Abstract
A widely cited working paper from the 2000s, now published (Jones, 2026), documents a substantial decline in trading costs and a positive relation between bid–ask spreads and excess returns from 1900 to 2000. We update the analysis and find that over the next quarter century, turnover surges, spreads decline sharply and subsequently cluster near zero, and excess returns become negatively related to spreads. Additional analysis attributes the reversal of the excess returns relation to distortions in the time‐series analysis created by the sudden reduction and clustering of spreads. We find similar distortions using the Amihud illiquidity measure and in simple cross‐sectional analyses, which suggest the effects we highlight might be pervasive.