Fraud Detection
This project analyzes accounting fraud using a logistic regression model based on established fraud detection scores—Beneish’s M-Score and Dechow’s F-Score—along with three additional financial indicators: debt-to-capital ratio, changes in short-term debt, and cash dividends. Using SEC enforcement actions as the fraud benchmark, we find that Dechow’s F-Score is consistently more predictive of fraud than the M-Score. Cash dividends show a statistically significant negative relationship with fraud, suggesting that stable dividend payments may signal financial health. However, results vary by year, highlighting the complexity and evolving nature of fraud detection
Disclaimers:
-
Variable definitions and structure based on coursework at the Wharton School, reinterpreted and summarized for independent presentation.
-
This analysis was originally developed as part of coursework at the Wharton School of the University of Pennsylvania. The opinions and conclusions are my own and do not reflect the views of the institution or its faculty.
-
This analysis is for educational purposes only and should not be interpreted as financial advice or a recommendation to invest.