本研究選取94支台灣認購權證為研究樣本，研究期間自1997年9月4日至2002年6月30日止，透過常態性檢檢定、異質性檢定、ARCH-LM檢定及不對稱檢定以確認符合異質性及不對稱性之樣本，並依研究期間長達六個月以上不變趨勢區分成多頭、空頭及盤整期走勢，依類股區分成電子及傳統類股，探討認購權證之訂價行為。主要結論如下： 1.僅36支權證標的股報酬具有異質性及不對稱性。 2.透過三種價格誤差衡量指標及迴歸分析後，發現二個方法皆一致顯示隱含波動性具有較佳的預測能力，若僅考慮時間序列模型比較，則無法明確指出何一波動模型具相對的優劣。 3.四種波動性模型下之理論價格與市場價格存在顯著差異，隱含波動性相對於市場價格有高估的現象；GARCH、EGARCH與歷史波動性皆相對於市場價格有低估的現象；Wilcoxson檢定法發現隱含波動性相對其它三種模型有改善錯價能力，且二類股訂價表現無顯著差異；若不論及隱含波動性時，則各模型間未有顯著改善錯價能力，且二類股訂價表現僅EGARCH相對GARCH有顯著差異，亦發現整體顯示多頭期有較佳的訂價表現。 4.符合變異數異質性及不對稱性之36支認購權證普遍不具有學習機能，的確是造成價格誤差原因之一；深處價外之權證，由於投資人認為其反彈機率較常態分配機率高，是造成價格誤差原因之二；隱含波動性迴歸係數近8成顯著為正值，驗證該波動愈高愈加大錯價程度，其餘三種時間序列波動各約3成係數顯著為負，驗證波動愈高有助於改善錯價，故整體波動性不佳則是錯價原因之三。 This study selects 94 warrants of Taiwan dated from 1997/9/4 to 2002/6/30. We use normality test, heteroskedastic test, ARCH-LM test and asymmetry test to confirm samples are coinciding with heteroskedasticity and asymmetry. We furthermore discuss pricing behavior through different index trends and categories. Summarized results are as follows： 1.Only 36 warrants shows heteroskedasticity and asymmetry. 2.Implied volatility has the best prediction power through two approaches, and no definitely significant results when taken into account the time series models only. 3.Significantly difference exists between theoretical price and market price under four volatility models. Price of implied volatility is a little higher than market price while the other models are reverse. Wilcoxon test finds that the implied volatility can improve pricing error than the other three models. No obvious improvement to be found among three models without implied volatility. Relative significant difference in two-category exists between EGARCH to GARCH models, and bull-period has the best pricing performance. 4.No learning organism of Taiwan‘s warrants is the first reason of pricing error. Investors think out-of-money warrants having higher rebound probability than normal is the second reason of pricing error. Coefficients of implied volatility are 80% significantly positive, this approves that higher volatility causes higher pricing error. While the other volatilities are 30% significantly negative, this approves that higher volatility is helpful to correct the pricing error, these are the third reason of pricing error.
