Comment on: Price Discovery in High Resolution

Abstract

The microstructure literature comprises a rich set of papers that seek to understand pricing dynamics at a granular level, commonly exploring the joint dynamics of bids, asks and last sale prices. Its focus is on identifying innovations in prices and separating permanent price impacts from transient effects. Hasbrouck (1995) provides a tool that has been extensively utilized in the literature to examine these dynamics in many different market contexts over the last two decades.1 However, the evolution of markets over this period, most notably the exponential growth in the volume of data and the increasing importance of trading speed has made the application of Hasbrouck’s (1995) method and other related tools discussed in Hasbrouck (2018) more computationally and econometrically challenging. Hasbrouck (2018) offers a new approach to help overcome these challenges.

In this comment, we briefly describe the evolution of markets and detail the challenges that these changes create for microstructure researchers and highlight the solution that Hasbrouck (2018) offers for these problems. We survey the literature that uses linear multivariate time-series models to understand high-frequency markets. We focus on three examples from the literature to discuss how estimation constraints have affected their modelling choices, describe the potential drawbacks of these choices and how Hasbrouck’s (2018) method can alleviate these constraints. We deliberately select papers that cover different asset classes: cash equities, fixed income and equity options. We hope that our discussion will help provide guidance about the costs and benefits of different modelling choices for future researchers confronted with a variety of methods to answer related research questions. We conclude by considering the implications of Hasbrouck’s 2018 paper for the current policy debate on market data costs.