Honors Theses
Advisor
Dr. Gayan Warahena Liyanage
Department
Mathematics
Publication Date
4-22-2026
Document Type
Honors Thesis
Abstract
Extreme Value Theory (EVT) provides a statistical approach for modeling rare and extreme events such as financial crises that can lead to bankruptcy. Traditional financial models often assume Gaussian distributions and rely heavily on historical trends, making them well suited for modeling typical corporate financial behavior. In contrast, models predicting municipal decline typically rely on urban indicators such as unemployment, housing prices, and population trends rather than corporate financial data. This study seeks to bridge these approaches by investigating whether extreme corporate financial events provide predictive information about municipal economic decline. Detroit, the largest U.S. city to declare bankruptcy, serves as a case study due to its strong economic dependence on the automotive industry. In this research, EVT is applied to the financial time series of Detroit’s three major automotive companies by fitting the Generalized Pareto Distribution to quarterly exceedances. Value at Risk and Expected Shortfall are computed from these fitted distributions and used as predictors in a regularized logistic regression model to identify periods of municipal decline. The resulting predictions are compared with Detroit’s Economic Condition Index using cross-correlation analysis and Granger causality tests to evaluate whether extreme corporate financial events provide early signals of municipal financial distress. The results provide a quantitative toolkit for linking corporate financial extremes to large scale economic downturns.
Permission Statement
This item is protected by copyright law (Title 17, U.S. Code) and may only be used for noncommercial, educational, and scholarly purposes.
Keywords
Undergraduate research
eCommons Citation
Coyne, Megan, "Extreme Value Theory and Corporate Financial Extremes in Predicting Municipal Decline" (2026). Honors Theses. 502.
https://ecommons.udayton.edu/uhp_theses/502
COinS
