recent years, lower competition and new regulations widened the safety margin of the banking system, resulting in a healthier and more resilient financial sector.
Degree
thesis:*- Grantor dc:publisher
- Democritus University of Thrace (DUTH)
- Year dc:date
- 2021
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Agrapetidou, Anna
- Αγραπετίδου, Άννα
Subjects
dc:subject × 14Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/50538