Jour Fixe - Frequent Batch Auctions and Informed Trading

Frequent Batch Auctions and Informed Trading

This paper studies liquidity provision by competitive high-frequency trading firms (HFTs) in a dynamic trading model with private information. Liquidity providers face two sources of adverse selection risk: Risk from trading with investors with superior information and risk from trading with other HFTs that respond to new public information more quickly and engage in latency arbitrage. The impact of the two different sources of risk depends on the details of the market design. We determine closed-form representations of equilibrium transaction costs in continuous limit order book (CLOB) markets and under frequent batch auctions (FBA). Our key measure of inefficiency, expected markup flow, captures markups that investors expect to pay per unit of time and allows for a detailed comparison between the two microstructures. In the absence of informed trading, FBA dominates CLOB in terms of markup inefficiencies, just as in the original paper by Budish et al. (2015). Surprisingly, this result does no longer hold with privately informed investors. We show that equilibrium prices in the FBA design allow liquidity providers to charge strictly positive markups and earn profits – even in case of risk neutrality and perfect competition. However, a slight variation of the FBA design in which traders may submit orders conditional on auction excess demand removes the inefficiency, yielding zero markups in equilibrium and thus lowest transaction costs for investors.