建構中小企業授信違約之信用風險因子模型

Abstract

銀行作為金融市場資金往來的中介角色，其經營的成敗良窳，不僅關乎其自身的存續發展，更影響全體存戶的權益與國家金融的穩定。若銀行未能審慎控管其授信放款品質，致生不良債權，無疑令銀行股東與存戶暴露於鉅大損失的風險之中。當銀行在審酌中小企業客群之授信業務時，往往面臨財務報表失真、股東資力薄弱、與經營者誠信與否等問題，如何運用客觀有效的衡量方法，作出最佳的授信決策，以避免後續逾期放款的發生，對銀行授信業務健全與否至為重要。 本研究以國內某商業銀行西元2010年至2011年間300筆中小企業授信案件為樣本，其中113筆為授信違約公司，另187筆為授信正常公司，除考量財務分析指標外，另納入公司治理與授信條件、經營者個人條件與信用共計三個構面、35項衡量指標，據以建立三種風險評估模型：分別為二元邏輯斯迴歸模型、決策樹演算法、及結合上述兩種方法之模型。由實證結果顯示，決策樹演算法具有較高的授信風險因子區別能力，其預測結果之正確率達到96%，並求得13項顯著的影響變數。藉由本研究之實證結果，希冀提供一個嶄新具有價值的觀點，供金融同業日後審酌中小企業授信業務時之參考。

Financial intermediaries such as banks play an important role to keep the flow of funds moving in financial systems, and therefore their performance not only affect their existence and development, but also affect the benefit of customers and the national financial stability. If banks fail to control the quality of credit loan carefully, it may result in non-performing loans. Without a doubt, it makes the bank shareholders and customers expose to great risk of loss. When banks review the credit loan of the small and medium-sized enterprises, they often face the problems of distorted financial statements, insufficient financial strength of shareholders, dishonesty business owners and so on. For the soundness of credit loans, it is important for banks to use effective and objective ways to make the best decisions in order to avoid the occurrence of overdue loans.

Three hundred scredit-loan cases of small and medium-sized enterprises from 2010 to 2011 were collected from one specific commercial bank in Taiwan. One hundred and thirteen companies were breach of loan cases, and one hundred and eighty-seven companies were not. We measured not only financial ratio indicators, indicators of corporate governance and credit conditions, but also indicators of the owners’ individual and credit conditions. Thirty-five indicators in three dimensions were included in the analysis of three risk assessment models: binary logistic regression model, decision tree, and two stage combined. The results showed that decision tree has higher differentiability among the credit risk factors, and its correct prediction rate is 96%. Thirteen influential indicators were identified in this study. The results of this study may offer a new perspective and valuable references for financial intermediaries’ credit risk management for loans of SMEs.