A Study of Conditional and Unconditional Distribution Types on Futures Hedging Performance
| Autor: | Che-Hung Liao, 廖哲宏 |
|---|---|
| Rok vydání: | 2005 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 93 This article mainly discusses the effects of data distributions on futures hedging performance. The estimated optimal hedge ratios (OHR) in previous research generally base on the assumption of normal distribution. Under the minimum variance hedging framework, this study possibly incorporates the real phenomenon that the distribution of financial asset returns is leptokurtic in the second-moments estimating model to enhance the hedging performance. Take the advantage of the assumption of the generalized error distribution (GED), the employed exponentially weighted moving average (EWMA) estimator further enable to fit various distribution types. We use the conditional and unconditional distribution type of the data to compute the OHR. The conditional way allows the distribution type varying with the recently available information; the unconditional way assumes the distribution type, which is decided by user arbitrarily or by data itself in this case, is invariant with time. The samples are five stock index markets including TAIEX, MSCI Taiwan Stock Index, S&P 500, NASDAQ 100, and Dow Jones. Base on the criterion of minimizing the variance of hedged portfolio returns, the empirical results indicate that choices of the best model differ from the underlying sample. Generally speaking, the EWMA estimator accommodating distribution types yields superior performance to the ones that base on normal distribution. Additionally, estimate the OHR conditionally is able to capture the possibly time-varying distribution type, but the unconditional one that decided by user generates better performance in most cases. However, as the statistic test is implemented to examine the significant difference of the performance, we find that these two approaches do not affect the hedging performance significantly in most cases as the duration of hedge is one day, but it is not the case when the duration of hedge is extended. The minimum variance OHR is a quotient between a covariance and a variance; this may offset the benefit gains form estimating conditionally and result in the hedging performance can not significantly superior to, or even inferior to the unconditional way. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|