Predicting Which Banks Can Survive The Threat of Bankruptcy using Pooled Cross Sectional Time Series Regression Analysis
Open Access
- Author:
- Cook, Justin Michael
- Area of Honors:
- Finance
- Degree:
- Bachelor of Science
- Document Type:
- Thesis
- Thesis Supervisors:
- James Alan Miles, Thesis Supervisor
James Alan Miles, Thesis Honors Advisor
Chris Muscarella, Faculty Reader - Keywords:
- pooled regression
binomial regression
Ed Altman's Z-Score
financial distress
prediction of bankruptcy
time series
SIC code - Abstract:
- This paper focuses on determining which financial measures and ratios are the best to use when predicting whether or not a financial firm will survive as a going concern or decline into financial distress, ultimately declaring bankruptcy. Ed Altman’s Z-Score is tested on a sample of 40 financial firms, 20 that survived the financial crisis, and 20 that did not, for every year for each of these firms over a timeframe of 2005-2009. Key profitability, debt management, liquidity, and other financial measures and ratio are chosen as independent variables to analyze. Bloomberg, Mergent Online, and Compustat from WRDS, are used to pull data for these variables over the timeframe, and excel is used to compile this data and run regressions using binary dependent variables, as well as used to run a pooled cross sectional time series regression that takes temporal aspects into considerations. Resulting analysis of Altman’s Z-Score showed that it is unsuccessful in determining the onset of financial distress for financial firms over the time frame selected, but is still successful at predicting it for non-financial firms. After analyzing the results of my simple regression model, there were a couple of variables that had a positive correlation with a firm’s level of financial distress, but overall, it was also not reliable in determining the onset of financial distress for financial firms. Through the use of the pooled regression, the results significantly improved and there were four independent variables that were statistically significant at the 95% confidence level, and two independent variables that were statistically significant at the 90% confidence level. I also have comments about sample selection; variable selection, regression analysis, and financial modeling methodology that will help someone conduct a more similar analysis in the future.