The tyranny of averages.
Laurent Fournier, manager, business analysis and statistics at NYSE Euronext, examines the reduction of average trade sizes on the European equity markets.
A paradox is a statement that is seemingly contradictory or opposed to common sense. Defined this way, the assertion that the recent downward trend in trade sizes equates to a proportionate reduction in liquidity certainly looks like a paradox. Although many have highlighted a reduction in average trade sizes since 2008, NYSE Euronext can demonstrate that the volume of fully executed large orders on our lit books has reduced less than this overall decline in trade sizes. Consequently, we hope to shed some light on the value of order size.
Exploding the paradox
Trade size is the result of the order matching method. The size of the partial output, i.e. the trade, is of equal importance to the size of the fully executed order. If one needs to trade 1,000 shares, whether the order is executed immediately in one transaction or through several transactions at the same price and time, the transfer of risk has occurred, albeit the resultant information to the market is different. The order can generate a number of simultaneous trades if there are several orders at the same price on the opposite side of the book. This is well illustrated by the difference between the average trade size in September 2010 on French CAC40(r)/ Dutch AEX(r) constituents at [euro]8,900 versus the average executed order size at [euro]15,100 (with an average of nearly two trades per executed order). This is shown in Figure 1.
From January 2008 to September 2010, the average trade size decreased by 63% from [euro]23,900 to [euro]8,900, whereas the average executed order size only fell 55%, from [euro]33,890 to [euro]15,100. Since January 2008, the divergence between the relative ratio of executed order to trade size has increased significantly, from 42% to 70%, as the number of small orders has increased and their size has further reduced. This trend is mainly due to market fragmentation, execution benchmarking and the extensive use of algorithms. However, the executed order distribution has not been completely diluted, as the number of very large executed orders has remained stable. As Figure 1 illustrates, for CAC40/ AEX constituents, more than 5% of the largest executed orders were above [euro]44,125 (and generated four trades on average) in September 2010, compared to [euro]52,840 in January 2008.
Furthermore, 1% of the largest executed orders were above [euro]142,550 (and generated seven trades on average), accounting for more than 26% of the total daily transaction value in September 2010, compared to [euro]155,250 in January 2008.
This resilience of the proportion of large executed orders can mainly be explained by the attraction to a very strong market depth at the best limit in our lit books. This displayed market depth at best limit remained on average above [euro]35,000 during September 2010, accounting for more than three times the average trade size.
If we concentrate on the executed order size analysis of the auction (opening and closing), then the average during September 2010 was [euro]120,100, with 5% of the largest executed orders above the Large In Scale threshold ([euro]500,000 as defined by MiFiD) and 1% of the largest above [euro]1,500,000. This demonstrates the strong added value of the auction mechanism on price-forming platforms.
In conclusion, it is important to focus on order size as well as on the average trade size trends. The decline in average trade size is not the cause of liquidity contraction on the lit books but mainly the result of greater frequency of small orders while large orders remained the same. The clustering of those small orders at the best limit on NYSE Euronext has three benefits:
The notion that opaque trading is the only way to execute large orders, simply because lit books demonstrate smaller trade executions, is part of the paradox; the reality is more diverse.