By forcing certain swaps trading onto SEFs, Dodd-Frank is asking the swaps market to make overnight changes that took the equities market 20 years. This means more risk from more directions, and Mark Brennan of ITRS Group argues that rigorous monitoring is essential to ensure the new empire doesn’t crumble at the start.
The new swaps market structure, driven in the U.S. by Dodd-Frank, is unprecedented in its disruption to market participants, and has inspired no shortage of debates and jeremiads on its various flaws and shortcomings.
That’s understandable: by regulatory fiat, almost overnight, swaps trading has had to move to an electronic market – a move that most other markets, notably cash equities, but also exchange traded derivatives, FX, and corporate credit, have been making at varying rates for years.
Whilst they’ve had the luxury of adding and altering systems as and when the rules and markets change, growing organically and ironing out kinks along the way, the swaps market has been granted no such luxury.
Sure, there may be some second-mover advantages – lessons learned from mistakes already made – but the swaps market is being asked to build Rome in a day, with all the difficulties that entails.
The new SEF trading mandate has imposed significant regulatory hurdles for banks, and caused them to re-assess entire business models.
But above all else, SEF trading imposes technological problems – across a variety of functions and workflows.
Banks must connect to SEFs, get market data, apply credit checks, execute trades, connect to an SDR (if a dealer), and connect to the clearinghouse.
The evolution of the cash equity markets in the late 90s and early 00s involved a steady march towards automation. More and more trading was done electronically; more servers were used to run complex trading environments, which sometimes replaced entire trading desks.
As infrastructure grew, and trading volumes increased, banks saw increased instability, resulting in large-scale, firm-wide initiatives to create technology systems and teams devoted to operating and monitoring their trading infrastructure.
It took a number of years for this “second-order” problem to succeed the core problem of building technology infrastructure to address client and market structure requirements.
Put differently, it took several years before banks’ business interests aligned in such a way that key stakeholders understood they needed to monitor their trading plant, not only for stability and capacity, but also for visibility into fault tolerance, as well as client flows.
Whereas historically server failures were IT problems, now resilience and overall management of “failure domains” was front and center a problem for the business, affecting P&L. Eventually, even the regulators took notice, giving us Reg SCI.
This shift was gradual, which afforded banks the luxury of addressing the first-order problems (building these complex systems) before creating and in turn addressing the second-order problems of monitoring and stability.
The SEF market structure, like equity markets, is horizontal and fragmented, where fungible swaps trade on multiple venues. The race for SEF winners and losers is not complete, but regardless of the outcome, the buy-side and sell-side will connect to multiple SEFs.
This implies the need to watch multiple venues and multiple price feeds. Ironically, in the absence of best execution regulations, banks transacting on SEFs need increased scrutiny of their flows.
Additionally, SEF execution methods include both RFQ and order books; and dealers may also utilize Request for Stream (RFS) mechanisms. Though the SEF will link the RFQ to the order book, these separate mechanisms imply multiple flows that the banks will want to watch.
If the immediacy of the new market structure and complexity of its technology were not enough of an incentive to monitor flows, there is the added dynamic that banks may find themselves transacting on SEFs with some very sophisticated technology players, who have embraced the new market structure as playing to their strengths.
HFT and “Chicago style” electronic trading is not new to other asset classes, but again, the issue is that elsewhere it evolved over time, and banks built out their electronic trading infrastructures concurrently with the growth of HFT.
The traditional bi-lateral swaps market was a closed, invitation-only market, transacted primarily over the phone.
Now, the much-touted fair (read: open) access that the CFTC demands of SEFs allows some sophisticated technology shops to automate price making, previously the purview of dealers.
Those banks running a modern rates business need to understand the quality and quantity of their prices, and the speed of their contributions and responses.
This implies a very rapid change in technological intermediation. The sell-side dealers are now potentially competing with automated trading shops.
What does this insight include? Monitoring system health (e.g. am I connected to the SEF?) should be a given for any modern technology infrastructure. Monitoring latency gives insight to the trading desk that the prices they are seeing, as well as making, are rational.
Traders will want to know if they are falling behind – either in SEF price feeds or, for example, in responding to client RFQs. And finally, analytics of SEF price and order data can give trading desks the confidence that they are adequately navigating the new market structure.
For example, some SEF order books may become “toxic”, as liquidity at certain price points evaporates. This insight is only available by monitoring these flows, and applying appropriate analytics.
Rome wasn’t built in a day. Arguably, a market structure isn’t as complicated as an empire, or even a city; but there are bound to be problems when you create a whole new market landscape at a stroke.
Swaps traders don’t have the luxury of figuring it out as they go, while the market evolves around them – as those in the equities (and other) worlds have.
That’s not to say that the new regulatory requirements or the speed of their implementation are good or bad; but the sheer complexity of the task makes it all the more crucial that systems are properly monitored and stable from day one.