Working Papers

WP448. Improving Realized LGD approximation: A Novel Framework with XGBoost for handling missing cash-flow data

The scope for the accurate calculation of the Loss Given Default (LGD) parameter is comprehensive in terms of financial data. In this research, we aim to explore methods for improving the approximation of realized LGD in conditions of limited access to the cash-flow data. We enhance the performance of the method which relies on the differences between exposure values (delta outstanding approach) by employing the machine learning (ML) techniques. The research utilizes the data from the mortgage portfolio of one of the European countries and assumes the close resemblance for similar economic contexts. It incorporates non-financial variables and macroeconomic data related to the housing market, improving the accuracy of loss severity approximation. The proposed methodology attempts to mitigate the country-specific (related to the local legal) or portfolio-specific factors in aim to show the general advantage of applying ML techniques, rather than case-specific relation. We developed an XGBoost model that does not rely on cash-flow data yet enhances the accuracy of realized LGD estimation compared to results obtained with the delta outstanding approach. A novel aspect of our work is the detailed exploration of the delta outstanding approach and the methodology for addressing conditions of limited access to cash-flow data through machine learning models.

WP352. Concept of peer-to-peer lending and application of machine learning in credit scoring

Numerous applications of AI are found in the banking sector. Starting from front-office, enhancing customer recognition and personalized services, continuing in middle-office with automated fraud-detection systems, ending with back-office and internal processes automatization. In this paper we provide comprehensive information on the phenomenon of peer-to-peer lending in the modern view of alternative finance and crowdfunding from several perspectives. The aim of this research is to explore the phenomenon of peer-to-peer lending market model. We apply and check the suitability and effectiveness of credit scorecards in the marketplace lending along with determining the appropriate cut-off point.
We conducted this research by exploring recent studies and open-source data on marketplace lending. The scorecard development is based on the P2P loans open dataset that contains repayments record along with both hard and soft features of each loan. The quantitative part consists of applying a machine learning algorithm in building a credit scorecard, namely logistic regression.

WP301. A comparison of credit scoring techniques in Peer-to-Peer lending

Credit scoring has become an important issue as the competition among financial institutions becomes very intense and even a slight improvement in accuracy of prediction might translate into significant savings. Financial institutions are seeking optimal strategies through the help of credit scoring models. Therefore credit scoring tools are widely studied. As a result different parametric statistical methods, non-parametric statistical tools and soft-computing approaches have been developed in order to increase the accuracy of credit scoring models. 

In this paper different approaches to classify customers as those who pay back loan and those who default on a loan will be employed. The purpose of this study is to explore the performance of two credit scoring techniques, the logistic regression model and neural networks. In order to evaluate the feasibility and effectiveness of these methods analysis is performed on Ledning Club data. Peer-to-Peer lending, also called social lending are investigated. On the basis of the results, we can conclude that logistic regression model can provide better performance than neutral nets.