企業信用風險預警模型之探討

當公司發生財務危機前，應有跡象顯示其經營不善，若可建立一套有系統且客觀的財務預警模型，預測未來發生信用貸款違約的可能性，以強化銀行徵授信系統分析能力，避免授信於可能發生財務危機之公司而導致呆帳，亦可以保障投資人、股東與銀行等之利益。臺灣自2013年開始採用新的國際IFRSs會計準則，故本研究以 2013 年至 2021 年間台灣發生危機的上市櫃公司為研究對象，選取非金融產業的 71 家危機公司，再選入同會計年度、規模相近且同產業的 142 家正常公司，進行兩者比例為「1：2」的配對。將銀行借款總額/資產總額、淨值總額/負債總額、營業利益/營收淨額、稅前純益/資產總額、稅後純益/利息費用、(EBIT-利息收入)/銀行借款總額、營收淨額/應收帳款、存貨週轉率、速動比率作為財務比例，透過共線性檢定、逐步回歸分析篩選後建立 Logistic 模型。實證結果表示，Logistic模型可建立有效的財務預警模型，預測發生危機前一年準確率為83.6%，可作為金融機構於授信放款前，風險防範及決策之參考。

關鍵字

財務危機；財務比例；財務預警模型； Logistic 模型

並列摘要

Before the company has a financial crisis, there should be signs of poor management. If a systematic and objective financial early warning model can be established to predict the possibility of future credit loan defaults, it can strengthen the analysis ability of the bank's credit system and avoid credit granting. It can also protect the interests of investors, shareholders and banks in case of bad debts in companies that may experience financial crisis. Taiwan has adopted the new international IFRSs accounting standards since 2013. Therefore, this study takes listed OTC companies in crisis in Taiwan from 2013 to 2021 as the research object, selects 71 crisis companies in the non-financial industry, and then selects 142 normal companies in the same fiscal year, with similar scale and the same industry, and carries out a "1:2" matching principal between the two. Taking total bank borrowings / total assets, total net worth / total liabilities, operating profit / net revenue, net profit before tax / total assets, net profit after tax / interest expense, (EBIT - interest income) / total bank borrowings, net revenue / accounts receivable, inventory turnover rate and quick ratio as financial ratios, and logistic models are established after collinearity test and stepwise regression analysis. The empirical results show that the logistic model can establish an effective financial early warning model, with an accuracy rate of 83.6% one year before the crisis, which can be used as a reference for financial institutions to prevent risks and make decisions before credit granting.