GARCH模型对期铜市场风险的研究

Research on Copper Futures Risk by GARCH Model

期货市场是一个高风险的市场,因此需要有效地控制并且监管风险。本文以上海期铜市场97年到04年的收盘价格为研究样本,通过拉格朗日检验,发现价格收益率序列服从ARCH过程,在正态、student-t和GED三种分布假设下,估计了GARCH(1,1)模型的参数,结果表明student-t假设下模型的拟和程度较好,然后利用EGARCH(1,1)-M模型检验了上海期铜市场杠杆效应和波动集群效应。最后在两种置信水平下,利用GARCH(1,1)和Risk Metrics方法计算了期铜市场每天的VaR,Kupiec检验表明基于t分布的GARCH(1,1)模型能更准确地反映上海期铜市场的风险。

Futures market is of high risk, so we need to control its risk efficiently. We examine the daily returns of the copper futures in SHFE（Shanghai Futures Exchange） from 1997 to 2004. the Lagrange Multiplier test verifie that the returns series of the copper futures is an ARCH process. Then we estimate the parameters of GARCH （1,1） with three distributions （normal, student-t, generalized error distributions）. Better results are achieved when using student-t ＆ GED distribution. EGARCH（1,1）-M model shows there exists the effect of leverage and persistence of volatility in the copper futures of SHFE. The daily VaRs of the copper futures in SHFE are computered with GARCH（1,1） and riskmetrics （95 and 99 percent confidence interval respectively）. The Kupiec test indicates that GARCH（1,1）-t model reflects the real market risk more accurate than GARCH （1,1）-N and risk metrics.