Machine learning and AI proved to be popular topics at the second Securities Finance Technology Symposium in St Paul’s, London, but some suggest that technology is still not being used to its full advantage
For the second year running, Securities Lending Times hosted its Securities Finance Technology Symposium in London. Hot topics included: data, machine learning, artificial intelligence (AI), collateral optimisation and management, liquidity platforms, and of course, the Securities Financing Transactions Regulation (SFTR). Technology was at the heart of the conference but the main consensus seemed to be that the securities finance industry is still not utilising technology to its full advantage and is perhaps behind in this area compared to other industries.
Kicking off the conference, speakers discussed data on a panel entitled ‘Data - The Fuel for Automation’. David Lewis, senior director at FIS, noted that there is lots of interesting data out there, which gets our attention, brings innovation and drives competition. However, he pointed out that “there is a lot to be done in the basement, we won’t get the basics right unless the industry sorts the foundations”.
He argued: “We talk about AI and machine learning but until we strengthen the foundations of our basic data, we won’t see transformational change.”
Lewis explained that the SFTR is a “helpful driver as purses are very much more open when you use a regulatory flag. People are investing in getting the regulations right which has the potential to dramatically improve the quality of data held and exchanged between counterparties”.
“I would focus on getting the building blocks right; we can play with the fun stuff but we can’t get the yield from the kind of automated efficiency that we are capable of without getting things like standard settlement structures right.”
The moderator, Andrew Dyson, ISLA CEO, asked the panel if the regulatory agenda gets more focus from a budget perspective. Dyson said: “Is it a constraint that there is not the budget for some of these things?”
In response, Paul Wilson, managing director at IHS Markit, explained: “You can’t avoid the regulatory agenda because by definition you have to otherwise you don’t have a business. My challenge to all of you is that there has been a huge amount of automation that has occurred but we have tended to automate existing processes.”
Wilson continued: “We have not stepped outside of the box and addressed Agency Lending Disclosure (ALD). The industry has spent vast sums on automation but we have missed a trick and ALD has the potential to drive some out of business.”
Panellists also discussed how they use data, and Ed Oliver, managing director, product development, eSecLending, said: “eSecLending is a big user of data and it is a core element of what we do. Over the next two years, we are going to become a huge giver of data as well. As users of data, we need to find ways of finding it interesting and relevant for what we need.”
In the SFTR Regulation and Reporting panel, Harpreet Bains, executive director, agent lending, global product head, J.P. Morgan cited: “The industry needs to keep the impact of getting SFTR wrong in sharp focus.”
The moderator, Val Wotton, managing director of Deriv/SERV, DTCC, asked panellists what lessons can be learned from previous regulation implementation and how those can be applied to SFTR.
Bains highlighted that whilst it’s understood that regulators have indicated they are sympathetic to challenges in the new regulatory environment, we can still expect to get severe fines for misreporting.
She said: “We have to be really careful that we are not copying or manufacturing data to make a fully matched report—a matched report is not necessarily an accurate or compliant report.”
In another panel on SFTR, Mark Steadman, executive director of European head of product development and change management, DTCC, commented: “History tells us that trade repositories are much better prepared for SFTR than other regulations in the past.”
Steadman discussed the difference in preparation to SFTR, compared to the readiness shown with the European Market Infrastructure Regulation (EMIR), indicating that this time around, the regulators might not be so forgiving as they were under EMIR on day one, as the industry has had a long time to implement SFTR.
Regarding SFTR, John Kernan, senior vice president and head of product development at Regis-TR said he would like to take a slightly more optimistic view concerning SFTR, hoping that national competent authorities (NCAs) are pragmatic and give a six month grace period when reviewing data quality issues.
The moderator, Pierre Khemdoudi, managing director of global head of equities, data and analytics products at IHS Markit, asked panellists what kind of expectations there would be from regulators on day one of SFTR implementation and how their firms were preparing.
Kernan highlighted that Regis-TR had recently put out its user acceptance testing (UAT) environment and indicated the roll out of the trading module is scheduled in London, Luxembourg, Frankfurt and Madrid over the coming weeks.
Craig Laird, executive director of Morgan Stanley said Morgan Stanley’s core repo and securities lending build is “progressing well”, although there remains a number of open questions concerning lifecycle events and specific fields that are pending regulatory clarification. He added: “We are learning from our individual data flows. We are also looking at the cross over with the second Markets in Financial Instruments Directive (MiFID II) reporting for ESCB trades and how SFTR fields will be mapped to MiFID II.”
Although he indicated the relatively low volume and the complexity involved in switching eligibility between the two regulations should not be underestimated.
Khemdoudi then went on to ask how ready the industry is for SFTR implementation and what areas still need the most improvement.
Seb Malik, head of financial law at Market FinReg, said: “With UAT testing, some issues we are facing [as an industry] are to do with booking of repos. Repos, we quite often find, are booked to securities lending. But nothing will match if you’re not booking it correctly.”
Khemdoudi then asked what the panellist would advise market participants if they indicated they were not ready for SFTR. Laird emphasised the importance of reference data accuracy and control as well as the significance of a robust control environment and having the right team of people to work with.
Malik mirrored: “Training is key, whoever you use is absolutely fundamental, from your IT team all the way up to management. Make sure everyone understands what is going on.”
He added: “Work with the working groups, such as the ISLA and International Capital Market Association.”
“We want to make sure we are all working to the same set of industry standards. Across every corner of Europe we want to be working to the same interpretation of fields.”
The conference also had a panel on post-trade, in which Rob Frost, global head of product development, Pirum Systems, argued that Distributed ledger technology (DLT) may solve individual problems in post-trade, but it won’t replace all current systems in the medium term.
Frost indicated that DLT cannot survive on its own and will need to connect with existing systems, especially when concerning the navigation of regulations such as the Central Securities Depositories Regulation (CSDR) and SFTR.
The moderator, Dan Barnes, editor of The Desk, went on to ask the panellist who would drive the adoption of DLT and similar technologies.
Laurence Marshall, COO at EquiLend, suggested that the adoption of technology is better today than it has been for the last few years.
Marshall said: “With new technology coming out all the time, and more and more data becoming available, there is an important requirement to look at different infrastructures.”
Frost explained that new technologies can offer a lot to help avoid fines and solve regulatory reporting issues. He said: “There was a struggle to get adoption of technology in post-trade in the past, but now the back office and front office’s obligations are converging, that is what we are seeing at Pirum, anyway.”
Marshall cited it was important to understand that “answers are not always found in technology”.
He highlighted: “The marketplace is becoming more complex and difficult with a wider variety of solutions available within post-trade, whether it’s to help with structures or digitalisation. But things are never as quick as you would like them to be.”
Barnes then went on to ask the panellists what were the main pressures that the post-trade sphere was facing in terms of technology.
Dave Grace, head of post-trade for the UK at Capco, said that CSDR will help “harmonise and standardise market practices, while it has already allowed T2S to survive and thrive.”
Grace explained: “The more the European Union regulations such as EMIR and CSDR foster a competitive environment, the more competitors there will be entering the game.”
He added: “In terms of technology, there will come a point where legacy technology will become horrifically old and unmanageable. Firms will have to decide if they want to continue spending money maintaining them. There are currently a lot of legacy systems holding back-offices together across the industry.”
Grace cited that regulation should be used as a catalyst for business model and technology change, specifically regulations such as SFTR and CSDR.
When asked what they would advise firms to prioritise in preparation for the growth of technology, panellists suggested treating coming regulations like CSDR and SFTR as an obligation to duplicate industry requirements.
Frost suggested to leverage connectivity and projects that are already underway, as well as to participate in working groups and make sure that all teams across post-trade lifecycle are aware of the tools that are available to them.
Later in the day, Tammy Phillips, founder and CEO of Asterisk Network Solutions, said: “Our peer to peer network enables beneficial owners or hedge funds to achieve direct market access or access via an agent. We have provided the opportunity for prime brokers to alter their legacy structure to move from a principal in a securities lending transaction to becoming an agent. One of the key reasons we did this was because the market has evolved in such a way that the vast majority of outstanding GC transactions are loss-making transactions for prime brokers.”
“The market has continued to recover since the financial crisis and we have adjusted to 100’s of new regulations, we have been doing well and there were collective pats on the shoulder… it’s often difficult to encourage an organisation to do something new when times are good and revenues are up year on year. However, over the course of the last year, that peak stopped, we have to change the status quo from a year ago today, the market is changing quicker than we are. We need to have the determination and courage to deliver something for the market, not for an individual organisation.”
Meanwhile, Matt Wolfe, vice president, business development, OCC, predicted the industry is moving towards a more open network if the proper privacy and protocols are in place. He emphasied that it is challenging because many in the industry have legacy systems, including OCC but they are working on an initiative to replace it. In terms of SFTR, Wolfe explained that the regulation is highlighting the problem that many beneficial owners “are not comfortable having their identity disclosed by their counterparty, which is a big challenge that needs to be addressed.”
Turning to technology, he said: “DLT has a lot of opportunities and areas that can be addressed. Standardisation is where everyone on the network agrees on a standard process as well as on the inputs. The results of that agreed upon set of inputs and processes are always going to reach the same results.”
Another speaker said that they are pessimistic about DLT: “There are a lot of challenges to implementing it. Think about how much effort there would be in trying to get everyone to that progress. There is also environmental concern around DLT—they are not efficient. The network itself takes up a lot of powerful engines. I think the adoption rate of DLT will be slow.”
Towards the end of the panel, Wolfe referred to a quote which suggests that computers are incredibly fast, accurate and stupid, while humans are incredibly inaccurate, slow and brilliant, and only together can they work really well.
Wolfe highlighted that there is a role for humans to be involved with machine learning, especially when the costs of the results being wrong are high or when data is sparse.
He concluded: “Change is happening faster and we need to take it upon ourselves to do a bit of self-learning [for technology]. I would encourage you all to pursue and investigate these ideas.”
