|
Under the limited scale in Taiwan, the operating environment of the banking industries has revealed a highly competitive trend. The difference of the interest rate of various banks between deposit and loan becomes smaller and smaller progressively and a cutthroat competition is formed. In order to strive for more customers and to seek for more profit, most often banks will reduce the credit check and credit granting quality and expand the credit loan. One of the reasons of the burst of the Asia monetary storm in 1997 is due to excessive expansion of granting credits by financial institutions. Following the detonation of the landmine stocks, collapse of enterprises became more and more and the excessive lending rate of financial institutions rose rapidly that eroded the profitability seriously. In order to diversify the risk and to increase the profitability, consumer finance business that belongs to high interest difference product has become the new domain that domestic banks are fighting amongst themselves to develop at the present stage.
In the consumer finance business, for the examination on small credit loans by financial institutions in recent years, mostly it has now been changed to utilizing credit scoring system as a supplement in order to increase the operation efficiency compared to basing on judgment and experience of the examiner in the past. However, with reference to the previous literatures, for the research on the general constructed credit scoring models, its sample source will only adopt the information of applicants of small loans that have passed the examination and do not include the rejected samples excluded in the beginning. However, the credit loan application scoring should be utilized in the evaluation of all loan applicants. Therefore if only the scoring model established on the samples that have passed the examination is used, then no matter whether it is the selection of variables or the weight design, there will be errors generated due to insufficient coverage. Therefore the research result will still exist in the selected sample errors. In order to modify this deviation and in order to form a complete sample source, therefore, the rejected samples are also added in so as to solve the problem of the deviation in sample selection. The utilization of sample selection model adopts two phases method and relevant variables are selected so as to establish an appropriate small credit loan application scoring model hoping that it can rapidly and objectively detect the high and low of the credit risk of the applicants.
Based on what is shown in the credit application scoring model that amongst the basic population statistics variables, the factors of probability of contract default by the applicants are 「age」, 「marriage」, 「education level」, 「job title」and 「any real estate」. In the variables of the bank credit transaction record, 「the total of number of loans of financial institutions」, 「number of possessed cash card」, 「number of deferred payment of credit card in recent one year」and 「number of bank inquiry for the recent three months」are also factors that will significantly influence whether the applicant will default the contract.
To the experiment samples, the overall prediction rate of the credit application scoring model after adding in the rejected samples is 84.0% and the overall prediction rate of the prediction samples is 70.6% and there is a drop of the prediction capability. However, for applicants with high risk of contract default, considerable level of identification capability can be provided. Therefore, there will be certain help on enhancing the examination efficiency on credit loan applications of financial institutions and on reducing the loss due to bad debt from contract default.
In addition, this research also transform the credit loan scoring model into the credit scoring sheet that the first line examination personnel can use directly so that it can be applied directly on the bank credit loan examination. In so doing, the credit scoring model and the credit scoring sheet can bring its practical value to full play. That means through the method of quantifying the information of loan applicants, this type of information can be applied to evaluate the size of the risk of the credit of the borrower and can serve as a reference for deciding on approval or rejection of the loan. Also it is hoped that the bad debt rate of the small amount credit loan can be reduced so as to enhance the operating performance of banks.
|