本文分析NASDAQ交易所於2001年4月9日，將1/16進位的分數報價變更為小數報價前後買賣價差組成份的變化情況。本文主要應用Huang and Stoll (1997) 在考量買賣壓力下的價差分解模型，並以GMM估計法進行參數估計。價差分解模型有二維、三維與買賣壓模型，以價差分解為逆選擇成本、存貨持有成本與委託單處理成本三種價差組成份。 實証結果發現，價差於改變小數報價後顯著地縮小，二維模型實證結果顯示價差縮小幅度與個股交易量大小呈正相關，且改變小數報價後逆選擇與存貨持有成本亦為減少。三因子模型實証結果顯示，大與小市值之逆選擇成本在改變小數報價後為減少的，存貨持有成本則為增加，中市值恰好相反；在考慮買賣壓後，逆選擇成本於小數報價後亦為顯著的減少，存貨持有成本則是增加。 最後，Panel Data模型加入虛擬變數探討與五個影響變數來觀察小數報價前後對價差組成份的影響，實證結果發現，在大與小市值中，平均交易間隔時間與成交量對二因子模型所估計出之逆選擇與存貨持有成本的影響程度在事後是減少的，在中市值中影響是增加的，價差對於二因子模型所估計出之逆選擇與存貨持有成本是影響程度減少；在大市值組中，平均交易間隔時間、成交量與價差對三因子模型中所估計出之逆選擇成本影響程度在事後是減少的，在中與小市值組中影響程度則是增加的；平均交易間隔時間對買賣壓模型所估計出之逆選擇成本和存貨持有成本的影響程度在事後是增加的。 The paper estimate the components of bid-ask spread around the decimalization of NASDAQ on April 9, 2001. We apply Huang and Stoll (1997)’s structure models to decompose of bid-ask spread based on two-/three way and buying and selling pressure models, estimating the components of bid-ask spread through the GMM approach. Furthermore, the bid-ask spread can be decomposed into three components including the order-processing, inventory-holding, and adverse selection costs. We show that the spreads is more decrease after post-decimal pricing. The empirical results of two-way model illustrates that the linkage between the declined level of the spread and their trading volumes are positive relationship. Thus the inventory and adverse selection costs would be decreased during post-decimal pricing. In three-way model, results show that adverse selection costs decreased and inventory holding costs increased after decimalization in the high-/ low- size, and contrary to the mid-size. For structure model considering buying and selling pressure, we find that the adverse selection cost is decreased significantly and the inventory holding cost is increasing after decimal pricing. Finally, we use Panel Data model with dummy variables and five agency variables to test the influence on the components of bid-ask spread during pre-/post- decimal pricing. The empirical results that the inventory and adverse selection costs are estimated by two-way model effect on average trading between time and trading volume is decreasing for high-/low- size samples during post-decimal pricing, but increased in the mid-size. The effect of spreads on the inventory and adverse selection costs for full sample would be decreasing. The empirical results that the adverse selection costs are estimated by three-way model effect on average trading between time, trading volume and spread is decreasing for high-size samples during post-decimal pricing, but increased in the mid-/low- size. The empirical results that the adverse selection costs are estimated by buying and selling pressure model effect on average trading between time is increasing for full sample during post-decimal pricing.