In the complex world of banking and debt recovery, making strategic decisions on how to assign delinquent customers to different recovery strategies is crucial. This project dives into a scenario where a bank employs various recovery strategies based on the expected amount they anticipate recovering from each customer. The primary goal is to assess, in this non-random assignment, whether the incremental amount earned surpasses the additional cost of assigning customers to a higher recovery strategy.
Threshold assignments, like the one in this project, are prevalent in various domains such as medicine, education, finance, and the public sector. Regression discontinuity, a powerful analysis method, proves invaluable in situations involving threshold assignments.
Project Tasks:
- Explore the concept of regression discontinuity in the context of banking recovery strategies.
- Conduct graphical exploratory data analysis to gain insights into the recovery strategy data.
- Apply statistical tests to analyze the relationship between age and the expected recovery amount.
- Investigate the impact of gender on the expected recovery amount through statistical testing.
- Visualize the distribution of recovery amounts for a comprehensive understanding.
- Conduct statistical analysis on recovery amounts, exploring patterns and trends.
- Develop a regression model without incorporating a threshold, laying the foundation for comparison.
- Introduce a true threshold to regression modeling, enhancing model accuracy and relevance.
- Fine-tune the regression model by adjusting the window, ensuring optimal performance.
Join this data-driven exploration of banking recovery strategies, employing regression discontinuity and statistical analysis to make informed decisions in the intricate realm of debt recovery.