台灣證券交易所為了提升市場資訊透明度，自2003年1月2日起將報價揭示從最佳一檔買賣報價改為揭露最佳五檔買賣報價。本文主要應用Sandas (2001) 的限價委託簿結構模型，以GMM方法對台灣證交所委託簿資料進行實證。研究樣本則分為揭露五檔的初期與四年後的近期兩區段。為了觀察市場透明度增加後市場流動性、波動性與交易成本間的關係，本文採用Panel Data模型來進行分析。 參數估計結果發現，市價買賣單委託數量符合市價單數量分配的特性，非常接近實際限價委託簿情況；委託單擲交成本為負，反映交易者理性的投單行為；市價單價格衝擊皆為正，隱含逆選擇成本為正值；另外，與委託單擲交成本及逆選擇成本相比，公共資訊對於股價變動的影響比較小。 實證顯示，市場透明度提升後，市場委託買賣單數量是減少的，委託單擲交成本是降低的，逆選擇成本是降低的，顯示台灣證交所實證揭露五檔措施能有效改善資訊不對稱的程度。此外，交易者較偏好擲交台灣卓越50成份股的委託單，提升其流動性，亦降低資訊不對稱所產生的逆選擇程度；而T50成份股委託單處理成本的負值程度較高，顯示交易者擲交T50成份股的委託單，其監視成本較高。 在以Panel Data模型進行實證結果顯示，透明度提升後委託單擲交成本是降低的，而個股流動性對委託單擲交成本的影響則為負；市場波動性對委託單擲交成本的影響則為正；個股波動性對逆選擇成本的影響則為負。 In order to elevate market transparency of the information, Taiwan Stock Exchange Corporation (TSEC) started to disclose the best five bids and asks quotations in January 2, 2003. We apply the structural model of Sandas (2001) and use the data of limit order book from TSEC. The empirical mothodology is based on GMM method and the research sample is divided into two different periods. Finally, we also use Panel Data model to observe the relationship between estimated parameter, volatility and liquidity. The empirical results represent that market order flow characterizing the distribution of market buy/sell order quantities. The estimated order processing cost is negative, but this could reflect rational behavior of orders by traders. We also find that the price impact of market orders is positive, presents adverse selection cost is positive. Furthermore, we show that the improvement of market transparency not only decreases the order submission and adverse selection costs but also reduces the level of information asymmetry. When traders submitted the order of Taiwan top 50 constituent stocks, the monitor cost of limit order book is higher. Finally, the results of empirical study on Panel Data model, we find that the individual stock liquidity impact on order submission cost is negative. Then, the market volatility impact on order submission cost is positive. However, the stock volatility impact on adverse selection cost is negative.
