應用存活分析建構台灣中小企業風險評估模型

Abstract

近年來國內中小企業受到經濟不景氣的影響，且隨著全球化競爭日漸激烈及產業外移，國內銀行等金融機構對於中小企業之放款態度漸趨保守，造成絕大多數的中小企業均面臨資金嚴重短缺的窘境。然而，中小企業的財務結構不若大企業健全，財務報表資訊可信度較低，無法真實反應公司實際經營狀況，此導致中小企業與金融機構之間存在嚴重的資訊不對稱（information asymmetry）問題，使得金融機構很難有效研判中小企業的償還能力，以致銀行或金融機構對於中小企業授信作業不易，放款政策不易訂出。目前國內、外文獻所提出的違約之風險評估多以上市櫃之大型企業為研究對象，對於中小企業違約風險評估之研究相對較少，因此現有的國內、外文獻所建議違約風險評估系統可能不適用於評估中小企業貸款的信用判定標準。因此，本研究針對以中小企業為主要借款客戶之金融機構，建構一套客觀且合理之風險評估模型，以預測中小企業是否會違約及預測違約的時點，以供銀行或金融機構之決策者作為彈性放款決策之參考依據。本研究所建構風險評估模型之過程，主要分為四階段：（1）選擇變數與蒐集整理資料；（2）利用主成份分析來消除變數間之共線性；（3）利用Cox模式來建構風險評估模型及建立中小企業在各個時點的存活機率表；（4）驗證存活機率表之有效性，以提供銀行或金融機構決策者作為彈性放款決策之參考依據。本研究蒐集了國內某金融機構所提供之中小企業客戶歷史資料進行模型建構，驗證本研究模型之可行性。

Recently, domestic small and medium enterprises (SMEs) are affected by the economic downturn, increasingly intense global competition, and industry relocation. Financial institutions such as domestic banks became conservative regarding to SME loans; as a result, most SMEs face the problem of fund shortage. However, SMEs’ financial structure is not as sound as large enterprises; the credibility of financial statements is relatively lower and cannot reflect the actual operational performance. As a result, the information asymmetry between SMEs and financial institutes is serious. It is hard for financial institutions to estimate SME’s repayment capacity and therefore difficult for them to design the loan policy for SMEs. Domestic and foreign studies on default risk assessment focus mostly on listed or over-the-counter companies which have a larger scale. There are not as much studies on the default risk of SMEs. As a result, the default risk assessment system in domestic and foreign studies may not apply to the credit criteria in SME loans. The study targeted financial institutes whose clients are mostly SMEs. We built an objective and rational risk assessment model to predict SME’s default possibility and time to default. Decision makers in financial institutes can use this as a reference for flexible loan decisions. There are four stages in constructing the risk assessment model: 1. To select variables and collect data; 2. To eliminate the colinearity among variables using the principal components analysis; 3. To build the risk assessment model and survival rate table for SMEs based on the Cox model; and 4. To verify the effectiveness of survival rate table, which is a reference for decision makers in financial institutions regarding flexible loans. The study collected historical SME data from a domestic financial institute to build and test the feasibility of the model.