基于ordered probit模型的我国上市公司信用评级的研究

发布时间：2018-01-06 03:15

本文关键词：基于ordered probit模型的我国上市公司信用评级的研究　出处：《大连理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：受国际经济形势的影响,我国的市场经济也在不断的发展和变化,随之而来的信用评级在经济的发展中扮演着越来越重要的角色。投资行业、银行业等对于准确估计企业信用等级也提出了更高的要求,而现有的信用评级体系和方法主要是建立在财务指标数据上得到的,通过借鉴国内外信用评级的研究成果发现,非财务指标对企业信用评级也有重大影响,综合考量财务指标和非财务指标能够更准确的估计企业信用等级。又因为信用等级是多分类有序变量,本文选用了有序概率模型,这种模型是估计有序离散变量最常采用的模型之一。本文收集了38家上市公司的财务指标与非财务指标作为样本数据,利用统计软件R语言对数据进行描述性统计分析、相关性分析及回归分析,使用逐步回归法简化模型。估计结果显示非财务指标对信用评级确有显著影响,逐步回归后得到的模型AIC值最小,拟合优度较好,模型中保留的变量对企业信用评级均有显著影响。有序概率模型是以随机扰动项服从标准正态分布为基本假定,得到如上估计结果,因此本文使用拉格朗日乘数检验法对该假定做出判断,检验结果显示不能拒绝模型中随机扰动项服从标准正态分布。
[Abstract]:The impact of the international economic situation, China's market economy is also in constant development and change, the credit rating plays an increasingly important role in economic development. The investment banking industry, for the accurate estimation of corporate credit rating is also put forward higher requirements, and the existing credit rating system and method the main is to establish in the financial index data, based on the research results of the domestic and foreign credit rating that non-financial indicators have a significant impact on the enterprise credit rating, comprehensive consideration of financial and non-financial indicators of corporate credit rating can be estimated more accurately. And because the credit rating is ordered categorical variables, the the ordered probability model, this model is one of the estimated model ordered discrete variables are most often used.
This paper collected 38 listed companies of the financial indicators and non-financial indicators as sample data, descriptive statistical analysis of data using statistical software R, correlation analysis and regression analysis, using stepwise regression method to simplify the model. The estimation results show significant influence on the non-financial indicators of credit rating is obtained after the stepwise regression model of AIC value the smallest, good fitting, the model variables retained significant influence on enterprise credit rating. The ordered probability model is perturbed obey standard normal distribution assumption in a random, such as on the estimation results, so this paper use the method to judge the Lagrange multiplier test assumes that the test results show that the model cannot be rejected in random the disturbance obeys normal distribution.

【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F832.51;F270

【参考文献】