實現波動率在DCC 模型上之應用

Abstract

本文針對不同投資組合風險值模型進行評比，並以S&P500 股價指數、那斯 達克股價指數和道瓊工業指數週資料為研究對象，實證結果發現將實現波動性基 礎下的DCC ( Dynamic Conditional Correlation ) 模型，在較高信心水準下，其準 確性較傳統報酬基礎下的模型良好，但整體而言，仍然以變幅基礎下的CCC ( Constant Conditional Correlation ) 模型表現最佳。 在研究中進一步改變了投資組合權重以及投資組合標的，其結果對於模型的 影響並不大，顯示出當投資權重改變或投資標的改變時不需要變動風險值估計模 型。

This paper investigates the difference portfolio Value-at-Risk models. We use the weekly data about the stock indices of S&P500 , NASDAQ , and DOW JONES for empirical analysis. The empirical results indicate that a DCC( Dynamic Conditional Correlation ) model based on realized volatility has better performance than the traditional return-based models. As a whole , the CCC( Constant Conditional Correlation ) model based on the range has the best performance. Furthermore , we changes the portfolio weights and the portfolio components. It doesn’t affect performance of the models. It showed that we don’t need to change Value-at-Risk models when the portfolio weights or components change.