We're sorry but this page doesn't work properly without JavaScript enabled. Please enable it to continue.
Feedback

Session II: Mortgage Market Design: Lessons from the Great Recession

00:00/55:12

Formal Metadata

Title
Session II: Mortgage Market Design: Lessons from the Great Recession
Title of Series
Number of Parts
12
Author
Contributors
License
CC Attribution 3.0 Unported:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
Identifiers
Publisher
Release Date
Language

Content Metadata

Subject Area
Genre
Computer animationLecture/ConferenceMeeting/Interview
Computer animationDiagramLecture/ConferenceMeeting/Interview
Diagram
Computer animationLecture/Conference
Computer animationDiagramLecture/Conference
Computer animationDiagram
Computer animationDiagramLecture/Conference
Lecture/ConferenceMeeting/Interview
Computer animationLecture/ConferenceMeeting/Interview
Computer animationLecture/Conference
Computer animationLecture/Conference
Computer animationLecture/Conference
Lecture/ConferenceMeeting/Interview
Meeting/Interview
Lecture/ConferenceMeeting/Interview
Lecture/ConferenceMeeting/Interview
Meeting/Interview
Lecture/ConferenceMeeting/Interview
Lecture/Conference
Transcript: English(auto-generated)
So, welcome back for the second session of the conference, which features two papers that cover pretty important grounds for those who are interested in the design or transmission
of policy. So, let me invite with a great pleasure Tomas Piskorski, Columbia University, to speak first on mortgage market design. So thanks a lot for having us here. This is a joint work with Amit Seru, who is at Stanford.
So the context of the paper is really the Great Recession. We know it started around the collapse of Lehman a decade ago. It has huge adverse impact on the economy with millions of foreclosures, job loss, and loss of output. What's sort of important also about the Great Recession was really followed by the unprecedented
stimulus measures. And just to sort of show you a little bit of a timeline of that, we had obviously a prolonged period of very low interest rates, both in US and in other places like in Europe. There was a run of quantitative easing in US, QE1, QE2, QE3, and I think what's sort of interesting in the European context, there was very two large, big debt relief programs
implemented at the federal level in US. For those of you from Europe who are not familiar very much with that, these two programs of home affordable modification program, there was a large federal program when government literally paid banks money to modify loans.
The original target was about 4 million borrowers. And the second even larger program was the home affordable refinance program. Essentially, the government allowed banks to refinance mortgages that had very little equity to still low interest rates. I'm going to be a little bit more precise what these programs meant, but I just want
to tell you that this monetary stimulus coupled with quantitative easing grants were also sort of done in conjunction with these two large federal debt relief programs. So we still are undergoing the recovery from the crisis. Many of the areas, as I will show you in a second, hasn't really fully recovered yet. And there's also very significant regional variation in the timing of the crisis and
the extent of the recovery. Now, this is a very big debate about the causes and consequences of the crisis. So let me tell you what we'll do in this paper and more broadly in our recent work, I'm actually going to show you some few slides that are not in the paper from some recent work we've done in this area.
So first of all, I'm just going to show you some summary statistics on the extent of the crisis in the U.S. They're going to be based on a representative panel of 10% of the U.S. population. This is consumer level panel when we tracked 18.5 million people monthly on more than a few hundred variables. Then I was just going to show you some summaries to some statistics, some empirical
evidence on the extent of recovery from the crisis across U.S. And in particular, I'll be interested in things like how much heterogeneity is there, how much time on average it took regions to recover. And then importantly, I will try to sort of, I will try to understand to what extent the variation in the extent of recovery relates also to implementation of
debt relief. And we will be focusing on three kind of main debt relief factors, mortgage contract rigidity, the ability of households to refinance their debt, and ability of the banks to restructure debt of household during the recession. This third part will very heavily draw from our prior work in this area.
We have a number of papers in this area that look on these factors. And kind of the main part of what I'm going to be talking about is drawing implications from this evidence and additional evidence coupled with a simple illustrative model to what we can learn from grand recession in terms of designing mortgage market going forward, thinking about better, smarter,
debt relief and macroprudential policies. So let me skip this. So first of all, let me just show you a few summary statistics based on the representative panel of U.S. population. This is really a 10 percent panel. We're really using a little bit less. Instead of using 18 million households, we use 13.5.
This is purely random because it's based on randomization of Social Security numbers. It's really representative. Let me first show you certain statistics on the extent of defaults in U.S. economy over the last decade. So the exercise here, we look on cohort of borrowers that were alive in
2007, and we see what happened to them over a decade of the period of time. So first of all, serious delinquencies. About a quarter of U.S. households went out home with a mortgage, were seriously missing their payments for at least two months or more.
Obviously, if you are in a bottom 10 percent of credit distribution, 70 percent of these borrowers were seriously delinquent on their mortgage payments. So almost a vast majority. Ultimately, not all of these borrowers end up being foreclosed, but in the U.S. during the last 10 years, one out of 10 households suffered foreclosure, and actually
this foreclosure was completed, the household lost home. Again, borrowers with multiple homes, the so-called mortgage investors, have much higher foreclosure rate, and again, in the bottom credit distribution, this is much bigger. These foreclosures led to the permanent loss of home ownership for the majority of these borrowers.
Only about 20 percent of borrowers regained home ownership in 10 years, and on average, it took them almost four years to regain home ownership. And also, there's a very huge impact on regional mobility and locational choices. The majority of borrowers that suffered foreclosure reallocated to different areas.
So these statistics have certain implications for the aggregate amount of foreclosures in the economy. So as of 2007, there was about 56 million households that owned homes with mortgages in the U.S. Of this household, about 10 percent suffered foreclosure. We're talking about almost 6 million foreclosures completed over the period of 10 years.
About a quarter of these foreclosures came from households that owned multiple homes. And as I already told you, these foreclosures imply persistent loss of home ownership rate for the majority of these borrowers. And in fact, the foreclosures can account for very substantial decline in U.S.
home ownership rate following Great Recession. Now let me show you now a little bit few statistics on the extent of the crisis at the regional level and the extent of regional recovery. So first of all, U.S. is also quite heterogeneous. It's also a monetary union with a set fund setting rates.
We have states, you know, Arizona is different from California. It just shows you that, you know, the house price decline, this is really, I should call it annual house price decline, 2007-2009, it's generally was negative. But the areas that are more red have a bigger, more pronounced decline in house prices. Most of you familiar with U.S. recessions are very surprising, usually same states,
coastal areas have bigger decline in house prices. They also experience bigger decline in unemployment and bigger increases in mortgage default. Now, if you look on recovery, the recovery from the crisis was also quite slow despite this unprecedented stimulus measure.
So very low interest rates, free runs of quantitative easing and this big, large federal debt relief program. So this is an average mortgage default rate, a fraction of loans that are essentially delinquent in a zip code over time. So this first period is this crisis build up and we call it recovery period starting from 2010 when quantitative easing was in place, when interest rates reached very low level
and when debt relief programs, refinancing and modification programs were in place. And all it shows, it took a number of years to get back to the pre-crisis level. Again, just to be very precise with language, when I say recovery, I'm just saying recovery to the pre-crisis level, not saying it's necessarily the right metric,
it's just a simple metric to describe it. So by now, by 2017, 74% zip codes recovered in terms of, they have the frequency rates at or below 2000 pre-crisis level, but it took on average 3.6 years. And there's a very significant regional standard deviation in the extent of recovery.
The same goes for foreclosure. By now, pretty much foreclosure crisis is over in vast majority of regions in US, but again, it took on average more than three years for those areas that recovered. House prices hasn't fully recovered yet. Only by 2017, only about half of zip codes recovered
in terms of the house price level. And standard deviation is actually very high still. If you look on standard deviation of house price changes across regions, there's still a lot of heterogeneity. Same goes for consumption, durable consumption measured by auto sales. Again, only about 53% of zip codes recovered to the level of consumption in pre-crisis
and again, there is very significant heterogeneity, but actually it didn't decline. There are areas that are doing great, there are areas that are not doing great. And the same goes for unemployment. This is as of 2016, so I don't capture the effects of Trump presidency, but as of 2016, only 40% of US countries recovered to the pre-crisis level of unemployment.
Others were worse. And again, there is a substantial standard deviation variation, but it's somewhat moderated. Just to show you one picture just graphically, this shows you that it is the areas as of 2010 that have lower consumption than before the crisis and this is the recovery. You see it gets better, but it's still a lot of areas that are still doing worse than they were doing
in 2007 in a pre-crisis period. So now what I'm gonna do, I'm gonna try to sort of provide you one set of factors that potentially can account for the extent of recovery across regions. It can explain the heterogeneity and it can explain the speed.
Of course, there are many reasons why regions recover faster, some regions recover faster, some regions recover slower from recession. But what I try to argue, and this is really heavily based on our prior work in this area, that an important part of the speed and extent of recovery is related to what we call debt relief friction.
So let me be very precise what I mean by that. So remember the picture I showed you at the very beginning, the interest rates are rock bottom. But vast majority of houses do not automatically benefit from that. If you have fixed rate mortgage, it doesn't mean that your mortgage payments go down and you get a stimulus. You only benefit automatically if you have adjustable rate mortgage with no flaws.
If you have an arm, there is an automatic pass flow, low interest rate and monetary policy. If you don't have an arm, you will have to refinance. But the problem in great recession is vast majority of households at risk of default or that were in delinquent, they didn't have any equity left. So they couldn't refinance to benefit from lower rates
because if they would have refinanced in normal private market, their mortgage premium would skyrocket and they wouldn't have any incentive to refinance. That's why the government put in place this home affordable refinance program to allow borrowers with very little housing equity left to still refinance to lower rate. But the only people that were eligible for this program
were people that had loans that were already, prior loans already guaranteed by government, the so-called GSE loans. Okay, if you don't get automatic debt relief because you have arm and if you cannot refinance, there is a third way that you can get help. Essentially, it is often in the interest of lenders to restructure your debt
because foreclosures are very costly, they can generate significant deadweight costs. There's big literature on this. But an important factor also that affected the extent of debt relief that ultimately went to houses was the ability of banks to restructure debt. In our prior work, we showed there's a lot of variation that some banks have in structure to modify loans,
some banks don't. And in addition, this is exasperated by these frictions that generally banks had very little incentives to modify loans that were securitized within the own balance sheet. So we use free empirical measures to capture these debt relief factors. One is the fraction of loans that are armed in the region. Remember, if a region has a lot of arms,
the pass-through of monetary policy will be quicker. On both ends, by the way, it goes both ends. The second measure is the measure of fraction of loans that are eligible for home affordable refinance program. If the region has a lot of loans that are eligible for this government program, this region will potentially get more help
because these borrowers will be able to refinance lower their mortgage payments, and as a result, the consumption can go up, the house prices can go up, and so on. And the third measure is the measure which we call high capacity share. It's the fraction of loans in a region that are serviced by banks that are more competent to conduct loan workouts as opposed to foreclosure.
For these measures to have any hope to explain some regional variation extent of recovery, there has to be a regional variation in these measures. And just to show you this is fraction of arms, there are some regions with 70% of loans are arms, there are some regions with virtually none of the loans are arms. And you can think about it, the redder areas
is a quicker pass-through of monetary policy. The greener areas have a slower pass-through of monetary policy, it's not as automatic if you have an arm. This is fraction of loans that are eligible for federal refinancing program. Again, there's a significant variation. And this is fraction of loans that were serviced by banks that have capacity to modify debt.
So let me first show you a few kind of difference in difference pictures. When more exposed are areas that have this debt for relief factors on average above than the median which means they have more adjustable rate mortgages, they have banks that are more likely to modify loans and they have more loans eligible
to federal refinancing program. This is without any controls. Obviously these measures are not fully randomly assigned by no means. But all these pictures shows that areas that were more exposed to debt relief experience after the stimulus period that we roughly say started in 2010, experience a relative improvement. In other words, a more pronounced decline
in default rate. Remember, this is a default rate. So the more exposed areas experience a bigger differential decline in defaults relative to the less exposed ones. The same goes for house price growth. On the opposite, you want more exposed areas experience faster recovery in house prices. In other words, if you have an area when more loans are adjustable rate,
when more loans are eligible for federal refinancing program and when more loans are in hands of banks that are less likely to foreclose and more likely to modify, you see essentially an improvement in house price growth and you also see an improvement in durable consumption in auto sales. Of course, this is without any controls. So you can run a kind of an interest
and I'm not gonna spend too much time. You can run a kind of formal difference in difference analysis, control for tons of socioeconomic factors at the zip code level. In addition, control for mobility flows in population and when you run this kind of regressions, there's a standard errors in parenthesis. You find that all these measures are very highly related even if you put them all together in regression
and this regression goes until 2016. So not only they're highly related to the extent of recovery but these effects are quite persistent. In other words, this shows that areas that have a lot of arms, loans eligible for refinancing program and services that can modify loans, this area experience much more pronounced
differential reduction in foreclosure rate, improvement in house prices, improvement in durable consumption and decline in unemployment. Decline in unemployment. And these effects are economically very meaningful. I will tell you a little bit what the cell aesthetic is implied but they're quite important. So let me just summarize what I just told you right now.
So what I'm trying to argue that part of the reason the crisis in US was quite severe is because there was this frictions in the past through stimulus, being at low interest rates, being at federal debt relief programs to households. So, and this is regardless of the narrative of what caused crisis in the first place.
So no matter whether you think the crisis was caused by optimistic beliefs, high leverage, world decline of interest rate and savings glass but push house prices and lending in US. Regardless of what caused the crisis, what we argue is the crisis would have been much less severe if these frictions, if the mortgage contracts in US
were a little bit more flexible, if banks were a little bit more willing or able organizationally speaking to modify loans and if there were a little bit more loans eligible for the federal refinancing program. In fact, if you take this estimates, obviously regional estimates are a little bit tricky because you have to think about aggregation to the general equilibrium level
but you can easily get something the crisis will be 50% less severe if we just have a little bit better, quicker pass through of these debt measures to the economy. And we are in the central bank. What I want to kind of, what this kind of research underscores is that household balance it and mortgage market regimes is a very important element
of monetary policy pass through. Traditionally, we thought a lot about banks, we thought a lot about other kinds of channels through which the monetary pass through operates. The most kind of direct channel, if you have an arm and I drop interest rates dramatically, this houses literally expand thousands of dollars of stimulus per year in terms of reduction in mortgage payments.
The same goes for credit cards and other forms of debt. So what we sort of, what this evidence shows, these factors are very important. So what I'm gonna do now, I'm gonna try to draw something from this evidence for the mortgage market design going forward. So I hope at least some of you could see the same way. What we're trying to argue is look,
these frictions have prevented to some extent implementation of debt relief, potentially make crisis worse. Could we somehow going forward, think about elevating these frictions? And they're really kind of two type of approaches. The ex ante approach is just change mortgage contracts or maybe put some automatic debt relief policies in sort of ex ante sense.
Exposed approach is wait until crisis happen and do something. So what we can learn about it. So one possible solution is ex ante approach, change the nature of mortgage contracts. Now, intuitively, you would think about some automatic adjustment of terms. So that in recessions, in severe recession, the houses get reduction,
temporary reduction mortgage payments, which will be compensated lenders in the boom, they could get higher payments. And as such, you can avoid deadweight costs of foreclosures, you can have a better risk sharing between investors and borrowers. Of course, you need some form of marketing completeness behind this for these contracts to be improving, but it's not very difficult to write a model
that will sort of generate this. And just not to be pretentious, anytime you see economies that something happened and I'm gonna change the nature of contracts and market, we actually did that after great depression. So prior to great depression, you know, in the 30s, mortgages in US were very short term maturity. Part of the wave of defaults in great depression was that it was the maturity default.
Borrowers were able to make payments, but they had to refinance and banks were belly up, nobody could refinance this debt. So one lesson of great depression, folks at that time said, this is horrible to have this residential mortgage short maturity, because it exposes us to the risk of maturity refinancing so we end up with 30 year fixed rate mortgage. So what I'm trying to argue,
we probably have much more foreclosures in US and default if we have shorter maturity mortgages. So there was a lesson learned in great depression, maybe the lesson learned in great recession is we might want to have a little bit more flexible mortgage contracts, which allow a quicker pass through of that relief. So there's a big mortgage design literature
to which you have some work in this area as well. Generally this literature tells you ideally, you would index to some measure of local house prices and income to the extent is possible. Of course you need to take a lot of factors into account, the empirically relevant frictions that complicate design of these mortgages. You need to seriously think about the impact on incentive of borrowers and banks once you,
because effectively this mortgage contracts that provide debt relief in recessions a form of the automatic stabilizer and insurance given to households. So yes, generally in response to such contracts, they will increase the debt levels, they will less incentivized to repay the debt. But there's a lot of work in mortgage design literature showing that even if you take this ex-ante incentive of ex-ante,
this context could still in some class of models improve kind of an overall welfare and adjustable rate mortgages can perform quite well in providing debt relief provided the interest rate to which they are indexed very highly commutes with income and house prices. You want essentially interest rates to be high in booms
and to be low in recessions. If this is the case when arm rather than being risky is effectively insurance to the households because you pay more when the economy is doing well, which you can sort of afford to make the payments, you pay less when economy is in recession. So just to show you, we have a simple stylized illustrative framework that just tries to make two points,
tries to understand how important is to use the right index for the design of mortgages and tries to also underscore the fact that it's very important to have a, if you want to go with ex-ante approach, change the nature of mortgage context, you have to have a really good ex-ante understanding of risks. So it's a simple model, there's a borrower that needs money to buy a house,
there's competitive lenders provide funding, lending happens before income and house price risk is realized. If borrower doesn't repay its debt foreclosure happens, there's some deadweight costs of foreclosure that generates efficiency gains of preventing them. They're sort of needed really for incentive reasons, if you're never foreclosed person that nobody would ever repay their debt. And this really compares to type of mortgages.
Fixed rate mortgage, your repayments are fixed, 90% of loans currently in US have a fixed rate mortgage type in terms of the origination or index rate mortgage, and mortgage that is indexed to some index. And a key thing what we want to do, it's very intuitive what I'm gonna show you is that obviously what this index is determines a lot the efficiency of such solutions.
So first of all, it's a simple kind of schematic thing. If there's some costs of originating this more complex index loans, they'll be kind of an interior solution. If your index is highly correlated with income and house price risk, then index rate mortgage, adjustable rate mortgage is just a special case of index
rate mortgage when as an index rate, let's say use one year treasury rate would be better than FRM. On the other hand, if this index is not that highly correlated with house prices and income and there's some implementation costs, you would actually do worse with this kind of mortgage. In other words, if you have an arm but your interest rate is not that highly correlated with the relevant income and house price risk,
as for example was in the inflation episode in US, then such contract does not really help you much in providing debt relief. This is kind of a three dimensional version of that when you recognize the fact that, because of lack of time, let me just tell you what it shows. It shows you that the benefits of index contracts are the highest for borrowers
that make the least down payment. They have to be a risky example. And the highest to the extent this income and house prices are highly correlated with each other and highly correlated with index you are using because you want this index to be high when house prices and income are doing well, this index to be low when house prices and income
to be low and you want house prices and income to come off together. Intuitively in a simple real business cycle model, that's what you will get. In the boom, real interest rate is high in the recession, real interest rate is low. But of course, we're talking about nominal interest rates and much more, in reality, this is not the case. And the last point I want to make before I show you some more evidence is that of course,
an efficiency of this kind of ex ante solutions when you want to index mortgage contracts to some index that will provide debt relief in the class is critically depends on our understanding of the distribution of economic risk. We can design the best contract possible given our ex ante knowledge of risk. If this risk expose us to be very differently distributed,
this context will perform poorly because essentially we have errors in modeling. Say we think house prices and income are highly correlated or we think fed fund rates is always very low in recessions. If we design mortgage contracts index to such indicators and this is not what happens, expose we will not necessarily. Again, this thing's something I've very intuitive. We just put a very simple framework
to kind of quantify it a little bit. So what I'm going to do in the last remaining five minutes of my talk, I'm just going to show you some evidence that speaks to these issues. If you would like to go with this approach, what kind of indexation you would use and what are challenges to do that? Let me just also acknowledge that Bob Sheeler has been thinking about this for decades.
He was a big advocate of indexing to house prices. He was big advocate. He was, I would say, probably a little bit ahead of his time because at that time we didn't even have zip code level house price indices. But his idea of indexing mortgages has been very alive and around for quite a while. So I will just provide you a little bit more evidence on it. But I've already showed you
there's a lot of regional heterogeneity in economic conditions that already suggests that indices that are decided at the national level are not that efficient because it's a one size fits all. The same problem you have in European Monetary Union. Same problem in U.S. U.S. is a monetary union. So if you have mortgages indexed to say one year treasury rate,
one year treasury does not reflect conditions well of Arizona and California at the same time necessarily. But I'm gonna show you some additional sort of evidence that speaks to this. The first thing, we just derive a measure of U.S. state business cycle using principal component. This just shows you 90% of states in U.S. They generally follow similar pattern.
Of course, they all kind of experience recession, but the shadow area shows you that there are areas that do better and there are areas that do worse kind of at the same time. Maybe better way to say it, that there's a kernel density distribution of correlation between national variables like real GDP growth, real house price growth, and state level measure of a business cycle.
And I'm just trying to say yes, in general, when U.S. GDP is doing well, states are doing well, but it's not a correlation of one. There's a lot of heterogeneity, and it's even worse with interest rates. So generally higher interest rates in U.S. means on average, U.S. states are doing well. So interest rates over the last 30 years
are generally high in the booms and lows in recessions. But again, this is very far away from perfect correlation. And a more sort of formal way to show it, there's a way to decide what is the absolutely upper bound of variation. You can explain at a given original level
with a coarser indicator. So I think the best way to understand just look on house prices. Suppose I want to explain zip code level, monthly zip code level house prices. If I'm allowed to use zip code level monthly index, I can explain 100% of variation by just using zip code level house price. If I move to the city level, because city has multiple zip codes, I won't be able to use a city level index
and explain 100% of variation in underlying zip codes. And this is absolutely an upper bound. This is not the, this is like, if you're the smartest person in the room, this is the maximum you can do. So as you keep on going up, you can see with national indices, for example, you can only explain about 30% of variation in house prices in levels, and only 13% variation for closure,
and the same holds for growth rates. So that shows you that essentially, national or even state indicators do not capture, there's just a lot of very granular variation in small U.S. regions like zip codes where this index doesn't really capture. In fact, you can have a city when one zip code experienced 20% house price growth,
and another is minus 20% at the same time in the same year. There are plenty of data points like that in U.S. So the last thing I just want to show you is that there's also a lot of variation in stability of economic relationships over time. This just shows you the cross-sectional annual regressions across U.S. countries of mortgage default on unemployment.
Generally, areas that have higher unemployment have higher default, so that's good. And in other words, areas that have higher default have lower house prices, that's good as well. But as you can see, these relationships are not stable over time. The coefficient is not constant, and the same holds for house prices and unemployment relationship. So in the remaining one minute and a half or two,
let me just sort of conclude. So what I show you, there's a very significant economic heterogeneity across space and time. One size fits all policies or indices is not likely to be very efficient. And there's some sort of instability of economic associations over time. Why the signs are usually the right one, they kind of change over time, and this sort of complicates the design
of contracts ex ante. So what are the implications for the ex ante mortgage design? Again, I've already sort of told you, the arms sounds like a cool idea, but there are periods like stagflation episode and even early in the crisis of 2007, eight, when the Fed was really raising crates during recession already, so it's not always great to have an arm.
In fact, if you have an economy with a lot of arms, the monetary policy will have to be kind of smart and recognize that the pass-through is very quick. In additional, an economy with a lot of arms imposes a political constraint. Raising instances can become very unpopular from sort of political, you can complicate price stabilization objective. So in general, to the extent possible,
there will be bigger gains of using more granular indicators, like regional level house price indicators. We currently have indices like that available, so there is some hope we can do it. In terms of expos debt relief solutions, they have advantage that you don't need to have such a good ex ante understanding of risk, because you can see what actually happens
and react to that. The problem, expos solutions are subject to implementation frictions. You might have to go through financial intermediaries that might not have capacity to implement that relief, and there are also political constraints. In fact, part of the reason that GSE and HARP was not fully implemented because of the opposition from Congress. In terms of future policies,
one kind of thing I want to say, there's some debate, I'll discuss it with folks in Washington. Maybe we should make a HARP permanent feature of GSEs, essentially have a policy that the regions that experience, let's say, more than 20 or 25% drop in house prices, these borrowers would be able to refinance. And I mean, just literally 30 seconds.
Other approaches, of course, we can always think about leverage limits, just don't allow people to be levered in the first place. We can rely more on rental as opposed to home ownership rate, and just few closing folds. There's this important trade-off between ex-ante and expos approach. Ex-ante approach allows quick, automatic implementation of debt relief,
but there are issues of what kind of indices you use, and it's likely to make errors, potentially errors in this design. Expos approach is cool, but it's subject to all these implementation frictions. And just to conclude last slide, directions for future work, I think there's generally with the data, we have more need to have in-depth analysis of relevant income and house price rates, can potentially thinking about developing indicators
and indices that can be used for the debt relief policies and that's why programs should think more about these frictions in the implementation. Anyway, thank you very much. Maria Suntas and Meti from Stockholm. I need my slides. Yeah.
So thank you for having me here, and can you find them? The net. Yes, perfect. Thank you.
Okay, so it was a pleasure to read Toma's paper, and so I actually, I read the paper that is for coming in the Brookings papers, and the paper is more based on the second part
of the presentation of Toma's. But in general, what Toma's and Toma's and Amit and different co-authors have been doing, has been to document how different type of contractual fissures or financial frictions like the ability by servicers
to renegotiate mortgages, may have aggravated the financial crisis. On the basis of this large body of work, Toma's in the Brookings papers is suggesting some possible solution for preventing
future housing market crisis, as long as they happen in the same way as they have happened so far, and as long as households having different contract do not adjust their behavior.
So what I will do in my talk is thinking more about incentives, and what are the incentives that may have prevented households from the financial intermediaries to get the victim of the frictions.
And this is also a bit based on my research. But just a very quick summary, what is the proposal of Piscorchia and Cixero? Just a small step back. Housing crisis generate huge externalities.
Why? There are price default spirals, so if there are some foreclosures, many other households are likely to go under waters and to default. How can we prevent this? Well, the recipe on the Brookings papers is, well, a different type of adjustable rate mortgages,
not only indexed to national conditions, but also indexed to local conditions, making sure that intermediaries are able to renegotiate. This is good, but I want to highlight
that my also come up with some costs, especially if we consider what are ex-ante incentives. So what are adjustable rate mortgages, even if indexed to local conditions? Well, these are re-sharing contracts.
What would be the ex-ante incentive for the household? Well, the households would take even more risk. So giving these type of contract might just increase leverage and put the economy in a situation in which while the households have even more debt
and might be even more inclined to default. Second, the problem is that, well, in this particular crisis, the crisis originated in the housing market, but housing prices might vary for many reasons. There might be also shocks affecting other sector
of the economy that propagate to the housing market. In this case, forcing intermediary to just renegotiate may actually cause problems for financial stability, especially when intermediary have capital losses.
So in this body of work, of course, Thomas was talking to us just about the housing market, but I would think that from a policy point of view, we would have to think how the housing market fits in the overall economy. Thomas was hinting at it also in the moment
in which we introduce adjustable rate mortgages, while they transmit the monetary policy in a much stronger way. And this is true also when a central bank increases the interest rate. Typically, the objective is that,
while there may be too many commercial loans to some sectors of the economies that might be in some geographical areas, but in the moment in which there are adjustable rate mortgages, geographical areas in which, well, there are just households borrowing
would be negatively affected and there could be a negative effect on consumption. So I am wondering whether we could try to decentralize more the equilibrium letting financial intermediaries choose. And the idea is that,
financial intermediaries should have lots of incentives to do so. Why? Well, defaults for closures have been documented to generate huge externalities. So a puzzle, I think, in this literature that has been not very well studied
is why intermediaries do not jump in and renegotiate and internalize these externalities. So I'll let you see how I think in some of my work and I will show you some other numbers
on the way in which different zip codes were affected by the housing crisis. So this is based on a paper with Giovanni Javara that appeared in the Journal of Finance in 2017. In this paper, we have a simple model in which, well, we abstract from the specific friction
of servicers or fixed interest rate mortgages and so on, but we argue that different lenders have different incentives to internalize
the externality of foreclosures. And the incentives depend on how much a default feedback on their balance sheets. Theoretically, what we show is that, well, a lender that has retained on its balance sheet a larger fraction of the mortgages outstanding
in a geographical area will have a strong incentive to renegotiate. Why? Well, if the prices drop even farther because of foreclosure, that will be even more default that feedback negatively.
Then, of course, this is my nice empirical ideas, but one might say, well, but at the end, even these lenders, to the extent that had the servicers that were unable to renegotiate, couldn't do so. So we put together a dataset to explore
whether indeed these high market share lenders with stronger incentive to internalize externality were more likely to renegotiate. So this is basically the micro evidence of our paper.
So the dependent variable is the probability that a loan that is in default, that is a 90 day delinquent, is foreclosed. What you see here is that the share retained is our measure of the lender incentive
to internalize externalities in a given geographical area. So the positive coefficient in the first column tells you that the lender will be less likely to foreclose a defaulting loan if it has retained lots of mortgages
in that particular area. This is not due to the fact that this lender has issued the different mortgages. Why? For mortgages that the lender had sold, the ones in the securities column,
the lender doesn't behave differently from other lenders. So maybe these are the lender that have better services. We argue not. Why? Well, remember, here we have a loan level data. So we can look at how a given lender behaves in a different sense of structure.
How? We have a lender fixed effects. So we exploit the differences in lender's behavior between different geographical area. Explore heterogeneity of lenders
to renegotiate or foreclose within a tiny geographical area, meaning that, well, we can absorb any differences in geographical characteristics using either zip code or even finer sense of structure fixed effects.
So it looks like that at least some lenders were able to overcome the organizational capability and started to foreclose less. So this is, so basically what do we get out of this?
Well, we get that the negotiation friction perhaps can be overcome. And second, it might be that these contract with the servicers that were not able to renegotiate is actually an optimal contract.
Does the behavior that I showed you has any economic significance? So in this table, I am running regressions that are very similar to the ones that Thomas showed you. But these are at the zip code level. And I am trying to explain the drop in housing prices
across the different zip codes in which the concentrate, there was higher or lower concentration of outstanding loans. So the economic magnitude of these effects tells you that in zip codes, in which on average, the mortgage lenders had retained
a higher fraction of outstanding loans, had a lower drop in housing prices, and we basically explain with our measure of concentration about one third of the standard deviation. So on the basis of this, what is the policy implication?
Well, I am not suggesting that we should favor concentration. I am aware that concentration has some cost. However, from an eccentric point of view, even in a more recent work, we show that in more general contests than real estate market,
concentration of outstanding loans acts as a stabilizer. Lenders that are important to an industry or to a geographical regions are more likely to provide the liquidity to avoid the downturns and propagation of shocks.
But on the basis of this evidence, we can design bailouts exposed, meaning that in the moment in which there are defaulting loans or defaulting mortgages, if we are going to create a bad bank, an intermediary that purchases the non-performing loans,
favoring some geographical or industrial concentration might help to internalize the externality. So overall, what is my conclusion is that
I have been an avid reader of Tom and Amit work on the real estate crisis. I just think that from the point of view of the policy implication, we should consider that, well, each crisis is different, can originate in a different part of the economy,
and that this intervention may come with some costs that are very salient when the capitalization of the banking system is a concern for a policy maker. And in general, an approach in which we put more attention on intermediaries,
organizational structure and incentives, of course, above and beyond the concentration of their claims might be fruitful for designing economic policy.
Do you want to add anything? Yeah, just thanks a lot for the discussion. I generally agree with everything you said. In fact, it's quite consistent with some of our older work, but generally banks don't have as much incentives to modify loans they don't own and they service. From the practical standpoint, 90% of loans in US are securitized. It's $10 trillion market.
So generally, these loans are not held on balance sheet of banks. So that's why thinking about some contracts that are just example, like adjustable rate, that's also the reason the justification for federal intervention in this market is these externalities that are not priced in the decisions of banks. Because first, even if they will cut them on balance sheet,
depends on the local concentration. And secondly, they generally don't keep this loans on balance sheet. So they have to be paid for doing that. If there are significant externalities of foreclosure. So these are very well taken point. I just want to say we don't have a very strong predictions of what should be done. We just want to document a set of frictions
that have a certain predictions for how we could, what we should do going forward. I think some of them are sensible, but obviously the ex-ante effects are very important. In fact, some work I've done in this area on the mortgage design, literature is very serious about private market competition, impact of informational asymmetries, ex-ante impact of changes in mortgage structure
on incentives of households to retain debt and to be indebted. And all of these are matter, of course, and they actually lower the efficiency of the solutions. But as there's some very interesting work on the macro area, like the general equilibrium models, kind of quantitative ones, that finds very similar implications. But yes, there is an endogenous response.
You have to think about incentives, but in this kind of macro calibrations, you generally find out that changing this context to be a little bit more flexible in certain way, of course, assuming reasonable distribution of risk that you have good understanding of that could generate some non-trivial gains. But these are very well taken comments. Comments, questions from the floor?
I thought it was a great paper. One question for Thomas, which I think is in line with what Maria Sunta said, which is within the ex-ante policies, you haven't talked about anti-cyclical provisions
for debt to income and loan to value, et cetera. And I was wondering why, because given the fact that in the States, they actually, the Federal Reserve does not have the possibility to implement those type of hopefully useful measures, neither the European Central Bank,
so what's your view in that respect? Thank you. I'd like to ask some questions. Let's see, yeah, yeah. Thank you, this is really interesting. I was wondering how you think about the regulation of financial intermediation in the context as to structural features of the economy.
So one of them would be labor market frictions, people who were delinquent borrowers also probably had worse labor market outcomes. And the second feature, even if it's not a structural problem, is sort of the presence of extreme agglomeration and clustering in sort of some cities where he'd harder than others along all dimensions, labor markets, housing markets, and so on.
So I also just have a quick comment. I thought the general spirit of the paper is, I think it's hard to disagree that somehow most contracts should have more contingency. But I think in the spirit of the discussion, also it would be important to understand
what frictions prevent these contracts from being signed in the first place. And here, I think the data you showed is interesting because you see a lot of variation, but it doesn't look entirely random. You see that there's clustering of certain types of contracts, and I wonder if you've taken a look, is this legal regulation, what determines, you know, in some areas, people to take more variable rates,
in some others more fixed, et cetera. And going back to your, you know, the political economy aspect of raising interest rate when contracts are indexed, let me tell you a quick anecdote. Last year, I negotiated a mortgage in Spain. So fixed rate, 20 plus years, just under 2%, very low rate. But of course, the variable rate is about 0.6.
So I tell the bank manager, I think the fixed rate looks pretty good. Are you crazy? Look at the variable, it's very low. Yeah, but rates may go up. Are you crazy? Everyone is indebted at variable rates. If they raise the interest rate, the economy will implode. It cannot happen. So which one did you take?
Are you crazy? So I wanted to push Tom Michael also a bit further on, but you're basically pushing for more debt relief. And I very much share that in general. In this part of the world, you often,
you know, incentives come to the fore and so often you hear, well, this will generate moral hazards. Here, this is mostly thinking about sovereign debt markets rather than mortgage debt markets, but sort of more generally, how would you respond to this type of questions? So let me quickly respond to,
I probably won't respond to all also in the interest of time. So, you know, in the debt design literature will tell you, you can do it in an incentive compatible way. Now, whether in the sovereign debt market, you get implemented politically, whether there'll be some kind of pressures for countries to deviate, I don't know. But in the debt design literature,
the way it really works, the contract is like literally coming to what you said. You get lower rates now because we're in recession, but you will have to pay higher rates when the economy is doing well. And if I put some prepayment penalty, a penalty from exiting this contract, essentially what you do, you shift in payments across states of the world. You're saying, look, you can look,
you can afford to pay more when economy is doing well. So I'm gonna charge you more when economy is doing well. When economy is doing worse, I'm gonna charge you less. If you do it as a ex-ante mortgage contract coming back to your analogy of interest, it will exactly work like that. You get a very low interest rates on arm, but sure, there's nothing for free. You'll have to pay more when interest rates go up.
But to the extent interest rates go up in a boom or when economy is doing well, you will be also your labor income will be higher and so on. Of course, in reality, it doesn't always have to be the case. And that's why we have to think really more seriously what kind of indices we use. And the same could be done in terms of debt relief programs. They're not completely unfunded. You can tax borrowers. That's why we have guarantee fees
in Fannie and Freddie's system in US. Obviously, there's big discussions here too low, but precisely the guarantee fees are there in place. It's like a deposit insurance. You pay it in a good time, you charge, borrow a certain fee, but we're understanding that there's a severe crisis, I'm gonna offer you debt relief. So kind of economic theory gives us a good set of ideas how to do it in an incentive compatible
and sort of fair way. I understand there's a practical political implementation of some of these solutions would involve, for example, having a prepayment penalty. And let me finish just with one thing about the financial innovation. In US, 90% of mortgages are government sponsored enterprise mortgages guaranteed by government.
Government essentially tells the market participant, this is three or four types of contracts you can use if you want to get subsidized financing. That pretty much suppresses any financial innovation. So everybody originates either a basic arm or an FRM. Any other loan you want to originate,
you would have often a hard time compete. And coming back to your question, the variation in arm share is largely related to, is this the area that has a lot of GSE loans based on house prices, areas that have so-called jumbo homes, you see much more arms there. Because these loans are not eligible for government financing, there's more of a private market there. Also areas with original subprime loans,
there was a lot of these loans were actually of an arm type, partly potentially for the kind of behavioral reasons you mentioned, the initial payment was low. So whenever the government is not involved, you see more variation in contract types. Whenever the typical loan made is subsidized by government, you end up with FRM. So by no means, of course, this variation is random.
So in some of our work, we exploit variation across borrowers, not across regions, exploiting very tiny discontinuities and you find the same effects on consumption defaults, because we're very mindful of the fact that in the aggregate, this is not obviously a random assignment. Maybe let me stop here, because I don't want to take too much time.
No questions? So coffee break. Please come back at three, 20 past three.