The association’s common domain model has completed its pilot phase. SFT speaks to members of the ISLA working group to discuss how it went and what’s next
Panellists
Saied Attarian, consultant- CDM modelling, REGnosys
Lee Braine, director of research and engineering, Barclays
Sunil Challa, director, group strategy, Barclays
Julian Eyre, managing director, Eyre & Co.
Chris Rayner, software engineer specialist, FIS
David Shone, digital transformation, consultant, ISLA
Ben Smith, product manager, Sharegain
Martin Walker, head of product securities finance and collateral management, Broadridge
Harry McAllister, information architect, BNP Paribas
The CDM pilot phase has just finished. What did that consist of and what are the expected results?
David Shone: The pilot, conducted with REGnosys, involved a sub-group of ISLA’s digital working group working together to define the framework for a cash-collateralised, delivery-versus-payment (DvP) settled, securities loan. The process consisted of conceptual modelling starting from the Global Master Securities Lending Agreements (GMSLA) legal template, discussion and agreement with the working group, before encoding the product and transaction into the Common Domain Model (CDM). The pilot also looked at two fundamental business events in our asset class
Martin Walker: Modelling some of the key characteristics and life cycle events of securities lending trades to the point where it is possible to translate existing data formats into the CDM format. It should be well defined enough to do that translation from data in existing multiple systems.
Chris Rayner: FIS was pleased to support this important pilot project to complete a meaningful proof of concept in a compressed time frame. It’s been a great example of how diverse players in the market can come together with a common purpose. The primary aim was to model the allocation process as applied by an agent lender during a trade execution. To keep the scope of the pilot achievable, only a cash DvP trade was considered. Initial thoughts about how to model settlement actions and collateral provisions were also discussed. By the end of the pilot, the working group has agreed upon a set of core concepts for trade execution and allocations that are modelled into the CDM.
Julian Eyre: While not responding to the details of the functionality that was delivered for the pilot, this phase validated the extensibility and flexibility of the core CDM model. The Rosetta application and the underlying artefacts in the platform enabled the rapid application development of a securities lending transaction model that starts to meet the requirements of the market.
How difficult was it to take a CDM for derivatives and adapt it for securities finance?
Walker: The work isn’t finished but it has been a very good start. There are plenty of more complex scenarios to model. There are a lot of similarities between derivatives and securities finance in terms of product characteristics and complexity. However, securities finance does have a much greater focus on securities settlement, which presented some interesting challenges.
Harry McAllister: The CDM was originally developed in the context of over-the-counter (OTC) derivatives, however, the CDM is a highly-granular model which by design represents products in terms of core economic components and operations, which can, in turn, be composed into any number of higher-level representations. From the modelling point of view, assets such as cash and securities are also essential to derivatives – likewise concepts such as contract formation, settlement and allocation. Thus many of the building blocks required to construct the securities financing model – together with the methods for deploying them – already existed at the commencement of the exercise. What the technical team didn’t have at the outset was a complete understanding of securities financing markets. The value of the working group has been as a forum allowing subject matter experts to review and critique the model in the process of construction – bringing experience and insight to bear in resolving issues of interpretation, emphasising which features are essential, which require adjustment and which are potentially redundant. The speed with which the group has been able to make progress is a validation of the CDM, combining a proven technical base with a modelling framework which allows small and medium enterprises to focus on defining business functionality.
Eyre: The core CDM developed by the International Securities and Derivatives Association (ISDA) for derivatives leverages attributes and data elements from FpML. While there is no core application programming interface/standardised messaging model for securities finance certain products (such as basic repo) can be supported by FpML hence the CDM. The success of the securities lending transaction pilot demonstrates the relative ease of extending the model for securities finance. Further exploration will be needed to validate that all securities finance processes and functions can be supported by the CDM. Given the ‘product’ neutral objectives of the CDM, the core cash flow models and business processes will expand over time to address far wider market functions such as a generic ‘lending payout’.
Rayner: A great question. There is a lot of commonality between the derivatives and securities financing worlds, and these aspects of the ISDA CDM have been leveraged and reused in the International Securities Lending Association (ISLA) CDM. This not only reduces the effort compared to starting with a blank sheet of paper, but it also means we can better connect the dots between securities finance and derivatives. It quickly became clear to the working group how securities financing specific elements should be added to the ISLA CDM to adapt to the specific requirements in this business and make it easier for ISLA members to implement the CDM.
What was also interesting was how often the group agreed that certain aspects of the business were not as well understood as they should be. For example, block trades, where an execution has multiple allocations associated with it, came under scrutiny concerning the actual terms of the agreement and practical implementation. This highlights another very important benefit that the CDM will give us as a market – consistency. Consistent terminology, consistent validation, consistent data. This standardisation will increase the ability of all market participants to automate processing, which in turn will reduce costs, trading risks and breaks.
Shone: One of the CDM’s primary design principles is one of composability and re-usability. I think this assisted momentum in the pilot, with the ability to pick up, extend or adapt existing components where there was an analogous concept. Where securities lending has a more developed concept of an event, for instance, the process of allocating a trade, our pilot resulted in the rewriting of the existing components to be able to cater for both derivatives and securities lending. This is an example where a fully-developed CDM, catering for multiple products, truly will allow for interoperability between distinct asset classes; this has to be invaluable for any firm trading multiple products, especially the smaller firms who may want one application that deals with all of their products due to licencing costs: if that application uses the CDM it can do that easily without conflicting data models for each asset class.
Is the adoption of the finalised CDM expected to be quick or gradual? Is there a critical mass that must be achieved for the merit of the model to be realised?
Saied Attarian: The pilot has demonstrated that adoption of CDM is a relatively straightforward exercise and can have a positive impact on the market. However, this was a proof of concept only covering limited economic scenarios. We aim to continue working with the participants of the working group to further develop the product model to cover additional scenarios, as well as the implementation of other functionalities such as re-allocation and billing. Once this is done the pace of adoption is down to market participants. However, the merit of the model can be realised immediately by an adopter through the standardisation of messages and functions
Shone: As with most technologies I expect that in the immediate term adoption will be incremental. Indeed, I believe firms will want to take advantage of the composable design to adapt parts of their infrastructure piece by piece, demonstrating real value add and cost savings at each increment. A big bang approach would most-likely be costly to adapting firms in both time and money, whereas an incremental approach allows for expertise to be developed, benefits to be cumulatively realised and ongoing continuous improvement. I suspect that we will initially see pockets of adoption, either within firms between two or more related systems or between firms, say, for example, a financial institution and one of their post-trade matching platforms. When those pockets start to overlap you will then have critical mass and we should then see a runaway effect. When that will be is hard to say and highly-dependent on individual firms’ resourcing constraints and strategic vision, and external factors such as regulation and market disruption.
Rayner: I would expect it to be gradual. The past few years have seen a lot of regulatory initiatives that are still ongoing and CDM will inevitably compete for priority with other initiatives. I expect take-up to gather momentum once regulatory projects such as the Securities Financing Transactions Regulation (SFTR) and the Central Securities Depository Regulation are behind us.
However, there are immediate benefits even within firms from adopting CDM even before any critical mass is achieved in the market. Consistent canonical data structures and processes will reduce friction internally within firms as well as supporting external flows. As a software vendor, we should feel its benefits very quickly as it will reduce the amount of interface work that we need to undertake and assist in the consolidation of data across silos.
Julian Eyre: Bank adoption of incremental CDM functionality will deliver business benefits for internal processes. This approach could also help any adopter accelerate familiarisation with the technology.
The adoption of CDM can work in various ways, partly depending on the maturity of the CDM:
One: through an industry infrastructure provider potentially incentivising CDM usage or through the emergence of a new market platform designed around the CDM model.
Two: internally within a bank business vertical (derivatives or securities lending/repo) to reduce E2E internal friction and costs
Three: horizontally across an industry (derivatives or securities lending for example) where like-minded counterparties recognise the value in reducing reconciliations and exceptions between them, and accelerating enhanced digitisation of the industry
Four: holistically across a bank to add value within a business vertical as well as risk (Fundamental Review of the Trading Book (FRTB) benefits), regulatory reporting, client integration (better client service through efficiency etc), especially when cross-product (i.e. derivatives, securities lending and repo)
The merit of the model can also be assessed in several ways. The most critical will be an adopter’s business benefit.
In the 1990s the FIX protocol developed, where adoption of the industry standard was initiated by a global asset manager and a broker to streamline end of day processes (allocations specifically). Further brokers accelerated adoption to benefit from the asset manager order flow (and cost benefits). In the case of securities finance and the CDM, the embracing of the technology by early adopters may demonstrate similar business benefits.
Walker: Widespread adoption is most likely to take a path that is initially gradual until you reach a critical mass. That is the point you are likely to see a lot of activity as people try to avoid being left behind. It is tricky to pinpoint what the exact tipping point will be but we will certainly know when we get there.
How successful has the ISDA CDM been for making derivatives transactions more efficient? Can that success be replicated or out-done for securities finance?
Rayner: The ISLA CDM and the pilot process have been construed in such a way to build on and learn from the experiences of the derivatives project. There is certainly great potential in the project and as a vendor servicing both derivatives and securities finance, the common DNA of the CDM across both businesses benefits both us and our clients. When the efficiencies of the modelling are recognised by more participants, I think we’ll see rapid uptake.
McAllister: I think it’s fair to say that CDM is still a work in progress for the OTC derivatives industry, but one which we expect to deliver over the coming year or so as we see the first commercial services referencing CDM coming to market, promising efficiencies in post-trade processing with integral regulatory compliance. There is no reason why similar benefits cannot be realised in the securities finance domain, where there is arguably even more potential to eliminate operational inefficiencies by leveraging standardised definitions of product features and lifecycle events.
Eyre: It is my understanding that the adoption model for derivatives is not clear. However, there is an active initiative to use the CDM for regulatory reporting (partly enabled by the EMIR rewrite).
CDM hackathon participants have looked at DLT solutions using the CDM as an efficiency mechanism. The challenge here is that unless the CDM standards are adopted, use of a DLT platform remains a significant hurdle in most cases.
Where does the most significant efficiency challenge lie for securities finance? Is it to pick off specific complex high-cost processes (allocations, legal contracts or settlements and payments?) Or to consider a wider lower friction end to end process, initially, implemented for a product sub-set?
Walker: It is still early days for the ISDA CDM. So, it is premature to talk about its impact. However, Looking at the fundamentals of the securities finance and derivatives markets, there is a great deal of potential in securities finance, possibly a great deal more. This is because various factors including regulation and the success of various market infrastructure providers have already driven a great deal of standardisation over the last 10+ years in derivatives. Clearing a derivatives trade or reporting a CDS to DTCC’s trade information warehouse, for instance, are both powerful agents of standardisation.
Shone: I’m not in a position to comment on behalf of ISDA’s membership, however, I think the ISDA CDM has had a strong year. Not only have at least two pilots been carried out at other trade associations (ICMA is the other), but there has also been several awards, white papers and announcements from other organisations that have begun to use the CDM in some capacity. For instance, Goldman Sachs recently announced its Legend platform which has been used to extend pieces of the CDM, while REGnosys and ISDA won the G20 Techsprint for regulatory reporting.
In the best-case scenario, what would the CDM do for the securities finance industry?
Ben Smith: I think one of the biggest trends in our industry is around using technology to democratise securities finance. But we can’t achieve that until industry participants are using the same data formats. The CDM is one way we can start to standardise the most basic building blocks that our industry is built around.
With that foundation in place, we can build stronger, smarter services for a much wider pool of beneficial owners. I’ve seen first-hand at Sharegain how having the right data structure and standards in place cascades through every layer of your software. An end-user may never see it, but their experience is ultimately defined by how well their data is structured.
This is the best-case scenario for the CDM: that it allows us to radically expand the pool of participants, provide them with a seamless user experience and, ultimately, to democratise our industry.
Attarian: The main benefit of using CDM is the introduction of data and function standardisation across the securities finance industry. During discussions with the wider working group, it was apparent that different methodologies are used across the market for parts of securities financing transactions and its collateral provisions.
This generates reconciliation over-heads and prevents the smooth interoperability of different point solution across the value chain. Wider adoption of CDM will help to bring unification and consistency across the market which will result in increased efficiency and reduction in post-trade operating costs. By providing a standardised input layer, the CDM will also facilitate regulatory alignment and data collection processes, which is a new challenge that the industry is facing with SFTR.
Walker: A securities finance industry based on more standard models, where it is clear what is the front-to-back cost of deviations, could sweep away a great deal of existing infrastructure and business processes. Not to mention drive considerably more automation. Whether that will come from the inside e.g. CDM or externally i.e. more regulation or a combination remains to be seen.
Rayner: By providing common terminology and data structure for securities financing agreements, collateral schedules, products, transactions and lifecycle events, the CDM will facilitate more automation in the industry, with less chance of breaks between counterparties. One key benefit would be the standardisation within CDM of collateral schedules. This would greatly improve the scope for optimisation of inventory and collateral across cleared, triparty and bilateral relationships. Overall CDM will inevitably lower costs and help remove redundant processes such as reconciliations and break resolutions, fails management and so forth.
Eyre: The best-case scenario is to remove more than 40 percent of the cost from operational processes. This is achieved through a selection of mechanisms:
• A reduction in errors as the data will be standardised across the industry.
• Improved client service as the adoption of CDM will improve data and quality of service – less time spent on managing fails and exceptions
• Improved reputational risk – CDM will allow lower cost, more accurate regulatory reporting
• Better risk reporting – CDM delivers a standardised representation of trades, cashflows and helps align with FRTB challenges
• A reduction in treasury costs – CDM can provide enhanced real-time treasury and funding processes
• Improved financial reporting – due to the better quality of data within the organisation
• New cloud-based (DLT?) shared service or interoperable platform to dramatically reduce IT costs
One of the benefits of the CDM is the opportunity to increase automation, which in turn could solve or mitigate several market pain points today. How much of an impact do you see this CDM having on costs and overall market efficiency over the next year or so?
Sunil Challa: The core benefit in leveraging CDM for post-trade services is for market participants to agree on a common processing standard that would allow for;
(a) reduced duplication through the use of authoritative data,
(b) externalise non-differentiated services, and
(c) unlock value through reduced friction points.
We see the first two in promoting operational, cost efficiencies which would in turn enable capital and funding efficiencies. By allowing these models to extend beyond functional and product silos, it provides market participants with a reference architecture to target for post-trade services. In such an environment where services operate with standardised data and process definitions, clients – FMIs, buy and sell-side firms, can seek best of breed solutions to fit their operating model and make decisions with the understanding that standards not only ensure interoperability across service providers but allow them an open path for future innovations in the marketplace.
Walker: A core driver for reconciliations and breaks in securities finance, is the diversity of data models. Every time there is a translation between models, there is the scope for error.
Widespread adoption of a CDM would reduce the scope for those types of errors. What you need though to allow greater automation to be built, including automation that can be built on top of core systems, is both data model standardisation and more open, standardised APIs.
Shone: I couldn’t agree with Martin more. One of the steps that ISLA would like to take following the pilot is to work with member firms on their pain points and help them in developing business cases for CDM adoption. With any number of possible routes to take through developing and adopting the complete CDM, firms must identify the path that gives them the optimum business benefit for their efforts.
Attarian: Adoption of the CDM can have a huge impact on the market. It will not only introduce standardisation across the market, but it will also further introduce efficiency especially during post-trade life cycle events. If market participants are using the same data standards to represent securities financing trades, the pain of on-boarding and reconciliation of trade activities will be eliminated. Furthermore, by using the same standards and methodologies, reconciliation and post-trade activities that range from settlement, re-allocation, mark-to-market, rerate, billing and more, will become far easier and less time will be spent creating complex inhouse solutions to perform these tasks.
Rayner: There is absolutely the scope for CDM to help reduce costs in the industry. Realistically we may need to look out further than the next year to see the benefits, however. This is an initiative that will have real and tangible benefits over the medium to long term horizon and FIS as a key partner to the market is happy to support and drive the initiative as it is completed and adopted into our products and services.
Eyre: To assess the impacts and business benefits the industry needs a clear understanding of the current state, and then to build a business case for CDM adoption.
In-year benefits will depend on the level of bank activity and buy-in; if the banks adopt a community approach efficiency can be accelerated, however, if the banks rely on vendors to implement the CDM the time to value will likely be quite extended. If the market/banks can assess benefits and implement the CDM model incrementally, then in-year benefits will be achieved.
Is the adoption of a CDM for securities finance an essential step on the road to the wider use of DLT and is that the ultimate destination? Or, will DLT become one of several common routes to market thanks to the CDM?
Attarian: DLT is developing at a pace all across the financial services industry. Simulations and prototypes are replacing theoretical concepts surrounding the distributed ledger technology. The first step in creating such technology in the securities finance industry is harmonisation and standardisation. CDM can be a solution to creating standardisation, and the first step in adopting DLT by the wider market. However, CDM in itself can be adopted as a market data standard to increase efficiency across the sector without the adoption of DLT.
Rayner: We expect that CDM is an evolutionary step towards the use of DLT more widely in the market. For example, it provides the building blocks of standardisation of data structures and processes that are needed for smart contracts. It would be rather more
Walker: The origins of the ISDA CDM had a strong theme of facilitating the growth of DLT based systems. I think the world has moved on since then and the general feeling is that a CDM can facilitate developments in infrastructure regardless of the underlying technology. Also, where we see progress towards production of DLT systems, they are part of a hybrid rather than a purely DLT model. Something I wrote about almost four years ago with Anton Semenov of Commerzbank’s blockchain team in “bridging the gap between investment banking infrastructure and distributed ledgers”.
Looking at Broadridge’s work in securities finance DLT, we have also adopted a pragmatic hybrid approach where the Broadridge’s DLT solutions are designed to work with existing infrastructure
Shone: It would be wrong of me to say that CDM is essential to implementing solutions through a DLT, however, I would say it would make it significantly smoother and easier to adopt such a solution if the data structure is standardised. Similarly, with smart contracts on top of a DLT, a common language to build them from is an advantage.
DLT solutions are possible without a CDM; HQLAx is a good example known to the industry where a DLT solution has been used to good effect. However, at the end of the day, a DLT is another type of network and entering into any communications structure is only truly effective if all nodes on that network are built using standardised building blocks and communicate with a common language.
Where processing happens within multiple nodes on a DLT, you must process the same event in the same way on all nodes within that network. CDM based lifecycle events support unambiguous consistency of output, applying a standard practice throughout rather than reliance on convoluted translations.
Eyre: I see the CDM approach as being a fundamental foundation for future securities finance platforms. Initially, CDM adaptors may provide a tactical technical solution to some pressing industry challenges. A longer-term trajectory towards DLT seems likely.
Without a standardised market CDM, there is a high risk that multiple DLT solutions may evolve, where none are interoperable. Or at least the interoperability will be inefficient and not solve for one of the key DLT benefits of mutualised platform costs.
If the CDM is adopted as the standardised model (ie CDM native), then DLT vendors, who build to the model, will deliver interoperability, and contribute to the success of DLT in the future.
Lee Braine: As DLT matures, its technology is being considered for more applications in financial services. There is a good choice of DLT platforms, including Fabric, Corda and enterprise versions of Ethereum. And there is a growing ecosystem of fintechs, bigtechs and consultancies with the capability to design and build enterprise-grade solutions on those DLT platforms.
However, the challenge often comes in constructing viable business cases, where several factors have to align across the industry. For example, having a sufficient number of market participants to achieve momentum in the market and providing adoption scenarios that allow market participants sufficient choice of how deeply to integrate with the new technology.
The CDM provides a common foundation, in terms of standardised processes and standardised data. Such standardisation is typically a predicate for the adoption of DLT because the technology often requires common ‘smart contract’ code and shared ledgers. This means the CDM can act as an enabler for DLT.
However, the key architectural step is to determine whether, for a particular use case, it is better to use a centralised model (such as a market infrastructure firm running common processes on a centralised authoritative data store and market participants accessing that data store) or a decentralised model (such as each market participant running common processes on its local authoritative data store).
