Personal Credit Scoring Model and Its Applications

Author: Shi Qingyan
Publisher:
Publish Date: 2006-01-01
Features: Since the 1990s, with the rapid development of China's economy, credit consumption in China has gradually emerged. Various personal consumer loans, such as credit card consumption, personal auto loans, durable goods loans, education loans, and mortgage loans, have been introduced. So far, although the proportion of personal loans in the total loan balance of Chinese commercial banks is not high, the growth rate of personal loans has been very fast. The rapid growth of personal consumer credit requires commercial banks to have a relatively complete credit risk management system. However, the current situation is that Chinese commercial banks still have significant shortcomings in personal loan risk management. The main manifestations are as follows: , a lack of a comprehensive understanding of the credit status of borrowers. Due to the preliminary stage of personal credit reporting system construction in China, commercial banks cannot obtain credit data of loan applicants from credit bureaus like developed countries, making it impossible for banks to fully understand the credit status of loan applicants. The problem of information asymmetry is particularly prominent when issuing personal loans. Second, a lack of scientific methods for evaluating personal credit. When assessing the credit risk of borrowers and deciding whether to issue loans, the main reliance is on the subjective judgment of credit personnel of credit-issuing institutions, thereby determining whether to grant certain consumers the right to credit consumption. Precise credit scoring methods are rarely used. The disadvantages of this subjective evaluation method are quite obvious: (1) It is highly subjective and cannot accurately measure consumers' credit risk; (2) At the same time that risk cannot be accurately measured, the cost of risk control is high. The characteristics of personal credit business are that the transaction volume of each single business is small, but the number of transactions is large. Therefore, credit evaluation and control methods that rely mainly on the judgment of credit personnel not only fail to accurately measure personal credit levels but also cannot effectively reduce the management costs of individual loans. The rapid growth of personal consumer credit and the backward credit risk management techniques and methods of Chinese commercial banks have created a stark contrast, urgently requiring commercial banks to improve risk management systems that match the growth of consumer credit. As Chairman of the China Banking Regulatory Commission, Liu Mingkang, said: "It is essential to pay great attention to controlling and preventing bad loans in the field of consumer loans. A perfect credit rating method must be established, either through a five-level classification or a scoring card method, but a complete and rigorous risk supervision system must be in place to address the fierce growth of consumer loans." Personal credit scoring is a quantitative analytical method for predicting the likelihood of default by loan applicants or existing borrowers. In countries with a more developed social credit system, such as the United States, since the method was introduced into the field of consumer credit management in the 1950s, personal credit scoring technology has made significant progress, and its application areas have become increasingly broad. So far, personal credit scoring has been widely used not only in consumer credit fields such as credit cards but also increasingly in mortgage loans. Before the 1990s, the application of credit scoring in corporate loans was relatively limited, one of the main reasons being the significant differences among different corporate borrowers, making it difficult to establish relatively accurate scoring models. However, with the emergence of new methods, the enhancement of computer capabilities, and improved data acquisition, the situation is changing. Since the mid-1990s, many banks have begun to use credit scoring methods to evaluate small business loan applications. Foreign experience shows that personal credit scoring has the advantages of quickly processing customer loan applications, low processing costs, consistency and objectivity in handling customer applications, and the ability to quantitatively assess customers' credit risk. Therefore, it plays an important role in consumer credit risk management and has been successfully applied in areas such as small business loan application evaluation, credit card fraud prevention, risk-based interest rate pricing, direct response scoring, and asset securitization. However, for a credit risk management technology that has proven effective abroad, domestic commercial banks use it less, and domestic academia has conducted relatively little research in this area. Of course, due to the imperfect social credit reporting system, establishing personal credit scoring models in China still faces many difficulties and challenges. Under current conditions, can commercial banks utilize data and information from existing information systems to establish customized scoring models for different products? Among the credit scoring methods used abroad, which methods work better in China, and which are less suitable? Among the available data, which explanatory variables have a greater impact on improving the predictive accuracy of the model? Are there regional differences in consumers' credit behavior? These questions all require systematic exploration and research. This book is an attempt by the author in this regard. Based on an introduction to the establishment, implementation, and monitoring of personal credit scoring models abroad, it explores some problems and solutions encountered in establishing personal credit scoring models in China. Some parts of the book are adapted from Shi Qingyan's doctoral dissertation. The arrangement of the book is as follows: Chapter 1 introduces the basic concepts of personal credit scoring; Chapter 2 discusses methods for personal credit scoring model data and preprocessing; Chapter 3 introduces methods used to establish personal credit scoring models, including both traditional statistical methods such as discriminant analysis and newly developed methods such as neural networks and genetic algorithms; Chapter 4 discusses how to test personal credit scoring models; Chapter 5 compares several commonly used personal credit scoring models using data from Chinese commercial banks; Chapter 6 explores the comprehensive application of multiple methods to establish credit scoring models; Chapter 7 proposes solutions to the problem of insufficient sample size commonly encountered in personal credit scoring model building for Chinese commercial banks; Chapter 8 introduces behavioral scoring models; Chapter 9 discusses the implementation and monitoring of personal credit scoring models; Chapter 10 extends personal credit scoring models by incorporating a "customer activity level" score into the credit decision-making process. Mr. Lin Junyue, Vice Chairman of the Credit Work Committee of the China Market Association, has always been concerned about and supported the publication of this book, and has put forward many valuable suggestions. The editorial team of China Zhengfang Publishing House has provided strong support for the publication of this book. The author sincerely thanks them for their enthusiastic support and assistance. Due to the author's limited expertise and the fact that the development of personal credit scoring models in China has just begun, this book is inevitably flawed and incomplete. We sincerely hope to receive criticism and corrections from experts and readers. If the publication of this book can serve as a catalyst for further research, it will have achieved the author's purpose in writing it. Shi Qingyan Qin Wanshun September 2005, Peking University