我国旅游上市公司信用风险的预测方法研究

发布时间：2018-12-14 12:48

【摘要】：信用风险是企业运行过程中会面临的主要风险之一,是研究企业管理领域不可或缺的一块研究内容。我国旅游业虽然起步较晚,但是作为我国的朝阳产业,旅游业具备很好的发展前景。然而在旅游公司大量涌现的同时,也出现了一些公司治理结构不合理、同行恶性竞争、内部管理混乱、信用出现危机等问题。尤其是信用风险不仅使得企业遭受巨大的损失,也影响了企业甚至整个旅游行业的发展。因此,如何有效的预测企业的信用风险,使得企业及时采取措施加以控制和防范,已经成为了企业和学者关注的重要问题。本文首先选取了在上海和深圳证券交易所交易上市的28家旅游上市公司作为信用风险预测的企业样本,并采用全样本的形式进行实验。结合指标变量的选取原则选取了15个财务比率作为样本变量,收集了企业2001年到2010年的数据形成数据集。经过数据的预处理,并对其作出充分的统计描述,分析了数据的均值和标准差的变化,同时进行了正态性检验,证明样本对信用风险预测方法进行实验的适用性。为了消除不同数量级的影响,进行了数据的标准化处理。其次分两种方法进行实证研究。第一种是将数据集进行四种不同方法的处理,分别是标准化、Random-sampling算法、流形学习、结合Random-sampling算法和流形学习。通过三种模型MDA、Logit、Probit的分析比较,得出预测准确率,从而验证了本文采用的Random-sampling算法均衡数据和结合流形学习中的ISOMAP和LLE算法降维处理的方法是有效的,且找出了最佳的预测模型为Logit模型,这种数据的处理方法和Logit模型的结合产生的预测准确率达到了最佳,且得出越接近ST的年份预测准确率越高。另一种方法是结合了因子分析和Logit模型进行预测,得到的模型有良好的模拟效果,最终也达到了较好的预测准确率,同时具体分析了我国旅游上市公司信用风险预测的影响指标,从而得出了企业在经营过程中应该重视的财务指标以及其波动情况,并给出了相应的建议,可以避免公司出现信用风险。本文对我国旅游上市公司的信用风险预测的方法研究,能够为企业开展信用风险预测提供技术支持,也丰富了预测的实验方法,具有一定的理论和实际意义。
[Abstract]:Credit risk is one of the main risks that enterprises will face in the course of operation, and it is an indispensable research content in the field of enterprise management. Although the tourism industry of our country starts late, as the sunrise industry of our country, the tourism industry has a very good development prospect. However, with the emergence of a large number of tourism companies, there are also some problems such as unreasonable corporate governance structure, vicious competition among peers, confusion in internal management, credit crisis, and so on. Especially credit risk not only makes enterprises suffer huge losses, but also affects the development of enterprises and even the whole tourism industry. Therefore, how to effectively predict the credit risk of enterprises, so that enterprises take timely measures to control and prevent, has become an important issue that enterprises and scholars pay attention to. In this paper, 28 listed tourism companies listed on the Shanghai and Shenzhen stock exchanges are selected as samples for credit risk prediction, and the experiments are carried out in the form of full samples. Combined with the selection principle of index variables, 15 financial ratios are selected as sample variables, and the data set of enterprises from 2001 to 2010 is collected. After data preprocessing and sufficient statistical description, the variation of the mean value and standard deviation of the data is analyzed. At the same time, the normality test is carried out to prove the applicability of the sample to the credit risk prediction method. In order to eliminate the influence of different orders of magnitude, the data are standardized. Secondly, it is divided into two methods to carry on the empirical research. The first is to process data sets in four different ways: standardization, Random-sampling algorithm, manifold learning, combined with Random-sampling algorithm and manifold learning. Through the analysis and comparison of the three MDA,Logit,Probit models, the prediction accuracy is obtained, which verifies the effectiveness of the Random-sampling equalization algorithm and the ISOMAP and LLE algorithms in manifold learning. The best prediction model is Logit model. The combination of the data processing method and the Logit model results in the best prediction accuracy, and the higher the prediction accuracy is, the closer it is to the ST. Another method is to combine factor analysis and Logit model to predict, the obtained model has good simulation effect, and finally achieved a better prediction accuracy. At the same time, it analyzes the influence index of the credit risk forecast of the tourism listed company in our country, and obtains the financial index and its fluctuation situation which the enterprise should pay attention to in the course of operation, and gives the corresponding suggestion. The company can avoid credit risk. In this paper, the research on the methods of credit risk prediction of tourism listed companies in China can provide technical support for enterprises to carry out credit risk prediction, and enrich the experimental methods of prediction, which has certain theoretical and practical significance.
【学位授予单位】：浙江师范大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F592.6;F224

【相似文献】