 Okay, welcome to the second session. I don't think I need to present the presenters. And the discussions were known. I don't think neither that I have to repeat the rules, though I think they must have been said so as to share the one hour we have and include a discussion, but I was asked to repeat that you should always speak into the speakers because there may be a YouTube film movie made out of this conference and therefore apparently people don't remember this detail. Okay, then let's start, Avin. Thank you very much for having me here. It's a great pleasure to be here. So this is work I've been doing with Adam Gurran and Tim McQuaid, Tim is my colleague at Stanford. And as the title suggests, it's about mortgage design. So before I go into the details of the model, I just wanted to maybe spend one slide just motivating why I think this is an important topic. So I think we all probably understand, you know, certainly from the recent crisis, if not from past crisis, that high levels of debt are a significant source of an amplifier during financial downturns. And when we think debt, again from the recent experience, it's debt both in the banking sector as well as the household sector. I have made a slide here where I'm drawing the contrast between how policy has thought about banks and households when it comes to debt. So if you just think about the macro-prudential tools that we have to deal with high levels of debt in banks versus households, there's actually quite a big contrast. So take banks. You know, banks get in trouble with high levels of debt. Exposed, there's a collection of tools to deal with that. So central banks can provide liquidity against high levels of debt. Governments have certainly in recent years recapitalized banks, so there's ways to deal with debt directly through government intervention, exposed. And then one of the things that we have seen post-crisis in terms of an ex-ante, you know, pre-crisis, is higher capital requirements, reducing debt levels in the banking system, as well as innovations in contract design. What do I mean by contract design here? I mean things like securities that convert into equity during bad times. All of this kind of increases the resilience of the banking sectors, high levels of debt during downturns. Now, if you look at the household side, on the other hand, my sense, and this is what I'm going to tell you, is that far less is done and has been done and is capable of being done. So first of all, if I think about the U.S. experience exposed with high levels of household debt, my read on the studies of mortgage modifications in the crisis is that they were largely unsuccessful. Institutional rigidities in the way in which modifications were pushed through meant that what happened was much, much, much less in terms of debt relief than was equivalently available, say, to banks. So there's little exposed ways of dealing with household debt. So think about ex-ante. Well, one of the things that has happened since crisis is, say, increased LTB ratios, reducing debt levels going into crisis, and that is a way of dealing with debt. But it's different when you think about households than when you think about banks, because, of course, households don't raise equity. So you can't just impose a high LTB constraint on a household and say that the household raised equity the way you would with a bank. The household has to finance it internally, and that immediately means that housing demand will be affected. So that means to me that the lowest-hanging fruit here that hasn't really been studied is contract design on the household side. And so that's just a pitch for why I think this is an interesting and important issue, and that's what I'm going to take you through. Okay, so with that, this paper is about mortgage design. And so it's about contract design. And the question for the paper is, how does different forms of indexation translate into household behavior, consumption behavior, default behavior, as well as home price behavior? So that's the question for this paper. Now, at kind of just a basic theoretical level, it's fairly obvious, I suspect, to everyone in this room, that adding state contingencies to mortgages must enhance welfare. If we think about the, in the US we have the standard mortgage, it's this 30-year long-term fixed-rate mortgage. So it just seems obvious that putting some state contingency there has to benefit things. And many people have proposed different forms of state contingencies, equity, indexation to home price indices, and other forms of indexation. So it sort of seems obvious that that should be beneficial. There's some questions about why that hasn't been done, and those would be theoretical questions. I'm not going to approach that. So the question for this paper is going to be, indexation is clearly should be beneficial. So how much, how beneficial might different forms of indexation be? So that's what I'm going to do. I'm going to take a model that is a reasonable representation of households, of the housing market, and I'm going to just throw in different mortgage contract designs, and I'm going to compute for you welfare benefits of different contract designs, and I'll try to rank them. That's the exercise in this paper. The contract designs I'm going to look at are all contract designs where indexation is off of interest rates. So monetary policy varies with the state. So a natural way of thinking about indexation might be to index off of monetary policy. So that's what I'm going to do. I'm going to study contract designs where the state contingency is all coming through indexing mortgage payments in different ways to interest rates, and I'll rank them as I said. I'm not going to go out and do things like principal indexation, say linked to home prices. We have looked at that, but I'm not going to show you those results in what I have today. So the exercise of this paper is I'm going to build a model. It's a heterogeneous agent macro model. I'll take you through it. It's kind of what you'd expect. It's a life cycle model of people buying homes, earning income, et cetera. We sort of build it in a way that we think we can reasonably, realistically represent household behavior. The model is mostly in partial equilibrium. There's one market which we'll close in general equilibrium, and that is the housing market. And the reason we're going to close that market is because we'd like to allow the model to speak about home price spillovers. So when people foreclose their default, that affects home prices, which then affects consumption, default, et cetera. So that's the only equilibrium object that we're going to have in there. As I said, we'll calibrate this in a way that we think is reasonable. I'll talk about it as I go through. And then what I'm going to show you is, as I said, a collection of quantitative simulations, and the simulations are going to be as following. I'm going to take the US downturn 2007-2009. I'm going to assume that what happened in 2007-2009 was coming from a world where everybody had long-term fixed-rate mortgages. And then I'm going to rerun that where I throw in a different mortgage design. So I'll just throw in a collection of different mortgage designs and I'll compute out how much better households would be under different mortgage designs. And I'll give you some numbers and we'll learn something about how mortgage designs can affect welfare. So there's really two main results that I'm going to try to show you today. The first one, and these are all really quantitative statements. The first one is that if you sort among different types of mortgage design, the best forms of indexation have this feature that they reduce, they relax budget constraints, particularly for young homeowners. So people who bought homes recently who are relatively constrained. So their savings is low. They've bought a home and they're kind of up against payment constraints. An indexation, a form of indexation that relaxes those payment constraints can deliver big welfare benefits. So it's going to be, you know, there are different forms of indexation. Almost all forms of state contingent indexation will deliver welfare benefits. But if you sort among different forms of indexation, this one seems to do a particularly good job. And I'll give you some intuition for why that's working and I'll show you some numbers about that. Okay, so that's one result that I'll try to stress in this presentation. And then another thing that I'm going to try to stress is, as I said, the state contingency in the set of mortgages that I'm going to consider here are all through monetary policy. So monetary policy state contingent, so we'll piggyback off of that in thinking about indexation. And I'll show you how different forms, how monetary policy interacts with the benefits of indexation. So for example, QE policies can, will affect how you think about different mortgage contracts because they'll affect the long end of the yield curve. And so mortgage mortgage contracts that are more pinned to the long end of the yield curve will have benefits tied to QE policy. So there's an interaction with monetary policy that I'll also try to explain. Okay, let me just go straight to the model. So here's the model. There's actually, I'm not going to show you any equations. So Monica showed you a few equations. Pierre showed you more equations. I will show you zero equations. But I think you can actually all sort of fill in the pieces in your head as I talk this through. Okay, so what is the model? It's, as I said, we're trying to build a model that is a realistic representation of a household that is going through their life cycle, making decisions about when to take on a mortgage, when to buy a house, when to pay down a mortgage, the set of collection of questions that you would think of as being relevant for mortgages. So the model is an OLG model. Agents in this world are going to live for 45 periods. They're going to earn income for 36 periods. As they go through their 36 periods of earning income, they're going to have a life cycle income profile that we're going to match in a way that is to kind of the best of what we know about life cycle income. Then they're going to hit a retirement period. So as they go through this, they're going to also accumulate housing. So at some point in their life, they'll choose to want to buy a house. They'll get some benefits from housing consumption. They could also choose to rent, but they'd get lower benefits from renting a house. So at some point they'll want to buy a house. When they buy a house, they'll want to take a mortgage. The mortgages that I'm going to consider are these mortgage designs. So just to explain this, let me talk about a long-term fixed-rate mortgage. At some point they'll want to buy a house. They'll lock in a long-term fixed-rate mortgage when they buy the house. That purchase of the house and the taking of this long-term fixed-rate mortgage will be subject to a collateral constraint. So there's an LTV constraint that a bank is going to impose on the household when they're making this loan. With the long-term fixed-rate mortgage, there'll be chances for the household, for example, to move, to prepay their mortgage, to refinance. So all of the sort of stuff that we think of in the world will be available to our household in this world. There's uncertainty in the world. The uncertainty affects a couple of different things. The first thing I'm going to talk about is just income. So as I said, households are going through their life, earning a life cycle income profile. Income is stochastic though, and it's stochastic in a couple of different dimensions. First of all, there's three aggregate states, an expansion, a recession, or a crisis. And so those states will also index income. So there's high income, low income, very low income. There's also an idiocycratic income shock in which particularly in the crisis state, income is going to be more left skewed. So a given individual could get a very bad income shock, basically gets unemployed, in which case income falls a lot. And that will also vary according to this aggregate state. So there's an aggregate level effect on income and there's an idiocycratic component to income that's going to move across states. All of this is in keeping with what I understand about the income processes for households. And as I said, we're going to calibrate this in a way that matches the world fairly well. So that's the household income profile. There's also two other aggregate shocks. The other aggregate shock is a credit supply shock. So banks are going to supply credit in this market. These mortgages that households are going to take. And the credit is going to be supplied subject to an LTV constraint. And I'm going to allow the LTV constraint to be either high or low. So in the model, in the calibrations could be either 95% or 80%. And that will move around stochastically. So the aggregate states are expansion recession crisis as well as this credit supply either high or credit supply low. So with that, that kind of describes the aggregate states. Monetary policy is just a set of numbers. For each of these states, I can say an interest rate. So in the expansion state, there's an interest rate. In each of these states, there's an interest rate. That sets the short rate. That short rate then feeds indirectly into mortgage rates. How does it feed indirectly into mortgage rates? It's basically off of a bank present value condition. A bank is making a loan in the mortgage market. They can fund it using the present value of interest payments. And they're just going to take a PV operator on their mortgage loans. And so that will set the mortgage rate subject to the bank making a minimum profit condition. And as we change mortgage designs, we're just going to do this present value operator for different payment streams and get the same present value profit for the bank. So that's the banking side. As I said, so agents go through their life. They take mortgages. They make payments. They mortgage amortize over their life to refinance. They might move. So in this model, people are going to get moving shocks. So they're going to get case shocks that make them sort of separate from their current home and move into a new home. And so there's going to be some churn in the housing market. All of this is kind of what you'd hope to put into a model. So it'll all be in there. And so that's the mortgage side. And as I said, there's two constraints here. So there's an LTV constraint when you buy a home that the bank is going to impose on you. And then the last thing I should emphasize is that the only thing that you can borrow against is your home. So households here are going to be liquidity constraints. They aren't going to be able to borrow against future income. They will be able to borrow against home. So home has this collateral value. So that is the model. I'm not going to show you any equations. Hopefully I've given you enough of a sense as to what is in the model that most of you can kind of fill in the details. There's two equilibrium objects in this model. So as I said, the model is mostly partial equilibrium in the sense income processes I'm just going to throw in exogenously. We're just choosing them to match lifecycle income profiles. There's only two equilibrium objects. The first one is home prices. So in the model, there's going to be a housing market clearing condition every period. There's going to be a bunch of guys who are going to be buying houses. They're mostly young guys in this model. Young guys who are transiting from renting to home ownership. Or there's going to be a small group of buyers who are going to be buyers who were foreclosed on previously or were defaulted previously who will have their defaults cured over some time and then they'll come back and they'll buy some houses. So that's the housing purchase side of the market. And then the housing supply side, who sells houses here? Well, it's a combination of things. It's old people who are getting out of housing. And then it's some financially distressed sellers. So households are defaulting and banks were foreclosing on homes. There's going to be some of that. And then there's going to be some movers who are churning the housing market as well. So that's the housing market equilibrium. As I've described this model, you can kind of see that at every point in time there are 45 cohorts of households. So the state space here is very large. So we have this usual very large state variable problem. And how are we going to solve it? We're going to solve it. And I guess what is now the standard way in the literature, we're going to follow this Paul Smith approach of sort of specifying a low dimensional function for how prices are going to move over time, just as a function of a few state variables. And we will iterate the model and converge on what that price function that best represents the evolution of prices is. And then you can do a bunch of checks, for example, a standard check that you want to do is we're going to be guessing a low dimensional forecast rule that's one period ahead of time. And then you can check you can do with whether that's a good representation of the forecast rule is see whether your forecast rule works 10 periods ahead. So forecast 10 years out with that one period forecast rule, does it actually match the simulation? And so we checked all of those things and the forecast rule generally works well. You can look at the paper for how well this does. Okay, so that's the housing market. And I mentioned this already. The other equilibrium object I should have mentioned is the mortgage spread. So how banks are breaking even on giving mortgage contracts. They set an interest rate on their mortgage contract based upon this break even condition. What we're going to do is we will, as we change mortgage designs, we'll just ensure that banks make the same profit. Okay, so that keeps all the mortgage designs on sort of the same footing. Okay, so calibration. We tried really hard to do a good job in the calibration and there's really two conceptual parts of this calibration. One part of it is just representing household behavior in terms of levels well. So match income profiles well. Match distributions of things like LTVs of wealth. Match those well in the data. Okay, so that's about just matching the levels. That's important because, for example, the main simulation I'm going to show you today is I'm going to rerun the 2007-2009 crisis with different mortgage designs and it's important to match the LTV distribution of the world in that exercise because depending upon the LTV distribution, a given set of shocks will have bigger or smaller effects. So we want to match that level fairly well and we do. The other thing that we're going to match is we're going to match what I'm going to call a slope. So as I've mentioned to you, the equilibrium aspect of this model is there's going to be a feedback between home prices, consumption, and default. So there's an equilibrium home price feedback mechanism and we're going to try to match that based upon fairly nice quasi-experimental evidence. So there are these nice papers by Fuster and Wollin who are able to isolate what the impact on different households who have different levels of, say, temporarily low income or temporarily tight budgets and how that impacts their default behavior. So that is about how, if I was to reduce someone's income, how they would change their default behavior. So we have nice experimental evidence from that. We can replicate that within our model. That is, we can sort of simulate their experiments in our model and we'll replicate that. So what does that mean? We've replicated now sort of a level of how households behave as well as how they're behaving with regards to temporary declines in their income or with increases in their payments. So with that, that's the calibration. The rest of it is fairly standard. As I said, there are these aggregate states, expansions, recessions, crises. There's interest rates. I'm going to specify through that so we can just simulate this model out as we go forward. All right. So I'm going to show you mainly one set of counterfactuals. So what I'm going to take you through is, as I said, I'm going to rerun the 2007-2009 crisis. This means I'm going to feed in income shocks that look like income shocks in the world. I'm going to tighten the LTV constraint from 95% to 80%. So I'm going to tighten the LTV constraint. That's going to decrease housing demand and push down home prices. And I'll show you a slide in the next picture, because those two things kind of reasonably represents what happened to the US 2007-2009. All right. That simulation that I'm going to show you is under the assumption that everybody in the world has a fixed-rate mortgage, a long-term fixed-rate mortgage. Then I'm going to do a collection of exercises. So the first exercise I'm going to do is I'm going to take everyone in the world. I'm going to swap out their long-term fixed-rate mortgage. And I'm going to imagine that at the beginning of the crisis, someone could sort of wave a magic wand and put everybody into an adjustable-rate mortgage. Why adjustable-rate mortgage? Because interest rates fall a lot as we go into the recession. And so there'll be some reduction in payments off of that. And I'll show you how much that benefits households. So I'll show you that. And then I'll do other exercises where I allow people to choose adjustable-rate mortgages ex ante before a crisis. And I'll show you how that impacts welfare. And then I'll show you other mortgage designs. And there's more mortgage designs in the paper that I'm not going to show you. First of all, this is just a check. As I said, I'm just going to simulate out the crisis with a tightening LTV constraint and a reduction in income. I'm plotting here. These are home prices. I'm going to suffer what everyone else... Okay, no, I have a red line here. So that's home prices. Year zero is the crisis. And then this is in terms of year. So home prices fall and then go back up. If you look at this fall, that actually about 30% fall is about what happened in the crisis. This is the share of households with negative equity. You can look at what happened in the world. That's actually quite close. This is... These are default rates. These are also quite close to what happens in the world. And this is consumption, air consumption of the households, which also is close to what happens in this world. I want you to think of this as a baseline. What I'm most interested in doing is I'm going to throw you... I'm going to add a different mortgage design. And I'll show you how these numbers look relative to this baseline. And then we'll compare gaps of how beneficial different mortgage designs might be. All right, so here is the magic one. So imagine that you were in 2007, so year zero of this crisis, and you could take everybody out of their fixed-rate mortgage and put everyone into an adjustable-rate mortgage. All right, what's the difference? Well, the fixed-rate mortgage... Long rates on the fixed-rate mortgage are down about 70 basis points. So long-term rates, because short-term rates fall in this model, long-term rates will also fall just via expectations hypothesis by about 70 basis points. So in the fixed-rate mortgage world, long-term fixed-rate mortgages are lower. However, households are going to be subject to a refinancing constraint, so they aren't necessarily going to be able to take advantage of that 70 basis point. So that shows up in that baseline dynamic that I showed you. So what I'm doing when I put people in adjustable-rate mortgages is short rates fall by about 3%. So a big decline in short rates, and it's effectively an automatic refinancing. Everybody can take benefit of that, because if my mortgage was indexed to short rates fell 3%, everybody who has an adjustable-rate mortgage will benefit from that. Rather than in the fixed-rate mortgage world, only guys who could refinance would benefit from that. So that's the experiment I'm doing here, and you can see here it has a beneficial impact, no surprise. House prices, the blue line here is the baseline, the purple line here is under this adjustable-rate mortgage, and again this is with short rates falling 300 basis points in the crisis. There's less default, prices fall less, and consumption falls less. Okay, kind of about what you'd expect. I'll translate these numbers in terms of consumption equivalent welfare in a few slides. What is happening here, it's mostly about young households, and you can see this by looking at default rates. So the blue line here is the fixed-rate mortgage, the red line here is this adjustable-rate mortgage, which I gave people in the crisis. The difference here is the default rates, the yellow line is the difference in default rates, and you can see here this is in terms of age of the model, so 10 years is you've been earning income for 10 years, and those are the guys, these guys who have been earning income for 10 years, and guys who have been earning income for about 20 years, these are guys who are at different stages of the life cycle who bought homes, these are the guys who really default less. That's one big impact of the adjustable-rate mortgage, and then the other impact is rental. So young guys who previously wore renting are going to be more inclined to buy because the payment that they face with an adjustable-rate mortgage is much lower than the payment they would pay for the fixed-rate mortgage. The fixed-rate mortgage rates are down 70 basis points, the adjustable-rate mortgage rate is down by 300 basis points. So that's really what's happening. The second exercise I'm going to do is I'm going to allow people to optimize ex-ante. So what I just did before was I just gave people adjustable-rate mortgages in the crisis. Now I'm just going to allow people ex-ante to choose between adjustable and fixed and play out the world as it does. And the adjustable-rate mortgage is still beneficial. It's not quite as beneficial. It's about a third less beneficial, and the reason it's about a third less beneficial is because what happens is with adjustable-rate mortgages, people anticipate that there's going to be insurance in the crisis. So they know that their payments are going to fall in the crisis, the adjustable-rate mortgage. So they lever up ex-ante. This is the leverage distribution by LTV with the adjustable-rate mortgage in orange and the fixed-rate mortgage in blue. And you can see this adjustable-rate mortgage makes people kind of push out further out. They lever up more because they know that their payments are going to fall, and that undoes some of the benefits. Again, this is kind of what you'd expect. This does suggest, by the way, that contract design could be well-paired and LTV-constrained, so sort of mixing macro-prudential tools with mortgage design and LTV-constrained could deliver all of the benefits that I showed you before by, say, replicating the blue line here. All right. As I said, the big benefit here is really coming from young homeowners, and you can see this up in this top panel here. The top panel, what I'm graphing is, I'm graphing by age the welfare benefit of the switch from the fixed-rate mortgage to the adjustable-rate mortgage. The welfare benefit here is computed as, suppose you're in 2007, you're going into the crisis, and I was to ask you, how much would you pay in an annual consumption benefit to go into the adjustable-rate mortgage for the rest of your life? That's the number I'm computing. And across all of our households here, the answer is 1.27% of annual consumption. So I'm going into a crisis, and I tell people, you know, you were in a fixed-rate mortgage. I could throw you into an adjustable-rate mortgage. How much would you benefit from that in consumption-equivalent terms? And the answer is I'll benefit with that at 1.27% of annual consumption per year for the rest of my life. Now, I shouldn't store it. This is conditional on the crisis. So a crisis is hitting, so I'm giving you a maximum benefit with a crisis hitting for sure. We also, in the paper, do a steady-state analysis, okay, so I'm also graphing for you those same percentages by a type of agent, and in particular by age and by whether an agent is a homeowner or a renter. And you can see here the real big benefit here, which is this orange line down here. I don't know if you can even see these numbers. This is minus 3, minus 4%. There's about a 3%, 4% welfare benefit off of a young homeowner by switching from the fixed to the adjustable. So if you go later on in an agent's life cycle, the benefits are considerably less. So it's really about young guys, and I'll show you why that is over the next two slides. Okay, let me skip this one. All right, so the next thing that I'm going to do is I'm going to show you another mortgage design, which is a mortgage design that is a fixed-rate mortgage but allows you to do underwater refinancing. Why am I doing this? Because the comparison I just did between a fixed-rate mortgage that we have in the world and adjustable-rate mortgage has the adjustable-rate mortgage beating the fixed-rate mortgage in two ways. It is an automatic refinancing mortgage. It automatically takes advantage of low interest rates as they come into the recession. The second thing that it does is it front-loads payment benefits because short rates fall more than long rates. So I'm showing you now a mortgage design that allows you automatic refinancing to a long-term mortgage rate. So imagine we could replay the U.S. experiment where many people went into negative equity territory but allowed them to fully benefit from the drop in long-term mortgage rates. So that's the exercise that I'm going to do. And so here is result one. It basically doesn't help. And when we first did this computation, it was surprising to us. So long-term mortgage rates are falling by 70 basis points. Agents are refinancing to take advantage of it. New homeowners are benefiting from the lower mortgage rate, but yet it really doesn't help anything. So this is the baseline. You can barely see the gap between the baseline and this new mortgage where people can refinance in an underwater mortgage. But it doesn't matter very much. And in welfare equivalent terms, there's a 0.13% consumption equivalent welfare gain compared to the 1.3 I was showing you before. And the reason for this is because the long-term mortgage is priced off of the long end of the yield curve. The long end of the yield curve falls by 70 basis points. The short end of the yield curve falls by 300 basis points. The agents who are really benefiting in this economy from our calibration are young liquidity constrained agents. So fixing the present value of transfers of state contingency that you give agents, it's always better to give it to them front-loaded rather than over the life of a mortgage. That's exactly what an adjustable rate mortgage does. It reduces payments in the recession, which benefits someone who's liquidity constrained tremendously, whereas refinancing to a lower fixed rate is beneficial, but it spreads those benefits over the entire life of the mortgage. That's not going to be as beneficial. I mean, you can kind of see that, but what is most surprising is it almost has no benefit. Almost all of the benefit is coming off of the welfare benefit of targeting consumption relief, or really what I think of this is budget constrained relief to households who are young and relatively new homeowners. So that is a result that we were surprised by, but really comes out very clearly once you start looking at this. The second thing that we did was we said, well, long end of the yield curve can benefit from QE. So long-term rates in the US came down in part because the Fed bought a whole bunch of mortgages. That is something that you would expect to benefit the long-end, the long-term mortgage design, rather than a mortgage design that is keyed off of an adjustable rate mortgage. And so we just went out and said, well, suppose you were able to lower long rates and additional 100 basis points below the expectations hypothesis long-term rate. How good is that? And the answer is it's good, but it's only still a welfare benefit of about 33 basis points, 0.33% compared to the 1.3% off of just the simple indexation to adjustable rate mortgages. So this is a way of saying that QE is useful, but the real benefit is cutting current payments. We do other stuff in the paper. I'm just going to show you one last exercise because I think I am running out of time or over time. And the last exercise I'm going to do is I've basically made a case for adjustable rate mortgages. That's really what I've made a case for. Now, as many people are probably aware, adjustable rate mortgages, the way they're working in this model is because interest rates are correlated with income. And in the downturn, income falls and interest rates fall. You can imagine periods, as we went through in the late 70s and the 80s, when that correlation flips. And you get high interest rates as income is falling. In that case, an adjustable mortgage is going to do quite badly. So we study one other mortgage design. This is a mortgage design that I, Jan, Eberli, and I wrote up in a Brookings paper a few years back, which is a fixed rate mortgage with the option to convert to an adjustable rate mortgage. So it's a standard fixed rate mortgage where you have a one-time option to convert yourself into an adjustable rate mortgage. It looks a little exotic, but what it allows you to do is it puts a cap on the interest rate because if interest rates go up, as they did in the 80s in the Volcker episode, you wouldn't convert. You would just keep your low fixed rate mortgage. So it has the flavor of, if you like your mortgage, you can keep it. But if interest rates fall, you can take benefit of this. So I'll show you what that does. So here's the benefit of that in the exercise I've just done, the exercise being the rerun the crisis. And the benefit of this E.K. mortgage is 0.92% in consumption equivalent welfare again compared to 1.3%. So it's not quite as good. Why is it not quite as good, basically, because banks are charging you for it? So you're not quite as happy. But we're running that in an experiment where we're just rerunning the U.S. crisis where income is falling and interest rates are falling. We also do another exercise where we resimulate the 1980s recession, the Volcker recession, where you would think that this would really benefit you and the answer is it does. The adjustable rate mortgage loses 1.34% in terms of welfare, whereas this cap mortgage only loses a half percent. So that gives you a sense as to whether this might be beneficial. There's more in the paper. We play around with monetary policy and there's a feedback here clearly between monetary policy and the benefit of a different mortgage design. We do this a couple different ways. I'm not going to show you all of that because I'm out of time. I think the main thing I wanted to say is here, at the first order point, I really think mortgage design is a huge first order effect of importance to welfare. So it's worth studying. We have studied designs which are indexed to interest rates. They seemed like the easiest ones both from the standpoint of understanding within a model as well as in terms of implementing. That is, they're within the space of current contracts. So they seem like relevant ones to look at. And if you want to know what I've learned so far from this, it's that mortgage designs that frontload payment relief are really what look like they deliver welfare benefits. Okay, so thank you very much. Thank you. Okay, well, as a previous discussion, I should start by saying that I probably know the most natural choice that we're discussing for this paper because much of what I know is on the corporate side knows the mortgages. I also don't spend my time doing equilibrium models. But now the laser paper was very interesting because of course it has a very important question at heart and Arvind didn't even need to use equations nor will I. So the starting point here is that build up of leverage in the economy is what makes a propagation of the crisis into the real economy. This is the key piece that facilitates the propagation of the crisis from the shock into the real economy. And so specifically focused here will be on the household debt and accumulation of debt in that sector. Now, once we acknowledge that point, of course it would be nice to have policy tools that allow us to act effectively on this outstanding debt during the crisis. Now, there are two ways to do that. There is exposed and exalted ways of dealing with debt overhang. Now, from what we've learned in the years following the crisis through different programs that we are trying to facilitate refinancings of the mortgages and generally easing the burden of the debt and the household sector is that the programs were not very effective and it seems that there are important implementation frictions. So what Arvind and his co-authors do here, they say, all right, well let's take a stab back and if we design contracts ex ante that already have features that connect contracts to the monetary policy, then that could be a solution that helps us out. Now, with that question in mind of thinking about what is a type of contract that would help us to maximize welfare before the crisis as well as help effectively to deal with the problem that once we run into the crisis. So this is the question about an intensive margin here. We are thinking about the debt that is already outstanding. So that question, once we articulate the question, the answer actually, as Arvind pointed out, is rather trivial and we don't need, the intuition for it is very simple. Of course, if you're going to introduce this element of contingency into these contracts and allow, especially if you give this option of conversion during the crisis into the variable rate, then adjustment conditional on the monetary policy being the ones at lower interest rates, then that would translate into lower payments for the household and that's a very effective way of dealing with that. So that's good news, but that's not where I see the main insight. The way to think about this paper, what it provides, the main contribution of this paper is that it provides a quantitative model that ties together consumption, defaults on housing prices, as well as has this element of realistic features over mortgage that, and this model allows us to throw different mortgage designs and come out with quantitative measures for thinking about which contractual features are the best ones. So there are two ways, given that this is a main contribution, this is a quantitative tool that is provided to us for thinking about mortgage design and affecting some more interior policies through this mortgage design. So to me, there are two ways to think about this paper. First of all, I think that the paper is really nice and it certainly satisfies the wish list of the welcoming speech of what the model should deliver, but given the job of the discussion, one way to think about it is currently the model delivers a large economic effect that favors adjustable rate mortgages and so one thing is to think about what are the assumptions that are wired into this model and what are the things that are missing that make this result large and do we believe in them? Probably we will, but nevertheless I'll highlight a couple. And the second point is, okay, well, worst-case scenario doesn't work. Worst-case scenario, the effect is economically small. What is the damage that we are doing by introducing this policy and by shifting the household towards the adjustable rate mortgages? All right. One thing that Irene didn't emphasize that much, but much of the discussion of the potential damage is the fact, of course, that by hardwiring the relationship between the monetary policy and mortgages and counting on the fact that interest rates will go down, there is this element, well, what if the interest rates actually go up? And one thing that you might consider to add more color to the discussions that you already have is that not every recession, of course, has the same implications for how you would be cutting rates. Right now there is just a view, well, they either go up or they go down and I think that there is elements that you can reflect on that. But there are other points and some of them are there in the paper, but for example, the first one is if people are shifted in the aggressive, in the adjustable rate mortgages and there is the initial problem is which you started where we're trying to solve macro fragility actually can be aggravated because everybody in expectation of that there will be benefit of low interest rate and downturns will increase the leverage. Now it's present as one of the analysis where they blend the adjustable rate mortgage and fixed rate mortgage, but if you take it to the extreme of policy actions as this could become a very important point. Now the other thing that is not sought through in the way in the model is okay well who carries the risk and currently the banking sector is this present value solving agent, risk neutral agent and it assumes no frictions and nothing else in the portfolio on the size of bank. And the question is well, if perhaps part of the answer as to why these products are not predominant in the current environment could be tied to the fact that something else could be moving on the side of the bank balance sheet and it seems like thinking in that direction might be something relevant. Now one other thing that is missing and perhaps it's worth reflecting about is that introduction of the adjustable rate mortgages or anything that layers up options than before or after it makes the product less tractable. I mean fixed rate mortgage easily understood by most of the household once we make the contract fancier it seems to me that we will definitely need a little bit of a tighter oversight of this product which is not impossible but definitely not free. Now on the size effects so this is as to what is the downside what if the upside is not that large what damage are we making. In terms of the size of the effect so currently we're saying okay well arms are winning and let's reflect on a couple of assumptions that are there that delivers that result. So first point is what about unconventional monetary policy. So the entire paper is the premise of this paper is based on this graph it's a tight relationship between the short term rates and the reference rate which is a prime on library and so you can see from this picture that most of the times they are very very tightly related in fact they always very tightly related and this picture already points out to the effectiveness of arm in the period because since 2009 this reference rates are flat right. So to me the way they approach this quantitative easing tools is a little bit of cheating because the three scenarios that evaluate they start with the fact that as well there is a lower short term rate by 100 bps plus something on the long term action and once you remove that short term rate 100 bps then really there is there is nothing that working that much towards the arms so it seems like the arms in this context might be a little bit overstated from that perspective and to add one point on the corporate side so the corporate side there is a senior secure credit is variable rate which is what actually arm in the context of the paper means here and so there is a paper that follows what happens to the outstanding balance of debt through the fact that senior secured debt is a variable rate for corporate sector. The conclusion here is that in fact this paper uses the quantitative easing period and the fact that the short term rate doesn't move any longer at the point to contrast that it loses its effectiveness. Now the second point on the measurement is the paper does a fantastic job of calibrating it to the default profile but I felt that there was a missing piece there calibrating it to the consumption and what we're looking here is a graph from the paper by Marco Di Maggio and Quarters which tells us well the effect of reduction in payment on the consumption is unambiguous but another thing that comes out of this paper is that the pass-through of the reduction to payment to consumption is actually not one-to-one it wasn't crystal clear what it is but it felt to me like it's between a certain 50% and so of course if you incorporate that parameter into the models the magnitude of this arm welfare effect will go instantly in an obvious way down. Now a short observation on the size effect this has more to do with how committed you are to this blank page approach that you're taking here to the design and if you are, so the paper looks ignoring things that exist we get to redesign the world from scratch and in that world we're going to come in and come out and design a perfect contract. Now to the degrees that you're going to work somewhat if that's comfortable with that approach you can ignore this comment you can work with contracts that are already somewhat out there and if you Google arm it actually doesn't tell you it's a variable rate it says well it's a contract that has a fixed period that has an initial period of a fixed rate and after that it's a variable rate so say that we introduce that element of that there is a period and by the way it's 5, 7, 10 years then what you're going to have is that starts to slow you down now Arvind mentioned that there are 45 generations of people in the pipeline taking mortgages all the time so it slows you down but not quite that much because not everybody if you have a slow down period and not everybody instantly with the magic one convert it into a variable rate but it's more traditional art forms and some people will enter with the delay but nevertheless with many, many generations stuck together that will work slow it will be a smaller effect but not that dramatically now if you allow for a feature that was crystal clear present in the corporate sector which is the fact that when you go into the boom there is plenty of refinancing and so once you're refinancing then you would actually would happen is that this pipeline approach with many generations just going one after another actually starts to bulk this is what it looked like in the corporate sector so this is a picture that reflects that it's not a pipeline you as you go into the crisis there is a refinancing constantly refinancing and so the second the credit shuts down what you observe is that most of the expirations actually sits in year 5 because corporate loans are 5 year maturity which would mean that if it translates this to the arm world with 5 year fixed period you land in a situation where you have 5 years and you cannot act on them but again as I said this is only to the point if you gonna somehow go for perhaps more practical approach and start working with products that already exist out there as opposed to taking a blank page approach my final comment so there is this element which also speak to the magnitude of the effect one of the fundamental underlying assumptions is this pride default spiral and this price default spiral is premised to a large degree on the fact that there is the rental market and the ownership market are segmented and so the rental market price is fixed and it's really when you the household defaults there is a massive effect on the price of the house now there is some academic evidence on this but it's rather not direct and it's a bit harder to reconcile with the anecdotal evidence unambiguous that there is a cost of liquidating a household but the question is does it have this large impact on the price and granted that there are frictions from conversion from a rental property into condominium but actually the frictions going in the opposite direction are higher price and you can rent it and there are no frictions on that that seems like the frictions that go in the direction of that you would need to offset the effect that is modeled here actually rather smaller at least of course very nice paper thank you so I would propose that we collect some comments and questions that you maybe answer in one go unless there are a lot of questions please