 Good morning everybody and welcome back to day two of the ECB CEPR conference on the macroeconomic implications of CBDC My name is Tony Arnold from the ECB and CEPR and I'm going to chair this third session on Central Bank digital currency and the banking sector We have two fascinating papers in the program just to remind everybody this split is 25 minutes for the presenter 10 minutes for the Discussing and then roughly 10 minutes for for Q&A at the end and I'm going to use Both signs and for the presenters also an alarm to just remind them of the time so the The first paper is presented by Maria Elena Okay Okay, good morning first I want to thank the organizer for putting together such an amazing conference and for inviting me To present this paper on the interplay between central bank digital currency and bank deposits I'm Maria Elena from Uppsala University and the Center for Monetary Policy and Financial Stability in Stockholm And this is a joint work with Humphang also at Uppsala Okay, so I don't think I have to spend too much time on the motivation For CBDC so the majority of central banks has been investigating the issuance of central bank digital currency There are several reasons I Just report to so there is this growing demand for digital payments methods for retail purposes. So precisely for the general public and this is also in a response to the rising demand of Like not, you know the the case for CBDC It's also in response to the rise of stable coins and other crypto assets such as bitcoins And also the gradual decline of the use of cash for transactions in many economies and I think Sweden here It's a very good example So one of the major concern for central banks when thinking of CBDC is the risk of household substituting away Bank deposit for CBDC So this is because this could potentially lead to disintermediation of the banking sector in terms of reduced bank profits and Negative real effects on the general economy. So this would lead to financial instability, which of course, it's a big concern So in this paper, we try to answer two questions So what is the potential risk of financial instability following the introduction of a CBDC? in particular In relation with the equivalence result in the literature, so we Answered this question by building a Real business cycle model with CBDC and bank deposit based on the work And then we are revisiting the equivalence result in the literature. So I'm not spending too much time telling you what's that, but it's simply the Fact that the private and the public Payments are equally efficient In the way that the central bank can pass back fundings to commercial banks upon introducing CBDC so We revisit These in two ways first we add a financial friction for central bank lending to banks namely the collateral requirement So this is done in the context of the discount window lending facility That it's extended from central banks to commercial banks behind some Collaterals and also we consider different degree of substitutability between CBDC and deposits So yeah imperfect substitutability So the next question is okay, but if there are real effects in the economy That's the substitutability between the two asset Impact the risk of financial instability. So we will answer this with a dynamic analysis Of the shift in the household's preferences so likewise, I'm not spending too much time on the literature because It's mainly done by people in this room But basically we're going to contribute in three ways so The literature on introduction of CBDC on commercial banks, which is even more it's even bigger than the papers I mentioned here Then yeah the equivalence of payment system and then this relationship between CBDC and bank deposit with which maybe it's a bit less Studied for now Okay, so just to give you a flavor of the results So we're going to study the equivalency in two scenarios So the first scenario is the CBDC and bank deposit as perfect substitute With this collateral constraint for central bank lending So we find that the central bank can replace the lost funding for banks Under more restrictive conditions than what those found in the literature Precisely because of this collateral constraint So there are no effect on financial stability However, when the two assets are in perfect substitute The central bank can post a loan rate, but this rate won't make the bank indifferent to the competition from CBDC because they their Profit change won't be zero. So eventually there will be some real effect in the economy So the equivalence Analysis is done Through a comparison between two city states one without CBDC of with a very little amount of CBDC and one with CBDC So then to answer to inform more what has the real effect in the economy We move to a dynamic setting and we study the Effect of shifts in the household preferences So what we eventually find is that the CBDC demand increase, but there is limiting limited crowding out of deposits But bank profit drops due to lower bank market power. So it's a result that Confirms most of the literature on this topic so eventually the substitability between CBDC and deposit it's key for Central banks when think of introducing CBDC, okay, so The agenda for today I'm going to go pretty quickly through the model Then the main part so the revisitation of the equivalence result then the dynamic analysis and then I'm going to conclude so As I said, we build the model on year's work. We have A non-competitive bank that investing capital reserves government bonds and from through deposits or borrowing from the central bank Subject to a collateral requirement in the context of the discount window lending So I Chose to focus on the collateral requirement because this is the addiction we make So the collateral requirement is formulated following the standard literature and It implies that the L so the central bank loan extended to banks must be lower than a fraction of the bonds hold by the banks So the fraction the collateral parameter is theta B Which of course is bounded between zero and one B is the amount of government bonds held by the bank Which is remunerated at a rate lower than the risk-free rate because of the convenience yield So commercial banks have Some benefit from holding government bonds because they can precisely use them as collateral to get funding from the central bank And our L is the central bank loan rate, which is the object of interest in the equivalence result Okay, so the rest of the economy It's composed by household that value good leisure and the liquidity of services provided by CBDC and deposits Neoclassical firms that produce using labor and physical capital and Consolidated government that collect taxes based deposits subsidies investing capital lends to banks against collateral and issues CBDC and reserves Okay, so In the context of the equivalence result We consider an initial equilibrium with a policy with deposit and reserves So there exists another policy in equilibrium with less deposit and reserves as CBDC central bank loans government bonds Different ownership structure of capital additional taxes on the household, but otherwise the same equilibrium allocation and prices so as I mentioned we start by Revisiting this result considering the case of perfect substitute ability between CBDC and deposit with the collateral requirement for central bank lending so The households real balances Z are a function of both CBDC M and deposit M Where here the weight lambda represents the liquidity benefit of holding CBDC relative to deposits So We derive this central bank loan rate RL that Passbacks loss funding for deposit to the banks so this loan rate Depends on the rate of return on the deposit RN The risk-free rate the rate of return on reserves are are the rate of return on capital and that of bonds And also on the operating cost of banks can I use the Well, whatever, so it's the new which is a function of the reserve to deposit ratio of The banks so set up and of the other banks set up bar So this is done to capture potential externalities from other banks affecting the representative bank Then here theta t is the subsidy rate of the central bank to banks While theta B is the collateral parameter So to zoom in on this result We call it the central bank equivalent loan rate in the context of the equivalent analysis so this rate It differs from the Loan rate without the collateral requirement precisely for this term at the denominator So this term It's positive Because from the household problem the rate of return on capital is not risky so it can be approximated with the risk-free rate and from the convenience here we know that the Rate of return on bonds It's lower than the risk-free rate and of course the collateral parameter responded between 0 and 1 So it follows that the central bank loan rate with a collateral friction So the collateral requirement is lower than without so the intuition is that when there is No collateral requirement for central bank lending The bank can borrow as much as you want from the central bank However, when the there is a collateral requirement, it must respect the central bank need to post lower loan rate to incentivize the bank to borrow as much as before so this is done because In this way, there is no change in banks profit and in the banks balance sheet So there are no real effect of introduced introducing CBDC So in the paper there is a graph reporting the relationship between the loan rate and the Collateral requirement It's I'm not reporting it here But just to mention that the loan rate that the central bank offer is lower with tighter collateral constraint So it depends precisely on the level of the collateral requirements set by the central bank So then we move to the case with imperface of stability between CBDC and deposit with the collateral requirements. So in this case household real balances are a CES aggregator of CBDC and deposit and here the new parameter Epsilon, which is positive Is the inverse of the elasticity of substitution between the two assets So in this case We can derive a central bank loan rate But this does not imply that banks profit aren't change So the the central bank loan rate posted does not imply that So the intuition is that when there is a change in banks profitability the new policy does not guarantee the same equilibrium allocation as before so Banks are not indifferent to the competition from CBDC and there will be some real effect in the economy So as I mentioned at the beginning of this presentation The equivalence result it's a comparison between two steady states So to inform how this How does an increase in CBDC demand affect the real economy and financial stability ultimately? We move on to dynamic analysis so CBDC and deposits are in perfect substitute. This is an assumption that has been done in the in the literature There might be different reasons. I remember yesterday someone asked, okay, why this can be the case It can be for example, I think Barder and Kumov mentioned this in their paper they use the They make the comparison with households switching their banks So households tend to stand to stay what we've they know better. So this can also be a Motivation So then We we solve the model we fund the equilibrium and then we shock the economy To study the responses to change in the household preferences for CBDC over deposits so we Do we study two shocks? So one is the positive shock to the liquidity benefit of CBDC, which is the parameter lambda t And the other it's a negative shock to the substitubility between CBDC and deposit Which is one over the epsilon parameter in the CES aggregator So I Just in the paper I have more Impulse responses here. I'm reporting the most interesting one. So The responses are reported as percentage point deviations from percentage point deviations and Here we we do the analysis assuming that the rate of return on C on CBDC. It's lower than that of deposit And we keep that rate fix in this in the Dynamic analysis so The the two impulse responses so the blue sorry the black one Referred to the shock to the liquidity weight of CBDC and The blue one to the substitubility so The two shocks means The in the terms of lambda it means that CBDC is becoming a more attractive means of payment Because it it's a positive shock to the liquidity benefit attached to it while the blue responses means that it's The substitubility between the two assets is diminishing So it's a negative shock to the substitubility So you see that the two Shock are quite similar in the responses For example, if we focus on the lambda shock so the black responses You see that there is Drop in the spread of CBDC deposit and reserve So the spread here is the surplus on the risk-free rate. So Since the rate of CBDC and reserve it's kept fixed this Drop is due to the drop in the risk-free rate, which is not reported here So the drop in the deposit spread is interesting because you see it's it's much larger than the other two This is because the deposit spread depends both on the CBDC and the reserve spread so the reserve spread represents the marginal cost of issuing deposit because of this Of the cost function of banks depending on the reserve to deposit ratio While the deposit spread depends on the CBDC spread because here banks have market power So they can charge a markup on their prices So yeah, the deposit spread Drops by large compared to the other two spreads and here you see the CBDC demand the first graph pop left increases This is pretty intuitive because the liquidity benefit of CBDC has increased so then the reason why banks profit drops is because as the demand for CBDC increases banks face a Reduction in their market power. So in their possibility to charge this markup on the prices So this will eventually lead to drop in banks profits To prevent the huge deposit outflow So the main difference between these two Exercises we do it's in the deposit demand response. So in case of Negative shock to the substitubility between the the two assets. So the blue line on the top right graph so you see there is a minor Decrease in the deposit demand so Recall that here the rate of return on CBDC is lower than that of deposit So this scenario it's analogous to one studied by Bacchett and Peratzin So they also assume this and they find the similar Similar result. So here the intuition is that it takes more of one asset to replace the other In this in this shock to the so negative shock to the substitubility between the two So actually, I think I'm a bit ahead of schedule So, yeah, I just want to conclude to say Yeah, basically With this study we found that it's important to consider the degree of substitubility between CBDC and deposits When evaluating the consequences of fishing CBDC Of course, it's not easy But it's important to keep in mind and also accounting for the collateral requirement Banks must respect when borrowing from the central bank Because this the central bank loan rate depends on the constraint restrictiveness So as I said the tighter the collateral constraint the lower the central bank loan rate must be This is of course in the context of the equivalence analysis However Even when we find that CBDC has real effect on the economy and negative effect on banks profit these effects seem limited so Yeah, that's basically what I have to say So thank you and I look forward to the discussion But some of the additional material on the extra slides or you want to move over to the discussion It's interested in the model. I put everything in this light because it's a fairly standard. Yeah Well, then maybe we Whatever extra time we have we keep it for the Q&A and makes it makes it more lively. Okay, so let's move On to this cousin Dirk Niepert is going to discuss the paper Okay. Thank you very much. Thanks again very much to the ECB for for making this possible This is a very ambitious paper here by Maria Elena and Han Feng about I think two questions Not directly related. Maybe it could be even two different papers one concerns the neutrality questions under what conditions can we conceptually in principle Think about the central bank completely Stereolizing the effects of retail CBDC on the banking sector on the real economy on everything And in particular what they're asking is whether this is possible in the world in Which if the central bank were to refinance banks, this is only possible subject to some collateral requirements and the other aspect that they focus on is whether this would be Possible this neutrality if there is some non-linear liquidity aggregation somewhere in the economy. They particularly focus on some non-linear Liquidity preferences some CES aggregate of liquidity preferences. That's one part of the paper And the other paper is about is more standard in the sense that they simulate an economy subject to those frictions or subject to those properties and See, you know how it responds to certain shocks that is more I think aligned with some of the other papers that we have already seen and I think that we're going to see also today Now the paper that at the model that they are using to study those questions is based on some work of mine And I feel a bit embarrassed but in order to sort of pinpoint what exactly the contributions of the paper are I would like to Contra positive to what these other models of papers have suggested So when it comes to the neutrality They they contrasted with this paper of Marcus is in mine and What we found in that paper was that as soon as you have some non-linear liquidity aggregation or more generally some non-linear Feature through which liquidity matters for the allocation. You cannot get neutrality And the intuition is the following that a sufficient condition at least for neutrality Is that all the different agents in the model they need to preserve their wealth position because if the wealth positions Changed and the choice sets change and then everything will change to some extent So you certainly need this to be the same in order to guarantee neutrality On top of this what you also need to guarantee is that all of the agents in the model maintain their liquidity positions Defined as how much purchasing power are we actually giving up in order to hold liquidity? So these things you need to keep constant Otherwise you will not have neutrality, but if you need to keep those two things constant And you also want to keep the price system in equilibrium constant, which you also need for neutrality Then you need to hold constant the liquidity positions of everybody and the relative prices of liquidity So how much of a spread does it cost you to hold deposits? How much of a spread does it cost you to hold CBDC? For example, and if you want to have all these things being constant Which you need if you want to guarantee neutrality then you cannot have some Aggregation that is non-linear because as soon as you start to shift How much liquidity you hold in the form of deposits as opposed to say CBDC Then the marginal utility so to speak of these different positions will change and therefore the relative spreads will change And therefore you cannot have neutrality So I think we know from this result that Non-linear liquidity aggregation necessarily undermines neutrality. So in that sense I think that is not a new result that that Maria Elena and Hanfeng show us that this will matter somehow the really new aspect that they bring to the table is that you could still have neutrality according to their result if You impose a collateral requirement. That's sort of the new aspect I think if you look at it from this side Then the other paper and that's basically where they built the model on is is my paper from 22 and Whatever it took until it was published finally So there is no collateral requirement indeed So this is completely new what they introduce here in the first one of the first versions I had linear liquidity aggregation and then I focused on neutrality positive aspects normative stuff Out of this is gone now because the referees didn't like it the final version in the JFE in the JF Still has no collateral requirement But it has both linear and non-linear aggregation and it just focuses on the normative aspects So against this perspective again the contribution of the paper to me the most interesting question is could we still have neutrality? even with such a collateral requirement and then the positive analysis That is also completely different from what I did in those papers Let me focus on these two contributions and see What to make of those The first one on the positive side As I already told you I'm I'm perfectly on board with arguing and Expecting and believing that if you have C as liquidity aggregation, this will matter No, no doubt about this. I think this is clear very convincing I'm in the positive analysis as they conducted they do not to my understanding really model a central bank loan and Collateral in the form of government bonds and the reason of course is that beforehand in the neutrality result They argued it doesn't matter. So it makes sense not to put this into positive analysis So I'm sort of fine with what they're doing on the positive side. I believe that When it comes to the neutrality result there, I'm a bit less on board I agree with the derivation that Maria Lena showed you about what such an equivalent loan rate would have to be Such that if the central bank is willing to refinance banks at that particular rate The banks would be able to maintain their profits From beforehand the introduction of CBDC. I think what they lose sight a little bit of is the following feature The bigger picture of this equivalence result was You know, what does it take for the central bank to intervene such that the banks Maintain their choice sets in total which means in particular That you continue to rely on the banks to extend credit to the real economy or in the language of that model That on the asset side of the banks their capital holdings that their holdings of physical capital are Exactly the same as the one in the world before CBDC when they were financed through deposits because what you want Is that the banks can basically maintain their business model as it was except that on the financing side The depositors are replaced by the central bank loan. That's the purpose. You would like to Change the payment system Therefore you have to change the liability side of the bank's balance sheet But you don't want to interfere in any way with what the banks are doing on the asset side of their of their balance sheet If you can maintain this and you also get the incentives for the banks, right? Then you have insulated the banks from the introduction of retail CBDC because then they can continue doing their business Exactly in the same way as they used to it's just that the financing has changed Now the question is this possible and the answer is no because what they are doing is that they are saying, okay Let me the central bank refinance the banks But only under the condition that the banks only holds such and such amounts of government bonds Which then will be the collateral for me the central bank to extend this loan to the banking sector But if the banks hold all these government bonds, then they will not be able to hold capital anymore So they will not be the ones to extend credit to the real economy anymore in the equilibrium that they show us I think it is the central bank now that has taken on all these capital holdings from the banking sector So now we have truly this but some people fear the neutrality result would suggest this big socialist central bank out there, right? That isn't the business of credit extension to the to the real sector So in that sense, I don't think that it's really in your charity result It's a result that says, you know that the central bank could take over extension of credit to Main Street and Still maintain the profits of the banking sector That's true, but it would not maintain the business model of banks in the sense that they are in charge of extending credit to Main Street Now you could say, okay Then let's rewrite the model a little bit and let's assume that rather than having then holding bonds the bank could sort of Provide the capital exposure to the real economy as a collateral to the central bank and And there's potentially some some way to go here But then again, this would not be truly a neutrality result because in the initial allocation This capital did not serve as a collateral to depositors if it has to serve as a collateral later on to the central bank Then this makes a difference unless in the first economy collateral wasn't scarce at all There was no role for collateral or whatever. So I don't think it's truly in neutrality result that they are showing here I think what they established is that you can insulate bank profits But you cannot insulate banks business model from retail CBDC let me conclude with What I think is so my perspective more generally on this collateral thing I think so When I argued that to a first Approximation the central bank could insulate commercial banks from retail CBDC and I get the response This doesn't work because of collateral. I buy this I think collateral undermines the neutrality result Now is this bad for the neutrality result or is it bad for the real world or whatever? I don't know my perspective somewhat is that in the current world We have a we have a monetary architecture in which depositors provide funding to Deutsche Bank Anticipating whether correctly or not that in the worst possible scenario the ECB would be backing Deutsche Bank So right now we believe as depositors somehow that the central bank is actually standing ready as a lend-off last resort to You know safeguard depositors. So the current system works without collateral and Depositors do not ask for collateral from Deutsche Bank when they're providing funds to Deutsche Bank in this world with CBDC We would just make this explicit It would be exactly the same arrangement except for the lend-off last resort arrangement would become an explicit arrangement in the sense That this refinancing through the central bank actually happens Now it is correct that if in this new world, which just makes explicit what is implicit today If in this new world we ask suddenly for collateral then there is no neutrality. That's true The question to me therefore that comes out from this non neutrality is rather Why don't we ask for collateral today in the current system? Why do depositors today don't ask for collateral when they provide funding cheap funding to commercial banks? Although in essence it is the central bank that makes sure that we are doing this at least we believe that it's the central bank that Safeguard us when we are doing this and give cheap funding to Deutsche Bank. So that's sort of my general Perspective on this. I think it's a it's a nice paper. I'm not fully on board in this particular aspect. Thank you Excellent presentation great discussion. So let's collect a few questions both from the audience here in the room And if there any questions online as well and then I Give the floor back to Maria Elena to respond to both discussions and questions. Yes, can we Anybody with a mic to circulate? So let's start with the first round three questions one two three Thank you. So two things. I start with the with the model so I think they're on top of the issue that Was saying so is is about where he's going the risk in the asset side of banks He's moving from banks to the central banks. So you have a distribution on risk So and and the central bank would be not the lender of last resort, but the lender of first resort. I Think there is another issue So this is a RBC. So you don't have nominal rigidities and you have these lending rate that is offsetting in some sense the Profitability of banks But imagine to have some nominal rigidities and the lending rate then is also an instrument for monetary policy and You would be in a classical not in bank Timberger rule where you have the same instrument that is used for two different objectives price stability and Let's say keep profitability of banks. So this is another reason why I don't believe that it would hold in this case these neutrality of This policy Yeah, thanks. I actually have a question. So We have this CDC and preference shock and this leads then I think if I if I saw correctly on the picture to and also to an Increase in in deposits. So maybe you can also explain a bit more on this channel. I had a similar question is more for me to understanding how the model works and Say the impulse response that you showed it had it had two shocks One of the two was a preference shock that was common to the CBDC and to the deposits The epsilon if I'm not wrong, no, but the CDC goes up and that was it goes down and I would my intuition would have Been the the opposite. No, that is a preference shock is common. They should behave kind of similarly If you can maybe I missed some point probably in the discussion. Thank you Thank you. So there was a lot of emphasis on the neutrality point My question is is neutrality a desirable aim? I mean from a practical perspective CBDCs will be introduced For a number of reasons that we have discussed several times You're introducing a new liability of the central bank as a result commercial Banks will have to adapt as such they will have to adapt their business model and This doesn't seem to me a very bad thing. I mean short of creating financial instability episodes, which nobody wants You just forced commercial banks to adapt to a new system and they Could find new ways of making money and become more profitable once CBDCs are introduced and potentially new digital banks will be introduced Potentially new digital assets can be issued by the commercial banks themselves. Thank you Well, thank you very much for the great discussion Yeah, I agree with all the points I think the point on the the comment on the neutrality result, which doesn't hold eventually. It's really interesting We can talk more about it So Regarding nominal rigidity Thank you. That's also very interesting comment about the Loan being an instrument for monetary policy so that also implies that maybe that this neutrality result does not hold So it will also look into that So in terms of the two questions on the dynamic in the in the model So the first question was why there is an increase in the deposit demand following the shock So This is the case for a shock to the liquidity benefit of CBDC So it's a positive shock which leads to an expansion in the liquidity in the system So since they are in perfect substitute in this case This will lend to the expansion of the demand for both assets With respect to the shock to the Substitutability parameter. So the second shock the one on epsilon the blue one So you were puzzled because they behave differently if I remember correctly So you mean in terms of the demand, right? Yes Yeah, which is the case for the epsilon. Sorry for the lambda shock But not for the epsilon because of the initial City-state values of the two rates So the rate of return on CBDC is lower than that of deposit which is something that can be Plausible So then as I said, it takes more of one asset to replace the other. So that's why we have this Minor decrease in the deposit demand And in terms of neutrality, I think I really need to think more about it So, yeah, I have a lot of thinking to do on that Say something On the neutrality Why is it that I get so fascinating about this? I think there's two important insights from neutrality results The first one is that we want to understand why CBDC matters in the first place It's not obvious and in particular when we write down more complicated models with all kinds of bells and whistles And we see herbs going up and down we want to understand is this because of CBDC Or is this because of some other some other talk assumptions in those models and often it's the latter So I think we need to understand exactly which condition of the neutrality result is violated So that we are sure that there's a Swedish CBDC proper and not something else which strives to results The second feature is that the immediate respond on the immediate thinking I think among many central bankers in Like five years ago was this is obviously crazy because it will change everything and the world will go under and I think the neutrality results with increasing sophistication show us It is not that obvious that the world will change dramatically as a consequence of CBDC The world might change dramatically as a consequence of particular central bank choices in response to CBDC Central banks have a lot of leeway to neutralizing these things if they wanted to of course They may not want to and I fully agree with you that we shouldn't aim at this This is not some normative objective to have neutrality But it's a lower bound if you know that in principle we could ensure a neutrality Then we know with very high likelihood We can actually do better than the status quo which should make us think about this option Let me abuse my world as chair to make a comment on on this Literature given a fairly imperfect understanding So I find it very useful as a theoretical benchmark to exactly see you know when it's violated It's a bit like maybe Modigliani-Miller on capital structure Nobody believes that Modigliani-Miller holds in the real world, but it really forces us to think you know What what are the assumptions are exactly violated such that the capital structure of of a firm or bank? Matters and that's why I find it, you know, it's a useful theoretical exercise, but but hopefully Central bankers around the world will actually not strive for neutrality But for an improvement in the allocation and you know increasing welfare by designing it in a way that it might for example in many papers We had which using imperfect competition in deposit markets over whatever the frictions are, right? Yeah, there's a very valuable benchmark like other benchmarks Okay, any any more questions comments If not Let's move on to the next paper a central bank digital currency and banking choices Presented by Andrew Archer from the Bank of Canada, which really seems to be a Bank of Canada effort of current or former Bank employees 25 minutes Thank You Tony and thank you for for bringing us here It's nice to reconnect after after a lot of years. I I think I was Tony's RA for a short period When I was at the bank as a research assistant So someone asked me yesterday, what did I learn from the talks yesterday and I said I learned it was a macroeconomic conference So I'm a I'm a structural IO guy So there's a bit of a bit of a story to why I'm presenting something that uses the methods of of IO To macro people. I think it's great So we're we're we're trying to take on these big questions about the future of money and the future of of the banking system I think pretty quickly we start talking about imperfect substitutability about market power About the choices of banks. I think these are worlds where Io economists and in more in general micro economists and micro data people have quite a bit to say So hopefully I inspire you to think about some of these things in a way That's not just macro and hopefully some of my answers will roll up and become part of Dirk or other people's papers thinking about how How substitutability might look So standard disclaimer doesn't reflect the Bank of Canada or governing council everyone here understands that sort of Thing so CBDC digital money. We're really focused on retail payment CBDC Despite this widespread interest, there's a lot of people who are worried about crowding out We're gonna call it. We're not gonna say disintermediation because we're really focused today on the deposit market We're not gonna have a lot to say about loans Okay, so we're gonna develop and estimate a structural model that's gonna be familiar to Io economists Maybe less familiar to to macro economists. I'll take you through some of the particulars One of the major things we want to highlight is that there's attributes that are gonna drive adoption and the response Other than the interest rate So a lot of people focused on remuneration. I think it's a key attribute There's other ones that matter and there's other ones that the seed that the central bank might control Particularly because a lot of people don't want to pay interest A lot of central banks are saying we're not paying interest. It's a different product Well, there's a lot of places that we should think about roles for the central bank or choices Firstly is the service location. How are people actually gonna get access to CBDC? They're gonna have to be onboarded. They're gonna have to get service. This is a very practical question And we're gonna compare schemes where it's done entirely online To done by the post office to done in actually in bank branches so imagining some sort of Intermediated model that is probably on the mind of a lot of central bankers is we're gonna get the the Bank branches to actually do the hard part and we're gonna offer the back end We're gonna think about what impact that would have on the demand Additionally, we're gonna think about the fact that banks don't just sell deposits So consumers when they're choosing their deposit bank care about the other products they might get at that bank A lot of people it turns out get their mortgage their credit card and they're deposited at the same place And that's gonna have a fairly big impact on the choice It's also something that the central bank probably isn't going to do Practically, we're not gonna offer mortgages to consumers I don't think that's surprising Okay Methodologically, so we're really focused on the consumer We're gonna build a demand side that uh, you're gonna choose your your main bank branch Based on the network based on the possibility of getting another product from that bank On the supply side, we're gonna have a fairly standard for i o Banks are gonna compete in differentiated basically Nash Portrait It's gonna be static And they're gonna compete in the deposit rate So in spirit, I think kind of similar to some of the other models We've seen that banks are basically setting a deposit rate and determining how much deposits they're going to get from a demand curve Uh, probably a little bit different than the approaches we've seen here in the last couple days We're going to estimate the model using household And branch level data. So we've got information on which bank which bank the household chose How much liquid assets we've got information on their portfolio of banking Banking choices We're going to learn something about demand parameters Basically, you're going to build a whole demand system preferences for rate of return bank branches preferences for other products And then with this demand curve and a notion of equilibrium We'll be able to basically back out what we need for a counterfactual Which is some notion of a marginal cost So if we have a demand curve a supply curve and a marginal cost we can play with the central bank digital currency In the counterfactual The challenge of course is that central bank digital currency doesn't exist But there are papers in i o particularly thinking about a minivan You can think about the demand for a minivan if you think about the demand for a sedan and a van Well, it must be somewhere in between some of those some of those attributes You can derive the demand for but there's always going to be some amount that we can't Learn until we actually issue Uh and on our paper the cbdc attributes or exogenous and we're going to vary them and think about how uh How the response of the banking system, but also the households particularly are going to look Uh a lot of literature some people are in this room Probably missing some people in this room's papers I want to focus however on the empirical work that's been done So jachi's previous paper, which I think is in the jme now It's going to think about just the household demand For cbdc relative to deposits in cash She's going to be using a similar or the same survey as us But she's really only focused on the household and trying to think about sort of the bounds and what goes into household choice of liquid assets Tony whitehead and a bunch of co-authors They're going to be quantifying the effect on lending in particular. They're going to have a structural model that focuses on Some of the things that your your macro models have focused on collateral The dynamics of deposits the dynamics of loan resetting Complicated model. We're focusing much much more on the consumer. We're putting the consumer in the driving seat What's saying? What does the consumer want? How are they going to adapt this product? There's a separate theory or a separate set of literature Kim P. Hoon One of his papers with some co-authors at the bank alex and someone else. I can't remember their name at the moment They're going to predict the adoption and usage of cbdc, but really at the point of sale So in a choice set with do I pay with my credit card or do I pay with cash? A complementary idea to ours. We're focused on how do I allocate my bank? So their their intercept is sort of our model and our intercept is sort of their model We're going to be on Adding this idea that this branch network Sort of a practical idea of how do people actually get cbdc? What drives their demand? To in-person services as well as this idea of complementarities Quick outline I'll take you through the model So we have three stages We're going to basically use backwards induction On the household to figure out what their eventual demand for Deposit bank is going to be what their utility from choosing any given deposit bank At the bottom There's some set of products k think credit card or mortgage or sort of investment product Where they're going to face a choice set That's going to depend on Whether they've chosen that as their home bank So they're going to get some amount of extra utility if I pick Deutsche bank for my deposits and Deutsche bank for my mortgage If I pick another German bank, I'm going to get slightly less utility The idea is going to be that I can figure out what my expected utility From getting a mortgages if I choose Deutsche bank Given where I live and how close the Deutsche bank and everyone else's branches are So we're going to roll up this choice problem into basically an expected Value in IO. We call that an inclusive value. It's including all the utilities of a choice that I'm going to make in the future Additionally, we're going to allow the household to basically allocate its liquid resources That are going to be fixed in our model between Cash so physical cash is going to remain and whatever digital product they've They've chosen to get either CBDC or deposits So we're going to allow the household to make this sort of continuous choice In a probably in a in a constant elasticity framework So it's something something a little bit similar to what we've seen in the previous paper Again, that's going to give you the indirect utility. You're going to take a maximum You're going to choose some interior amount of cash and deposits and that's going to give you a indirect utility Uh, essentially Then those indirect utilities are going to enter into basically a first stage utility function where I care about The branch network and also what I'm going to do later They're I'm going to care about the possibility of getting these complementarities But also care about the fact that I'm going to make this a liquid asset allocation and there's going to be basically Parameters that weight those those those relative importance those relative indirect utilities Okay, uh, essentially that allows us to integrate over households And get an aggregate demand curve. So given all the given all the parameters that we've estimated We get an aggregate demand curve suppressing all the the household level or all the the other Inputs to the function it depends on bank j's deposit rate, but also every other bank's deposit rate Uh, we're writing down a very simple profit function Uh, which is just there's some exogenous loan rate minus some deposit rate minus some marginal cost of doing both processes So I got to take the deposit which might cost me something but then I also have to make the loan This is sort of partial equilibrium way of thinking about the The profit function of the of the bank because we're holding the loans sort of exogenous Uh, take a first order condition You get this markup That is going to be equal to the inverse semi-elasticity This allows us to basically Take data data and data and invert and find marginal cost Which is sort of a trick that IO people have used for a while, but allows us to do these counter fractals So introducing central bank digital currency. It's going to essentially be a new product at the top level choice So you're going to be saying should I allocate it to deutsche bank or to cbdc or to another german bank Uh Introducing cbdc is going to have a direct effect Some people are going to like it. We have this uh, this error the of unobservable that says, uh You know any product of greater than minus infinity utility is going to have some choice probability Uh, so it's going to directly be a competitor Uh, but there's going to be these these choices or these attributes that the central bank has chosen Uh, that are going to impact this probability Uh, the interest rate is going to enter directly into the liquid asset allocation Uh, the complementarity is basically going to be turned off for cbdc So that whole term is going to be zero For cbdc for banks people are actually going to like that quite a bit Cbdc is not going to offer it and then we're also going to be uh differing the service location So how how uh attractive cbdc is based on how much uh Service locations and how how near they are to the consumers Uh, take you very quickly through the data Uh, we're using something called a kd financial monitor It's a wonderful survey because it has this micro data on not only where you live, but which bank you chose Some other surveys will tell you where you live and how much deposits you have But we know what bank you have and that's really important for For how we set this paper up Uh, also got location information on deposits, mortgages, credit cards, giscs I know the american term is basically a Uh cd so take a bit of a deposit So it's a insured but term sort of instrument that households uh Buy I think it's just a term deposit in job. It's just a term deposit here. Yeah But I guess I guess canadians don't go to their bank and say I'd like a term deposit. They get a gic So it's just a just terminology Uh, and then we have the location and then we got a bunch of rich You know, how many members are there in the household the racial characteristics the vocational characteristics even language characteristics we have We're going to be merging this with some data sets on where the banks are Where the post offices are and that's going to be our our variation or our our Our choice set for where to where's it going to be? distributed by we're using interest rates from A firm called cannex. We're using the demand deposit rates. So we're really looking for This sort of basic transactional deposit and we're basically assuming that that's the only thing that's a substitute to cbdc So other papers particularly the ones that focus on interest rates are going to talk about term deposits and cbdc We say if it's not really paying interest, we really only want it to be the The transactional deposits that are a close substitute Uh five year closed mortgage for our loan rate. That's the most common product in canada Uh closed just means you pay a fee to uh refinance if you do it before the five years And I think when I talk to american audiences, they're shocked that Not everyone in the world gets the wonderful 30 year mortgage But we're stuck with five year five year to 30 five year term, but 30 year amorts in canada is the is the normal So for demand estimation To be a little bit of details on how we actually do it We've set it up so that it's sort of done in stages This is for attractability issues and you can buy the results of one stage if you don't necessarily buy the results of another stage that's In contrast to how some other IO people will set up these problems For the bank choices We basically Run a discrete choice model pretty standard discrete choice model and we find a very strong preference For the home bank So if I've chosen it in my deposit bank, I'm probably going to get my mortgage from there Uh portfolio allocation uh We write down a optimization problem. We take the first order condition. I think we take logs We do some transformations. It turns out that it's a linear problem in Uh An unobservable so we can just run oos And there we find which we're pleased about higher deposit rate means you hold more deposits Relative to cash. That's that's pretty good for our our our Validation uh deposit bank choice. So we take these two as given roll them up into uh another linear discrete choice model Uh, we find that both of these terms matter quite a bit which is positive that our our theory has some predictive power Not only do do people care about their uh home bank When they're making the mortgage choice, they seem to internalize that at least in the context of our model when they're when they're making their deposit choice Strong preferences for better branch network. Basically, that means you like having more branches near you And you like to be close to the branch. So we're including how many Deutsche banks there are in your market, which is about two kilometers for urban customers and about five kilometers for rural customers And also where's the nearest branch? So just uh, just a great circle distance. How far are you from the the nearest branch and both of those matter a lot Uh supply estimation So we basically take this first order condition and do an inversion Take you through some of the averages very quickly Mortgage rate taken as exogenous is pretty high. Deposit rate is very low markup Which is the this inverse semi elasticity two percent The banks appear to have some market power And our marginal cost is about uh two point seven percent which we take as exogenous as well Uh in all our counterfactuals Uh, their return on assets for the largest banks in Canada in this period is about 1.9 So we're not that far off in terms of the markup And you might imagine that they're all integrated broker deal with their broker dealers So this is probably the we the the least profitable part of their business But we're at least in the in the ballpark of a potential profitability Oh, okay In counterfactuals, uh, we're going to hold these demand and cost primitives the same Solve a Nash per trend game basically in the uh deposit rates CVDC rate we're going to basically vary between zero and the average interest rate Which is only 10 basis points. I think if we went to the overnight rate Which is now above five percent. We have pretty different results But we don't think that's the relevant policy experiment at the moment Uh, we're going to allow the CVDC branch network to either be no network Basically an only online onboarding Uh, Canada post again, we don't know what the relationship would look like We're just saying we're somewhere where there's a federal office in a lot of small towns the post office And then we're also thinking about the union of all bank branches as sort of what is the extremum of the uh Of the best possible branch that could be available And then obviously switching off the complementarities Uh, the headline result on uh adoption So this is the share of CVDC in liquid assets If there's no service location not very many people like it They like the post office a bit better They like the bank branches even more And they like the super set of those even better And within each it appears moving a pretty small amount of interest In most of our models of 10 basis points doesn't have that large effect Which I don't think is that surprising Okay, average reduction in deposits So average across banks. So a slightly different way of aggregating that previous result We see a very similar picture That uh without a service locations Very few people are going to withdraw their deposits And if we make the service locations very attractive There's going to be much more and this is about the range of one of the largest banks Sorry, this is about the range of one of the largest banks in Canada uh Additionally, we're able to look at sort of the increase in the uh interest rates across banks Uh, so I won't spend too much time on this basically banks with the largest local market shares Are also those that have the highest markups and they're going to be able to respond more So they're going to lose for their deposits. So the guys that have really tight markups aren't going to respond to what and they're going to lose relatively more of their market share So we're able to think about the heterogeneity across banks They're going to focus too much on that what I want to focus on because I'm at the ECB And I want to I want to say a little bit about this is the holdings So I think uh Fabio Panetta yesterday was talking about how important it is to calibrate Some of these CBDC holdings We're going to we're going to be able to say a little bit about that in our model, which is exciting In con so the bank of england we were discussing has a much higher possible limits that they've at least discussed in speeches and And uh policy discussions am I that far off will be discussed And 20 000 pounds The ECB is much much more conservative And I think we're going to argue that's a little too conservative I Essentially our model if you choose CBDC You're going to allocate liquid assets between CBDC and cash And if you're constrained you got to hold the remainder in cash That's actually going to make CBDC pretty unattractive for someone who wants to hold a water liquid assets So you're going to find the the the adoption and the utility is quite a bit lower so here Not a lot to walk too far Here we're going to vary between Different holding limits in CAD and dollars obviously a hundred dollars is a ridiculously low limit But that's just to fix the ideas that yeah adoption would be pretty darn close to zero If there was no limit we'd have the results I showed you before That about 13 of households under the most extreme network would adopt or would hold their liquid assets in CBDC uh What's really important and really interesting though I think for for policy makers Is if you were to set the limit to quite a high value to 25 000 So after accounting for exchange rates more like the bank of england world than the ECB world you can reduce the Reduction in in deposits by more than half This seems a little counterintuitive if you think that everyone's the same The issue with these deposits though is that it's an extremely right tail distribution So there are some households that hold a couple thousand dollars and allocate them to another Liquid asset or they invest in stocks and then there's some whales who have hundreds of thousands of dollars in In deposits and it turns out that's who the banks care about And that's what matters for the aggregate share So it's extremely right tail distribution Uh, only 13 of households would be constrained by the limit of 25 000 And those households make up the most of the demand deposits So I think the 90th percentile is something like, uh, 31 000 dollars And that's 52 of all the liquid assets held by those households So I think there's a there's quite a bit to learn here Uh, I also want to encourage that if we're going to think about calibrating these limits You got to look at the micro data Not just averages So I think I discussed with this ECB Someone from the ECB previously that they thought about averages across households I think when we're thinking about this you really got to look at the whole distribution Of of uh a liquid asset holdings Uh So unless you my my my boss francisco had had said this can't be right There can't be people who hold hundreds of thousands of dollars in deposits their Opportunity cost is massive He's not wrong. We don't have an answer for why they're doing this, but they're doing it And I've confirmed this in another candy and survey So I recommend looking at as as many surveys and seeing I think this fact is probably true in In in most household surveys Uh to conclude because I think tony's about to give me one minute Uh, we can also look at the changing consumer surplus Uh Basically across different designs. It's hard to think about what the changing consumer surplus for just introducing a CVDC is but uh, we see households on average aren't hurt that much by a Pretty big change in the uh, or by a very large deposit women They go from 25 basis points to about 20 basis points average additional happiness something like a compensating variation In conclusion, we've built this model Uh someone from another policy institution asked me what a policy institution should take away from this I'm going to basically summarize this as don't worry too much about disintermediation I think as dirk said, there's a lot we can do There's a lot of levers to press on one of them is you basically control how attractive it is to the consumer And then you can use this limit You can even set the very limit very high and you probably shouldn't worry too much about disintermediation from deposits because The the biggest depositors aren't going to leave Uh, just an aside a final aside is I think it'd be really interesting. I think in Tony's paper to think about the Uh flow limits. It's very hard to find data To appropriately calibrate or appropriately find those those, uh, those flow limits But it's something that's also in this discussion. Not just the holding limit All right. Thank you. Uh, I'll pass it over to nikola The discussion this, uh, nikola pavanini So let me start thanking the organizers for including me in the program and asking me to discuss this very interesting paper um I'll just wait for the slides to come Okay, perfect Okay, so I'm going to be briefing summarizing the paper because I think andrew did a great job and and I'll spend most of the time on my comments So just a few big highlights on my take on the paper So there's a couple of research questions that the authors are after here It's trying using a model to understand how cbdc would affect demand and supply Of bank deposits and in particular they focus on two features of bank deposits that make this product differentiated Which are the ability of banks to provide financial products and the fact that they rely on a branch network Which consumers might like so how how to what extent is to These two characteristics influence the effect of cbdc on demand and supply So they build a model that can be showed you On deposit demand and supply whereas the supply is basically banks setting interest rates and competing patronage And households can choose cash versus deposit holdings and then conditional on deposit They choose which bank they want to deposit their money in and they value branch network and the fact that they get complementary financial products And then they use the model to run counterfactuals in which they introduce a cbdc And they do it under alternative scenarios with no or some interest-bearing some remuneration and with alternative degrees of service location that cbdc wouldn't jar They use this data as as and you described from from canada in which they have information on on households deposit and bank accounts And the loans that they get from from survey evidence They know the allocation of liquid assets cash versus deposits and they know the location of this depositors which allows them to match this to the location of branch networks and and Define a choice that the households have based on some travel distance time limit That defines how far they're willing to go to Deposit their money and then they have information on bank level interest rates on deposits and mortgage loans And the key takeaways is that um the take up of cbdc heavily depends on the service location that will be assigned to it so that Without any service location fully digital just like andrew showed us The take up would be 0.7 percent Market share of deposits and it can go up to almost 12 percent With the maximum amount of service location And complementarity with financial services matters a lot, right? So in a world without it to kind of reinterpret the results that andrew showed you cbdc could actually reach 38 percent market share So this means that this plays a significant role And then he they look also at holding limits and and they show how for for example 25 Thousand canadian dollars holding limit will reduce the share by half across different scenarios in terms of service location So this is just a couple of figures from the paper that and you already showed you on the left hand side You see the share cbdc across four different Degrees of service location and so it can go up to 12 percent and most and on the right hand side You see how banks would respond in terms of Deposit interest rates to kind of contravail the the loss of of market shares and and the numbers here are pretty small Right so in in the best scenario the largest service location banks would increase deposit rate by 3.5 basis points So that's that's a pretty small number Last thing I want to show you is the effect on consumer services is something and you didn't show you And again, you know the four set of of columns here represent The different degrees of service location vertical axis is the change in consumer surplus And the breakdown of this bar tells you to what extent the gain in consumer surplus come from three different potential sources In gray higher deposit rates in in purple better service network and in and in Dark gray more choice varieties. So the vast majority of this comes from a better service location through the cbdc, right? So I have four set of comments here and suggestions The first is on this role of the complementary financial products, which is is a key part of the paper The second is on on The fact that this model's basically allows for a single bank choice for deposit. No not multi-homing Then I have something on identification and that kind of big picture question zone Where the model could be taken to and this kind of addresses some other questions other papers have addressed in this Conference, but doing it through the eyes of an IO model I think you could still look at stability competition and transmission monetary policy and and I have a few suggestions on this So let me start from the first. So the complementarity financial products, they they're important, right? But but they're kind of exhaustionously given within the model and of course you could think that banks might respond adjusting loan rates as well to the introduction of cbdc and and And we and co-authors have data on loan rates So that's something that that could be used to Could be enthogenized within the model and I think it's relevant because if you look at the literature Borrowers are are much more elastic to loan rates as depositors are to deposit rates So it would be better for banks to adjust by little, you know That the the loan rate reduced it to keep the depositors in because they value this complementarity And it's also not obvious that loan rates would go up and down in response to bdc Right because they could be reduced to attract borrowers to the complementarity But it goes to be increased right if if the cbdc for example makes deposit funding more costly as as they show that it does So data is there right from loan rates It's used for bank profits But it's not used to Determines the benefit that the borrowers get from financial products. So I think that's that's an easy thing that could be adjusted and and also you could think of estimating at the amount for bundles rather than then this kind of sequential choice that that they model in which How to choose deposits and loans as as kind of in a simultaneous way And Matt chance code as an interesting AR paper that started this literature I was also a little bit puzzled by the fact that there's a parameter in the model that basically determines the Fraction of households who obtained a loan right? This is all held as fixed And and I wonder, you know to what extent this is demand or supply So a little bit of more discussion on that would would would help The second point I have is on single bank for deposits, right? So the way these models are geared is to basically allow for mutually exclusive alternatives So this means that you can only choose one bank as your deposit, but you know, we I would like to understand to what extent this is during the data So I don't know if for Canada I look for data from the Netherlands and I Just had evidence from Relatively old survey which shows that 22 percent of households actually deposit across multiple banks And these are the most wealthy ones, right? So in terms if you think of volume adjusted This is potentially a non-negligible fraction of money that is deposited by the same person across across the firm banks And and I I'm also thinking that you know going ahead if I think about the Keynes Beach You know if we have this big tech firms that offer digital wallets may be Multihoming in terms of deposits is is the trend that they were going towards so perhaps Catering the model to this type of of behavior of consumers might be an interesting extension and and you could think of you know People depositing across multiple banks cbdc and a private bank and this would change interpretation that andrew gave in terms of How this affects constrained households because of the holding limit, right? And it would be also more realistic I think if you're allowing the cbdc to use the service location of branch networks Right because then you're already going to the branch for cbdc Then you might as well open an extra deposit account there And connect it to this if if cbdc uses the branch network And then I think there could be you know kind of revenues the bank should expect for offering this service that could be including to the model to to kind of Taking to account of their response I have a few comments on the identification right now the key parameter here in the model is Basically depositor's demand elasticity because that tells you the substitution that that happens in the data Um, and so here I think it would be nice to compare this the numbers that you get with the literature It's a little bit different the way you estimate demand relative to what other people have done But I think we have a bunch of numbers out there for this class of value models So I think it would be useful to to know how you compare to those It wasn't also not entirely clear to me What kind of variation in the data allows you to separately identify the fact of the branch networks have on The three decisions that households are making in this model right deposits cash versus deposit And this this financial products and especially that you know these three decisions might have correlated on observables And and perhaps modeling them Sequentially might lose a little bit of that Um, I was also a little bit puzzled that you know, you don't use any instruments here for for To to recover the slope of the demand curve and which is kind of standard in the literature And and the last point I have here is that you focus on households with our insured depositors presumably But there's a fraction of how of of depositors that are uninsured, right? If I look at the eurozone data that I've worked on that that's kind of 25 percent of total overnight depositor uninsured So these are first potentially and what the literature shows is that Uninsured depositor large depositors are less sensitive to deposit rates and are likely benefiting more from Complimentarity relative to households because they're more likely to get a financial product And so this might mitigate the impact that you're finding even further So I I wonder whether you know, you could have at least some aggregate data on uninsured deposits that you could that you could include um And then the last point I have is is kind of on a bigger picture view of how this model could be used or taking Taking ahead and and extend it to understand the um, yeah Uh, how cbdc affects the banking sector through the lens of an io model So for stability, they're just type paper by by egan or taxable models. They actually endogenize banks default and and and Still have the positive competition and it's not obvious, you know from from from your model how cbdc would affect stability It could be a negative effect if they possibly become more costly But it could be positive right if for example banks balance sheet shrink and they become less systemic Of course, there's consequences for the real economy But but at the same time it could also be that you know if cbdc deposits are lent to banks by lending facilities This is something that the model could could quantify right this could have a positive effect on on stability I was also interested by the effect on concentration that you show right in one of your counterfactuals You show that larger banks those with a larger markup are the ones that software the la the least from the introduction of cbdc So this could have some consequences from Competition policy as well and and if we think about concentration This is also something that mediates the transmission of monetary policy And that's something that you could look at in in your paper right because If the banking center becomes concentrated this you know through a market power effect could affect negatively the transmission But you know if if the central bank uses cbdc To directly implement monetary policy this could you know ease the transmission and and there's a nice paper On that that you could rely on So let me just conclude to say that I really like this paper. I very much enjoyed reading it I think it's it's you know, it's a good take and and it's a use of IO models Which is slightly different from most of the papers that we've seen Today and yesterday. I think highlighting the importance of product differentiation You know service location and complementary financial products is something that tells us about the you know Kind of the pass through with the effects of of cbdc on the positive markets What I would work on it is you know focusing a little bit more on perhaps making the Loan rates and dodges in the model I think that's not too hard and perhaps also including this other class of depositors the uninsured ones that Um, you know could also be run prone for example So I have a some effects on on stability too And then I would you know overall think of you know How this model could be extended for other papers to look at stability concentration and transmission of monetary policy So thank you again for asking me to discuss this and Thank you very much. Nicola before giving the Giving entry a chance to respond. Let's collect a few questions from the audience and perhaps online as well Hi, thanks a lot. Um, I fully agree with you that I think um the service location and the network is very important However, I think this was very focused on the physical worlds and I think I mean we're talking about the future of money So I don't really care how close the next post office to me is or how Close the next bank branch of this But I much more care about the user experience and kind of the digital convenience and how easy to onboard Kind of by video identification So I was wondering kind of how how kind of what would this change if we also look at this digital kind of the digital I mean The digital digital convenience of it and and what it would change and how it could be onboarded And I don't know also how I mean, I guess it's much more difficult to find data on this So how we can bring this into the model and what would it change? Thanks Thank you It was a fantastic paper. First of all And I have two point one is more about the discussion I I totally agree that the issue of multiple accounts in different banking is crucial and one thing that there is a literature mostly on I know a couple of paper on on Denmark where they show that regulation on insurance of deposits is crucial in order to see This kind of behavior So when you have a limit in a single bank, you can have regulation where you are allowed to have Let's say more than account and In the limit is is is not binding in this way second point is is on the On bank side so on the supply side of deposits as I know about Europe. I don't know very well about the the The situation in Canada when you did your estimate But here is in order to look for the degree of competition on deposits It really depends on on the unconventional policies that we had during the last 20 years So we know that the the the QE had increased a lot the amount of deposit reducing the The the competition So I don't know how robust your results are on this Thanks. So my question is about your results on holding limits So for one, I'm not so sure about how you actually identify The effect from the data and then I also wonder About which type of hoarding hoarding limits we're actually talking about So is it the type of hoarding limits that prevent you from using it or the cbdc in payments? Or is it rather the type that? It just appears bad you as a customer see it's 300 euros and it looks bad and you don't want to hold that So I ask it because at least talking for the digital euro there are solutions That would allow you to still make a payment even though you breach the holding limit a great paper. I think So I kind of agree with what was asked before The onboarding would be made by commercial banks. I mean It's not that we are going to go to some physical place and demonstrate who we are to be open now I mean wallets would be offered by banks in the tier system KYC is already done. That's the quip pro quo between central bank and commercial banks Right, so you commercial banks you do kind of the dirty job for me of Offering cbdc is the question is what they're going to get in return two things first in fact it was asked by nicola Remuneration of course is going to be a fee most likely The question is how to calibrate the fee second is The Offering as panetta said yesterday that cbdc's will be Depotentiated from what they could do offering zero remuneration And not programmable So these opens the space For commercial banks to offer something new even compared to the standard deposits That you're talking about and is absolutely correct That what Make you choose one bank versus the others is the financial products attached So this new thing could be either a tokenized deposit Or a stable coin issued by commercial banks With service attached to it and then the competition will be among commercial banks and cbdc's will only be the basis of a payment system to ensure that the In the digital asset world The way they want to do it then they can be turbocharged one, you know 10 years from down down the line, but not now so How your model would change if this new form of bank Instruments such as tokenized deposit or stable coin were to be introduced. Thank you um, so perhaps one thing Concerning the the uptake that you get that this very small amount if you have only the Up the the on-board not going through banks or post offices So, how would you define? cbdc to be successful because basically you don't have I think payment use for payments in your model So would you also consider cbdc to be successful if you have a very small percentage holding overall, but lots of Transactions going through or would you require sufficient amounts of holding to qualify it as a success? Andrew if you could maybe limit it to two three minutes at most or a few minutes in response So we Enjoy enjoy the coffee out here. Sorry. You're giving me a lifeline then. Okay. So thank you for the discussion and the questions I think I'll work in reverse order Since that's what I have open on this last point about defining a success far above my pay grade I think you have reached a really interesting question though about differentiating between holding limit and usage in payments I also want to encourage you to think about maybe it's an outside option that it could be a success if both are zero It could be a case that you're changing the attributes that are offered in the private sector and we could be having a massive important effect even if we observe zeros So I I I've struggled to try and answer that when my boss asked what would be a success I struggle again to know what a success would be we can tell you what the what the positive answer is the normative answer is much much trickier The onboarding so I think there's two people that asked that why are we using a physical location? I My answer is going to be in the survey data The households care a great deal We still have 80 something like 65 percent of households going in every month to their their branch location I think most the people I talked to People in this room probably don't go to their branches that much a lot of people do Uh, I think one of the ways we could think about moving towards the future is we have age And I think we've done some work to look at the younger people are less sensitive to branch location But they're still sensitive and that sensitivity is very high So so it it's it's onboarding or its service they can get or it's even Awareness and and understanding of where they could get service if they needed it. So it could be the option value of service In terms of purely online We have purely online banks in canada. They have a very low market share So our model is going to say people don't like that very much Uh holding limits which type of holding limit? The one that can work in our model, which is very stark, which is if you go over you just have to hold that in cash And I think that does speak to the the multi-homing question that I'm getting from several people I think we're working on multi-homing It reduces a couple more parameters That the model is sensitive to we think that The single homing assumption is going to uh, sort of be the starkest result in terms of the holding limits That uh, if you introduce multi-homing people are going to hold even less in cbdc They're going to basically balance between cbdc and and the bank account There's also econometric issues. How do you Learn who's going to be or how do you how do you differentiate those people that are going to be multi-homing versus single homing? In response to to the discussion on that. Yeah, it's not zero in canada. It's probably similar to the netherlands uh question about how to model it Uh, there was a question about the supply side unconventional monetary policy and qe of households We have household data We didn't do quite as much qe in canada and we certainly never had negative interest rates So it's probably a little bit different. I also think you're probably talking about the wholesale And the larger deposits. We're looking at households So The 99th percentile is a million dollars. So I think we're talking about different groups It's probably a problem if we want to think about the european context Maybe less so for the canadian context, but it's a fair point uh Yeah, so great comments concentration goes up I think this is a problem when people confuse concentration and competition I usually think in terms of markups when you think about concentration You could have a negative impact transmission of monetary policy stability are great questions It's how to add them how to make it uh, tractable, but also, uh, applicable Identification So we're relying a little bit on we have microdata and fixed effects I think we're you're right. We got to explain that more at least and we're going to do some sensitivity analysis around using the egan hortax you instrument Uninsured versus insured Uh Something to look at as well multi homing. We're working on again. There's there's into implementation questions That we should discuss Uh And then the endogeneity of of lending, uh, the challenge I think we're facing is how much do we want to make? Uh deposits numerically equal to loans Do we want to allow different sources of funding that gets very quickly into the questions of of tony whitehead at all? So we're uh, we're Opening another can of worms with that shall I say but thank you for the comments. They're they're very helpful and I want to discuss them more Awesome before I let you go and true, uh, there's one more question from the online audience so I'm I'm just reading it out loud. Um Elena febrile from the spanish treasury asks whether you've contemplated to explicitly Create money demand functions in your model Uh, satrauski shopping time models something along those lines I Know so I think I think you're talking about micro foundations of money demand. I think we're estimating a money demand function in a sense Uh, we haven't It is a good point. We could probably get some variation there in terms of distances Maybe for the shifting the the consumers demand We're allowing there to be a household specific demand for cash relative to deposits But there's probably some things we could add in the spirit of those models But we probably won't explicitly have one of those models But thank you for the comment. That's that's something to think about