Algorithmic Trading and Market Quality: International Evidence of the Impact of Errors in Colocation Dates

Abstract

This paper examines evidence on colocation dates and their impact on market efficiency. International colocation dates can be sourced from a number of avenues including: [1] an 'exchange's news announcements and reports, [2] news media, and [3] by direct communication with the officers of an exchange. Boehmer et al. (2021) report colocation dates based on [1] and [2] and do not reference prior work that reports colocation dates that are primarily sourced from [3]. The consequence is that the discrepancies between prior studies and Boehmer et al. are significant and economically meaningful: the errors average 12.75 months with the largest being 46 months. This paper documents these discrepancies and provides evidence of how these differences in colocation dates matter for evidence of their impact on market efficiency. We document that colocation is not exogenously determined and highly correlated with the changes in market quality measures, counter to claims of Boehmer et al.