本文主要探討新上櫃股票(IPOs)期初累積報酬中所隱含私有資訊交易。我們以台灣新上櫃股票市場為研究對象，利用私有資訊交易模型來估測IPOs上櫃後所含的資訊訊息，取代一般文獻中以間接的代理變數來衡量IPOs期初報酬中資訊不對稱的狀況。同時並以非涵蓋模型(Non-nested model)檢測，私有資訊因子的單一廻歸模型與代理變數的複廻歸對IPOs期初報酬的解釋程度，何者為優勢模型？實證結果發現台灣新上櫃股票市場確實存在期初報酬，在IPOs期初報酬的廻歸分析模型中，私有資訊交易模型中所求出代表資訊不對稱的因子，與新上櫃股票期初報酬的模型解釋力達顯著水準；而一般間接代理變數中以中籤率和上櫃前股東人數之解釋能力為顯著。此外，我們將以一般代理解釋變數的廻歸模型，與私有資訊因子的廻歸模型作非涵蓋模型的檢定，比較何為優勢模型?證據顯示T統計量顯著拒絕代理變數模型為優勢模型之虛無假說；反之我們再進行相對應的檢定，發現T統計量並不拒絕資訊交易模型為優勢模型之虛無假說。因此，我們推論本研究所推導之私有資訊因子的模型確實優於傳統的代理變數複廻歸模型。 In this thesis, I use a new empirical model, Informed Regression Model (IRM) to estimate the probability of trades generated by privately informed traders in the IPO market of Taiwan OTC. Meanwhile, I also follow the literatures on the IPO subjects, using Explicit-Variable Regression Model (EVRM), to find out the factors affecting the initial returns of the IPO securities. Appling the Non-nested Model testing method, I compare the IRM with the EVRM and find several empirical results as follows: (1) Evidence shows that the IPO market of Taiwan OTC existing initial returns. (2) The factors affecting the initial returns of the IPO securities are the drawing percentage and the number of shareholders before IPO. (3) The empirical results show that the factor of the informed traders has a significantly negative impact on the initial return of IPOs. (4) The result of the Non-nested model testing shows that the IRM is superior to EVRM.
