Investigating Asymmetric Mean-Reversion of Stock Return, Inter-market Effect and Contrarian Strategy－Using ANST GARCH Model

• Main Authors: Lu, Chih-Chiang, 盧智強
• Other Authors: Goo, Yeong-Jia
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/33486572309531669580

博士 === 國立臺北大學 === 企業管理學系 === 92 === Although the profitability of contrarian is obviously related to the Mean-Reverting Behavior of return, few theses have explored it in terms of dynamic time series but elaborated on where it comes forth, how winner-loser investment combinations are successfully set up, or why difference persists in return calculation. This paper stands out in that it adopts time series ANST-GARCH model, which features a leverage effect in both conditional mean and conditional variance equations, to investigate whether asymmetric reverting pattern can be developed in monthly, weekly, and daily excess return and in volatility dynamic process of TSEC, OTC, TAIEX future markets, to examine the factors behind it, and to further established it as one of the sources of contrarian. In this paper, a great effort also goes to fashion a bivariate ANST-GARCH model to analyze if, in each two of the 3 markets, the dynamic process of daily excess return exhibits autocorrelation or volatility asymmetric reverting pattern and if this pattern exists in inter-market transmission and volatility spillover. The results are as follows: 1.All monthly, weekly, and daily excess returns for both TSEC and OTC show persistent asymmetric reverting pattern while only daily excess return for TAIEX future shows the same pattern. Besides, leverage effect goes with the volatility of all series except for monthly excess return of TAIEX future. 2.Under time varying rational expectation hypothesis, the daily excess return for both TSEC and OTC and the weekly excess return for listed stocks still display asymmetric reverting pattern. The asymmetric return pattern of daily excess return for all three markets and the weekly return for OTC can be explained by rational expectation hypothesis. 3.The empirical relationship between data frequency interval and volatility effect is highly sensitive, where the dynamic process speed varies with the interval. 4.When factors like Jan. & Feb. effect, the day-of-the-week effect or trade volume rate is considered, the asymmetric reverting pattern varies with data frequency interval and market efficiency. 5.A month-interval is recommended for contrarian in TSEC and OTC, whereas a shorter daily strategy is preferred for TAIEX future operation. 6.When contrarian is practiced on a daily basis, the inter-market data transmission and spillover effect on asymmetry can still be a reference although with less strength and speed than on a monthly or weekly basis. When TSEC is targeted the prior TAIEX future return over positive information can serve as a leading indicator. As for TAIEX future investment the explainability (predictability) of return asymmetry over either positive or negative information for TSEC and OTC can be put into consideration.