对于预测上海股市股指波动性的数量模型的比较

目前对于随时间而变化的波动性预测模型主要有两类:一类是假设波动性的变化是一个确定性过程的GARCH模型,另一类是假设波动性的变化是一个随机性过程的随机波动模型.本文同时将这两类模型(GARCH(1,1)模型及其特例RiskMetrics模型,随机波动模型)应用于上海股市中,通过对样本外区间的预测准确度来衡量这两类模型对我国股市波动性的预测能力.通过4个预测准确度指标,MS模型对于上证指数在样本外区间的预测相对于另外两个模型而言都是显著最佳的.RiskMetrics模型的表现则要优于GARCH(1,1),且RiskMetrics在实际运用中要方便的多.

The Comparison of Models Forecasting Shanghai Stock Market Time-Varying Volatility

There are two types of models commonly applied to measure time-varying volatilities: one is the GARCH model which assumes that volatility follows a deterministic process,and the other is stochastic volatility model which assumes that the changing volatility is a stochastic process.We here apply these timevarying volatility models,the GARCHG(1,1) with its special case RiskMetrics model,and stochastic volatility model to Shanghai stock market to compare the performance of these two models via out-of-sample forecasting.Our results show that the MS model significantly outperforms the other two models for all the 4 measurement criteria considered.In addition,though the RiskMetrics model performs slightly better than the more general GARCH(1,1) model,it is much easier to implement than the latter.