Over the past couple of years, the issue of ‘illiquidity’ and how to address it has dominated the discussion in the corporate bond market. This has resulted in a ‘rush for gold’ mentality, with every firm that feels it has something to offer looking to provide a technology solution for the ‘liquidity problem’. With a plethora of new trading systems and incumbents all vying for attention, it is worth pondering who the likely winners will be and what the market will look like five years from now once this latest infusion of capital has all played out. Having provided sophisticated workflow tools that enable sell-side corporate bond desks to find the other side of the trade for a dozen years, Codestreet is well-placed to observe the evolution in the corporate bond market.
A common thread amongst all the new platforms is the focus on new trading protocols. Each platform distinguishes itself with some nuance on how to match buyers and sellers who are likely counterparties, with each system implying varying degrees of change to present market structure. What many of these platforms are missing – beyond the capital needed to sustain themselves long enough to generate sustainable liquidity – is a recognition that price discovery is at least as big a problem as the challenge of matching logical counterparties. Without incorporating a convincing form of price discovery within their proposed solution, these new platform operators will most likely fail. You can find the appropriate counterparties, but you can’t make them trade.
What leads to a trade is a ‘believable’ pricing context that allows both parties to feel confident in the price they are accepting. Only the platforms that provide a means to achieve satisfactory price discovery will prevail.
Pricing context
In any illiquid market where a tradable price is not readily available, such as the institutional corporate bond markets, availability of a pricing context is critical for determining whether a price is acceptable to a trade participant. This idea goes well beyond the corporate bond market and is true for just about any illiquid market, such as art. The request-for-quote (RFQ) process has been successful in the corporate bond market, not only because it is efficient, but because it provides a believable pricing context that allows the buy-side participant to feel comfortable with the price he is about to accept. As time goes on, RFQ pricing responses for larger trades are becoming more believable reflections of the best available price, thus the volume of larger trades done via RFQ increases.
A ‘believable’ pricing context is key to the success of any platform that expects to be a significant part of any future corporate bond market structure. The qualification ‘believable’ is key, because central limit order books (CLOBs) and streaming pricing, while providing the appearance of reflecting market pricing, are not believable at size in an illiquid market (no one will leave an ‘on the screws’ price at size available to be picked off) and this is exactly the reason that CLOBs and streaming pricing protocols have not been successful in the corporate bond market.
Similarly, synthetic pricing from data vendors is insufficient to provide true pricing context and cannot ever be more than a lagging indicator. This is why it is so difficult to see pure buy-side to buy-side platforms succeeding. As well as the obstacles of buy-side workflow, regulation and behaviour change, the core problem with buy-side to buy-side trading is that price formation amongst a group of buy-side firms must become believable and reliable. This outcome does not seem likely given that buy-side traders are not generally expected to provide on-the-money pricing of bonds with immediate response.
Buy-side to buy-side platforms will support opportunistic trades where an item that is on a watch list becomes available and one of the participants is willing to pay or accept a counterparty’s proposal without competing bids and offers to provide context. But this will always be just a subset of the market. More generally, market participants are not determined buyers or sellers without regard to market price; rather, they decide to buy or sell based on the price that is offered within a pricing context that enables them to evaluate whether to trade or not.
The need for a pricing context is not just an e-trading problem. Voice brokers have long known that to complete a trade they need to paint a pricing context for the participants on the phone by bringing in other parties. This fundamental principle of trading psychology dictates that only platforms that can create a pricing context from believable, numerous and consistent pricing sources are ultimately likely to emerge victorious.
So if establishing a believable price context is key to driving trading activity, it’s fair to anticipate that the successful platforms will be those that best support believable price discovery.
This is why issuing an RFQ to a set of valid pricing sources is such a compelling protocol for an illiquid asset like corporate bonds. An RFQ response returned from a validating set of reliable pricing sources provides the pricing context that generates the necessary comfort and security for a price taker to move forward with the trade. The RFQ protocol has the virtues that responses are current, are focused on the particular trade at hand and are delivered within what is presumed to be a competitive context. It’s tough to top that.
An intelligent RFQ
So if RFQ contains the necessary ingredients for price discovery, what form will RFQ take going forward?
Currently, RFQ is a very inefficient process. It thrives in its present form because there is no compelling alternative. This is why traders will tolerate hundreds of irrelevant pop-ups that contain requests of little interest to them. This is a burden to the sell-side trader, but also a problem for the buy-side in that pricing fatigue reduces meaningful participation on any given RFQ response.
What is needed is a more intelligent RFQ process that encourages broad price discovery from high-quality pricing participants while maximising the efficiency of both those providing the pricing and those requesting the pricing. The path to achieving these goals lies in more effective use of the data available on the trading floor. Trading data can be used to help both the initiator and the responder to the RFQ. For the initiator, a broad set of data can be leveraged to identify the logical set of pricing providers to receive the RFQ request. This minimises information leakage for the initiator, while also reducing the noise that the responder needs to filter out, increasing the odds of getting a suitable response. For the responder, the availability of deep trading history along with sophisticated data processing tools enables an easily digestible context to be extracted from which to formulate a pricing response to an RFQ. This context can be used by either a human trader or an automated responder, both similarly benefiting from understanding the opportunity to realise a profit from the position a winning RFQ response would create.
The intelligence that can be extracted from a comprehensive set of trading floor data can also be leveraged by a wide range of matching protocols designed to help identify a logical set of responders to an RFQ request. Dark pools and other matching systems all have their place, but that place is to help identify a suitable subset of participants within an RFQ process.
Systems that improve RFQ targeting and enable an optimised set of pricing participants to respond to a request are a benefit to all participants. Targeting reinforces the perception of an accurate pricing context, thereby encouraging the buy-side to act on the response. Likewise, it encourages participation by sell-side firms as they are being asked to respond only to requests they care to respond to.
To enable the collection of the data necessary to offer accurate targeting of RFQ responders, the RFQ must be part of a workflow that allows the RFQ system to gain access to the data that will both drive RFQ addressing and help RFQ responders to optimize their responses. Workflow systems that best capture the relevant data will be the true competitive differentiators.
The future will look like an optimised present. Gone will be the blast RFQ, to be replaced with a directed RFQ that is initiated when the pricing sources are known to be in a position to respond. In turn those responders will be armed with all the data they need to optimise their response to RFQs. Welcome to the corporate bond future.
A data-driven future
One of the concerns with respect to the future of RFQ is about the withdrawal of sell-side balance sheet support. Ultimately this will be resolved by the market. With rates low and balance sheet expensive, dealers need to get paid to potentially step in front of a train. That is going to be true for any market participant. Similarly, in panic situations or severe market gyrations where uniform sentiment driving prices down, the market will find a new level at which to trade regardless of what platforms are out there. This is independent of market structure and is a consequence of the increase in market homogeneity arising from increased and consolidated information flows that lead to shared sentiment. Dealers with the infrastructure to guide suitable RFQ requests towards their desk and the ability to lay off the risk will be able to offer the most competitive pricing and will prevail in this future market. It’s a data-driven world, even on the corporate bond trading desk. The future belongs to those who best capture and process the data needed to optimise their participation in the new, targeted RFQ trading process.