以往文獻指出市場上充斥著無法由效率市場假說加以合理解釋的「異常現象」,因此80年代出現一個新理論-「雜訊交易理論」,De Long et al.,(1990)認為市場上普遍存在之雜訊因素是影響股價的重要原因之一,但討論雜訊來源及其對股價影響性之相關研究相當有限且研究結論也相當分歧,因此本文之研究目的為針對雜訊因素作整合性的探討,在股價資料可能存有時間序列相關之考量下,應用變異數比率法,進行融合時間序列橫剖面迴歸模式估計,以驗證台灣股市雜訊交易之存在性、來源及其對股價之影響性。實證結果證實國內股市確實存在著雜訊交易行為,不同產業間存在著不同的雜訊來源,且不同的雜訊因素對股價亦有不同程度之影響性。
There is considerable evidence that stock price can diverge significantly from fundamental value. It implies that many investors do not follow fundamental values to buy and hold the market portfolios. They often pick stocks through their own research or sentiment. Therefore, DeLong et al., (1990) proposed ”Noise Trading Theory”, which pointed out noise factors - investors' sentiment can affect stock prices. However, there are few studies concerning noise trading in which the coverage of noise factors and results are not quite consistent. The purpose of this paper is to incorporate relevant noise factors in the literature into our empirical study in order to verify the existence, sources and impact of noise trading in Taiwan stock market. The variance ratio test was incorporated with the estimation of Time Series Cross-Section Regression model to account for the possible serial correlation for the data series. The results confirm the existence of noise trading. The aspects of noise trading and their influence diverge for assorted industries.