Repo markets have relied on email and chat messaging since these tools have become widely available. While manually sending post-trade instructions to a human reader has been useful, there continue to be important opportunities for automation either by capturing data from text exchanges or by moving transactions onto Straight-through Processing (STP) platforms. Email and chat messaging read by people may not be dead for repo transactions, but nor should they be widespread going forward for the same purposes. A guest post from MarketAxess.
Has email died for many firms as a way to exchange order and confirmation messages for repo? Are we going to be hearing 2010 calling sometime soon, asking for their emails back? The answer to both these questions may be “yes”, as technology advancements have led to more efficient processing methods. The difference from ten years ago is who, or what, is reading the email.
It’s more time-efficient for computers to parse emails and chat messages for order-related information than it is for people to read emails and transpose into electronic systems. But both human and computer readers introduce a potential new element of error into the trade process as information is moved from emails to the next generation of information management platforms.
Manual trade and price negotiation has benefits
Our market investigation on email usage found differences between price negotiations and post-trade processing, differences between small and large firms, and a further divergence in practice between dealer-to-client and dealer-to dealer-business. Some asset classes, for example High Quality Liquid Assets (HQLA) repo, are more prone to using electronic markets without the need for email while emerging markets or corporate bond repo rely on more human interactions. Whenever there are more discussions about credit utilization or negotiated haircuts, manual messaging is more of a requirement than for government bonds.
In all cases, price negotiation is still high touch. Salespeople communicate with clients and their traders via emails, Bloomberg chat and other messaging protocols. This shows that what may be considered a commoditized market is not that commoditized at all: baseline pricing exists but basis points matter, and negotiating the price requires engagement. Further, when clients are rolling a position, they would prefer to keep the securities in place and not go through the hassle and expense of moving the paper. Traders know this and can offer slightly less competitive pricing while still getting the trade done.
There are other reasons for sticking with emails or chat messages: dealers may want to be aggressive in pricing if they need particular securities; dealers want to avoid showing their hand to the market; and books may need to be highly controlled across statement dates or netted off the balance sheet. The expertise of price management is not well suited to algorithmic or automated trading. Haircuts may vary based on asset volatility, asset concentration, capital allocation and market sentiment. These unstandardized factors need to be negotiated along with rate and term.
Post-execution, text automation and capture are lagging
There appears to be a gap between automated extraction of order data from emails and chat messages vs. extraction of post-trade data. Large firms have invested in tools to capture the trade but post-trade confirmation emails and chat messages are still largely the domain of human readers. Large firms have responded by investing in STP platforms to manage confirmations, comparisons, margin calls, trade rebalancing, substitutions, and move securities without an email at the beginning of the process. Still, enough information is sent by email and chat that must be cut and pasted by both large and small firms that opportunities for error creep into the system. These errors can impact inputting delivery instructions, manually adjusting positions for substitutions or trade close-outs, tracking settlements, or any of the other tasks that are managed without human involvement at firms with more advanced STP.
The imminent arrival of Europe’s Central Securities Depositories Regulation (CSDR), which affects any firm trading European securities, provides an important push for all entities to review the accuracy of their post-trade processing mechanisms. This is happening first in Europe but has global implications: large dealers that operate in Europe, the US and Asia are re-assessing their processing across the firm at the same time.
CSDR includes penalties for late matching and settlement fails that range from 0.15 bps to 1 bps on the value of the trade, depending on the cause. The more that the use of email and chat messaging is reduced for post-trade, and firms shift those communications to standardized matching platforms, the more they can see a direct reduction in potential CSDR fines. Smaller firms that operate outside of Europe are much less impacted by CSDR and do not have the same incentives to move to STP; this translates into a higher cost to do business both internally and for counterparties subject to potentially higher fail rates.
Improving emails vs. migrating to STP post-trade
Given the prevalence of email and chat messages, there is an argument to be made that firms should stay with this protocol and work to minimize errors. A consistent formatting of emails for trade details would go a long way toward improving accuracy when using automation to scrape texts. Likewise, machine-readable spreadsheet attachments would provide greater consistency if the same vocabulary were used by all counterparties. But should column orders change or unfamiliar abbreviations or short cuts be used, the process could grind to a halt. The objective is to have exception processing in place only – but exactly where the bar is placed on what constitutes an exception in this context will be the source of debate.
On the order capture side, dealers are increasingly using artificial intelligence (AI) programs to read emails and translate information into machine-readable form. From there, orders can pop up on a trading screen at which time a rate can be put on it. In a trade like that, terms matter. “I have cash to invest for…” may mean the same thing economically as “Offer me paper for…” but an AI system must provide that translation. In order to be effective, when systems are dependent on machines reading human intentions, there must be some way to catch errors very early. This work is largely complete at the biggest firms.
A gap still remains after the order is complete however. In a perfect world, the next important step is to remove the email trail in favor of both sides entering or capturing trade details in an electronic system that assures post-trade accuracy. Trade discrepancies and exceptions found in a post-trade matching platform should be resolved by chat, then rebooked and matched. This model can extend from trade matching through to matching a pair-off or partial pair-off for settlement.
In the meantime, email and chat messages remain a part of the repo trade and post-trade process. They are vital for relationship management and other firm communications. The more that dealers and clients can reduce variation in text strings and attachments and link these messages to an automated platform for post-trade reconciliation, the more that firm-wide and industry-wide efficiencies can be generated. Removing this manual intervention should lead to more time dedicated to other improvements, decrease settlement and fees, and help to avoid CSDR-related penalties in the future.
About the Author
Colleen Stapleton
Business Development – US Lead
MarketAxess Post-Trade
Colleen Stapleton is the US Lead of Business Development for MarketAxess Post-Trade. Colleen actively works with new and existing clients to improve their Post-Trade repo automation, as well as to develop innovative and creative ways to address client’s needs. Prior to joining MarketAxess she spent 13 years at J.P. Morgan where she held many roles across the Fixed Income Markets Middle Office. Most recently, she was Vice President of Client Strategic Initiatives where she focused on improving the client experience for many of J.P. Morgan’s largest investment manager clients. Colleen graduated from Fordham University with a B.A. in Political Science and Business Administration. She also competed as an NCAA Division I Cross Country/Track & Field athlete. Her passion for racing continues today as she tries to keep up with her three small children, Augustus, Hazel and Ellis.