Essays on institutional trading and order flow competition in equity markets

Abstract

This dissertation examines order flow competition and institutional trading in modern global financial markets. It consists of three essays addressing issues on the performance drivers of an alternative trading platform, the impact of HFT activity on institutional execution cost, and broker trading behaviour and market share in the primary market. The empirical findings in these essays provide valuable insights into a number of market participants, such as sophisticated and institutional investors, brokers, regulators, exchange market operators and academics. The first essay, presented in Chapter 4, examines the determinants of Chi-X’s market share in a global context. The introduction of alternative trading venues that compete with listing exchanges has been brought about by regulatory reform and technological advances over the past decade. This essay is the first study into the performance drivers of a prominent alterative trading venue (Chi-X) in an international context. The findings in this essay contribute to the understanding of factors that influence a trader’s order routing decision and ultimately a venue’s market share. The results indicate that Chi-X's market share is negatively related to trading fees and latency relative to the primary exchange, while differences in adverse selection risk have an immaterial effect. Chi-X's market share is found to be positively related to liquidity relative to the primary exchange, measured by bid-ask spread, depth and percentage time that it displays best bid or offer. Venue market share is negatively related to order-to-trade ratio and positively related to average trade size, which suggests a trader’s preference to interact with natural and accessible liquidity. Furthermore, trading tends to concentrate on the primary exchange during market downturns and when volatility is high, while tick constraint stocks tend to trade more on Chi-X to avoid queuing at the touch on the primary exchange. Moreover, there is an improvement in market liquidity in regions where Chi-X was introduced during relatively normal market conditions. The second essay, presented in Chapter 5, examines the presence of High Frequency Trading (HFT) around large institutional trades and their price impact on NASDAQ and NYSE. The findings contribute to the understanding of HFT around institutional trades and its effect on realised execution costs. The focus on institutional price impact adds to the current literature around HFT's impact on conventional market quality metrics, such as spreads and depth, while addressing concerns that HFT adds to the difficulty of executing institutional orders, despite its perceived association with liquidity provision. The results in this chapter indicate that the impact of HFT participation on institutional execution costs vary between buy trades and sell trades. This essay also documents that institutional buy trades present higher execution costs, when the level of HFT participation is high. In contrast, institutional sell trades could be executed at better prices with a high level of HFT participation. The contrasting results can be attributable to the difference in the nature of HFTs and their respective ability and costs to short sell. Moreover, there is no evidence that HFT activities are beneficial or detrimental for the largest five percent of institutional trades. The third essay, presented in Chapter 6, studies the trading volume and market shares of brokers surrounding Seasoned Equity Offerings (SEOs) in the Australian primary market based on a unique broker ID dataset provided by ASX. Further regression analysis is conducted in this essay to examine the key determinants of the behaviour of affiliated and unaffiliated brokers around SEOs. The findings contribute to the understanding of broker affiliation around SEOs as well as providing insights into the key factors that influence a broker’s market share. The results suggest that broker affiliation has a significant impact on broker trading volume and market share on both the announcement days and issuance days. However, there is no significant evidence showing that lead managers or co-managers outperform other underwriters. Rather, the results indicate that co-managers are not expected to gain any abnormal volume or market share. Moreover, broker reputation, market capitalisation and relative offer size of the offering firm are shown as the primary characteristics influencing broker performance.