Applying the Grey Models to Binary Data - A Study about Factors Impacting on Default Risk of the Issued Cash Cards

K.-T. Hsu, T.-M. Yan, K.-Y. Lo, and F.-H. Chen (Taiwan)


grey model, cash card, GM (1,N) model, GM(0,N) model.


The response variable in this study, i.e. occurrence of overdue loans is binary variable. Traditionally, logistic regression has been one of the most preferred methods applied to binary dependent variables. Based on the results of logistic regression analysis, we try to find out the way for applying grey models in binary data and test the applicability of grey models. Before issuing cash card, the issuing banks evaluate the possible default risk according some selected factors. The factors, which are employed by the issuing bank observed in this study, are sex, family status, age, sources of application, income type, occupation, ownership of residence, job seniority and education. Whether these factors are in practice irrelevant and induce huge loss of the issuing banks observed in Taiwan, is the first of our main problems. The factors enter in the final model of logistic regression are income type, ownership of residence and education. Based on these results, we find that the applications of GM (1,N) model and GM (0,N) model in original sequences are not satisfactory. By means of applying GM (0,N) in the grey relation generating data, we can reach the most appropriate results. The results illustrate that the influence of six of nine evaluation factors used by bank personnel are not consistent with expectations.

Important Links:

Go Back