The debate about which securities market trading structure will best enable Europe to achieve its economic goals continues. Three issues of crucial importance are fragmentation, internalisation, and transparency. A brief summary of the current state of research into how these phenomena affect market performance and best execution is provided here.
Fragmentation
Apart from a few notable exceptions, there is now a wide consensus that it is beneficial to have competition between exchanges, and other types of market participants, for the trading in a particular security. Seven main and closely related themes have been raised in the debate over whether all orders in a particular security should be consolidated on a single trading system, or whether orders should be allowed to fragment to competing trading systems. These themes are discussed below.
Competition between exchanges and other types of trading systems
Such competition keeps the services that exchanges and trading systems offer at a low price, including transaction, clearing and listing services, and provides an incentive for them to innovate and enhance their product lines, for example by automating various aspects of their systems and trading different assets.
Investors go where they believe their orders have the greatest chance of being executed. Order flow thus tends to attract order flow, which means that trading in any particular security is likely to be concentrated on a single dealing mechanism, or at most a small number of competing systems. If, however, trading is fragmented between different trading venues for whatever reason, it has been argued theoretically that this dispersal of orders lowers the probability of execution at each location and may therefore reduce liquidity and also increase the volatility of transaction prices. Other theoretical analysis of the competition between exchanges identifies the possibility that different trading systems may co-exist, if traders have diverse needs and desires, and if each trading system satisfies the trading preferences of a particular group of investors. Some theoreticiansshow that competition between trading systems is not always beneficial, and that it can sometimes decrease, as well as increase, liquidity.
In the European context, a range of empirical analyses have been made of the competition between the national exchanges following the Big Bang in London. Notably, this period was characterised by a bout of intense innovation at the various exchanges concerning their market structure, their technology, and the services they offered. Most of the measures of market performance - for example, in creating a diversity of trading venues, in linking the markets, and in being able to deal in greater size at the available spreads - show that such competition was beneficial.
A large amount of empirical research on the effects of competition between trading systems has also been undertaken in the US. Much of this work concludes that increased competition in securities markets leads to tighter bid-ask spreads, and, with some exceptions, has also had a range of other beneficial effects. This evidence is discussed at greater length later in this article.
Different preferences of different traders
The existence of different trading systems with distinct market architectures brings choice to investors. Different benefits are available from different types of trading systems. The transaction costs of demanding immediacy of execution, for example, can be significantly higher that the costs incurred in trading patiently. Large cost savings can also be achieved by using automated trading systems over traditional intermediated markets.
The existence of multiple trading venues, and the diversity amongst the preferences of investors, has put pressure on what best execution really should mean. Contrary to the historically accepted approach, some recent research suggests both that the concept of a best price is not well defined with multiple trading systems, and that attempts to mandate best execution as a consumer-protection device can limit competition between trading systems, and thereby not give the result that the consumer wants.
Inter-dealer competition
Such competition keeps the services which dealers offer at a low price, provides an incentive for them to enhance their products, and can also reduce the market power of price-setting agents and thus result in better execution conditions.
Price priority
With price priority, the highest bids and lowest offers get executed before other orders. A lack of price priority may lead to a reduction in the speed and accuracy with which information about a company is incorporated into the price of its shares, and in the quality of price formation of a market. Although it may be more difficult to deliver price priority in a market with competing trading systems than in one centralisedtrading system, market participants themselves have an incentive to ensure that they receive the best price, and thus that price priority obtains.
In order to understand much of the evidence, it is important to appreciate the difference between two types of orders: so-called 'informed orders'and 'uninformed'orders. 'Informed'orders are said to come from market participants who have some knowledge about the future value of a security they are trading, whereas 'uninformed orders'are from market participants who have no special information about the future value of the security. The execution of uninformed orders presents a lower risk to an intermediary than the execution of informed orders. This is because when an intermediary trades with an informed order, the likelihood is that the price of the security will move in favourof the informed trader, and against the interest of the intermediary. The dealer is thus said to face an 'adverse selection'problem when trading with informed traders.
Much early empirical work concludes that the NYSE offers better prices than those available on other trading venues. This evidence has been used to suggest that trades executed off the NYSE may therefore not obtain best execution, in terms of receiving the best available execution prices. It has also been widely argued that payment for order flow diverts low risk 'uninformed'orders away from the primary trading venue, and leaves it only the more risky 'informed orders'for execution. On this analysis, 'cream-skimming'as it has been called may undermine the process of price discovery because the diverted orders do not contribute to the price discovery on the primary market. Furthermore, if quote matching is employed, diverted orders may be executed at worse prices than would have occurred if those orders had been entered on the primary market. There is, however, growing evidence suggesting that market quality is not any worse when trading takes place off the primary market, and furthermore that the quality of such executions, when factors other than price are taken account of, may indeed be better.
A key difficulty in assessing the costs and benefits of fragmentation is that competition between orders may work best in a single consolidated market on which all orders can interact against each other, whereas competition between exchanges, trading systems, and OTC trading, and all the benefits such competition yields, implies fragmented markets. The two forms of competition may not be compatible with each other.
Secondary priorities
Time priority gives market participants an incentive to submit limit orders early to a trading system, because by doing so they are more likely to have their orders executed. Time priority is difficult to enforce in a world of competing trading systems. It has been suggested, theoretically, that a lack of time priority may cause spreads to widen, and may also reduce the chances of limit orders being executed. Advances in automation may, however, decrease the importance of having an enforceable secondary time-priority rule. This is because a similar outcome can be achieved by monitoring the market closely and submitting a market order should the price of the relevant stock move to where the limit would previously have been placed. This strategy also minimisesthe costs to investors who submit limit orders of providing a free option to other investors to trade against these limit orders when prices are likely to move against them.
Consolidation of information
The consolidation of information means that quote and trade information about the trading in a particular security is readily available from a single source. It is sometimes argued that the consolidation of such information is easier in a consolidated market, although private solutions to consolidating such information across markets have also been developed in many contexts.
Transparency is widely considered to play a key role in consolidating markets, by ensuring that prices are equalised over all trading venues. It allows arbitrageurs to trade whenever prices on one trading system are inconsistent with prices on another trading system. The view that full transparency is required in order to ensure price equalisation across trading systems is not, however, universally accepted.
Public good nature of various exchange services
Such services include price stabilisation, market surveillance, self-regulation, and quality certification. The existence of multiple centresfor trading an asset may reduce the provision of the public goods that exchanges provide, given the difficulty of funding them. It may also be easier to share regulatory costs on a fair basis in a consolidated market. However, various regulatory solutions have been proposed and implemented to address the problems of ensuring that trading on all available venues is adequately regulated, and that such regulation is funded appropriately.
Internalisation
Given that 'internalisation', the execution of trades in-house rather than on an exchange's order book, is a form of fragmentation, many of the themes discussed in the context of fragmentation are repeated here. The early analysis of internalisation stressed two key potential costs associated with it. First, internalisation was shown, theoretically, to have the potential to harm the price discovery function of a market, because it stops all orders in a particular security from competing directly against each other on a single order book. Such fragmentation, it was said, might also lead to larger bid-ask spreads and greater price volatility.
A second key problem identified theoretically with internalisation is that it may divert 'uninformed', low risk, trades away from a primary exchange, leaving only the more 'informed', and higher risk, orders on this primary exchange. This 'cream-skimming'may lead dealers to set wider spreads on the primary exchange in order to protect themselves from being picked off by investors with better information. If the diverted 'uninformed', and low risk, orders are executed at prices taken from the primary market, they may also then be executed at worse prices than if they had been entered on the primary market.
Two key benefits with internalisationhave subsequently been noted. First, it might allow alternative trading venues to compete with a primary or central market, and such competition could lead to enhanced pricing. Second, it would give market participants a greater diversity of choice of where to execute their orders, rather than having only the single option of sending them to the central market. This might benefit particular classes of investors. For example, the execution of retail traders'orders via internalisationmay let them obtain better prices and lower dealing costs than on a primary exchange, if their orders are low risk 'uninformed'orders. If dealers compete for retail orders, they will set smaller spreads for these low risk uninformed orders than would occur in a consolidated market, given that the dealers do not have to protect themselves against being picked off by dealing with investors with better information.
In the section on fragmentation, it was shown that much of the early empirical work on internalisation(almost exclusively about the US markets) concluded that the primary exchange (the NYSE) offers better prices than those available at other trading venues. However, a growing body of more recent evidence about internalisation and the practice of some trade execution venues in the US of paying brokers to execute their orders at the execution venue (known as 'preferencing') highlights the benefits available from internalisation, and questions both whether better prices are in fact available from the NYSE, and also more generally whether better execution is available. Using various different measures, internalisation and 'preferencing' have been shown not to harm the execution of market orders or limit orders, and may indeed improve it. One example is that market orders traded via 'preferencing' on regional exchanges tend to trade more favourably relative to the NYSE than market orders placed on 'non-preferencing' regional exchanges; another is that limit orders have a greater probability of executing on regional exchanges than on the NYSE. Internalisation also seems to have little short-run effect on posted or effective bid-ask spreads.
There is conflicting data about whether internalisation leads to cost competition or to the problem of 'adverse selection' noted above, when a dealer is likely to lose money, and therefore set higher spreads, if he deals mainly with informed traders. The most recent theoretical analysis of how internalisationfunctions takes account both of the prices that investors receive and of the commissions they pay for their executions. The conclusions about the effects of internalisationon best execution are more complicated than previous analysis suggests. In particular, it is shown that payment for order flow can lead to an increase in execution quality, taking into account both the price and commissions paid, which is consistent with cost competition. At the same time, however, it can give rise to an increase in the proportion of higher risk informed trades on a primary exchange, which is consistent with cream-skimming.
There is empirical evidence that when best execution takes account of factors other than merely price, the internalisationof orders can lead to better executions than are available on the NYSE. For example, while execution on the NYSE appears better than internalisationusing measures of trade-price quality, internalisationprovides more timely executions and produces more liquidity enhancement than at the NYSE. If potential reductions in commissions are taken into account as well as execution prices, previous results showing that better execution has been available on the NYSE than elsewhere may need to be revised.
A range of theoretical reasons have been put forward for why institutional traders may like to use a particular form of internalisationknown as the 'upstairs'market (a US term referring to the activities of members of an exchange, namely brokers and dealers, when they search for counter-parties for big institutional orders off the floor of the exchange). Upstairs markets may help institutional investors locate trading counterparties, and thereby execute large trades without fully revealing their orders to the 'downstairs'market. 'Upstairs'counterparties may be able to filter out large 'informed'orders. Upstairs trading may facilitate the collection of information about the unexpressed supply of, and demand for, securities. Upstairs trading may also help risk-sharing amongst market intermediaries, thereby lowering transaction costs.
Empirical evidence from a range of different contexts shows the benefits of upstairs trading, which occurs not only in the US, but also in many other countries including in Europe. On the NYSE, upstairs trading appears to be used by investors who can signal credibly that their trades are 'liquidity'motivated, and not motivated because they have better information. The upstairs market may thus enable transactions that would otherwise not occur in the downstairs market. In Australia, off-exchange trading -including ECN, upstairs and after-hours trading -is shown to benefit those traders in a position to switch trading venues, by lowering their trading costs. In Canada, the upstairs market-makers on the Toronto Stock Exchange provide a vehicle for screening out orders from investors who have better information, and for executing large liquidity motivated orders at a lower cost than the downstairs market.
Transparency
The key effects of transparency on securities markets, and on the participants in such markets, are complex and contradictory.
Almost all the evidence shows that greater transparency improves the speed and accuracy with which information about a company is incorporated into its share price. It may allow traders to select which trading system delivers the best quoted price, thus facilitating arbitrage between different systems, ensuring price priority, and enhancing the price discovery process. In turn, this is often believed to enhance best execution.
There is growing evidence, however, that greater transparency may also harm market performance in various ways. Not all investors are willing to expose their orders publicly, given the risks they run by doing so. Enhanced transparency can reduce their willingness to participate in the market, and has been shown to decrease liquidity in various contexts.
Traders with superior information are likely to prefer trading systems with less pre-trade transparency, so that they can keep confidential their trading intentions. However, uninformed traders, with no particular information advantage, prefer greater transparency. No single transparency regime will therefore be seen as optimal by everybody.
Transparency may make bid-ask spreads widen because market-makers have less incentive to pay to capture the information that a trade with an informed trader will bring. Alternatively spreads may decrease because information about transactions reaches all market participants, or because dealers are aware of each other's positions and compete more strongly with each other.
Transparency may encourage stabilising speculation that helps absorb order flow imbalances between 'buy' and 'sell' orders, and reduce volatility. Alternatively it may exacerbate market participants' strategic behaviour towards each other, with the possibility of increasing volatility.
The level of transparency in a market affects how trading systems compete against each other. Several commentators suggest that trading systems should own the property rights in the information arising from their trading systems, and that they have the appropriate incentives in most circumstances to determine the appropriate levels of transparency for their trading systems.
Footnotes:
The opinions expressed in this article are those of the author alone, and not necessarily those of the Oxford Finance Group, or any of its clients. This article reprints, with permission, a section of Ruben Lee, 'Capital Markets that Benefit Investors: A Survey of the Evidence on Fragmentation, Internalisation and Market Transparency', which was prepared for Association of Private Client Investment Managers and Stockbrokers/European Association of Securities Dealers, British Bankers Association, Danish Securities Dealers Association, Futures and Options Association, International Primary Market Association, International Securities Market Association, International Swaps & Derivatives Association, London Investment Banking Association, Swedish Securities Dealers Association, and the Bond Market Association (30/9/2002).