 So we can go on now with session two on the effectiveness of monetary policy. And I would like to ask the first speaker and the discussant to come up here and sit at the table and that would be Jan Hanes-Lang and Antonia Natoli Segura. So Jan is an economist in our micro-potential policies division here at the ECB and you work on micro-potential policies both on I think both on borough based and capital based measures and a lot on issues that are related to real estate markets. And yeah, and Anna Natoli is economist at the Financial Stability Directorate at the Bank of Italy. Welcome to the ECB and you also work on micro-potential policies, stress testing and bank resolution frameworks. So thanks very much, Jan. I would give the floor to you. Do we have the... And I think there's a pointer. Yeah, so first of all, thanks a lot to the conference organizers for inviting us. So the paper I present today is called the State-Dependent Impact of Changes in Bank Capital Requirements. And this is joint work with Dominic Menow from the Bundesbank. So he's also here in the audience. So if I say something you don't like and you can go to him after the talk. So I think the paper we're presenting today actually speaks to a couple of the questions that were raised yesterday in the policy panel. So in particular, in the current environment, is it still possible to go ahead with tightening micro-potential buffers? That's one of the questions. And the second one is how high are the costs in general of increasing micro-potential buffers? And I think our paper provides reasonable and intuitive answers to both of these. A spoiler already. So I think the key message would be, in our view, you can still go ahead given high profitability of the banking sector. And also the costs, if you do it right, phasing in the higher capital requirements can actually be extremely low. So therefore, you probably want to err on the side of higher buffers in normal times than rather too low buffers. So I hope I caught your interest in our paper and I'm going to provide you some more details of how we get to these conclusions. So given that we work at central banks, the usual disclaimer applies that these are our views and not those of the institutions. So for the talk today, there are four sections. So I will start with a brief motivation and an overview of the key results that we have. Then I will show you the structural model setup that we use for our analysis. And then in section three, I will provide the key results of our model, both analytical results and quantitative results. And in section four, I will provide a brief conclusion and sort of implications for policy. So what's the motivation for the paper? Our key motivation is really that the impact of changes in bank capital requirements seems to be state dependent. So if we look at empirical evidence, there's evidence that there's close to zero impact of increasing capital requirements when you're in good states of the macroeconomy. And you can reduce lending by a few percentage points if you increase requirements in bad states. And you can see this illustrated in the left hand chart. So here we summarize the findings of some key empirical papers. And you can see that basically, for example, when capital buffers are high, the output gap is positive or when banks are profitable. The impact of increasing capital requirements is very low, so very close to zero. But if these factors are adverse, then actually lending is reduced quite a lot if you increase capital requirements. And the second key motivation for our paper is that actually this feature of state dependence is missing from many of the standard macro models, especially the ones that we use for policy evaluation at central banks. And you can see this illustrated in the table on the right hand side of the chart. So we borrowed this from a recent ECB paper, which summarizes basically the quantitative implications of various DSGE models of a one percentage point increase in requirements on lending. And you can see in the first column that basically most of these models would lead to a reduction of roughly one percentage point in lending for one percentage point increase in capital requirements. So this is really the motivation for why we study state dependence in our paper. So let me walk you through sort of the key results and what we do in the paper. So our key result is really that there's strong state dependence and the impact of changes in bank capital requirements can differ by up to two orders of magnitude, depending on whether you're in good macro financial states or in bad macro financial states. So how do we get to these conclusions? So we build a structural nonlinear banking sector model and there are a few key ingredients. So we have monopolistic competition, but that's actually not crucial to derive the results, but it helps in some of the modeling. But we assume that bank equity is always more costly than debt. So this is actually an assumption that makes it or tends to make it actually hard for increasing capital requirements to have low costs. But even with that conservative assumption, we derive the results that I will show you. And then within that framework, we add two occasionally binding constraints. So the first one is a potentially time varying capital requirement. And the second one is that banks cannot issue equity so they can only increase equity through retaining earnings. So this is actually also a standard or an assumption that is done quite often in macro models. And what we show is that the interaction of these two occasionally binding constraints really introduces this very strong state dependence of the impact of changes in bank capital requirements. So when banks are in normal states and we basically term something as normal when banks hold voluntary capital buffers and they make positive profits so that they could actually cope with moderate capital requirement increases. So when banks are in normal states, the impact of increasing capital requirements is actually extremely low. So it's roughly minus one percent in loans for one percentage point increase in capital requirements. And what we show is that the impact can even be lower than this. However, if you're in bad macro financial states, so meaning states where banks don't have voluntary voluntary capital buffers or very low voluntary capital buffers and where they make big losses, the impact of changes in capital requirements can be extremely large. So we show that the impact can be up to 10 percentage point more loans if you are able to release capital requirements by one percentage point in these states of the world. And let me quickly illustrate sort of the intuition behind these results because I think it helps a lot to understand the more technical results that we present later. So in normal states when banks have voluntary capital buffers and are profitable, something that we call a pricing channel is present. So banks have the equity to fulfill higher capital requirements, but because equity is more costly than debt, the funding cost of loans increases. But this impact on funding cost is actually very small. So it means you only move up a little bit on the loan demand curve and the impact on equilibrium loan quantities is extremely small. However, if you're in the bad states, something is present which we dub the quantity channel. And it simply means in these bad states, banks are already equity constrained. So if you are able to actually change the requirement because banks are highly leveraged, that requirement directly affects the loan quantity that banks can supply to the market. And therefore basically it basically means you're controlling something like a financial accelerator. And if you're able to basically alleviate that constraint in the bad states, you can have a big supporting impact on loan supply. So in the interest of time, I will actually skip the presentation of the related literature. But in our view, the key contribution of the paper is really that we spell out clearly these two different state-dependent transmission channels, so pricing channel and quantity channel. And we derive some very clean analytical results regarding the quantitative implication of these transmission channels. So let me briefly run you through the structural model setup that we use. So we assume fairly stylized, but in our view realistic bank balance sheet and profit and loss account. So basically on the asset side, banks hold loans, so there's only one asset that they have, and they finance that via deposits and equity. So the first equation just shows you the balance sheet identity. And then the profits of the bank are defined as net interest income minus the cost of risk, so provisioning and minus operating cost. So in the equation, let me briefly explain the different terms. So net interest income is simply the interest rate that the bank gets on loans times the loans outstanding minus the interest rate on deposits times deposits. And we substituted out deposits by using the balance sheet identity and substituting in loans minus equity. Then cost of risk is simply we assume that a time varying fraction theta of loans default and they need to be written off through the P&L. And then operating costs, we simply assume that they scale up linearly with the size of the bank. So let me mention two important aspects regarding the modeling. So the loan interest rate is endogenous in our model. So we have a monopolistic competition and we have an aggregate loan demand function that is downward sloping with a constant interest rate semi elasticity. It's basically illustrated in the annex, but in the interest of time, I will not show the equations, but it's important to understand that basically the interest rate on loans depends on the loan choice of the bank and on the aggregate loans supplied in the economy. And then the impairment rate already mentioned we assume that it's stochastic and time varying and we assume it follows a log AR1 process. And the reason for assuming a log AR1 is that in the data we have actually a fat right tail of provisioning so that features needed to mimic the data. So as I mentioned, we impose two occasionally binding constraints. But before explaining the constraints, let me quickly explain how equity is built. So banks can only build equity through retaining profits after dividend payouts. So what you can see is basically next period equity of the bank is equal to starting period equity plus the profits during the period minus the dividends that they pay out. And while the dividends are the choice, optimal choice of the bank, we assume that banks cannot issue equity. So the way we impose that occasionally binding constraint is that dividends have to be greater or equal to zero all the time. So some of you might think, okay, that's an extreme assumption to assume that banks cannot issue equity at all. But our reply to that would be, I mean, first, as I mentioned, it's often assumed in macro models. And in the annex, we also show evidence that actually banks rarely issue equity in the Euro area, even when they make losses. So for unlisted banks, even in 90% of the time where they make losses, they don't issue new equity. For listed banks, it's a bit more, but even their 50 to 70% of the time they don't issue new equity when they make losses. So in our view, it's an extreme assumption, but let's say it's a high level first pass approximation. Then the second occasionally binding constraint is that potentially time varying capital requirement are and that must be met by the banks at all times. So the capital ratio, which is defined as equity over risk weighted assets. So Omega is our risk rate that we assume this must be greater or equal than the capital requirement all the time. So the objective of banks in our model is that they maximize the present discounted value of expected dividend payments. And the discount rate that they use is determined by the required return on equity, which we denote by role. And we assume that raw is strictly greater than the debt funding cost of the bank. So we make it basically in principle, we make it costly for increasing capital requirements. And the decision problem of the bank can actually, so this dynamic maximization problem can be represented by the Bellman equation that you see on this slide. Basically the value function has four state variables. So you have the credit risk shock theta, loans of the bank, equity of the bank and the aggregate loans in the economy. Because as I mentioned, the interest rate and therefore also the payoffs depend on the aggregate loans in the economy. And what's important to note is basically the two occasionally binding constraints that we introduced in the second line of the equation. So the first one is the capital requirement constraint. And here we denote the associated Lagrange multiplier with Chai one. And the second one is the equity issuance constraint where the Lagrange multiplier is denoted by Chai two. So the equilibrium of our model is determined by the first order conditions. And after you take the first order conditions to impose the representative bank assumption. So basically the idiosyncratic loan choices and the aggregate choices need to be consistent given that we assume a unit mass of identical banks. So given that I've presented now the model setup, let me briefly run you through the key results. So we have analytical results and we have quantitative results for model that we solve with global solution methods and we calibrated to Euro area data. So the first result is that there is a pricing channel of changing bank capital requirements. So proposition one states that in the absence of an equity issuance constraint, the equilibrium loans respond to changes in capital requirements by a pricing channel. So the percentage change in loans is equal to the right-hand side expression. And let me run you through the intuition behind the right-hand side and this pricing channel. So if you increase capital requirements, given that equity is more costly than debt, this should increase the funding cost of loans. And that's basically captured by the square brackets. So raw minus id is simply the equity premium and omega times delta r is the additional equity that you need to fund the loan. But the key insight is that actually the funding cost impact of higher requirements should be very low. So if you plug in reasonable numbers like an 8% cost of equity, a 2% cost of debt, 50% risk weight and a 1% point increase in requirements, that should increase your funding costs by three basis points. So we're talking about one-eighth of a standard monetary policy rate increment. So this is actually super low. And then given that we assume monopolistic competition in the model, this increase in funding cost is passed on to borrowers with a markup. So that's why you have mu over mu minus one in front of that funding cost increase. And the impact on equilibrium loan quantities then depends on how elastic loan demand is. So that's why you have the epsilon there, which is the interest semi-elasticity of loan demand. And if you look at empirical estimates, they are usually estimated around three this epsilon. So if you put basically the numbers together, you end up with this minus 0.1% decrease in loans for one percentage point increase in capital requirements. So these results are for the model where you don't have the equity issuance constraint, but importantly what we show is for the full model with the issuance constraint in states of the world where banks actually hold voluntary capital buffers before and after changing capital requirements, a similar pricing channel is active and importantly the impact can even be lower than this minus 0.1 on lending. So the key message regarding the pricing channel is really that the impact on lending is very small. And here we illustrate this with simulations from the full model. So what you see in the chart is on the y-axis is the decrease in loans in response to an increase of capital requirements from 10% to 11%. On the x-axis you see basically the relative frequency of these impacts. And we only look at states where banks hold voluntary capital buffers before and after this increase in capital requirements. So focus first on the red bars. So if banks actually hold voluntary capital buffers but they don't pay dividends before and after, then you get this minus 0.1% drop in loans for one percentage point increase in requirements. If banks also pay dividends before and after, the impact is virtually close to zero and that's illustrated by the blue bars. So the impact can even be much smaller than the 0.1 and can basically be almost nothing. The second key result of the paper is that there's also a quantity channel can be present for changing capital requirements. And this is illustrated with Proposition 2. So in states of the world where the equity issues constrained and the requirement constrained are both binding, equilibrium loans respond to changes in requirements via this quantity channel. And the quantity channel simply states that the percentage change in loans that you get in equilibrium is equal to minus the percentage change in the capital requirement. And given that capital requirements tend to be low, so let's say 10% roughly, if you change it by one percentage point, that roughly means a 10% point change in the requirement. So you can already see the quantity channel whenever that is present will have a big impact on lending. So the intuition is really that. If banks are equity constrained, so basically if both constraints bind, they don't hold voluntary capital buffers, they don't pay dividends. So loans are simply determined by equity plus profits and the requirement and equity and profits are given. And you can see from that equation, if you change the requirement, the only thing that can adjust is loans and given the high leverage, the adjustment in loans will be very, very large. So changing requirements when the quantity channel is present can have a big impact on loan quantities. And we illustrate that also with our model. So basically the quantity channel impact on lending can be very large and it can be basically up to 10% point for one percentage point requirement release. And we illustrate that by shocking our model economy with a very big credit risk shock and then comparing the evolution of loans for two different economies. So one where we keep the capital requirement constant at 11%. That's the red line in the charts. And one where we decrease the requirement from 11% to 10% upon the impact of the shock. And the shock basically is the same in both cases. So what are the mechanics? When the bad shock hits, banks make big losses that eat up all the voluntary capital buffers. They're actually forced to deleverage and you see that by the red line in the right hand side chart. So loans would need to drop by almost or by more than 12% to still meet the requirement given the heavy losses that banks incur. But if the regulator is actually able to release the requirements to 10%, this gives banks basically space or available space to absorb losses and mitigates the deleveraging pressure. So you can see in the right hand side chart that the blue line drops much, much less. So roughly 9% points less than the red line. So this illustrates the big impact of the quantity channel. So finally, we ask what kind of capital requirement rules can help you prevent that the quantity channel is present. And in proposition three, we derive such generalized rules. So policymakers can avoid the quantity channel with any time varying capital requirement rule that satisfies that the requirement never turns negative and it satisfies the following condition in all states. And this condition is fairly intuitive. So the requirement that you set for next period has to be lower than the current capital ratio of the banks plus the return on risk-weighted assets and the whole thing is adjusted for a growth rate factor. So G star is basically the desired long growth rate of the banks in the absence of the equitation constraint. And what's the intuition? It's basically the return on risk-weighted assets and long growth together give you the speed limits for how quickly banks can increase their capital ratio. So when profits are positive, a gradual buildup of the capital ratio and therefore also of the requirement is possible without constraining banks. But of course, if banks make losses, you need to release and reduce the requirement because there's basically downward pressure on the capital ratios of banks and to accommodate that, you need to reduce the requirement. So finally, we use our quantitative model to actually implement such a simple state-dependent rule that is consistent with Proposition 3. And what you can see illustrated in the chart is basically the transition dynamics from an economy where you have a constant 10% capital requirement to an economy where basically the capital requirement is varied between 10% and 15% depending on profitability. So when there are positive profits, their gradual continuous increases in the requirement when there are losses, the requirement is reduced. And the blue ranges show the percentile ranges across 100,000 simulations of the economy. And what you can see is that there are actually big gains at rather moderate costs of such a state-dependent capital requirement rule. So after five years, if you focus on the right-hand side chart, basically the far-right tail of credit drops is gone. So credit crunches or severe credit crunches are eliminated from the economy. And you do that at rather moderate costs because the average and upper percentiles of the long-growth distribution are barely affected. So they only reduce very minimally. So let me just provide a brief conclusion so I will not recap the findings but say what we think are the policy implications. So yesterday there was already a lot of mention of positive neutral CCYB and in our view the results that we derive in the paper clearly support such a macro-potential strategy because the cost of building buffers when banks make profits, if you do it gradually, should be very low. But at least you have something that you can release in times of stress to really give banks breathing space and continue supplying loans to the economy. So yeah, how and when should capital requirements be increased? Just do it when banks make profits. It's easily observable. How should you do it? You should do it gradually. So speed limits are given by return on assets and the long-growth. This should be super easy to implement and you will impose almost no costs on lending and the economy if you do it in the simple and transparent way. When should you release capital requirements? Clearly when banking sector makes losses. So recession is not enough. If banks still make positive profits in the current environment, profitability is very good. Not time to release at the moment. How should you do it immediately of sufficient size so banks can actually absorb the losses that they realize. But of course this is a bit harder to implement because you have this observation lag between losses actually being borne by the banks and then being reported. So of course some type of preemptive release might be needed. But I'll leave it there in the interest of time and thanks a lot for your attention and looking forward to Anatoly's discussion. Thanks Jan. So thank you to the organizers for inviting me to discuss this paper. It has been a pleasure to read it. So let me start with a very brief overview of what the paper does. So the authors developed a dynamic, a linear equilibrium model of the banking sector in a context in which banks face two occasionally binding constraints. One arising through a regulatory minimum capital requirement and the other one is by the assumption that banks cannot issue equity. And the results of the paper are that the effects of changes in capital requirements are state dependent. When no of the constraints binds, then the impact of the change in capital requirements is small. And when the both constraints bind, the impact is very large, is 100 times bigger. So this paper provides a very strong policy message going to the direction. It is good to build macro potential space when banks have ample capital headroom and they are doing strong profitability. This comes at a very low cost and has a very sizable potential benefit if losses realized later on in the future. So in my discussion, what I'm going to spend some time in describing the mechanisms and the calibrated results and then I will do some comments. So the model, this is a discrete time infinite horizon. You have a continuum of banks that issue one period loans that are subject to some risk. Some of the loans may default. That is aggregate risk. And this follows and there is some persistence in this aggregate risk. And there is also downward sloping demand for loans. Now banks are funded with some funds, with equity that is expensive and cheaper deposits. The banks face two constraints. They have to satisfy a minimum capital requirement and they cannot issue equity. They can only pay dividends. So essentially the model, the way I see it, is a sort of discrete time version of the Brunner, Meyer and Sannikov AR paper with the capital requirement. So what are the mechanisms in this paper? So let's think before on how this economy responds to shocks. And this is very much in line with what we know from Brunner, Meyer and Sannikov. So when there is a very large shock in this economy, banks are going to hit their capital requirement constraint. Now at that point, the lending supply in the economy is going to be determined by the bank capital. So shocks to the economy that are translating to equity losses for banks then are transmitted and amplified through the funding constraint. So there is amplification of losses when negative shocks are very large. Now banks anticipate that and they dislike hitting their constraints because they realize that when there are bad shocks, aggregate lending supply will be constrained and interest rates will be very high. So banks realize that they would be able to do a lot of profits during crisis and that's why they want to have voluntary buffers. There is a model in which banks endogenously have voluntary buffers in order to protect themselves from the implications of aggregate shocks. Now, given that they have voluntary buffers, if there is a small negative shock, it is not going to deplete entirely the voluntary buffer and this shock is going to be transmitted to the economy but it's not going to be amplified. So this is Brunner, Meyer and Sannikov. And what the main results of the paper are is just they don't look at the impact of shocks but they look at the impact of changes in capital requirements that one might think there are a shock to the banks as well. And the results are new but they are in a way a mirror image of the results in Brunner, Meyer and Sannikov. So when the changing capital requirements is done when banks don't have voluntary buffers, then the capital requirement is binding, lending supply is determined by the capital requirement and there is amplification. There is amplification of the changing capital requirements. When instead banks have ample voluntary buffers, the changing capital requirements is no amplified. So there is an impact but it is very small. So this part of the paper, the analytical part of the paper is very transparent and the analytical results are very intuitive and they're very persuasive. So let me get to my comments. My first comment is on how to place this paper in the context of the existing literature. So there are some quantitative microfinance papers that have occasionally binding financing constraints for the banks and that look at the optimality of state-dependent policy rules. And there are some papers that are even richer in the sense that they have also firms that are facing occasionally binding financing constraints. And let me mention the two papers that come to my mind. One is by one of my co-authors, Viagorta, and this paper, the interplay between banks and firms balance sheets is going to generate two types of resets when there are shocks. The other type of crisis, these are crises in which firms balance sheets are most hit and these crises have a very fast recovery. Then there can be also banking crises that are crises in which the bank's net worth is mostly hit and these crises have a particular feature that the recovery is very slow. And in that paper, Viagorta looks at optimal policies that have not changed the capital requirements. What is the optimal recapitalization rules in this environment? And it is also state-dependent. He shows that it is optimal to direct recapitalizations to the most hit sectors. There is also a paper by Vadim Levin and co-authors published in Econometrics. It's a very rich macrofinance model calibrated for the US and they look, the purpose of the paper is to look at the optimal liberation of capital requirements in the US and they show that cyclically adjusted capital requirements of state-dependent capital requirements dominate fixed capital requirements. So it's important to place the paper in the context of these contributions. My second comment is that the paper should strengthen its normative implications. So the way it is written so far, it has a positive approach to the analysis of capital regulation. And they look at what is the effect of changing capital requirements and the results are in line with what we knew from the literature of the impact of shocks in the presence of financial restrictions. So what is missing is a normative approach. So why are capital requirements? Why are there capital requirements in this economy and how to optimally set them? I would give some suggestions for the authors to try to improve the paper in this front. So a natural question that came to my mind is is the model equilibrium constrained efficient? If it is not constrained efficient, there would be a rationale for policy intervention and potentially for capital requirements. So notice that in this model, the bank's capital accumulation choices affect equilibrium in the loan market. And the equilibrium in the loan market in turn affects how much profits banks do during crisis and those profits enter the financing constraint. So it's the typical setup in which there could be pecuniary externalities. So there could be excessive leverage by the financial sector or too little. We don't know, it depends. And it would be interesting to explore that. Then another thing is that in the model, the authors assume that deposits are insured but still in equilibrium, the calibrated model, so that in equilibrium there is no bank default. Now bank default and deposit insurance is the traditional motive for regulation of the banking sector both in reality and in many models. So I would suggest the authors to allow for bank default that would give rationally for capital regulation in the model and that is also realistic. And at the very least, if that is too complicated, assume a sort of mean variance social preferences for bank lending so that the traders between maximizing aggregate lending and minimizing volatility of lending can be studied with a bit more of a structure. Then my last comment is that, I mean, the authors also push for a particular state-dependent rule for setting capital requirements. It was not presented in the presentation but the way that the proposed rule works is for given banks' capitalization, the capital requirement is set in order for banks to be able to accommodate average lending in the economy and still keep a buffer. With the banks under that rule, when the banks suffer losses, the capital requirement is reduced and when they make profits, the capital requirement is increased. Now, some questions that this type of proposed rule raises to my mind is that in the context of this model it's not so clear that this rule would be so different from the standard macro-publicial buffers that authorities have at their disposal. For example, a counter-cyclical capital buffer in this economy in which there is only one source of aggregate risk and which is the loan impermanent rate, when the loan impermanent rate is low, there are lending expansions and banks make profits. So a counter-cyclical capital buffer would push also towards increasing capital requirements. So the particular situation we are facing in many economies now that credit is contracting and banks are doing a lot of profits does not arise in this model. And this is the particular situation in which many authorities don't know what to do. So I think it would be necessary to introduce some additional shock to break the equivalence between the two policy rules. And to conclude, an important... I think the paper has the potential to speak to something we are very interested in. Vice President Deguindos in his keynote dinner speech concluded with the reference to the discussion between buffer-releasability and buffer-usability. This is a model, there are not so many models there, that has endogenous capital buffers. And the model has the potential to speak to how different are through the lens of macrofinance models usable buffers from releasable buffers. So I think this is something that you might want to explore. And let me conclude with this. Thank you. Thank you very much. And I totally saw lots of ideas, I think, to further work on the paper. We can take one or two questions. And I have Alastair, then there was Stefan. Okay, then here. But please try to be concise in your question. Thank you. Very interesting. But you can observe in the market today a 17% cost of equity being applied to European banks, so that you'd be interested how that would impact the model. And one of the things as a market participant I've observed is that if the central bank treats bank profits as its rather than as fundamentally according to shareholders in most cases, shareholders apply higher and higher discount rate to bank earnings. So what's actually happened is the cost of equity has risen in Europe as the capital levels have risen because the capital's been continually diverted. So, you know, that 17%, I'm interested how your model would work if you increase the cost of equity as you capture the capital, which is observably what's happened. And I'd expect it would be less beneficial. Thank you. So thanks a lot for that presentation. In a sense, it's a very regulatory perspective on the banking system and the role. And I think it's great to have the distinction, no, the state dependence there. But one thought I had was what about changing over time kind of uncertainty? I guess if you have kind of you're in a regime of high uncertainty and you have your banking system and you have your regulations in place and you expect in the future it goes down, the uncertainty goes down or the opposite. Can you say anything now ready or is this a next paper? I guess it's half similar. I will have two comments. The first one, I think you assume that somehow the return on equity gets adjusted as the requirement increases. I think in the real world a lot of the loans are 30-year mortgages. So in reality, it has increased the requirement the return on equity goes down and that's what's happened. You have other businesses which are not able to reprise, which are not loan-based. So in practical terms, the return on equity goes down. As an equity investor, your capital gets trapped in the business because you think you are getting let's say 8 or 17 or whatever you think you should be getting but in reality your remuneration is very, very long. And in the real world, the release of requirements doesn't work because in a crisis nobody wants to be running. If the standard is 15, nobody wants to be running with 10 because basically they will get destroyed by the fixing of market A and B because everybody knows they will have to go back to 15. So in practical terms, nobody is willing to go down to 10 even if technically they could, which is what happened with COVID. So the release doesn't work in practical terms. It only goes up and up and up but there are no step-backs. Thank you very much. Just a question about the timing of capital requirement announcement. So you know that the CCYB is usually announced one year in advance. So I was wondering whether a timeliness and unexpected announcements would change the impact of capital requirements. And would banks price in these early or would it give like a mitigated amount on impact and how does the level of competition or concentration impacts your results? Thank you. Thanks very much. So please proceed. Thanks a lot for the discussion and also for the interesting questions. Now I think you did a really good summary of the paper and indeed I think Bonamaya Zanikov is closely related and we really see our contribution in focusing on the capital requirement and clearly spelling out the state dependent transmission channels of changes in requirements. So indeed I mean the modeling framework itself is not entirely new but I think the perspective of how we use it adds valuable insights especially for the policy debate. So I mean your first related comment would be placing it into the literature and they fully agree we can still do a better job to clearly specify what we do differently and what we focus on differently. In terms of sort of more normative implications these are definitely valuable suggestions and we thought of some of them and I mean we need to see what's actually realistic to model because one of the key strengths of our current paper at least as we see it is that we're actually able to derive very clear analytical results and in our view that actually helps a lot to have a transparent debate if you don't always need to resort to numerical solutions. So we will explore what we can actually add and still stay within let's say a clean and neat framework. Then maybe on your final comment on the relation to the existing CCYB rules they actually disagree because in my view existing CCYB rules depend on a measure of cyclical risk. So you need to first estimate something that is unobservable. Our proposed rule is only based on observable variables capital ratio and profits and it's a completely different approach. You basically say don't care about risks because if you increase gradually and slowly and predictably capital requirements when times are good the costs are very low. I mean you want to err on the side of caution because you don't do much harm by having let's say one or two percentage points of buffer in addition and I think therefore this type of rule deviates quite a lot from this risk based perspective where you say no, no, it's really costly to increase the requirements and therefore you really want to be sure that you need them for high risks. But in my view this type of approach has not really proven that impactful and effective in reality. Some of people who have seen David's talk yesterday would speak also in that direction. Then I mean regarding the questions so indeed I mean cost of equity estimates have gone up also the cost of debt has gone up so what really matters in the model is the difference between the two because that's basically the additional cost that you impose on banks. Of course if you assume that this sort of equity premium goes up this will increase the pricing channel impact but even if you increase that by 50% the equity premium your three basis point funding cost impact will go to 4.5. So in my view of course it will have an impact it will make it more costly but it will not lead to a different order of magnitude compared to what we have shown and then actually I mean my view would be the following what we actually show here is that being simple and transparent and it's not that we're taking away profits from banks the equity stays in banks and even with our rule we basically ensure always that banks can pay dividends even in crisis times because the release will always be bigger than actually what the hit to bank capital ratios would be so the rule we implement in the model allows banks basically to continuously pay out dividends we simply impose that they gradually need to increase requirements ratios when times are good On Stefan's question regarding uncertainty I don't really have a good answer I haven't thought about that part there was a question on potential ineffectiveness of release in times of stress so I didn't mention it here but I think what's important to keep in mind that what I've shown you only works if the market imposed leverage limit or capital requirement is below what the regulator sets so of course if let's say in good times your requirement is not high enough so that actually the losses that banks incur in bad times would force the regulator to reduce capital requirements below a market imposed capital requirement then no effect of release there I fully agree but the flip side of that is in the good times you need to increase the requirement way above what's market imposed so that in bad times you can reduce the requirements sufficiently that you're still above a market imposed leverage requirement final question on timing so in our model the CCYB or the requirement is announced one period in advance and actually what it does is given that banks anticipate this and if you have a clear rule it actually will lead to lower voluntary buffers because there's less self-insurance mechanism in the model because the regulator takes out the insurance motive to some extent because they reduce the requirements okay I'll leave it there thanks very much Jan so I saw a number of hands still up so maybe you can use the coffee break which will start now I would like to thank the presenter and the discussant very much and then we resume at 11.15