無模型假設波動度之預測能力

Abstract

本篇研究首先檢驗了無模型假設波動度的預測能力，接著我們將研究重點延伸到高階動差資訊的探討。我們的實證結果指出，三階動差含有部分無模型假設波動度所沒有的資訊，但四階動差則不能有效預測未來波動度。另外，本篇研究也將無模型假設動差進行拆解，並比較系統性和非系統部分的預測表現。我們的研究結果顯示，非系統性部分波動度之個別預測能力優於系統性部分，不過另一方面，系統性及非系統性部分波動度具有不同的資訊含量，兩者都可以用於預測未來波動度。

In this paper, we first examine the forecasting power of the model-free volatility, and then we extend our study by investigating the information content of the higher moments. Our results suggest that skewness contains some information other than the model-free volatility while kurtosis seems to be an inefficient forecast for the realized volatility. This paper also decomposes the model-free moments and compares the predicting performance of the systematic and the unsystematic parts. Our findings suggest that unsystematic volatility generally outperforms the systematic part of the volatility. However, the results also show that both parts of the volatility contain useful information that can be used to predict the realized volatility.