|dc.description.abstract||Having not yet considering much on consumer loan management, the banking set few functions on debt-collection for past due loan customers but lack such functions for review management and payment behavior analysis. Nowadays, the banking should know how to manage, explore and apply his unique internal information related to large credit underwriting customers under the circumstances of unstable profit and intensely competition, especially when the consumer loans contribute major profit for him. Because consumer loans have the characters of small amount and numerous accounts, the trade-off between risk management and profit development has become primary issue for banking under the consideration of lowest cost and highest efficiency.
This research studies the risk factors of past due loan based on one commercial bank data. Three methods are used for this research. There are logistic regression analysis, discriminate analysis and probit regression analysis. The results shows that academic degree, loanratio, wordingyear, cardusesum, latetime, checknum, loanyear affected a lot on such kind of past due loan. Besides, by using above-mentioned methods to attest the accuracy of factors, the correct percentage of each model are closely, that are 72.2% of logistic regression analysis, 71.15% of discriminate analysis and 72.2% of probit regression analysis. The difference among these analyses is very little, but the logistic and probit regression analysis performs the same predict ability.
The hit rate of contract violated ratio model could apply to the banking practical application to get the adequate balance between credit policy and risk control. The bank fully discloses the risk level of loan quality management in order to stabilize the performance and compliance the regulatory.