本研究以我國1990年至2007年間總計976檔初次公開發行的股票為研究對象,將此長達18年之研究期間,依承銷制度的不同細分為三個區間,以探討我國新上市櫃股票異常報酬是否因為不同的承銷制度而有所差異;最後再針對短期異常報酬,探討是因為何種原因造成我國新上市櫃股票具有正的短期異常報酬,以及異常報酬是否因為負債比率、發行市場等因素的不同而有所差異。 實證結果顯示,我國新上市櫃股票存在顯著大於零的短期平均異常報酬,但在上市櫃20日後之長期平均異常報酬則不顯著異於零,由此可判斷在排除短期異常報酬後,就不會出現長期異常報酬,說明我國新上市櫃股票存在短期異常報酬是由於承銷價被低估所致。將三種不同承銷方式作比較,發現三個期間的新上市櫃股票短期平均異常報酬均顯著大於零;但是長期績效三者卻不同,可見不同承銷制度會影響新上市櫃股票之長期異常報酬。 迴歸分析方面,在全部樣本中,中籤率與短期異常報酬呈顯著的反向關係,而公司規模與短期異常報酬間呈現顯著之正向關係,此結果符合信號理論;而上市公司的短期異常報酬較上櫃公司小,符合資訊不對稱假說。 The subjects for this study were 976 initial public offerings from 1990 to 2007 in Taiwan. We separate this 18-year-long sample period to three periods by underwriting systems, in order to compare the abnormal returns of initial public offerings under different underwriting systems in Taiwan. Finally, we explore what caused the initial public offerings in Taiwan having positive short-run abnormal returns, and compare the short-run abnormal returns of initial public offerings under different debt ratios, IPO market and other factors. The empirical results show that IPO stocks in Taiwan exist significantly positive short-run average abnormal returns, but the long-run average abnormal returns are insignificant different from zero after issuing 20 days. We can judge that IPO stocks would not exist long-run abnormal returns when the short-run abnormal returns were excluded, which means IPOs exist short-run abnormal returns are caused by the underpricing of offer price. To compare three kinds of underwriting systems, the short-run average abnormal returns are all significantly greater than zero; but the long-run performances of three periods are different, it means various underwriting systems would affect IPOs’ long-run abnormal returns. All the samples in regression analysis, odd ratio and short-run abnormal returns show significant negative relationship, while the size of the company and short-run abnormal returns show significant positive relationship. The results confirm to signal theory; and short-run abnormal returns of the listed companies smaller than the Otc companies confirm the information asymmetry hypothesis.