本研究以中部某公營銀行於2006年~2011年所承做336件中小企業貸款之授信案件為研究對象,從銀行授信審核及承作條件觀點,參酌銀行授信企業信用評等表及銀行往來狀況,選取20個變數分析影響中小企業貸款提前清償之關鍵因素,以提供銀行於承作中小企業貸款業務及擬定授信策略與風險的控管之參考,以避免資源成本的浪費,更期望能對銀行實質的收益及整體的經營績效有所助益。 透過卡方檢定與羅吉斯迴歸模式,分析影響中小企業貸款案件提前清償相關因素與發生機率之差異性。實證結果發現產品市場性、稅後淨利率、存貨週轉次數、應收帳款週轉率及貸款利率等變數,對於中小企業貸款案件提前清償因素有顯著影響,其中產品市場性、稅後淨利率、存貨週轉次數與貸款利率對中小企業貸款案件提前清償有顯著正向影響,而應收帳款週轉率對中小企業貸款案件提前清償則有顯著負向影響。 應用混淆矩陣預測,對未提前清償戶、提前清償戶及整體中小企業貸款案件之預測準確率分別為96.8%、69.0%及89.6%。然而使用資料期間將影響預測準確率,當刪除不穩定期間資料,將使預測準確率分別提高為98.1%、72.7%及91.4%。
The research objects are the 336 small and medium enterprises (SMEs) loans undertaken by a government-owned bank during 2006-2011. According to loan auditing, lending policies, the credit ranking of the business and records of interaction with banks, 20 variables are selected to analyze the critical factors of affecting the prepayment of SMEs loans. The results can offer banks handling SMEs loans business and framing loan policy and risk control strategy, then can ultimately avoid wasting resources, besteading physical revenue and overall operational achievement. Chi-square test and Logistic regression analysis model will be applied to exam the differences the probability among relevant variables that has significant influence on the prepayment of SMEs loans. The results suggests that “product market”, “net after-tax profit margin”, “inventory turnover”, “Interest rate”, and “accounts receivable turnover” are significant influence on the prepayment of SMEs loans. The former four factors have positive impact while the last one has the negative influence on the prepayment of SMEs loans. Applying the confusion matrix prediction, the prediction accuracy rates on “no prepayment cases”, “prepayment cases” and “overall SMEs loan cases” are 96.8%, 69.0% and 89.6% respectively. However, the prediction accuracy rates will be affected by the data stability. Those will significantly increase to 98.1%, 72.7% and 91.4%, respectively as deleting the fluctuating period of data.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。