质押式回购利率的风险度量研究——基于ARMA-GARCH模型的实证检验

文章对2002年1月4日至2009年3月31日我国银行间质押式回购市场进行实证研究,结果表明:(1)t-分布和g-分布下的模型能更好地捕捉回购利率序列的尖峰厚尾性;(2)回购利率波动具有显著的非对称性,利率上升时的波动更大;(3)ARMA-PARCH-M模型是估计回购利率VaR值的理想模型,t-分布下的模型适合多头头寸VaR值的预测,而g-分布下的模型适合空头头寸VaR值的预测.这说明我国回购市场的利率风险较高.

Risk Measurement of Pledged Repo Interest Rate:Empirical Study Based on ARMA-GARCH Model

The paper makes an empirical study on the inter-bank bond pledged repo market in China from Jan.4,2002 to March 31,2009.The empirical results show that models with t-distribution and g-distribution can better portray the heavy-tailed feature of leptokurtosis of repo interest rate series.The repo interest rate volatility is significantly asymmetric and is bigger when the interest rate rises.ARMA-GARCH model with t-distribution is fit for the estimation of VaR of long position,and ARMA-GARCH model with g-dist...