基于已实现SV模型的动态VaR测度研究 本期目录 >>
Title: Study on Dynamic VaR Measures Based on Realized SV Model
作者 吴鑫育;周海林
Author(s): Wu Xinyu; Zhou Hailin
摘要: 基于日内高频数据构建的已实现波动率测度在金融计量经济学文献中引起了学者们的广泛关注. 将已实现波动率引入传统的SV模型(基于日度收益率), 同时考虑金融资产收益率与波动率的“有偏”、“尖峰厚尾”以及“非对称效应”等典型特征事实, 构建融合高频与低频数据信息的已实现SV(RSV)模型, 与有偏广义误差分布(sged)相结合来测度动态风险值(VaR). 为了估计RSV-sged模型的参数, 提出基于有效重要性抽样技巧的极大似然方法. 采用上证综合指数和深证成份指数日内高频数据进行的实证研究表明, RSV-sged模型能够有效地刻画中国股票市场的波动性特征, 并且展现出优越的风险测度能力.
Abstract: Accurate measurement of financial market risks plays an important role for the survival and development of financial institutions and the stability of the whole financial system. The recent 2007-2009 global financial crisis causes a broad impact on the real economy, highlights once again the necessity of financial market risk management. During this turbulent period of high volatility, accurate risk measurement and assessment are even more critical and encountered tremendous pressure and challenges, since there is a widespread risk of global financial instability. The most widely used market risk management tool is the so-called Value-at-Risk (VaR), which is widely used to assess the risk exposure of investments. It measures the worst expected loss over a given horizon within a given confidence level. Numerous financial institutions, risk managers as well as Bank for International Settlements (BIS) have adopted VaR as a first line of defense against market risk. VaR has become a standard risk measure used in financial risk management owing to its conceptual simplicity, ease of computation, and ready applicability. It is well-known that the latent volatility of asset returns is a crucial factor in accurately estimating VaR. There is a remarkable amount of empirical evidence that financial market volatility is not a constant but in fact changes over time. In addition, volatility clustering has been observed in financial return data. The most popular models used to capture these empirical stylized facts of volatility are the GARCH-type models and stochastic volatility (SV) models. Traditionally, VaR is computed based on these volatility models. However, the traditional GARCH-type models and SV models use only daily returns for modelling the volatility dynamics. Clearly an individual return observed on a given day can provide only limited information about the volatility of asset return. High-frequency financial data are now widely available and many authors have recently introduced a large number of realized volatility measures, such as realized volatility, bipower variation, realized kernel, and many others. These measures are far more informative about the current level of volatility than is the daily returns, which would provide a consistent estimator of the latent volatility in the ideal market condition. This motivates us to extend the traditional volatility models that use only daily returns also to take advantage of additional volatility information provided by high-frequency intra-day data to measure the financial market risks. It is now well-documented empirically that the volatility of asset returns respond asymmetrically to market news (good news and bad news), which is known as the volatility asymmetric effect. Moreover, the distribution of financial asset returns generally exhibit skewness, leptokurtosis and heavy-tails. It is important to specify these empirical stylized facts of financial asset returns as fail to do so can result in a substantial bias in VaR estimates. And several studies have shown that the skewed generalized error distribution (sged) can capture skewness, leptokurtosis and heavy-tails of financial asset returns successfully and yield more accurate VaR estimates than the alternatives. Based on the above analysis, this paper proposes the realized SV model with sged distribution (RSV-sged model), which incorporates the standard SV model that only use daily returns with realized volatility which is constructed by using the intra-daily high-frequency data and empirical stylized facts of financial asset returns (skewness, leptokurtosis, heavy-tails and volatility asymmetry effect), to dynamic VaR measures. The efficient importance sampling technique is proposed to implement the maximum likelihood method for our proposed RSV-sged model. Empirical results for intra-day data of Shanghai Stock Exchange composite index and Shenzhen Stock Exchange component index demonstrate that our proposed RSV-sged model can describe efficiently the volatility dynamics of Chinese stock markets and produce more accurate VaR estimates than other models.
关键词: 已实现SV模型;VaR;有偏广义误差分布;有效重要性抽样;极大似然
Keywords: realized SV model; VaR; skewed generalized error distribution; efficient importance sampling; maximum likelihood
基金项目: 国家自然科学基金资助项目;教育部人文社会科学研究资助项目;安徽省自然科学基金资助项目;安徽省高等学校省级优秀青年人才基金重点资助项目;国家自然科学基金资助项目
发表期数: 2018年 第2期
中图分类号: 文献标识码: 文章编号:
参考文献/References:

[1] 余素红, 张世英, 宋军. 基于GARCH模型和SV模型的VaR比较[J]. 管理科学学报, 2004, 7(5): 61-66.

[2] 蒋祥林, 王春峰. 基于贝叶斯原理的随机波动率模型分析及其应用[J]. 系统工程, 2005, 23(10): 22-28.

[3] 孙米强, 杨忠直, 余素红, 宋军. 基于随机波动模型的VaR的计算[J]. 管理工程学报, 2004, 18(1): 61-63.

[4] 周孝华, 董耀武, 姜婷. 基于EVT-POT-SV-GED模型的极值风险度量[J]. 系统工程学报, 2012, 27(2): 152-159.

[5] 周孝华, 张保帅. 基于SV-GED模型的极值风险度量研究[J]. 管理工程学报, 2014, 28(1): 171-178.

[6] 吴鑫育, 马宗刚, 汪寿阳, 马超群. 基于SV-SGED模型的动态VaR测度研究[J].中国管理科学, 2013, 21(6): 1-10.

[7] Andersen T G, Bollerslev T. Answering the skeptics: Yes, standard volatility models do provide accurate forecasts[J]. International Economic Review, 1998, 39: 885-905.

[8] Andersen T G, Bollerslev T, Diebold F X, et al. The distribution of realized stock return volatility[J]. Journal of Financial Economics, 2001, 61: 43-76.

[9] Barndorff-Nielsen O E, Shephard N. Non-Gaussian Ornstein-Ulhlenbeck-based models and some of their uses in financial economics[J]. Journal of the Royal Statistical Society, Series B, 2001, 63: 167-241.

[10] Barndorff-Nielsen O E, Shephard N. Econometric analysis of realized volatility and its use in estimating stochastic volatility models[J]. Journal of the Royal Statistical Society, Series B, 2002, 64: 253-280.

[11] Zhang L, Mykland P A, Aït-Sahalia Y. A Tale of two time scales: Determining integrated volatility with noisy high-frequency data[J]. Journal of the American Statistical Association, 2005, 100(472): 1394-1411.

[12] Hansen P R, Lunde A. Realized variance and market microstructure noise[J]. Journal of Business & Economic Statistics, 2006, 24(2), 127-161.

[13] Bandi F M, Russell J R. Microstructure noise, realized variance, and optimal sampling[J]. Review of Economic Studies, 2008, 75(2): 339-369.

[14] Bandi F M, Russell J R. Market microstructure noise, integrated variance estimators, and the accuracy of asymptotic approximations[J]. Journal of Econometrics, 2011, 160(1): 145-159.

[15] Barndorff-Nielsen O E, Hansen P R, Lunde A, et al. Designing realized kernels to measure the ex-post variation of equity prices in the presence of noise[J]. Econometrica, 2008, 76(6): 1481-1536.

[16] Barndorff-Nielsen O E, Hansen P R, Lunde A, et al. Subsampling realised kernels[J]. Journal of Econometrics, 2011, 160(1): 204-219.

[17] Hansen P R, Lunde A. A forecast comparison of volatility models: Does anything beat a GARCH(1, 1)?[J]. Journal of Applied Econometrics, 2005, 20(7): 873-889.

[18] Hansen P R, Lunde A. A realized variance for the whole day based on intermittent high-frequency data[J]. Journal of Financial Econometrics, 2005, 3(4): 525-554.

[19] Takahashi M, Omori Y, Watanabe T. Estimating stochastic volatility models using daily returns and realized volatility simultaneously[J]. Computational Statistics & Data Analysis, 2009, 53(6): 2404-2426.

[20] Dobrev D P, Szerszen P J. The information content of high-frequency data for estimating equity return and forecasting risk[R]. International Finance Discussion Papers 1005, Board of Governers of the Federal Reserve System, 2010.

[21] Koopman S J, Scharth M. The analysis of stochastic volatility in the presence of daily realised measures[J]. Journal of Financial Econometrics, 2013, 11(1): 76-115.

[22] Shirota S, Hizu T, Omori Y. Realized stochastic volatility with leverage and long memory[J]. Computational Statistics & Data Analysis, 2014, 76: 618-641.

[23] Venter J H, de Jongh P J. Extended stochastic volatility models incorporating realised measures[J]. Computational Statistics & Data Analysis, 2014, 76: 687-707.

[24] Zheng T G, Song T. A realized stochastic volatility model with Box-Cox transformation[J]. Journal of Business and Economic Statistics, 2014, DOI:10.1080/07350015.2014.918544.

[25] Takaishi T. Bayesian estimation of realized stochastic volatility model by Hybrid Monte Carlo algorithm[J]. Working paper, Hiroshima University of Economics, 2014.

[26] Hansen P R, Huang Z, Shek H H. Realized GARCH: A joint model for returns and realized measures of volatility[J]. Journal of Applied Econometrics, 2012, 27: 877-906.

[27] Hansen P R, Huang Z. Exponential GARCH modeling with realized measures of volatility[R]. Working paper, European University Institute & CREATES, 2012.

[28] Watanabe T. Quantile forecasts of financial returns using realized GARCH models[J]. Japanese Economic Review, 2012, 63(1): 68-80.

[29] Liesenfeld R, Jung R C. Stochastic volatility models: Conditional normality versus heavy-tailed distributions[J]. Journal of Applied Econometrics, 2000, 15: 137-160.

[30] Theodossiou P. Skewed generalized error distribution of financial assets and option pricing[R]. Working paper, Rutgers University, 2000.

[31] Broto C, Ruiz E. Estimation methods for stochastic volatility models: A survey[J]. Journal of Economic Surveys, 2004, 18(5): 613-649.

[32] Richard J F, Zhang W. Efficient high-dimensional importance sampling[J]. Journal of Econometrics, 2007, 127(2): 1385-1411.

[33] Koopman S J, Shephard N, Creal D. Testing the assumptions behind importance sampling[J]. Journal of Econometrics, 2009, 149: 2-11.

[34] Kupiec P. Techniques for verifying the accuracy of risk measurement models[J]. Journal of Derivatives, 1995, 3(2): 73-84.