基于GARCH模型的人民币汇率风险的VaR方法研究
|
摘要
自从2005年人民币实施浮动汇率机制以来，人民币汇率的波动较为频繁。2010年，我国进一步推进了人民币汇率形成机制改革，增强了人民币汇率弹性。随着汇率市场化改革的进行，人民币汇率波动日渐市场化，我国涉外贸易投资主体、商业银行、中央银行等各大经济主体所面临的汇率风险也日益凸现。在此背景下，加强人民币汇率风险管理已成为摆在各大经济主体面前的重大课题，而其核心和前提是实现对人民币汇率风险的有效度量。
本文以2005年7月25日至20013年3月22日人民币汇率数据为样本，先用GARCH模型计算对数收益率方差，然后计算其VaR值,测算人民币汇率风险，最后通过Kupiec失败频率检验法检验模型的有效性。实证分析结果表明，人民币中间汇率对数日收益率相比于正态分布更适合t分布，对数收益率具有平稳性，不存在序列相关性，但具有异方差性，适合GARCH模型研究前提；通过比较不同阶数的GARCH模型，四种对数收益率均是GARCH(1，1)拟合优度最佳；所计算出的VaR值能很好的反映四个对数收益率序列的波动情况，GARCH模型在95％的置信水平下通过了失败频率检验法。
Since the implementation of a floating RMB exchange rate regime in2005,the RMB exchange rate fluctuate more frequently. In2010, China further promoted the reform of RMB exchange rate formation mechanism, which enhanced the RMB exchange rate flexibility. With the exchange rate market-oriented reforms progressed, China's main foreign trade and investment, commercial banks, central banks and other major economic entities faced risk increasingly apparent. In this context, to enhance the RMB exchange rate risk management has become a major economic agents placed in front of a major issue, and its core and the premise is to achieve an effective measure of the RMB exchange rate risk.
This paper empirically analysis the variance of the logarithmic rate of return from July25,2005to March22,2013,firstly use GARCH model to calculate the logarithmic yield variance, and then calculate the VaR value, finally test the validity of the model with failure frequency Kupiec test method. Empirical results show that the RMB exchange rate compared to the normal rate of return is more suitable for the t-distribution, logarithmic yields have smooth, there is no serial correlation, but with heteroscedasticity, GARCH model suitable premise; By comparing the model with different order,GARCH(1,1) is the best model;the calculated VaR values can be well reflected in four pairs of sequence numbers yield volatility situation, GARCH model at the95%confidence level through the failure frequency test method.
本文以2005年7月25日至20013年3月22日人民币汇率数据为样本，先用GARCH模型计算对数收益率方差，然后计算其VaR值,测算人民币汇率风险，最后通过Kupiec失败频率检验法检验模型的有效性。实证分析结果表明，人民币中间汇率对数日收益率相比于正态分布更适合t分布，对数收益率具有平稳性，不存在序列相关性，但具有异方差性，适合GARCH模型研究前提；通过比较不同阶数的GARCH模型，四种对数收益率均是GARCH(1，1)拟合优度最佳；所计算出的VaR值能很好的反映四个对数收益率序列的波动情况，GARCH模型在95％的置信水平下通过了失败频率检验法。
Since the implementation of a floating RMB exchange rate regime in2005,the RMB exchange rate fluctuate more frequently. In2010, China further promoted the reform of RMB exchange rate formation mechanism, which enhanced the RMB exchange rate flexibility. With the exchange rate market-oriented reforms progressed, China's main foreign trade and investment, commercial banks, central banks and other major economic entities faced risk increasingly apparent. In this context, to enhance the RMB exchange rate risk management has become a major economic agents placed in front of a major issue, and its core and the premise is to achieve an effective measure of the RMB exchange rate risk.
This paper empirically analysis the variance of the logarithmic rate of return from July25,2005to March22,2013,firstly use GARCH model to calculate the logarithmic yield variance, and then calculate the VaR value, finally test the validity of the model with failure frequency Kupiec test method. Empirical results show that the RMB exchange rate compared to the normal rate of return is more suitable for the t-distribution, logarithmic yields have smooth, there is no serial correlation, but with heteroscedasticity, GARCH model suitable premise; By comparing the model with different order,GARCH(1,1) is the best model;the calculated VaR values can be well reflected in four pairs of sequence numbers yield volatility situation, GARCH model at the95%confidence level through the failure frequency test method.
引文
[1]Akgiray and Booth.Model ing the Stochastic Behavior of Canadian Foreign Exchange Rates[J]. Journal of Mult inat ional Financial Management.1990,1:43-71.
[2]Hsieh.Testing for Nonlinear Dependence in Daily Foreign Exchange Rates [J].The Journal of Business. V01.62.No.3,1989:339-368.
[3]王佳妮,李文浩.GARCH模型能否提供好的波动率预测[J].数量经济技术经济研究,2005(6),17-21.
[4]池启水,刘晓雪.人民币汇率波动特征实证研究[J],统计与决策,2007(23),120—122.
[5]朱孟楠,严佳佳.人民币汇率波动：测算及国际比较[J],国际金融研究,2007(10),54-61.
[6]余波,罗辉.关于人民币汇率波动的研究——基于GARCH模型的分析[J],当代经济,2009(7),140-141.
[7]Baumol.WJ.The transactions demand for cash,An inventory theoretic approach[J],Quantitative Journal of Economy,1952,66(4):545-556.
[8]J.P.Morgan. Risk Metrics Techonolgy Document [M], New York.4thedition Risk MetricsGroup,Dec.1996,47-55.
[9]Marten M. Forecasting daily exchange rate volatility using intraday returns [J],Journal of InternationalMoney andFinance,2001(20):1-23.
[10]骆殉,吴建红.基于GARCH模型的人民币汇率波动规律研究[J],数理统计与管理,2009(2),295-300.
[11]王德全.外汇风险度量研究——基于GARCH类模型及VaR方法[J],2009(8),11—15.
[12]Philippe Jorion风险价值VaR-金融风险管理新标准[M].1996,87-121.
[13]Robert Engle.Autoregressive Conditional Heteroskedasticity with Estimates of the Variance of UK Inflation.[J],Econometrica.(1982)50:987-1008.
[14]Kupiecs Paul H. Techniques for Verifying the Accuracy of Risk Management Models[J]. Journal of Derivatives.1995,5(2),671-693.
[15]范英.VaR方法及其在股市风险分析中的应用初探[J].中国管理科学.2000,8(3)：26-32
[16]林美艳,薛宏刚.VaR模型及其在上海股市中的应用[J].辽宁大学学报.2006,33(1):6-10
[17]张术林,魏正红.RiskMetrics模型评估与扩展[J].数学的实践与认识.2009,3:34-43.
[18]徐中华.基于VaR历史模拟法的中国股市风险研究[J].复旦大学硕士学位论文.2008:21-30.
[19]李腊生,孙春华.VaR估计中的概率分布设定风险与改进[J].统计研究.2010(10):28-31.
[20]陈旭辉.基于VaR模型的我国商业银行的市场风险度量研究[J].湖南大学硕士论文.2010:4-20.
[21]杨帆.GARCH族模型的VaR方法在我国股市风险测量中的应用研究[J].哈尔滨工业大学硕士论文.2007:25-40.
[2]Hsieh.Testing for Nonlinear Dependence in Daily Foreign Exchange Rates [J].The Journal of Business. V01.62.No.3,1989:339-368.
[3]王佳妮,李文浩.GARCH模型能否提供好的波动率预测[J].数量经济技术经济研究,2005(6),17-21.
[4]池启水,刘晓雪.人民币汇率波动特征实证研究[J],统计与决策,2007(23),120—122.
[5]朱孟楠,严佳佳.人民币汇率波动：测算及国际比较[J],国际金融研究,2007(10),54-61.
[6]余波,罗辉.关于人民币汇率波动的研究——基于GARCH模型的分析[J],当代经济,2009(7),140-141.
[7]Baumol.WJ.The transactions demand for cash,An inventory theoretic approach[J],Quantitative Journal of Economy,1952,66(4):545-556.
[8]J.P.Morgan. Risk Metrics Techonolgy Document [M], New York.4thedition Risk MetricsGroup,Dec.1996,47-55.
[9]Marten M. Forecasting daily exchange rate volatility using intraday returns [J],Journal of InternationalMoney andFinance,2001(20):1-23.
[10]骆殉,吴建红.基于GARCH模型的人民币汇率波动规律研究[J],数理统计与管理,2009(2),295-300.
[11]王德全.外汇风险度量研究——基于GARCH类模型及VaR方法[J],2009(8),11—15.
[12]Philippe Jorion风险价值VaR-金融风险管理新标准[M].1996,87-121.
[13]Robert Engle.Autoregressive Conditional Heteroskedasticity with Estimates of the Variance of UK Inflation.[J],Econometrica.(1982)50:987-1008.
[14]Kupiecs Paul H. Techniques for Verifying the Accuracy of Risk Management Models[J]. Journal of Derivatives.1995,5(2),671-693.
[15]范英.VaR方法及其在股市风险分析中的应用初探[J].中国管理科学.2000,8(3)：26-32
[16]林美艳,薛宏刚.VaR模型及其在上海股市中的应用[J].辽宁大学学报.2006,33(1):6-10
[17]张术林,魏正红.RiskMetrics模型评估与扩展[J].数学的实践与认识.2009,3:34-43.
[18]徐中华.基于VaR历史模拟法的中国股市风险研究[J].复旦大学硕士学位论文.2008:21-30.
[19]李腊生,孙春华.VaR估计中的概率分布设定风险与改进[J].统计研究.2010(10):28-31.
[20]陈旭辉.基于VaR模型的我国商业银行的市场风险度量研究[J].湖南大学硕士论文.2010:4-20.
[21]杨帆.GARCH族模型的VaR方法在我国股市风险测量中的应用研究[J].哈尔滨工业大学硕士论文.2007:25-40.