Empirical Analysis Of Credit Scoring Based On A Bi-probit Model

The credit cards business has become the focus area of the future developmentof the commercial banks. Without a relatively perfect system for risk measurement andcontrolling, credit issuers have to focus the risk management on the target customerselection and the credit audit policy. As a result, researches about credit scoring becomemore and more important, and to construct an appropriate credit scoring model forcredit markets in China is of great need.This article intends to find out the factors affecting the granting decision of theissue banks and the default probability of the credit card applicants. A bi-probitapproach is adopted to conduct the modeling. In this article, both the credit grantingprocess and the loan performance process are taken into consideration. The data in thisarticle were collected through questionnaire surveys. The website wenjuanxing.com andthe Shenzhen north station are chosen to conduct the survey. The model in this articlewas estimated by Maximum Likelihood Estimation (MLE).The conclusions that can be drawn from the analysis are:(1) People with asavings account in the issue bank are more likely to apply for a credit card successfully;(2) People have their own estate are more likely to apply for a credit card successfully;(3) Students are more likely to get a credit card;(4) People with a relatively highmonthly income are more likely to apply for a credit card successful;(5) people with arepayment schedule and the―%disposable income customer is paying on loanrepayment‖is lower than30%are more likely to have good performance in loanrepayment process;(6) People have delinquent history are more likely to default in theloan repayment process.