Find us on Facebook Find us on LinkedIn Follow us on Twitter Subscribe to our YouTube channel Instagram

Modelling Order Flow, Price Impact, and Market Resiliency in the Australian Stock Market

Kevin Lo

This dissertation examines the effect of trading on prices and price impact costs in an empirical and quantitative manner by using a number of econometric and computational models. Four research studies are described: the first three of which examine price impact costs and the dynamic relationship between price changes and order flow, and the fourth considers the context of contrarian portfolio strategies. By modelling order and transaction data, these studies attempt to bring greater insight into the dynamics between prices and order flow activity, and illustrate the importance of managing price impact costs to researchers and market practitioners. The first research study extends the existing literature on the price impact of trades. This study begins by estimating the functional relationship between price change and a number of variables that measure trading activity and trading direction, including order imbalance, trading volume, and the ratio of trading volume to liquidity supply. The main findings are that the relationship between price change and trade direction variables is nonlinear and concave, as price change increases with trading activity but at a decreasing rate. This study then specifically examines the dynamic relationship between price change and order imbalance by applying a bivariate vector autoregressive (VAR) model. The result from this model supports the claim by Hasbrouck (1991a) that trading activity not only affects price instantaneously, but also has persistent impact on future prices. The full price impact of trades arrives with protracted lags.

The second research study examines the effect of large trades on future order submissions.This study suggests that trades not only have persistent impact on prices, but also influence the composition of market and limit orders in the order flow. This study contributes to the literature by measuring market resiliency, one of the three dimensions of liquidity proposed by Kyle (1985) and the one that so far has received the least attention in the literature. This study firstly measures market resiliency by investigating the changes in price and liquidity after the submission of aggressive orders. A Bayesian multivariate time series count data model is then applied to quantify the dynamic relationship between order submissions and changes in price and liquidity. This study provides empirical evidence that suggests the limit order book is resilient, as temporary imbalances in the liquidity supply caused by trades are shown to attract new orders which restore both prices and liquidity. The third research study introduces market simulation as a rather innovative method to estimate the price impact and execution costs. The simulation models the order flow characteristics of individual stocks in the context of the trading mechanisms of the Australian Stock Exchange. This method offers a number of benefits, including the ability to directly measure the price impact of trade packages, which are otherwise difficult to estimate empirically without access to proprietary data sets, and the ability to create both resilient and non-resilient markets for comparison. The results suggest that the price impact and execution costs are smaller in a resilient market than in a non-resilient market. Additionally, as the resilient simulated market satisfies the no-arbitrage condition suggested by Huberman and Stanzl (2004), the results suggest arbitrage opportunities are eliminated if the market is resilient.

The final research study extends beyond the effect of trading on prices and considers the context of contrarian portfolio strategies. This study aims to contribute to the literature in three different ways. Firstly, the profitability of contrarian strategies on an intra-day level is examined. Secondly, the relationship between the degree of price reversal and order flow activity is analysed. Finally, this study investigates how order flow information improves contrarian profits and then assesses the viability of contrarian strategies which exploit order flow imbalances. Empirical evidence from this study suggests that although order flow based contrarian strategies generate positive revenues, the net profits after transaction costs are not economically significant.


Richard Coggins