Real-Time GARCH*
提出一种新的GARCH模型，将波动率描述为过去和当前信息的混合，连接了GARCH与随机波动率模型，能改进波动率预测、尾部拟合和调整速度，并提供了规范检验框架。
Most GARCH-type models follow Engle’s (1982) original idea of modeling the volatility of asset returns as a function of only past information. We propose a new model, which retains the simple GARCH structure, but describes the volatility process as a mixture of past and current information. We show how the new model can be interpreted as the special case of a stochastic volatility (SV) model, which provides therefore a link between GARCH and SV models. We show that we are able to obtain better volatility forecasts than the standard GARCH-type models; improve the empirical fit to the data, especially in the tails of the distribution; and make the model faster in its adjustment to the new unconditional level of volatility. Further, we offer a much needed framework for specification testing as the new model nests the standard GARCH models.