價格群集(price clustering)指金融資產的報價或成交價格之尾數特別集中在某些數字(偶數、整數)的傾向，嚴重的價格群集可能代表市場缺乏效率或價格遭受扭曲。金融資產報價中均有最小升降單位的規定，而使得價格均為升降單位的倍數，因此在觀測價格尾數時，升降單位的大小將影響到價格群集於何種數值，本研究觀察升降單位(tick size)之變動對價格群集的影響。多數的財務模型都事先假定金融資產價格為連續的，並不會特別發生在某特定數值，實際上因為有升降單位的規定，連續定價的假設便難以成立，才有價格群集的發現。 本研究發現在台灣股票市場中不論是何種價格區間或升降單位如何變動，仍舊存在價格群集的現象。此結果與Hameed and Terry (1998)之結果相同，任何一種價格區間內都存在著明顯的價格群集情況，而價格尾數為0與5的出現頻率皆高過於其餘價格尾數的出現頻率，而價格尾數為0的出現頻率又高過於價格尾數為5的出現頻率。本研究並發現價格群集與價格水準呈現正相關，價格水準越高價格群集程度越高；價格群集與交易量呈現負相關，交易量越大價格群集程度越低；而價格的波動程度則與價格群集沒有明確的關係；升降單位的變動可能會影響到價格群集的改變。此結果大致支持Harris(1991)的協商假說(Negotiation hypothesis)與Ball et al.(1985)的價格解析度假說(Price resolution hypothesis)。 Price clustering means the last digit of the quote of financial asset or trade price has a tendency of centering on certain numbers (even numbers, round number etc.). Serious price clustering could represent the market lacks efficiency or the price has been distorted. There are usually rules of the smallest tick size in financial asset quote so that the price can be the multiples of the tick size. Therefore, when observing the changes of the tick size, the size of it will affect which value the price clustering falls. This research is to observe the effects the variation of the tick size on price clustering. Most financial models presume financial asset price is constant so that the financial asset price will not fall on a certain number. In fact, because there are rules of tick size, the assumption of constantly fixed price is hardly existed; therefore, price clustering was discovered. This research finds that in Taiwan’s stock market, no matter what kind of price intervals or changes of tick size, price clustering still exists. This conclusion is equivalent to that of Hameed and Terry(1998); the frequency that the last digit of the price is 0 and 5 is higher than any other numbers. And the probability of the last digit being 0 is higher than that of 5. This research also finds that price clustering and the price level are positive correlated; the higher the price level, the more intense the price clustering is. On the other hand, the price clustering has negative correlation with the trade volume; the larger the trade volume, the less intense the price clustering is. Moreover, the price volatility and price clustering are not necessarily related while tick size might affect price clustering. These conclusions support Harris’ Negotiation hypothesis (1991) and Ball et al.’s Price resolution hypothesis(1985).
