The Post-IPO Flipping by Institutional Investors and The First-Day Return：Predictive Factors for Long-Run Returns

本研究以台股初次公開上市櫃公司IPO後首日股價報酬熱度及機構投資人首月脫售持股情況兩指標，試著預測IPO後一年期股價表現。本篇論文蒐集自2010年起至2014年3月31日止252筆在台灣初次上市櫃公司資料，並將其依首日報酬熱度由低至高依序分為Cold、Cool、Hot及Extra-Hot四種群組。研究結果顯示，Cold IPOs長期傾向擁有較佳超額報酬，而首日報酬熱度最高的Extra-Hot IPOs長期則表現最差。也證實了IPOs股票折價承銷理論中所提，公司若在IPO初期出現較大幅度的折價讓利，長期將導致公司因降價求售股票而遭受價值損害。 另外，本文也以三大法人脫售持股比例為分組，探討法人是否具有長期股價預測能力。過去一些國內外文獻皆以全部樣本直接討論法人脫售行為對長期績效表現的影響，本研究進一步將252筆公司資料分為：(1)初次上市櫃前曾在興櫃市場掛牌交易及(2)上市櫃前未曾在興櫃市場掛牌交易 兩群組。結果顯示，機構投資人只有在對IPO個股具有優勢資訊時(曾在興櫃市場交易)，才能顯著對公司優劣做出判斷，進而對長期績效表現做出預測；反之，若是未曾興櫃之IPO個股，機構投資人對其一年期股價表現則無顯著預測能力。本研究最後以首日報酬率、承銷商聲譽、IPO時大盤指數、法人首月脫售持股比例及IPO時個股市值等五大因素，做出IPO一年期超額報酬預測式。結果發現只有IPO時大盤指數、法人脫售持股比例及首日報酬熱度顯著影響一年期超額報酬率。其他變數如承銷商聲譽及IPO時公司市值皆只具經濟意義，對一年期超額報酬並無顯著影響。In this article, we are trying to predict the 1-year excess returns of the IPOs by observing two indicators：The first-day return and the flipping level by institutional investors. We use the 252 IPOs data in Taiwan stock market from 2010 to March 31th, 2014 and divide it into 4 different levels by the first-day return：Cold IPOs, Cool IPOs, Hot IPOs and Extra-Hot IPOs. And it turns out the Cold IPOs are inclined to have better long-run performances on the stock price, while the Extra-Hot IPOs have the worst performances in the 12-month excess returns. This study can prove the mispricing theory：The IPOs will suffer in losses in the long-term because of the underpricing behaviors. This thesis also analyzes the predictive power of the flipping level of institutional investors. We divide our sample data into two different groups：(1) IPOs who have been traded in the dealer market (Emerging Stock Market, ESM) (2) IPOs who traded in the order driven market directly (without the dealer market experiences). It reveals that the predictive power is prominent only if the institutional investors have superior information about the IPOs (with dealer market experiences); In contrast to the first group, the predictive power seems to be weak when the IPOs have no dealer market experiences. In the end of this study, we are trying to construct a regression model with five indicators：the rank of the underwriters, the IPOs’ market capital, the market index during the IPO period, first-day return and the flipping level of institutional investors. We find out that in addition to the IPOs’ market capital and the rank of the underwriters, all other indicators are statistically significant to predict the 1-year excess returns of IPOs