本文首先以BDS統計量檢定發現OTC股價具有非線性相依。其次,應用非線性的波動轉換GARCH模式來描述股價與波動行為。實證結果發現(1)OTC股價的變動有條件異質性,而且條件波動呈現高度的持續性,顯示目前的波動資訊可以預測未來的波動。(2)條件波動具有不對稱性與不對稱性的反轉現象。(3)條件波動具有週日效果。此外,本文設定三種不同波動函數型態的GARCH-M模型,探討報酬與風險的關係。實證結果發現條件平均數與條件變異數、條件標準差與對數變異數沒有顯著的正向關係,隱含變異數與標準差或對數變異數,並無法適切描述投資者所考慮的風險,建議研究其他衡量風險的方法。
In this paper, BDS statistic is applied to test the nonlinearities in Taiwan OTC stock prices. Empirical evidences show that stock prices behave nonlinearly. Therefore, we use a nonlinear volatility-switching GARCH model to describe the behavior of stock prices. We find that the change of stock price is heteroskedastic and the conditional volatility is highly persistent. This indicates that current volatility can predict future volatility. Second, there exist the asymmetry effect and the inversion of asymmetry effect of the conditional volatility. Third, The day-of-week effect exists in the conditional volatility of stock return. Furthermore, we use three alternative specifications, linear, square root, and logarithim, for GARCH-M models to discuss the relationship between return and risk. We find an insignificantly positive relation between conditional expected return and the three alternative specifications of volatility measures. The results show that we may study the other risk measures which are used for consideration by investors.