金融风险管理的重中之重在于对金融资产实际波动率的预测.因为汇率市场的复杂性以及多变性,汇率波动率数据具有极强的异方差性.文章着重研究在异方差环境下,如何正确地使用最小二乘模型平均法来提高实际波动率的预测精度.文章以异质自回归(HAR)模型为基础,以不同的滞后项构建出多个候选模型.最终模型是所有候选模型的加权平均.而通过为每个候选模型配给不同的权重,模型平均法可以灵活动态地调节最终模型的结构.文章首先证明了所提出的最小二乘模型平均法具有渐近最优性.在随后大量实证中,发现所提出的方法在汇率实际波动率的预测精度方面优于很多同类方法.

Modeling and predicting the volatility of financial assets is an interesting issue in risk management. Recently a new approach to modeling volatility dynamics has relied on improved measures of ex post volatility composed from high-frequency daily data. This paper investigates least square model averaging approach under heteroskedasticity to forecast realized volatility. Candidate models are constructed from taking a full permutation of all of the possible lag terms of the conventional HAR model. By assigning differential estimated model weights to each candidate model, we achieve the effect of a flexible lag specification of the HAR model through model averaging. Furthermore, we prove that the proposed estimator is asymptotically optimal in the sense of achieving the lowest possible mean squared forecast error. Applied to exchange rate volatility over several forecast horizons, the proposed least square model averaging under heteroskedasticity provides very competitive forecasts, compared to the HAR model and the model averaging method assuming homoskedasticity.