to talk a little bit about some research that
I’ve been working on. And I’m going to present
one finished paper, and then also some of
the additional stuff I’ve been thinking about. Most of this,
basically all of this is work I do jointly with
my co-author, Atif Mian. And we’ve also been
working with a PhD student at Princeton named Emil Verner. So I just want to
give them credit right away, because
they’re also researchers on all of this agenda. So from the very first time
I started really thinking seriously about research when I
was a PhD student, what is it, 15 years ago now, I was always
interested in this joint kind of issue. And that is, how
does finance interact with the macro economy? That’s kind of the focus of how
I even teach my MBA course I teach, the PhD course I teach. I think it’s the issue that,
to me, is most interesting. Because economists
generally think of these as two different spheres. When we think of
macroeconomics, we think of what we call
the real economy. And what I mean
by that are things like GDP, household spending,
employment, investment. Those are kinds of the things
that we think ultimately matter, at least partially,
for welfare, for how well– the well-being of people. Finance, in and of itself, is
not really about those things. It’s about how we transfer
resources, financial resources, from some people in
the economy to others. And so when we talk about the
financial sector, what we mean are things like banks,
mortgages, bonds, stock markets. These are claims on cash
flows, but they’re not directly, at least, affecting
the real economy, OK? And so there’s always been this
fascinating question to me, and that is, what is the
interaction of these two? How does finance itself,
or the financial sector, interact with the real economy? And a lot of my original thesis
work was in corporate finance. It was thinking about
how different types of financial contracts affect
investment of companies. So that was linking the way
companies finance themselves to the way they invest. So that was that linkage. And then as I moved on,
especially when I started working with Atif
Mian, we started thinking more about this
from the household side. And that is, how does it
change spending, employment, investment patterns if
people, for example, have a lot of mortgage
debt and house prices fall all of a sudden? So those are the
kinds of questions, generally, that I’m
very interested in and I’ve been working on. Basically, that’s the one theme
you can see through, basically, all the research I’ve written
over the last 10, 12 years. And the recent research
I’ve been doing, which I’m going to
talk about today, basically has three main
themes, or arguments. One is that credit supply
shocks matter a lot for economic fluctuations. So let me define what I mean
by credit supply shocks. Generally, and
again, we’re still kind of forming what we
exactly mean about these terms, but in layman’s term, I would
say a credit supply shock is when the financial
sector, for reasons unrelated to the
actual underlying performance of companies or
the income of households, decides to lend either more
or less to those households. So again, the idea is
that for some reason, the financial sector
decides, all of a sudden– we can talk about why, what
the underlying shock may be– they decide to lend more, OK? People in asset
pricing have been thinking about this question as
the time varying risk premium. So that’s some of the
arguments my students here have heard me talk about a lot. Other people will call this
credit market sentiment. Why, all of a sudden,
do the interest rates charged to risky
borrowers go down? But that’s what I mean
by a credit supply shock. And that’s not the traditional
approach in macroeconomics. The traditional approach
in macroeconomics has been that the source
of economic fluctuations is what are called
technology shocks, shocks to the productive
capacity of the economy. Think about all of a sudden
companies become better at producing goods for a certain
amount of capital and labor they employ. That’s been the traditional way
people have thought of that. And we are taking an
alternative route here. We’re saying, look, we
think that the evidence is more consistent with what
drives economic fluctuations is not so much coming from
the technology side, but instead, is coming from
the financial sector side. Further, what we’re
showing in our research is that these
shocks tend to work through the household sector
as opposed to the firm sector. And that’s kind of
a surprising result, because most of
the modeling that is thought about this
credit supply side, for example, the very
influential research of Ben Bernanke on the
Great Depression, where he argues that a
negative shock to bank– basically, the ability of
banks to intermediate credit, what I would call a negative
credit supply shock, was a big reason for
the Great Depression. It has emphasized mostly firms. And so the argument
is, OK, good times are when firms are able
to get the financing they need for investment. And bad times is when banks
basically pull back on credit and firms can’t invest,
they can’t employ workers. Our argument is that the
firm channel is actually secondary to the
household channel. That the household
channel seems to be what’s driving these
economic fluctuations. So all of a sudden, subprime
mortgages become available in 2004, ’05, and ’06, and
that, in some sense then, leads to house prices going
up, which then leads to a lot of people borrowing against
their homes to buy things, which provides a boost
to economic activity. So you can see the
difference in the way we’re thinking about it relative
to the traditional channels. And then finally, we
argue that these shocks amplify the business cycle. So that the fundamental
shocks may be something like technology shocks. They may be shocks
to other things. But the financial
side actually acts, at least in the short
run, as an amplification. It kind of amplifies
the good shock, makes the boom bigger than
it otherwise would be. And it exacerbates
the downturn when things start to turn south. So I want to be very clear
from the beginning of the talk. We’re not saying that finance
is bad or is evil or anything. We’re just saying that, at least
in the business cycle sense, it seems to be amplifying
these business cycles, even though in the long
run, it may be great right. In the long run, it may
be great to have a better financial sector, a
bigger financial sector for intermediating credit. Our work is really on
this kind of medium run. Let me tell you why
this is so important. This is work by Jorda et al
who did this very nice research on 17 advanced countries, taking
them all the way back to 1870. And this is just the total
ratio of bank lending to GDP, so how many loans are being made
by banks for both non-mortgage and mortgage lending. And what you can see over
the last 20 to 30 years– this picture, as you see,
goes all the way back to 1870. That’s what’s very nice about
this data they’ve collected. And what you can see
from this picture is that there’s been this
dramatic relative rise in mortgage lending over
the last 40 years, OK? To the degree that now mortgage
lending as a ratio of GDP is far above
non-mortgage lending. And when I say
household debt, I’m including, of course,
mortgage lending. That’s the biggest part of it. And so it kind of begs the
question, what’s going on? Like, what’s changed? Why are banks making so many
loans against real estate today relative to the past? It also leads to– you know, our
traditional models of banking are usually about, they
invest, they kind of monitor, and examine
firms, but that doesn’t seem to be a lot
of what banks are doing. They seem to be switching
a lot more to being lending against real estate directly. So let me talk a little bit
about this research paper that’s completed,
more or less done. And what we basically
show in this paper is that sudden large increases
in household debt in a given country– the sample is 40 countries
over the last 40 years, so it’s a big sample of
mostly advanced countries, but we have some
emerging economies like Thailand and Turkey,
and Korea we have in there– that sudden large
increases in household debt systematically predict a
subsequent decline in growth, OK? So the timing here is
if debt in a country, if household debt in a country
all of a sudden goes up, over like three to four
year period, that actually systematically predicts
lower economic growth in the subsequent
three to four years. So if you want to
think about the US, there was a dramatic
rise in household debt between about 2003 to 2007. And we all know what
happened after, right? So that’s the kind of timing
that we’re thinking about this. In particular, these
increases in debt are associated with consumption
booms and an increase in the imports of consumption
goods from overseas. This is especially true
in emerging markets. Those of you who may follow
Latin America, if you think about where Brazil is today,
Brazil, from 2009 to 2011, saw a massive
household debt boom, a huge import of durable
goods, especially autos. So that’s the kind of
thing to keep in mind. And then they went through
some pretty difficult times after that. We find that these
household debt booms are associated with very
low interest rate environments, OK? So that’s kind of why we think
that this is a credit supply shock. You could tell a story where,
oh, household debt goes up because households
want more credit. They believe house prices
are going to go up. That’s fine. All of that could be true. But it has to be the case that
the financial sector seems to be expanding credit. Otherwise, interest rates
would go up, not down. If all of a sudden, households
all want to get more mortgages and the banks don’t change
their beliefs at all, you’re going to see
higher interest rates, not lower interest rates. And so that’s one of
the points we make. We suggest that there
is, maybe, perhaps, a role of flawed
expectations formation, because as I’ll show
you, what the research shows is that economic
forecasters systematically underpredict– I’m sorry, systematically
overpredict GDP growth at the end of these booms. So we can take a factor, that
is the increase in household debt over the last
three years, and we can show that that
systematically predicts forecasting
errors by forecasters at the IMF the OECD, OK? So there is a factor
they should know, and they should be taking
into account in their models, and they’re just not. So that suggests that
part of what’s going on is that during
these credit booms, people don’t understand
that they could end badly, or that on average,
they end badly. Then we show some evidence
on the global debt cycle, which I’ll discussed before. OK, so I don’t want to
get too technical here, but let me just tell you the
basis of some of the findings. So this picture
here represents what we call an impulse
response function, which is from, essentially, time
series regressions, actually panel regressions,
using this big sample of 40 countries over 40 years. So what we do with
that is we basically use that and we explore
the relationship between three variables– GDP, firm debt,
and household debt. And the goal of a
vector autoregression, which is what produces
this, is to kind of just understand what’s the
dynamic relationship between these three variables. I view it as purely descriptive. I don’t think it’s a cause– I mean, it’s not telling
you a causal relationship. It’s just kind of describing
what happens in the data. What this impulse
response function represents is, after
estimating these models, if I all of a sudden,
within this framework, and having these
estimated coefficients, if I shock household debt
in a country, if I say, all of a sudden, let’s have
an increase in household debt, what is going to happen to GDP? That’s what this picture
is going to show you, OK? So what happens? If you shock household debt,
you see one, two, three, basically a three to
four year rise in GDP. So you see a boom, OK? GDP growth increases. And that’s driven
primarily by consumption. But then what do you see after? You see a reversion. And this is the kind of focus
of this paper, that from year 3 to year 7, you basically
see a recession. You see that growth falls. In fact, growth falls so much
that by the end of the period, you’re at a lower GDP than
you were when you started. Now, I want to emphasize this
latter result of ending up at a lower point is not really
the focus and is not so robust. So for example, if you took
out the Great Recession, you wouldn’t get that
lower long run effect. But what is incredibly robust
is this medium run decline, OK? That’s not just about
the Great Recession. I can even do this regression
excluding everything after 2006 and I’ll find the same result. That’s important because– I follow this stuff so much I
think it’s common knowledge, but it’s pretty well shown
that the Great Recession was more severe in
countries that had bigger rises in household debt. So we kind of know for
the Great Recession that household debt is a
big predictor of recessions for that recession, OK? But I’m saying, this is
true even in the past. It wasn’t just the
Great Recession driving this relationship. Here’s a scatterplot
which can kind of give you a sense of just the reduced
form relationship, just the basic core relationship. What I have on this x-axis
here, the horizontal axis, is the change in the
household debt to GDP ratio from four years
ago to last year, OK? And then what I have
on the vertical axis is GDP growth from this year
to three years going forward. So I’m asking for every
country year in my sample, how is the household
debt to GDP ratio change from four years ago to
last year related to growth from this year going forward? And what you see in
these scatterplot, when you run a regression, which
is what this line is, is you see a negative relation. And it’s quite
statistically powerful. So some of the usual
suspects are on here. I don’t know, those of you
sitting the front may squint. You can actually see
which countries these are. This is Ireland in 2008. So anyone that follows
this knows Ireland had this enormous
increase in household debt between 2003 and 2006 primarily
driven by a housing boom. And then they saw a really
severe recession, right? Greece, which of course, is
in the news all the time, is down here, 2007, 2008. So these are countries that had
big rises in household debt, sharp declines in GDP growth. But what I want to emphasize
is that this is not just about those usual suspects
in the Great Recession. For example, here’s
Norway in 1988. I don’t know if we have any
Scandinavians in the audience, but it’s pretty well known
that in the late 1980s, the Scandinavian
countries, Sweden, Norway Finland in
particular, had big increases in household debt
and then subsequently lower GDP growth. People generally refer to
this as a banking crisis, but it’s important– that, to me, is
almost an outcome. The banks got in trouble because
they were lending too much to the households. Then the households
started defaulting. Then you get a banking crisis. So what we argue
in this, and that’s kind of what I was
hinting at the beginning, is that this research
and these results seem most consistent
with the notion that these shocks are being
driven by credit supply. And in particular, I want to
emphasize this point again, because I said it pretty fast. It’s pretty common during
these booms in household debt that you see
interest rates fall. And in particular, spreads of
risky debt to riskless debt, they fall during these times. So those of you who work
in corporate finance, one of the common measures we use
is something like the interest rate on, say, high yield
debt minus the interest rate on investment grade debt. That’s a measure of risk. The risk premium we call
it in the credit market. That is usually falling
during these periods when household debt is going up. Likewise, other scholars
have used the share of debt going to low credit quality
individuals or firms. Some people call this
the high yield share. So how much total debt is
being originated by high yield firms versus
investment grade firms, that also is going up
during these booms. So all of this suggests
that the financial sector is playing a role in this. It’s not just the households
changing their beliefs, because again, if the households
alone just got optimistic about house prices, they
wanted to borrow more, and the financial sector
didn’t change, then what has to happen? Interest rates have to go
up, if anything, right? I mean, if you hold
credit supply fixed and credit demand goes up,
interest rates have to go up. So then the question is,
where do these come from? Like where are these household
debt shocks coming from? Where are these credit
supply shocks coming from? Well, I wish I had a better
answer on that right now. So that’s what we’re
working on now. The emerging market
literature has talked about this for a long time. This is not new in
that literature. If there’s something
new here, it’s really more about the
advanced economies. The emerging market
literature, it’s usually some argument about
the influx of foreign capital. So you can think about Latin
America in the early 1980s. You can think about the East
Asian countries in the ’90s. There’s some shift in what
I would call global credit supply, or global
appetite for risk, that leads to money flooding
into these countries and then household debt
is where it shows up. You can also think
about deregulation of the financial sector
as a potential source of these shocks. That you all of a sudden
allow the financial sector to do more lending,
and so then you get an expansion
in credit supply. There’s also a lot of recent
research on behavioral biases in the credit market, what has
been known as credit market sentiment. That’s another source. And as I mentioned, the
prediction from all of this is that the expansion of debt
is associated with low interest rates and increased credit
to low credit quality borrowers, which is
everything that we basically show in the paper. Then how do you get a downturn? Well, the argument
then is eventually people realize they’ve
extended credit too much. Maybe when defaults
start creeping up. Most people think the
subprime mortgage default rate crisis started as a kind of
the very last loan that I make is the first loan
to default. That’s like the classic symptom
of a credit boom gone bad. I’ve extended credit to
such a low credit quality borrower that they literally
default within two months after making the mortgage. And so you can imagine that’s
when the whole process reverts. And then we show evidence
that things like rigidities on wages and
rigidities on prices seem to be at least partially
responsible for why, when the demand
shock comes, when we pull back all that credit
and households no longer can be taking out home equity
loans to buy TVs and furniture, why does that lead to
an economic downturn? Because prices have a
hard time adjusting. Wages have a hard
time adjusting. And that’s what you get,
at least in the short run. Now, my colleague
Erik Hurst has been doing some really
interesting work on that these
kinds of debt booms may also distort the real
economy in a way that’s not just about short run frictions. For example, he shows
that a bunch of people stop going to community
colleges during these booms and they start working in
construction and retail. That has devastating
long term consequences, because when the debt boom
ends, they tend to not go back to school. So you can argue that that’s
foregone human capital accumulation because
of the debt boom. A lot of people say that’s
what happened in Spain during the recent boom. A lot of people dropped
out of high school, dropped out of college to
go work in construction. And when the boom ended, they’ll
end up being less educated, and therefore, less productive. This result is
actually strong enough to generate what I call a
global household debt cycle. So our analysis so far
that I’ve described is really about specific
countries, right? So this country
has a big decline– or a big increase
in household debt. And therefore, sees a big
decline in subsequent GDP. But what we started
showing, and this is how we kind of came
to this, is that when you look at the
components of GDP after the household debt boom– so the components
of GDP are basically consumption,
investment, net exports, and government spending. Not much interesting is
happening on the government spending, so we don’t talk
much about that in the paper. But the one margin
that seems to be going in the opposite
direction is net exports. So after the big boom in
household debt, consumption falls dramatically,
investment falls dramatically, but net exports
actually goes up. And those of you who
remember your macro, net exports is
exports minus imports. And what really is happening
is imports are falling, so net exports kind
of mechanically go up. That’s exactly what’s
been happening in Brazil over the last three years. I mean, you cut– all
of a sudden stop buying imported cars, your net
exports are going to go up. So that got us thinking, oh,
well, the net export margin helps countries cushion
this blow in some sense. But then what happens if
everybody goes through it all at the same time, right? If everyone is all going
through a big household debt cycle at the same
time, and everyone’s trying to cut imports
simultaneously, well, that obviously
affects everyone’s exports. And so that margin
won’t be there. And so we basically
find that countries that have their own
household debt cycle more correlated with the
global household debt cycle, they end up seeing an even
worse decline in growth after a boom in household debt,
which kind of makes sense. They don’t have that
net export margin. In particular, a global
rise in household debt reduces the scope for
exporting out of a downturn. And in particular, recent
changes in household debt matter for each country. And it generates what I call
a strong household debt cycle. So here is the picture. So here now, I’m not using
country-specific evidence. I’m just using
all the countries. I basically smoosh them all into
one observation for each year. So it’s just a time series plot. Every year on this
picture shows you how the global household
debt to GDP ratio changed over the
last four years. And then it shows you
subsequent GDP growth on the vertical axis. So the usual suspects
are out here. I think this is pretty
well known at this point, that there was this massive
increase in global household debt between 2004
to 2006 and ’07. And then that led to a
decline in global GDP growth. That’s the kind of
world global recession. What’s not so well known
is that if I take out these observations, the
regression line actually looks identical. Meaning that it wasn’t a
new phenomenon, actually. This magnitude of how big the
household debt boom was new. But this is something
I’m realizing in my recent research. If we go back and study a
lot of the lessons, actually, of the Great Recession,
we could have learned by actually studying the 1980s. But it just wasn’t
on a scale that was large enough that I think
people really understood. But it’s basically, the
1980s look like a mini-2000s. From 1982 to 1989, there
is a sharp increase in household debt in the
US driven by credit supply. And then you see a
severe recession. The same thing is happening
in a lot of countries. There I think you can make
a more direct connection to financial deregulation, OK,
in the 1980s in particular. So what are some other examples? And then I’ll wrap up
and take questions. So we’ve been looking at this. And all of this is, at this
point, pretty anecdotal. At that point, I
don’t think we’ve done enough serious research
to back the claims that I’m about to show you. But I think it’s an
interesting idea. Michael Pettis, who’s an
economist that works in China and has been thinking a
lot about China recently, in one of his recent
articles he had this quote, which got me fascinated. And I started to read a lot more
about the Latin American debt crisis. It’s amazing how much we
can learn from that episode. He basically says,
“In early 1970s, as a newly assertive
OPEC drove up oil prices and deposited their
massive surplus earnings in international
banks, these banks were forced to find borrowers to
whom they could recycle flows. They turned to a group of
middle-income developing countries, including
much of Latin America.” So his story, which is
fascinating actually, and I’m starting to
buy into a little bit, probably more than I should. I should first look at the data. But his story is that
when we talk about things like the current account
deficit and trade balances, that we should actually think
of the flipside, which is the capital account imbalances. And that those are really the
fundamental driver of things. That the reason the US– like
having trade restrictions against Mexico makes no sense,
because both Mexico and the US are running a trade deficit
with the rest of the world. And the reason they’re
running a trade deficit with the rest of
the world is not because of anything
on the goods market. It’s because China, essentially,
is exporting so much capital overseas. And so it has to
show up somewhere. So in some sense, he’s
focused on the financial side, and saying that the flows
on the financial side are really what’s
dictating a lot of what’s going on on the real side. And again, I think people aren’t
used to thinking like that. They’re usually thinking
the real economy moves first and finance follows it. But we’re making the
argument that maybe it’s going the other way around. Financial deregulation
in Scandinavia in the mid to late 1980s. There they– this quote is
from a survey of that episode. “The boom-bust
process there starts with deregulation of
financial markets leading to a rapid inflow of capital
to finance domestic investments and consumption.” So again, this is
the kind of story that we’re trying to tell. In new research
that we’re hopefully going to be able
to make publicly available within
the month, we’re testing this hypotheses in the
banking sector in the United States in the 1980s. And in particular,
state laws were deregulated at a much different
frequency in the 1980s. And we can kind of
use that variation to see whether states
that deregulated their financial sectors earlier
had bigger booms and bigger busts in that cycle. So again, the argument here is
there’s some aggregate credit cycle going on. And there’s a boom-bust
phenomenon happening more in the states that deregulate
their financial sector earlier. So these pictures are
really preliminary, so let me just kind of
try to explain them. Think of this as 25
states which deregulate very early versus 25 states
that don’t deregulate until like 1989, ’90. And then I’m showing you
the relative boom and bust in house prices, unemployment,
real GDP per capita, and residential investments. So this represents
the difference between the early
deregulation states and the late
deregulation states. And so what you see is this
very striking boom-bust pattern, that especially in house prices. House prices go up a lot more
in states that deregulated their banking sectors early. And then you see a crash. And here on residential
investment, you see again, this boom-bust phenomenon. You see it in unemployment too. Unemployment goes down
and then goes back up. This is very
preliminary research. But again, the argument
we’re trying to make here is that the business
cycle ends up getting amplified because of
this easier credit supply. So let me wrap up. So fluctuations in credit supply
working through the household sector seem to be
an important source of economic fluctuations. We argue that shifts
in credit supply tend to amplify the
business cycle, leading to severe boom-bust episodes
in some circumstances. But I want to be careful to
avoid normative statements. Economic growth in the
long run may be higher because of shifts
in credit supply, even if it generates
more volatility. And I’ll give you an anecdote
that one of our discussants showed, which I
thought was very nice. So he showed two pictures of
growth in Thailand and India. And those of you who work in
emerging markets may know, Thailand has a reputation for
being a bit like the Wild West in terms of its
financial markets. Whereas India is a really
heavy regulated system, OK? And what he basically shows
is exactly our phenomenon, that Thailand goes like
this in terms of its GDP. And India goes like this. But the average growth
rate is much higher in Thailand versus India. So that’s an anecdote. I don’t think people
have taken that seriously and really estimated
if that’s the case. But you can imagine
that’s easily true, that what we’re
showing you in our work is that the business
cycles tend to vary more. But overall, the average GDP
growth is higher in the areas where the credit supply kind
of moves more dramatically. I will say, though, that the
Great Recession has got people questioning that, in
particular, one of my colleagues and some others. You see an increasing
amount of research saying that economic growth,
even in the long run, may be lower if these booms
caused by the financial sector distort human capital, for
example, or the industry mix. So you can imagine what happens. That’s some of what we’re
showing in our new work. In these states that deregulate
early, what kind of industries do you think expand in
terms of employment? Retail, construction. What kind of industries
are stagnant? Tradeable goods, exporting. So you can imagine
that that’s partially what’s been going on in the US
over the last 20 years, right? That we’ve been focusing
on retail and construction because we’ve had
these housing booms, and that leaves the tradeable
part of the economy, the exporting part of
the economy weaker. Again, that’s not my
research, and I still think the evidence
still needs to be shown. But that’s basically where
people are starting to head. The big open question
I still think is, what is the source of
these credit supply shocks? Like, why do they come? I think there’s a general
argument you hear, something about financial excesses. So the Latin America
example I like because it’s very
clean in that sense, that you know exactly
what happened. These OPEC countries. Now that, of course,
means somebody else is paying a lot of money
to those OPEC countries, so it should offset
in some sense. So it’s still an open question. Maybe it’s related
to monetary policy. People have argued
that wealth inequality, because the rich
when they get richer put more money into
the financial system. So then you have the
credit supply shock. But I think we’re still pretty
far from figuring that out. But I think that
we’re making progress, so hopefully, people can start
to try to answer that question. And I’m happy to take questions. Thanks a lot for listening. AUDIENCE: I remember the dot com
recession about the early odds. And by the name
I’m giving it, it wasn’t related to
household credit. Or at least it seems that way. Are you saying– is your
paper saying that it is, or– but the element was there? AMIR SUFI: No, I
want to be clear that we’re not
trying to say this is the only source of
recessions, for sure. I will add that that
was almost not really a recession, in
terms of if you look at the magnitude
of that recession, it’s far smaller than
’80, ’82, or ’90, ’91, or the Great
Recession, for sure. But yeah, I think that is
a very different recession than these other three, and
from most of the recessions we’re talking about. That’s primarily focused, I
think, on capital investment fueled by equity, not debt. And so it’s a different animal. Totally agree with that. AUDIENCE: So when you
showed the one chart where it’s consolidated,
just a calendar year, it’s going back over
time, it shows– and you mentioned that
the order of magnitude sort of explodes in the most
recent financial crisis. In the United States, how much
of that was due to subprime? So you talked about high
yield share earlier. Homeownerships
going from 63%, 69%. How much of that was
driving the magnitude here? AMIR SUFI: So that
channel of what I would call the extensive
margin of subprime mortgages allowing people that previously
weren’t able to get mortgages to buy a home, I think that’s
an incredibly important part of triggering what happened. But in terms of explaining
the aggregate rise in debt, it’s not that large. And that reason is because,
like you said, I mean, 63% of the people
already owned homes. So the channel
that really matters is, at least we argue, that if
the fundamental shock is credit supply, that ultimately then
pushes up house prices, right? Because anyone who’s
bought a home knows this. I mean, what is the
broker going to tell you? What interest rate can– what
monthly payment can you afford? Well, if interest
rates are lower, and it’s easier to get credit,
then the monthly payment for a given price that you’ll
be able to afford will be lower. So you basically are saying,
I’ll buy a bigger house or I’ll buy a more
expensive house. So the real thing that explains
the aggregate rise in debt is then existing homeowners
turning around and borrowing against home equity. That’s really the thing, by far,
that explains the magnitude. And that was disproportionately
strong among low credit quality individuals. So it’s not– I wouldn’t say that that’s
not exactly subprime. But even prime individuals,
people all the way up to the 80, 85th percentile of
the credit score distribution, they responded
quite aggressively to that higher house prices
by borrowing aggressively. It’s very interesting,
the people at the top 10% of
the credit score distribution who are homeowners
are almost completely unresponsive. They do not go out and do a
cash out refi or increase home equity when their
house price goes up. So I think when we’re thinking
about which borrowing increased the most, the subprime stuff
was important for getting the whole thing going, I think. But it was really
everyone, more or less, except for the very top of
the credit score distribution, borrowing against their homes. AUDIENCE: Hi. So in your presentation,
you mentioned India versus Thailand. And one had more
boom-bust, more or less, because of more regulation. So from those
findings, how do you think you can form our current
discussions in the United States about
further deregulation of the financial
sector, like Dodd-Frank? AMIR SUFI: Yeah,
so I think there’s these short run and long
run effects that have to be kind of carefully considered. And I think that what I hope
our research is doing is just– when you read the traditional
economics literature on financial
deregulation, they’re telling a very
specific story, which is we need to improve the
allocation of capital to firms. That’s really the story
they’re telling about how you get better growth. So there’s some firm out there
that desperately needs capital. They’re small. They’re really productive. They have a great product. We all want to buy it. But they don’t have the
capital to buy it, OK? That’s the traditional argument
that financial deregulation has had. I think our research,
our new project, is really showing
that that doesn’t look like what’s going on. It looks like what
actually happens is that at least
in the short run, it tends to work through
the household sector, through more mortgage lending,
through more credit card lending. And you could tell the argument
that that improves consumption smoothing, or you could
tell that improves household insurance provision, or
you could tell the story that that just boosts household
spending in a way that might be dangerous if households
have behavioral biases. And so my worry in the
current environment is that are we just trying to
deregulate the financial sector in order to get them to boost
household lending more or not? Now, I want to be careful. There’s a lot of people that
argue that household lending is too tight, that restrictions
on household lending are too tight. And I think I have some
sympathy for that view, as someone who’s done a couple
of refis over the last two years. It is kind of
remarkable now how much you have to show and
the documentation. So I think there’s some of that. But I do think– I think that we, at least– I’m not saying the firm story
is completely wrong in the US. But at some point,
I mean, we need to ask ourselves a question. In a country like the
US, this advanced, with this advanced of
a financial system, do we really think there’s
a ton of firms out there that can’t get a
loan, and that’s why they’re not expanding? I mean, maybe. But I don’t know
if that channel is that big, at least in the US. I think when we’re talking
about emerging economies, it’s a different picture. I think the emerging
economies, there is a credible
story to say, look, India needs to have financial
deregulation because there are small firms that are
starving for capital. I think that story makes
sense for me for India. It’s not obvious it makes
sense for me for the US. AUDIENCE: Is there a defense
in your series of mortgage debt versus non-mortgage
debt for a country like Canada, where there are
less tax advantages for people holding mortgages? AMIR SUFI: Yeah, although,
I mean, as you know, Canada’s had a huge boom
in mortgage debt now. I think the tax
benefit is important, but I think that
it’s not so important that it would lead to
materially different patterns. But I haven’t looked
on it, to be honest. It’s a very good question. I mean, I think in the US,
when the interest deduction, originally, did lead
to people switching from borrowing on credit cards
to borrowing on their home. So that is an important channel. AUDIENCE: Should the
[INAUDIBLE] should prevented? And if so, how? AMIR SUFI: Oh, gosh. That’s way beyond my pay grade. You know, I don’t know
if they can be prevented. I mean, I don’t even know
what the source of them are. So I don’t know if
it’s even feasible. I think the normative
implications we make in our book are not so
much about these credit supply shocks, but about financial
structure, which takes me back to my roots as a corporate
finance person, which is really what I am at the end of the day. And that is, I believe
that the financial product of debt itself is part
of the reasons you get these amplifications. And to the degree that the
government very actively promotes the use
of that contract, I think that that’s a
really bad decision. Like I don’t understand
why the government does it. They give the mortgage
interest deduction. They basically penalize
banks for having claims that look too much
like equity when they lend. So that’s something I
think is part of there. So let me just explain
that intuition, because it’s not so obvious. But I think debt
has this ability to fuel asset price
bubbles because of the nature of the contract. And it basically
leads to a situation in which people who are not so
optimistic about house prices, for example, may be
willing to lend to people who are very optimistic
about house prices, because they feel safe. Because I have the senior claim. So if house prices fall, as
long as they don’t fall by much, I’m fine. And so you get this shift. This is basically the work
by John Geanakoplos at Yale, I think I really buy into. You get this shift
of buying power from more pessimistic people
to more optimistic people, which I think tends
to fuel the booms. This is an old idea. I mean, Charles Kindleberger,
the MIT economist, I mean, he talked about
this all the time, that one of the primary
features of asset price bubbles is that they are fueled by
some new financial product that makes the lenders feel they
have something as safe as money. And he wrote that in like 1982. I mean, you can see
subprime triple-A mortgage backed security. And it’s exactly what
he was talking about. So I think there’s
something about debt specifically that we
need to think about. AUDIENCE: You haven’t
mentioned this, so I guess it isn’t important. But at this time,
weren’t the banks securitizing these mortgages,
selling them around the world, getting those loans off their– and continuing to do
that, because it was just wonderful thing to do? AMIR SUFI: Yeah,
that’s a great point. And I should maybe
be more fair and say, like on this slide, when
I talk about the sources, I should just say, you
know, fraud as one of them. [LAUGHTER] And I have research
on that, for sure. I mean, I have research
showing that there was a lot of fraudulent
practices going on associated with
securitization. I guess at the end of the
day, I think of the bank– I mean, I think I am convinced
by the behavioral economists that the fraud would
not have been possible unless someone had some
incorrect beliefs somewhere, right? Someone was willing to
buy into this story. So that you need a
combination of it. It’s both. You need the fraud and I think
you need the beliefs to be changing as well. Because the way the fraud works,
at the end of the day, is you basically say, I
think this is kind of fishy what they’re doing,
but as long as house prices keep going up, who cares, right? So it’s a combination
of the both. So I don’t mean to absolve– and certainly, anyone that knows
my research well will know, I do not absolve
the financial sector for the stuff they
did from 2002 to 2006. And that’s a very important
part of the story. So I should have it here as
one of the potential causes of these credit supply shocks. AUDIENCE: Right. So I mean, on one hand, you’re
saying OK, in the long run, we would expect that the
financial sector and credit markets do a really
good job at– AMIR SUFI: I didn’t say that. I said that could be the case. I don’t know. AUDIENCE: But I think
it’s fair to say that this is broadly, probably,
still the consensus, right? On the other hand
it’s saying, OK, if you have these problems
in the medium term, that could very well lead– as also your colleague
here has shown– lead to problems
in the long run, precisely because you
might have some allocated inefficiency that even in firm
outcomes, as you were saying. You know, think about Spain and
construction workers, right? So what’s your hunch about this? AMIR SUFI: You know, I
think I would split it in between advanced economies
and emerging economies. I think my hunch is
for emerging economies, the traditional notion is
correct, that yeah, fine, you get more booms and busts in
Thailand, but in the long run, you get better
capital allocation. And I think in
advanced economies, again, I think I’m increasingly
skeptical of this idea that capital is massively
misallocated across firms. And so when you see the
financial sector growing rapidly, you have to think
about what exactly is it that they’re doing? And I think it’s like
one of these things that I think about a lot. That when you have
an economy that is incredibly efficient and
fluid, I think most of us at the University of Chicago
think that’s a good thing. Like economies should
be very flexible. They should be able to react. But the problem in
that kind of situation is if something gets distorted,
everyone moves so fast and takes advantage of
the distortion, right? You see this in construction
and with the financial sector. And so it’s like
you say, I mean, I think I do have a hunch
that this misallocation point is quite important. I know Erik, for example– maybe he doesn’t
want me saying this. I guess it’s on camera too. But I know he’s doing research
thinking very seriously about the allocation of
talent from undergraduate, from elite undergraduate
institutions, into finance from
2002 to 2006 away from sciences, for example. Like that’s an
interesting question. And I think that my hunch
is that prior to 1980, the advanced countries
probably looked more like emerging countries. But today, when I see
these big booms in credit, it’s hard for me to imagine
that there’s even long run good consequences of that. AUDIENCE: Could it
be that volatility in advanced economies is
becoming just a bad thing long term because of
how long people must– how much time people must spend
developing their human capital? 100 years ago, there could
be a construction boom, doesn’t matter. They’re going to go into
the next job the next day, or the next year, as soon
as the economy returns. They can’t do that now. If you haven’t spent that
four years or five years in an undergraduate
institution, you’re shut out of the economy
in a lot of ways. Could it be that we need
to start making that trade, pulling away from volatility,
to start guaranteeing even just slow growth? AMIR SUFI: Wow, that’s
a hard question. I think there’s a
lot of stuff mixed up in that question
in the sense that– I mean, there’s these
long term secular trends that we know about,
about employment middling out, the disappearance of
these kinds of other jobs. How does that– I
guess you’re asking how does that longer
term secular trend interact with these
boom-bust cycles? Yeah, I don’t know. I have to think about it. Honestly, I don’t
have a good way of trying to conceptually
link those two things. I mean, I think one thing I will
say that’s just a fact that I find quite fascinating is
whenever you see these long run secular trend pictures,
so for example, a common picture many of you
may have seen in a newspaper is the decline in manufacturing
employment as a fraction of total employment in the US. Like we’ve all– well, I’m an
economist, so I’ve seen it. I won’t say we all have seen it. But people have seen
that picture, right? It accelerates in downturns. It’s kind of very interesting
that you see this decline, and then it accelerates
every recession. So that’s where I
would start in thinking about the answer
to that question that you just gave, but I’d have
to think a little bit about it more. That somehow these
big shocks end up being more costly in this world
in which secular trends are leading to it being
harder to get a job if you don’t have human capital. But I have to think
about it a little bit. AUDIENCE: When we see
regulation, supposedly, it’s going to
loosen up recently, and similar to what
happened in the ’80s, what happened in early 2000. And other than that, we
have similar monetary policy with easy credit that’s been
going on for an extended period of time. And then we have the price
of housing right now is going back to 2006 level. And then on the
financial market, you said risk premium is
declining, exactly what we are seeing today. S&P is all time high right now. And then word on
the street, everyone is trying to
overplay risk assets. Are we in a perfect storm again? AMIR SUFI: Well, one
of the things I always look at when I try to think
about these predictive type comments, which are
always very hard to do, is it seems to me that the
booms that end badly are the booms that distort the
real economy in some way. It’s something you
can point to and say, this easy credit is causing
this in the real economy. So from 2002 to 2006, it
was very easy to know. I mean, if you look at the
fraction of total retail sales that was in furniture,
appliances, and electronics, it’s stunning. It just goes up dramatically
during those four years. So there’s some way
you can see, OK, people are borrowing
against their homes, and they’re buying
a lot of goods. So that’s kind of what I look
for when I try to think, oh, is there some real effect
of this credit boom? And I’ll be honest, in the
current environment, other than in the auto loan
sector, I would say, like in subprime
auto loans, I don’t see so much obvious distortion
of real economic activity. What I see is a bunch
of companies refinancing their debt, like
every other month, to take advantage of
low interest rates. But they’re not really investing
and they’re not hiring workers to a large degree. And so I’m not saying
there won’t be a crash. I’m just saying if you think
about the real economic effects of such a crash, I
mean, the cynical view, maybe the optimistic
view, is OK, there’s a crash in
high yield bond prices. OK, so some rich people
lose some money, right? But is it going to really
affect the real economy in a dramatic way? I don’t know. I think we sometimes
think oh, anytime there’s going to be a drop
in asset prices, it’s going to hurt
the real economy. But I’m not so sure that I
see that channel this time. I could be wrong. It’s always dangerous to
make such a statement. Now, if you’re asking
about asset prices, I’m not going to try to do that. But I will tell you there
are these pretty strong predictive relationships,
well, strong– I don’t know how statistic– Lars probably
wouldn’t believe it. Whether they’re statistically
significant or not, I don’t know. But that when the credit
risk premium is really low, it tends to predict a
decline in asset prices over the next two
or three years. Now, those models
have been blown out of the water for the
last eight years. Now, many people will say
that’s because of the Fed. Because the risk premium has
been low now, since what, 2011. Like really low. In credit markets. AUDIENCE: Do you think
with the Great Recession and what’s happened,
your model is going to be as
predictive of what’s going to happen in the future
because of the increased scrutiny on housing
debt and so on? AMIR SUFI: Yeah, I mean,
that’s a great question. I don’t know. I know that the Bank of
International Settlements, which is kind of the central
bank of the central banks, and they put together
the regulations for global national
banks, and they try to coordinate the global
banks, which they’re very upset with the current
administration because they’re saying they’re going to undercut
all of the stuff they’ve done. I know they have zeroed in on
household debt, like hugely. And they– in some
sense, we kind of like– the story we’re
telling in our book, they’ve been saying
for 30 years. I mean, there have
been economists there since the 1980s
pointing this out. So we’ll see. I mean, I think it’s
a great question. I think that the
nature of these crises is that they always show up
in different manifestations of what they look like before. But I do think
that real estate is the only asset in
the economy that’s collateralizable enough
and owned by households enough that if regulators really
focus on real estate lending, then you could imagine less
boom-bust kind of stuff going on. I’ve always been amazed at
just watching in Hyde Park, for example, how many new
rental buildings are going up. That when I first moved
to Hyde Park in 2005, those would have been
condo buildings, not rental buildings. So that kind of changes the
whole dynamic of the way the stuff works that
we’re talking about. If 50% of people– like this
stuff we’re talking about doesn’t happen in Germany when
home ownership rates are low. It doesn’t, because Germany
just looks like this. So I think that is
related to that. All right, well, thank
you very much, everyone. [APPLAUSE]