本研究以Granger and Terasvirta (1993) and Terasvirta (1994)提出之非線性平滑移轉(Smooth Transition Regression, STR)模型，自非線性角度出發，找出門檻及平滑移轉參數，進而探討IPO定價過程。以2005-2010年間臺灣證券交易所(Taiwan stock exchange, TWSE)與臺灣櫃檯買賣中心(GreTai Securities Market, GTSM)公佈之初次上市櫃公司為研究對象，深入研究發現： 一、以掛牌前一季之每股盈餘做為門檻變數，探討門檻值上下區間分別對承銷價差(ΔP)與首日收盤價差(IR)進行平滑移轉模型檢定，發現承銷價差之結果在每股9.7595元時發生平滑移轉，但門檻值與轉換速度皆呈現不顯著；而首日收盤價差之結果在每股盈餘14.6601元時，發生平滑移轉，且門檻效果顯著。 二、以承銷價差為應變數，發現高於門檻值(每股9.7595元)時，與圈購前一週之週報酬對承銷價差有負向顯著影響，初始價格較接近掛牌價；主辦承銷商之承銷市占率、圈購前第二週之週報酬對承銷價差則有正向影響。 三、以首日收盤價差為應變數，發現變數之一的承銷價差對其有正向影響，符合過去文獻中兩觀點，1.「贏家的詛咒」理論，發行公司與承銷商為了吸引及保障沒有資訊的投資人來參與IPO市場，必須在承銷價上給予折扣。2.Benveniste and Spindt (1989)指出券商會以折價方式來補償機構法人所給予的私有資訊的論點。 四、在承銷商訂定詢圈區間時，公開資訊大多被併入承銷價中，除高盈餘公司較易產生不充分效率定價的狀況發生；然而，臺灣IPO股票在掛牌上市後，實證發現與盧正壽和劉維琪(2005)在臺灣市場發現一致，即本質不好公司模仿本質好的公司之定價策略產生混合均衡情況，使最後贏家往往是機構法人及專業投資人。 In the study uses Smooth Transition Regression Model proposed by Granger and Terasvirta (1993) and Terasvirta (1994) to investigate whether the earnings per share (EPS) before the IPO has nonlinear impact on the efficiency in IPO pricing. To verify the nonlinear impacts between (1) the efficiency of IPO pricing process, whether the information to be incorporated into the price update between initial price range and offer price (2) the initial return whether the information to be incorporated into the spread between offer price and first-day closing price (3) EPS before listed whether to exist smooth transition effect under different levels of transition variables. According to the empirical research, when the EPS is the transition variable that has two key findings. First, in the filing period, IPO pricing is almost efficient in the lower EPS companies. Second, the IPO firms that high quality firms and low quality firms take the same pricing strategy, which is consistent with the predictions of Lu and Liu (2005). In summary, the final winner is often the corporations and professional investors in the IPO pricing process.
