我国上市公司财务危机预测实证研究

发布时间：2018-01-27 05:56

  本文关键词： 财务危机 描述性统计 逻辑回归 预测模型 实证研究　出处：《西南财经大学》2012年硕士论文　论文类型：学位论文

【摘要】：一、研究背景 为了健全社会主义市场经济体制,完善资本市场体系,拓宽公司融资渠道,我国于1990年末相继成立了上海证券交易所和深圳证券交易所。来自中国证监会网站的数据显示,截止2011年底,境内上市公司数(A、B股)共计2342家,股票市价总值214758.1亿元,为我国的经济发展做出了重要贡献。 同时,为了促使社会资源的优化配置,每个市场主体都必须接受市场竞争“优胜劣汰”的法则。自从沪深两市投入运营以来,为了规范上市公司行为,保障投资者权益,监管部门和交易所制定了一系列的法律法规和规章制度,对财务状况已经出现异常的上市公司股票实行通报警示。这些由于各种原因经营不善的上市公司,面临退市风险,给利益相关人带来损失。 因此,对上市公司的财务健康状况进行有效的评估,科学地预测上市公司面临财务危机的可能性,具有现实意义。财务危机预测模型为不同的利益相关人提供相应的决策依据：有利于公司的管理者查缺补漏,防止公司陷入财务危机;有利于公司的债权人对公司的信用进行有效评估,防范信贷风险；有利于投资者做出理性投资决策,规避投资风险。 二、本文结构及观点 全文分为五章,内容及观点如下： 第一章是绪论。阐述了研究的背景和意义,回顾了国内外对财务危机定义的探讨,确定了本文实证研究的财务危机上市公司为我国各年首次被ST或*ST的上市公司,并对国内外的财务危机预测研究文献进行综述。 第二章介绍了财务危机预测研究的理论与方法。对公司财务状况的影响因素进行归纳,梳理了财务危机预测规范研究的理论之后,着重介绍了财务危机预测实证研究的方法,并作出比较评价。 通过对财务危机预测实证研究方法的归纳总结来看,自从对公司财务危机的预测引入数理分析方法以来,各类建模方法层出不穷,主要可分作统计方法和非统计方法。就当前而言,经过众多研究证明和实践检验,且被广泛接受的统计方法有多元判别分析和逻辑回归分析,非统计方法是人工神经网络模型。 第三章是我国上市公司财务危机描述性统计分析。运用非财务数据进行描述性统计分析,定性分析宏观经济、行业差异以及公司治理结构等因素对上市公司财务危机的影响情况。 本章搜集自1998年实行上市公司异常状况警示制度以来,各年首次被实行ST或*ST的上市公司,定义为“财务危机”公司,搜集这些公司被警示前两年的数据进行描述性统计分析,定性分析宏观经济、行业差异以及公司治理结构等因素对上市公司财务危机的影响情况。为建立预测模型,选择作为对照的正常公司样本时,是否考虑年份、行业、规模的对应提供依据,同时为预测模型中非财务指标的选取提供参考。 第四章是本文的逻辑回归预测模型的实证研究部分,介绍了实证研究的数据采集来源、样本选取方法,指标的检验及筛选和模型的构建及检验。由于2004年沪深两市修订上市规则以后,“最近两年连续亏损”的将直接被*ST处理。本文选取了2005到2011年七年间233家沪深交易所实行*ST的上市公司作为财务危机的样本公司,同时采用同行业同期间随机选取的原则,取得相同数量的非ST或*ST的公司为对照的正常公司。然后把2005到2009五年的公司作为训练组进行逻辑回归建模,而2010年和2011年的作为应用组进行独立样本测试。 在预测指标方面,本文选取了公司偿债能力、盈利能力、营运能力、现金流量能力、发展能力、风险水平和公司治理七个方面的指标体系,共36个指标作为初选预测变量。先检验指标的正态性,然后利用非参数检验对指标进行差异显著性的检验,对通过差异性检验和需进一步讨论的指标,再进行相关性度量,看是否需要在建模时考虑变量间的相互作用。最后在逻辑回归时采用逐步回归法,从而完成指标的筛选。 在进行逻辑回归之后得到包含总资产净利润率、资产周转率(长期、短期和总资产)、现金流量比率和债务保障率六个指标的预测模型。并对模型进行了回代检验和独立样本验证。从回代检验来看,模型对训练组两类公司总的判别准确率为76.03%。由于这是对训练组样本t-2、t-3、t-4三年的总体数据的预测结果,可以说模型具有较好的判别能力。而独立样本的检验表明,模型在财务危机公司被通报警示前4年都具有判断能力,而前3年内则有良好的判别能力。 从模型的检验效果看来,模型在t-2年和t-3年的识别能力分别为85.7%、83.6%,t-4年为69.4%。从t-4年开始模型对财务危机公司的识别能力明显降低,但仍有微弱预测能力,而对正常公司的识别能力并不随预测期的前推而降低。 第五章对本文的研究进行了总结,对后续研究进行展望,并就加强我国公司财务危机预测提出了若干建议。 从本文的研究可以得到的结论是：上市公司公开的财务报表信息能有效的反应公司财务健康状况,可以利用一定的财务指标建立针对上市公司整体的财务危机预测模型；总资产净利润率、资产周转率、现金流量比率和债务保障率几个方面对于公司陷入财务危机的可能性有明显的指示作用；逻辑回归预测模型从t-2到t-4年都具有预测能力。 同时从本研究得到如下启示：是否对上市公司实行通报警示主要看其盈利能力；公司出现财务危机的可能性受公司盈利能力、营运能力、现金流等方面的综合影响,要保持公司健康的财务状况,’必须注意公司各方面的运行情况。 三、本文的主要创新 本文在前人研究的基础之上,对财务危机预测的实证研究进行了深入的探讨,建立了相应的财务危机预测模型,在研究中本文在以下几方面进行了重要的修正和创新。 (1)对我国1998年至2011年各年财务危机公司,运用非财务数据进行描述性统计分析,定性分析宏观经济、行业差异以及公司治理结构等因素对上市公司财务危机的影响情况。为预测模型中,选择作为对照的正常公司样本时,是否考虑年份、行业、规模的对应提供依据。同时为预测模型中非财务指标的选取提供参考。 (2)在建立财务危机预测模型时,财务危机公司样本的选取,剔除了非财务原因被交易所实行通报警示的公司,主要考虑到审计意见否定或信息披露方式不合规定的公司,其财务信息的真实性值得商榷。 (3)在数据选取上,考虑到各年宏观经济环境的差异,与不同年份被首次通报警示的财务危机公司对照的正常公司,相关指标数据也选择相同时期的数据。 (4)无论财务危机公司,还是正常公司,都是只在沪深A股主板上市的公司,避免信息披露要求的差异导致可比性存在问题。 (5)在指标差异性检验上,考虑了T检验的数据正态性前提。先用K-S方法检验指标的正态性,然后再选择适当的差异显著性检验方法。 (6)将总体样本分作训练组和应用组,2005至2009年样本用来建立预测模型,2010至2011年样本用来测试预测模型的有用性,独立样本测试更具说服力。
[Abstract]:First, research background
In order to improve the socialist market economic system, improve the capital market system, broaden the financing channels for the company in China at the end of 1990 have been established in Shanghai stock exchange and Shenzhen stock exchange. China from the SFC website data show that as of the end of 2011, the number of domestic listed companies (A, b) a total of 2342, the stock market value of 21 trillion and 475 billion 810 million yuan that has made an important contribution to the economic development of our country.
At the same time, in order to promote the optimal allocation of social resources, each main market must accept the market competition of "survival of the fittest" principle. Since the Shanghai and Shenzhen two put into operation, in order to regulate the behavior of listed companies, protect the rights and interests of investors, regulators and exchanges has formulated a series of laws and regulations, the financial situation has been abnormal the shares of listed companies to implement notification alerts. Due to various reasons these companies face delisting risk, cause losses to the stakeholders.
Therefore, to effectively evaluate the financial health of listed companies, the possibility of scientific prediction of listed companies facing financial crisis, is of practical significance. The financial crisis forecasting model to provide the appropriate decision-making basis for various stakeholders: help the company managers check network, to prevent financial crises in favor of the company; the creditors to evaluate the company's credit, to prevent credit risks; help investors to make rational investment decisions, avoid investment risks.
Two, the structure and view of this article
The full text is divided into five chapters, and the contents and views are as follows:
The first chapter introduces the research background and significance, reviews the study on the definition of financial crisis at home and abroad, the empirical research of financial crisis of listed companies of our country every year for the first time by the ST or *ST of the listed companies, and the domestic and foreign financial crisis prediction research literature are reviewed.
The second chapter introduces the theory and method of financial crisis prediction, summarizes the factors that influence the company's financial status, and combs the theory of financial crisis prediction, then focuses on the methods of empirical research on financial crisis prediction, and makes a comparative evaluation.
Through summarizing the empirical research methods to predict the financial crisis of the induction, since the prediction of financial distress by mathematical analysis method, various kinds of modeling methods mainly emerge in an endless stream, can be divided into statistical and non statistical methods. At present, many research proof and practical test, statistical methods and widely accepted with discriminant analysis logistic regression analysis and multivariate, non statistical method is the artificial neural network model.
The third chapter is descriptive statistics analysis of financial crisis in Chinese listed companies. Descriptive statistics analysis is conducted using non-financial data, and qualitative analysis is made on the impact of macroeconomic factors, industry differences and corporate governance structure on the financial crisis of listed companies.
Since this chapter collected since 1998 the implementation of the listed company abnormal warning system, the first implementation of ST or *ST of the listed companies, defined as "financial crisis", these companies have been collected two years ago warning data descriptive statistical analysis, qualitative analysis of the macroeconomic situation, industry differences and influencing factors of corporate governance structure the financial crisis of the listed companies. In order to establish the prediction model, selected as the normal control samples of the company, whether to consider the year industry, provide the basis for the corresponding scale, and to provide reference for the selection of non-financial index prediction model.
The fourth chapter is the empirical research part logistic regression prediction model, this paper introduces a data collection source of empirical research, sample selection, construction and inspection index inspection and screening and model. After the 2004 amendments to the Shanghai and Shenzhen two city listing rules, the last two years of continuous losses will be directly handled by *ST. This paper selects 2005 to seven years in 2011 233 in Shanghai and Shenzhen Stock Exchange to implement *ST listed companies as the financial crisis of the Sample Firms, the industry in the same period were randomly selected to obtain the principle of the same number of non ST or *ST for normal control. Then the company from 2005 to 2009 in five years as the training group for logistic regression modeling in 2010 and 2011, and as the application group of independent sample test.
In the forecast indexes, this paper selected the company's solvency, profitability, operation ability and cash flow ability, development ability, risk level and corporate governance index system of seven aspects, a total of 36 indicators as the primary predictor variables. To test the normality index, and then test the significant difference of the index the use of non parametric test, through difference test and discuss the index, then the correlation metric, see the need to consider whether the interaction between the variables in the model. Finally, using stepwise regression method in logistic regression, and completed the selection of indicators.
Include net profit rate of total assets after logistic regression, asset turnover (long-term, short-term and total assets), prediction model of cash flow ratio and debt guarantee rate of six indicators. And the model of the back substitution test and independent sample test. From the regression test, discriminant model of total training group of two companies the accuracy of 76.03%. because it is on the training samples T-2, T-3, the prediction results T-4 three years of data, can be said that the model has good distinguishing ability. And the test of independent samples showed that the model was informed in the financial crisis warning 4 years ago has the ability to judge, and the first 3 years there was good discrimination capabilities.
From the point of view of the model test results, model in T-2 and T-3 recognition ability were 85.7%, 83.6%, T-4 69.4%. T-4 years from the beginning of the model of the recognition of the companies in financial crisis decreased significantly, but there is still a weak predictive power, while the normal company recognition ability is not with the forecast period before push reduced.
The fifth chapter summarizes the research of this paper, looks forward to the follow-up research, and puts forward some suggestions on strengthening the financial crisis prediction in our country.
Can be obtained from the conclusions of this study are: public financial report of the listed company information can be healthy and effective corporate financial situation reaction can use financial indicators, the establishment of the listed company's financial crisis prediction model; net profit rate of total assets, asset turnover ratio, cash flow ratio and debt guarantee rate aspects significant role for the possibility of financial crisis; logistic regression prediction model has predictive power from T-2 to T-4.
At the same time get the inspiration from the study: whether to implement the notification warning mainly depends on the profitability of listed companies; profitability, the possibility of company financial crisis by the company operating capacity, cash flow and other aspects of the comprehensive effect, to maintain a healthy financial condition of the company, "must pay attention to the operation of each aspect of the company.
Three, the main innovation of this article
Based on previous studies, this paper has made an in-depth discussion on the empirical research of financial crisis prediction, and established a corresponding financial crisis prediction model. In this study, the following important amendments and innovations were carried out in the following aspects.
(1) in China from 1998 to 2011 each year, the financial crisis, using financial data descriptive statistical analysis, qualitative analysis of the macroeconomic situation, industry differences and corporate governance structure and other factors on the financial crisis of the listed company. As the prediction model, selected as a normal company control sample, consider whether the year the industry, provide the basis for the corresponding scale. At the same time provide a reference for the selection of non-financial index prediction model.
(2) in the model of financial crisis, the financial crisis companies selected sample, excluding the non financial reasons by the exchange to implement reporting warning of the company, taking into account the main negative audit opinion and information disclosure irregularities, the authenticity of the questionable financial information.
(3) in data selection, considering the difference of macro economic environment in different years, the relevant index data of normal financial companies with different warning dates were also selected in the same period.
(4) whether financial crisis companies or normal companies are listed on the main board of Shanghai and Shenzhen A shares, the difference between them will lead to comparable problems.
(5) in terms of index difference test, we consider the normality of data in T test. First we use K-S to test normality of index, then choose the appropriate difference significance test.
(6) the total samples were divided into training group and application group. From 2005 to 2009, samples were used to establish prediction models. From 2011 to 2011, samples were used to test the usefulness of prediction models, and the independent sample test was more convincing.

【学位授予单位】：西南财经大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F275;F832.51;F224

【参考文献】

中国期刊全文数据库 前10条

1 张鸣,张艳;财务困境预测的实证研究与评述[J];财经研究;2001年12期

2 杨华;;财务危机预警模型中的非财务指标应用研究[J];财会通讯(综合版);2007年05期

3 杨轶;张友棠;;企业财务预警国内外研究述评[J];财会月刊;2008年20期

4 邓晓岚,王宗军,李红侠,熊银平;企业经营失败评价方法研究进展述评[J];系统工程;2005年08期

5 宋鹏;张信东;;基于Logistic模型的上市公司财务危机预警研究[J];经济问题;2009年08期

6 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年06期

7 郭兆;;上市公司财务危机预警Logistic模型实证分析[J];金融经济;2009年12期

8 陈静;上市公司财务恶化预测的实证分析[J];会计研究;1999年04期

9 李玲;谭滔;;企业财务危机预警模型构建的理论解析[J];科技与管理;2008年01期

10 宋秋萍;开展财务预警分析,增强经营者忧患意识[J];生产力研究;2000年Z1期

中国博士学位论文全文数据库 前2条

1 孙洁;企业财务危机预警的智能决策方法研究[D];哈尔滨工业大学;2007年

2 刘彦文;上市公司财务危机预警模型研究[D];大连理工大学;2009年

中国硕士学位论文全文数据库 前1条

1 卢贤义;我国上市公司财务困境预测的实证研究[D];厦门大学;2001年

，

本文编号：1467756