企业财务危机预警模型——基于商业银行信贷决策的分析
|
摘要
国外商业银行信用风险度量、管理日趋成熟，国内商业银行在此
方面由于这样或那样的原因处于落后状态，国内在此领域的研究成果
也十分匮乏，面对新巴塞尔协议草案的逐渐完善和国内游戏规则与世
界的接轨，我们的银行业将面临巨大的挑战。
对于商业银行而言，企业贷款总是被列为最重要的贷款之一。银
行发放企业贷款时，几乎容不得出现闪失。多数企业贷款数额巨大，
如果一旦成为坏帐，银行自身要冒很大的风险。另外，银行纷纷围绕
信誉良好的企业展开竞争，因此贷款收入减去银行的资金、劳力、税
赋和一般管理费用等发放贷款必须的成本，所得的差额越来越小，这
样的结果就是使得利润/风险比率较小，因而只要几笔贷款出了问题，
就会严重影响到银行总的经营利润。多数信贷官员十分重视企业的财
务报表分析，研究企业的各种财务指标，希望从中挖掘出有利于其决
策的信息。但是,庞杂的财务指标体系导致获得的信息错综复杂，严
重干扰信贷官员的分析与决策。如何从众多的指标体系中筛选出能够
准确反映企业财务状况的指标，并建立数量化的模型便是本文的中心
所在。
财务危机（Financial Crisis）又称为财务困境（Financial
Distress），最为严重的财务危机就是破产（Bankruptcy），许多对
财务危机的研究就是从企业破产着手的。由于利益的不同，商业银行
从债权人的角度看待企业财务危机的观点，有其特殊性。它既不同于
一般的权益投资者，也不同于内部经营管理者。从单纯财务的角度来
看，商业银行对信贷企业所看重的，首先是盈利的可能性，再次就是
债务的偿还能力。商业银行在处理企业贷款时主要面对的是企业的信
用风险，实际上就是违约风险。文章参照了新巴塞尔协议草案，同时
考虑到我国的特殊情况：《企业破产法》 虽然其早在1986年颁布，
，
1988年11月1日开始试行，然而迄今为止，还没有一家上市企业破产
的案例，最后认为商业银行的财务危机观应界定为企业发生很大违约
１
可能性：1）连续两年亏损或者第一年亏损使股东权益低于股本；2）
企业上市两年内发生亏损的排除在外；3）企业上市第三年发生亏损
达到股东所有者权益小于股本也排除在外。排除理由是这两类上市企
业有包装嫌疑而且企业数据的时段较短，不具有分析性。
文章选取了反映上市企业偿债能力、盈利能力、资产管理能力、
经营发展能力的40个财务指标作为备选变量。基于1993年是会计改
革以及会计信息规范的开始，我们的研究对象选自于 1993－2001 年
沪深交易所的上市企业。样本数据取于香港理工大学中国会计与金融
研究中心和深圳市国泰安信息技术有限公司的数据光盘，财务指标计
算依据上市企业的年报。从1130家上市企业中初步选择了117家企
业作为模型的样本。作为配对标准的控制因素一般包括会计年度、行
业和资产规模，这些配对标准用来控制由于财务危机组和控制组之间
由于报告的季节性、行业特征和企业规模的差异所可能带来的偏差。
为了克服这些偏差，文章选择与财务危机组117家企业同行业、相近
资产规模和同年度的117家非财务危机企业作为配对样本。
文章在进行实证研究时，主要使用了描述性分析、单变量分析、
多元判别分析，多元逻辑斯蒂回归四种统计方法。这些统计方法在国
内外进行财务危机预警分析中采用的频率较高，而且实践证明了这些
方法预测所得的判别准确率较高。此外，为了克服多重共线性，文章
还有所创新，首先对原始数据进行主成分分析，然后选取若干主成分
作为自变量进行多元逻辑斯蒂回归分析。总的来说，三种多变量方法
效果好于单变量分析，在多变量分析中，通过比较误判率和判别的稳
定性，明显可以看出逻辑斯蒂回归要好于多元判别分析，基于主成分
的逻辑斯蒂回归好于基于原始数据的逻辑斯蒂回归。最后，文章选择
基于主成分的逻辑斯蒂回归模型作为商业银行信贷决策的财务危机
预警模型。
商业银行的信贷官员只要把申请借款企业的财务指标代入模型
中，如果得到的概率大于 0.4680，则认为该企业是财务危机企业，
反之，如果得到的概率值小于 0.4680 则可以认为是健康企业；如果
得到的值较大，信贷官员可以要求申请借款企业增加抵押品；如果得
２
到的概率处于 0.4680 附近，这时应该谨慎决策或者选取其它的评价
手段，比如专家评价法。另外，模型还可以用于对已经得到贷款的企
业跟踪审查，进行过程监控，若发现客户财务状况恶化，银行可以及
时采取补救措施，提高银行信贷资产的安全性。此外，模型虽然不是
针对单一贷款或组合贷款的评级模型，但可以作为贷款评级的参考依
据。
文章的创新在于①应用上的创新。从商业银行信贷管理的角度出
发，提出商业银行的财务危机观。一般学者建立财务危机模型主要是
从投资者和监管当局的立场出发，选择样本，建立模型。本文从商业
银行角度出发建立针对性强的企业客户信贷分析预警模型。②方法上
的创新，使用基于主成分的逻辑斯蒂回归建立财务危机的预警模型。
文章比较了目前使用频率较高的三种多变量分析
In the measure and management of credit risk, overseas commercial
banks have been gradually maturing, however domestic commercial
banks fall far behind for this or that reason. In this field the achievement
of domestic researches is deficient. In face of gradual perfection of the
new Basel Agreement and the entry of WTO, our banking will be
confronted with enormous challenge.
Many commercial banks attach the greatest importance to enterprise
loan. They do not allow the little accident to happen. The amount of most
enterprise loan is very large. If one of them becomes unregainable, the
bank will shoulder agreat risk. Moreover, commercial banks compete
with each other for the companies with good credit. So the ultimate profit
is thin. The profit/risk value is smaller, which affects the total profit
seriously. Most credit officials put stress on the analysis on financial
reports. They hope to dig out the information that can help them in
decision-making. However, the cumbersome financial index system
causes the complexity of information obtained, which seriously interferes
with the credit official’s analysis and decision-making. How to pick up
the sensible and accurate financial indexes that can reflect the enterprise’s
financial situation from numerous indexes and establish the positive
model are the main aims of this paper.
Financial Crisis is called the Financial Distres. The most serious
financial crisis is financial bankruptcy. Many researches on financial
crisis start from the enterprise bankruptcy. Because the interest of
commercial banks is different from the outside investors and inside
governor, commercial banks think a lot of the possibility of payoff and
the liquidation ability. When handling enterprise loan, commercial banks
１
face primarily credit risk of the enterprise, which in reality is default risk.
The paper refers to the new Basel agreement draft, at the same time
considers the special circumstances of our country and finally assumes
that “commercial bank’s financial crisis view” should be defined as the
big default possibility of the enterprise.
The paper selects 40 financial indexes that reflect the capability to
repay, to make profit, to manage property and to develope. The paper
chooses 117 listed companies as the samples of positive model. For
overcoming some deviation, the paper chooses 117 healthy listed
companies which have the same industry, close property scale with the
117 bad companies.
The paper mainly exploits profile analysis, single variable analysis,
multi-variable analysis and logistic regression model. These methods are
frequently adopted in domestic and abroad study of financial crisis. The
accuracy of these methods can predict the exact result. To overcome
common linearity, the paper makes an innovation. The paper offers a
main component analysis and makes a logistic regression based on
them .Finally, the paper chooses logistic repression model based on main
component as early warning model of commercial bank’s credit risk.
Credit officials of the commercial banks input financial index into models
and if the probability received is greater than 0.4680 , we may draw a
conclusion that the enterprise is enterprises with financial crisis, on the
contrary, if the probability value received is smaller than 0.4680 , we can
regard it as healthy enterprises; If the value is relatively large, credit
official can command that the enterprise should increase the pledge; if
probability received is in the neighborhood of 0.4680, credit official
should make a prudent decision and choose other appraisal means at this
moment, such as expert's evaluation. In addition, the model can also be
used to supervise enterprises that have already got the loan. If commercial
banks find that customer's financial situation worsens, the banks can take
２
the remedy in time.
The innovation of the paper lies in ①Application innovation. From
方面由于这样或那样的原因处于落后状态，国内在此领域的研究成果
也十分匮乏，面对新巴塞尔协议草案的逐渐完善和国内游戏规则与世
界的接轨，我们的银行业将面临巨大的挑战。
对于商业银行而言，企业贷款总是被列为最重要的贷款之一。银
行发放企业贷款时，几乎容不得出现闪失。多数企业贷款数额巨大，
如果一旦成为坏帐，银行自身要冒很大的风险。另外，银行纷纷围绕
信誉良好的企业展开竞争，因此贷款收入减去银行的资金、劳力、税
赋和一般管理费用等发放贷款必须的成本，所得的差额越来越小，这
样的结果就是使得利润/风险比率较小，因而只要几笔贷款出了问题，
就会严重影响到银行总的经营利润。多数信贷官员十分重视企业的财
务报表分析，研究企业的各种财务指标，希望从中挖掘出有利于其决
策的信息。但是,庞杂的财务指标体系导致获得的信息错综复杂，严
重干扰信贷官员的分析与决策。如何从众多的指标体系中筛选出能够
准确反映企业财务状况的指标，并建立数量化的模型便是本文的中心
所在。
财务危机（Financial Crisis）又称为财务困境（Financial
Distress），最为严重的财务危机就是破产（Bankruptcy），许多对
财务危机的研究就是从企业破产着手的。由于利益的不同，商业银行
从债权人的角度看待企业财务危机的观点，有其特殊性。它既不同于
一般的权益投资者，也不同于内部经营管理者。从单纯财务的角度来
看，商业银行对信贷企业所看重的，首先是盈利的可能性，再次就是
债务的偿还能力。商业银行在处理企业贷款时主要面对的是企业的信
用风险，实际上就是违约风险。文章参照了新巴塞尔协议草案，同时
考虑到我国的特殊情况：《企业破产法》 虽然其早在1986年颁布，
，
1988年11月1日开始试行，然而迄今为止，还没有一家上市企业破产
的案例，最后认为商业银行的财务危机观应界定为企业发生很大违约
１
可能性：1）连续两年亏损或者第一年亏损使股东权益低于股本；2）
企业上市两年内发生亏损的排除在外；3）企业上市第三年发生亏损
达到股东所有者权益小于股本也排除在外。排除理由是这两类上市企
业有包装嫌疑而且企业数据的时段较短，不具有分析性。
文章选取了反映上市企业偿债能力、盈利能力、资产管理能力、
经营发展能力的40个财务指标作为备选变量。基于1993年是会计改
革以及会计信息规范的开始，我们的研究对象选自于 1993－2001 年
沪深交易所的上市企业。样本数据取于香港理工大学中国会计与金融
研究中心和深圳市国泰安信息技术有限公司的数据光盘，财务指标计
算依据上市企业的年报。从1130家上市企业中初步选择了117家企
业作为模型的样本。作为配对标准的控制因素一般包括会计年度、行
业和资产规模，这些配对标准用来控制由于财务危机组和控制组之间
由于报告的季节性、行业特征和企业规模的差异所可能带来的偏差。
为了克服这些偏差，文章选择与财务危机组117家企业同行业、相近
资产规模和同年度的117家非财务危机企业作为配对样本。
文章在进行实证研究时，主要使用了描述性分析、单变量分析、
多元判别分析，多元逻辑斯蒂回归四种统计方法。这些统计方法在国
内外进行财务危机预警分析中采用的频率较高，而且实践证明了这些
方法预测所得的判别准确率较高。此外，为了克服多重共线性，文章
还有所创新，首先对原始数据进行主成分分析，然后选取若干主成分
作为自变量进行多元逻辑斯蒂回归分析。总的来说，三种多变量方法
效果好于单变量分析，在多变量分析中，通过比较误判率和判别的稳
定性，明显可以看出逻辑斯蒂回归要好于多元判别分析，基于主成分
的逻辑斯蒂回归好于基于原始数据的逻辑斯蒂回归。最后，文章选择
基于主成分的逻辑斯蒂回归模型作为商业银行信贷决策的财务危机
预警模型。
商业银行的信贷官员只要把申请借款企业的财务指标代入模型
中，如果得到的概率大于 0.4680，则认为该企业是财务危机企业，
反之，如果得到的概率值小于 0.4680 则可以认为是健康企业；如果
得到的值较大，信贷官员可以要求申请借款企业增加抵押品；如果得
２
到的概率处于 0.4680 附近，这时应该谨慎决策或者选取其它的评价
手段，比如专家评价法。另外，模型还可以用于对已经得到贷款的企
业跟踪审查，进行过程监控，若发现客户财务状况恶化，银行可以及
时采取补救措施，提高银行信贷资产的安全性。此外，模型虽然不是
针对单一贷款或组合贷款的评级模型，但可以作为贷款评级的参考依
据。
文章的创新在于①应用上的创新。从商业银行信贷管理的角度出
发，提出商业银行的财务危机观。一般学者建立财务危机模型主要是
从投资者和监管当局的立场出发，选择样本，建立模型。本文从商业
银行角度出发建立针对性强的企业客户信贷分析预警模型。②方法上
的创新，使用基于主成分的逻辑斯蒂回归建立财务危机的预警模型。
文章比较了目前使用频率较高的三种多变量分析
In the measure and management of credit risk, overseas commercial
banks have been gradually maturing, however domestic commercial
banks fall far behind for this or that reason. In this field the achievement
of domestic researches is deficient. In face of gradual perfection of the
new Basel Agreement and the entry of WTO, our banking will be
confronted with enormous challenge.
Many commercial banks attach the greatest importance to enterprise
loan. They do not allow the little accident to happen. The amount of most
enterprise loan is very large. If one of them becomes unregainable, the
bank will shoulder agreat risk. Moreover, commercial banks compete
with each other for the companies with good credit. So the ultimate profit
is thin. The profit/risk value is smaller, which affects the total profit
seriously. Most credit officials put stress on the analysis on financial
reports. They hope to dig out the information that can help them in
decision-making. However, the cumbersome financial index system
causes the complexity of information obtained, which seriously interferes
with the credit official’s analysis and decision-making. How to pick up
the sensible and accurate financial indexes that can reflect the enterprise’s
financial situation from numerous indexes and establish the positive
model are the main aims of this paper.
Financial Crisis is called the Financial Distres. The most serious
financial crisis is financial bankruptcy. Many researches on financial
crisis start from the enterprise bankruptcy. Because the interest of
commercial banks is different from the outside investors and inside
governor, commercial banks think a lot of the possibility of payoff and
the liquidation ability. When handling enterprise loan, commercial banks
１
face primarily credit risk of the enterprise, which in reality is default risk.
The paper refers to the new Basel agreement draft, at the same time
considers the special circumstances of our country and finally assumes
that “commercial bank’s financial crisis view” should be defined as the
big default possibility of the enterprise.
The paper selects 40 financial indexes that reflect the capability to
repay, to make profit, to manage property and to develope. The paper
chooses 117 listed companies as the samples of positive model. For
overcoming some deviation, the paper chooses 117 healthy listed
companies which have the same industry, close property scale with the
117 bad companies.
The paper mainly exploits profile analysis, single variable analysis,
multi-variable analysis and logistic regression model. These methods are
frequently adopted in domestic and abroad study of financial crisis. The
accuracy of these methods can predict the exact result. To overcome
common linearity, the paper makes an innovation. The paper offers a
main component analysis and makes a logistic regression based on
them .Finally, the paper chooses logistic repression model based on main
component as early warning model of commercial bank’s credit risk.
Credit officials of the commercial banks input financial index into models
and if the probability received is greater than 0.4680 , we may draw a
conclusion that the enterprise is enterprises with financial crisis, on the
contrary, if the probability value received is smaller than 0.4680 , we can
regard it as healthy enterprises; If the value is relatively large, credit
official can command that the enterprise should increase the pledge; if
probability received is in the neighborhood of 0.4680, credit official
should make a prudent decision and choose other appraisal means at this
moment, such as expert's evaluation. In addition, the model can also be
used to supervise enterprises that have already got the loan. If commercial
banks find that customer's financial situation worsens, the banks can take
２
the remedy in time.
The innovation of the paper lies in ①Application innovation. From
引文
[1]Beaver W H：Market Prices，Financial Ratio and the Prediction of
Failure .Journal of Accounting Research，1968。
[2]Carmichael D R：The Auditor’s Reporting obligation. Auditing
Research Monograph No.1。
[3]George Foster：Financial Statement Analysis. Prentice-House New
Jersey,1986。
[4]Edward I Altman：Financial Ratios Discriminant Analysis and the
Prediction of Corporate Bankruptcy ,Journal of Finance 1968 ,23。
[5]Richard Morris: Early Warning Indicators of Corporate Failure –A
Critical Review of Previous Research and Further Empirical Evidence.
Ashgate Publishing Ltd, 1997。
[6]Ross J E: Corporate Management in Crisis. Prentice-Hall: Englewood
Cliffs, N.J。
[7]Gordon：Towards A Theory of Financial Distress. The Journal of
Finance,1971。
[8]Wruck：Financial Distress ，Reorganization and Organization Efficiency.
Journal of Financial Economics,1990。
[9]Kose： Managing Financial Distress and Valuing Distress Securities:
A Survey and A Research Agenda. Financial Management,1993。
[10]Brealey, Myers and Marcus：Fundamentals of Corporate Finance.
McGraw-Hill Inc,1999。
[11]Fitzpatrick P J：A Comparison of Ratios of Successful Industrial
Enterprises with Those of Failed Films. Certified Public Accountant,
1932。
[12]Beaver W.H ：Market Prices, Financial Ratio and the Prediction of
Failure. Journal of Accounting Research,1968。
[13]Mark E.Zmijewski： Predicting Corporate Bankruptcy: An Empirical
Comparison of the Extant Financial Distress Models .Working Papers. State
６２
University of New York,1983。
[14]Edward I.Altman：Financial Ratio, Discriminant Analysis and
Prediction of Corporate Bankruptcy. Journal of Finance,1968。
[15]Edward I.Altman. Corporate Financial Distress: A Complete Guide to
Predicting, Avoiding and Dealing with Bankruptcy .Journal of
Finance,1983。
[16] Peter S.Rose: Commercial Bank Management,4th Edition, McGraw-Hill
Companies ,Inc，1999。
[17]Marc J.LeClere: The Occurrence and Timing of Events: Survival
Analysis Applied to the Study of Financial Distress. Journal of Accounting
Literature, 2000.Vol.19。
[18]陈静：上市公司财务恶化预测的实证分析[J].会计研究，1999（4）。
[19]陈晓：中国上市公司财务困境[J]。中国会计与财务研究，2000（9）。
[20]吴世农，卢贤义：我国上市公司财务困境的预测模型研究[J]。经济研究，
2001（6）
[21]李华中：上市公司经营失败的预警系统研究[J]。会计研究，2001（10）。
[22]章彰：商业银行信用风险管理[M]，第 68－69 页，中国人民大学出版社，2002
年。
[23]周首华、杨济华、王平：论财务危机的预警模型－F 分数模型[J]。会计研
究，1996年（8）。
[24]吴世农、卢贤义：我国上市公司财务困境的预测模型研究[J]，经济研究，
2001（6）。
[25]李小燕等：企业财务危机预警分析模型的比较研究[J]。金融科学，2001(2)。
[26]张文彤：SPSS 11 统计分析教程高级篇[S]。北京希望电子出版社，2002 年。
[27]卢纹岱：SPSS for Windows 统计分析（第 2 版）。电子工业出版社，2002
年。
Failure .Journal of Accounting Research，1968。
[2]Carmichael D R：The Auditor’s Reporting obligation. Auditing
Research Monograph No.1。
[3]George Foster：Financial Statement Analysis. Prentice-House New
Jersey,1986。
[4]Edward I Altman：Financial Ratios Discriminant Analysis and the
Prediction of Corporate Bankruptcy ,Journal of Finance 1968 ,23。
[5]Richard Morris: Early Warning Indicators of Corporate Failure –A
Critical Review of Previous Research and Further Empirical Evidence.
Ashgate Publishing Ltd, 1997。
[6]Ross J E: Corporate Management in Crisis. Prentice-Hall: Englewood
Cliffs, N.J。
[7]Gordon：Towards A Theory of Financial Distress. The Journal of
Finance,1971。
[8]Wruck：Financial Distress ，Reorganization and Organization Efficiency.
Journal of Financial Economics,1990。
[9]Kose： Managing Financial Distress and Valuing Distress Securities:
A Survey and A Research Agenda. Financial Management,1993。
[10]Brealey, Myers and Marcus：Fundamentals of Corporate Finance.
McGraw-Hill Inc,1999。
[11]Fitzpatrick P J：A Comparison of Ratios of Successful Industrial
Enterprises with Those of Failed Films. Certified Public Accountant,
1932。
[12]Beaver W.H ：Market Prices, Financial Ratio and the Prediction of
Failure. Journal of Accounting Research,1968。
[13]Mark E.Zmijewski： Predicting Corporate Bankruptcy: An Empirical
Comparison of the Extant Financial Distress Models .Working Papers. State
６２
University of New York,1983。
[14]Edward I.Altman：Financial Ratio, Discriminant Analysis and
Prediction of Corporate Bankruptcy. Journal of Finance,1968。
[15]Edward I.Altman. Corporate Financial Distress: A Complete Guide to
Predicting, Avoiding and Dealing with Bankruptcy .Journal of
Finance,1983。
[16] Peter S.Rose: Commercial Bank Management,4th Edition, McGraw-Hill
Companies ,Inc，1999。
[17]Marc J.LeClere: The Occurrence and Timing of Events: Survival
Analysis Applied to the Study of Financial Distress. Journal of Accounting
Literature, 2000.Vol.19。
[18]陈静：上市公司财务恶化预测的实证分析[J].会计研究，1999（4）。
[19]陈晓：中国上市公司财务困境[J]。中国会计与财务研究，2000（9）。
[20]吴世农，卢贤义：我国上市公司财务困境的预测模型研究[J]。经济研究，
2001（6）
[21]李华中：上市公司经营失败的预警系统研究[J]。会计研究，2001（10）。
[22]章彰：商业银行信用风险管理[M]，第 68－69 页，中国人民大学出版社，2002
年。
[23]周首华、杨济华、王平：论财务危机的预警模型－F 分数模型[J]。会计研
究，1996年（8）。
[24]吴世农、卢贤义：我国上市公司财务困境的预测模型研究[J]，经济研究，
2001（6）。
[25]李小燕等：企业财务危机预警分析模型的比较研究[J]。金融科学，2001(2)。
[26]张文彤：SPSS 11 统计分析教程高级篇[S]。北京希望电子出版社，2002 年。
[27]卢纹岱：SPSS for Windows 统计分析（第 2 版）。电子工业出版社，2002
年。