Exchanges on alert?
2003 is proving to be a relatively bumper year for securities, derivatives and commodity exchanges worldwide on account of unprecedented volatilities. Industry observers might be in disagreement as to whether order sizes are trending up or down, but the fact is that many market facilitators are enjoying rude health. However, the days when exchanges could enjoy the benefit of multiple sustainable revenue streams emanating from the business function of 'making the market' are long gone. And the argument that there is a trade-off between the need to reach a wider transaction community with greater externalities vs. the risk of fragmentation is gradually losing its power too with the arrival of new market standards and interoperable solutions.
But before we get into that, it's worth noting why exchanges shouldn't rest on their laurels, however tempting the prospect. Historically, exchanges could depend upon seven main sources of revenues according to Ruben Lee, author of the seminal text "What is an Exchange?" Exchanges could look either to membership subscriptions, and/or to fees for listing, trading, clearing, and settlement, and/or charges for the provision of company news, and quote/trade data. There were exceptions of course, where exchanges could derive added revenue streams from technology solutions for their customers, shared services and even levying fines. But for the most part, transaction fees proved to be the primary revenue source for most exchanges.
Membership fees would continue to be paid as long as exchanges continue to have members - but given the seemingly unstoppable trend towards demutualisation, this source of revenue is likely to be short-lived because most exchanges will depend upon customers for order flow, not merely member firms. Even those exchanges who choose to continue as mutual organisations will discover that competitive pressures will limit the extent to which membership fees can be raised. This competition will arise not only from other exchanges, but also from intermediaries such as broker-dealers or even the larger investment managers who might seek to 'internalise' trades, and which thereby compete with exchanges because these firms do not charge membership fees. As the larger global houses encourage the internal consolidation of price delivery and price discovery, the exchange entities that function as stand-alone exchanges will have more to fear than most.
This trend will compel exchanges to raise their game and become more efficient by way of driving transaction costs down and providing better quality liquidity. While advances in computer and communications technology have already changed the way securities exchanges operate, the implications are not fully appreciated. Once captured electronically, the marginal cost of executing an extra trade across an electronic trading system or of delivering quote and trade data to further customers is close to zero. Together these advances are likely to revolutionise the business models that some securities exchanges follow, by radically increasing the importance of revenues from the sale of the quote and trade data not to mention reference data emanating from such exchanges. Last winter's announcements by the London Stock Exchange to revisit the issue of SEDOL numbers and to price upgrades commercially bears this logic out.
Liquidity and value remain the key to exchanges ensuring that their members continue to co-operate in partnership with them. Efficient execution is important, but liquidity is more important, and sustaining value from customers through value-added services and market externalities such as practices and brand-building are key. But above all, as exchanges evolve, it is important that they maintain the liquidity of their book. To attract participants on their trading systems, exchanges should look to XML-based connectivity and message protocol standards such as FIX to allow the maximum number of participants including life companies and corporate pension funds access to exchanges in a cost-effective manner. Exchanges also need to make customer access and membership as simple as possible, so that all types of participant are able to connect inexpensively.
Remote membership will continue to be a growing trend as exchanges look to cross-border business expansion by adding non-local players, and exchanges will need to address the issue of membership agreements with these newer members if exchanges are to expand their liquidity. From the point of view of economy of scale, the good news is that exchanges are considerably more efficient than they might have been a decade ago. However, the traditional revenue streams of exchanges may be diminished by a new threat - the arrival of virtual trading places equipped with smart order routing. The challenge for exchanges will be how can they continue to attract and retain liquidity (maintain stickiness) if customers can, through a single screen on their desktops, easily tap into liquidity worldwide or switch between rival execution venues?
Exchanges will already be aware of the larger buy-side and sell-side firms deploying their order management systems (OMSs) to gain access to a broader range of liquidity pools to manage their order flow better. Their larger customers are hard-pressed to manage margins through continuous cost reduction, cost containment and cost avoidance. Rationalising their pipes to myriad execution venues will become an ever more attractive prospect because it will enable customers to strip fixed costs out at a time of uncertain volatile earnings. Customers will therefore increasingly look to single front-ends to tap into liquidity presented by alternative execution venues.
Internalisation - opportunity or threat?
Should exchanges fail to spot the trends, they would be extremely foolish. Technology allows liquidity not only to be fragmented but, nowadays, to be reconsolidated. Many exchanges rightfully keep their rivals under close scrutiny, while others look to ECNs or ATSs as sources of potential contention for order flow. But they would do better to look at moves among their larger sell-side members to develop full-spectrum services which might by-pass the need for an exchange. They should look to the innovations devised by integrated houses such as Citigroup, UBS, Deutsche Bank and J P Morgan, and even more worryingly, self-crossing on the part of the larger buy-side firms themselves, lest the food supply dry up!
The trend towards internalisation at the large broker-dealer and investment management firms in particular should raise serious issues for exchanges. This is no accident - as exchanges open up for business from new customers, there will be a race between exchanges looking for direct investor liquidity and the distribution strategies of the larger broker-dealers such as Merrill Lynch reaching out even further towards the end-investors. Sell-side entities are unlikely to charge membership fees to their customers, but might instead leverage their considerable distribution to offer end-customers a full-spectrum service. For example, they might consider offering front-end access to a myriad of virtual trading spaces - a concept known as consolidated virtual matching (CVM) - in order to forestall moves towards buy-side only clubs and peer-to-peer trading initiatives such as eCrossNet and Liquidnet from taking significant market share.
Internalisation itself can be viewed as a pool of liquidity which sell-side firms can offer back to their buy-side clients, so it boils down to an issue of liquidity provision, just as in the case of the stock exchanges. Client institutions should be able to tap into a global house's order flow, cross off what needs to be executed and then move on, thereby keeping closer watch on where the action is with regard to particular execution venues. Internalisation is attractive to many institutions because it delivers arbitrage opportunities and also eliminates market fees, which will have a positive effect on the price they give to end-clients. In principle, internalisation also allows organisations to tackle the entire post-trade process in-house - after all, there should be no central counterparty or settlement fees to pay.
That's the theory, and the key for providers (who would typically be the larger broker-dealers) would be to have the technology that enables firms to aggregate order flows sufficiently to offer that liquidity out to their clients. The arrival of distributed trading platforms comprising custom-developed components coupled to a supporting reference data model (covering instrument and legal entity identifiers) will equip select sell-side firms with the ammunition to support FIX messaging over multi-vendor portals. The platforms could provide price delivery and discovery functionality; the more so if workflows supporting exchange-traded/OTC instruments, dynamic price streaming, multi-user access, view+do watchlists, entitlements/permissioning and price tiering/history are integrated into what could be referred to as an 'exchange-in-a-box' architecture.
Several brokers are already moving into the direct market access business by acting as full-spectrum service providers for their buy-side clients, which differentiates them from vendor front-end 'single-window' type solutions. Broker-dealers have an in-depth understanding of how to extract the maximum value from utilising a given asset through committing capital, collateral management, self-clearing, risk netting, and providing access to specialist sources of liquidity than their clients are likely to. As some of the larger buy-side firms will wish to trade blocks and programme trades themselves, brokers may decide to respond by providing tiered levels of service directed at the retail investor: hedge funds, active-managed and passive-managed funds for instance.
Possibly the single most pressing issue around internalisation is the degree of transparency in the way the order flow has been internalised and the latter could be of concern to some regulators. Generally it might be argued that it is better for investors to have good pre- and post-trade transparency in seeing where supply and demand meet. If the arguments in favour of internalisation are economic, there are equally concerns about where might the real price be, and who might provide a true benchmark? What might 'Best Execution' look like in an even more fragmented liquidity environment comprising exchanges, ATSs and internalised matching? Would a sell-side firm offering the price discovery function have the interests of the buy-side client at heart? Or might proprietary trading considerations get in the way, particularly as sell-side firms typically commit capital to provide immediacy of execution?
The best way of answering these questions is to examine the merits and pitfalls associated with Best Execution and capital commitment.
Best Execution - best for whom?
Notions of 'Best' Price and 'Good/Best' Execution have become increasingly important as margins shrink and investors - be they institutional or retail - seek better returns. 'Best Execution' has also come to the fore in many regulators' minds in Europe on account of the Investment Services Directive. But equally, the search for 'Best Execution' as such has proved elusive, despite the many assurances otherwise. Definitions of 'Best Execution' appear in marketing pitches, RFPs, and due diligence reporting, but a generally-accepted and/or a legally-satisfactory definition of 'Best' Execution does not exist. Should the scope merely cover market prices as such, or should it be more expansive and cover overall end-to-end transaction execution extending to consideration of trading, clearing and settlement costs?
By 2000, Arthur Levitt (then Chair of the Securities & Exchanges Commission or SEC), kick-started the process of articulating new standards for Best Execution. The SEC Office of Compliance Inspections and Examinations offered this definition of Best Execution: 'In placing a trade, the trading desk will seek to find a broker-dealer or alternative trading system that will execute a trade in such a way that the trader believes will realise the maximum value of the investment decision'. By 2002, the SEC revised its view to state Best Execution as "A duty to seek the most favourable execution terms reasonably available given the specific circumstances of each trade", which is helpful but open to challenge under the trading conditions within a fragmented market. After all, if there is no central reference indicating where the most favourable terms are likely to be (e.g. because they are internalised without any transparency of information granted to investors), then sourcing 'most favourable execution terms' may not be feasible.
The implications of these guidelines are :
- Significant infrastructure spending to build record-keeping and reporting systems to track and audit trading information appropriately. Negotiation of acceptable commission ranges and documentation of the variance between negotiated and actual commission rates will become necessary.
- Trade management oversight committees will need to be established and the internal documents prepared for those committees will be audited by the SEC, with real and potential conflicts of interests documented.
- The choice of a particular trading system must be supported and the review and evaluation of trades, broker selection and execution performance can be expected.
The SEC guidelines recommended the establishment of trade management oversight committees that would be responsible for developing a trade management policy and a process to manage the efficacy of trades. Would you be getting what you paid for? How do you evaluate the service you receive? Are you evaluating the providers of that service? In principle, this is a breakthrough. But could the multi-faceted concept of Best Execution ever be measured, especially when the movement of large blocks into and out of the market from programme trades can affect both the bid-ask spread and the behaviour of the price chart? For example, were the latter market impact not a feature, the notion of tying Best Execution to volume-weighted average pricing (VWAP) would make a lot of sense, but many traders viewing the higher volatilities in share prices near the open or just before the close are likely to disagree.
So a Best Execution strategy would depend upon exactly when a trader decides to step forward to trigger a trade, not to mention the costs of the other trades on a chart. But how should this be computed when an order is fragmented into smaller tranches and executed over a series of trades, resulting in liquidity and market impact dispersal? With the growth of ECNs and internalisation in the global houses creating markets within themselves, how could the investor ever have prior knowledge of where the liquidity is and what is the best price? And without consistent benchmarks for market quality as a function of price delivery, much less price discovery, can Best Execution therefore ever be a viable candidate for regulatory oversight alone?
As there is no standard for measuring execution costs, the risk in following this route is that regulation might be used like a sledgehammer to crack a nut. Some sell-side players would be concerned at any open-ended costs in providing all the audit trails required against how they demonstrably achieved Best Execution. This is because Best Execution is not just a price but a process, itself difficult to define given different prices quoted for different types of order, different needs for immediacy, different portfolio weightings etc. As there is no such thing as same best price for every client or every order, some industry observers have argued Best Execution is better addressed as a best practice by market participants rather than centrally.
From the market practitioners' perspective, both the Investment Company Institute (ICI) and the Association for investment Management and Research (AIMR) were also asked to convene best practice groups to help define Best Execution. AIMR proposed Trade Management Guidelines on Best Execution consistent with those of the direction of the SEC, and AIMR's 2002 definition was that Best Execution might be defined as "The trading process most likely to maximum the value of client portfolios". This definition clearly emphasised attaining the best investment results, but not the lowest costs. It is clear that AIMR Best Execution is a procedure or process rather than a metric that one can apply on a trade-by-trade basis. The definition says nothing about best quoted price, lowest commissions, cheapest execution, trading cost analysis (TCA) nor whether any metrics would be assessed against a performance benchmark such as VWAP.
It is equally apparent that implementing investments is a process spanning portfolio manager to trader, and equally apparent responsibility for securing Best Execution is shared by many. However, whether firms use Plexus Group, Elkins McSherry, Capital Research Associates or VWAP, part of the answer to achieving Best Execution lies in having the process to measure it, according to Robert Schwartz, Professor of Finance at Baruch College, CUNY. The contention lies in the means, not the end. If the goal is superior portfolio performance, for example, to what extent is providing Best Execution a responsibility of market centres and other trading facilities? Do you get better radar in your cab or improve the infrastructure of the road system in the city?
Seen through these lenses, perhaps there is a role for regulation, although perhaps of processes, definitions of compliance and using an agreed form of TCA rather than regulation of products. This will allow firms to entertain different dealing and reporting styles in the same manner as they entertain different investment styles.
At what price immediacy?
Historically, accepted wisdom stated that securing Best Execution depended on the broker's willingness to commit capital during times of market distress to grant the buy-side immediacy in execution. The sell-side would assume the risk that they could locate a natural counterparty to the trade from their extensive knowledge of trading styles and relationships on the street. However, the work on 'Implementation Shortfall' highlighted by Andre Perold of Fidelity (Journal of Portfolio Management 1988) and finessed by Marc Gresack and Tony Kirby of Instinet (1992) addressed the issue of timing of orders and subsequent impact of orders on market prices and the role of broker-dealers in this process. Immediacy was a boon, but it was granted at the cost of leaking information from the buy- to the sell-side, with adverse affects on price.
The process of order execution is a source where performance resides, especially in a low-return environment, and buy-side firms choosing to make trades on ECNs such as Instinet or BRUT or through buy-side-only facilitation routes such as Liquidnet do so because they need anonymity. If the street sees a buy-side firm trading in a name, the other sellers and buyers are interested in the size and investment style. This information allows the street to infer why the buy-side firm is trading which is how the street anticipates price action - reflected in market impact.
It's a moot point how big a deal market impact is when compared to costs associated with commissions, delays and missed trades (opportunity costs). Plexus Group provided an overview to an industry meeting last November where they illustrated the 'iceberg' of transaction costs - 13 b.p. for commissions; 34 b.p. for market impact; 58 b.p. for delays and 18 b.p. for missed trades, so clearly the sums aren't trivial. But the buy-side's call on street capital for execution immediacy should be an insurance against the identity of the firm, the investment style of the fund and the size being transmitted to the street, or against front-running of the order. This process reflects the asymmetry of intent within a sell-side institution; the broker sales trader is the trusted party as far as the institutional sales trader, but additionally, the 'upstairs' or proprietary trader has the job of trading the firm's block capital. The latter may act in a manner which is adversarial to the trading intentions of the buy-side client, resulting in noticeable loss in performance against capital committed to ensure immediacy of execution.
The reason why this process is tolerated is partly behavioural and partly commercial. Sales traders pre-Myners are not necessarily compensated to make decisions that really work to achieve Best Execution, and indeed some might even have disincentives in doing so. They have soft dollar commissions to be paid to compensate for much-needed vendor services such as market data feeds or analytics. Many run in fear of demanding portfolio managers who are very focussed on mitigating opportunity costs and foster a culture of blame directed at traders if they delay or miss a vital trade (see Plexus figures above). Lastly, the broker sales trader often works to compensate the buy-side for losses or errors, thus providing a measure of assurance for the risk-averse buy-side trader from the point of view of 'knowing how to trade'.
From a commercial point of view, brokers don't operate as charities and therefore executing a block trade with broker capital commitment is no free lunch. Brokers traditionally 'rent' capital when trading a block because natural counterparties are said to occur only 20% of the time, according to Harold Bradley, SVP of American Century in a definitive paper titled "Views of an 'Informed' Trader" published in 2002 by AIMR. Brokers regularly make capital for block facilitation available only to payers of the largest commissions - functionally equivalent to offering a commission discount. Then if they lose money on trades, they can earn full rents form smaller full-commission players, so they are making up the difference with commissions from the smaller firms that do not have the same kind of leverage with the broker.
To conclude, when sell-side liquidity was abundant, the use of capital to guarantee the buy-side immediacy was relatively free and easy. However, most houses over the last couple of years started to look at the cost impact of committing capital. The consequence is that sell-side firms have become much more selective in terms of choosing who to commit capital to, and this is accompanying a greater reluctance on the part of the sell-side to commit capital to markets at large. Trends towards associated demands of financing, shifts towards programme trading, block trading, financial engineering involving derivatives and so on, compete for capital commitment as higher-margin businesses, and thus take over vanilla executions. The demands in terms of how to price capital in other words are becoming more intensive.
Where next for exchanges?
Best Execution is viewed differently by the buy-side, the sell-side and the regulators, and each view will be determined by the transparency of processes and the availability of underlying information to support each process. The problems with achieving Best Execution cannot be separated from the existing economics of trading systems, or the reluctance of portfolio managers to change the way they approach the trading function. Best Execution is not just about measuring trading costs (Trading Cost Analysis), but about clearly understanding and defining the implementation process, about having an implementation process that is consistent with the underlying management style, and about measuring the process and gathering feedback to make changes. There could be a central role for exchanges in helping provide reference market price information alongside critical reference data elements such as client account (legal entity identifiers) per transaction, and charging for the same.
But not before time. If payment for order-flow already exists in the brokerage market, it may not be long before this culture takes root in the exchange market. Exchanges could well be paying for the privilege of executing orders on their trading systems in future, and looking to incentivise existing mechanisms such as market-making to ensure that this continues to be the case. Historically, brokers acted as 'small order aggregators' working off negotiated block transactions in small increments over the phone. The arrival of ECNs during the 1990s eliminated the risk premium that institutions pay to trade, and determined the 'real' cost of a trade in terms of what an ECN was prepared to charge in cents per share, as opposed to the standard sell-side commission. As technology replaces capital in the aggregation process in order to reduce search costs for liquidity, then capital will be far too precious a resource for the sell-side not to ration.
So exchanges will have no choice but to address not only connectivity issues, but equally those of providing a central 'Best Execution' function if revenue streams continue to come under challenge and investors require compensation for order flow. My best guess is that this process will take effect post-implementation of Myners in the UK, as buy-side firms seek to derive revenue to compensate for the shortfall in softed trades. Once a common definition for Best Execution is more widely accepted, exchanges should contemplate mechanisms for comparing market prices at various execution venues in order to assess who might be offering Best Execution at any time. This is ideal territory for a market infrastructure to defend because of the need for a central reference price (needed, for example, for margin or mark-to-market calculations).
It is also an area ideal for exchanges to defend because, as large transaction processing organisations operating under attractive economies of scale on behalf of members and customers, they could offer the technology to make such comparisons. It is a concept which could develop into a revenue-generating opportunity for exchanges and is certainly an answer to the threat from internalisation and the paying for order flows. The signs are that Deutsche Boerse's XetraBest is a step in the right direction because private investors are guaranteed to receive a better price for their order than the best price available at that time in the central Xetra order book. Investors are able to review XetraBest trades from the previous five days and compare the 'improved' prices against the prices that were available in the Xetra order book at the time of execution. It is a vital first step towards equipping investors with access to consolidated virtual matching across any execution venue.
Market infrastructures that are able to incorporate clearing and settlement and/or cash plus derivatives price discovery will profit from greater economies of scale and scope to spread their costs and risks in the current volatile trading environment. But their focus will have to be one of value, and how to keep delivering that value to end customers profitably. Any fragmentation of liquidity will be supportive to exchanges; while the buy-side might wish for anonymity in some instances in order to avoid the impact of order sizes on the price discovery process, there will equally be economies of scale and scope to be had for offering counter-services that centralise and consolidate both order flow and information. Market infrastructures could therefore choose to run a suite of business models enabling them to price according to clearing and settlement arrangements or per risks assumed.
As mentioned earlier, exchanges need to get fully behind transactional standards such as XML-generic protocols (FIX), as well as embracing reporting protocols such as XBRL which are becoming more accepted by regulators and accounting standards bodies. Exchanges cannot afford to punt proprietary standards at the buy-side community, and instead should engage in every opportunity to encourage both the institutional investors as well as the end investors to transact their cash and derivative instruments. The evidence is that exchanges which encourage an open approach to connectivity, coupled to significant value creation through leverage of information, will have found how to be successful predators. Those who take the reverse route of developing proprietary standards and compete on offering vanilla services to a diminishing number of old-guard member firms will surely be prey.