財務危機預警模型―產業預測衡量

Abstract

企業發生財務危機時往往伴隨著巨大的破產成本，且外部效應導致位於供應鏈的每一家廠商都受到牽連。因此，過去數十年當中許多學者致力於企業財務危機預警的研究當中，期望找出固定的預警模式讓不論是金融機構、企業或一般投資大眾能提前預知到財務危機的到來，以免蒙受損失。變數的引用方面，從一開始的財務會計比率、市場資訊到近期關注在總體與非財務面，而模型數學方法建構也從一開始的敘述統計觀察、Z-Score與Logistic迴歸模型等到近年來利用電腦開發更複雜的演算方法，如資訊包絡法等，財務危機預警模型發展至今已經越趨成熟。以往的研究在樣本選取上多利用總產業的資料加以建構，探討預測變數對於整體企業財務危機的預測能力，較少將樣本資料分成不同產業面向討論；然而，不同產業各自擁有不同的產業特性，這樣的產業特性將會影響一家企業在財務比率上呈現不同的樣貌，進而對其財務危機發生的可能產生重大的影響。因此，本研究中採用受到廣泛應用的Logistic迴歸方法，並利用逐步迴歸法找出對於各產業危機發生有顯著影響的預測變數，分別建構總產業、紡織業、營建業與電子業的財務危機預警模型， 期望使模型有更高的預測能力。 實證結果顯示，三大產業與總產業建構的財務危機預警模型在總預測正確率上差異不大，但分產業建構之模型能有效的提高危機公司的預測正確率。紡織業中，短期內公司的償債能力、現金流量運用效率與營業費用相對於營業收入的比例將影響財務危機的發生機率，而有息負債利率(X4)在短期與長期都為影響財務危機發生與否的顯著指標；營建業方面，短期內應注意企業的獲利能力，而長期除了獲利能力外，企業的經營能力也左右著財務危機的發生；電子業中，影響財務危機發生的層面廣，財務結構、經營能力與獲利能力在危機發生的前一年、前二年與前三年都有重要的影響，然而在成長動能高與資本密集之下，獲利能力在為電子業中最關鍵的因素。

Company’s financial distress has been an important issue for decades because it is a huge cost for no matter corporations or nation’s economy. For solving this problem, previous scholars tried to find out a pattern to predict financial distress’s coming, for investors, to avoid the losses, for business owners, to avoid bankruptcy. In terms of prediction variables in financial distress prediction model, studies from accounting based variables, market based variables to macro and non-financial factors. As for methodology in modeling, researches form descriptive statistics, multiple discriminant analysis to Logistic model and more complicated algorithm by using computing. In this study, we will use the data from three different industries in the period of 2008-2014, textile, construction and electronics industry, to construct financial distress prediction model by Logistic regression. We hope that the models constructed by industry data set, textile, construction and electronics industry, would have higher power in predicting than which constructed by total samples. Our research result shows that, models constructed by the samples of textile, construction and electronics industry have higher predictive power in identifying distress companies than constructed by total samples. In textile industry, debt-paying ability, operation ability and profitability is the key factors in affecting the probability of being in distress in the short run. In the long run, interest expenses to debt ratio is the only significant variable. In construction industry, company’s profitability ability is the key in the short run. In the long run, operation ability is crucial in predicting financial distress as well. In electronics industry, financial structure, operation ability and profitability are the important factors over previous three years. However, due to the capital-intensive and high-growth features, it is necessary for companies in electronics industry to be well-performed on profitability to sustain it.