基于Copula-Garch模型的我国保险公司最优投资比例研究

发布时间：2018-05-20 23:33

本文选题：保险投资组合 + Copula-Garch模型　；参考：《浙江工商大学》2013年硕士论文

【摘要】：当前,我国保险业竞争日趋激烈,保险公司的承保利润持续下降,保险公司想要获利必须依靠保险投资业务。但是我国保险投资的收益率仍然较低,严重影响了我国保险业的健康发展,而导致保险投资收益率偏低的一个重要原因就是投资比例的不合理。由此可见,如何优化投资结构,提高投资收益已成为各保险公司面临的一个重大考验。目前,保险学术界研究的一个热点问题即为如何求得保险公司的最优投资比例,为此,学者们构建了各种不同的模型。本文创新性地将Copula-Garch模型和均值-CVaR模型进行有机结合,引入到保险投资比例的研究上来,为这一领域的研究打开了新的视角。我们知道,Copula函数是衡量相关性最有效的工具之一,因为传统的相关性分析建立在资产收益率符合正态分布的假设下,而现实的金融市场数据往往不符合这一特点,但Copula函数可以准确求得非正态、非线性条件下多项资产的联合分布,从技术上解决了这一难题。与此同时,CVaR方法已经成为当前金融风险管理中最热门的风险度量方法之一,它能充分考虑到小概率极端事件的发生,使风险衡量更为安全、有效。本文假设保险投资组合中含有银行存款、国债、企业债、基金、股票五项资产,选取Shibor隔夜拆借利率、国债指数、企业债指数、基金指数和股票指数来分别模拟其收益率。首先,利用Garch (1,1)-t模型来拟合单个资产的收益率数据,求得单个资产的边缘分布特征；然后利用Copula-t函数来求解各资产之间的相关系数矩阵,从而得到多个资产收益率的联合分布；接着用CVaR (?)代替方差作为风险度量标准,建立均值-CVaR模型,以在一定收益率水平下令CVaR最小为目标,求得最优投资组合中各资产的比重；最后简要分析了实证结果,并提出改善建议。
[Abstract]:At present, the competition of insurance industry in our country is becoming more and more intense, the underwriting profit of insurance company is declining continuously, the insurance company must depend on the insurance investment business to make profit. However, the return rate of insurance investment in China is still low, which seriously affects the healthy development of the insurance industry in China, and an important reason leading to the low return rate of insurance investment is the unreasonable investment ratio. Thus, how to optimize the investment structure and improve investment returns has become a major test faced by insurance companies. At present, one of the hot issues in the research of insurance academia is how to get the optimal investment ratio of insurance companies. For this reason, scholars have constructed various models. This paper innovatively combines the Copula-Garch model with the mean Cvar model, and introduces it into the study of the proportion of insurance investment, which opens a new perspective for the research in this field. We know that the Copula function is one of the most effective tools to measure relevance, because the traditional correlation analysis is based on the assumption that the return on assets conforms to the normal distribution, and the real financial market data often do not conform to this characteristic. However, the Copula function can accurately obtain the joint distribution of multiple assets under the condition of non-normal and nonlinear conditions, which solves this problem technically. At the same time, CVaR method has become one of the most popular risk measurement methods in current financial risk management. It can fully consider the occurrence of small probability extreme events and make risk measurement safer and more effective. This paper assumes that the insurance portfolio contains five assets: bank deposits, treasury bonds, corporate bonds, funds and stocks, and selects Shibor overnight borrowing rate, treasury bond index, enterprise bond index, fund index and stock index to simulate the return rate respectively. Firstly, the Garch model is used to fit the yield data of a single asset, and the edge distribution characteristic of a single asset is obtained, and then the correlation coefficient matrix of each asset is solved by using the Copula-t function, and the joint distribution of multiple asset returns is obtained. CVaR) Instead of variance as a measure of risk, a mean value CVaR model is established, in order to order the minimum of CVaR at a certain rate of return, the proportion of each asset in the optimal portfolio is obtained. Finally, the empirical results are briefly analyzed, and suggestions for improvement are put forward.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：F842.3;F224

【参考文献】