Hello, and welcome back to the second section of module six Stanford GSB Gen 544 how software eight finance and today we are talking about clearing settlement and then at the end regulation. Right now, I'm going to take you through the clearing and settlement for two kinds of securities that's US treasuries and US stocks. Obviously there's hundreds of different kinds of securities in different jurisdictions. So these are just two example. It's not so important that you follow every step. What I want to leave you with is a sense of the labyrinth theme complexity and the risks. The amount of time and energy it took from a large number of people at many firms working across the front office and the back office to understand these processes in these risks. And then for me to distill them and present them as is considerable and really just want you to come away with the notion that it's just as complicated, just as interesting, and just as risky. Maybe not, maybe even more so than the traditional front office execution process. The part of the firm where I grew up. So let's share the screen. All right, here we go. I'm going to start with US Treasuries. And again, this is just clearing and settlement, not execution. So, settlement processes are benign. It's like plumbing, we don't think about it on most days, but when there's a problem with the plumbing, it's the only thing we can think about. The settlement processes are benign until they're not. Given the global systemic importance of Ttreasury's settlement has to be seamless, especially as we go through crises and we saw this in the COVID-19 crisis. Because treasuries are in so many respects the bedrock of the global financial system. The settlement process these just have to work if they don't, there are immense consequences for financial stability and systemic safety and soundness and capitalism. That's really the the religion of confidence and so it's important that the settlement processes work and provide a foundation for that confidence. We've gone through this but to put the market into perspective, there's $14 trillion in outstanding US Treasuries that amount, obviously going up. That was the figure from February 2020. And about half a trillion dollars worth of treasuries are traded every day. On the right we have an old Treasury. Certificate. That one's from the 1980s. I haven't seen one of these in a long time. And the number of treasuries that exist in this paper certificate form is Vanishingly small, it's effectively zero. They all exist in electronic book entry form. They've been immobilized and then later dematerialized. I'm going to consider two use cases for treasury clearing and settlement. One is dealer to customer and then the other is dealer to dealer into dealer. So dealer to customer is 0.29 trillion let's just call it 0.3 trillion per day and almost all of that is traded on two platforms Bloomberg and Tradeweb. Believe it or not a considerable part of the work flow is still voice on the telephone. Also electronic requests for quotes we've touched on in previous previous lectures is an important channel as well. More and more of this is getting electrified, in spite of the fact that more and more of the execution is electronic, clearing and settlement processes really haven't changed all that much over the years. Most trades are clearing bilaterally over on the right, we have an example of a Bloomberg customer Balder for trading US Treasuries. Today I'm going to talk in some detail about the interdealer, dealer to dealer part of the market and the clearing and settlement inter dealer for US Treasury trades. Dealers trade a little under a quarter trillion dollars per day. Not actually a big surprise, that's almost the same amount as as what's traded dealer to customer because then of course dealers are laying off the risk with one another. And that trading, the inter dealer trading happens in a highly automated fashion through inter dealer brokers. Why? Because the dealers want to trade anonymously and the interviewer brokers provide that anonymization suite. Roughly three fourths of the interviewer broker trades clear bilaterally. Over on the right, we talked about some of the platforms where this Into dealer trader trading happens. We have any x, which owns broker tech, that's one of the major venues. And then we have Nasdaq fixed income, which bought a speed. There's a whole history there. It was owned by BGC. Which itself spun out of cancer in 2004. And then e speed was sold by BGP to Nasdaq in 2013. And then the last interview or venue over there on the right is Dealerweb, which is a part of Tradeweb. Now going back prior to 2000, all of the the platform users on these interdealer platforms were members of a Central Counter Party, a clearing house. And that particular Central Counter Party, CCP, with the government securities division of the DTCC Fixed Income clearing Corporation so they keep coming back to BTCC. IDB platforms themselves and a number of their participants clear and federal to this day through the CCP. However, there are many other platform users including the Principal Trading Firms, we talked about PTF. In session two, that account for majority of the interview or broker trading volume and those entities the PTS clear and federal bilaterally with the IDB. And you may remember that historically, many prefer to call these firms, high frequency traders. But to my mind, it's not really the frequency of latency that is the salient characteristic this area salient characteristic is that they do a lot of trading intraday. And generally go home with their positions flat, though some do carry risk across the trading day for multiple days. We've talked about the benefits of central clearing novation multilateral netting of exposures and margining So there are hidden risks all over the place the clearing and settlement of US Treasury trades. When these trades are bilaterally cleared and many of them are the intraday and the overnight credit risk remains with the original trading counterparties from trade execution all the way up until settlement. Indeed, the risk mitigation and management practices are much less standardized in this bilateral clearing and settlement process. Then they are with the CCP. By construction the CCP will have a standard process and bilateral pairs have invented their own bespoke processes over the years, and they're all different, like snowflakes. And that difference actually creates risk. And it's hard to see how it actually generates any value is the legacy. See, also there is a challenge here that the underlying asset is the definition of the risk free asset. And so the asset can be risk free, but the clearing and settlement could be full of risks and one hypothesis that because the underlying asset is risk free. Market participants might not be applying the same level of rigor to the clearing and settlement of their treasury activities as they do on for instance, their derivatives businesses. And we certainly saw an example of that becoming a huge problem for the financial system in 1998 when it came time for LTCM to unwind its leveraged treasury position. So there are many different patterns for treasury market clearing and settlement. And I'm not going to go through all of them but I wanted to give you a flavor of the main cases. So we have the order customer with bilateral clearing. We have interdealer where both sides are members of the CCP or clearing house and therefore of course there's central clearing and that's an interdealer trade. And then we also have the case where two dealers are trading through an inter dealer broker to preserver their anonymity. But both of the dealers CCP numbers, and then we have the case where there are two dealers. And they're not members of the CCP, and they are trading through an inter dealer broker. And then the last case we have it's one of each. So the buyers, the CCP member in this case and the seller is a non member of the CCP. And there again is an inter dealer broker. So there's a lot of complexity just in coming up with this taxonomy in the first place. So now, again, without going through it in immense detail, I want to give you a sense of all the risks and the vast complex complexity here. And so, the gray inner circle. And the diagrams I'm about to reveal the note stages in the transaction life cycle. And the stages are starting of course with execution and ending with final settlement and in between we have firming the trades matching the trades and clearing and settlement. Some would call affirmation matching part of clearing and others might call it part of execution. I've called them out here. And then we have many different market participants and they're all color coded. So we have the buyers and sellers clearing banks, the CCP or central Counterparty Clearing House, the inter dealer broker. Sometimes there are multiple disclosed participants in the transaction in the custody bit bank we have prime brokers of agent clears. And it goes on. And so I'm going to enumerate all of the cases here. So we have a CCP member trades with the CCP non member and that diagram is relatively simple. And you can see who's got the wrist on T plus zero. And who's got the wrist On t plus one as we go through the four stages execution, first and second affirmation matching, third clearing and settlement or final settlement. And then next we have a different case which is a CCP member trades with the CCP non member on behalf of multiple disclose participants, and you can see that diagram looks yet a little bit different. And then down in the lower left we have two members of the CCP trade and it's pretty clear where the CCP steps in. You can see that in the yellow, handling the affirmation and matching and that happens between T plus 0 and T plus 1. And then we have two CCP members trade by an inter dealer broker and you can see where the inter dealer broker has risk in the trade through execution. And then as soon as it goes on to affirmation and matching, that risk transfers over to the CCP, and that again happens in between t + 0 and the start of t + 1. We have the CCP non members trade by an interviewer broker. We have a CC member and a CCP non member trade by an inter dealer broker. In the last one, one of the more complex ones is to CCP non-members trade through an inter dealer broker. And there's also a prime broker in the mix. And you can see that you start getting lots of different circles and lots of overlap. And it's really important to lay out who owns what risk rigorously at every point in time. So I'm going to just go through two of these examples. This is two CCP members trade. And you can see execution trade booking happens on people 0 and then the affirmation submitting the trade details matching and novation and netting are all happening late in t plus 0. And then some of that is going into the overnight process and then going into t plus one. Now affirmation and confirmation is over and we're into clearing and settlement, and we're getting into fedwire processing and delivery versus payment. So someone's got to get cash and someone's gotta get the securities. We got final settlement. And here we can see who owns which kinds of risk at different times in the process. Now, let's go through a more complicated use case. And this one is a principal trading firm. It's buying from a dealer who is a member of a CCP. And so we're going to start with execution on and again that's our trade booking and that's happening on 50 zero. And then we go into affirmation and confirmation. And we can see that the buyer is holding the overnight risk but also there's some inter dealer broker overnight risk, and some CCP, overnight risk and the seller too. So it's super complicated. Then as we go into t plus one into clearing and settlement, we can see the roles of the CCP and CCP, the IDB and the agent Claire's in the Actual processing of the cache and it gets more and more intricate. And then we go into the final settlement which looks like final settlement and all the other processes so, not so important that you understand in gory detail what's going on here, but that you carry away the sense that it is complex. It's labyrinthine, many of these properties are bilateral and therefore they're bespoke or customized. They're all different. And it's not clear that everybody completely understands how they're different. And it's also not clear that the software equals the business in other words are all of these risks and processes precisely modeled in software, or they do the details exist in the form of some legal agreements signed today. Decades ago. It's in the Manila folder in some storage cabinets that nobody has seen for a long time. And sadly, that is literally the case for some of these agreements, not so much with the PDFs because they're relatively new with people have been trading for a while. So now we can step back and see the complex clearing and settlement process for this particular segment and see it in all of its glory. So now let's talk about a very different kind of asset, which is the US stock trade. And how clearing and settlement happens for US stock trades. You'll remember that not that very long ago, stock trades settled on T+ 3 as compared to T+1 settlements for treasury trades. And it was a major industry effort to and took many, many years to take that T +3 down to T+2 in as many discussions about taking it further down to T+1 and then the aspiration is T+0. It's going to be a long road and I don't think anybody knows how long it's going to take to travel down that road. So, pre trade to trade capture we have the broker sending the trade to the market. And then we have the NSCC stepping in. And there's a number of components and the NSCC, which is the clearinghouse for us equity trades. We have the UTC universal trade capture, which validates all the trades that are submitted by the exchange and does so during the course of the trading day. And then that's going to feed trades into the CTS, the Consolidated Trade Summary. And then the Consolidated Trade Summary is going to decide whether the trade goes into one of two places. First, the CNS the Core Netting System, or alternately into the obligation warehouse. And you can see there's limits monitoring and many other pieces of software and business processes that also interact with the prime broker. And then after the core netting system or the obligation warehouse trades then go into a cast which is the automated customer account transfer service. And this is a service that moves securities again they have been immobilized and de materialized but moves them virtually, from one broker to another. In the general case when the customer on both sides of the trade were trading in street names through different brokers. And then ultimately, it goes to DTC. And there are a few other settlement locations for the atomic delivery versus payment. So again, super complicated, a lot of moving parts, and something that nobody thinks about. But I can tell you that for instance, when a trade gets hung. Some part of this process or maybe a whole bunch of trades get hung. Everybody wants to understand exactly what's happening at every step. And who owns the risk at every step. And it can change minute by minute. And you want all those risks, operational risks of various kinds. But also clearing credit in settlement risk, you want them all precisely modeled in your software. So, we went through the various parts of the NSCC, and again, is immensely complicated in detail here. And once you get a sense of the overall picture and how they all fit together. And so, as I mentioned, the US equity market moved from T+3 to T+2 not very long ago in 2017. And so you can see all of the pieces of the pie that allowed that to happen. And exactly who is owning which risk. And what's happening on these various days. So T+1 has matching affirmation, which is, you see from the treasury example, all got done on day zero. That's happening on day one for stock processing, and you can see that the trades get to the CNS really between T+1 and T+2. And then ultimately settlement happening on T+2. So another view of what's happening between T+0 and T+2. So now let's talk about an epic event in the markets, one that I will never forget. I was in London at the time I remember the 2012 Olympics were happening. And as you'll hear from this description, you can get a vivid sense of what I was doing on that day. And the answers, I was not at the Olympics for$ that day or the days, the days immediately following it. So let's hear about the Knight Capital Group Trading here. Let's talk about the $440 million software error at Knight Capital Group on August 1 2012. A day I will never forget. I was co head of the Goldman Sachs equities trading business in that year. In 2012, Knight was the largest trader of US equities with a market share of 17% or $21 billion per day. A combination of poor software design, reusing a boolean variable for an altogether different purpose. And poor software deployment practices. Manually deploying code to redundant servers, leaving out one server by accident. Triggered a piece of code that was meant to be used only in assimilated test environment. Remember that 80s movie, WarGames? The broken software caused a major disruption in the prices of 148 NYSE-Listed Stocks. For 212 incoming parent orders processed by the defective code. Knight Capitol sent millions of child orders, resulting in night paying $7 billion across 4 million executions on 379 million shares. Of a 154 different companies in approximately 45 minutes. All of those trades were unintended. Knight would have had to pay $7 billion of US dollar cash on T+3 to settle all those trades. Knight didn't have $7 billion. According to SEC rules, trades on securities, whose prices moved more than 30% can be canceled under the clearly erroneous policy. Only six stocks had moves and more than 30% though. Knight's trades on those stocks were canceled. All the other trades stood. Therefore, Knight entered the day as a clearing member in good standing. By the end of the day, having eroded so much of its capital. It was no longer a clearing member in good standing. And so those trades were going to clear Goldman Sachs stepped into Knight's role on all of the error trades. Taking on the entire block of trades. At a 440 million discount to the closing valuations on the day. Goldman Sachs worked out of the risk over months. It took 17 years to build Knight. Knight destroyed itself in 45 minutes losing $10 million per minute. So an example of what happened when. Something that you expect. Let's talk about the 440 million. Alright, and to occur in the standard way doesn't occur. Namely, someone comes into the day as a clearing member in good standing. And then because of some execution error is no longer a clearing member in good standing. And I remember that's when everybody went to the rule books and everybody wanted to know exactly what was going to happen at what moments in time in the entire market. Hanging because of course, people who've done these trades needed to know for purposes of their own P&l and their own risk management, whether the trades done or were going to be cancelled. So when you look at the clearing and settlement process industry wide. From back office of one firm to back office of another firm, through the clearing out of duty exchanges, through the interdealer brokers, the prime brokers, and all of the software infrastructure providers. Here is what you see, tremendous complexity, very few people understand at all. I don't think anybody understands all of them, no one person. And beyond that, there's pointless extensive replication. Actually, everyone's back office does pretty much the same things. And so the question is, why does every Wall Street firm have its own back office? Well, you could ask though the automobile industry a long time ago, standardized around tire gauges. And so they're few manufacturers of tires, and everybody goes to those manufacturers and tires in the old days. Automobile manufacturers were vertically integrated, even down below the tires into owning their own rubber plantations. And yet, margin pressure and the competitive evolution of that industry forced the industry to break apart again. According to standards and supply chain with standard pricing and standard formats for tires and many other things, and that kind of disaggregation just hasn't really come to the wall street cell site. It's been predicted for a long, long time, but it hasn't really happened. And if you just look at the cost, and the complexity, and the operational risk inherent in these back offices. And the extra operational risk and complexity occasioned by the fact that all of these pairs of dealer to customer, and dealer viewer have potentially different flows and different processes and different documents. It's just this endless proliferation of complexity and that proliferation is itself a source of risk and cost. And you can imagine, hypothetically, a shared industry back office, this is really the holy grail, providing these services and doing it in an excellent way for the whole industry. You've got the beginnings of that with the clearing houses, that the clearing houses are really too low. And the stack for whatever reason they haven't gone up the stack to provide standard clearing and settlement processes to everybody, where everybody could just plug in. Another consequence beyond the operational risk occasioned by the tremendous complexity, and pointless and expensive replication is that there's credit risk everywhere. And as we saw, as just one example in the case of Knight Capital Group On August 2012, there's fragility in the technology and in the processes and that translates into broader systemic fragility. Now, there are some emerging technology applications, and there is a burgeoning interest in what are we going to do about all of these processes. And if you talk to people who are at the vanguard of this, probably there's a greater initial opportunity for innovation. And bilateral derivative rather than insecurities currencies and physical assets. Meaning can we deploy technologies such as the blockchain to streamline standardized share. Some of these clearing and settlement processes, and so there's a lot going on, especially in derivatives. Citi and Goldman Sachs are working with a startup called Axoni. And, Goldman has an investment in Axoni in the organization called principal strategic investments that I looked after for many years. If you look at an equity derivative, a standard equity derivative, an equity swap, it's got immensely complicated life cycle, in part because companies do all kinds of things. They take corporate actions the various times, and so this equity swap has to continually more. Then change into a new equity swap occasions by all of these corporate actions and also the two entities that traded. The stock can step out in favor of multiple new participants. And so it's a transaction that evolves and takes many forms, and many participants need to understand the form of the transaction right now, and they have to agree. And so obviously I've led to a description of a blockchain and indeed, there is a blockchain that's inspired by ethereum and MGF. And Citi have actually conducted the trade on this blockchain. So to my mind, that's an incredibly exciting development is very, very early days. There's a number of other similar in spirit kind of medium term pilot projects, to reduce the complexity and also the reconciliation and therefore. The fails and breaks between various security transfer and payment systems. The idea is to pay dividends to reward long-term shareholders an improved tracking of ownership position, type voting rights to the term of ownership. There's an incredibly interesting example. Some very early work that's happening to issue commercial paper. And trade commercial paper on a blockchain that's moving away from derivatives to a security. What's interesting about commercial paper or CP, two super important part of the plumbing of the market. You might have heard about that during the COVID-19 crisis as another example of some fragile corners of the market. What's interesting about Citi is that in many respects, the simplest possible security that doesn't require very many pieces of information. To fully specify a piece of CP. And so therefore, it seems like a good candidate to start the reference implementation, and putting the entire process of execution, clearing, and settlement, bilateral and multilateral on a blockchain. And they're one of the incredibly complicated problems is the final step, which is settlement or DVP delivery versus payment. The question is, well, let's say we've got the security the CP, represented in tokenized form on a blockchain. What about the cash, we need a stable coin of some sort or CBDC, or something that is also a digital asset. Where you can do this atomic transaction of delivery versus payment, either party A receives the security, party B receives the the digital cash simultaneously. Or nothing happens at all guaranteeing that you never enter into the case where one side got the security, but the other side didn't get the digital cash. So work in progress, pilot projects, super early days, and an exciting area for Innovation. I want to say a little bit about industry utility. There are a lot of them. I'm not going to go through these in detail, but here are some of my favorites. There's tri optima, which compresses derivatives portfolios. There's SWIFT, the society for worldwide interbank financial telecommunication. Manages about half of all the high value cross border payments in the world. In the foreign exchange market, there's a continuous linked settlement. Or CLS, which is a multi-currency cash settlement System that greatly mitigates what's called her stat risk, which is different currencies are obviously issued by different sovereigns in different time zones. And so there's different cutoff windows for the trading movement of those security of those, Sorry of those cash instruments. And you want to avoid the case where in an FX trade one side for instance gets the dollars but the other side doesn't get the yen. Because of this difference in the timing of the settlement windows and CLS does magnificent job of mitigating risk. Bank of New York Mellon, Boney, a for profit, obviously, and an incredibly important part of the financial system infrastructure that everybody uses. I would say JP Morgan Chase also provides custodian and payment services that are at the heart of the financial system. Now we've talked many times that DTCC, which, clears and settles quadrillions of dollars worth of securities transactions annually across, treasuries. And stocks and then others that other instruments such as agency mortgages. Euro clear is the European analogue of DTCC. And then over on the right, we have Bloomberg privately held software data media company that just about everybody uses. $9 billion of revenue per year. Mike Bloomberg on almost all of it. And there's an example of continuous innovation where Bloomberg has, successfully defended its price point which is the neighbourhood of $25,000 per Bloomberg user, per year, across a financial ecosystem in order of 350,000 users. And it defended that price point by providing more and more services and capabilities in analytics and data, through that one Bloomberg terminal. And then an early stage competitor to Bloomberg, really competing with very different Acquiring different users and that Symphony, we'll be happy to talk more about that, with one of our guest speakers and Symphony notice this act. There's that although those 50,000 users pay a high price point annually to Bloomberg service. It's a small fraction of the total number of people that work on the sell side and the buy side. And so the question for Symphony is can it provide services at a much lower price point to a much larger number of users. Back Office innovations while I described some early stage innovation, some of them using blockchain, but really compared to the importance. And the complexity and the size of the opportunity, and the dream of shared industry back office. There's a lot less innovation than one would expect. And I know that there are huge and hugely important companies to build here. Let me give one example of a company that is and innovating Bank of New York Mellon are about me. It is increasingly exposing all of this custodian services, hugely important part of the back office through API's. And you can you can chart a course for the evolution of shared industry back office here. But yet it's early days. So what's missing? Well, there's more more missing that actually exist. So we talk a great deal about know your client and and anti-money laundering. And yet, these processes are so important, so pivotal, so fundamental, so much complexity. And so much reputational and compliance risk are highly artisanal and bespoke have been multiple efforts to build shared industry utilities here. And they really have been only partially, partially successful and we really have to ask ourselves why they're just not enough margin. Is it just that hard to herd the cats and create standards. And of course, as I mentioned, the shared industry back office remains the Holy Grail. Seems to me that somebody needs to define the API, for a shared industry back office, and then begin adapting. The back office of a particular company so that it exports that API, and then set up those back office components in a multi tenant Cloud format. So that they can begin providing these shared back office services to everybody. And that to my mind is the only way we're going to tackle an important part of the cost side of the equation for the sell side banks. They hold a lot of capital that's appropriate to ensure systemic safety and soundness. But to get the returns on equity back to an attractive level. Rationalizing these back office costs across the industry is important. And it's not just a cost reduction exercise, not by any means. It's a a risk mitigation and complexity reducing move and one that will be of huge value to the industry. I'll pause there before going on to our last section, which is, regulation.
