Investigation of the forecasting model for the loandefault from local small and medium enterprises-Taking the authentication accounts in a commercial bank as an example

• Main Authors: Yeh, Chun-yu, 葉春有
• Other Authors: Fang, Kuo-Jung
• Format: Others
• Language: zh-TW
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/39638124256040489608

碩士 === 實踐大學 === 企業管理研究所 === 93 === Since 1990, the government has started deregulation on finance and internationalization such as the establishment of new banks. In order to expand their business territory, many Taiwan financial groups started to raise funds for bank operation. An excessive number of financial institutions brought significant changes to the management environment. Because of imbalance of supply and demand of fund market, banks have already lost the leading situation on the market as in the past, that led to aggravating competition of this industry. Under performance pressure, banks lowered their credit granting conditions, and consequently, the quality of credit granting became worse. Meeting with economic recession and Asia financial crisis, the past-due loan issues appeared one after another. The ratio of loan overdue by banks in Taiwan was 4.15% by the end of June in 2004, the high ratio and low stock price had a significant influence on banks operation and investors’ rights. Small and medium enterprises (SMEs) are one organization form of domestic corporations. Taking the end of 2003 as an example, Taiwan SMEs were of 97.83% among all enterprises, which shows that the influence of SMEs’ development in the future on the society is really extensive. So the study is based on real life examples of bank loan. Factor Analysis and Discriminate Analysis(Mahalanobis’ Method) are the tools used to get some important structures from variables of SMEs’ operation environment. Data are divided into "model group" and "contrast group" and they are verified in two stages. The first stage takes data from the model group which is processed by Factor Analysis to determine factor scores, then uses the Bayesian approach with Mahalanobis’ Method and combines conditional probability density function with priori probability to construct Mahalanobis’ distribution function, also probes into the forecast ability of conditional probability density to develop a reliable model of SMEs’ Mortgage Default probability. The second stage takes data from the contrast group. Again, Factor scores are produced based on Factor Analysis, and Discriminate Analysis (Mahalanobis’ Method) is used on the discriminate model built during the first stage to determine the accuracy of the model forecast. The result of this research finds that of the 19 behavior variables used by credit granting practice for SMEs credit grading standards, through the use of Analysis Factor, Discriminate Analysis (Mahalanobis’ Method), a 98.44％ hit ratio can be achieved to effectively predict the probability of SMEs’ mortgage default. Under the concept of “pre-protection is always better than after remedy”, it is hoped that this model can help to set up a set of economic and effective system for SMEs mortgage default forecast. Therefore, precautionary measures can be taken when necessary to avoid non-performing loans.