基于BP神经网络的上市公司财务危机预警研究

发布时间：2018-01-04 02:12

本文关键词：基于BP神经网络的上市公司财务危机预警研究　出处：《南华大学》2014年硕士论文　论文类型：学位论文

【摘要】：目前，我国证券市场的不断发展导致目前关于我国上市公司的相关政策和法规不断趋于完善，这对于我国上市公司而言既是机会也是威胁。机会在于法律法规的不断完善使证券市场变得更趋于规范化；威胁在于这也大大的约束了上市公司，稍不留神公司也许就会被“ST”或“*ST”。因此，在现代激烈的市场竞争下，企业若想生存、发展和获利就必须要加强对企业财务风险的控制和财务危机的防范。鉴于此，建立一套行之有效的上市公司财务危机预警系统迫在眉睫。本文首先从BP神经网络和现金流量指标的选取两方面对国内外相关文献进行了简单回顾，然后介绍了财务预警相关的理论基础，界定了财务危机的含义及实证中财务危机样本的类型，也详细阐述了BP神经网络相关理论。随后在前人研究的基础上，本文从偿债能力，运营能力，盈利和获现能力，成长能力，财务弹性以及现金流量结构等六个方面共选取了26个指标，建立了一套以现金流量指标为主财务预警指标体系。本文的实证部分选取了2007年到2012年间沪、深两市A股共77家ST公司，并按照同行业、同时期、同规模的原则选取77家非ST公司做为本文的训练样本，用样本被ST前一年、二年、三年的数据对指标进行筛选，并用筛选出来的指标建模。本文用神经网络的方法建模，实证分析的结果表明BP神经网络方法所建立的财务预警模型在上市公司被ST前三年的预警准确率分别达到了96.08%、88.24%和78.92%，，从而证实了神经网络财务预警模型的优越性及其预测的准确性。本文得到的结论如下：（1）本文所建立的以现金流量为主的财务预警指标体系具有很好的财务预警效果；（2）本文用BP神经网络的方法建立的财务预警模型预测精确度高，具有很强的应用价值；（3）本文构建的神经网络模型，当模型输入数据在[0,1]之间时，能够取得较为稳定的预测结果。
[Abstract]:At present, the continuous development of the securities market in China has led to the improvement of the relevant policies and regulations of listed companies in China. This is both an opportunity and a threat to our listed companies. The opportunity lies in the continuous improvement of laws and regulations to make the securities market more standardized; The threat lies in the fact that this greatly restricts the listed company, and the company may be "St" or "St". Therefore, in the modern fierce market competition, the enterprise wants to survive. In order to develop and make profits, we must strengthen the control of financial risk and the prevention of financial crisis. In view of this, it is urgent to establish a set of effective financial crisis warning system of listed companies. In this paper, the BP neural network and the selection of cash flow indicators of the two aspects of the domestic and foreign literature were reviewed, and then introduced the theoretical basis of financial early warning. Define the meaning of financial crisis and the types of financial crisis samples in the empirical, but also elaborate the relevant theory of BP neural network. Then on the basis of previous studies, this paper from the solvency, operational capacity. A total of 26 indicators were selected in six aspects, namely, profitability and cash flow structure, growth capacity, financial elasticity and cash flow structure. Established a set of cash flow indicators as the main financial warning index system. The empirical part of this paper selected a total of 77 St companies in Shanghai and Shenzhen A shares from 2007 to 2012, and according to the same industry. At the same time, the same scale of 77 non-St companies as the training sample of this paper, with the sample of St one year, two years, three years to screen the indicators. Modeling with the selected index. This paper uses the method of neural network modeling. The results of empirical analysis show that the financial early warning model established by BP neural network method in listed companies before St three years of warning accuracy reached 96.08% 88.24% and 78.92% respectively. The superiority of neural network financial early warning model and the accuracy of its prediction are verified. The conclusion of this paper is as follows: 1) the financial forewarning index system based on cash flow established in this paper has a good financial early warning effect; 2) the financial early warning model established by BP neural network has high accuracy and has strong application value. 3) the neural network model constructed in this paper, when the model input data in [Between 0 and 1, a more stable prediction result can be obtained.
【学位授予单位】：南华大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP183;F832.51;F275

【参考文献】