中小企業財務預測分析--以銀行移送信保之授信為例

The Analysis of Financial Prediction – with the Example of Crediting Transferred by Banks for Credit Guarantee

李國賓(Kuo-Pin Li)

指導教授 ： 蘇文斌

朝陽科技大學/管理學院/保險金融管理系/碩士(2009年)

摘要

摘 要
台灣的中小企業一直是國內的重要經濟命脈，中小企業亦是台灣經濟發展的基石，具有彈性調整靈活進出市場特性，不管從過去還是現在，從台灣經濟的發展貢獻來看，對促進經濟成長貢獻有目共睹，因而中小企業亦是奠立起經濟起飛的重大基礎。不過，中小企業財報不透明與資本過小的產業特性，讓中小企業的融資需求，有可能銀行因為考量授信風險，讓企業主的融資需求遇到瓶頸，而面臨融資的困境。本研究對象為台灣的中小企業，樣本的研究母體主要針對國內某商業銀行對國內民營中小企業移送信保的授信貸款資料，利用取得的企業財務狀況之台幣授信案為樣本資料，其中包括已違約戶和仍是正常戶，共計308戶。藉由申請貸款之前三個年度的財務報表資料，評估其是否可做為有承辦信保授信之銀行放款的依據。就採取的樣本中，以正常繳息企業與違約倒閉企業兩者於核貸前之近三年財務狀況分析是否具有顯著性。本研究針對中小企業在銀行貸款融資方面，已有中小企業信用保證基金保證之財務報表，利用近三年可取得之財務比率，做為判定企業是否於移送信保之授信後，未來的營運是否仍會面臨財務問題，亦即公司是否會遭受違約倒閉或仍可持續正常營運。故本研究的目的如下：
1.以銀行業者所提供之中小企業參加信保授信部份相關資料，探討實務上，財務分析比率是否對信保戶授信違約風險有顯著影響。
2.找出可作為中小企業信保授信決策品質的因素。
本研究利用羅吉斯廻歸模式(Logistic Regression Model)檢定實證結果發現前二年流動比率、前二年總資產週轉率、前三年總資產週轉率、前二年款貨資產比、授信比等5個變數對於企業的經營現況具有顯著的差異。本研究經由迴歸模式的預測分類結果，對日後是否成為倒閉案件的預測正確性為50.7%，對是否成為正常案件預測的正確性為97.9%，整體之正確性為87.7%。整體模式適配度（Goodness of Fit）檢定的χ2=123.209（p=0.000<0.05），達到顯著；而Hosmer-Lemeshow檢定值為6.297 (p=.614>.05)未達顯著，表示前二年流動比率、前二年總資產週轉率、前三年總資產週轉率、前二年款貨資產比、授信比等5個自變數所建立的中小企業信用保證授信模式適配度非常理想。因此，授信人員可利用此模式在最短期限內，依申請戶的條件預測該筆借款未來成為倒閉案件或正常案件的機率。
本研究可導出下列授信模式：
Z =4.551＋0.019×前二年流動比率－1.111×前二年總資產週轉率＋0.797×前三年總資產週轉率－0.051×前二年款貨資產比－0.039×授信比
銀行在進行中小企業放款時，勿因中小企業信用保證基金可分散風險，而失去了授信原有的嚴謹度。授信政策仍應兼顧風險及收益考量，因此，追求利潤的同時，若能依廻歸模式分析案件的未來性，相信可避免剔除優良案件，提高銀行收益，同時，可篩選出高風險案件，兼顧風險控管。

關鍵字

中小企業信用保證基金 ； 羅吉斯廻歸模式 ； 風險控管

並列摘要

Abstract
The small and medium-sized enterprises have been one of important economic lifelines and cornerstones of economic development in Taiwan. They have characteristics of flexible adjustment and access to the markets and their contributions to the economic growth are obvious to all from the point of view of the economic development in Taiwan in the past and now. Therefore, SME are the significant base for economic taking-off. However, The industry characteristics of the intransparency of financial statement and too small capital often cause a bottleneck for the demands of loans from the enterprise owners who are mired in the difficulties due to the consideration of credit risks by banks. The subjects of research were the enterprises in Taiwan. The parent body of samples was the credit loan application data by SME that a commercial bank transferred to the authorities for credit guarantees. The samples of NT dollar credit cases together with the financial situation of applicant enterprises that the researcher had obtained were total 308 accounts including default accounts and regular accounts. On the basis of the financial statement data for three years before the application, we assess whether the data could be the basis of loans for the bank that received the loan application. The research also analyzed the significance of the financial situation for recent three years before the ratification of loans between regular repaying interest accounts and default and collapsed accounts. Using the financial statement and the information about ratio of finance that were available in recent years in the loan applications by SMEs that had obtained credit guarantees from SME Credit Guarantee Fund, the research would determine whether the applicant enterprise would still face financial problems, that is, whether the company would default, collapse or sustainably and normally operate in future after ratification of loan that was credit guaranteed by the Fund. So the purposes of the research are as follows:
1.Based on the data provided by the bank about the SME that had applied credit loans and credit guarantee, it explored if there were significant effects of financial analysis ratio on the default risks of credit guarantee accounts.
2.Find out the factors that could be the quality of credit decision-policy for SME credit guarantees.
This research used Logistic Regression model to test the empirical results. It found that the five variables of the two-year current ratio, previous two-year total asset turnover ratio, previous three-year total asset turnover ratio, previous loan-asset ratio and credit ratio had significant differences in the current situation of operations. The forecast accuracy of the results of forecast classification through the regression model in the research on the closedowns of enterprises was 50.7% whereas the forecast accuracy on the normal case was 97.9%. The overall accuracy was 87.7% whereas the tested χ2=123.209（p=0.000<0.05）of overall goodness of fit of the model and reached the significance level. But the value of Hosmer-Lemeshow Test was 6.297 (p=.614>.05) and did not reach the significance level. That means the goodness of fit of the credit model of SME credit guarantee that was constructed on the bases of the five variables of the two-year current ratio, previous two-year total asset turnover ratio, previous three-year total asset turnover ratio, previous loan-asset ratio and credit ratio was ideal. Therefore, credit officers can use the model. Therefore, credit officers can use the model together with conditions of the applicant company to forecast in the shortest period the probability of collapse case or normal case in consideration of a loan.
This research can thus induct following credit model:
Z =4.551＋0.019× previous two-year current ratio－1.111× previous two-year total asset turnover ratio ＋0.797× previous three-year total asset turnover ratio－0.051× previous two-year loan-asset ratio－0.039× credit ratio
When a bank is considering loans to SMEs, it should not lose their strictness and carefulness that they should have and their credit policy should pay attention to both risks and profit. Therefore, at the time of seeking profit, if a bank can analyze the futurity of the case by regression model, we believe that they can avoid ruling out good cases and raise profit for the bank. Meanwhile, it can also screen out high risky cases and give consideration to risk control.

並列關鍵字

SME Credit Guarantee Fund ； risk control ； Logistic Regression model

參考文獻

35. 曾信超、黃新宗(2005)。中小企業融資授信實務與風險轉嫁之實務探討。華人前瞻研究，1(2)，43-64。

30. 黃愷婷、楊素柳（2003）。銀行授信風險管理架構之探討，真理財經學報，頁93-150。台北：真理大學財經學院。

16. 林左裕（2005）。利用未償債務之扣除進行逃漏遺產稅之研究---Logit模式之應用，交大管理學報，25(1)，205-229。台北：交大管理學報編輯室。

40. 劉美芳、李國賓(2007)。創業貸款與財務風險之探討。修平技術學院2007年卓越管理學術暨實務研討會，台中縣。

22. 陳逸文（2001）。銀行授信之風險管理。台灣金融財務季刊，1(1)，87-97。