| Summary: | 碩士 === 元智大學 === 財務金融學系 === 96 === In assessing VaR for multi-period holding returns, the square-root-of-time rule is undoubtedly the most easy and popular one within the scale of estimation methods. However, this rule pre-assumes several un-realistic assumptions such as (iid), normal and homoskedasticity distribution of the return series, which may against the stylized facts of most financial returns. In this dissertation, we examine and reconcile different potential bias factors of financial return series from the literature to see how bias the square-root-of-time rule without considering other return characteristics. We consider time series specifications including serial dependence, volatility clusters, heavy tail and jump components in asset return to examine the volatility of scaling by square-root-of-time rule. The ultimate goal is to set up a standard recipe for practitioners to assess the size and direction of bias compared to the true multi-period VaR they may potentially commit for different return features. Finally, we attempt using the method of variance -ratio test to evaluate the feasibility of the square-root-of time rule.
|
|---|
