[[alternative]]NON-PERFORMING HOUSE LOAN FORECAST MODEL

[[abstract]] 本研究之目的為減少銀行發生不良債權，導致無法回收的現象發生，及為銀行者提供簡單之篩選核貸原則，減少銀行本身因債權無法回收導致銀行本身的資金造成呆帳，進而間接造成整體之金融危機發生的可能性，擬進行相關之不良債權之研究，以提供銀行業者在進行核貸的參考依據及建議。 本研究採用羅吉斯模型為研究分析，以某一銀行北部某分行及南部某分行，民國84年至90年貸放之住宅貸款為抽樣母體，隨機抽取正常戶200件、催收戶110戶，共310件作為樣本。研究結果發現，影響住宅貸款發生不良債權的影響變數為是否配偶連帶保証、是否專案貸款、借款金額、借款利率、有無車位及是否越區承作有顯著影響，正常繳款戶可達86.5%及預測逾期繳款戶的正確率可達77.3%。[[abstract]] The purpose of this study is to reduce non-performing loans which may be unrecoverable and to provide simple rules for credit sieving and approving for the purpose of reducing bad debts resulted from unrecoverable loans that damage the capital of banks, which may indirectly lead to the possibility of financial crisis for the entire financial system. Therefore, this study plans to probe relevant issues concerning non-performing loans for banks’ reference when they examine and approve their credit applications. This study adopts Logistic Regression model for analysis. One branch in the north and one branch in the south of a selected bank were chosen as the subjects of this study. The population is the hosing loans made in the two branches during the period between 1995 and 2001. Random sampling is adopted for 200 samples from normal customers and 110 samples from overdue customers, which totals 310 samples. The results find that the variables which lead to non-performing loans are whether jointly and severally liability hold by spouse, whether a project loan approved, borrowing amount, borrowing interest rate, whether a parking space included, and whether a loan approved beyond the region, all of which showed significant influences. Accounts with normal repayments reaches 86.5% and the accuracy of prediction for overdue repayments reaches 77.3%