Search

SEFs and clearing: managing the operational challenges

1 November 2014

By

It’s been more than a year since swap execution facilities went live, and the swap execution landscape continues to evolve. This is particularly true for the firms involved in clearing the trades. While swap clearing has been around for more than a decade, the introduction of SEFs has dramatically changed the workflow and introduced new complexities to the clearing process. Earlier this year, several experts in the operational aspects of clearing swaps came together as a group to address these issues collectively. The FIA Cleared Swaps Operations Committee began discussing how the industry could move towards a new model for clearing swaps in the new execution landscape. 

As five members of the committee explain in the following roundtable discussion, most trades are in fact cleared without a problem, but the complexity of the infrastructure and the lack of common practices causes delays and increases the risk of broken trades. It is a regulatory requirement that all trades must be checked against risk limits, but there is no standard process for achieving this. Some firms are linked directly to the SEFs while others are using credit hubs, centralized databases where credit limits are stored. Some are using a “ping” model for checking limits, others a “push” model. This is making it difficult to manage credit risk accurately in real-time, particularly when clients are trading across multiple SEFs and bilaterally as well. 

Equally important, these conditions are discouraging end-users from trading on SEFs and slowing down the transition to more efficient electronic trading. For some trades, the limit checking process can take a fraction of a second, for others it may take hours before a trade is cleared. 

The committee, which now consists of representatives from more than a dozen clearing firms, is in the process of discussing an initial set of operational recommendations with key stakeholders in the markets, and in particular with SEFs and clearinghouses. One of those recommendations, which is discussed in detail below, involves the use of “credit tokens” to confirm that trades have passed a clearing firm’s risk checks all the way through the trading and clearing process. According to the roundtable participants, a standardized approach to determining who sets these tokens and how they are attached to trades would go a long way to improving clearing certainty of trades, a key concern for the buy-side. 

This highlights one of the key lessons learned from SEF trading so far. As the roundtable participants explain, there are important differences in how swaps and futures are traded and cleared, and as a result all of the participants in the cleared swaps markets need new ways of processing trades and managing risks. 

Given the number of entities involved in the clearing process—SEFs, clearinghouses, credit hubs as well as the clearing firms themselves—the landscape is likely to continue to be extremely complex. But the committee is hopeful that an industry-wide dialogue on a new model would be a good place to start moving toward a more harmonized and efficient process for clearing trades. That would help reduce costs and risks for all market participants and provide customers with greater certainty of clearing.

FI: All of you are involved in the operational aspects of clearing swaps. What kinds of operational issues does the SEF landscape pose for futures commission merchants like yours and how does that contrast to the operational issues that you face in the futures markets? 

SHAPIRO: I think there will definitely be a harmonization over time between futures and OTC, but I do think that there will continue to be differences based on two primary facts. First, the products are different. In general, futures are more standardized, trade in smaller sizes and tend to trade more frequently. There are also differences in the regulatory landscape. For example, with OTC you have LSOC (legally separated but operationally commingled), with futures you don’t, and having a different segregation model means different funding requirements. 

Second, the market structure is different. In the futures model, when you execute a trade on behalf of clients, you have confidence that it is going to clear due to the vertical integration model. In the OTC space, there is no guarantee that the trade is going to clear. When I say, as an FCM, “yes” to a trade on a SEF platform, I am guaranteeing that I will clear that trade. However, the clearinghouse is still going to do a credit check on me, and if they reject that trade, it’s void ab initio per the CFTC guidance. Basically, it’s as if that trade never existed. That’s a primary difference between futures and OTC. 

Having said that, I think that we will see a convergence over time. As businesses tend to organize themselves around the similarities, we will see a push to have both products grow and evolve, and they will become more and more similar. But I don’t think that there will ever be an identical regime. 

TORRETTA: You touched on a really important point in terms of the differences in how trades are executed and screened across cleared derivatives. In the futures space trades are executed via an agency model and the executing FCM performs the screening of the orders, whereas in the OTC clearing space it’s the clearing FCM that’s responsible for performing the credit check against the counterparty. That’s a major distinction in terms of how the whole clearing process is then facilitated technologically. 

It also highlights the fact that the credit check being performed in OTC clearing is in line with the clearing limits that we’ve established with our clients and reflects our appetite to clear that activity. In contrast, for futures the limit check is an order-screening process. I think there’s an opportunity to look at how we are screening cleared trades across OTC and futures and determine the best practices for managing that risk. 

DENNE: I would add that globally this is still a work in progress. Given the size of the swap market outside the United States, I don’t think we’re going to see the full impact of these changes until we see similar sorts of legislation and regulations put in place in Europe and Asia. 

The other part of this evolution is the shifting economics between the bilateral market and the cleared market. There’s a discussion now on the requirements for initial variation margin on bilateral swaps, and that will start to normalize the economics of when you clear a trade or don’t clear a trade. Clearing probably will look relatively more attractive than it’s been in the past, and we might start to see a voluntary shift toward swap clearing, even from those institutions that are not impacted by the U.S. regulations. 

FI: In the SEF landscape, trades can take various routes to clearing. How does that affect your risk management process? 

WILLKEHR: It goes back to the question of who is responsible for applying the risk check. For voice trades we do internally, we have the ability to control the amount of credit that we extend. For SEF trades, on the other hand, the SEF applies the limits that we establish and there are a lot of variations in how these are applied. Some have the ability to do this internally. Others may outsource that to a third party or a credit hub. Even with the SEFs that have the ability to calculate it internally, the methodology may differ. So when you’re establishing a limit for a client and then split it out among all the different SEFs, the shape of that limit may differ at the different SEFs. That causes problems when you try to aggregate the information and determine how much of that credit has been used at the different SEFs or off-SEF. It’s kind of like combining apples and oranges. 

SHAPIRO: I think the different models are based upon what clients ultimately want. If a client wants to trade quite frequently and is concerned about latency, they may want us to push a limit to a SEF. That’s especially important if we move to order book trading, because that gives them the quick dynamic of getting the orders approved in a timely fashion. Some clients care a little bit less about time and more about full utilization of their credit line. In that model, maybe the SEFs are pinging a hub where the FCMs house their limits, or maybe they’re pinging the FCM directly, or maybe they’re pinging a hub that’s pinging the FCM. I think every FCM in the market has a little bit of a different approach, but regardless of the differences, we are all working with our clients to understand which model, push or ping, works best for them and then supporting them accordingly.

FI: If the market continues to have these different client segments, does that mean that there will continue to be multiple models for the clearing workflow? 

SHAPIRO: I think that’s correct. There are multiple models today, and I don’t think the multiple models are going away. We want a highly automated and flexible infrastructure that accommodates all shapes and sizes of clients. 
To add to that, I think we want competition. You don’t get the best product unless you have competition and people coming up with new thoughts and ideas. But at the same time, as these new thoughts and ideas are introduced to the market, we don’t want to have broken process flows, manual workarounds, and people looking at spreadsheets. With the time parameters that the CFTC has put on us, and the parameters and the standards our clients hold us to, we need to make sure that good operational workflows are in place. 

FI: How long does it take to run SEF trades through the credit checks and to get a result? 

TORRETTA: It depends on the model that the FCM adopts. It will differ depending on if they’re internalizing that credit check versus allowing an electronic trading venue or a hub to perform the credit calculation. At a minimum, the credit check is being performed sub-second. This means that a client goes onto a venue, requests to introduce an order, and within less than a second they know that they have credit approval from their FCM to clear the trade. That is the process referred to as the ping model, where the order actually gets routed to the FCM, or to a hub that’s facilitating the screening of the order for them. That’s an order-by-order type of screening that’s happening outside the SEF, and that’s happening sub-second. 

WILLKEHR: We can’t forget that we still have a number of participants on the other end of the spectrum that support request for- quote trading via manually calling someone up on the phone. We’re talking milliseconds on one side and we’re talking half an hour on the other side, depending on when emails get sent around. 

DENNE: People assume that SEF trading is shorthand for electronic trading, but it’s not. There are still trades that are going to be voice brokered with intermediaries on both sides. Those are the areas where it may take minutes or longer. 

SHAPIRO: I think the speed requirements are, again, an evolution. When we initially started clearing trades, it could take hours for a trade to reach the clearinghouse and ultimately to the FCM to get cleared. As the infrastructure for clearing was standardized, clearing timelines improved and now regulation is in place that requires FCMs to approve or reject off-SEF trades within 60 seconds and perform pre-execution credit checks on SEF trades. 
The question always becomes, well, when does that time clock start? For off-SEF trades, the time clock starts at the point that I, as an FCM, receive the trade. So a bilateral trade could happen between a dealer and a client four hours ago. It still has to get into the middleware platform, it has to be matched, it has to go to the clearinghouse, and then the 60 seconds starts when it comes to me. Yes, things are clearing a lot faster, most FCMs are well within the 60 second timeframes, but there are still some processes higher up in the work flow that may take a while. 

The other critical piece is the clearinghouse credit check in the SEF workflow. The regulation says that a clearinghouse has to approve or reject a trade within 10 seconds. But just like the off-SEF flow, that 10 seconds does not start until the clearinghouse receives a trade. When a trade is done electronically on a SEF, there are multiple paths that it can take to the clearinghouse. The SEF could submit that trade directly to the clearinghouse or it could route it through a middleware platform, which is still fast but adds a hop. Some of the SEFs also support a manual affirmation process, where the client will go into the middleware platform and manually affirm the trade, and then it will be submitted to a clearinghouse. Currently, there’s no specific timeframes enforced for the manual affirmation. For example, a transaction may sit in a middleware platform for four hours, which means the clearinghouse is not clearing the SEF trade within 10 seconds, they are processing the trade for clearing within four hours and 10 seconds. 

So when we look at the regulatory timeframes, it’s important to remember that there are different workflows in the industry. As we’ve moved to the SEF landscape, they’ve improved, but we’re still learning along the way. 

FI: Let’s talk about a specific example of the operational challenges that you face. FCMs have started using something called a credit token to indicate which trades have passed the credit check. How does this work and what are the challenges to making this work well? 

WILLKEHR: Essentially, a credit token is supposed to be a mechanism that allows the FCM to monitor and ensure that the credit limit utilized on a SEF platform actually equals that of the cleared trade. It’s really a control to ensure that our credit is being used properly. 

SMITH: I see credit limits as they exist today as really nothing more than a yes/no parameter. Credit tokens will go further in that they work like a digital signature that passes all the way through all of the parties. It shows that you’ve approved credit for that specific FCM to stand by that transaction. 

A good analogy is the way banks monitor how people use their credit cards. When you swipe your card, a message is sent to your bank with information about how much credit has just been used. And if you’re over the limit, a message comes back pretty fast saying no, the transaction is not approved. 

So that’s the idea behind the credit token. But we won’t have that type of function until we standardize how credit tokens are generated, and do that in a way that persists all the way through the workflow. 

DENNE:  We don’t want to get to a world where it is all entirely vertically integrated with absolute standardization and no possibility of innovation in the process. That is not what we would want the market to evolve into. What we want is a solution where it doesn’t matter who issued the credit for that transaction, or where that transaction has been executed, or which firm is clearing it, or whatever path it takes to get to the clearinghouse. All we need to know is that the transaction has been approved by someone via an unambiguous message unique to that transaction that is traceable throughout that whole process. 

FI:  Are all the trades that you are now clearing getting these tokens? 

SHAPIRO:  No, they are not. It goes to the complexity of the workflow. As we said before, there are multiple SEFs, multiple clearinghouses, multiple FCMs, and multiple clients. So one of the first principles that needs to be addressed is who should set the token. What we see is that sometimes the FCM is setting it, sometimes it’s the hubs, and sometimes it’s the SEFs. 

For example, we might set a token on a trade. Then, when it gets to a SEF that has a token functionality, they overwrite our token with their unique token. Then that goes to the clearinghouse, and then the clearinghouse sends it to me. I set my token as 123 and it comes back 456. There’s no way for me to confirm the credit check I did on this transaction and tie it to the trade that actually got executed. 

DENNE:  One final twist in this is that in the pre-SEF world, all the credit checking was being done after the trade was executed and before it cleared. When the SEF framework came along, the FCM community didn’t get a second look at that trade. It’s assumed that the trade has been screened by us before it’s executed. So we need to have that token that says, yes, we really did issue credit for that transaction. It’s not just a breakdown of the workflow between the SEF and the clearinghouse. 

FI:  Why do you think this is so hard to accomplish? 

SMITH:  I think it’s the timeframe. With mandatory clearing, we all had a long lead-time to get comfortable with the process and onboard clients in a semi-orderly fashion. With the implementation of SEFs, I just don’t think there was enough lead-time to make sure that we had this right and put a robust architecture in place. 

SHAPIRO:  I would add that there are two other factors. One is the number of parties that are involved. When you look across the SEF landscape, you have probably 40 or more different parties that need to implement the token. Remember it’s not just the FCMs, it’s also the SEFs, the CCPs and the middleware platforms. Second is that while everyone may agree in concept that a token is needed, it needs to be standardized across all these platforms. That doesn’t mean we should enforce these requirements or try to make everyone alike. It means that we need standardization for basic requirements so that this is beneficial for operational ease and regulatory compliance across the FCM community. As a first step, let’s set out some principles on how tokens should be used, and then let’s all work together to get a process flow that’s best for the industry. 

WILLKEHR:  Going back to your question, the reason why it’s so hard is because people have built their systems already, and they don’t want to go back and rebuild those systems because there are so many other areas that are demanding attention. But it’s important to keep talking about standardization. That’s all this is. Going back to Tony’s analogy, that’s why the credit card reader works. There are different readers and different service providers, but it all works because it’s a standardized methodology. 

DENNE:  The fact of the matter is it’s a highly competitive landscape now among the CCPs and the SEFs, and indeed among the FCM community. So everybody has a strong vested economic interest in expanding their product sets. If they want to solve next month’s problem or next month’s opportunity, rather than patch the process from last month’s error, then you know where the priority is going to be. 

SMITH:  I’d echo that sentiment entirely. But we’re all realists around the table as well. We’re not expecting all of this to be done by the end of the year or some other date. I think all we want to do is have an open dialogue with all of the participants in the market about a more robust market infrastructure. We think this doesn’t just benefit FCMs. It benefits clients and CCPs and SEFs. We all have a vested interest in making sure this infrastructure is as robust as possible. 

FI:  What are the benefits for the buy-side from these improvements in the clearing workflow? 

DENNE:  Execution certainty. That has to be the key concern of the buy-side. Having the certainty that when they’re done, they’re really done, and they’re not going to receive a phone call from an FCM or a dealer saying, oh, sorry, that trade a half an hour ago or four hours ago in fact is not done. And, by the way, the market’s moved five basis points since then. I think that’s probably the number one concern for the buy-side. 

TORRETTA:  I’d agree. And that’s even more of a concern as this evolves. Right now, we’re in a world where you know who you’re trading with. In the event that something goes wrong, you can have a gentleman’s agreement and work it out. But as we move to a central limit order type of world with anonymous execution, nobody is going to stand behind those trades. If something doesn’t get accepted for clearing, there’s really nothing to be done at that point in time. This highlights why the best market structure is one that provides executing participants with pre-trade certainty. This requires both the FCM vs. client check and the CCP vs. FCM check to be performed at the point of execution. 

FI:  How does the industry get closer to that certainty? 

DENNE:  Through a variety of practical or technical ways. The token is definitely one of them. We’ve also talked about some of the controls around making sure the transaction that is actually submitted for clearing can in no way possible shape or form be anything other than the trade that was approved for execution. Again, we’ve seen various deficiencies in the infrastructure that haven’t been able to eliminate that possibility. 

SHAPIRO:  I think the other point is to have accuracy of limits and messaging across platforms. Suppose I set a limit on a client of a hundred dollars. A trade comes in with a value of 20 dollars. Now I’m saying that the client can only trade eighty dollars more. Well, if I’m not told that they canceled that 20 dollar order and I should release that limit, it falsely impacts the client’s ability to trade. So these operational conundrums aren’t only about getting trades that have already been executed through the pipes. They’re also about allowing the future trades to go through because all the systems are linking together and giving an accurate picture of that client’s standing. 

I also think that the FIA Cleared Swaps Operating Committee helps to foster regulatory compliance. I don’t necessarily think that all the parties in the industry recognize the regulatory obligations of the other parties. I don’t think every SEF can say they know what the FCMs’ obligations are, and vice versa. What’s really important is to bring people together to find a common ground, to facilitate execution certainty for clients, support regulatory compliance, and ensure accuracy of data flows. 

DENNE:  To be fair, I think we all experience very high straight through processing rates and I don’t think any of these problems manifest themselves on an hourly basis. But the number of transactions in this market is very different from the futures market or the cash equities market. It doesn’t take very many transactions to go wrong for those to be relationship-ending and even business-ending type mistakes. I don’t think there is a tolerance for any significant rate of error. 

WILLKEHR:  I think some of these deficiencies in the downstream clearing infrastructure are definitely impacting the rates of execution on some platforms and the ability to move to where I think regulators want us to go in terms of increased electronic execution. It’s a mix of things. FCMs might feel somewhat reluctant to push out larger limits on so many different types of platforms because those controls aren’t robust enough. Some clients might be frustrated with figuring out where they have limits and not being able to consolidate them to know where they have head room.

FI: Looking forward in time, what changes do you expect in the work flow? 

SMITH: If we look at where we are today with regards to rates and credit, the regulations for mandatory clearing and SEF execution are largely in place or in motion. The next step in the U.S. is going to be new products like foreign exchange NDFs (non-deliverable forwards) and potentially some of the push controls that we’re going to see with central limit order books on swap execution facilities. But I think we’re now in a position where we can take a little bit of stock of what’s happened. The last three and a half years have been a bit of a maelstrom, really, when you look at regulation in the U.S. I expect that European and Asian regulation will throw up its own challenges. But we would all prefer consistency in workflow across regions and hopefully, we can take some of these lessons learned out to the European and the Asian side. 

SHAPIRO: When we look at the SEF landscape, the thought originally was that client activity was going to be spread across 16 different venues. I think what we’ve seen in actuality to date is that activity has centralized on one or two dealer-to-client platforms and clients have mostly become direct members of those platforms. By comparison, in the futures space, almost everyone is using an agency model. It will be interesting to see how the models change as the market evolves and as clients change their trading strategies and how they access the market. 

DENNE: I think we touched on something in the conversation around the different execution models and whether we’re going to see this transform into an agency or remain a largely principal-based market. The next six to nine months is when we’ll start to see whether the industry moves to that model or whether it stays as it is today. I also think that clearing has created a demand for solutions to problems that we didn’t have before. For example, compression trades. How do clients deal with swaps that can’t be netted? That’s created a problem for the buy-side and we’re starting to see SEFs play a part in solving that problem, bringing up packages of compression trades for dealers to bid on. For example, CCPs have started to come up with more imaginative ways of netting down transactions that are economically similar but not precisely the same. I think we’re going to see more of those types of developments as people become more experienced with client clearing. 

WILLKEHR: I think we all need to remember that it’s taken quite some time to get where we are today with clearing. I mean, the dealers were clearing swaps for 12 years ahead of the mandate. As Tony said earlier, we had the time to develop the infrastructure to extend those protections to clients, and we’re still working on perfecting that. Contrast that to how rapidly we pushed execution onto the SEFs. They haven’t really had the time to properly build out their infrastructure and evolve their platforms. 

To be fair, some of the infrastructure problems we’re seeing have been addressed by some of the participants, but they need to be addressed by all of them, front to back. For example, the clearinghouses have rolled out controls around bunched orders and the allocation process. Clearinghouses have specified the message structure needed to ensure that a bunched order trade is allocated properly and not over-allocated. But unless the SEFs send that information down, the process doesn’t work. That has an impact on some clients because they feel there’s too much risk around that process and as a result they may not be executing block trades. They continue to execute in end accounts. Those are smaller trades and that will impact the price. 

Going back to your question, looking ahead to where the market is evolving, it’s key to iron out some of these clearing infrastructure issues now in order to offer some of those services that Marcus just talked about and bring new products onto the market. For example, the MAC swaps. MAC swaps are a standard product that is well suited for order book trading. Right now they’re largely traded on RFQ (request-for-quote) but one or two SEFs are offering them or planning to offer them in the next month on their CLOB (central limit order book). 

So we really need to address these infrastructure issues now. That’s why it’s so critical to agree on certain basic principles like the credit token we spoke about before. I think there’s good buy-in for these principles. There’s a lot of work still to do in terms of implementation but those should be the next steps.

  • FIA