Matthew Scott is head of global rates, securitized assets and currency trading on the U.S. taxable bond trading desk at AllianceBernstein, an investment management firm with more than $500 billion in assets under management. He has been trading interest rate swaps for 13 years and has seen tremendous changes in his business, especially since the financial crisis. A large part of the changes came about as the result of the regulatory response to the crisis and the requirement that most swap trades be cleared.
But another strand of the transformation has been a technological revolution in the trading process. Scott describes how AB has developed proprietary software that applies data analytics to each trade before it happens, rather than only relying on post-trade analysis. "I have a pre-trade transaction cost analysis that I run on every single trade that we do regardless of where it is traded," he says. Scott also discusses the trend towards greater automation of trading, the pros and cons of clearing, and the transition away from Libor.
MarketVoice: At AllianceBernstein, what is the main purpose of swaps? Are they used to hedge portfolios against interest rate risks or mainly as a way to express a portfolio manager’s view on the direction of interest rates? Maybe a combination of both?
Matthew Scott: We use interest rate swaps to gain exposure to duration in many markets. We trade every single market for interest rate swaps, be it developed markets or emerging markets.
In developed markets such as the U.S., we trade and invest opportunistically in swaps, futures and cash bonds, based on our interest rate hedging needs, duration management and alpha source, and we toggle between the three products based on where we see relative value.
In a lot of the less liquid markets, getting exposure to duration via the cash bond markets is challenging from a liquidity perspective. In New Zealand, for example, the government bond market only has nine available bonds across the entire curve. If you want to get exposure to that market, the cash bond market is not as liquid and neither is the futures market, so we typically will express our duration view by using interest rate swaps.
We also use swaps to make alpha or beta calls on certain markets, or to hedge out interest rate risk on a bond portfolio. A corporate bond portfolio may own a few hundred different corporate names and we will use interest rate swaps to hedge out the interest rate risk in the portfolio.
MV: How much trading you do in interest rate swaps in terms of notional value?
Scott: I can’t give you a precise number but in the G10 [currencies] alone it’s over $100 billion annually notional in U.S. dollar equivalents.
MV: Can you give me an idea about the size of your staff, how many people are trading on the trading desk, and where they’re located?
Scott: We have a total of 21 people in both taxable and nontaxable trading. We have four traders in non-taxable all based in New York and 17 in global taxable trading. The taxable traders are based in New York, London and Hong Kong.
Most of the swaps that we trade are done out of New York. That’s a function of the fact that much of the interest rate trade idea generation happens in New York and the majority of assets that we manage in fixed income are managed in New York.
MV: Tradeweb recently announced that its volume of swap trades in December was twice as much as in the same month of the year before. Is that a sign that the market is becoming more electronic and moving more trades to electronic platforms such as Tradeweb, or is it more a function of a market-wide increase in trading?
Scott: The electronification of the swaps market has been going on for quite some time now. It is important to note, though, is that some of e-trading was legislated after the financial crisis. Post Dodd-Frank, if you are trading U.S. dollar-denominated swap, a euro-denominated swap, or a sterling-denominated swap, you have to centrally clear that and trade it over a swap execution facility. Dodd-Frank was signed into law in 2010 and fully implemented in 2012 and 2013. Since then you’ve really had a huge portion of the market that was legislated to trade electronically.
There’s also been a structural shift toward electronic trading. The electronification of the swaps market has helped compress bid-offer spreads, so that brings down the costs that our clients pay to transact. Second, from a trader's perspective, it is very convenient to trade swaps electronically because you get the benefits of straight-through processing where traders can send a trade from an order management system via FIX or API connection to an execution management system. When you trade on one of the electronic venues, the execution details of the price, competing quotes, and all the underlying economic details get fed back into your system automatically. Traders do not have to manually book trades into their internal systems any more.
MV: How much of your swaps trading is electronic?
Scott: We trade as much electronically as we possibly can. The super majors—U.S. dollar, British pound and euro-denominated swaps—are traded on SEFs such as Tradeweb and Bloomberg. On the other hand, Scandi [Scandinavian currencies], New Zealand dollar, Australian dollar and Canadian dollar trade mostly on institutional chatrooms like Bloomberg or Symphony in request-for-quote format. A lot of people trade over chat rooms. I don't know if that fits your definition of electronic trading, but they are an electronic form of communication. These chatrooms are persistent, the trading data is stored, and the history can be reviewed at a moment’s notice.
MV: One of the effects of electronic trading is that there is a lot more data available to buyside traders. Does your electronic trading technology include things like pre-trade analytics and transaction cost analysis?
What we’ve done here at AB is we’ve made pre-trade TCA part of our investment process. Before we implement one of our trade ideas, we will systematically analyze how much it is going to cost us to do a trade, normalized in fractions of basis points.
Scott: We’ve invested a substantial amount of resources putting pre-trade intelligence tools in place at AB. We have a pre-trade transaction cost analysis tool that is run on every single rate trade that we do regardless of where it is traded. This tool covers all global rate products including swaps, futures, and cash bonds. What we’ve done here at AB is we’ve made pre-trade TCA part of our investment process. Before we implement one of our trade ideas, we will systematically analyze how much it is going to cost us to do a trade, normalized in fractions of basis points.
In the past, transaction cost analysis was always done post trade. Post execution you would go in and you see how much you paid, either the execution price versus the cover price, or you look at the bid offer at the time you traded, strike a mid and perhaps estimate slippage.
Now, if I want to trade, say, $500 million five-year interest rate swaps, I know it’s going to cost us x basis points from mid to do the trade. If I want to do that in five-year futures instead of swaps, I know exactly how much it’ll cost us to do it. If I want to do it in five-year cash bonds instead of swaps or futures, I know how much that option will cost. The pre-trade intel helps us with instrument selection and implementation.
MV: Where do you get the information to make that kind of analysis, and how do you bring it all together?
Scott: Our own trade database [and] transaction history. Also, we’ve also invested in a proprietary system called ALFA, short for Automated Liquidity Filtering and Analytics. ALFA is where our trading desk aggregates millions of different data points from electronic trading venues, TRACE, dealer inventory, messages, our own transaction data, and trading axe indications received from our trading counterparties on the street. ALFA monitors all of these liquidity pools and data points simultaneously and lets us know if there is a counterparty looking to engage on a bond or swap in which we have a position or interest. This provides the trader with liquidity and depth available before we engage the market with a trade. Again, this puts the liquidity engine and pre-trade market intelligence at the front of the process.
MV: What about margin costs? Is that also part of the pre-trade analysis?
Scott: We focus on actionable trade intel, market liquidity, depth, bid-offer, aggregation of trading axes in the market, use of algos vs. block trades, and so on. The post-trade costs like clearing fees and margin on cleared swaps are mostly static for cleared derivatives and do not change from trade to trade. Margin on uncleared over-the-counter derivatives is governed by the individual account’s Credit Support Annex that is in place with the executing broker. This can vary a great deal from one account to the next, so it's hard to generalize. For example, some accounts can post only USD cash, while others can post multiple forms of collateral like GBP, EUR, corporates, munis, etc. That collateral is priced on a different discount curve which will alter the upfront price.
MV: Given how much trading has moved to electronic platforms or communications networks, it seems like the stage is set for greater automation of the trading process. How much of your swaps trading is automated?
Scott: Less than 10% of our global swap trades are automated. For comparison sake, in cash U.S. Treasuries, 44% to 47% of our trades are automated. These cash trades are low touch; we route the orders directly to an EMS [execution management system].
Automation in the interest rate swap market is well behind cash and futures at the moment. Why? Much of this is due to the market structure in swaps where new USD, EUR, GBP spot starting swaps must be traded over SEFs, where the predominant method of execution for the buy-side to dealers is still RFQ. The other currencies where interest rate swap markets exist are less liquid than those three majors, they're not as actively traded over electronic platforms, and therefore they're less conducive to automated work flows. Also, spot starting new swaps represent just a portion of the swap volume as there are forwards and cleared offsets, which are the terminations of existing swaps already at the clearinghouses that have rolled down the curve. Cleared offsets trade on an NPV [net present value] fee, not a rate, which must be then split across multiple line items in some cases.
MV: In light of that, what do you think is needed to achieve a greater degree of automation in swaps trading?
If swap volumes increase, you’re going to see the desire to handle that flow more efficiently by implementing smarter algorithmic pricing of swap offset packages and streaming liquidity of spot and forward swaps.
Scott: One of the key catalysts is the amount of volume going through. If swap volumes increase, you’re going to see the desire to handle that flow more efficiently by implementing smarter algorithmic pricing of swap offset packages and streaming liquidity of spot and forward swaps. This will get investors more comfortable with automated workflows and add leverage into the business by moving buyside traders up the value chain in the investment process.
MV: Are you seeing any trends in liquidity for interest rate swaps?
Scott: I’ve seen a pick-up in interest to quote less liquid currencies such as Norway, Sweden, Canada and Australia. Those four currencies in particular, there’s definitely been an uptick in inquiry electronically and the desire to trade these more electronically.
MV: You also mentioned the requirement for clearing. What proportion of your swap trades are cleared?
Scott: 99%.
MV: Is there a cost or other benefit of cleared versus uncleared swaps?
Scott: Yes, you no longer have to take credit counterparty risks. The financial system is a much safer place as a result. If you’re trading a centrally cleared instrument, you might be executing with XYZ bank, but you’re not taking on the credit risk of having to exchange coupon payments with them throughout the lifecycle of the swap. All of that is being done with the exchange and your FCM [futures commission merchant].
MV: You seem quite happy with how Dodd Frank changed the swaps trading business. Are there any downsides to what they did?
Scott: I think that the CFTC has been very intelligent about how they’ve gone about implementing the rules of Dodd Frank. There are some increased costs with respect to clearing. In the past you’d do a trade, the client would get executed at a price and then the cost of all of the post trade services, be it collateral movements or booking the trade out, would happen between the bank and the client. Now, you have other intermediaries that are involved in the trade, such as an FCM or a clearinghouse like the CME or LCH, and they extract their fee along the way to process the trades. The cost structure from a post-trade perspective is higher. That’s one downside, but you’re also getting something for that because our end users, the clients that we manage money on behalf of, are no longer on the hook for credit counterparty risk.
MV: Libor’s going away. Are you taking steps to use swaps-based alternatives to Libor like SOFR?
Scott: Yes. That has been a big focus for us for a few years now and remains a huge project for us internally in 2019. It’s not just for interest rate swaps, it’s for other products like loans and preferreds [preferred stock]. There’s a whole number of different asset classes that are going to be affected by this migration from Libor to SOFR [the secured overnight financing rate]. We are taking a wholistic approach to navigating the Libor transition.
MV: Do you think there’s sufficient liquidity in those SOFR swaps?
Scott: Not yet. The SOFR rate is trading obviously, you’ve probably seen there is volume going through on CME in the contracts, but there’s not yet a developed market for, say, a 10-year swap that references SOFR. Maybe one or two trades have been done total in the entire marketplace. There hasn’t been a very deep pool of liquidity yet for a fixed float swap that references SOFR on the floating leg.
Remember, you’re going from an unsecured Libor world to a secured world, so any model that references swap spreads is going to have an embedded credit component to it. That’s going to be going away. We’re focused on that as well. We have to make sure that our curves are completely updated so we can value these instruments appropriately.
There’s a lot of work that needs to be done ahead of the transition between now and 2021. I’m hoping the market will be mature enough by the second or third quarter of this year, or early next year, worst case scenario.
