Clustering in US Stock Prices after Decimalisation
利用2001年美国股市从分数报价转为十进制报价的自然实验，发现股价在5美分和10美分处显著聚类，且程度远超传统解释，暗示人类对突出数字的偏好。
Abstract Early in 2001, US equity markets transitioned from trading in discrete price fractions to a smoother decimal format with a tick size of one penny. Theory suggests in an unconstrained world, stock prices should be distributed uniformly, particularly if the cost of defeating time priority is low. This regime change provides a natural experiment to test whether investors prefer to trade at particular price points even when their choices are essentially unconstrained by regulation. Instead of uniformity, we find widespread evidence of price clustering at increments of five and ten cents (nickels and dimes); the overall magnitude of clustering is double in scale of what is otherwise expected. Previous studies which documented clustering around even‐eighths argued that these patterns were a rational market response to trading impediments. We report consistent findings, but also find that the overall level of post‐decimalisation clustering is far more extensive than is reasonably explained by prior hypotheses. The evidence instead suggests a more fundamental human bias for prominent numbers as discussed in the psychology literature. Contrary to previous studies, we find no difference in price clustering, ceteris paribus, between the Nasdaq and NYSE after decimalisation. Should regulators choose to revisit the notion of tick size, our evidence suggests that for many stocks there would be only minor impact between the transaction prices that prevail now and those that would occur if the tick size were increased to five cents.