Design of Default and Profit Model Allocation for Different Risk Levels

• Main Authors: Ho, Yun-Yi, 何韻儀
• Other Authors: Kao, Ming-Sung
• Format: Others
• Language: zh-TW
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/7k3ju7

碩士 === 輔仁大學 === 金融與國際企業學系金融碩士班 === 107 === P2P lending has the problem of information asymmetry. In the past, the literature proposed a model based on predicting the probability of default. Later, some literature noticed the profit of investors and platforms and then proposed a model for predicting IRR. However, these documents use the overall data as the source of modeling data, and the prediction ability for the high-risk interval is poor. Therefore, this study will divide the risk interval according to the Lending Club risk level (A~G), and each interval contains different risk levels. Establish multiple default probability models with machine learning-XGBoost and test the results of seven risk levels to improve the high-risk interval prediction ability and design the profit model with the target of calculating the expected return rate and profitable amount, respectively, with the probability of default. And the two aspects of profit, looking for the best configuration combination for analysis, giving investors or the platform to choose the optimal allocation combination according to the set goals. The study will be divided into five chapters, namely, introduction, literature review, research methods, empirical results and analysis, conclusion and suggestion.