Abstract:

The market for mass debt portfolios is among the most dynamically developing segments of the financial market, and the precise valuation of acquired portfolios constitutes one of the key factors determining the financial performance of debt collection entities. This article proposes a model for the valuation of mass debt portfolios based on clustering and historical operational data, integrating a quantitative approach with the decision-making logic characteristic of operational management in debt collection entities. The model divides a historical portfolio of receivables into homogeneous segments using an unsupervised machine learning method, assigning to each cluster historical financial parameters, including the recovery rate and the costs of the debt collection process, and then determines the present value of a new portfolio as the discounted sum of the differences between recoveries and costs for each segment. The empirical study was conducted on a dataset of 120,000 real mass debt claims obtained from the Polish debt collection market. The identified clusters differ in terms of debtor characteristics, which is reflected in the financial parameters of the segments and justifies their separate valuation. The proposed model is characterized by interpretability and operational simplicity and explicitly accounts for debt collection costs as a component of valuation, which corresponds to the actual decision-making logic of the management boards of debt collection entities and securitization funds that acquire mass debt portfolios.