 Greetings, and welcome back, everyone. Let me start by thanking Michael and Christie and Jenny and the rest of the Michigan team for making this possible and having us here on stage. I'm going to introduce everybody really quickly and then sit back down for a bit. Mark Newman, so we didn't have a cultural anthropologist, but we do have a physicist. So Mark Newman is going to take us out of the same thing. I suppose. He's going to take us out of our comfort zone and give us the big picture view of System Think. I'll point out that when there was a group of us at the OFR that decided to form a reading group around networks, network analysis. And the first place we started was by each buying a copy of Mark's textbook, which I can strongly recommend. I'll publish it. Thanks, you. And this publisher, thanks me. Then I'll talk. And then Adair Morse, I'm going to talk. Well, you'll see what I talk about. Adair Morse is going to talk about applications of intake to financial inclusion. And then Jared Sawyer at the end will talk about the regulatory response and what's going on in Treasury. So without further ado, I will sit down and give Mark a chance to speak. Thank you. Well, I hope this won't be too uncomfortable. So I am a physicist, so I don't work on financial risk. But what I do work on is networks of connections between people and organizations and what the structure of those networks can tell us about many things that we're interested in, but including certain types of risk that I'll talk about. So here's an example picture of the kind of thing that I work on. This is an example from my own work. This is a network of collaborations, in this case. It's collaborations, in fact, between a group of scientists. So the nodes in this picture represent a group of scientists. And the connections between them, the lines, represent who's been collaborating with whom. And the idea is that we're interested in the way the structure of a network like this could affect things like, for instance, the spread of information. So if you're a person on one side of this network and you want to communicate with a person on the other side of this network, it's going to take a long time for information to diffuse from one side to the other. Let me give you a few other examples. This is a picture of a friendship network. So this is a group of students in a US high school. And the nodes represent the students in the school. The connections between them represent who's friends with whom. You take a look at a picture like this. You can clearly see that there's some interesting structure going on. There are these four groups of nodes in the network. So what's going on with that? This is a network that comes from a study by my friend, Valdis Krebs. And it shows patterns of close physical proximity between people. So the nodes are again people. And the lines represent who's been in close contact with whom. And this is of interest because we're interested in the spread of diseases of various kinds. So I will talk about this more in a moment. This is one kind of risk that we can use to understand how networks affect risk. So if you're concerned about the spread of a disease, like for instance, the flu, then the flu spreads. If I have the flu, I can give it to you if we come sufficiently close to one another, if we're in physical proximity with one another. So there is some network of who's been in proximity with whom and the disease spreads over that network. So if you want to understand how the disease spreads, and in particular, if you want to understand what is your risk for catching this disease, then you need to know what the structure of this network is. I'll give you one more example. This is kind of a fun one. This again represents a group of high school students in a US high school. The nodes here are the students. They are color-coded, blue and pink, the traditional baby gift color code to represent the boys and the girls. And there is a connection between any two nodes if they dated during their high school career. So, right, so well. So there's that. So this is a fun network. I think many of us would like to have seen a network like this for our high school. I think many of us wanted to be that kid there, right? And we can see that one right there. So this is fun, but it also has a serious side to it. So this comes from a large multi-year study funded by the NIH. It's a multi-million dollar project involved interviewing tens of thousands of kids in schools all across the country. And why did they put all this effort into studying this? They put it in this effort because they're concerned about the risk of HIV and other sexually transmitted diseases. And if that's the risk that you're concerned about, then this is the network that you need to be looking at. OK, so that's an example of some of the kinds of things we study. So I want to think about risk in these networks. And the first example I want to give is of the spread of disease. It's one that's been very well studied. There is a long history going back about a century to the early part of the 20th century of people mathematically studying epidemiology, the way disease is spread, and more recently making computer models. So it's a very well-developed field, one of the few fields in which I feel like we really have a very strong theoretical understanding of the nature of risk. And so it's a good starting place for making analogies with how we might be concerned about network effects on other kinds of risk. So suppose I'm this red dot in the middle of this figure here, and I'm concerned about my risk for catching the flu, for example. So certainly that depends on who I have contact with. So in this particular case, I have direct contact with three people. That's the sort of first ring around me in this figure. So it's clearly the case that if I have contact with a lot of people, then I'm at greater risk for catching the flu than if I have contact with only a few people. If I'm a hermit that lives in the mountains and never talks to anybody, then there's zero chance that I will catch the flu. However, it's not just how many people I have contact with that matters. It's also who I have contact with, because it depends on what their level of risk is. If I have contact with people who are themselves high risk individuals, then that makes me a high risk individual as well. Conversely, if I have contact with only with people who are at low risk, if the people I have contact with are very unlikely to have flu, then there's very little chance that I'll get it. I can't catch the flu if the people I have contact with themselves don't have it. It doesn't matter how many people I have contact with. If they don't themselves have the flu, then I can't catch it either. So it matters not only how many people, but I know also what their level of risk is. And their level of risk depends in turn on the people they have contact with. That would be the second circle out from me in this figure. And their level of risk depends on the people they have contact with and so on add in for an item. So it really is a network effect. It doesn't just depend on who I know or my local environment within the network. It goes on, add in for an item, and hence it depends on the structure of the entire network. In order to understand my level of risk, I need to understand the structure of the entire network. So it really is a systemic effect. Even though I'm only concerned about a local thing, me personally, it's a systemic effect of the entire structure of the network. So as I say, this is now a well-developed field. There's been some very sophisticated work in this field. Here's one example from a study by Steve Eubank and his collaborators at Virginia Tech in which they constructed the entire network of connections between people in the city of Portland, Oregon. So they made a detailed micro simulation of the entire city, including every street, every car, every person, every building, and used it to work out who comes in contact with whom and then to predict how diseases would spread through the city. Here's another example from the work of Victoria Kalitsa and collaborators. This shows simulation results from a model of a hypothetical outbreak of the flu that starts in Asia and spreads worldwide. And here they're calculating levels of risk in individual cities within Europe. So an example of a classic result in this field is the following. The risk to an individual with catching the disease clearly depends on how likely the disease is to spread. There's a fundamental parameter in these models which is so-called transmissibility or transmission probability, which is the probability that a person who is infected spreads the infection to their uninfected neighbor in the network. So when the transmission probability is very low, the disease doesn't spread very far and when it's high, it spreads very easily, as you would expect. But it's more interesting than that. What you see is so-called phase transition behavior for low values of the transmissibility. Nothing happens at all. The disease simply doesn't spread. It might spread to one or two people, but it's just gonna fizzle out and it just dies out and then there's no disease and it doesn't get anywhere else. But if you have high values of transmissibility, then it spreads and you have a big epidemic outbreak and the transition between those two happens, it's a sharp transition. It happens at a specific point. Below that point, just nothing and above that point, you have worldwide spread of this disease. So there's a sharp transition where you just step one foot over the line and suddenly you have disaster. Your disease is spreading worldwide and you have sort of systemic failure in your system. So this is a classic result in the field, been well known for a long time. It's maybe suggestive for other kinds of risks that you might be interested in. However, to be fair, I don't think that this simple model of biological infection is a good model for other kinds of transmission like sort of social infections of various kinds that might be sort of more applicable to the kinds of issues we're talking about on this panel. When people talk about social infections, I'm thinking of things like panic infecting a community. They often use a different model called a complex contagion model. So in standard biological contagion, you catch a disease from one other person whoever infects you, that's the person who gave you a disease. In complex contagion, you can catch a disease from more than one other person. So an example of this would be, suppose you hear a rumor that something happened. So one of your friends tells you something and you're like, okay, that's weird. That doesn't sound like something real. I'm not gonna believe this rumor. It just sounds very implausible to you. You ignore it. But then later on, another of your friends completely independently tells you the same thing. Hey, did you hear that this happened? You're like, well, that's weird. Okay, maybe this is real. And now you believe it. So it's a different kind of infection in which you have to get infected by two people before you yourself get infected. So this is commonly used as a model of the spread of rumors or information or panic in social systems. And it has an interesting feature. So it still has that sudden transition from no spread to spread. But now the transition becomes much sharper. You go from nobody infected, sudden jump, lots of people infected. In fact, in certain parameter regimes of this model, the jump is what we call discontinuous jump, meaning that you go from nobody infected down here to suddenly almost everyone infected discontinuously. It's just a sudden jump. And this is a particularly insidious form of a risk because there's no signal that it's going to happen before you step over the line. Everything looks completely normal. And then you just take that one last step across the line and suddenly a huge fraction of people in the system are infected. So you could ask, what can we do to prevent this kind of systemic failure? So this will be my last slide. There are standard techniques that we use for preventing this kind of thing. So one, so in the biological regime, one is just mass vaccination. We just give everybody the flu vaccine. And this works, I mean, obviously it works, but it's immensely inefficient. You have to give virtually everybody the flu vaccine in order to prevent the flu from spreading. An alternative that's been recently suggested is a more targeted vaccination strategy where you only vaccinate the people who are at highest risk for an infection. And if you do that, it turns out that's enormously more effective. Instead of having to vaccinate 90% of the population, you only have to vaccinate 10% of them or something like that. So it would be a lot less expensive and a lot more efficient. The only catch is that we don't know how to identify the people who are at highest risk. We have no way of finding out who those people are. If you knew the entire structure of the network, perhaps you could do it, but normally we don't. So a third possibility is what we might call reactive prevention where basically what you do is, when it looks like you're getting close to an outbreak, you tamp it down. You say, oh, I see something bad going on over there. Let's go over there and vaccinate people. If you see something bad going on over here, you go over there and you vaccinate people. So an example of this would be the ring vaccination strategy that was an important part of the elimination of smallpox. That every time you see an outbreak, you go and treat people over there to prevent it spreading. So there has been work on that kind of thing as well. And it does work, but there is a catch. So this plot here shows, as the transmission probability increases, there's your traditional epidemic threshold where an epidemic disease breaks out. Here's what happens if you do this sort of reactive prevention, you do much better. The transmission probability can go much higher before the disease breaks out. But the catch is that when the disease does break out, it's much worse. There's this huge jump to suddenly lots of people infected. The analogy I like to give here is to forest fires. In forest fires, they practice fire suppression where you put out any little fire that starts. And this prevents fires from happening, but it has the negative effect that dry wood tends to build up in the forest. There's a lot of brush in the understory of the forest that's building up over the years because there haven't been any fires to burn it out. And then when a fire does happen, it's particularly bad. There's all this dry fuel there and you get a very bad fire. So this is kind of an analogous thing that you can prevent these things from happening, but it's not necessarily a good thing to do because the net result is that when something bad does happen, it's especially bad. Okay, I'm out of time here, so I should stop talking. Hope I've given you some food for thought here. We'll move on to the next speaker. Thank you, Mark. Well, they're setting up the slides. I'll give you the disclaimer. Save us a little bit of time. So what I'm about to say are my thoughts alone, I'm not speaking on behalf of the OFR or the US Treasury. They don't let me do that. What I'm gonna do is I'm gonna start by providing a working definition of fintech and then run through a handful of examples to, there are so many dimensions to fintech and financial stability, can't possibly cover them all. So I'm just gonna hit a handful of what I hope are illustrative examples and catch some of the themes. And I do need my slides. It's okay. So I can start with my working definition of fintech. I actually don't have a slide for that. So the way I've been thinking of it is that fintech is not simply a combination of finance plus technology, sort of two great tastes that taste great together. Of course, we've had combinations of finance and technology for centuries. Dick mentioned the telegraph earlier. Actually, technology literally goes back to earliest recorded history as a key component of finance. So Kenea form writing was built out in large part to support bookkeeping in Sumerian granaries back in the day. So fintech is about the, or I'm sorry, financial technology is about the earliest thing. We know about, my definition of fintech is gonna be a little bit different and try to capture the fact that we didn't talk about fintech until very recently. This is a new phenomenon. So in my definition, fintech is a financial service that is enabled by a computational technology. In keyword there is enabled. So the notion is that the service would be infeasible if you didn't have the computational technology. So for example, counter example, ATM machines give us the ability to withdraw cash on the weekend. I'm old enough to remember life before the ATM. We could in fact withdraw cash from our bank accounts on the weekend. We did it by cashing a check at the grocery store or the convenience store. There were mechanisms for this. The sorts of things that I'd like to count as fintech are approaches designed to address big data scalability problems in finance. So fintech is our response to the big data revolution. In other words, you can't use linear technologies to solve exponential problems. Fintech is the exponential technologies that we use to address the big data problems. And that's why you see, for example, at London Club, a third of the headcount is computer scientists because you need that kind of power. So I knew that Mark was gonna talk about phase transitions. So I came up with one of my favorite little pictures. This is a phase transition from financial markets. It's also closely related to a financial stability question. So you'll see that something happened there about two thirds of the way through the time series. That something was the failure of Lehman Brothers in September 2008. And it also, the way I want to interpret it coincides with the beginning of the fintech revolution. And I'll talk about that in a second. First of all, what are we looking at? So the blue line there is a market price that is the interest rate charged on federal funds. These are overnight interbank loans. The money on these transactions settles in the reserve accounts of member banks at the Federal Reserve. And a key point to observe there is that because it's held at the Federal Reserve, there is no chance that the bank holding your money overnight is going to fail. The alternative is that you might deposit it at another commercial bank and those banks could in fact fail. The blue line is the interest rate. The orange line is the aggregate balance in those reserve accounts over time. And you'll notice that for a long period, the aggregate balance was hugging close to zero. And in fact, there are minimum reserve requirements and the banks would keep their reserves as close to the minimum as they could. And in fact, if you trace that back, I look back as far as the early 1950s, it never budges over the flat line. It stays down there. And the reason is that the Federal Reserve doesn't pay interest. So if you've got money that you need to hold, you can put it someplace at interest or hold it essentially for safekeeping with no interest. Well, obviously something changed here, right? I think what happened was something similar to the complex contagion story where all of a sudden there was large scale distrust in the system as a whole. There was a fight to quality. And even though you weren't earning interest, you were willing to hold large sums of money in your reserve account at the Federal Reserve. An interesting question, given that these balances have persisted at a significant level well after the crisis is whether we're in a disequilibrium situation. So there's an argument to be made from sort of standard Keynesian models that the interest rate is constrained at the zero lower bound. The market current price is negative. We can't get there and therefore allocation of resources is suboptimal. Or are we in some new equilibrium that we've just never seen before? Is this a brave new world? Don't have the answer to that. I think it'd be an interesting thesis topic for a student in the Ford School. But there it is. How does fintech tie into all this? Well, one of the things that brought us to the crisis was of course subprime lending. And the subprime pipeline was an enormous, many splendid thing. But so there are a lot of technologies that enabled it. I'll just mention two. And in honor of Jillian's remarks about acronyms, I'll give them an acronym form. First is MERS, the Mortgage Electronic Registration System. So when loans go into securitization trust they have to hop through a number of different ownership stages that was operationally impractical to do its scale until the mortgage industry concocted MERS. MERS has a lot of interesting stories around it. I won't go into, but it was a technological shift that enabled large scale securitization. The other is algorithmic pricing for CDO tranches. So a lot of complicated math went into that. Again, the large scale securitization, the distribution of the credit risk would not have been possible without those pricing algorithms. Okay. Quickly. So we just had a panel on HFT. So I'm not gonna spend a lot of time on this. A really interesting topic. Clearly HFT is an activity that is impossible without the technology. I can, I remember pre ATM. I also can remember one of my first jobs after I got my undergraduate degree was in a foreign exchange trading room. We still had at least one person on the desk who had come up straight out of high school onto the trading desk. And I was led to believe that that was a fairly common place. Why would you put an 18 year old on a trading desk and let him or her make million dollar trades on a regular basis? Why does this make sense? It makes sense because the fastest computational engine we had back at that time were 18 year old brains. So FX trading is ultimately not that complicated. You just need to have good reaction times. What we're seeing now is an arms race effectively in transactions latency. One thing about arms races is the participants seem to want to keep racing long after everyone else thinks they've crossed the finish line. And there's a policy question in there about whether it makes sense to intervene in some way. Couple of technical issues that didn't come up and I'm gonna be looking to some of the folks from the HFT panel to maybe inform me on this, but regulatory time stamp resolutions in these markets are orders of magnitude slower than the transaction latencies. And what's even scarier to me is that we're not there yet but if trends continue, we keep getting faster, we will hit the physical limitations of our best temporal measurement technologies which is common view satellites. And question I have is what does this imply for policy and behavior around front running and not front running in the sense that John mentioned, but illegal front running where you get a client order and trade in front of it because you know the client order is gonna move the market. If we can't time stamp the trades, how can you possibly prove front running? John and Michael both pointed out that there's a tendency for the slow money to migrate to markets where they're less likely to be taken advantage of. Is it possible and what would be the conditions that would trigger a large scale withdrawal of participation from markets if you knew that the system was rigged? I don't know the answer to that. I'll be again curious to hear what the experts have to say. Another question that strikes me in this context is the problem of tight coupling and whether the operationally things will move too fast across markets in ways that were unforeseen. Another example is blockchain. Blockchain is not a fintech by my definition. So blockchain is not a financial service. It's a computational technology, but obviously blockchain is an input to a lot of financial service approaches. So it's worth considering. One thing that blockchain does is to make clever use of dependency to its benefits. So typically in software architecture, dependencies are a problem, not a solution, but blockchain turns the tables and it uses inter-temporal dependencies. So the hashes in the blockchain accumulate over time with the effect that you get a lot of stability in the recorded information. You can't disagree with the blockchain without disagreeing with the entire history of the blockchain. The other thing that blockchain does closely related is it creates this cross-sectional dependency. There's a consensus formation mechanism. So participants in a particular blockchain have to buy into the chain, literally. And that creates what I'm calling epistemic closure. If you don't know what that is, it's a nice cocktail party phrase for groupthink. Basically each blockchain is a Truman show. And you have to, if you're gonna participate in the chain, you have to agree to the consensus and agree to everything that's in the Truman show. I think that that has lots of interesting economic and mechanism design implications that we've only begun to consider. The Bitcoin blockchain started this off and we were at it before we knew what we were doing, really. Now there are lots of blockchains and with a range of mechanisms for permissioned access and other things. But I don't think that all the implications have been explored or thought through. So governance around blockchain and hard forking in particular is an interesting and important policy question. Another one that we haven't confronted yet because the landscape is still relatively mature is if you've got multiple blockchains. So imagine two central banks in different jurisdictions that disagree on whether a trade is settled or not. How do you reconcile that? Whose blockchain is right? Don't know. I don't think it's been addressed. And then lastly, this is an example of a larger issue of information centralization in financial markets. So one of the things that's come out of the crisis is centralization of clearing. So with that comes a centralization of information. The benefits for margining and default management are pretty clear in that centralization. The side effects involving information centralization, I'm not sure have been fully thought through. I'll give one particular example which is the Equifax breach. So the Equifax business model was created long before there was an internet and the cybersecurity issues were essentially non-existent but the internet did occur and the digitization of credit risk information occurred and what had been an operationally efficient business model became a glorious honeypot for cyber thieves and we need to take that into account. And then lastly, one example, I could probably talk for an hour about this picture. I will keep it very brief. This is an example of the supervisory response. So another thing that's happened since the crisis is we've started getting fairly elaborate network data that simply did not exist in one place before. These are network statistics calculated for CDS obligations. So we have weekly snapshots of the US CDS market and we're able to construct a network of who's obligated to whom and then we're calculating for each network snapshot, what is the complexity? Well, what does complexity mean? We're looking at the complexity of unwinding or resolving the network if we had to and the key thing we are looking for there are cycles. So if an obligation comes back around through various participants back to its starting point, we argue that that creates a coordination problem. And so cycle rank complexity is one way of counting the cycles in the network. There are a lot of other ways. That's what we're doing there. You see across the bottom the sort of crashing waves of cycle rank complexity for individual vintages of a particular index CDS contract and then the blue and green lines across the top are the complexities of the market as a whole calculated two different ways and that is going through the London Whale crisis. So complexity is growing as the crisis builds and in the middle of the picture, JP Morgan starts unwind things and complexity starts to drop again. And with that, I will. Thank you. So thanks, thanks for having me here. So I tried to be like a lawyer today. I have one slide. So perhaps that was not wise because I'm gonna miss my slides here in a second. But okay, I'm gonna talk about four points and actually the one slide partially is so that I can, you can have these four points sink in. I'm gonna talk about the lending side more than the trading side. And the panel, there's just one. Yeah. Do you know what that makes you our favorite? Okay, thank you. The panel is supposed to be about systemic risks. And so I'm gonna talk about areas where our consumer facing the loans and the credit markets and so forth where we should think about the future and where the capital is coming from it and these sort of issues. So the first one, which when we were organizing this our organizer said, well, that's not, you need to tie that tighter to systemic risk. So here we go, we're gonna try. The idea here is there's some natural data monopolies. Okay, and so let me start with an example that's not here in the US. IMPESA and the telecoms in parts of Africa, right? IMPESA owns the mobile banking, everyone mobile banks in Kenya and a bunch of other places. The telecoms are the provider. Okay, these are natural monopolies, if you will, of data. Well, what does it mean for credit? Well, they're also the natural providers of credit because you can see all the transactions. You can better credit score. And they are the bank, essentially, right? So that's a natural credit provider. Okay, the government's in Africa in different, to a different degree, of course, opening up of these mobile payment facilities to other providers, but it's a natural monopoly. Okay, so that we should be concerned about natural monopoly for a number of reasons, but in the case of thinking about systemic risk, it means where's the risk? Who's holding the capital? Who's providing the capital? That means there's a natural monopolization of the capital going into the credit markets and all these consumer-facing instruments in Africa or Olegopoli, if you will, with the different telecoms. So I gave a keynote on fintech and Kenya to a bunch of governments and they're like, what, no, yeah, I'm in there. And I think, though, if we bring that home here, or this may not be home, but we bring it here, what is the parallel? Well, there's lots of parallels we haven't thought about. It's about data, it's about information. So I can better credit score you, any of you, based on your observables, income and FICO score, if I had your Facebook page, why? Well, your Facebook friends are predictive of your risk, things I can't see, right? That's a natural monopoly, why? Because Facebook is a natural monopoly. PayPal, again, payment streams, Amazon, I can keep going, right? These are natural monopolies over information. There's a paper, this, Jorgensen has a paper on how you can credit score people based on consumption items that buy, right? And so these things, this is the future. You know, all these things are collapsing. And if you don't believe me, we need to go to our leader country, which right now is China, right? China is way ahead, Alibaba is way ahead, right? They own payments, consumption and credit. And they can do a better job of scoring everybody. Okay, so now I'm supposed to talk about systemic risk. Well, systemic risk is where's the capital? Who's holding the risk of these natural monopolies? What's the risk of the data? There's lots of things that are systemic here when you have natural monopolies over information. Okay, so that's point number one. Point number two, which is related. The capital for platforms. So as we heard from the Lending Club earlier, the peer to peer model, the word peer, the second peer in that is not really peer anymore, it's peer to hedge fund or peer to banks, right? The investors in peer to peer are institutional, largely, not all, not all, but you know, this is one of three sources or one, three models of where the money comes from for the consumer facing platforms on lending. And it's the original one, and it's the one that appealed in the disruption, particularly coming out of the crisis, even though these existed before the crisis. So the idea there in the peer to peer original is that the money is diffused. Well, the reality is the money is not diffused that's coming in to fund these platforms. And so the three models are here, I don't know if I, yeah, I wrote them up there. And a lot of the models, most firms are some sort of hybrid of these three. So there's the peer to peer. And I should mention on the peer to peer, by the way, the default risk and the platform risk is different for retail clients than it is for institutional clients. Then my regulator here that's gonna come after me. It's an interesting fact that I didn't realize that the exposure of a retail investor to Lending Club is different than exposure of an institutional client to Lending Club, which is an interesting artifact that the SEC tells me. The second one is what I called a platform pooler or an ABS pooler where the platform takes loans and pulls them into an asset backed security. And then those go off to institutional investors. So eliminating one financial intermediation in pulling them themselves. And here we, again, who holds this capital, right? Who's the holds this risk? Who's providing the capital and holding the risk? And I don't think we have any idea. Lending Club may have an idea in terms of their risk and each of the platforms will have an idea but I don't think we know who's got this risk in the economy. Certainly the big investment bank type entities which we don't have investment banks anymore but these entities are big participants in this market. And but where is this risk sitting? I don't think we have a good idea. So that is the second type. The third type, which many of the platforms have moved further this way, self liquidity funding. I don't know what I called it. And the idea here is that the platforms get either loan facilities for a potential ABS or they get loan facilities that are on their own balance sheets. So again, where's the risk, right? Is it sitting on the balance sheet of some of these platforms? Is it Goldman Sachs holding a lot of this? I don't think we have any idea. Or at least I don't. Maybe some of the regulators hopefully might know but I don't think we have any idea. Now the volumes thus far have been small relative to thinking about the total 13 trillion in consumer finance float but they're getting bigger. And certainly rocket mortgage is an enormous entity but of course that has their own mortgage side but I think we need to get ahead in understanding this risk where this risk is to think about what is the regulatory frame? Is there one? Is there a need to be one? And so here's something we're thinking about. Okay, number three. Number three is about web bank essentially. The banks that do the funding, the banking process for the platforms tend to be these industrial banks like web bank. This is the weakest of my knowledge sets here but these banks have what I believe called a light touch regulatory framework. And so they're banks but not regulated as much as some other banks. I mean, I don't know those details but I wanted to bring up another issue that came up earlier as well. So we've transferred, think about what these banks are doing. One thing they're doing is you import the state banking regulation of web bank or the industrial bank to the platform. So Lending Club for example, the regulatory framework for the banking and that's the usury laws and things like that come from Utah, okay? And that's fine. The credit card companies have long done this. It's not like this is new. The difference though, I do think we need to think about that in context of moving the part of consumer debt from credit card revolving float to installment. And the reason I'm bringing that up is, as we talked about that discovers and that was interesting that discovers now a platform doing installment loans. Well, Lending Club maybe should be a platform doing discover as well and you may owe me millions later for having said that. But the, when we hit a downturn, you've taken a customer that had $16,000 paying higher interest rate on a discover card or something else on a credit card and you've moved them into installment product and they get out of debt quicker. However, if they don't have money for the payment the payment is much larger than it would have been on the credit card. So does that become systemic in terms of this because we have a framework that is largely installment based and largely based on a state banking regulation that's perhaps one of the more flexible in terms of the rates and things like that. So bringing these together, I think there is a systemic, as we see more and more consumer debt moving into installment, which again, I like the benefits of that getting people out of debt quicker, but it comes with some costs in a downturn and I don't think anyone has talked about that at all and I think it's worth talking about. Now the final one, I could talk about, which I don't even know if you can see, I could talk about for an hour, discrimination and democratization. So I gave a paper here yesterday at Michigan and the Finance Group on discrimination and whether the algorithmics credit scoring does better or worse in terms of discrimination than having loan officer touch where you can see faces essentially. Okay, algorithms you can't see faces. Okay, on the other hand, algorithms, you add more data and adding more data you may be putting in discriminatory data. And so then the question becomes, what do you mean discriminatory data? And we get into this legal term of disparate impact which has already come up once today. And so some co-authors and I including a lawyer from Berkeley Law have written a paper and we try to connect the idea that economists talk about a statistical discrimination to the idea of disparate impact and what the courts are struggling with. And I think it's really very important because the courts don't have a framework now for as we move into big data, what it means, what is disparate, how could you possibly test for disparate impact? The current court terminology is that a lender can, if there's discrimination in the raw statistics, a lender can say, no, no, we're using variables that are legitimate business necessity. That's the term in the court. Well, legitimate business necessity, actually the economists have something to say, what that means. I can write down a model from life cycle model of repayment risk that comes out of your understanding income and wealth, your expense levels in your area where you are, you just write down an economic model of repayment risk for a person. And variables in that repayment, in that model, are legitimate scoring variables. Variables outside that model are illegitimate. However, we can't see all the variables inside that model that I wanna write down. For example, family wealth. You can imagine some minority groups have different family wealth structures than some of the majority groups. And so how are we going to, going forward, give, as we move more and more into big data, and algorithms to credit scoring, how are we going to move, are we going to regulate the idea of lending club and others to a better job of sorting people because we can lower the interest rate on average for everyone if we do a better job of sorting. This doesn't work out best for everyone, but on average it does. And so the idea in what we've mapped in doing the study is that if a variable, the correlation of a variable with the hidden variable that you can't see, if race or ethnicity or gender comes into that correlation only through the hidden correlation with wealth say, that should be legitimate statistical discrimination. The extent that you're over adjusting race beyond the hidden wealth variable or whatever you're trying to proxy for, that should be illegitimate statistical discrimination under disparate impact discussions or illegitimate business necessity because it's not a business necessity if it's not in the fundamental risk model. So we think we're going to push knowledge on that and by the way, the study finds that the fintechs are doing better than the lenders now, and this is in the mortgage market, then the traditional lenders more so on the smaller traditional lender side because it must be because of a lack of facial recognition. So we still are having that loan officer in group bias, whether they know it or not. But I think the future we need to think more deeply. So finally, I know I'm out of town. I must say something about democratization and inclusion. The idea, it came up in the nice comments from Richard from Lending Club about inclusion. So the lending platforms are not being more inclusive of people at this moment, at least in the United States. So in other words, I don't think Lending Club would say we're reaching out and getting to more Americans on the consumer side. What they would say is we're reaching out with more inclusive products, okay? Those are two different things. We're not seeing that technology has yet democratized on the consumer side, and they may push back on the small business side. But on the consumer side, the people that are going to the platform loans, the large ones, are taking credit card debt and getting lower interest rates and getting an installment product that they presumably prefer by revealed preference. Okay, so I'm not saying that's not good. That is good. That is that we have a more inclusive product offering that fits people better. That's what they want if they're moving that way. But it's not that we're reaching out and democratizing finance, right? We have not done that on the consumer side with these products as of yet, to my knowledge. And so as we talk more and more about FinTech and inclusion, which is an important word, we need to understand what is the geography? What is the geography in a bunch of different ways? Demography of FinTech and inclusion. And how can we understand what are the hindrances and how to provide models that are inclusive with, again, proliferating the product offering, which I think appeals finding better fits for better people. So I think there's a lot more research that needs to be done on this particular point and the fact that Lending Club is talking about it is a wonderful step, and we need to do more on the academic side to get to these points. Oh, I was supposed to put it in terms of risk. I think up at the political risk, right? And which is true, right? We've seen it in the elections. People are angry of feeling, we had this conversation last night, left out or whatever lack of opportunity or however you wanna think about the anger in the population. I think inclusion is a really important topic. Anyway, out of time, thanks. Well, thank you very much for having me here today. I'm Jared Sawyer, I serve as Deputy Assistant Secretary for Financial Institutions Policy at the US Treasury. And I promise to keep my remarks pretty brief so you guys can ask questions and we can have a moderated panel because I know there's a lot of fascinating research that's been presented here today. But what I thought I would kind of do is provide a little bit of an overview of what Treasury has been doing in the financial regulatory space since the Secretary assumed office and how we think about financial regulation and how that trickles into a discussion on financial technology. So a lot of our work can be traced back to an executive order that was issued by the President in February that you'll hear me refer to is called the core principles executive order. And really what the core principles executive order did was it outlined certain core principles that the President had in mind and how he wanted the financial regulatory framework to look like and to match up with certain kind of core beliefs. And those core principles focused on issues like consumers having informed choices, being able to save for retirement, making sure that American companies could be competitive internationally, making sure that we have effective and efficient regulation and that American taxpayers are protected from any kind of systemic risks. And so that core principles executive order certainly outlined the vision for a financial regulatory framework under this administration. But it also directed the Secretary to study the current regulatory framework and evaluate whether the current framework was consistent with those core principles. And the mandate to us was if we found gaps or we found a misalignment of principles, we were to make certain recommendations. And so starting in February, the staff undertook a very rigorous stakeholder engagement process meeting with lawyers, market participants, consumer groups, think tanks, other researchers to understand the current regulatory framework and what we decided to do instead of issuing a single report, we said, okay, let's take a step back and let's be very comprehensive in our study. And so we decided to break the report up into four separate reports so we could be comprehensive. And it focused on certain key sub-sectors of the financial sector. So the first report came out in June and it focused on banks and credit unions. And really what that report did was it looked at regulations focused on capital and liquidity and stress testing of financial firms, kind of some of the core tenants of the Dodd-Frank Act and many of the rules that we saw that came out of international standard setting bodies post-crisis. It also looked to a certain extent at some of our consumer financial services regulations. And we made a number of regulations where we thought that the regulatory framework could be realigned with those core principles. And so we focus on regulations that can allow community banks to be a little bit more nimble and to hopefully slow the concentration that we're seeing in many of our rural areas in the banking sector. Making sure that regulations are appropriately calibrated. So in many cases, not calling for a full repeal of a regulation, but saying, okay, let's make sure that the regulations are tailored and appropriately calibrated to consider the size of the institution, the products and services they're offering and their risk profile. The second report that came out in October was focused on our capital markets. So it's how does the regulatory framework impact the flow of capital and our market functions? And so that focused on market structure, the availability of companies to go public and reducing that burden. It focused on issues about clearing of derivatives and central counterparties and small business capital formation. So some really key topics when you think about the United States having the most robust and dynamic capital markets in the world. The third report that's been issued and the third and last one that's been issued as of today focused on the asset management and insurance industries. And both of those were in a single report. And so in that report we focused on issues like what is the appropriate evaluation and methodology for measuring systemic risk in those two sub-sectors. It focused on how should the United States engage in international standard-setting bodies because many of the regulations affecting asset managers and insurance companies have originated in international forums. It certainly focused on issues about effective regulation and efficient regulation. So it looked at how can electronic funds transfers, ETFs, how can the approval process be streamlined through new SEC regulations? So those are some of the core functions and work streams that Treasury has undertaken since the Secretary assumed office and that executive order was issued. The fourth report, which I probably is of most interest to this audience, will focus on non-banks, but also financial technology, fintech and innovation. And we will probably spend a lot of time focusing on those last two issues. Before I kind of dive into what you might expect in that last report, I think it's really critical to kind of take a step back and say, okay, if that last report averages the same length as the first three, we're looking at a roughly 800 and 900 page analysis of the US financial regulatory framework with hundreds of recommendations. And I think it's a very thoughtful analysis and study and one that I hope will ensure that the United States has a regulatory framework that is appropriately calibrated and promotes financial stability, innovation and competitiveness for US firms and allows certain market access for foreign firms. So focusing on financial technology, obviously financial technology is a loaded word as we heard and it's very hard to define. You can think of reg tech or regulatory technology. You can think of consumer facing technology. You can think of kind of core processes of financial services firms. And those are all issues that we hope to cover in that fourth report. We're in the process right now of scoping out the work we wanna do and some of the core key questions we wanna ask ourselves. But heading into this report, especially this report, we're taking a posture of we really need to learn and be thoughtful in what we do in this next report. This report will be a forward leaning, forward thinking report in that it hopefully will set out our vision of a regulatory framework that promotes innovation and allows kind of a robust and dynamic financial technology marketplace to develop while still having consumer protections and system protections. And so some of the questions and key questions we may ask ourselves is how would a given development fundamentally alter a market structure or financial institution? How does the relationship of financial technology and incumbent financial institutions change the regulatory relationship? And how does a given fintech firm impact the infrastructure of a incumbent financial services provider? So those are some of the key questions we'll be asking ourselves in this report. Kind of taking a step back and thinking of, okay, so what is Treasury's role other than kind of studying these regulatory frameworks? Well, the panel discussion is on systemic risk and the Treasury Secretary chairs the Financial Stability Oversight Council, the FSOC, which is charged with evaluating and monitoring financial stability of the United States. So we always look at new developments through a lens of that role and his leadership of that council. We need to evaluate and appropriately evaluate financial stability and systemic risk. So that's always a lens that we focus on when we're evaluating regulations or new frameworks. And then also we have an incredibly important role and I hope we'll touch on today in cybersecurity. Treasury serves as the sector-specific coordinator for the financial sector when it comes to cybersecurity. So we're not a regulator, but what Treasury does is it sits in a coordinating role across all of our federal regulatory agencies and tries to coordinate responses, monitoring, recovery and prevention of any cyber attacks that might occur. What we've certainly seen over the last couple of decades, but particularly this last decade, is an increasing reliance on technology by financial services firms, which has really promoted a very strong financial sector, but it also results in increased complex and layered cyber risk because you have such a reliance on financial technology in these firms. And so what we really try to do in the posture we take at Treasury is there needs to be a public-private relationship in addressing cybersecurity. And so we spend a lot of time working with the private sector from a resilient standpoint, going through exercises to figure out how you might respond if a cyber event were to occur. We push the sector to think about as you're building and implementing new technologies and new processes, are you thinking through the cyber risks and how do all those different technologies interplay with your legacy systems? And so these are all kind of key questions that we think about, and that's just another lens that we use when we think about financial technology. So with that, thank you for having me here today and look forward to having a panel discussion. So, Brian Knight, Raketa Center. I was just curious whether or not you thought, and I guess this question is sort of, on first impression for Dare and then the rest of the panel if they're interested, St. Louis Federal Reserve Governor Bullard talked about the risk that fintech might pose to banks by slicing off all the high value services, leaving banks as basically just dumb pipes and making them more brittle. Do you view that as a potential systemic risk to the system and if so, it would seem perverse to me that the answer would be, well, then we can't let the banks become brittle. Like, don't we need some method of unhitching the system from the fate of banks? I don't think the banks are gonna let themselves become brittle, I don't listen to me. In 10 years, I don't think we're gonna see much distinction between any of these financial institutions, right? The banks will look like fintechs, the fintech name. I mean, it's all collapsing in my view, whether it's collapsing via relationships that we already have. The fintechs are providing CRA coverage for the banks already in terms of getting loans to the small businesses and people in CRA districts. The relationship with the discoverer going from being a credit card to a lending platform, Capital One is a bank, right? I mean, I'm not worried. I think it's gonna take care of itself. Thank you. I'll jump in on that one too. I think a couple of things that distinguish banks from the fintechs we're seeing. One is the level of capitalization. So banks are typically much more significantly capitalized and also more heavily regulated. So a lot of the fintech startups are regulatory arbitrage plays. And I would expect the force of capital to, especially when they go through the first downturn, right? So if you think about what happened to subprime loans, it was a great game until they saw the first downturn. The fintech ecosystem has not been through its first downturn yet. I would expect to see a lot of consolidation in the vein that Adair suggested. So I think I just wanna sort of push back and follow up on that one point. In the sense that to the extent it's a regulatory arbitrage play, there's also a significant difference in business model with the federally insured deposits. And so that, and I think your point is well taken that everything that arises might converge. But on the other hand, to the extent that there is a need for something that happens outside the depository space, niches or whatever, then we may see some sort of independence there. But with contracts, right? We're in the law school. We have to talk about contracts, right? These things, you can form the relationships with contracts and still have the regulatory arbitrage you're talking about. As I listened to Jared's description of the changes in the regulations, obviously the existing financial reforms, Dodd-Frank, not perfect, not necessarily well adapted to different kinds of institutions. But having lived through the savings and loan crisis which was a step toward deregulation, greater efficiency, greater competitiveness leading to disaster, then we went through the derivatives deregulation, can't touch those, that's the free market, it'll all work itself out, even bigger disaster. So I do have to confess I shudder a little. When I hear you talking, you're only talking in general terms, so I can't be critical really, but I shudder a little. When I hear you talking about more innovation in the financial sector, because I'm not sure, I mean so much of that innovation has been at the direct cost of obvious or hidden systemic risk. Certainly, and that's a fair point. I would say really the approach that we tried to take and I think that you'll see in looking at the three reports that are out so far, in many cases it's a focus on tailoring regulation to the different and myriad of financial institutions and types of financial institutions we have in the United States and it's a recalibration of rules. It's not wholesale repeal in many respects. It's saying what we've seen, this is a good moment in time, nine years post-crisis to say what is working and what's not and do we need to tweak the calibrations on some of these rules to make sure that the market is completely efficient to the extent it can be completely efficient, right? I think one of the mantras that we take when looking at innovation as well is you also wanna have responsible innovation. You don't wanna have innovation that creates risk to the system or puts the consumer at harm. And so as we evaluate what is the appropriate regulatory framework for innovation, those are also principles that we will keep in mind. If I can add a thought, so in my response to the last question, I think as we, the overlay, I think as we look forward, we need to start from the idea that this intermediation took a layer out at least in the sector that I was talking about, took a layer out of financial intermediation in the ABS market. So there's work, but Tomah, Philippon, that each layer is 2%, right? So you have a 2% to work with. So there's 2% returns there that someone when it was either the consumer's paying or more interest or the investors are not getting as much to work with in terms of getting the structures right. So it's not that I'm, I think the innovation helps with efficiency, but I do think we have to get the regulation right. I'm not against regulation, you know, and it reflects the same as what Jared's saying. It's just we should see the benefits of the innovation and embrace those at the same time we're looking at what are the implications, some of the pictures that we saw first thing this morning. And so we need to think about the implications, but that doesn't mean that we need to say, wait, stop, because we have some benefits just in terms of removing one layer of financial services, which is worth 2%. And so I think that is a benefit to the people I care about the households, right? And understanding what that benefit is and how to control other risk is really very important. My question is for the last speaker who discussed about core principles, executive order. How did you decide that you will be talking about liquidity and Dodd-Frank Act in the first report that was published in June? Why did you decide that those were the first priorities? So I guess the question is why was the first report on banks and credit unions? Yes, okay. It's a very good question. You know, that's where we started. You know, I think when you look at the United States and you look at the number of financial institutions, particularly banks and credit unions, you're talking about several thousands. They're in every single community across the country and they're core to the American financial sector and financial sector and way of life, delivering products to citizens in urban environments, but also in many rural environments as well. And so it just seemed like a good place to start. And we followed from there how regulations at entities, banks, credit unions can also impact the markets. And that's why I think it was a natural flow to the capital markets report. And was there any follow-up from your recommendations from wherever you recommended that? So certainly, you know, we don't want the reports to be the end game. You know, right now we have, I forget what the total number is, but you know, a couple hundred recommendations across all those reports. And what we're hoping is we'll see a number of them implemented and that they will match up with our independent agencies' priorities as they move forward with their regulatory agenda. There's certainly some action Congress can take and there's been some news and developments up on Capitol Hill on some policy objectives and changes that they're considering. And so, you know, this is really kind of a comprehensive way forward. You know, regulatory changes and congressional. Thank you. Vis-a-vis mobile wallets, Venmo, Hickier non-bank provider, and I do realize how those monies and where they sit, they are not protected under FDIC insurance, depending, it is somewhat qualified depending upon how they've set up those accounts. Why are we not contemplating as we think about a non-bank charter to actually structure this so that those consumers monies have the similar protections that our bank accounts have? And how do we think about global systemic risk if we don't have that in place and we do have these monies flowing? Sure, so I'll comment just briefly. I think that's a fair question. You know, I think what we're gonna try and do to the best of our ability in this last report is to really be comprehensive in the questions that we ask. And I think that's a fair question and one, you know, certainly wanna engage with different stakeholders on that point to better understand, you know, what are the pros and cons? Yeah, I'll jump in and say the same thing. Excellent question. Equality of treatment is under the regulation is important but also for small investors, the protection has clearly been a regulatory priority for a long time. Again, we don't know what the loss experience is gonna look like until it's too late but it's something that we should be thinking about. So if I heard the question correctly, I'm not sure I did. We understand the network structure in banking fairly well. What do we understand about the network structure of FinTech and what are the implications for risk? I would say at this point, we don't have a very good picture of the network structure for FinTech. I look at the conference offerings for FinTech conferences fairly frequently and the number of providers and sponsors that show up on those conferences is astounding to me. This is a very fervent ecosystem. There's a lot going on. People are trying a lot of different things. Exciting place to be. We don't have good central data collection on all the things that are going on there. It's a very interesting space. So maybe one more question. Hi, what's the Treasury's position on regulating robo-advisors, especially in the context of DOL fiduciary rule and whether or not you think that there's some risk opposed by robo-advisors going forward? Sure, so robo-advisors in that issue in particular, what is the appropriate regulatory touch is something that we're going to be studying in our fourth report. I don't want to prejudge exactly where we'll end up, but we certainly are very familiar with the issues that have been raised. It's about all I can say at this point. So we're right at 330. Please join me in thanking the panel. From here, we're going to take a 15-minute break and then we'll reconvene for the keynote from Federal Reserve Governor Lail Brainerd. In almost all cases, I think also, so at least we're not trying to study the structure of these networks because that's, is that we aren't very advanced in the stuff we do in these experts, but it helps me go, you know, assuming to understand how these things are.