基于Panel Logit模型的上市公司财务困境预警研究

发布时间：2018-03-14 13:47

本文选题：熵权法　切入点：面板数据　出处：《太原科技大学》2014年硕士论文　论文类型：学位论文

【摘要】：上市公司陷入财务困境不仅威胁到公司自身的生存发展，而且为债权人和投资者带来巨大的损失。为了减少债权人和投资者的经济损失并且使上市公司能够健康发展，需要建立能反映上市公司各方面风险的财务困境预警模型，有效地预测企业经营状况，以防范企业陷入财务困境，这对于提高企业经营管理质量，保护相关利益者利益以及促进我国资本市场的良性发展具有重要意义。关于上市公司财务预警的研究，国内外学者都做了许多工作，但以往的研究大都基于不同方法以单期截面数据预测公司是否陷入财务困境，属于静态预测研究。企业的财务状况具有持续性和累积效应，公司陷入财务困境是有一个逐渐演变的过程，并且企业财务状况暂时偏离正常值不应被归为困境公司。为了尽可能全面客观的反应企业财务状况，，本文从财务和非财务两个维度进行预警指标选取，采用能有效处理面板数据的PanelLogit模型对我国上市公司的财务困境进行预警研究，弥补了传统截面数据研究不能动态反应企业财务状况演变的不足，较客观地反映公司财务状况发展的动态事实，使预警模型更具有实际意义。本文以沪深A股制造业上市公司作为研究对象，首先选取2011-2013年首次被特殊处理（ST）的30家上市公司作为财务困境的公司样本，按照1:2的比例选取相同时间段从未被ST过的60家制造业上市公司作为财务正常公司样本，并初步选取了28个财务指标和10个非财务指标作为预警指标；其次，利用熵权法对38个初选指标进行分析，根据每个初选指标所提供的影响财务困境预警模型的信息量的不同，最终选取21个财务指标和8个非财务指标；第三，对筛选的29个变量指标进行因子分析，消除变量指标多重共线性对预警模型估计的影响，提取出对企业财务状况具有重要影响的7个财务公共因子和3个非财务公共因子，并将7个财务公共因子作为仅基于财务指标的PanelLogit财务困境预警模型的解释变量，将7个财务公共因子和3个非财务公共因子作为财务指标与非财务指标相结合的Panel Logit财务困境预警模型的解释变量；第四，运用Hausman检验，确定采用随机效应的Panel Logit回归模型。仅基于财务指标建立的Panel Logit财务困境预警模型实证结果表明，偿债因子、盈利因子、资本运用因子、资本结构因子对企业陷入财务困境有重要影响，财务指标与非财务指标相结合建立的Panel Logit财务困境预警模型实证结果表明偿债因子、盈利因子、资本运用因子、资本结构因子、股权集中度因子、评价因子均是影响企业陷入财务困境的重要因素；最后，本文对已建立的财务困境预警模型做了样本外预测检验，预测结果表明正确率分别达到86.67%和91.11%，并将在财务困境预警模型中加入非财务指标和仅包含财务指标的财务困境预警模型的预测效果做了对比，结果表明，引入非财务指标有助于提高财务困境预警模型的预测能力。
[Abstract]:Listed companies in financial distress is not only a threat to the company's own survival and development, but also bring huge losses to creditors and investors. In order to reduce the economic losses of creditors and investors and listed companies to make a healthy development, need to establish the financial early-warning model can reflect all aspects of the risk of listed companies effectively predict the condition of business, in order to prevent the financial crisis of enterprises, to improve the quality of enterprise management, to protect the interests of stakeholders and promote the healthy development of the capital market of our country has important significance.
Research on financial early warning of listed companies, domestic and foreign scholars have done a lot of work, but most of the previous studies are based on different methods to single section data of Forecast Ltd into financial distress, belongs to the static prediction. The financial situation of enterprises is persistent and cumulative effect, the financial distress is a gradual evolution the financial situation of enterprises, and temporarily deviate from the normal should not be classified as distressed companies. In order to reflect the financial situation of enterprises comprehensively and objectively as possible, the early warning index is selected from the two dimensions of financial and non-financial, using the PanelLogit model can effectively deal with the panel data to study the early warning of financial distress in China's listed companies. To compensate for the lack of development of the financial situation of the traditional research on dynamic response of enterprises cannot cross section data, objectively reflect the dynamic development of the company's financial situation In fact, the early warning model is more practical.
In this paper, the Shanghai and Shenzhen A share listed companies in manufacturing industry as the research object, firstly selected for the first time in 2011-2013 years by the special treatment (ST) of the 30 listed companies as the financial distress of listed companies, 60 manufacturing according to the ratio of 1:2 to select the same time period has never been ST listed companies as the normal financial company and the preliminary sample. We selected 28 Financial Indicators and 10 non-financial indicators as early warning indicators; second, the 38 primary indexes were analyzed by using entropy method, according to the different amount of information in financial distress prediction effect of each primary index provided by the selected 21 Financial indicators and 8 non-financial indicators; third, to the factor analysis of 29 variables selection, eliminate variables multicollinearity of early-warning model estimation, extract has an important influence on the financial situation of the 7 factors and 3 Public Finance Non financial factors, and the 7 factors as the only public financial PanelLogit financial early-warning model of financial indicators of the explanatory variables based on the 7 financial factors and 3 non-financial public factors as the Panel Logit financial early-warning model of financial indicators and non-financial indicators combined with the explanatory variables; fourth, the use of Hausman test, determined using random effects Panel regression model Logit. Only Panel Logit financial early-warning model of financial indicators to establish the empirical results show that based on the solvency factor, profit factor, capital factor, capital structure factor has an important influence on the enterprise into financial distress, financial indicators and non-financial indicators combined with the Panel Logit financial distress the empirical results show that the model established solvency factor, profit factor, capital factor, capital structure, ownership concentration factor, evaluation Factors are important factors for financial distress; finally, the sample for prediction of financial distress prediction model has been established, the prediction results show that the correct rate of 86.67% and 91.11% respectively, and in the early warning model of financial distress prediction effect add non-financial refers to the standard and contains only the financial early-warning model of financial the indexes were compared. The results show that the introduction of non-financial indicators can help improve the ability to predict the financial distress prediction model.

【学位授予单位】：太原科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F276.6;F275

【参考文献】