在沒有市創者的設置下，台灣股市的雜訊交易是否為重要的流動性來源？乃為值得探討的議題，本文將以台灣股票市場為對象，研究分析市場流動性與雜訊交易的關係，從而觀察分析其對市場流動性的衝擊。雜訊交易在資訊不對稱相對低與高時較大，而在資訊不對稱相對中等時，雜訊交易相對較小；交易量亦有相似的趨勢，由於一般流動性交易者於資訊不對稱較高時交易意願較低，較高的交易量所需之流動性較可能來自於雜訊交易。資訊不對稱相對中等及高時，市場深度相對較高，若依五分鐘之資料，資訊不對稱於中等時有顯著最高的市場深度。此證據似乎顯示較高的雜訊交易在資訊不對稱較高時可能較有助於市場的深度。資訊不對稱相對低時，有較高的報價價差，有效價差及實現價差。在資訊不對稱低時，較高的雜訊交易似乎對此交易成本的降低並無幫助。另外，當資訊不對稱較高時，有效價差及報價價差亦較高，但實現價差則較低。較高的雜訊交易於此時較可能有助於流動性的提供，降低實現價差之交易成本。 The Taiwan stock market has relatively high trading volume and turnover rate without designated market makers. How the market provides liquidity would be an interesting issue. This study examines whether noise trading plays a role of providing liquidity in the market and also investigate the relationship between noise trading and liquidity under different degree of information asymmetry. The evidence shows that when information asymmetry is relatively low or high, noise trading is high. However, noise trading is relatively low when information asymmetry is modest. Trading volume exhibits the similar pattern. Since discretionary liquidity traders tend to be unwilling to trade during the period of higher information asymmetry, liquidity required for higher trading volume is more likely from noise trading. In addition, when information asymmetry is relatively modest and high, market depth is higher. Based on 5-minute data, market depth is significantly higher when information asymmetry is modest. The evidence seems to exhibit that higher noise trading might help market depth during the period of high information asymmetry. Moreover, the results exhibit higher effective spread, quoted spread, and realized spread when information asymmetry is relatively low. It seems that higher noise trading might not be able to help reducing transaction costs in the market with lower information asymmetry. However, when information asymmetry is relatively high, effective and quoted spreads are also higher, but realized spread relatively low. The evidence seems to imply that higher noise trading might help liquidity providing in terms of reducing realize spread when information asymmetry is relatively high.