Crossing Network Trading and the Liquidity of a Dealer Market: Cream-Skimming or Risk Sharing?

Abstract

This article investigates the relationship between the trading activity of a crossing network (CN) and the liquidity of a traditional dealer market (DM) by comparing data from the SEAQ quote-driven segment of the London Stock Exchange (LSE) and internal data from the POSIT crossing network over two 6-months' periods. This exploratory study, which is the first one to use internal CN data, provide new insights into market competition between traditional exchanges and alternative trading systems in Europe, and opens further research leads. Based on a cross-sectional analysis in a sample of UK mid and small cap stocks, the findings support that CN-trading does not significantly increase adverse selection and inventory risk on the central market. It appears that the competition between market makers strengthens with the CN activity, and CN-trading gives dealers a risk sharing opportunity that leads them to improve quotes. Crossing Network Trading and the Liquidity of a Dealer Market: Cream-Skimming or Risk Sharing? During the last decades, new electronic trading systems, designated by the SEC as Alternative Trading Systems (ATS),1 have developed all around the world, in response to the need expressed by institutional investors as well as broker-dealers2 for alternative systems that would help reduce the cost of trading. These ATSs divide in two categories: Electronic Communication Networks (ECNs), which work as anonymous electronic order books, and Crossing Networks (CNs), which cross unpriced buying and selling interests. Like ECNs, CNs generally promise anonymity and lower transaction costs, but the main difference with ECNs is that participants do not enter the prices at which they wish to trade, so that no price discovery takes place in a CN. Instead, at designated cross times, interested buyers and sellers are matched and the price at which the trades execute is taken from an exchange. This price can be the central market mid-quote, or, in some cases, the preceding closing price or the volume-weighted average price over some period. In a CN, trades execute with no market impact, yet execution is not guaranteed. In such, CN address the needs of a certain type of traders, ready to sacrifice immediacy and execution guarantees so as to obtain lowcost execution. ECNs have gained substantial market share in trading volumes in the US, since the opening of Instinet3 in the 70s, and now account for approximately 40% of the NASDAQ trading volume.4 As for CNs, the main ones currently operating in North America are the Reuters’ Instinet Crossing Network, ITG’s POSIT5 and the New York Stock Exchange’s afterhours Crossing Network, the most prominent being POSIT. POSIT developed quickly in the United States. 35 millions of shares are presently traded daily on this system, which represents around 2,5% of volumes. The competition coming from these new trading facilities have changed the structure of financial markets, and probably also the role of intermediaries on these markets. The implications for liquidity are of much interest for academics, regulators and investors. Several articles have already addressed this issue by using ECN data, but no work has been done yet with CN data. This paper focuses and provides new evidence on the consequences of the trading activity of a CN, by testing market data from the London Stock Exchange dealer market segment (SEAQ6) and private data from the POSIT European crossing network.