Summary: | 碩士 === 國立交通大學 === 工業工程與管理系所 === 106 === Taiwanese economy has always been dominated by small and medium-sized enterprises. The credit of financial institutions can provide sufficient funds for SMEs. When conducting credit, financial institutions need to comply with the Basel Capital Accord III and add 0 to 2.5% of Capital Conservation Buffer. If the amount of Capital Conservation Buffer is not enough, financial institutions cannot resist the fluctuations of the economic cycle. If the amount of Capital Conservation Buffer exceeds the real demand, it will reduce the funds employed available to financial institutions and the loss of investable capital. Therefore, financial institutions need to understand the trend of defaults and the total number of defaults in the future of SME loan cases, in order to be able to prepare appropriate Capital Conservation Buffer. The purpose of this research is to construct a model for predicting whether SME loans will default. This model considers the time characteristics of different SME loan cases and uses two-stage data screening to find the model data that meets the time characteristics of each SME loan case to be predicted. Then use logistic regression to establish their respective prediction models. A set of real data of SME loan cases by a financial organization in Taiwan is used to demonsttrate the reliableness of the propose model. The results show that he default trend predicted by this research method can be close to the actual default, and better than the Cox Proportional Hazards Model. This research method can provide financial institutions with effective and reliable information, help prepare appropriate Capital Conservation Buffer, and increase the flexibility of investable capital.
|