The fixed income markets are going through a period of significant transition. While capital constraints and a long period of low volatility have led to challenges in how firms source liquidity, recent increases in bond volatility have led to renewed profitability – and investment – in the markets. And with regulatory reforms forcing more and more trading into the electronic domain, new liquidity venues are beginning to proliferate. Firms across the buy side and the sell side are now having to assess what this all means from a technology perspective. In this Financial Markets Insights report, Dan Barnes and Mike O’Hara of The Realization Group investigate the evolution of the fixed income market, with help from Stephane Malrait of ING Bank; Sassan Danesh of Etrading Software; Constantinos Antoniades of Liquidnet; David Bullen of Bullen Management Ltd; and Brian Cassin of Vela.
Fixed income trading, particularly in corporate and municipal bonds, lacks the centralised infrastructure common in exchange-traded markets. Investment firms are reliant on the dealer community – who are commercially incentivised to support clients – to provide prices and execution directly.
However, dealer balance sheets and leverage have both declined since 2008, according to the report, ‘Dealer balance sheets and bond liquidity provision’ published by Nina Boyarchenko and Or Shachar, economists at The Federal Reserve, and Tobias Adrian, financial counsellor of the Monetary and Capital Markets Department of the International Monetary Fund, in the Journal of Monetary Economics on 7 April 2017.
Capital adequacy regulation has made it more expensive for banks to carry risk, which has made the trading of capital intensive, low-yield assets – such as corporate bonds – unprofitable for buy-side clients. In their recent analysis of liquidity in the corporate bond market, Boyarchenko, Shachar, and Adrian wrote, “We link directly the trading behaviour of market participants to their balance sheet constraints. We find that post-crisis regulation has had an adverse impact on bond-level liquidity.”1
Under MiFID II best execution rules will drive investment firms to check their execution against a range of market prices, driving them to aggregate quotes from dealers. With the sell side withdrawing from the market, and absent of centralised infrastructure, the technical challenge of gathering and assessing prices is even harder.
Making prices and electronic trading in many bond markets is increasingly being driven by two other sets of players.
The technology used by buy side, sell side, market operators and AT traders is very diverse, creating a technical challenge for any firms wishing to build an accurate picture of the market. The missing link is an electronic infrastructure. In equity markets exchanges, these have formed the lynchpin for faster and resilient infrastructure development. In FX, datacentres provide cross connects to support superfast trading.
To move to the next level of electronic trading, the fixed income markets have to build an appropriate architecture in order to facilitate order routing, data transfer and access to execution venues.
Issuance of fixed income instruments does not take place on an exchange, nor are the instruments perpetual, meaning they can be issued more frequently and for different lengths of time – tenors – than other assets.
“The number of bonds is quite large – relative to equity or FX assets – and you have to find liquidity that you may want to execute electronically,” says Stephane Malrait, Global Head of eCommerce for Financial Markets at ING Bank. “Electronification is trying to solve those two problems; liquidity provision and execution transparency provision.”
“Electronification is trying to solve those two problems; liquidity provision and execution transparency provision.”
Stephane Malrait, ING Bank
The importance of these processes to the overall value proposition to customers varies not only between industries but also firms, of course. As such, the level of investment a firm makes in digitising its fulfilment process or migrating service delivery to a mobile app, for example, will be informed by its wider impact on the bigger strategic picture.
The varied electronification of trading between different categories of assets based on issuer and type – listed derivatives are largely electronic, municipal cash bonds barely at all – means that it is unevenly distributed. There are currently around 50 venues trading corporate bonds globally.
Brian Cassin, Head of Business Development at Vela, says, “This proliferation leads to an implementation problem for market participants, as each venue is leveraging a different technology, protocol and reference keys, cross venue transparency becomes very challenging. Add to this that many of the fixed income group’s technology is not as mature as their equity counterparts, so this transformation will take a lot longer, without real investment.”
However, one should not be blinded by the size of the mountain to climb in markets with limited electronification. In corporate bonds, three major trading venue operators dominate in terms of trading volume while only a handful of the new providers are having any success.
Sassan Danesh, Managing Partner at Etrading Software, says, “The other big liquidity source beyond venues are the dealers. So it’s not just venues. There are too many liquidity sources to have point-to-point activity and hence the buy side has to make a decision as to which of these liquidity sources they connect to, because they’ll only be able to connect to a relatively small amount.”
In government bonds there are fewer options, partly because these low risk, highly liquid instruments are still well supplied via bilateral dealer-to-client trading. However, as regulation encourages the electronic trading – requiring quantification which is so much harder via voice trading – relationship trading may take a hit in these markets as well allowing platforms to grow.
The revised Markets in Financial Instruments Directive (MiFID II) is placing demands upon fixed income trading – electronic capture of decision making, trades, price publication, quantification of execution, and best execution – so that electronic and automated trading will be made easier as voice and manual trading is made harder.
“The very liquid treasury securities – on-the-run treasuries, 10-year bunds – could be on central limit order books (CLOBs). We’re looking at the top 15-50 ISINs in fixed income and rates,” says David Bullen, Director, Bullen Management Ltd. “I think MiFID II will supercharge a highly-liquid, highly-transparent market in a limited set of securities. Because of the data which will be pumping out, there’ll be just a clearer definition of which ISINs in fixed income should fit that CLOB model.”
“There are too many liquidity sources to have point-to-point activity and hence the buy side has to make a decision as to which of these liquidity sources they connect to.”
Sassan Danesh, Etrading Software
He notes that if the most liquid 15-50 were to be included on a rotating basis, that would be hard to define and manage. Given the current levels of information on instruments however, that could be about to change.
“I think the MiFID data will drive an independent way of deciding what is a super-liquid fixed income security and which should be listed and de-listed; although the integrity of that process needs to be rigorous and not have someone keep on expanding the number of ISINs because they want to get more flow. It needs to be entirely data driven,” he adds.
As electronic platforms gather enough liquidity in particular instruments to reach a critical mass of users, it will be incumbent upon trading firms across the buy and sell side to ensure they can connect with them. That means trading and investment firms have to be actively engaging with their own proprietary systems to ensure that they are able to work with as many options as possible
“What our clients are increasingly interested to hear about are how will intelligent execution help them uncover additional liquidity, and how will data augment their trading decisions?”
Constantinos Antoniades, LiquidNet
Constantinos Antoniades, Global Head of Fixed Income at Liquidnet, says, “What our clients are increasingly interested to hear about are how will intelligent execution help them uncover additional liquidity, and how will data augment their trading decisions? The combination of the two can create opportunities for significant alpha generation and performance differentiation. Given the low yield and low return environment, the additional alpha from using better data and sourcing better liquidity can make a difference at the fund performance. Buy-side traders are increasingly having access to tools that can help the trading desk enhance returns for the fund.”
Banks and dealers have been actively engaging with the new landscape, albeit with less enthusiasm, than the asset managers. A shake-out of the business has seen several large dealers drop out of specific fixed income markets. Despite the raised cost of capital for trading bonds, their 2017 first-quarter profits from fixed income trading rose for the first time in five years, according to analyst firm Coalition’s ‘Coalition IB Index’ released in May 2017.
The sell side has seen interdealer brokers (IDBs) begin to open access to the buy side as sell side activity has lagged, creating a tension between the IDBs and their dealer members. Pre-trade data services, which harness both buy- and sell-side data, have seen lower levels of sell side participation than buy side activity. Consequently, asset managers have built their own pre-trade data aggregators, which their dealers implicitly support every time they offer clients’ prices. One of these, AllianceBernstein’s ALFA system, was acquired by pre-trade data provider Algomi. Automation of individual processes, if not the entire trade lifecycle, is becoming normal across businesses.
“The MiFID data will drive an independent way of deciding what is a super-liquid fixed income security and which should be listed and de-listed; although the integrity of that process needs to be rigorous.”
David Bullen, Bullen Management Limited
“Pretty much the entire marketplace now needs to have the concept of a rules engine because the regulations are so complex that relying on either a salesperson or a trader to understand this in real time, whilst the trading is taking place, is just not something that is practical,” says Danesh. “We’re seeing a secular trend towards the embedding of rules engines within critical workflows so that trading can be guaranteed to be conforming with the required regulations.” The economics for buy-side and sell-side firms are different. Asset managers need to reduce costs while developing greater in-house capabilities so they can get control over decision making, prove their value to investors and lower the cost of finding alpha.
Sell-side firms need connectivity that can keep them plugged in at a time when they are being disintermediated, with new offerings to attract client liquidity.
Brian Cassin says, “Vela is engaged in a number of opportunities to implement low latency market data and market access technologies. The focus from the fixed income community has been on achieving cross venue price transparency, building the book, and leveraging market access tools to execute on those venues. Bringing all of this technology together will drive flow to their trading desks – a change that will allow them to reach a larger pool of trading partners than classic phone broking.”
Firms are being driven to develop electronic trading for economic reasons. The regulatory environment is aligned with this aim. However, there is a gap in technical ability and demand. If execution opportunities are distributed across multiple different trading venues for an already diverse set of assets, liquidity aggregation is getting harder, not easier.
“As there are a hundred different venues, and of deciding what is a super-liquid fixed income security and which should be listed and de-listed; although the integrity of that process needs to be rigorous.”
Brian Cassin, Vela
“As there are a hundred different venues, and each is marketing themselves to capture flow and create liquidity,” says Cassin. “How does a trader know if they are getting a best quote or execution? They are having to call around. That is the problem we want to help solve.”
As that scale increases, management of the trading landscape is moving further away from the capabilities of human traders alone. Thankfully the convergence of factors is working in favour of electronic progress.
“It’s true that because there is diversification of venues and reference points you could see an increase in the cost of the market data and the complexity to aggregate, normalise, store and then analyse this market data,” says Malrait. “On the other side, if you look what MiFID II will potentially bring from January next year, a lot of this market data will be free to use after 15 minutes because it will be published by the reporting Approved Publication Arrangement (APA) to the market.”
To overcome this, smart venues and dealers are employing machine-learning and big-data systems. They are capable of handling the complex universe of assets that makes up fixed income, and thrive on large data sets.
“You have to potentially deploy intelligence to be able to create a price which is relevant and try to use all the information you can gather to find that relevance,” says Malrait. “So at ING we have produced machine-learning algorithms to be able to find what the most relevant price is for our clients.”
Antoniades believes that in the future venues will also have more intelligent tools to help users to execute an order that depends on the order characteristics. “A €30 million order should have a very different execution approach than a €2 million order,” he says. “We expect firms will be sending orders to a venue that is intelligent enough to suggest the best execution strategy and make intraday recommendations about how that may change given market events. At Liquidnet, we are moving toward using technology to arm the buy-side traders with such tools.”
At a practical level, firms in the market must ensure they have the components to build or connect systems in order to pull the data together from venues, from dealers and from clients. “Historically, these technology stacks have been started by a business unit,” says Danesh. “As workflows converge driven by regulatory requirements, firms implement a common technology stack across business units. The front office is too diverse to be able to typically use just one tech stack, so the obvious way to segment is high-touch/low-touch.”
An intelligent approach is needed to data aggregation for both market data and market access. For the sell side, getting past the challenges around connecting with different venues opens up opportunities, to provide consolidated pricing and quoting for the buy side.
In the absence of a central limit order book, it is incumbent on market participants to build the technological infrastructure to pull in prices and build a picture of the market on their trading desks to support best execution. Asset managers’ newfound independence will not disappear. It is driven by the withdrawal of dealer activity and smart brokers or liquidity providers can fill the existing gap in support.
From 2018 onwards, venues, dealers and asset managers will need to aggregate data from a far larger set than exists today. They will need to use that data more effectively than today, and with intelligent technology to make decisions. That will augment the fixed income traders, not replace them, and allow them to see the world in more detail than ever before.
This article was first published by Vela (a client of The Realization Group): Making the market: Building prices in fixed income