基于FIGARCH-EVT-Copula模型的外汇投资组合风险测度研究

发布时间：2018-01-06 10:26

本文关键词：基于FIGARCH-EVT-Copula模型的外汇投资组合风险测度研究　出处：《重庆理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着人民币汇率体制的改革和外汇市场机制的不断完善,外汇投资现已成为继股票投资后的又一重要投资领域。近年来,又由于各国资源开始重新配置,加剧了外汇市场波动的频繁和幅度,导致外汇风险不断加大。在这一背景下研究我国外汇市场的风险,并对其进行度量和管理已尤为重要。本文主要运用FIGARCH-EVTCopula模型基于美元、欧元、日元和港币兑人民币汇率样本数据对我国外汇投资组合风险进行测度研究。首先运用经典R/S分析和修正的R/S分析法检验四种外汇波动率序列的长记忆性,结果表明:我国外汇市场波动率序列存在显著的长记忆性特性,其中美元/人民币波动率序列的长记忆性最强,欧元/人民币、日元/人民币和港币/人民币的长记忆性强度依次递减。针对四种外汇波动率序列的长记忆性特征,引入能准确刻画波动率长记忆性的FIGARCH模型,并结合能很好刻画尾部风险的极值理论,构造出FIGARCH-EVT模型,同时引入传统的多元正态Copula函数和多元t-Copula函数,得到FIGARCHEVT-Copula模型。然后运用该模型对四种外汇投资组合风险进行测度研究,进一步计算出四种外汇在等权重投资下的风险值,并与GARCH-EVT-Copula模型下的风险值进行对比发现:在同一置信水平下,GARCH-EVT-Copula模型得到的风险值都小于FIGARCH-EVT-Copula模型,说明GARCH-EVT-Copula模型低估外汇风险,相对来说FIGARCH-EVT-Copula模型对于测度外汇投资组合风险更准确。为了更好地刻画高维相关结构,本文在Copula函数的选取中,又引入了多元藤结构Pair Copula,构造出FIGARCH-EVT-Pair Copula模型。运用该模型对四种外汇投资组合风险进行测度研究,计算出在风险最小情况时,四种外汇的最优投资比例以及相应的组合风险值。并将之与传统的FIGARCH-EVT-t-Copula模型进行对比发现:在相同置信水平下,FIGARCH-EVT-Pair Copula模型得到的组合风险值都高于FIGARCH-EVT-t-Copula模型;然后再运用Kupiec和Christoffersen检验法对上述两种模型的VaR预测效果进行稳健性检验,检验结果表明:在95%和99%置信水平下,FIGARCH-EVT-t-Copula模型不能通过稳健性检验,而FIGARCH-EVT-Pair Copula模型则通过,进而说明FIGARCH-EVT-Pair Copula模型在多元外汇投资组合风险度量时比FIGARCH-EVT-Copula模型更具优越性。同时还研究发现无论是哪种类型Copula函数的最小风险投资组合系数相差都不是很大,目前外汇投资仍然集中在美元,其它外汇投资较少。本文运用FIGARCH-EVT-Copula模型对外汇投资组合风险进行测度研究,为进一步研究外汇投资组合风险提供了一定的参考价值。
[Abstract]:With the RMB exchange rate system reform and the foreign exchange market mechanisms continue to improve, foreign investment has become an important after the stock investment after. In recent years, because the resources in different countries began to re allocation, exacerbated the frequent and the amplitude of fluctuation in the foreign exchange market, foreign exchange led to increasing the risk. The risk of China's foreign exchange market in the a background, and has been particularly important to measure and management. This paper uses the FIGARCH-EVTCopula model based on the US dollar, euro, yen and Renminbi exchange rate data measure research on Chinese foreign exchange investment portfolio risk. By using classical R/S analysis and modified R/S analysis method was used to test four kinds of foreign exchange volatility series long memory, the results show that the fluctuation of China's foreign exchange market rate series has long memory characteristic, the dollar / Renminbi volatility sequence length The strongest memory, EUR / RMB / yen, RMB and HKD / RMB long memory strength decreasing. According to the four kinds of foreign exchange volatility series of long memory characteristics, introduced FIGARCH model can accurately depict the volatility of long memory, and can well describe the tail risk of extreme value theory, construct FIGARCH-EVT at the same time, the introduction of the traditional model, the multivariate normal Copula function and multiple t-Copula function, FIGARCHEVT-Copula model. Then the model is used to measure research on four kinds of foreign exchange portfolio risk, further calculate four kinds of foreign exchange in the weight of investment risk and risk value, and the GARCH-EVT-Copula model were compared with values found in the a confidence level, the risk GARCH-EVT-Copula model obtained values are less than the FIGARCH-EVT-Copula model, GARCH-EVT-Copula model underestimated relative to foreign exchange risk FIGARCH-EVT-Copula model for foreign exchange portfolio risk measure more accurately. In order to better describe the high dimensional structure, this paper selected the Copula function, and introduced multiple rattan structure of Pair Copula, constructed FIGARCH-EVT-Pair Copula model. The model is used to measure research on four kinds of foreign exchange portfolio risk is calculated under the minimum risk when the four kinds of foreign exchange the optimal investment proportion of portfolio risk and corresponding value. And the FIGARCH-EVT-t-Copula model with the traditional comparison: in the same confidence level, the portfolio risk FIGARCH-EVT-Pair Copula model values are higher than the FIGARCH-EVT-t-Copula model; and then use the Kupiec test and Christoffersen of the two kinds of model VaR forecasting. For the robustness test, the test results show that: in the 95% and 99% confidence level, FIGARCH-EVT-t-Copula model Type not through the robust test, and the FIGARCH-EVT-Pair Copula model by FIGARCH-EVT-Pair, and that of Copula model in multi currency portfolio risk measurement model is more advantageous than FIGARCH-EVT-Copula. At the same time also found that either type of Copula function of the minimum risk portfolio coefficient difference is not great, the current foreign investment is still concentrated in the U.S. other foreign exchange, less investment. This paper uses the FIGARCH-EVT-Copula model of foreign exchange portfolio risk measure research, to provide a certain reference value for the further study of the risk of foreign exchange portfolio.

【学位授予单位】：重庆理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F830.59

【参考文献】