It has been 15 years since compression emerged, and it continues to evolve at a rapid rate. At a forum convened by Risk and sponsored by Capitalab, our panel discusses current hurdles, upcoming challenges and whether innovation is keeping pace with the activity, concluding with what each party can do to improve this ever-changing landscape.
- Gavin Jackson, Co-founder, Capitalab
- David Bachelier, Co-founder, Capitalab
- Allan Guild, Global Head of Execution Services and Regulatory Change,
- FX and Commodities, HSBC
- Mike Sweeting, Global Head of Risk Mitigation, BGC Partners
- Philip Staddon, EMEA Head of XVA Management, Credit Suisse
- Duncan Lillie, XVA Trader, Citi
- Edward Ground, XVA Trader and Markets Collateral Manager, JP Morgan
- Magnus Lindahl, XVA Trader, Nomura
Risk: Could you provide some background on capital and margin optimisation, and in particular compression?
David Bachelier, Capitalab: Compression is an activity that began in 2003. At the beginning it was more an IT and operational tool to reduce trade processing and the number of payments on derivatives; there was no leverage ratio. In 2008, with the Lehman Brothers default and the subsequent crisis, governments were forced to act. The Group of 20 Pittsburgh summit in 2009 – which led to the Basel III leverage ratio framework – forced banks to set aside Tier 1 capital in proportion to the size of their derivatives books.
At that point, compression became a core business activity and, to date, we have seen quadrillions of compressed notional. A lot of new techniques have appeared in the market – including at Capitalab, which was founded in 2015 – with the compression of new products such as swaptions, which were untouched by compression techniques at the time. We have also seen new techniques from a growing number of vendors – trade replacements, notional amendments and coupon blending. Clearing houses have also developed compression services; for example, solo compression at LCH.
So the landscape of compression activity has changed dramatically in recent years and it’s evolving fast. We need to ask ourselves – what will be the next challenges, what are the current hurdles? Is there enough innovation and how can we improve the landscape?
Risk: Is the compression process fully commoditised or is there scope for further innovation in this space?
Allan Guild, HSBC: We need to split innovation into two parts. First, innovation on the modelling side – companies are seeking to achieve a better result, but I don’t think there is consensus on what the target is. Second, innovation on the IT and operational side.
Picking up on what David said about how compression started as an IT and operational function, when you look at the primary driver now – the desire to reduce capital – the algorithms have become more complicated. From HSBC’s point of view, we almost want to go back to the start, back to technology and operations. We are interested in innovation around this area – how compression can become less time-consuming, how it can reduce operational risk and become less error-prone, and how it can be more of a straight-through process.
Philip Staddon, Credit Suisse: It is true that, if you increase the efficiency of compression, by definition it gets less operationally intense because there are fewer trades.
David Bachelier: There are two separate topics – first, how can we improve the operational process of compression? Second, how can we improve the efficiency of compression algorithms and techniques ?
One interesting development being studied is a direct connection from vendors to banks’ back-office systems, rather than asking the banks to submit the trade and/or the risks, which is resource-intensive. If we could do the background work and push propositions of compression without requiring banks to be involved in every preparatory step that would be a big saving. It’s an ideal scenario, and unlikely to happen soon. It’s a long shot, but financial products markup language (FpML) technology can help that. There are a lot of angles that we are trying to explore in that direction.
Philip Staddon: We talked about the benefits to having a central repository, but that’s the reverse of what you’re suggesting – putting a load of trades from legacy positions on to a central repository. That is a lot of work for banks and I can’t see a huge upside to it. We’ve had to do it for new trades under global regulations from the International Organization of Securities Commissions (Iosco), but the legacy portfolio is always going to be going down. So I’m not convinced that it’s worth the effort.
Coming back to efficiency, one of the biggest things we found – again, it depends on your portfolio because it has changed over time – is ensuring you’ve done everything internally as efficiently as you can. Originally, banks were fairly siloed; for example, you might find that a dollar trader in each region would be doing their own submissions rather than doing it in a co-ordinated fashion. So we centralised it around that desk and I’m assuming that’s what other banks have done, and that it’s achievable without too many problems. In effect, it means you make one submission instead of five. Your process is constrained by the number of silos because each one has to be risk flat. So if you only make one submission, you will achieve more compression. By doing that, we double the reduction, so if banks aren’t doing that then I would suggest this is the first step to take. I would characterise this as a ‘partial hubbing method’.
Risk: How else can we make compression more efficient?
Duncan Lillie, Citi: When you run a large compression cycle, you need to make sure that the majority of the trades are automatically processed, because manually processing more than 100 or so is too expensive, operationally.
Risk: Are you happy with the ratio you achieve from compression runs?
Magnus Lindahl, Nomura: When we run multilateral systems, the certainty of execution often leads to people being very conservative. They set fixing tolerances daily out to two-and-a-half years because they might not trust their own curve build, but there are many reasons why people are conservative in a process that is done automatically.
Duncan Lillie: In terms of the effectiveness of an algorithm, having multiple vendors should help because you may find that one vendor is able to achieve a greater reduction than the others. The main barrier to entry for vendors is the congestion in the compression calendar; I think an industry-wide discussion on co-ordination may be required.
Mike Sweeting, Capitalab: We have found issues with co-ordination. The availability of slots at central counterparties (CCPs) is limited and that tends to cause issues for when we can schedule activity versus when another provider may have something scheduled. So we might want to make a more multilateral run with swaptions and swaps together, but then there might be another vendor’s swaps run in the middle that gets in the way.
Risk: Is there any way of tackling that from either the bank or the vendor side or is it the CCP that needs to do something?
Allan Guild: Fundamentally, the runs are manually intensive, which means they take as long as they need to. Theoretically, if everything was automated and straight through, a compression run should take a matter of minutes, maybe an hour or two, and the verification step is taking up all of that time.
Mike Sweeting: We should focus on reducing the pairing stages and the manual processes running up to that. One of the problems a compression provider has is that banks don’t all use the same systems.
Magnus Lindahl: The banks might also have different operational preferences. One bank might want to opt for ‘end of the day’, one bank might opt for ‘intraday’. Depending on how their risk systems are set up, somebody might take a feed from MarketWire, somebody might book straight into the risk system and feed in the other direction.
Certain more advanced techniques that look at combining compression and initial margin optimisation, for example, or risk reduction on the back book, are quite far away. This is because the dealer community is also still quite a way from being able to support straight-through processing (STP).
Mike Sweeting: We understand that the more products we can combine, the better for participants. Also the faster we can move it from the back end – one of the pleasures of working at Capitalab is that we can leverage existing post-trade STP systems from our BGC parent when we book new trades.
Allan Guild: And the methodologies increasingly use trade events, such as amending or terminating rather than just booking new trades, so there’s very limited capability to book automatically.
Philip Staddon: If we consider slots with CCPs, one possibility is – rather than doing dollars, euros and sterling on different days – why not do all currencies together in one day? It would be preferable to sign off three processes in parallel rather than doing one thing one day, one the next and another the next. It would be more efficient that way.
David Bachelier: It’s a limitation at the CCP – they can’t really do every day.
Gavin Jackson, Capitalab: The CCP can support two or more per day, but the restrictions are more around where traders are based for risk sign-off.
Philip Staddon: If it’s hubbed, you can deal with it – if everybody is hubbed, you only need one person signing off. But it comes down to the way banks are set up.
Edward Ground, JP Morgan: I don’t think hubbing is complete yet. Some parties are much more advanced than others. So do you need critical mass to help make this work?
Philip Staddon: No, because you’d always exclude some trades if you wanted to, and still gain efficiency.
Edward Ground: But what if you exclude a large fraction?
Philip Staddon: Then efficiency comes back to hubbing.
Risk: How does hubbing work in this context?
Philip Staddon: Imagine you have four traders at a bank, all of whom have euro swaps. If they all submit to Capitalab, they would be submitting in four silos even though they work for the same company. They will reduce notional across the market, but it will be less efficient because if they can submit it all in one go, it gives more flexibility to the algorithm that is doing the optimisation, so you’ll get better reduction.
It also means that these silos will have different tolerances. So you’ll have to do some fine-tuning on risk tolerances in the short end. Ultimately, it means you’ll get very little done, however, if it goes to one central desk that does the whole thing for all four traders; it can take the netting effect of the whole portfolio and deal with all the tear-ups and the notional reductions.
Allan Guild: I agree, with the exception of the premise that the four desks all work for the same company. We said we’re doing this for leverage ratio, which is on an entity basis. So could a multi-entity firm do it more efficiently by combining and then distributing?
Philip Staddon: Maybe you can only do it within a legal entity due to practical issues.
Allan Guild: Yes, but a lot of us are running multiple events. Another thing you can do is a form of virtual hubbing and then distribution. But coming back to some of the points we had earlier, it’s complicated.
Edward Ground: You’ve got foreign exchange in flow, hybrids, multi-asset books – they’re not necessarily together.
David Bachelier: So, on that topic, we do offer the possibility to include internal trades within the compression cycle – simulating a single hub with multiple legal entities, even though you have two hubs and the internals. This is quite efficient for compressing multiproduct trades submitted across multiple entities of the same bank.
Duncan Lillie: An innovation I have discussed recently is that booking new trades between banks could be difficult – how do you agree the valuation of the trades? But, if you’re creating trades between desks within a bank, you’re much more likely to reach a consensus on the value. I agree with David, in that including legacy internal deals with a siloed submission can help the algorithm to achieve good results. Allowing it to propose new trades between those silos could also provide a boost to efficiency, particularly in cases where there’s a long tail of silos with small trade volumes.
Risk: Working on the assumption that a greater number of vendors is better for building efficiency, is there an appreciable difference in how much can be compressed between the vendors? Would it make you want to use one vendor over another?
Duncan Lillie: At the moment, different vendors are at different stages of their development, so it’s not necessarily fair to compare them. It’s quite hard to do a scientific test. It’s not as if I could submit an exact trade to this vendor and the same trade to that vendor, which would give you a feel for what you’d expect to get back.
Risk: They are not directly comparable – each has its own unique selling point. In that case, is it about balancing what works for individual banks in certain circumstances?
David Bachelier: My view is that, if you are a bank and you have, say, four different offerings in compression with different algorithms, tools and attributes, the best way to benefit at the maximum level from those four is to give liquidity to each. Coming back to the difficulty in scheduling the LCH runs, it’s now quite crowded – there are so many cycles that banks are unable to submit all their runs fully to each and every vendor. To get around it, they should give liquidity to each of the vendors in a systematic way – one vendor doing a combined compression, another doing a trade replacement, a third doing something else. That’s how you get the most benefit from all the different solutions. Of course, it is then up to the vendors to deliver the best results to keep that liquidity.
Risk: What is the status of bilateral compression processes within banks?
Duncan Lillie: The need for bilateral interest rate swap compressions has reduced dramatically due to the restrictions on new bilateral business, the compression cycles that happened in the past and of course the natural roll-off rate. We do negotiate compressions of bilateral swaps bilaterally, and we do take part in occasional multilateral cycles of bilateral trades. But a more efficient route is to backload those trades to a clearing house and compress them in multilateral cleared cycle. Compression of non-clearable products mostly happens in multilateral cycles, as the efficiency is so much higher.
Risk: Is it because they’re not in the multilateral runs and therefore not compressing as much?
Edward Ground: If you have a portfolio that has an obvious solution, it’s easier to do it bilaterally.
Mike Sweeting: Capitalab does bilateral and trilateral cycles on an ad hoc basis. We use bilateral and trilateral compressions to onboard a customer and we can do bilateral compressions for particular opportunities, not least because from a regulatory perspective under the Markets in Financial Instruments Regulation (Mifir), we have all the post-compression reporting in place. In some instances, banks may find it easier for a provider to do a small run, especially if the provider can get it rolling with onboarding a new customer.
Gavin Jackson: The other benefit is from a valuation perspective. I imagine it can sometimes be difficult to reprice your portfolio bilaterally, whereas if you have a vendor as an independent, trusted third party, we can arbitrate.
Risk: How do banks allocate their teams for compression across multilateral and bilateral? Any views on cost versus benefits?
Duncan Lillie: We don’t operate in silos. We have a single front-office team, which works on optimisations across many measures, supported by a dedicated middle-office resource.
Magnus Lindahl: We’ve kept unilateral, bilateral, multilateral, initial margin optimisation and compression together. We’ve tried to sponsor as many IT tool builds as possible internally to enhance processes in order to handle 10 runs rather than five. That’s been done over the past couple of years.
Allan Guild: If you talk about prioritisation, I think there’s a more interesting discussion over how much time you invest in performing the runs today, and how much time and money you invest to try and make the process more efficient. That’s the challenge that we find ourselves weighing up on a day-by-day or week-by-week basis. Similarly, we are trying to find ways in which we can make it more efficient.
We keep coming back to benefit versus cost. I came here with the view that there are different benefits from different providers, but actually the challenge to the industry is reducing the cost element – including operational risk. I’d love someone to take the other side of that and say the interesting piece is increasing the benefit.
David Bachelier: Is this especially true in forex?
Allan Guild: Yes. Coming at it from a forex background, you have more trades, more participants. Having said that, when speaking to colleagues in fixed income many of the same issues arise.
David Bachelier: It is even more the case in forex when you try to do forex options and swaps. One of the difficulties we found in compressing forex is the fragmentation of the market in terms of systems and STP. In rates, there is almost a single market-wide practice, but in forex, everyone has their own solution and way of booking/processing the trades and unwinds. This is certainly an area of improvement for multilateral forex compression.
Philip Staddon: Is there a way to create some shared information?
Gavin Jackson: There’s shared language – FpML – which many banks have implemented internally for representing their trades and you can export in that format. The trouble is there are different versions of FpML, but that’s an option we’re looking at to try and make the process easier for you to take FpML off any version.
Mike Sweeting: A lot of the post-trade STP for new trades is written in that language anyway.
Duncan Lillie: Much of the time spent on compression cycles – which, as we’ve said, is the major cost of them – is in the submission: working out what the population is, what the notionals and rates are. If that information was stored centrally and updated in real time, or at least once a day, we wouldn’t have these lock-out periods for several days while the vendor works out who owns what. If could be an almost continuous process – when all the dealers give the green light it could just take a few minutes to compute and process.
Allan Guild: In 10 years’ time, perhaps the idea of a trade will be old-fashioned and instead we have ledgers of activity on a blockchain. Is there any crossover here with disruptive fintech such as blockchain?
Mike Sweeting: There’s no reason why affirmation couldn’t be done on a blockchain and there’s no reason why the whole trade lifecycle couldn’t be kept on a blockchain. It’s disruptive to existing STP providers – they’re quite large and it would be difficult to implement – but it is something that has been mooted a few times in the past year.
Duncan Lillie: If we ever get to a point where there’s a historical ledger of what trades we’ve all done with each other, then hopefully by that stage industry participants will no longer be using notional as a measure of how risky a business is.
Allan Guild: I’m not sure you would need these guys any more at this point because those things are also compressed at a point.
Mike Sweeting: Not without agreement and not without the algorithms.
Allan Guild: It depends. I think you can do riskless compression instantly. The risky compression is clearly where the algorithm comes in.
Philip Staddon: With market developments so far, the algorithm doesn’t seem to be the difficult part.
Risk: Moving on to initial margin optimisation, there are quite a few vendors in this space – is there much differentiation between the vendors in what they offer?
Magnus Lindahl: They all have tended to start in the obvious place – forex. It has the simplest products to deal with and was the one that was going to grow the quickest, almost by definition. What we have seen is that the other asset classes have slowly but surely been growing, and the services have been lagging but are beginning to come up to speed. They are at different stages but are generally approaching the problem in the same way.
Duncan Lillie: In some ways, initial margin optimisation is more open to new entrants than the compression business because there is an agreed framework of exchanging risk files in a specified manner. Anybody can write a few hundred lines of computer code that draws line between different parties. That’s why the competition is healthy – which is good. They all have different approaches and levels of efficiency but the real differentiator is the speed and ease of execution.
Philip Staddon: There is a multitude of factors you would look at to assess whether you want provider A, B or C. But what no one can do is an objective test across all providers to say: the data is there, go and do it and tell me what you can do.
Risk: How would you do it?
Philip Staddon: All the data is in AcadiaSoft or the end-to-end technology common risk interchange format files you are referring to that everyone submitted. You can run it on that. You can run it across all providers and get an industry-wide view of the efficiency of the initial margin optimisation, but we haven’t done that.
Risk: Do you think that would be useful?
Mike Sweeting: It could address some of the issues we are discussing here, but it’s not the only measure of a provider: you have to consider what platform they use for processing trades, and so on. Rates are one thing, but once you get into forex there are all sorts of different settlement systems, which makes it more complicated. So it’s not the only measure.
Allan Guild: What we like about initial margin optimisation is that there’s a certain simplicity. We can look at a certain portfolio and see how much initial margin we need to lodge pre-optimisation, and how much initial margin we need to lodge post-optimisation, we can get the cost of funding from the balance-sheet team and easily calculate the dollar number that we have just saved. Getting to that answer I think is easier in initial margin optimisation compared with other areas. But you need to set that against the cost of how many people I’ve had to take away, and the operational risk faced by the middle-office team because of the manual trade bookings, and so on.
And, certainly, when we look at the different pieces in foreign exchange initial margin optimisation, the dominant factor is the network effect. You can have a marginally better algorithm, but it’s about how many people are participating in the runs and the trade portfolios they’re putting into those runs. That determines the amount of initial margin that we save in a run.
Philip Staddon: That is one of the problems with optimisation. Your ideal solution is to have one vendor because then you get the maximum contribution from everyone and the maximum benefit for everybody involved. But that’s not ideal from a competitive perspective and it’s not ideal from an operational risk perspective either. You need more than one.
Duncan Lillie: What if on a particular day, you get all the vendors at the same time, send off the files and then whoever gets the most efficient result wins?
Mike Sweeting: We like to think we provide something different on the basis that we can do delta, vega and curvature on the rates side. As a result, we think that would be the differentiating factor towards our rates and future forex set-ups.
Risk: Some vendors like Capitalab can optimise both cleared and non-cleared in the same runs, which is a new part of the process. How useful is this?
Magnus Lindahl: It is useful; if you want an outlet for your delta in the rates space, it’s the obvious place to put it. So it makes perfect sense. It has probably been delayed by the CCPs rather than by the willingness of dealers to get it working but it’s certainly beneficial.
Allan Guild: We look at compression, optimisation and clearing as being part of the same solutions set. Optimisation is a bracket that covers all of them, and we don’t consider it to be an accident of regulation that the cost of bilateral trading portfolios is much larger than it was pre-crisis.
Many banks are looking at how to reduce bilateral exposures by the measures that regulators have set. Part of the challenge, from a technology and operational point of view, is how to look at clearing, compression and initial margin optimisation in a more joined-up way. The clearing houses at times represented themselves as a comprehensive solution, but I think they’ve stepped back a little. You’re never going to get to a point where everything clears and it’s all solo compression. In a sense that would be ideal, but it’s not a reality.
Edward Ground: Banks have different priorities – they are shepherding scarce resources, so coming out with a grand, unified optimisation plan is a challenge.
Allan Guild: It only works if everything is joined up and the whole market agrees on what the target is, and we’re a very long way away from that.
Edward Ground: The market might never agree. One bank might think a leveraged balance sheet is more important than funding, or vice versa.
Philip Staddon: But that’s what gives you an opportunity. One bank has cheap funding and another is trying to get its leverage down. You’ll take initial margin costs, or the other way around, for the cost of leverage and you can trade it off. You’re optimising across things other than just the one siloed element.
Risk: Is there anything more dealers can do to help the optimisation process?
Duncan Lillie: If dealers had visibility on the impact that their risk constraints had on the efficiency of the cycle, it might encourage them to be a little more flexible. The vendor could supply results for a hypothetical unconstrained run, for informational purposes.
Mike Sweeting: Operationally, that’s achievable. It could be a very interesting exercise for the dealers.
Edward Ground: There’s a mix of new in scope trades and legacy grandfathered trades. It’s a problem to optimise across the sum of the two in various ways. The old portfolios are still pretty big at the moment, but maybe in three or four years they won’t be.
Risk: Are they currently being optimised separately and not across both?
Edward Ground: You can adjust old portfolios and you can put trades between portfolios, but you can only take trades out – there’s a one-way ticket in the old portfolio. So it’s a scarce resource from that perspective.
Mike Sweeting: The Commodity Futures Trading Commission (CFTC) could also help here. If you mix up pre-uncleared margin rules (UMR) trades with post-UMR trades in one portfolio for a compression, when out of the back end a partial rip-up results, according to strict reading of the Dodd-Frank rules some compliance departments would say you need to ensure it’s now a post-UMR trade, so valuations have changed. As a result, we see people pulling trades out of a compression because of this, which can’t have been regulatory intention.
Risk: Recent initial margin optimisation runs had lower participation from the dealer community, which may be creating higher funding burdens, with older optimisation trades expiring and not being replaced. Is this an issue?
Mike Sweeting: We didn’t see any difference from the December and January runs, they were similarly impressive, but then again we would think that. I can understand that some global systemically important banks (G-Sibs) might not want to load up on new trades before the end of the year, but we didn’t see any sign of it in our results.
Magnus Lindahl: There are so many conflicting regulatory frameworks whether it is funding in standard initial margin model or notional in G-Sib balance sheets. One day it makes sense to optimise and others it doesn’t. It creates a headache for these types of services if you lose some of the funding benefit.
David Bachelier: Funding is something that costs you over a long time, and around year-end you get some cuts in certain metrics. So maybe there’s an incentive to worry less about funding and worry more about other metrics around year-end.
Philip Staddon: But if it’s to do with the roll-off of hedges, what’s stopping you from making the hedges another two weeks longer?
Magnus Lindahl: I think it’s more to do with the G-Sib regulation.
Mike Sweeting: It possibly helps from a G-Sib perspective that they’re not live trades at year-end.
Magnus Lindahl: If multiple banks are doing the same thing, that results in good participation rates within the network and therefore the optimisation is achieved.
Risk: Not having to worry about that helps non-G-Sibs too. The hedges roll off but the banks are left with the increasing initial margin. Is there any way to fix it?
David Bachelier: If it’s just two weeks of funding costs once a year, it’s probably relatively immaterial from that perspective. You can even make the optimisation trades expire on the 30th of December and the next initial margin optimisation on the 1st of January, for example.
Risk: Some banks use synthetic forwards or zero-wide collars in the optimisation process to avoid the clearing mandate. Have compliance teams grown more comfortable with this practice?
Allan Guild: I’ll talk about the synthetic forwards because that’s a forex piece and something on which we feel reasonably settled. There’s a continuous conversation about regulation, wanting to follow the spirit of regulation and needing to follow the letter of regulation. And regulators talk about not constructing processes designed to avoid regulation. It’s pretty clear from the initial margin regulation that the primary products for forex delta are out of scope. Iosco published a second paper saying that, for cross-currency swaps, forex delta is not meant to be in scope and so we’re taking it out.
I think one of the purposes of this regulatory framework is about better classifying and managing our risks. As synthetic forwards can be used as part of a strategy to move forex delta out of an initial margin portfolio into a separate portfolio that is being well managed, I struggle to see that as problematic. We understand that other banks are worried about it, we’ve heard the same from vendors and through the media. When we take it back to first principles, it feels like it’s completely in line with the spirit and letter of the regulation.
This is not because you’re moving risk from bilateral to central, you’re moving it from one bilateral pot to another. There are a number of ways of doing this: taking a synthetic forward against a physically settled forward – you can also do it with non-deliverable forwards (NDFs) and move it into the clearing house.
We talk about initial margin being driven by forex. You start breaking initial margins down for forex, and a large part of that is forex delta from instruments other than cross-currency swaps and forex forwards. So there’s the NDF piece, which you can pull out and clear, but then there’s the delta in Group of 10 products coming out of derivatives. That is a risk measure that Iosco is fairly clear it didn’t want to charge initial margin on, if you look at how it has classified products and what Iosco has done with cross-currency swaps. So for us, that’s part of optimising the portfolio; I’m less familiar with it from a rates product point of view.
Edward Ground: The optimising community has become comfortable with a similar approach in fixed income as well. The question lies in whether you’re in breach of mandatory clearing.
Risk: Are people comfortable doing it now?
Philip Staddon: You trade rate delta using put call parity and swaptions. You’ve constructed a product that looks like a swap, which wasn’t the case that you were going through.
Mike Sweeting: I don’t believe the regulations are particularly clear on that from the bank’s perspective, are they. The broker community and Capitalab would only look at it from a trading mandate perspective. There’s only one thing in Mifir that even relates towards initial margin optimisation – that is a post-trade risk reduction system. There isn’t anything that we can see with regard to clearing and initial margin optimisation, so I guess people have to be comfortable when their compliance departments then see those synthetic swaps on their books or those swaptions on their books via swaptions and say, hang on a second, what is this and where does it come from?
Risk: Would you see whether people are allowed, are able to indicate whether they want to or are able to use them or not in the process?
Mike Sweeting: Yes. Within the process, you can define whether or not you want to use them, but obviously you get better results if you do.
Risk: Are more people getting comfortable with them?
Mike Sweeting: Different people, different compliance departments, different comfort zones.
Gavin Jackson: The vast majority are comfortable, with a couple
Risk: Let’s discuss novations of some of the larger directional hedge fund trades from the prime brokers, where the executing dealer is now facing another dealer after the novation, which creates initial margin spikes between the optimisation runs. What is the best way of dealing with it? Would you take it straight to clearing or wait for the next run?
Allan Guild: Going back to answers that I’ve given to previous questions around improving technology, seeing it as a kind of optimisation, we certainly see those challenges. They occur fairly frequently in foreign exchange and prime brokerage. Hedge fund plays a big part and initial margin spikes on novations are real. To the point of how that is calculated – are you calculating that as a funding cost to the next time you expect to do an initial margin optimisation run? In which case it could be two days worth of funding, so it’s actually not that expensive. Do you know whether the counterparty you’ve effectively taken the trade from is also going to participate in that run?
There’s uncertainty around it and it feels like it’s a technology systems-led solution. That’s what we were talking about earlier. Which algorithm you are going to meet the counterparty with when you do the initial margin optimisation run is not part of the conversation. Nor, frankly, is which vendor. The question we ask ourselves is whether we are going to be part of a run that gives us an opportunity to reduce this.
Philip Staddon: What would provide more comfort is a more frequent optimisation cycle, which we don’t currently have for reasons discussed earlier. If it was completely STP, there’s no reason why you couldn’t do an end-of-day initial margin optimisation between four and five down the road. And then those kinds of issues go away, subject to the conditions of the compression or optimisation.
Magnus Lindahl: It comes back to spikes; if you reduce your initial margin but you have spikes, you lose all your benefits. Then, in theory, you end up having to plan for holding more buffers. So do you gain that much from participating in the first place? Just as important as reducing the amount of initial margin and risk, we ought to be conscious about reducing the spikes. The solution to run end-of-day initial margin optimisations, obviously that would be one solution. Moving straight into CCP is also a possibility.
Risk: Is going straight into CCP a viable and practical alternative?
Philip Staddon: It depends on the details and the products of that triangulation. You can always ask the client to hold off on the novation until just before a cycle, but it’s down to the specifics of the parties involved.
Risk: In 2020, as more entities come under the scope of the Basel Accord for initial margin, is it worth trying to get them into the multilateral runs?
Edward Ground: This is a big question. We need to understand the glide path of initial margin because we are uncertain how clients will react, and there are thresholds that might neutralise much of this for smaller clients. So the first issue is to work out how much initial margin the 2020 community is going to generate, and we’re working on that. There are large error bars around it but, if it’s a small number, then it’s probably not a significant problem.
Philip Staddon: Setting aside issues around documentation, when you say it’s not a problem, do you mean they won’t need to optimise because there won’t be a requirement?
Edward Ground: Yes. What is the cost benefit of trying to optimise a client with a small posting requirement – maybe it’s not that great?
Allan Guild: Looking at the alternatives: if we consider the buy side and separate it into investors and corporates, the corporates are out of scope and the investor community is going to have a lot of choice. The threshold topic is important. If an investor manages a selection of funds and one of the funds invests in emerging markets, the investor does their analysis and realises the fund has to post initial margin. How many investment managers will decide against having one emerging market fund, and instead have a Latin America fund, an Asia fund or a Europe emerging market fund? They could decide to separate them all out into different legal entities.
It’s hard to quantify on the sell side. What is the cost to an investment manager of splitting an existing fund into three funds or changing the dealer panel, and so on? How are buy-side firms going to look at the cost of that?
Philip Staddon: What are they saving by doing that, apart from a funding saving? Investors tend to be sitting on a lot of cash, especially if they’re doing derivatives. They’re taking long positions and sitting on the cash. They can post the initial margin, so I don’t think that’s a constraint. It’s more about going through the processing and doing the calculation.
Allan Guild: It’s the operational cost as well as the financial cost, definitely.
Edward Ground: It’s not clear whether dealers can do the calculations or would even want to do the calculations with their clients because there’s another regulatory risk around that. You’re effectively calculating both sides of the margin requirement and, if you get it wrong, that creates a regulatory problem.
Mike Sweeting: There’s also a balance for providers to try and find out how many you have to touch to make it work for the few.
Risk: We know that the methodology for euro swaptions pricing is changing in July – how will this affect the way you approach optimisation? Will you look to swap out your old swaptions for the new ones to aid optimisation?
Duncan Lillie: That could be challenging. Personally, I doubt that developing a dedicated platform for that purpose would provide much benefit for the market. Dealers should continue to compress down the legacy trades, of course, but the majority of the unwinds will be against trades with the same settlement method. Trying to compress legacy against new will be tough. It would be cleaner to trade out of any settlement basis in the open market, then rely on the compression services to reduce like-for-like trades. Ongoing compression, and of course the passage of time, should mean that within a few years the majority of trades will be settled with the new methodology.
Mike Sweeting: I wonder if how this is handled will be a preface to what the market does to prepare for the end of Libor. The brokers can help here because they can use their matching systems to process old trades for new. Plus, as previously mentioned, compression can help an enormous amount.
Thinking back to a recent industry questionnaire from EY, most of the questions focused on linear trades. It’s as if the industry had forgotten about swaptions or anything that might be difficult. We’d like to think compression has a large role to play in both the cash internal rate of return issue for euro swaptions, which happens at the start of the summer, and wherever the authorities go with regard to Libor, which is not clear to anybody at the moment.
Edward Ground: One question is how different an old swaption will be from a new one. If they’re similar then it will be a lot easier to deal with a big portfolio of old and new – maybe they’ll be very similar.
Gavin Jackson: I believe they will be very different, valuation-wise at least.
Philip Staddon: Doesn’t it make third-party vendors’ lives a little easier in the sense that, what you exercise into looks very much like a cleared product in terms of valuation? So from an optimisation perspective, things could be a lot easier; there’s less conflict and fewer technicalities in the model. And it removes any kind of arbitrage between the two sides that people may be playing, so it makes things clearer.
The panellists were speaking in a personal capacity. The views expressed by the panel do not necessarily reflect or represent the views of their respective institutions.