 Hello everyone, and welcome to this month's CB and DC webinar event. I think we have a really exciting program this month, and I'm going to start things off by turning it over to our moderator, Sebastian Krenslen from the Swiss National Bank. Thank you very much, Todd. So I would like to welcome you to today's virtual seminar series on central banking and digital currencies. And we have a very interesting paper today, namely on the question, will central bank digital currencies disintermediate banks? Tony will be presenting the paper. Tony with it is professor of business administration at the University of Michigan, a very impressive resume. Among others, she received her PhD in economics from Princeton University, and she's an expert in multiple areas, among others, finance, econometrics, macroeconomics, and now, of course, on central bank digital currencies. She's also co-editor of the Journal of Finance, Financial Economics. And then we will also have Itamar Drescher, who is professor of finance at Wharton, also a very proven expert on monetary policy, transmission, and also on the interest rate setting, and he will be discussing the paper for today. So we will start with Tony. 25 minutes are allocated. Then we will have Itamar for a discussion around 10 minutes. And then, of course, I invite you to do the Q&A. For that, we have roughly about 20 minutes. We have two formats. We have the panelists who can unmute themselves. Please raise your hand. And then I will ask you to ask the question. Otherwise, all the others, please use the Q&A area to ask the questions. And what I also would like to mention is that the conference is being recorded. So the video will then also be posted on the website. And with that, I would like to invite you, Tony, to start presenting the paper. Very much looking forward to it. OK, thank you so much. I'm really looking forward to the feedback. This should be fun. Let me share my screen. And let me move something out of the way. OK, now I can go. So this paper, you already know the title. It's co-authored with Yu Feng, who is here as a panelist. It's a co-authors from Columbia University. There we go. So what's a CBDC? It's a country's official currency and digital form. So when I've talked to folks about this paper, they say, well, don't we already have digital money? But it's different from digital money of such as bank deposits because a CBDC is a direct liability of the central bank rather than a commercial bank. It's also different from other central bank liabilities like bank reserves because the CBDC can be held by the general public, not just banks. And CBDC has been increasing in popularity. This is an interesting map. It shows that over the world, 11 countries have launched CBDC, 15 are running a pilot, 25 are developing a CBDC, 46 are doing research. So this is something that many countries around the world are thinking about actively or have already implemented. But it's not without controversy. So we found some interesting quotes from some of the Federal Reserve publications. This is a good one. So a widely available CBDC could reduce the aggregate amount of deposits in the banking system, which could in turn increase bank funding expenses and reduce credit or raise credit costs for households and businesses. So that doesn't sound good. And as you will see in the rest of this presentation, that's kind of true. But the answer is not whether it's yes or no, it's how much because this is a quantitative question. And that motivates, oops, there we go, that motivates our central reverse question, which is not just how. So not just a qualitative question, which a lot of folks writing theory papers have already looked at, but how much. So a quantitative question, does CBDC influence banks deposits, their credit provision and their stability? So that's what we're going to do. So we're going to do a dynamic, hang on one sec, one sec, just a sec. What did I do wrong? Ah, okay. Sorry about that. Here we go. I had the wrong set of slides up. That's embarrassing. Let's keep going. So we're going to answer this question in three steps. So first we're going to establish an irrelevance result. We're going to say that in a frictionless world, deposits and loans are entirely separable. So we're going to say that in a frictionless world, the CBDC should have no impact on bank lending. That's a starting point. And then we're going to build a quantitative banking model. So it's going to contain frictions that allow a CBDC to affect bank lending. And we're going to fit the model to US data. So we're going to get some empirical estimates. And so then we can actually quantify things. Then we're going to counter factually, we're going to go back to that. And then we're going to go back to that. And then we're going to go back to that. And then we're going to go back to that. I'll give you more details later, but for now you can just think of this as. Analyst analogous to an IO analysis of how an entrant might change market structure. Okay. So briefly, what do we find? So what are first, what are the effects on money creation? First, that quote that I showed you is not entirely wrong. It's not about money creation, i.e. deposits. And the impacts bigger of CBDC pays interest. And in both cases, banks become less profitable. But here's what's really interesting. Is that the effect on credit supply is three times smaller than the effect on deposits. Why? Banks replace lost deposits with wholesale funding, usually in the US for money market mutual funds. This is not that expensive. And so they replace a lot of the lost deposits with wholesale funding. We'll show that the effect differs across large and small banks with small banks getting hit much harder. There's an interesting implication for bank stability. And this goes back to some of my discussants work. The Schnabel, Salaf and Drescher point out in the recent years, there was a paper that because of deposit market power, deposit rates are sticky. And of course, loan rates are sticky because loans are long-term. And so, in some sense, the textbook maturity myths match that we teach undergraduates is kind of not there because both loan rates and deposit rates are sticky. And so this limits the natural head of interest rate risks. So what happens with CBDC? So all of a sudden banks replace deposits with this interest sensitive wholesale funding. And that limits this natural hedge. We'll also show it slows recapitalization following negative shocks. So there's some negatives too. I'm not going to go into the whole model. It's really complicated. But I'm going to try in this short presentation to just give you an intuitive sense of what's going on. So here's the skeleton. It's dynamic, discrete time, infinite horizon. And there are three players. There's depositors, borrowers and banks. Depositors and borrowers are really simple. They solve these little static problems. Depositors have a dollar and they choose where to invest their wealth. Borrowers are also really simple. They want to borrow a dollar and they choose whether and how much to borrow. So this is it. They just solve these simple little static problems. They are not complicated players. The complicated player is the one that we care about, which are the banks. And they make dynamic optimization decisions. How does CBDC fit into this? You haven't said anything about CBDC. Well, we're going to estimate the model using US data. So obviously CBDC is going to be absent when we estimate the model because there's no CBDC in the US. But then we introduce it counterfactually when we examine model predictions. Okay. So now let's move on to the depositors and borrowers. Remember they solve simple static random utility problems. And if you do some algebra, you get regressions that are derived directly from the model. And so we estimate these deposit demand, this deposit demand and these loan demands using Barry Levinson and Paiskis. So remember these come from these little static optimization problems. And what we get is we get logistic deposit regressions. And what do they contain? They contain deposit rates. They contain non-rate characteristics like the number of branches, the number of ploys per branch. They contain issuer and time fixed effects. So we'll get things like cash fixed effects, transaction deposit fixed effects, things like that. And then a transactions convenience dummy, which is one for transactions deposits, zero for savings deposits, one for cash, things like that. So we run these regressions and we harvest two things. We harvest demand elasticity. I'll show how those get used later. But then we harvest the fitted value of these other things like the non-rate characteristics, the issuer and time fixed effects, the transactions dummies. And we're going to call that quality. So think of quality as just the fitted value of the other stuff in these demand regressions. Logistic loan regressions are similar. And we once again harvest demand elasticity. So those play less of a role in our counterfactuals, a little bit of a role, but a less of a role. Okay. So now we've done depositors and we've done borrowers. Let's do banks. So banks have a very standard ballot sheet. They have old loans. They have new loans. Those are a function of the interest rate that they charge. So the banks are monopolistically, monopolistically competitive and they choose rates. They have reserves. And they can also invest in government securities. They can have excess reserves. So what are total assets? They are old loans, new loans, reserves, government securities. What are the liabilities? They're deposits, non-reservable borrowing. So this would be wholesale funding. And then we have equity. And so assets equal liabilities plus equity. And reserves and government securities are in the federal funds rate. So pretty standard bank balance sheet. Deposits, let's move your eyes down here. Deposits are insured, but non-reservables are not, as in real life. And so default is possible under insolvency. So when bank value goes to zero, the bank gets auctioned off. The lenders, the wholesale funding lenders, don't particularly like this. And so they charge a spread over the federal funds rates such that they just break even because we assume that these lenders who are outside the model are perfectly competitive. Before I talk about the actual problem that the bank solved, I want to backstep a little bit and talk about a very simple problem that I think will solidify the intuition. So let's pretend there are no financial frictions, no regulatory constraints, no maturity mismatch, no time dimension, what would banks maximize? They would maximize the return on loans times loans. Loans are a function of the return because there's imperfect competition. They would minus the return on deposits times deposits, which are a function of the deposit rate. And then minus the federal funds rate times. This is not a function of its times. Loans minus deposits. So this is the, how much they pay if they do not get enough deposits to make all the loans they want to make. Okay. So you could take the first order conditions and screw around with them and you get this. So the lending rate is the lending rate is the federal funds rate plus something that looks an awful lot like a demand elasticity markup. There it is. And the return on deposits is the federal funds rate minus a markup because these spreads are spreads below. What's important here is that if you look at this equation, there's no D there. And if you look at that equation, there's no L there. So deposit taking and loan origination are separable. And in this case, if CBDC were to crowd out deposits, it wouldn't do anything. It would mess up this first order condition, but it wouldn't do anything to lending. So what's the intuition? So if banks can frictionlessly access wholesale funding, then loans are just priced at the market interest rate. If banks can have access to cheap deposits, then that certainly makes them more profitable, but it doesn't do anything to the lending side of the market. And there's an argument given that the loan to deposit ratio is one to one, every dollar that migrates from deposits to CBDC is one less dollar of lending. That's from also from a federal reserve publication. That's conceptually wrong. And so this irrelevance result then lets us isolate the various frictions that do allow a CBDC to affect the banking system. What are they? External financing frictions. Wholesale funding is expensive. You can't get it at the federal banks, can't get it at the federal funds rate. There's regulation, there are reserve requirements. And more importantly, there are bank capital requirements. There's also interest rate risk for maturity transformation. Banks as I said before, because they have market power, deposits are effectively long duration and CBDC makes that shorter duration, their liabilities because they have to use this interest-sensitive wholesale funding. Let's go back to the problem. So what are banks maximizing? They choose loan and deposit rates. And here's where the Barry Levinson-Pakess regressions come in. These estimated demand elasticities give us the quantities. Yay. They pick reserves and government securities. They pick non-reservable borrowing. So that's a Bellman equation that you have to solve. Taking the behavior of other banks as given. That's another fixed point to solve. Subject to constraints. What are dividends? They're profits, more profits and plus the change of inequity. They're subject to capital regulation. They're subject to reserve regulation. And they can't issue equity. That turns out truly, really not to be a big deal. We can put equity issuance in there. The quantitative predictions are almost the same. The same, this just makes it easier to solve the model. And banks default and is auctioned off when the value is zero. We estimate the model parameters. I'm not going to go into this because I really want to focus in a short presentation on the predictions. As I said before, first we estimate the loan demand systems with Barry Levinson-Pakess. We get rate sensitivities. We get the value attached to the non-character, non-rate characteristics, and we use cost shifters as instruments. The second, then we plug these elasticities into the dynamic model. And then the second stage, we estimate the remaining parameters using, including these financing costs. And we use SMM, pretty standard by now. So now here's the fun part of the paper. What's a CBDC? So we model a CBDC as a bundle of characteristics. It offers some rate of return. We're going to start with zero, and then we're going to mess around with that. It allows households to make transactions. So it has that characteristic of transactions deposits. We're going to assign it a issuer fixed effects that is the same as the cash fixed effect, because it's just another way of having cash. And then Kyron has a really nice student coot, and she looked at what happened when a bank introduced a digital app, and the deposits went up by 20%. So we're adding a 20% digital premium that the number comes from this paper. And then these attributes define a CBDC, which we then introduce into the demand system. And then we can trace out the effects of CBDC because we've estimated the parameters of this dynamic banking model. There's a lot of uncertainty in the quality of CBDC. I mean, how's it going to be implemented? How are depositors going to react? We don't know. So we vary quality. Remember quality is just the fitted value of those characteristics. And we check to see what happens with different qualities. And here's the first counterfactual. I know it's a table full of a lot of numbers, but I will tell you what to focus on. Column one, column five, and column six. Column one is no CBDC. Column five is the quality of CBDC is 100%. And six is the sensitivity of whatever this line is to the introduction of CBDC. So if there's no CBDC, of course there's no CBDC share. Transactions deposits are about half savings deposits are about half of that. And then we have loans, cash, deposit spreads, et cetera. Move over to your eyes over to the last column. And what about this second number? So it says that if you introduce a dollar of CBDC, transactions deposits go down a lot. So the sensitivity is negative 0.64. So that's a lot. And savings deposits go down a lot too. What's very interesting. So this is a lot. These are big numbers. Loans go down, but not by nearly as much. So what we see happening here is we estimate a cost of accessing wholesale borrowing, but it's not large. We estimate that it's an empirical result. And so banks are substituting lost deposits with wholesale funding, and they don't cut their loans one to one. That's the takeaway from this slide. Next slide. Let's look at what happens when CBDC pays interest. Same table format. Once again, look at the last slide, or the last column, excuse me. So transactions deposits go down by a lot. Savings deposits go down by even more. Loans are hit harder before it was 0.1, not negative 0.19. Now it's negative 0.27. But as before, the effect is much smaller than when, than the effect on deposits. So there you go. So just to pound it into your head. The effect on lending is three times smaller than the effect on deposits. There's a margin of substitution that banks use. We estimate that the frictions related to borrowing be a wholesale funding are small. So that's the rate on wholesale funding minus the federal funds rate. Why? It's actually much smaller than something you'd get for industrial companies because deposits are insured. And so banks do have a low default probability, despite their very high leverage. I think they have huge leverage. Their default premium must be huge, but it's tiny. And one thing that's interesting that's not in the previous tables is that the effect on aggregate credit is even smaller. This is where estimating the loan demand comes in interestingly because what we find from our empirical elasticity estimates, that firms would substitute into bonds. So even though they can't get loans, they'll substitute into bonds. And so the aggregate effect on, the effect on aggregate credit is truly tiny. Big banks versus small banks. Take a look at the first line. This is the estimate of the external financing cost for small banks, 0.04, and for big banks, 0.01. What's important here is obviously it's much more expensive for small banks to access external funding than big banks. So what we get is the CBDC sensitivity for small banks, for deposits, and large banks is about the same. But the loan sensitivity for small banks is quite large. The loan sensitivity for big banks is quite small. And so what we see is that CBDC would hurt large banks much less than small banks. And so it also has implications for the industrial organization of the banking sector. Just a sec. There we go. Consumer and producer surplus. This is fun. We're estimating demand systems. So we can calculate consumer and producer surplus because we can calculate triangles. That's all these are, are triangles from the demand systems. CBDC benefits depositors. It gives them another product. CBDC hurts borrowers because it reduces bank credit supply. So here's depositor surplus. Here's borrower surplus. It goes down. It also decreases the banking sector surplus by a lot. But the net surplus overall is positive. This is obviously not a comprehensive welfare analysis, but it's a good starting point. If you want to think about major stakeholders. And then the last thing I want to emphasize. Is without CBDC. What we do is we look at a 1 percentage point monetary policy shock. That is, we shock the federal funds rate up by 1 percentage point. And we find that that leads to a 2% negative 2% bank stock return. And once again, this is small because banks effectively don't have a maturity mismatch. When there is market power. This goes up by 3 fold when CBDC is introduced because when CBDC is introduced, bank substitute into wholesale funding, which is interest sensitive. And then they really do have a maturity mismatch. So monetary policy shocks hurt banks more. And I'm done. So we want to look at a framework to quantify the impact of CBDC on banks, depository services, credit provisions, cyclical behaviors. We find that it hurts deposits. It hurts loans much less. And that there are several channels. There's it lowers the deposit base. It reduces market power. It lowers bank profits. And it increases banks funding costs. And I'm done. Well, thank you very much, Tony. That was like Swiss punctuality to be honest. I know I always do that. Great. So then we have it tomorrow. Who will do now the discussion for 10 minutes? And then we can have the Q&A. Okay. You can see my screen. Yeah. Yeah, we can see it. Okay. Thank you very much. So. So, yeah, I really appreciate the opportunity to discuss the paper. I will say usually when I discuss papers, I really like to get into the nitty gritty of everything that's going on, but given that 10 minutes and given that as Tony said, there's a lot going on here. I sort of gave up on that and instead thought that. I'll internalize what I learned from it and just think about CBDC in that context and sort of give you some, some of my analysis or thoughts on it, but there's really a lot. It's a, it's a very, you know, it's an apparatus that has all the parts that you need. I could definitely quarrel with a lot of the parts, but that's, you know, and it would be interesting to do that or at least debate them. So I'll mention them as I go through because I think it's quite an important and complicated problem. So, and one that will deal with one way or the other, because it seems like this is happening. So what is CBDC? You're replacing paper currency with digital currency. I think paper currency is pretty much an anachronism at this point. So this is what we should do. What are the, the question is, how do you do it? But what are the advantages? I think it'd be more convenient for payments. Paper money is not a great technological solution at this point. It's more secure. It can't be counterfeited. It's hard to rob. If it's done in a decentralized way, it can be traced. So that is a disadvantage for underground activities, crime, terrorism, which is a good thing to not have. And it, in principle, allows for interest on central bank currency. And the question in this paper is what impact would it have on banks? So I think it, like I said, it's an interesting and important question. So first design choice I was thinking of here is whether this, the paper doesn't really talk about this aspect of it. Maybe it's not that important. Maybe it is. Do you want to make it centralized or decentralized? I'll tell you what I mean by that. Which is kind of what the paper has in mind is every entity has a digital currency account at the central bank. So basically a checking account at the Fed. And if you have that, you can pay interest on CBDC decentralized, which I think is the kind of the minimal deviation from paper currency would be that the central bank verifies what is a digital currency, but it's not keeping track of the ownership. So that is still need custodians for like, like, like you have a paper currency, treasury bonds or stocks. Then, you know, there's not like companies, you know, it's not like there's a, the government keeps track of those or of where the treasury bonds are. You need a custodian for a need a broker or somebody who interacts with the system. So that could be provided by companies like a checking or brokerage account. And so that's kind of like very similar to just replacing paper currency. In this case, it's hard to think of how you would directly pay interest on CBBC since you're not keeping track of who has it. So what would minimally disruptive CBDC be? And I'm not saying that we want to make it minimally disruptive. I'm just thinking through it. So you would replace paper currency with the equivalent digital currency. Now how much of this is there current value of currency is $2.2 trillion. In this case, banks would continue to provide custodial or checking accounts. So that wouldn't even dis-intermediate their checking accounts. There'd be no interest on it, just like paper currency. So it would still be dominated by deposits or bonds as long as interest rates are above zero. And it wouldn't really be a substitute for interest paying accounts. Okay. So just a substitute for currency. That's kind of the minimal change in demand for currency and would have the minimal impact in central bank balance sheet. Would be more convenient for transactions, more difficult to counterfeit, potentially more difficult to hide or steal, maybe. How much currency is there? You can see here, most currency is $100 bills. Naturally, consider the biggest bill. Most of it's held abroad. So there's really actually not that much of it in the US. It's mostly people either abroad wanting dollars or people hiding money or drug dealers. So more disruptive CBDC would be centralized now, but not interest paying. So the central bank would provide transaction accounts. This would, I think, dominate bank zero interest checking accounts. Most bank checking accounts pay very little interest, but they could increase it a little bit. Now, how much is that right now? Checking deposits shares around 15%. You see it actually increased a huge amount during COVID, around $3.5 trillion. So I'm going to use that to talk about them a little bit. But that's about how much of banks deposits it's been. I actually didn't understand the number in the paper, which in the model has a much larger share of deposits in checking or demand deposits. It didn't look that big to me. What would the central bank buy in exchange for providing these checking accounts? I think the least disruptive thing would be to buy whatever banks are financing with checking accounts. Otherwise, there's going to be a shift within the financial system. So in our paper and the banking on deposits paper, we show that in general what banks do is very carefully match the interest rate betas of their deposits and assets. These are the sensitivities of their cash flows to changes in the Fed fund rates. So this is from our figure. This is a cross banks, the interest expense beta and the interest income beta of different banks. So banks that have more interest sensitive, meaning less market powered deposits by shorter term, more interest rate sensitive, more cash flow sensitive assets and vice versa. So checking accounts have zero beta. Obviously they don't pay any interest or very little. So banks mostly use them to fund long-term fixed rate things like MBS. So you think of them buying them, the central bank would have to buy long-term treasuries or MBS. In a paper with Dominic Superos, my student who went to Columbia, we're looking at actually the natural experiment of the huge inflow of COVID era checks into checking accounts, which was that three and a half trillion dollars. We show that banks bought a lot of MBS with it, which it can explain why MBS spreads and MBS spreads and MBS rates were so low last year. You can also explain why they went up a lot now because they know that they're going to lose all the deposits and just point blank, stop buying them. What are the pros and cons of having this? So I think the pros are that there would be a uniform payment system. If you had this, it would make digital payments easier. They're surprisingly hard in the U.S. And this is, I think in part because they're monopolized by Visa and MasterCard, which are just actually Visa is the biggest financial services firm in the world. Or in the U.S., it's bigger than JP Morgan. So it's worth $400 and something billion. That's because there's $138 billion a year of interchange fees paid, of which 93.9 goes to banks and the rest goes to Visa and MasterCard. And in a paper with two of my students, we look at why credit card banks are so profitable and they are very profitable, much more profitable than banks in general. Their ROAs and ROEs are about three to four times higher. The question is though, is this how we're supposed to get rid of, if we believe they have monopoly rents, are we using CBDC as an antitrust tool? Shouldn't that be a tool for regulators? I don't really like the idea of introducing it in some way just as an antitrust tool. It sounds like we're not doing our job. And indeed China has widespread easy digital payments, but not because of CBDC. And then on the cons side, I think it's not totally appreciated how expensive it is to run a retail bank, even just for checking accounts like Goldman has run into serious problems building Marcus, which it wanted. And it in fact seems like it's gonna not give up on that. And it's estimated that the cost is about $250 to $400 a year per checking account for all the administrative crap that they have to do. The other thing is the Fed would become larger and larger owner of MBS. The most disruptive CBDC would be if they provide savings accounts that pay competitive interest rate. That is with the rate beta of one. And I wanna point out that market power of banks is not the same thing as rents. Banks have a lot of market power, but it's not clear that they make much rents at all because they get this market power by incurring fixed costs, not variable costs of two to 3% of assets a year. And there are a way then once you subtract that it is not obviously very high. It's about 1% and that includes credit risk premium. The question is do we want to kind of push them out of this or can we? And that depends on whether they value whatever these costs pay for, the quote unquote services that people get via deposits. So the key question I think is does CBDC eliminate the need for these services that these guys are investing in? Or is it just like capturing them via advertising? Is that what they're doing? Maybe then we want to push them out. If not, if it doesn't eliminate these services, does it provide some advantage, some efficiency gain in providing them? If the answer to that is also no, then you wonder why whether this is at all a good idea. We already have money market funds that pay a competitive Fed funds rate and they have not supplanted bank deposits at all. Okay, so I'm almost done here. So what's the biggest effect I think would be on long-term credit? So we showed that banks need to do this rate matching. Their main technology, their sort of reason for existence is that they can get these low sensitivity retail deposits. And then they must match them up with long-term assets. It's not just a matter of like they want to, they have to. Otherwise they would have very large maturity, duration mismatch. And the fact that they're matching them so carefully suggests that they really don't want to take that mismatch. If they were to lose this low beta funding by say having to pay competitive rates, their long-term credit supply would decrease substantially. It would be like a very big version of what we've just seen in the mortgage market when they know they're going to lose deposits because rates have gone up so much and they just gave up on, they stopped buying mortgages and the spreads went up so much. So does that mean we shouldn't have high interest-paying CVDC? I don't know, it depends on the answer to the previous question. Does it provide efficiency gains in providing deposit services? My last point is that I don't think banks can replace deposits with wholesale funding. I really think they could do a little bit, but banks are overwhelmingly built on deposits. It's really the origins of their business. That's what allows them to do long-term lending and hold illiquid assets. When they tried to do wholesale funding, that was the ABCP conduit pre-2008. It ended in a run on those things precisely because they're not insured and therefore highly runnable, just like in the pre-FDIC days when we had bank runs every 15 years. Okay, so conclusion, I think this paper is on a really interesting and important topic. It gets all the right channels in there, all sort of quibble or even quarrel with the importance of various ones. I think it's dangerous to think that this is an anti-monopoly solution because that's really a regulatory problem. I do think we have that. I think Visa and MasterCard are monsters. They are in huge rents. We need to make payments easier. Another thing is if CVs provide the liabilities that take over, they'll have to make the capital allocation decisions. We need to think carefully whether we want that or not. Thanks. Thank you very much, Itamar. For that, maybe just two points that I would also like to emphasize that you also mentioned. One is on the efficiency and effectiveness of payments. I think that's really one of the important aspects that we have also been focusing on in Switzerland. How expensive are payments and how can we also increase competition or ensure that there are alternatives to, let's say, credit cards? This is also why we are now enhancing and we're not the only ones among the central bank community. We're enhancing the payment system with instant payment functionalities, allowing that you could then also do point of sale transactions with your checking account, for instance. That's one of the elements. The other one, and that will be directly a question to you, Tony, is what also Itamar mentioned, is banks will have to substitute deposit funding with wholesale funding, meaning that it will become shorter term, meaning that if you look at the net stable funding ratio or the liquidity coverage ratio, so the liquidity regulation will effectively lead to it that they will have to have longer-term funding. The question is how will this dynamic then evolve over time? Because I think wholesale deposit will not be or funding will not be a perfect substitute to deposits. You, Fung, do you want to do it or do you want me to answer? Okay, I can try. So first, I would like to thank Itamar for the excellent discussion. I think this is really useful and really almost helped and we really appreciate it. So a few things. First, as you mentioned, I think we are not arguing that the CBDC are introduced for the purpose to lower market power. I mean, I think this is not the point and I don't think if, I mean, first of all, we don't think market power is necessarily a bad thing, as you mentioned. Second, even if the central bank wants to lower market power, we are not convinced that CBDC is the tool that the central bank should rely on to do this. I think there are many other purpose of introducing CBDC, like crimes, as you mentioned, cross-border payment is a very important function that a lot of, why a lot of countries have it right now and also financial inclusion, especially if the CBDC wants to pay interest rate. So where our goal is more likely, we want to see what's going to happen if CBDC is introduced, how this is going to influence the bank's deposit funding and the other services. This may not be the goal in the first place, but if CBDC is introduced, this is likely to be the consequence. It's likely to influence the bank in this very specific way, as our model predicts. And I think we need to understand that this consequence first before we come back to this very important question, that is, whether we should have a CBDC or not. Second, you mentioned that there are some features. For example, whether the bank should track the identity, whether the CBDC should come with a card like master or visa and things like that. I think these are also excellent questions. Unfortunately with the US data, it's kind of hard for us to quantify these features, especially given that our paper has a quantitative focus. I think, for example, in Canada, they have better quality survey data, and maybe using that data we can see more about these more specific features, more specific implementation. Your last comment on banks, can banks replace deposit with wholesale funding? I think the answer really depends on how much they're doing it. I mean, we are 100% on the same page that if the banks are going to rely entirely on wholesale funding, this is going to be a totally different system. But I think our argument is more regarding whether banks can do this on the margin, given that, for example, the non-interest paying CBDC will capture 7% of the market, given that this is how much deposits that banks are likely to lose can they do something at a margin? And I think our answer is yes. So I think these are two questions. On average, I don't think this is going to happen, and I don't think this is possible, but at a margin, there is a lot more that banks can do. And regarding Sebastian's question, regarding the liquidity regulation, we totally agree this is something that we don't have in the model right now. And it's certainly related to the large discussion on how liquidity regulation is going to influence banks, not just when CBDC is introduced or not. And this is something that we should, we will definitely work more on. Okay. That's my answer. Thank you very much. See if Tony has anything to add. No, that was perfect. That was awesome. Thank you. Can I make one comment just overall? I'm not, I don't think I came across, but also don't mean to come across as claiming that just because banks may lose deposits and their business is going to dramatically change, maybe they're going to lose their importance, that that by itself suggests we shouldn't do it. I think the question is, what advantage technologically would this provide? What you wouldn't want to do, for example, and I don't, nobody's suggesting it, I think, but is to have a central bank that offers high paying, high interest paying deposits that cost a ton to run. It takes a lot of interest rate risk that sits on the balance sheet of the government, but then you don't acknowledge that. And so there's all these costs you're just like blind to, you have to manage a ton of assets that you're not good at managing, and then you don't see it. And so then it, then that seems like a good idea, right? You would want to do it if, if it, if, you know, just like I think deposit insurance is a good idea. If it has like a technical, once deposit insurance came around and banks didn't have to depend on what is effectively called wholesale funding today, you stop having these disastrous bank runs. And that was clearly like a humongous improvement in the ability of the financial sector to manage. So I think, I think that's like some 30,000 feet view. That's, that's the, that's where it boils down to. I'm not, I don't want to be seen as like a water carrier for banks. They need to go, they need to go. I don't care. That, I didn't get that, that, that makes a lot of sense. One thing I think we need to do more in our paper that you emphasized is they just at least talk about some of the advantages of CBDC in facilitating payments, because payments are kind of ridiculously hard in the U.S. relative to many other countries, including like developing countries where the phone companies are essentially doing payments. I agree and I don't, and maybe Sebastian can help us to understand this. I still, I've talked to a lot of people and I still don't really understand why payments are so hard or why Visa and MasterCards position is, is so entrenched. Given that we have other paint, other credit card systems, and given that direct access from the phone to your account should be very easy. I know there's issues of security and you don't want it to make it easy for people to transfer money out. You could see what happens with the crypto, but nevertheless, Visa is a $400 billion company, MasterCards like a $300 billion company, and in China, people transfer money from phone to phone without them at all. I don't think it's a technologically difficult problem, so I don't really get that. So it's less about the technology, but as I understand also other central banks, it's more a question of sovereignty. So how can you ensure that you, or differently said, payments are the backbone of the economy. So you want to ensure that payments can be done at all times, that you can choose between different payment instruments, be it banknotes or digital payments, and that the costs of payments are not prohibitively high. This is currently the case, but you have a certain dependency on payment providers and central banks across the region have identified that you have a dependency which could lead to higher costs and if you have no banknotes anymore, then of course you're very dependent on be it foreign payment providers, and that's I think one of the ideas. Why can't with the phone we can check what's in your bank account and debit your credit and like a checking account or any kind of account, but you could check, you know, it used to be in the past that Visa solved the problem before internet that you went to a store and they don't know whether you have any actual money or not. And so Visa kind of stood there and provided this technology. At this point, any phone can check that immediately. Why? I don't understand. Why do we still need them? Why are they still so important? Can I just add to this? If we think about the context of Chinese payments, for example Alipay, you know, any kind of payment transactions within Alipay are updates on their internal database. That's the same way in which Venmo works. So in principle, if vendors accepted Venmo, that would be a competitor to how payments are transacted. So I think it's more of kind of frictions in terms of adoption and how it spreads. And we are seeing movements towards that. Like in the bank side, I mean, Zell is an example of that where now you will make payments between banks that doesn't go through necessarily the retail transaction rails that we would typically expect when we're doing it, doing a retail payment. So I think there are innovations. It's a market structure issue, I think more than a technological issue in terms of why we still have them monopolizing a large segment of the market. The stock market doesn't see a threat to Visa somehow. It's a mystery to me, but they don't see it. So we have Dirk, one of the, who has a question. Yeah. Thanks very much for the talk and the discussion. Just on that discussion that we heard just now, I think the example that is being mentioned these days on that discussion is PIX in Brazil, right? I mean, it's clearly not a technological issue. It was just the central bank intervening, coordinating, and suddenly everybody has an account and transactions are super cheap. So as I'm totally with you, it's not technology, it seems. Thanks. Yeah, I agree on that. I just think that if I go abroad and I want to use my Swiss payment app, then I'm not necessarily, it's not necessarily possible to do the payment. So I think that's also something that when you come to cross-border usage, then that might be one of the use cases where you still need a credit card because they allow for this interoperability. But Todd, you have one question. You raise your hand. Yeah, I wanted to change topics a little bit. Going back to the paper, Tony, I think the results on how a CVDC would affect the composition of bank funding and sort of the allocation of funding within the financial system are super interesting. Where I have trouble is going from there to the aggregate level. Like if you say, okay, thinking of the commercial banking system, it's not such a big deal if they lose deposits because they can replace it with wholesale funding, maybe subject to some caveat. But then I think, well, where does the wholesale funding come from? If that's coming from money market funds, are money market funds expanding? If so, where does the saving come from or are they shifting out of other assets and then who's losing funding from there? How far, you kind of made some statements about in terms of the aggregate credit, it's not going to be important, but I didn't quite follow how I could see that from your results. That's because it was a 20 minute presentation. Okay. No, so we also estimate the, do a similar demand estimation for borrowers. And so we know how they substitute between bonds and bank loans, depending on the price. And so then we can see when we introduce CVDC how that substitution has. That's as far as we can get. So in a precursor paper to this, in an appendix, which you've made, we have a whole DSGE model, where we have the equilibrium conditions. I don't, we should probably do that again, because I don't think you can answer your question. We can answer it qualitatively, but not qualitatively, unless we actually write down a big, huge model. Yeah, I see. You don't want to add anything? We don't want to do a whole DSGE model again, but that's a great question. So we did something in the paper, which is we're thinking the US is more likely that once the central bank gets the money, they are going to invest in the treasurer's market, which is going to influence the treasurer premium. So we use some reduced form equation to sort of capture this effect. And there's a fixed point. We see how this is going to influence the treasurer premium, and that in return, how this is going to influence the household choice probabilities of the different investment options. And we find that first, as you suggest, it is going to influence how much market share that CBDC can capture, because you are really influencing the attractiveness of the different options. But at the same time, we find it is pretty robust that the transmission from the deposit market to the lending market, or this sensitivity, is a quite robust result, saying that less than one third transmit from the deposit to the lending market. So this is a quite robust result. But again, we don't want to interpret this as saying that CBDC is not going to influence bank's loan provision at all. It's not as big as one to one that people typically argue, and we show that it is going to have important cross-sectional implications, which should be emphasized as well, because it changes the market structure, and it influences the funding access by different types of firms. So I think this is an important prediction as well, that we don't want our paper to be read as we're saying that, well, it's not going to disintermediate the banks at all. Yeah, okay. That makes sense. And if I can just go one step further and then think about, because all of this seems important for the welfare implications as well. We don't even want to call that welfare implication. Yeah, I was going to say, you called it surplus, which makes sense. But if there's a way to connect all of those, then you can get into welfare. We can say something. We just want to be super careful that's the reason why we are not calling this a welfare analysis. But I think you're right. Especially with the exercise that we endogenize the Treasury premium, we can say something about what's going to happen to different participants, different major participants. Yep. Okay, thanks. Right, so we have Russell and then Itamar again. Yeah, thank you, Tony. Thank you, Itamar. So I have something maybe related to Itamar point, but go into the opposite direction. So a lot of discussion about this intermediation rely on the bear market power, which Itamar tries to phrase it. It also depends on the competition between the CBDC and the deported. But in the popular proposal of CBDC, we have circulating right now, it is what we call the intermediate version of CBDC. And this means that the central bank, maybe the Fed or other major central bank will maintain account for everyone. So the account will be maintained by bank or other financial institute. So they will do all the custodium and maintenance. So in this case, it could be portable that after this intermediate CBDC system, the bank market power is increased, right? Because in the year, it still rely on the bank network and for this intermediation of the CBDC product to people. So how that will show up in the model. So the implementation are like that thing. In this case, it become a commitmentarity. So because when people in the CBDC, they need to have account at the bank to have assets with this one, right? So in the model, we do have things like administrative costs for maintaining a deposits. Because as Yitamar pointed out, they're not zero. And one thing that you've found I might get this wrong because I'm not doing the demand side part of the paper at all. I think we could introduce a different type of CBDC that was intermediated through banks. That might be an interesting thing to do. And that would have the specific feature that it would be. It would also cost them something. And so I think that if we could attribute those costs that we get from deposits to CBDC, we might be able to say something though that might be a bit of a reach to think that the set of costs would be the same. I'm not sure. Okay. That makes sense. Good. Then we have Yitamar. Two quick points. One regarding loans and bonds. Just want to point out one thing. It's something I didn't use to know. And I appreciate a lot more now, which is, well, the first thing I did know, which is, you know, CNI loans are not only like 15% of banks balance sheet. So they're very, very important, but they aren't most of what they do. On the other hand, there's only a very, very small fraction of companies that can issue bonds that are even rated. There are only about 1,000 companies that are rated. There's about 700,000 companies that have more than 20 employees and are not rated. So it's not that easy for the companies to switch. Most companies simply have no access to the bond market whatsoever. The other thing was, so it is not CNI. The vast majority of their balance sheet is, is like, well, the biggest category is mortgages. So is it important if they can't fund long-term mortgages? I don't know. Maybe people should switch to the floating rate, you know, to arms. I don't know. The question is how is that, how important is that? And why are, you know, what are the fixed costs in the background? You know, all these fixed costs that they're hedging with this, like how important is it to have those? That's what there was calling the services, but it's really rather than focus so much on like the, lending to firms. I think that's the question. Like mortgage rates went up a lot. Do we care about that? Maybe people should switch. Like I said, the arms. So I think that's kind of the big question that would happen if they were to lose the access to the special funding. You Feng, you take this one. Okay. So first, I think it is true that in our sample only a very small fraction is CNI. And in the model now what we do is we are not differentiating this different type of borrowers. We could have two different demand systems, right? We can have the demand from CNI and demand for mortgage and see how that's going to make the banks are going to react to the two types of loans differentially. We are not there yet, but I think it's a good proposal. Another thing is regarding like it's, once there is this disruption, how banks are going to react, maybe they're going to do something to this. This reduced deposit and loan interest rate risk catch. And we think this is possible. So one thing that we're doing a paper is actually we decompose the 19% decreasing loans by different channels. We ask how much this is really driven by this loss of the deposit. We ask how much this is driven by the interest rate risks and different things. And by having that exercise, I think we can sort of speak to our concern that because we're able to put balance on the different channels. So if the banks are able to do something regarding this channel and lower this influence, how likely this is going to influence our overall result of this 19 cents. This is something that we try in the paper precisely because we have the same concern. And I think one element which also turns in is to what extent the central bank would then do the funding for the banks, you know, because if all the banks lose deposits because of the CBDC, then the central bank will have to do, you know, the open market operations and provide the wholesale funding. And I don't know if that's the direction we want to go. No, then that exposes the central bank to risk, which is maybe not the best of ideas. Exactly. Great. So actually we're at the end of the session. I would really like to thank you, Tony, but also Itamar and Eufeng for the lively discussion and everybody who participated. And I would like to hand over now to Top for the outlook. Okay. Thanks very much everyone for participating in a great session. I just want to quickly advertise next month's session, which is four weeks from today, December 16th. Tony Honard from the ECB is going to be presenting and Rod Garrett from the BIS will be discussing. Okay. Thanks again, everybody. Hope to see you next month. Thank you too. Bye bye.