Financial Crises

Understanding Financial Crises: Lessons from
History
Alp Simsek
MIT
Alp Simsek (MIT) Lessons from the History of Crises 1

Roadmap
1 Course logistics
2 The subprime crisis of 2007-2009
3 Some lessons from the history of crises

Alp Simsek: Introduction
Associate Professor of Economics

Raised in Turkey
MIT undergraduate in mathematics and computer science
MIT Ph.D. in economics, 2010
Specialize in macroeconomics and finance

Why should you take this mini-course?
We will take an academic look at financial-banking crises:

How often do they happen?
Why do we care? From Wall Street to Main Street?
Why do they happen?
How do they get out of control?
What to do about them?
General mechanisms. Subprime crisis as case study/empirics

Course resources and requirements
Readings
No textbook.
Lectures: 8 lectures.
Grading: Pass/Fail
Participation: 30%
Take-home exam: 70%
Distributed at the end of the last lecture. Due in 24 hours.

Roadmap
1 Course logistics

Financial system channels resources to uses

Mortgages: Loans collateralized by houses
I&ippliers of capital
•.... melXNs _ S3Virqs
· RIms ..... C<I!!h
IUsers of capital
·FiIms tINI1 .,..",.
>Go> Ul II"ots IfIaI. SP8Ild
......... do:b W)'Ing a hou,..,
Subprime mortgage: Borrowers with lower credit ratings.

Solution: Securitization redistributes mortgage risks

Aside: Derivatives and securitization
Derivative security: Value derives from another security.
Financial innovation created new derivatives in recent years.
An interesting example is collateralized debt obligations (CDOs).
These are constructed in two steps:
Pool underlying securities (mortgages, but also corporate bonds,
loans etc).
Sell claims to parts of the cash fiows on the pool (“tranches”).

Structure of a CDO
Consider a bond with promise (or face value) of $100.

Suppose (for simplicity) it pays $0 in case of default.
Construct an equally weighted portfolio of many such bonds.
Create tranches by seniority:
The most senior tranche has a face value of $70. It pays in full
unless over 30% of the bonds default, in which case it pays the
remaining value of the bonds.
The next most senior has a face value of $15. It pays in full
unless over 15% of the bonds default, in which case it pays
whatever remaining value is above $70.
And so on until you reach the equity tranche, which has a face
value of $3 and pays only the value of the bond portfolio above
$97.

Structure of a CDO
Equity Tranche
Mezzanine Tranche
Senior Tranche
Super Senior Tranche
70 85 97 100
Bond Portfolio’s Ability to Pay
Image by MIT OpenCourseWare.

Why CDOs?
Credit rating agencies rate bonds according to probability of
paying in full.
There is a scarcity of the bonds with the highest rating (AAA):
These bonds account for only about 5% of the supply of

corporate bonds,
But many institutional investors are restricted to hold only
high-rated bonds.
CDO creates a supply of AAA tranches even if no individual
bond is rated AAA.
The low-rated tranches can be sold to hedge funds and other
investors who are looking for high yield and can tolerate high
risk.

CDO alchemy
Courtesy of Efraim Benmelech and Jennifer Dlugosz. Used with permission.
Figure: From Benmelech and Dlugosz (2009).

Are CDO ratings reliable?
Pitfalls with CDO ratings:
1 Unlike AAA bonds, AAA tranches of CDOs are “optimized”so
that there is just enough collateral to ensure AAA rating.
Riskier than a AAA bond (marginally AAA).
2
In view of diversification, the risk of AAA tranches

depends on the probability of a negative aggregate
shock (recession, falling house prices etc.) that affects
many underlying securities simultaneously.
3
Rating agencies are good at modeling idiosyncratic default risk.
Not so good at modeling aggregate shocks (and correlations).

Back to story: Subprime mortgages securitized
Subprime Subprime Percent

Total
Subprime Share in Total Mortgage Subprime
Mortgage
Originations Originations Backed Securitized
Originations
(Billions) (% of Securities (% of
(Billions)
Dollar Value) (Billions) Dollar Value)
2001 $2,215 $190 8.6% $95 50.4%
2002 $2,885 $231 8.0% $121 52.7%
2003 $3,945 $335 8.5% $202 60.5%
2004 $2,920 $540 18.5% $401 74.3%
2005 $3,120 $625 20.0% $507 81.2%
2006 $2,980 $600 20.1% $483 80.5%
This is vulnerable to a drop in nationwide house prices. Why?

House prices rose and then fell...

Defaults and delinquencies increased
PAST DUE MORTGAGES
30
25
20
Sub-Prime
Past Due
Percent
15
10
Prime Past
Due
5
0
98
99
00
01
02
03
04
05
06
07
08
09
19
19
20
20
20
20
20
20
20
20
20
20

Markets recognized risks in AAA tranches
Figure: From Brunnermeier (2009).

Courtesy of Markus K. Brunnermeier. Used with permission.
The spreads are calculated from CDS prices. They provide a
measure of the default probability for corresponding tranches.

Aside on credit default swaps (CDS)
A CDS is an insurance contract on the default of a particular
bond.
For example, suppose you own a corporate bond from company
XYZ with principal $1,000. If company XYZ defaults, you might
get back $500 instead of $1,000.
You may buy a CDS for XYZ from someone (CDS seller). In this
case, you will definitely get $1,000.
If XYZ defaults, the CDS seller pays you $1000 (in exchange for
the bond) so that your total of $1000 is guaranteed.
You “swap” the default risk with the CDS seller.

Example
In October 2008, the 5-year CDS rate on Morgan Stanley debt
with face value $10,000 was $1,000.
This means that you could enter a swap where you paid $1,000
a year for five years, and in return you get payment $10,000 if
MS defaults (in exchange for the MS bond).
This price provides a measure of the probability that MS will
default. For example, if the recovery rate on MS debt is 50% (in
a default, MS would only pay fifty cents on the dollar), this
(roughly) implies:
20% chance that Morgan Stanley would default in the next year,
About 70% chance of default in the next five years.

CDS during the Euro-debt crisis
© Cable News Network. All rights reserved. This content is excluded from our Creative
Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/.
Estimated probability of default on sovereign bonds over the next five

years in September 2011 (CNNMoney article on September 16).

Back to story: Markets recognized risks

Key aspect: Some financial institutions made losses
Krishnamurthy (2010), “How Debt Markets Have Malfunctioned
in the Crisis.”
Courtesy of the American Economic Association. Used with permission.

Their default risks increased

Some of them became bankrupt
Some others (Bear Sterns, Freddie, Fannie, AIG...) were bailed out
with government support.

Stock market crashed

The US economy entered the Great Recession
US Real Yearly GDP
13700
Billions of Chained 2005 Dollars
13200
12700
Recession
12200
11700
11200
0
1
2
3
4
5
6
7
8
9
0
1
2
201
200
200
200
200
200
200
200
200
200
200
201
201
Date

Economic activity fell below potential
Courtesy of the Congressional Budget Office. This work is in the public domain.
Triggered strong policy response by the Fed and the treasury.

Monetary policy appears to be constrained
Courtesy of the Board of Governors of the Federal Reserve. This image is in the public domain.
Triggered unconventional policies: Quantitative easing etc.

Triggered also bailouts and stabilizers, which raised deficits....

Fiscal policy steps in, raising government deficits
Courtesy of the Congressional Budget Office. This work is in the public domain.

The shock seemed small relative to damage
• One feature of the subprime crisis is that the initiating shock seemed to be
small relative to the ultimate damage it caused.
• Blanchard (2009, “The Crisis: Basic Mechanisms and Appropriate Policies”)
notes that:
o The estimated losses in the U.S. subprime market in October 2007
was around $250 billion dollars
o The cumulative world output loss relative to trend between 2008
and 2015 (based on IMF estimates) was around $4700 billion
dollars. About 20 times the initial loss in the subprime market!
o The cumulative loss in the world stock markets from July 2007 to
November 2008 was about $26400 billion. About 100 times the
initial loss!

The mystery of the subprime crisis: Whodunit?
The subprime crisis features many candidates for a culprit:
1 Extension of subprime loans by banks, e.g., lax lending standards.
2 Securitization and the CDOs.
3 Rating agencies.
4 CDS (looks innocent so far, but still a key character)
5 Large financial institutions that made the losses.
6 Government (Fed+treasury) suport or bailout of banks
In fact, books written (movies made) about each candidate.

But economics is about prioritizing & focusing on first order.
Where should we focus our efforts? Some history could help...

Roadmap
1 Course logistics

Lessons from history: Crises are “universal”
Allen and Gale (2009): Crises are “universal”phenomena.

They happened in different periods and in different countries.
They happened in developing and developed countries.
AG discuss Bordo et al. (2001), who analyze the incidence of
crises
1
in 21 countries over 120 years.
2
Banking crises: Erosion of most banking capital.
3
Currency crises: Forex attacks and devaluation (not our focus).
Twin crises: Both at the same time...

Bordo et al. (2001) find that:
Banking crises are relatively common in most time
periods (except for 1945-1971---highly regulated)
They also happen in developed countries--albeit
less frequently
Their aftermath is typically associated with severe
output losses.
Schularick-Taylor (AER, 2012) analyze the relationship

between banking crises and output more systematically...

Schularick-Taylor (AER, 2012):
They date 79 banking crises (denoted by year 0 in the figure) and
analyze the evolution of investment and output in their aftermath.

Courtesy of Moritz Schularick and Alan M. Taylor. Used with permission.
Severe drops in investment and output, partial recovery.

Cerra and Saxena (AER, 2008) do a similar analysis as
Schularick-Taylor using an alternative data set that covers 190
countries between 1960-2001.
Courtesy of Valerie Cerra and Sweta Chaman Saxana. Used with permission.
They find much more persistent effects, little recovery.

Lessons from history: Optimism/bad news
• Cerra-Saxena also find optimism during the crisis. Growth

forecasts systematically revised downwards as crisis unfolds.
• In their popular book, "This Time is Different," Reinhart and

Rogoff also emphasize optimism before and during crises.
• As we will see, optimism was also arguably widespread before

and during the subprime crisis.

Examples from recent history
Read Allen and Gale (2009) for a brief discussion of recent crises:
Scandinavian crises (Norway, Finland, Sweden) of early 1990s.
The Japanese crisis of early 1990s
Asian crises of late 1990s (Asian “dragons” & “tigers”):
Russian default of 1998 and the LTCM mini-crisis in the US.
The Argentina crisis of early 2000s.

Lessons from history: Summary
History suggests:
Banking
1 crises are common. Can happen in developed economies.
The following features are also quite common in crises:
2. Crises are typically followed by large drops in output.

3. Crises are associated with ex-ante optimism/ex-post bad
news.

Back to the subprime crisis: Whodunit?
Recall that we had the following candidates for blame:

1 Extension of subprime loans by banks, e.g., lax lending standards.
2 Securitization and the CDOs.
3 Rating agencies.
4 CDS (looks innocent so far, but still a key character)
5 Large financial institutions that made the losses.
6 Government support of bailout of banks

Common denominator: Banks and the government
Universality suggests points 1-4 might not be so central:

Banking crises happened without subprime, CDOs, rating
agencies, CDS....
In contrast, point 5 appears to be a common feature of crises.
Severe financial events that don’t involve banks need not
generate crises. The bust of the NASDAQ bubble in 2001 wiped
out a lot of wealth (more than subprime). But banks were not
exposed. Only mild recession.
Point 6 is also in the mix. Most episodes feature gov support.

Roadmap for the rest of the course
We made a tiny bit of progress, but several issues remain:

Transmission: How do banks’problems affect the economy?
Amplification: How do “small” shocks generate large damage?
Causes: What are the ultimate causes of bank losses?
Panics: Why are crises often associated with a panic?
What is the role of optimism/bad news in losses or panics?
Solutions: Optimal policy during a crisis? How about before?

Roadmap for the rest of the course
Lecture 2: Borrowing constraints and the net worth channel
Lecture 3: Leverage, fire sales, and amplification mechanisms
Lecture 4: Understanding banks’losses: Moral hazard or
mistakes
Lecture 5: Liquidity, part 1: Maturity mismatch and banking
panics
Lecture 6: Liquidity, part 2: Debt, information-based panics, and
fiight to quality
Lecture 7: Interconnections and complexity.
Lecture 8: Optimal policy: How to mitigate or prevent crises?
For review, read the chapter by Allen and Gale (2009).
For tomorrow, read the intro of Holmstrom and Tirole (1997).

MIT OpenCourseWare
http://ocw.mit.edu
14.09 Financial Crises

January IAP 2016
For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
48
14.09: Financial Crises
Lecture 2: Borrowing Constraints, the Net Worth
Channel, and the Credit Crunch
Alp Simsek
Alp Simsek () Net Worth Channel 1

What is the role of financial institutions?
Financial institutions (or “banks”) seem central to financial crises.

They intermediate credit between ultimate savers and borrowers.
Broadly, can think of them as investment specialists/experts.
They can identify firms with good borrowers, ensure repayment etc.
They can also collect information and “correct”asset prices, thereby
indirectly infiuencing investment (e.g., trading type activities).
Next: A model of banks and their borrowing constraints, based on
Holmstrom and Tirole (QJE, 1997).

Roadmap
1 Borrowing constraints and the net worth channel
2 Empirical evidence on the net worth channel
3 The credit crunch
4 The net worth channel more broadly

A stylized model of banks
Consider a model with two periods t ∈ {0, 1}.

Two types of agents: banks (B) and financiers (F).
The H-T model has three agents: Firms, banks (which they call
monitors), and financiers. I am presenting a simpler version.
Both types have linear preferences, C0 + C1 .
Financiers are money rich but idea poor:
Large endowments at both periods. Happy to make loans as long as
they break even. The interest rate is, 1+ r = 1.
They don’t have ideas for profitable investment.
Banks (our focus) are the opposite: idea-rich but money poor...

Banks have investment opportunities
Each bank starts with some initial cash, denoted by N.

It chooses how much to invest at date 0, denoted by I .
Suppose each unit of investment generates payoff R at date 1.

Suppose R > 1 so bank wants to invest as much as possible...
It invests her own cash. But she might also borrow from Fs.

Banks face borrowing constraints
Suppose the bank borrows on “per asset” basis.

For each unit of investment I , it can borrow ρ.
Imagine the bank as using its assets as collateral to borrow.
E.g., a mortgage with 20% downpayment features ρ = 0.8.
For a given ρ, the bank’s budget constraint can be written as,
I = N + ρI .
The bank’s investment can then be written as,

1
I = N .
1−ρ
We can also illustrate this on a balance sheet.

The instructor's calculations based on the Federal Reserve Statistical Releases.

Need: A theory of the leverage ratio
Recall our investment equation

1
I = N .
1−ρ
The ratio of total investment (or assets) to net worth, 1/ (1 − ρ) is

known as the leverage ratio.
What is the leverage ratio of commercial banks in 2008?
Might ask: What determines ρ, and thus, 1/ (1 − ρ)?
Why is ρ < 1? What would happen if we had, ρ , 1?

Need: A theory of borrowing constraints
These are deep questions. The field of corporate finance.

If we had ρ , 1, the bank would not be borrowing constrained.
Can do as much investment as she likes with little or no capital.
In practice, banks as well as firms (sometimes) seem constrained.
Literature emphasizes constraints driven by information frictions.
We next illustrate this using a version of Holmstrom-Tirole’s (1997)
model based on moral hazard– a particular information friction...

Moral hazard: The bank can misbehave
Suppose the project either succeeds and yields R, as before, or fails

and yields 0. Suppose also two versions of the project:
Project Normal Shirk

Private benefit to the bank 0 cI > 0 .
Prob. of success 1 q<1
The normal version succeeds for sure as before.
The shirk version requires less effort (skip due diligence etc).
Might fail, but it generates private benefit for bank insiders..
Moral hazard: Insiders (managers/workers/owners) might misbehave.

(Information friction: Fs do not observe whether bank insiders shirk).

Skin-in-the-game, to provide incentives
A borrowing contract divides the output from each unit of the project (in
case of success), R = R F + R B , to meet two objectives:
1 Financiers need to break even,
R F = ρ.
2 The bank insiders need to behave (incentives):
R B ≥ qR B + c.
The second condition can also be rewritten as,
c
RB ≥ .
1−q
For good management, B must have “skin in the game.”

Key implication: Limited borrowing
Combining the two conditions with R = R F + R B , we obtain
ρ = RF = R − RB
c
≤ R− .
1−q
So ρ cannot exceed an upper bound,
c
ρ=R− .
1−q
Suppose c is large (incentive problems severe) so ρ < 1.

We have thus obtained a theory of limited borrowing, ρ.

Broader interpretation: Limited pledgeability
This is a highly specific and stylized model.

But some features of the model are more general.
The bank is constrained since cannot promise (or pledge) all of the
value from investment, R, to potential lenders.
c
Can only pledge up to a level, ρ < R − 1−q .
This feature, limited pledgeability, is more general...

Broader interpretation: Limited pledgeability
An alternative and broader interpretation of limited pledgeability:

less tangible or more “risky”
R= ρ + R −ρ .
|{z}
more tangible or less “risky” assets (collateral)
Less risky =⇒ More pledgeable since it varies less with info (including
unobserved action), and thus, is subject to fewer frictions.

Limited pledgeability on collateralized loans
In collateralized lending, each loan is backed by specific asset.

In this case, you might imagine ρ as applying asset by asset.
In this context, ρ is known as the loan-to-value ratio.
Residual, 1 − ρ, is known as the margin/haircut/downpayment
The amount B would have to pay out of its pocket to buy the asset.
The inequality, ρ ≤ ρ, becomes a LTV or haircut constraint.
We do observe margins on collateralized loans in practice....

Courtesy of the American Economics Association. Used with permission.

Limited pledgeability on collateralized loans
Repo is a particular type of collateralized loan. Will come back.
The table shows that riskier assets have greater haircuts/margins.
It also shows haircuts are (typically) low on many financial assets.

One reason why banks can have so high leverage—will come back.

Back to model: Investment with limited pledgeability
With limited pledgeability, the bank’s budget constraint becomes:

1
I = N with ρ ≤ ρ.
1−ρ
The bank chooses how much to borrow and invest subject to ρ ≤ ρ.
The bank’s return from choosing particular ρ can be written as,
R −ρ R −1
RI − ρI = N= 1+ N.
| {z } 1−ρ 1−ρ
net return after paying back financiers
Since R > 1, the bank borrows and invests to the max (intuition?):
1
ρ = ρ and I = N.
1−ρ

Implication: Net worth channel of investment
1
I = N.
1−ρ
Key implication: Greater bank net worth, N, raises investment, I .
This is known as the net worth channel of investment.

The logic of the result is in fact more general. As long as:
1 The borrower has profitable projects (captured by R > 1),
2 The projects have limited pledgeability (captured by ρ < 1).
Then, it is reasonable to think that the borrower’s net worth, N (more
broadly, internal funds) would affect its investment. Why?

A caveat: What if banks’net worth could grow?
The more general interpretation also raises a warning fiag.

In a dynamic setting, banks would accumulate net worth (e.g., by
realizing returns from past investment).
Eventually, their net worth and investment could be so high that they
could run out of projects, R , 1 (already invested in all).
If this happened, additional N wouldn’t affect investment. Why?
These considerations are built into more sophisticated models.
Let’s look at some results (pictures) from Brunnermeier-Sannikov
(AER, 2014), “A Macro Model with a Financial Sector.”

Courtesy of Markus K. Brunnermeier and Yuliy Sannikov. Used with permission.

Net worth channel in a dynamic setting
η in the Bru-San model is the analogue of N in the static model.

q is the analogue of investment, I .
R −1
θ captures the marginal value of additional net worth (~1 + 1−ρ ).
The second plot illustrates diminishing returns: As η ↑ η ∗ , the banks
run into diminishing returns (θ ↓ 1 similar to R ↓ 1).
Banks’assets are subject to shocks, so η moves around...

Net worth channel in a dynamic setting
When η hits η ∗ banks pay out dividends to their owners (who value
receiving them). Don’t need funds, so might as well pay out.
This keeps η ≤ η ∗ . Within this range, there are two regimes:
1 Most of the time (normal times) η remains around η ∗ .

Is the sensitivity of investment to η higher or lower in this range?
2 But negative shocks can lower η substantially below η ∗ :
Is the sensitivity of investment to η higher or lower in this range?

Courtesy of Markus K. Brunnermeier and Yuliy Sannikov. Used with permission.
(In practice, η ∗ might also be even lower for reasons outside Bru-San).

Back to the static net worth channel
1
I = N.
1−ρ
More broadly, imagine this equation as saying: Net worth affects

investment, especially if the bank net worth is suffi ciently low.
What are some policy implications of the net worth channel?

Implications of the net worth channel
1 Financial shocks that lower B’s net worth lower investment.

Shocks to Bs assets, e.g., subprime shock, can trigger a crisis.
2 Heterogeneity and distribution of wealth matters.
Transfer of wealth from Fs to Bs raise investment– especially during a
severe crisis that depletes B’s net worth. Why?
A bailout can be thought of as a transfer from Fs to Bs.
So the NW channel provides one justification for bailouts.
But is the channel true? Let’s look at some empirical evidence.

Roadmap
3 The credit crunch

Courtesy of Gabriel Chodorow-Reich. Used with permission.

The graph plots the rate at which banks lend to one another.
Imagine this as a (very rough) empirical counterpart of R or θ.
When do we expect the net worth channel to be more relevant?
Recall also that banks indeed realized losses in 2007 and 2008.
What would the net worth channel imply for banks’investment– i.e.,
their loans to firms or consumers– over this period?

Ivashina-Scharfstein (2010): Bank lending declines
Courtesy of Elsevier, Inc., http://www.sciencedirect.com. Used with permission.
Syndicated loans: Large loans originated by one bank and held by

multiple banks. Main source of bank loans for large corporations.

Ivashina-Scharfstein (2010): Bank lending declines
Syndicated loans provides as an empirical counterpart to I .

Total syndicated loans fell by 79% from 2007Q2 until end of 2008.
They fell by 47% at the peak of crisis, 2008Q4 (Lehman shock).
The timing is quite consistent with the net worth channel.
Chodorow-Reich (QJE, 2014): Further analysis with more data:
1 Variation in the lending of individual banks: Did banks that were

more exposed to the (negative) subprime shock reduce lending more?
2 What happens to the economic activity of firms whose credit is cut?

CR: Aggregate lending declines (replicating IS)

CR: Individual lending is consistent with the NW channel

CR: Individual lending is consistent with the NW channel
2nd column: More exposed to subprime CDOs =⇒ Lower lending.

3rd column: Lower trading revenues or greater charge-offs (realized
losses) on real estate assets =⇒ Lower lending.
(Ignore the 1st column and the deposits entry of the 3rd column for
now. These will make a come back next week.)
Consistent with the net worth channel (though not conclusive).
The more important part of Chodorow-Reich concerns firms’activity.

To address this, we have to expand our “model” to think of firms...

Roadmap
3 The credit crunch

Incorporating firms into the analysis
We haven’t explicitly modeled firms– only Bs and financiers.

Recall banks intermediate credit between financiers and firms.
Consider a firm who has a relationship with a particular bank.
Assume:
A1. The firm doesn’t have suffi cient net worth or internal funds of its own.
A2. The firm cannot directly borrow from financiers (see the more general
model in H-T for a relaxation of this assumption).
A3. The firm cannot easily switch banks and start a new relationship.

Credit crunch: Firms lose bank financing and cut activity
This type of firm must borrow from its bank to spend.

What would happen to the firm’s borrowing as its bank loses NW?
What about the firm’s economic activity (investment, employment...)?
These effects are known as the credit crunch (firms’perspective).


Credit crunch “reduces” firms’borrowing
ΔLi ,s is the total amount of loans made (to all firms) by the bank s
from which firm i received syndicated loan prior to Lehman.
−ΔLi ,s can be thought of of as a measure of credit crunch faced by i.
Results show: Credit crunch reduces the firm’s probability of
borrowing. Supports A3 above: Firms can’t easily switch to another
bank.
The effects are economically large (see the last two lines).


Credit crunch “reduces” firms’employment
Firms whose pre-Lehman banks cut back lending more (e.g., due to a
greater drop in net worth) reduced employment more.
The effects are economically very large (last two lines).
CR finds (rough estimate) credit crunch can account for 1/3-1/2 of
the decline in employment in small-medium firms between 2008:3 and
2009:3.
There are concerns with causality (as usual) but still very interesting.

Differential effects on small and large firms
CR also finds the effects of a credit crunch differ by firm size.

Significant for small and medium firms, but not for the largest firms.
This finding can also be reconciled with the general model in H-T.
Large firms have a plausible alternative to bank financing: They
might be able to borrow directly from financiers.
E.g., IBM can issue and sell bonds to raise funds.

This might not be feasible for small firms– they need bank financing.
So small and medium size firms would arguably be affected more.
H-T formalize this by allowing for direct financing (relaxing A1-A2).

Differential effects on small and large firms
In the raw data, employment of small firms fell much more.
CR: Much of this drop can be “explained” by the credit crunch.

Roadmap
3 The credit crunch

How about firms’borrowing constraints?
So far, we emphasized banks’borrowing problems as being central.

But firms might invest to some extend with their own funds, N firm .
Large firms, such as IBM, might also be obtain direct financing.
H-T model considers both firms’and banks’borrowing constraints.
Firms face borrowing constraints, ρfirm < 1, just like banks do.
Their net worth, N firm , can also affect investment, similar to above.

The net worth channel more broadly
The more general insight is: The net worth of the borrowing sector
of the economy matters for economic activity.
Banks’net worth, N bank , declines: Credit crunch.
Firms’net worth, N firm , declines: Collateral squeeze.
The shocks to N bank are typically amplified more (next lecture).
But shocks to N firm could also matter and trigger a financial crisis.
Next: A different source of evidence that is consistent with some version

of the net worth channel– without pinning its exact location.

An alternative source of evidence: CFO surveys
Campello, Graham, and Harvey (2010): Survey of 1050 CFOs in 39

countries conducted in Fall 2008: US (574), Europe (192), and Asia
(284).
Questions about financial constraints and future plans.
Results:
All firms planned cuts in investment over this episode (no surprise).
Firms that report to be financially constrained planned much deeper
cuts in investment– consistent with borrowing constraints (a drop in
N firm or N bank or both).

In Fall 2008, firms across the globe planned to reduce
various forms of investment (and employment)
Courtesy of Murillo Campello, John Graham, and Campbell R. Harvey. Used with permission.
Figure displays all firms’(not just constrained) change in the policy

variable (% per year) as of Fall of 2008.
Firms reported they faced borrowing constraints
Borrowing constraints: Are you affected by diffi culties in accessing

credit markets?
In the US sample: 244 indicate unaffected, 210 indicate somewhat
affected, 115 indicate very affected.

Main result: Constrained firms in the US planned much
larger cuts than unconstrained firms
They control for many aspects of the firm except for CFOs report of
financial constraints and find similar results.

Taking stock: Transmission mechanisms
So why care about banks’problems (or financial shocks in general)?

Because they affect economic activity via the net worth channel:
Borrowing is constrained (due to frictions) so borrowers’internal
funds/net worth matters.
Decline in banks’net worth =⇒ Credit crunch for firms.
Firms’net worth also matters (albeit to a lesser extent).
Channels supported by empirical evidence from the recent crisis.
Tomorrow: Amplification mechanisms. Read the LTCM case study.

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
53
Lecture 3: Leverage, Fire Sales, and Amplification
Mechanisms
Alp Simsek
Alp Simsek () Amplification Mechanisms 1

Crises and amplification mechanisms
• Banking crises are often triggered by events that seem

"small" in retrospect.
• In the recent crisis, estimated initial losses in the U.S.
subprime market were less than $500 billion. This is not
too large: comparable to the losses in the U.S. stock
market on some bad days.
• But these losses triggered a worldwide financial crisis.
They were associated with extremely large declines in
economic activity (see Blanchard (2009, "The Crisis: Basic
Mechanisms and Appropriate Policies").
• This lecture: How (and when) do small financial shocks

have amplified effects?

Roadmap
1 Amplification from high leverage
2 Amplification from procyclical leverage
3 Fire sales and amplification
4 The LTCM crisis of 1998

Leverage and amplification mechanisms
Recall the Holmstrom-Tirole model from the last time

N0
Invest I0 = by borrowing ρ0 I0 .
1 − ρ0
We put the time subscripts to emphasize the timing. Bank made

investments at date 0 and returns are realized at date 1.
The realization, R1 , will induce a new net worth for banks, N1 .
Suppose banks will reinvest at date 1 given N1 , denoted by I1 .
New question: How would the realization R1 affect N1 , and thus I1 .
Let’s work out an example with initial N0 = 1 and ρ0 = 0.95...

The bank’s initial balance sheet
The bank’s initial leverage ratio is high, 1/ (1 − ρ0 ) = 20.

The bank’s realized balance sheet
Consider the balance sheet at date 1 after R1 is realized.

Leverage amplifies losses (and gains)
So in this example, we have
N1 = 20R1 − 19.
If R1 = 1, then bank breaks even and N1 = N0 = 1.

Suppose instead R1 = 1.01. What is N1 in this case?
Suppose instead R1 = 0.99. What is N1 in this case?
1% change in R1 has a large (20%) effect on N1 . Why?

Leverage amplifies losses (and gains)
More generally, the bank’s realized net worth can be written as,
R1 − 1
N1 = R0 I0 − ρ0 I0 = 1+ N0 .
1 − ρ0
1
Note that high leverage, 1−ρ0 , amplifies gains but also losses.
Going beyond the math, what is the economic intuition for this result?

Leverage amplifies due to debt being a constant promise
The bank’s debt is fixed regardless of its realized returns R1 .

It has to pay ρ0 I0 regardless of returns R1 are high or low.
Put differently, all the changes in returns are absorbed by N1 .
This feature of debt creates amplification of losses (and gains).
But should Fs’claims be necessarily be fixed regardless of R1 ?

So far, we simply assumed it. This is a deep issue. Will come back.

Leverage amplifies initial losses (and gains)
The drop in banks’net worth, N1 , reduces its new investment, I1 .

Suppose the same model is repeated at date 1 so that,
20% drop 20% drop 1% drop
�� 1 ��
I1 = N1 , where N1 = 20 R1 − 19.
1 − ρ1
1% drop in R1 translates into an amplified drop in N1 .
This translates into an amplified drop in I1 . Why?
In practice, maximum leverage ρ1 also tends to be procyclical...

Roadmap

© Dow Jones & Company, Inc. All rights reserved. This content is excluded from our Creative

In practice, leverage also tends to be procyclical
The bottom panels illustrate a measure of ρ1 for different assets.

Leverage ratios seem procyclical in the sense that they are high in
good times (with high R1 ) but low in bad times (with low R1 ).
Geanakoplos (2009), “The Leverage Cycle” proposed a theory of
procyclical leverage based on changes in uncertainty.
Bad times =⇒ Uncertainty =⇒ Nervous lenders=⇒ Less leverage.
We will come back to and formalize this argument next week.
For now: Procyclical leverage creates further amplification. Why?

Procyclical leverage creates further amplification
Going back to our example, suppose ρ1 = 0.9 < 0.95 after the loss.
60% drop 20% drop
��
z}|{ 1 z}|{
��
I1 = N1 , where N1 = 20R1 −19.
1 − ρ1 (R1 )
|� ��
{z �}
drops from 20 to 10 =⇒ 50% drop
Procyclical leverage ratio creates further amplification!
Borrowing becomes more diffi cult (ρ1 collapses) precisely when banks
make losses and need to borrow the most!

Roadmap

Another source of amplification: Fire sales
Another problem in practice is that R1 is also endogenous. Many

banks (or bank like institutions) invest in financial assets. For
such assets, the realized return can be written as,
Q 1 + D1
R1 = .
Q0
Here, Q0 , Q1 denote the price and D1 denotes dividend/payoff at 1.
Imagine buying the asset at date 0, receiving D1 , and selling the asset
at date 1 to cash out. Then your return would be R1 .

Another source of amplification: Fire sales
In financial markets, prices Q0 , Q1 are typically “endogenous.”

This makes R1 endogenous: What the bank does can affect R1 .
In particular, low investment I1 can lower Q1 , and ultimately, R1 .
This amplification channel is known as fire sales.
To illustrate, we need a little bit of background on asset pricing...

Forced mango sales would reduce the mango price
Can we draw an analogy from goods to assets? In theory? In practice?

Basic asset pricing theory: Forced sales don’t matter
Fair value, Q1 , is present discounted value of future dividends/payoffs.

The discount rate incorporates risk, but is largely unchanged with sale.

Asset pricing in practice: Somewhere in between
Asset prices in practice are somehwere in between. Imagine

Q1ideal , Q1old , Q1new are different but close to one another.
There are (various) advanced asset pricing theories that capture this.

Fire sales: Considerable market impact (larger than usual)
The result heavily relies on specialists being out. Why are they out?

Why are specialists out? Common shocks/distress
Shleifer-Vishny (1992): “Liquidation Values and Debt Capacity...”
First to emphasize/formalize fire sales. Illustrate with a parable:

Consider indebted farmer with low current cash fiow.
Cannot reschedule debt or borrow more.
Must liquidate (sell) the farm to pay back its creditors.
Potential buyers:
High valuation (specialists): Neighbor farmer.
Low valuation (non-specialists): Convert to baseball field
Neighbor is simultaneously distressed (industry wide shocks) and
thus she is also borrowing constrained.
The farm is sold to low valuation user at a fire sale price (much
lower than the value at best use, Q1ideal ).

Fire sales happen in real assets such as farms or houses
Campbell, Giglio, Pathak (AER, 2011), “Forces Sales and House

Prices.”
They analyze house sale prices in Massachusetts between 1987-2009.
They investigate sale prices around events that might plausibly
involve fore sales: the death or bankruptcy of owner, the foreclosure
by a bank.
They compare the sale price relative to price of comparable houses...

Courtesy of John Y. Campbell, Stefano Giglio, and Parag Pathak. Used with permission.

Timing of plausibly forced house sales

Plausibly forced sales are associated with lower prices
The timing for bankruptcy sales is especially indicative of a fire sale

mechanism (as opposed to other factors, e.g., poor maintenance).

The price discount is particularly large for foreclosures
Especially large discount for foreclosures. This supports fire sales, but
read the paper for other contributing factors (vandalism etc).
How about financial assets? Slow moving capital
Fire sales can also happen in financial assets such as bonds, CDOs...
With financial assets, another contributing factor to fire sales is that
there might be few specialists (or neighbors) to begin with.
This is especially the case for niche and complex financial assets that are
harder to price (require more effort/research etc).
We expect there to be enough specialists to absorb “reasonable”sales
or purchases that happen in regular days.
But an unusually large sale could create havoc.
Other specialists would eventually come in, but this takes time.
This is known as slow-moving capital (referring to specialist capital).

Evidence on fire sales/slow moving capital
Empirical studies show there can be fire sales also in financial markets.
Next slide is an illustration from Mitchell, Pedersen, Pulvino (2007).
Convertible bonds: Complex asset with a formula for fair valuation.
Convertible hedge funds: Specialize in valuing these assets.
In 2005, they had to lower their positions due to financial problems.
How did these fire sales affect the price of convertible bonds?

Courtesy of Mark Mitchell, Lasse Heje Pedersen, and Todd Pulvino. Used with permission.
Convertible arb HFs reduce their positions due to losses and redemptions.

Courtesy of Mark Mitchell, Lasse Heje Pedersen, and Todd Pulvino. Used with permission.
Other investors did not immediately step in (since highly specialized) and
the price fell below the theoretical value for an extended period.
Multi-strategy HFs eventually step in (previous slide) but takes time.
Fire sales generate further amplification
To learn more about fire sales, read the survey by Shleifer-Vishny.

Kiyotaki and Moore (1997), “Credit Cycles” illustrate that fire sales
can generate further amplification.
We can illustrate this in our framework. We had the example,
I1 = N1 where N1 = 20R1 − 19.

1 − ρ1
We also mentioned that

Q 1 + D1
R1 = .
Q0

Now let us add the assumption that asset prices are given by:
⎛ fire sale infiuence
⎞
�z ��
}| �{
Q1 = Q1 ⎝1 + GQ (I1 ) ⎠.
��
|{z}
factors exogenous to our model
Here, GQ (·) is an increasing function. Low I1 captures forced asset

sales by the bank, which lowers GQ (I1 ) and the price.
Normalize so that if bank were to break even at that 1 (so that
N1 = N0 ), then we also have GQ (I1 ) = 0 and Q1 = Q 1 .

Taking Q0 , D1 as constant (not our focus) this also implies,
supply-demand infiuence/fire sales
D1 + Q1 z� }|
�� {�
R1 = = R1 + G (I1 )
Q0 ��
|{z}
exogenous factors
Q 1 G Q (·)
for some constant R 1 and for some function G (·) = Q0
.
Note that G (·) is also an increasing function. What is the intuition?
Earlier analysis is special case with G (I1 ) = 0 and unresponsive to I1 .
The responsiveness of G (·) captures the severity of fire sales.
What is the effect of a 1% shock on net worth in this case?

⎛
1% drop =⇒ ⎞
large drop (why?)

��
z}|{ ��
z}|{
⎜ ⎟
N1 = 20⎝ R 1 + G (I1 ) ⎠ − 19
.
|� ��
{z }�
More than 1% drop. Why?
|� ��
{z }�
More than 20% drop. Why?.
What is the intuition behind the bottom two lines?

So fire sales create further amplification.
The drop in N1 further reduces investment I1 (why), repeating the
mechanism and triggering downward spiral...

Fire sale and NW channels trigger a spiral

Summary of amplification mechanisms
Putting everything together, we have (for the example):
1 1
I1 = N1 = 20 R 1 + G (I1 ) − 19 .
1 − ρ1 1 − ρ1 R 1
This illustrates three amplification mechanisms (find them!):
1 Leverage generates amplified losses (or gains).

2 Procyclical leverage, ρ1 R 1 , generates further amplification.
3 Fire sales, G (I1 ), generate further amplification.

Figure: From Brunnermeier (2009), “Deciphering The Liquidity And Credit

Crunch 2007-2008”.
Next: Case study (LTCM) that illustrates some of the mechanisms.

Using Jorion (2000), “Risk Management Lessons from LTCM.”

Roadmap

A cautionary tale of high leverage: LTCM
The hedge fund, LTCM, was founded in 1994 by star trader John
Meriwether and other traders who left Salomon Bros. after 1991.
Bob Merton and Myron Scholes (Nobelists) joined them.

They did relative value arbitrage: Find two very similar assets, buy
the cheap one, and (short) sell the expensive one:
On-the-run vs. off-the-run Treasuries,
Mortgage-backed securities (MBS) vs. treasuries,
High-yield vs. low-yield bonds in the Euro area.
These strategies make profit if the prices of two assets converge.
But they (temporarily) make losses if prices diverge further. Why?

LTCM had great success in early years
LTCM strategies deliver high returns only if leveraged.

Capital $5-7 billion in 1996-97.
Total assets about $125 billion, so 25:1 leverage.
Fees were 2% of capital + 25% of profits ($1.5 billion in 1997).
Prices indeed converged and they had a great run for a while.
But this also meant they ran out of investment opportunities.
At the end of 1997, LTCM returned $2.7 billion to investors...

History of spreads
© John Wiley and Sons. All rights reserved. This content is excluded from our Creative

LTCM’s leverage and asset growth
Returning capital (lowering N0 ) brought leverage ratio back to 25.

In retrospect, this was not a very prudent thing to do. Why?
Crisis in 1998: The shock
Preliminary problems in May and June 1998 from falling MBS prices
(16% loss from end 1997).
August 17, 1998 Russia announced surprise debt restructuring.
“Flight to liquidity”: prices of most fixed income assets declined.
=⇒ Additional losses (by August, 52% loss).
LTCM had about N0 =$5B. Invested in about I0 =$125B.

How much should R1 fall (from 1) for LTCM to lose about 50%?
This is the first amplification channel we discussed!

Leverage amplified LTCM’s losses
After 50% loss, LTCM’s net worth is down to about $2.5B

These losses force LTCM to reduce its asset holdings:
I1 = N1 ' $62.5B, much smaller than $125B.

1 − ρ1
(This assumes ρ1 ' ρ0 ' 25. If leverage ratio was procyclical, then
the required reduction is even larger. Why?)

Raising capital is another possibility but not easy
As an alternative to reducing I1 , LTCM also tried to increase N1 by

raising capital, i.e., bringing in new owners to the fund.
We ruled this possibility out in our model.
Also diffi cult in practice, especially in times of turmoil.
LTCM sought additional capital but obtained none.
No-one wants to put money into a fund that just lost 52%!
It did not help that they force-returned capital earlier.
So LTCM has to reduce I1 , i.e., sell about $60B of its assets....

Large asset sales reduce prices further
Asset sales by LTCM further reduce prices. Why?

Price falls might have been accelerated by predatory trading.
Business Week (February 26, 2001): “If lenders know that a hedge
fund needs to sell something quickly, they will sell the same
asset– driving the price down even faster. Goldman Sachs & Co. and
other counterparties to LTCM did exactly that in 1998.”
Predatory trading might be another reason why specialists were out.

But this is somewhat speculative (Goldman denies allegations).
Even without this, we would expect the price to drop. Why?
The declining price falls further increase LTCM’s losses.
September 23 bailout organized by Federal Reserve Bank of New York
(92% loss).

LTCM dramatically loses almost all net worth

Taking stock: Some lessons from LTCM
High leverage creates fragility, even if you have a genius team!

Small price movements/unexpected events can generate large losses.
These losses induce fire sales and induce price drops/further losses.
Procyclical leverage might further exacerbate these losses.

Taking stock: Some lessons from LTCM
Courtesy of the U.S. Securities and Exchange Commission. This image is in the public domain.
A view of the subprime crisis: Many LTCMs failing simultaneously!

Next time: Why do financial institutions make losses?
But why did LTCM or other institutions make these losses?

We will continue our discussion of LTCM next time.
Finish reading the LTCM case. Also start reading the Bear
Stearns/JP Morgan case. Can read pages 1-8 (until the section on
financial stresses).

Taking stock: Amplification mechanisms
There are more amplification mechanisms that we will see next week.
Leverage, procyclical leverage, fire sales are important ones.
Also help us understand why shocks to banks particularly damaging.
Banks are leveraged and subject to fire sales (specialized assets).
Tech bubble bust much milder since investors in tech stocks (like you
and me) are not highly leveraged. Losses are contained.
Subprime crisis much more severe since the losses are amplified!

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
53
Lecture 4: Understanding Banks’Losses: Moral Hazard
or Mistakes
Alp Simsek
Alp Simsek () Understanding Banks’ Losses 1

Why do financial institutions make losses?
Key question for policy: The causes of bank losses.

Two competing hypotheses:
1 Mistakes: Optimism and neglected risks
2 Moral hazard (of various forms).
Today: Formalization, and comparison in the context of the recent crisis.

Let us set the stage by analyzing LTCM’s losses...

Roadmap
2 Moral hazard and reckless risk taking
3 Empirical evidence on insiders’beliefs
4 Revisiting moral hazard: Franchise value
5 Revisiting moral hazard: Looting

What caused LTCM’s losses?

LTCM’s strategy: Relative value arbitrage
Recall that LTCM used relative-value trades with high leverage.

Find two very similar assets, A and B, that trade at prices pA < pB .
Buy A and sell B. Cash in when the prices converge.
But shouldn’t similar assets already trade at similar prices?
That would be the case in an ideal world, but reality is a bit messier...

Relative arbitrage opportunity: Siamese twins
Lamont and Thaler (2003), “Anomalies: The Law of One Price...”

Royal Dutch and Shell are stocks of the same company Royal
Dutch/Shell.
Courtesy of American Economic Association. Used with permission.

Relative arbitrage is risky business
Prices can be driven apart by preferences of investors for one stock

over the other (e.g. because one stock is an index constituent and the
other is not).
Arbs like LTCM can try to profit from this. But not completely
riskless.
It would be riskless if the arbs could hold the position forever.
But HFs have quite short horizons (redemptions in case of losses).
As Keynes said, “in the long run we are all dead” (including the Arbs).
So the risk is that mispricing will worsen before prices eventually
converge.
The risk is quite serious for stocks (as you see from the graph)
because there is no fixed date at which we can anticipate correct
valuation.
Bonds that have fixed maturity are a little less risky but still risky.

LTCM’s strategy: Relative value arbitrage
The success of LTCM’s trades relied on two principles:

1 Hedging: Correlations within a pair, (A, B), being high. Why?
So that pA and pB move together and pA − pB does not fiuctuate
much.
2 Diversification: Correlations across different such pairs being low.
This ensures that “you did not put all your eggs in a single basket.”
Risks can be much larger if these correlations are mismeasured, or if

they change.

LTCM trades not well hedged
It seems that correlations within pairs were overestimated.
This leads to much larger risks than thought: A reduction in long

asset prices not matched by a reduction in short asset prices leads to
large losses in view of leverage.
LTCM trades not well diversified
Returns explained by a single factor =⇒ Correlations across pairs

were underestimated.
So why did LTCM make losses?
So LTCM made mistakes– large blunders in risk management.

Narrative accounts, e.g., Jorion, suggest that this was due to a form
of optimism: They neglected subtle risks (too much focus on recent
data etc.)
But there is a more ominous alternative: Perhaps LTCM was careless
because it anticipated a government bailout.
Let us formalize the bailout argument and contrast with mistakes...

Roadmap

How about moral hazard?
The moral hazard arguments take various forms:
1 Compensation contracts: Poor incentives for traders.

2 Borrowing contracts: Risk taking at the expense of financiers.
3 Government guarantees: Poor incentives for the institution as a
whole.

How about moral hazard?
1 and 2 are always an issue (MH is a fact of life), which is why the market
has also devised various checks against them:
1 Traders’activity monitored by managers/risk management divisions...
2 Risk taking monitored by banks, restricted by collateral/debt
covenants...
(Recall the H-T model: MH restricts ρ but does not result in risk
taking!)
The government has no comparative advantage in dealing with 1 and 2.
Focus on 3: Moral hazard caused by the government’s presence itself.

A framework for thinking about moral hazard
For simplicity, suppose ρ = 0: The pledgeability is so low that the

bank invests only its own money. We will discuss ρ > 0 later.
The bank insiders start with N0 and choose the type of project:
Safe Risky
High state R1 R1H > R 1 .
Low state R1 R1L = 0
The normal project gives R 1 regardless of the state
The risky project returns higher in H but lower in L.
R1L = 0 for simplicity, but can also imagine as capturing amplification.

Let π ≥ 0 denote probability the bank assigns to low state L.

Interpretation of the framework
One interpretation for the recent crisis:

“Safe”: Holding prime mortgages or other safe investments.
“Risky”: Holding subprime mortgages.
Low state: A large decline in the nation-wide house price index.
(See Gerardi, Sherlund, Willen (2008), “Making Sense of the Subprime
Crisis.”: They claim Banks understood that a price drop would create
losses, but they underestimated the probability of a large decline.)
An equivalent interpretation:
“Safe”: Holding subprime mortgages plus CDS insurance.
“Risky”: Holding subprime mortgages without CDS insurance.
You can also provide similar interpretations for the LTCM episode.

Warm-up to MH: Without government, banks behave
To formalize moral hazard, suppose the parameters are such that
R1H − R 1
R 1 > (1 − π) R1H or equivalently π > . (1)
R1H
Banks assign suffi ciently high probability to L that, absent government

intervention, it is not profitable to keep a risky balance sheet.
So without government, banks don’t take risk. No crisis!
Let us introduce government into the analysis.

Bailouts: Government injects funds into banks
Imagine state L is realized so that the bank’s net worth is, N1L = 0
What happens to banks’investment, I1L ? What happens to loan to
firms?
Suppose the government can transfer wealth from financiers to banks.
Would the government want to do that? Why?
Bailout: Suppose the government transfers money to bank in state L,
which raises the bank’s effective return to some R1L,bail > 0.
The transfer is financed by taxing financiers.
(In practice, bailouts work somewhat differently. Will come back.)

With government transfer: Banks might misbehave
Suppose the parameters also satisfy,
R 1 < (1 − π) R1H + πR1L,bail .
The bailout is suffi ciently large to make the risky project attractive.
Anticipating the government bailout, the bank chooses to take risk.
Government bailout creates (or at the very least exacerbates) crises!
This is also known as “heads I win, tails you loose”principle.

A strict no-bail-out-ever policy could prevent crises
If we could commit to setting R1L,bail = 0, the crises would be averted!

This type of commitment is diffi cult since it would be very costly for
the economy in state L. In economics lingo, the policy is
time-inconsistent.
But suppose you could get around time-inconsistency, say by enacting
laws that make bailouts very diffi cult. Would you want that?
What are some potential problems with this no-bail-out-ever policy?

The policy is not robust to the alternative: Mistakes
Policy is not robust to the presence of neglected risks/mistakes.

In fact, neglected risks/mistakes provide a natural alternative to MH.
To illustrate this alternative using the above framework, suppose the
converse of assumption (1) holds, so that
R1
R 1 < (1 − π) R1H or equivalently π < 1 − .
R1H
The insiders believe state L is so unlikely that holding risk is profitable.

They choose “risky” even if R1L,bail = 0– since they view L as unlikely.
They choose risky if they expect R1L,bail > 0. Whether there is a
bailout has a small impact on the bank’s decisions since π is low.

Roadmap

How to test optimism vs moral hazard?
Key difference between the two explanations is the bank’s belief
about crisis, π.
Moral hazard: high π, deliberate risk taking. Mistakes: low π,
neglected risks.
In the subprime context: The probability of a nation-wide decline in

house prices.
Shouldn’t banks have seen the collapse of the housing bubble coming?

Wasn’t the housing bubble obvious?
How could the banks have missed this? Surely, π must be large.

Beware the wisdom-after-the-fact
The bubble is obvious only with hindsight. At the time, few people
anticipated a large nation-wide decline in house prices.
Watch “The Big Short”. How many pessimists were there? How were
they treated by others– the conventional wisdom?
During the boom phase, there are always justifications for high prices.
Reinhart and Rogoff (2009) call this “This Time is Different.”
For a narrative account, read the paper by Gerardi, Foote, Willen
(2010): “Reasonable people did disagree: Optimism and pessimism
about the US housing market before the crash”
We need to do something more systematic to gauge the insiders’π.

Lessons from Titanic: A AAA-rated ship
From Shleifer (2011), AFA address:

When built, Titanic was described as the safest, largest ship ever.
Insiders and financiers were on board: They believed it was safe.
Radio operators ignored warnings of icebergs nearby.
Many lifeboats on board, enough for 1/3 of passengers. Consistent
with regulation.
1500 people died. Some rescue boats were not full. Almost all the
crew died.
The insiders of Titanic, as well as the regulators, seemed to have low π.

Were insiders on board during the subprime crisis?
A similar analysis can help us to gauge π during the subprime crisis.

Look at insiders’own portfolios, i.e., what they did with their money.
Moral hazard, high π, suggests would be careful with own portfolio.
Mistakes, low π, suggest would take risks also with own portfolio.
Cheng, Reina, Xiong (AER, 2014) analyze personal housing
transactions of midlevel managers in securitized finance (CDOs etc)
in 2004-2006.
They use equity analysts and lawyers as comparison groups...


Divestiture: Sale of a home (could be second or first home),
Securitization managers were less likely to sell in 2004-2006.

They were also more likely to buy a second home, or swap up into
more expensive home, well until 2008.
The results are robust to various econometric checks.

Some insiders seem to be on board during the crisis
This suggests securitization managers were at least as optimistic

about housing
o as the general
o public,
E insiders Q1house ≥ E public Q1house .
This rules out π ' 1: Insiders didn’t see an imminent crash.
It does not definitively prove that π is low (which is not easy).
But reconciled more easily with π being low than π being high.

How about top management?
Ma (2013): a similar exercise with bank CEOs.

Main idea: More optimistic CEOs are less likely to exercise their
bank’s stock options, as well as to sell their stocks, essentially
speculating on their bank’s success.
Proxy for optimism by using CEOs’exposures to own bank’s equity
in 2002-2006.
Mistakes/neglected risks suggest: Banks with more optimistic CEOs
(indicative of low π) would have greater real estate exposures before
the crisis, and make greater losses during the crisis.
A case study that compares two banks, US Bancorp (less optimistic CEO)
and SunTrust (more optimistic CEO), illustrates Ma's more systematic
results.

Courtesy of Yueran Ma. Used with permission.


More optimistic CEOs’banks increased real estate loans by more.

Stock prices of the banks with more optimistic CEOs fell by more.
The more on board CEOs were, the deeper the ship did sink!

Most insiders seem to be on board during the crisis
The evidence on beliefs is more consistent with mistakes (due to

optimism and neglected risks) than moral hazard.
There are also other, more theoretical, reasons to be skeptical of the
earlier version of the moral hazard argument.

Roadmap

Problems with the basic moral hazard argument
The basic MH argument is too extreme for a couple of reasons:
1 Owners/shareholders do not necessarily benefit from a bailout...
Read the HBS case study of JP Morgan and Bear Stearns.
What happened to Bear’s shareholders’wealth after the bailout?
The bailout did benefit Bear’s creditors/financiers. Could formulate a
more sophisticated version with ρ > 0, and financiers make low
interest loans to Bear.
More reasonable. But still requires high π (by financiers as well
insiders).

Problems with the basic moral hazard argument
2. Dynamic considerations might push against reckless risk taking.
Franchise value: Future stable profits due to banking services.

Banks’such as Bear Stearns and JP Morgan build franchise-reputation
over time to provide (highly profitable) intermediation services...

Bear Stearns’revenue breakdown
© Harvard Business School. All rights reserved. This content is excluded from our Creative

Bear Stearns revenue breakdown
The revenues from fixed income (CDOs etc) are the largest.
However, there are sizeable revenues also from other divisions, e.g.,
investment banking, clearing services, wealth management...
If Bear didn’t make losses in fixed income portfolio, it would
presumably continue to make these other revenues year after year.
Think of the franchise value V F as the present discounted value of
these revenues.
After a failure and bailout, Bear insiders lose V F . They are transferred
to JP Morgan (to the extent that they did not disappear altogether).
Would franchise value make moral hazard more or less severe?

Franchise value can provide discipline against MH
Think of the Bear management as comparing V F + R 1 N0 , with
(1 − π) R1H N0 + V F + πR1L,bail N0 .
Choose the safe action as long as

net profits from fixed income (one-time)
πV F > (1 − π) R1H + πR1

L,bail − R 1 N0
potential loss of franchise profits (forever)
When π > 0, then V F pushes strongly against risk taking. The moral
hazard argument more diffi cult to sustain.
When π ' 0, V F has little effect. Mistakes due to neglected
risk/mistakes can cause crises even with high V F .

Why did franchise value not discipline Bear Stearns?
Page 12 of Bear Stearns case: “Within the bond business,...,key

lieutenants were in fierce disagreement over how best to manage the
extent of Bear’s mortgage related securities holdings. Bear’s head of stock
sales and trading, as well as the company’s head of propriety trading,
argued that the head of Bear’s mortgage division needed to reduce his
holding: “Cut the positions, and we’ll live to play another day,” said the
head of proprietary trading. Schwartz, however, was reluctant to unload
billions of dollars worth at prices that seemed to be unreasonably low and
possibly not refiective of their true value.”
This paragraph illustrates the crux of the franchise value argument.
And how FV discipline can be lost due to mistakes by top
management.
Schwartz replaced Jimmy Cayne as CEO in January 2008.

JP Morgan revenue breakdown
© Harvard Business School. All rights reserved. This content is excluded from our Creative

Did franchise value discipline JP Morgan?
The franchise value items appear to be even greater for JP Morgan.

This could be one reason why, Jamie Dimon (CEO), insisted on more
careful risk management and a “fortress balance sheet”.
Liquidity/capital exceeded regulatory requirements (and
competitors’)!
Note also that JP Morgan acquired Bear assets (its franchise value)
cheaply: There are rewards to being strong in a crisis!
This provides further discipline against moral hazard. Why?

Did Dimon see the crisis coming?
Even JP Morgan made some losses in the mortgage market.

Page 7: “Looking back, Dimon noted that, “Our biggest mistake was
assuming that home prices would go up for a decade without losses.
We loosened up our standards. Once prices stopped rising, losses
mounted. We need to write a letter to the next generation saying
that there’s a reason why we loan only 80% LTV on a home.”
So even Dimon did not fully see the crisis coming.
But he seemed to have the sense that something could go wrong (π
relatively high) in a complex economic environment and company.
Ma (2013): Heterogeneity in π also refiected in CEOs’own portfolio...

Ma (2013): “The patterns in Table 10 are also consistent with (the

earlier results for commercial banks). CEOs at the worse-performing
banks generally had larger ex ante increases in equity holdings, while
CEOs at the better-performing banks had smaller increases.”
The exception is Dick Fuld of Lehman Brothers.

Taking stock from franchise value and MH
V F provides discipline against moral hazard, especially when π is high.

Makes it harder to have crises with high π and reckless risk taking.
This also suggests moral hazard can be a real concern when V F is
low.
V F could be low, for instance, if the bank is almost sure to fail
regardless of its actions (it had already made big mistakes and too
late to save).
In these cases, a more sinister version of MH can apply: Looting.

Roadmap

Looting: An extreme version of MH
Looting (loosely): Extracting value directly (paying dividends to

owners, transferring money to friends etc) from a bank that is very
likely to fail.
Extreme MH: Transfer at date 0 as opposed to state H of date 1.
Akerlof and Romer (1993): Looting might be relevant for the Savings
and Loan crisis of 1980s (as well as other crisis episodes in other
countries).

Looting/MH during the S&L crisis?
Many thrifts (small banks) were or close to being bankrupt in 1980s.

They were also not profitable due to increased competitiveness earlier.
These observations suggest that their V F might have been quite low.
Regulatory action was delayed, and the ultimate cleanup happened
only in late 1980s and early 1990s. MH is a real concern.
Combined with a lack of accountability, looting could well be an issue.
Lessons:
Beware of moral hazard in situations with low franchise value.
MH during a crisis (as opposed to before) might be a bigger concern.
Swift regulatory action/bailout can actually mitigate moral hazard!

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
Lecture 5: Maturity Mismatch and Bank Runs
Alp Simsek
Alp Simsek () Lecture Notes 1

Liquidity and financial panics
So far: Focus on leverage and amplification channels.

Our story is missing a key feature of crises, “panic.”
Panic can be thought of as a sudden shift in financiers’behavior that
induces them to reduce their lending to banks/institutions.
Thus, the spotlight on the next few lectures will be on Fs:
1 They might need “liquidity” (Diamond-Dybvig model)
2 They might seek “safety” (Holmstrom-Gorton-Geanakoplos...)
3 They might be institutions, with own problems (Shin-Caballero...)
We will formalize 1 today, and will come back to 2 and 3 later.

Roadmap
1 Maturity mismatch and illiquidity
2 Liquidity pooling and bank runs
3 Potential solutions to bank runs
4 Applying bank runs in modern financial markets

Two mismatches: Risk and maturity
The instructor's calculations based on the Federal Reserve Statistical Releases.

Two mismatches: Risk and maturity
Bank balance sheets can be viewed as featuring two mismatches:
1 Risk mismatch: Relatively risky assets, relatively safe liabilities/debt.

We saw how this can create fragility (especially with high leverage).
22 Maturity mismatch: Longer-maturity assets, shorter-maturity
liabilities....

Maturity mismatch
Bank assets typically make small payments spread over a long

horizon. Imagine receiving 30-year mortgage loan from the bank.
Bank liabilities typically make promises over a much shorter horizon:
Imagine making a deposit in a checking/savings account.
The discrepancy is known as the maturity mismatch.
The maturity mismatch also creates the possibility of illiquidity...

Liquidity of an asset: Ease of converting into cash
Recall that the value of a financial asset with a long horizon is,
Q1ideal ' Present discounted value of future payoffs.
Liquidity of a financial asset (broadly speaking): How easily can the

asset be converted into cash in a short amount of time.
How close a value can I get to Q1ideal over a short amount of time.

Liquidity is arguably low for nontraded assets
So liquidity is closely related to the fire sales that we discussed.

For traded assets (such as stocks, bonds, houses), fire sales is a
situation in which liquidity “dries up” relative to normal times.
For assets that are not regularly traded in financial markets, liquidity
is a greater concern. Hard to sell even in normal times.
Historically, bank assets (mortgages etc) not regularly traded.

Liquidity can also be low for traded bank assets
Trading costs dramatically declined in recent years (IT revolution etc).

This has converted more bank assets into marketable securities.
An example is securitization, which converts traditionally illiquid bank
assets (e.g. mortgages/loans) into marketable securities (CDOs etc).
Note that bank assets are typically debt-like (someone else’s debt).
Debt-like securities can be liquid in normal times, but their liquidity
can also dry up very quickly at times of uncertainty. Will come back.
Thus, illiquidity is arguably a concern even if bank assets are traded.

Danger zone: Maturity mismatch with illiquid assets
Combination of maturity mismatch and illiquid assets an be
dangerous.
Diamond and Dybvig (1983): Formal analysis of this danger zone:
1 Why do the banks feature a maturity mismatch to begin with?
2 How does this arrangement create fragility/runs/panics?
3 What should the policy should do about this?
Our goal today: Illustrate the DD ideas using a stylized model.

Roadmap

A model of banks and financiers
Consider a model with a number of banks (B) and a large number of

financiers (F) as before.
In this case, there are three periods, {0, 1, 2}.
Date 0 is when investments are done as before.
Dates {1, 2} (previously date 1) when returns might be realized.
Fs have resources at date 0 as before, say 1 dollar each.
Fs have no resources at other dates (for simplicity).
Bs have no resources (can incorporate N0 but not necessary).

New ingredient: Fs are subject to liquidity shocks
Previously, we assumed the preferences were, C0 + C1 + C2 .

We now assume Fs might experience a liquidity shock at date 1.
With probability λ (say 10%), must consume at least C 1 at date 1.
Hard constraint for simplicity– imagine very low utility if C1 < C 1 .
Assume also that C 1 = 1. Not necessary but simplifies the math.
With probability 1 − λ, linear preferences as before C0 + C1 + C2 .
The liquidity shocks are independent draws. With a large number of
Fs, the fraction of Fs that will have a liquidity shock is exactly λ.

Project choice: Liquidity-return trade-off
Suppose there are two types of projects at date 0:

Liquid (e.g., cash): Investing 1 dollar yields 1 dollar at the next date.
Illiquid (e.g., loan): Investing 1 dollar yields R > 1 (say 1.5) at date 2
if completed, but it yields l < 1 (say 0.5) if liquidated at date 0.
First suppose there are no banks and each F is on its own.

Can hold cash or (unrealistically) make loans. What would she do?

Without coordination, liquidity is wasted
On her own, each F would put C 1 into cash, and 1 − C 1 into loans.
This ensures she can consume C 1 = 1, when she needs.
In the numerical example, invest everything in the liquid asset!
But the liquidity shock happens only 10% of the time. Most of the
time, she would have excess liquidity at date 1 that she doesn’t need.
If you were a social planner, what would you ideally want to do here?

Ideal arrangement: Liquidity pooling
Invest only the minimally necessary amount in cash, λC 1 .

In the numerical example, invest the fraction, λ = 10%, in cash and
the remaining fraction, 1 − λ = 90%, in loans.
Return from loans (at date 2) is given by, R (1 − λ) = 1.5 × 0.9.
We could give this to Fs without liquidity shock, so that:
Fs who receive a liquidity shock at date 1 consume C 1 = 1.
Fs don’t receive a liquidity shock consume C2 = R = 1.5.
Recall that C2 (1 − λ) = R (1 − λ) so the resource constraints hold.
(Other arrangements for date 2 also possible since linear utility).

Banks can implement the ideal outcome
The agents in our model can also implement this outcome.
Suppose the banks are perfectly competitive (for simplicity).
They compete to provide the most attractive contract to Fs.
We can then show there is a competitive equilibrium in which:

1 Each B offers the following (ideal) contract to each F:
F can deposit 1 dollar to the bank at date 0.
Can withdraw C 1 = 1 at date 1.
Or can wait and withdraw C2 = 1.5 at date 2.
2 Each F accepts a bank contract (better than investing herself).
3 Each bank attracts a large number of Fs (so has suffi cient liquidity).
4 Fs withdraw only if they actually receive a liquidity shock.

Discussion: Bank liabilities and liquidity pooling
The contract in this equilibrium resembles various types of bank

liabilities in practice: saving deposits, term deposits, short term debt...
So the model illustrates another function of banks: liquidity pooling.
Sometimes referred to as liquidity creation/maturity transformation.
Banks help the economy to invest in less liquid but profitable projects,
while meeting the liquidity demand from consumers/firms etc.

Liquidity pools, investment specialists, or both?
In practice, banks are also investment specialists as we discussed
before. They can identify/monitor projects better than Fs.
But the two roles don’t confiict with one another.

Investment specialist is a function (largely) on the asset side of banks.
Liquidity pooling/creation is a function (largely) on the liability side.

Resulting maturity mismatch creates a source of fragility
Liquidity pooling provides an explanation for the maturity mismatch.

But the mismatch also makes the bank somewhat fragile. Why?
This is nicely illustrated in the popular movie, It’s a Wonderful Life.
See for yourself:
https://www.youtube.com/watch?v=EOzMdEwYmDU

Can we capture bank runs in our model?
Recall in the equilibrium we had:

4. Fs withdraw only if they actually receive a liquidity shock.
Could it make sense for Fs that don’t need liquidity also to withdraw?
To address this, we need to specify payoffs in these contingencies.
Let λ̂ ≥ λ denote the total withdrawals. λ̂ − λ is unforced
withdrawals.
Recall the bank promised C 1 = 1, C2 = 1.5 for early&late withdrawals.
But it might be unable to meet these promises if λ̂ exceeds λ.
Let C1 λ̂ , C2 λ̂ denote the actual payoffs conditional on λ̂....

Banks’decisions after unforced withdrawals
The bank has cash (λ) just enough to meet forced withdrawals.
The bank can also liquidate loans to obtain l = 0.5 < 1 at date 1.
Suppose a bank that faces unforced withdrawals liquidates loans to
meet them, as long as it is possible to do so.
We can calculate the payoffs, (C1 ( λ̂ ) , C2 (λ̂ )), explicitly...

Late payoffs after unforced withdrawals
When λ̂ − λ is not too high, we have C1 λ̂ = 1 but
ˆ−λ
1−λ− λ 1
l
ˆ
C2 λ = R

1 − λ̂

λ̂ − λ 1
=
1.5 1 −
− 1
1 − λ̂ l
⎛ ⎞
unforced withdrawals
⎜ :
⎜ 1 1
⎟ ⎟
=
1.5
⎜1 −
λ̂ − λ − 1
⎟ (1)
⎝
l
1 − λ̂
⎠
:
cost of withdrawals
The bank meets the unforced withdrawals but at great cost to

depositors that withdraw late. Why is the cost so large?

Late and early payoffs after unforced withdrawals
When λ̂ − λ is even higher (here, λ̂ ≥ 0.55) the bank is forced into

liquidating everything.
In this case, the late depositors receive nothing,
ˆ = 0
C2 λ
and the early depositors receive less than promised,
ˆ = λ + (1 − λ) l = 0.55
< 1.
C1 λ
λˆ λˆ
The math is not important. What matters is the qualitative effects,

which we can illustrate pictorially (combining both cases)...

Unforced withdrawal decisions are complementary
C1 λ̂ − C2 λ̂ captures the incentive of an F to withdraw unforced.
Note C1 λ̂ − C2 λ̂ is increasing in λ̂ (in a neighborhood of λ).

In economics jargon, unforced withdrawals are complementary.
The more I do an unforced withdrawal, the more I force (ineffi cient)
liquidations on the bank, and the more I incentivize you to do so.
This type of complementarity can lead to multiple equilibria...

Bad equilibrium formalizes bank runs/panics
So the earlier liquidity pooling equilibrium still works.

When no-one withdraws, it doesn’t make sense to withdraw.
But our analysis uncovered an additional possibility: When everyone
withdraws, it becomes indeed reasonable to withdraw!
The bad equilibrium can be interpreted as a bank run.
It provides one formalization of a financial panic.
It also suggests some potential policy interventions...

Roadmap

Potential solution: Suspension of convertibility
Suspension of convertibility: (Used quite a bit in history.)

Suppose, the bank temporarily shuts down after λ withdrawals.
Gives C 1 to first λ withdrawals but then stops serving withdrawals.
The bank reopens at date 2, and pays the remaining Fs C2 = 1.5.
In the model, this kills the bank run equilibrium. Solves the problem!
In practice, this “solution” is problematic for many reasons. Why?

Potential solution: Lender of last resort (LLR)
Lender of last resort (LLR): (Many examples in history.)

Suppose the central bank (e.g., the Fed) has resources at date 1,
ultimately backed by the taxation power of the government.
CB lends to the bank at rate, 1 + r ∈ [1, R], say r = 50%.
When the bank faces withdrawals, λ̂ − λ, it now chooses to borrow
from the CB as opposed to liquidating projects. This implies,
R (1 − λ) − (1 + r ) λ̂ − λ
ˆ
C2 λ =
1 − λ̂
λ̂ − λ
= R+ (R − (1 + r ))
1 − λ̂
= 1.5.
Unforced withdrawals are less costly in this case. Why?

LLR eliminates the bank run equilibrium
The bank run equilibrium disappears altogether!

The presence of the LLR circumvents ineffi cient liquidations– ensures
the Fs that withdraw late will not lose their shirts.
Once we switch to the good equilibrium, there are no unforced
withdrawals, λ̂ = λ. LLR facility is not used at all!
So the main role of the LLR is preventive. Knowing that there is a

LLR in place ensures depositors don’t panic.
(This is of course extreme. In practice, LLR could be used, e.g., if
realized λ unexpectedly turns out higher etc.)

LLR and the Bagehot’s principles
The LLR function has been well understood– and used– historically.
Bordo (1990) on the reading list summarizes Bagehot’s (1873)

principles for the Bank (of England) to observe as LLR:
1 Lend, but at a penalty rate: “Very large loans at very high rates are
the best remedy for the worst malady of the money market...”
2 Make clear in advance the Bank’s readiness to lend freely,
3 Accommodate anyone with good collateral (valued at pre-panic
prices),
4 Prevent illiquid but solvent banks from failing..

Does the LLR help to prevent panic?
Bordo (1990): Historical analyses of LLR and the runs/panics.

Bank of England: Accepted its role as LLR after the panic of 1866.
The US in the second half of the 19th century: No federal LLR.
Compare the incidence of runs in the US and the UK between
1870-1933...

Table 1 from Bordo, Michael D. "The lender of last resort: Alternative views and historical
experience." FRB Richmond Economic Review 76, no. 1 (1990): 18-29 removed due to
copyright restrictions. Please see https://www.richmondfed.org/-/media/richmondfedorg/
publications/research/economic_review/1990/pdf/er760103.pdf.

Does the LLR help to prevent panic?
The table compares historical data from the US and the UK during severe
business cycle downturns.
Macroeconomic aggregates such as real output growth from peak to
through relative to trend (first column) are qualitatively similar.
The table suggests that large recessions/downturns are associated
with crises or panics in the US, but not necessarily in the UK.
Similar evidence for France, Germany, Sweden, and Canada: They
all had LLRs and had no banking panics:
“In France, appropriate actions by the Bank of France in 1882,
1889, and 1930 prevented incipient banking crises from developing
into panics. Similar behavior occurred in Germany in 1901 and 1931
and Canada in 1907 and 1914.”

Federal Reserve’s role as the LLR
The Federal Reserve was founded in 1914 to serve as the LLR.

Its role as the LLR precedes its role as price/macro stabilizer.
Gorton and Metrick (2013) argue that the Fed might have prevented
the panic that would have otherwise happened in 1920-1921
recession...

Graph of Federal Reserve Credit extended from 1917-1935 removed due to
copyright restrictions. Please see Gorton, Gary, and Andrew Metrick (2013).
“The Federal Reserve and Panic Prevention: The Roles of Financial Regulation
and Lender of Last Resort.” Journal of Economic Perspectives 27, no. 4, p.45-64.

Federal Reserve in the run-up to the Great Depression
Read Gorton and Metrick (2013) for an interesting discussion of the
Fed’s early attempts at figuring out the LLR function.
Initially, loans were underpriced, so banks started to borrow
continuously from the Fed– even when they were not in trouble.
Is this consistent with the Bagehot’s principle?

In mid-1920’s, Fed created a “reluctance to borrow” (via rationing
etc).
This also created stigma: Borrowing from the Fed itself became a
negative signal about the bank’s quality, because– given the
reluctance policy– only weak banks would borrow!
Runs can become more likely when the perception of R falls
(appendix).
So the stigma is a real concern– has also come up in the recent crisis.

Federal Reserve during the Great Depression
• Gorton and Metrick (2013) argue that the Fed failed to

prevent the bank runs that happened in the 1930s
(during the Great Recession), in part because of the
stigma it created in the 1920s.
• Banks were cautious and reluctant to borrow from the

Fed. The LLR function did not quite work as intended.
• After this experience, the U.S. resorted to a stronger

solution to prevent runs: deposit insurance...

Potential solution: Deposit insurance
Deposit insurance: (Example, FDIC in the US implemented in 1933)

Suppose the government has resources at date 1 as before (thanks to
its taxation power), and it guarantees deposit contracts.
To be specific, let’s say that the bank “fails” if it cannot pay at least
1 to its early or late depositors (i.e., if C1 or C2 falls below 1).
If the bank fails, the government takes the bank over and redeems the
original promises to depositors.
Early withdrawal receives C 1 = 1, late receives C2 = 1.5....

Potential solution: Deposit insurance
As in LLR, deposit insurance kills the bank run equilibrium.

Once we switch to the good equilibrium, withdrawals are λ̂ = λ.
The deposit insurance is not used at all!
So like the LLR, the main role of deposit insurance is preventive.
(This is of course extreme. In practice, deposit insurance could be
used, e.g., if realized λ unexpectedly turns out higher etc.)

In the U.S., deposit insurance seemed to work as intended, FDIC was
created in 1933. Bank failures dramatically declined and remained
low---until the S&L crisis of the 1980s.
Courtesy of Gary Gorton. Used with permission.

Deposit insurance and LLR can be potentially costly
So far, we painted a rosy picture of deposit insurance and LLR.

In the model, we assumed R > 1, the bank has a valuable project
(the bank is solvent) but it can fail due to runs (or illiquidity).
In practice, hard to know if bank is insolvent, illiquid, or both.
An insolvent bank, R < 1, would also experience runs. In this case,
the deposit insurance (as well as the LLR) will make losses. Why?

Deposit insurance, LLR, and moral hazard
This also implies deposit insurance creates some moral hazard.

As we discussed in the last lecture, a bank can deliberately take risks
and end up with RL < 1 in some states. FDIC absorbs the losses.
Franchise value provides some discipline, but MH is a concern.
Thus, deposit insurance is typically complemented with insurance fees
paid by banks as well as regulation of bank risks/leverage.
These are sound policies– up to a degree– regardless of whether
mistakes or moral hazard is the ultimate cause of bank failures.

Roadmap

Shin (2009): A case study of Northern Rock
Starts as a traditional mortgage bank. Rapid growth in assets starting

in 1998 (following the decision to go public).
Not exposed to the subprime mortgages– mostly in prime
mortgages.
But nonetheless, experiences a bank run in 2007.

The first bank run in the UK since 1866 (Overend Gurney episode).
At first glance, looks like the textbook Diamond-Dybvig model.
There are, however, some major challenges when applying the theory...

Deposits are only small fraction of liabilities
Courtesy of Hyun Song Shin. Used with permission.
Growth in assets financed by nontraditional types of liabilities.

Securitized deposits not the main problem since long term.
A decline in both retail deposits and wholesale funds. But...

Decline is in mostly nontraditional deposits (not branch accounts).

Retail deposits do not seem to be the real problem
The timing of the crisis is also inconsistent with deposits playing a
central role.
Stressed start on August 9 when the short-term collateralized debt
market almost froze.
Depositor run much later (on September 14), and only after the
public announcement of support by Bank of England.
“Irony of TV images of depositors queuing at the branch offi ces was
that it was the branch deposits that were the most stable.”
“The damage was already done well before the run by its retail
depositors.”
“Although retail deposits can be withdrawn on demand, bankers have
been heard to joke that a depositor is more likely to get divorced than
to switch banks.”

Short-term debt is the real problem
Shin (2009) also argues that the real problem was in

short to medium term debt (which correspond to the
wholesale funds in the earlier table).
Northern Rock saw large outflows of wholesale funds

as maturing short to medium term loans and deposits
were not renewed.

Some lessons from the Northern Rock
We will come back to this case for additional lessons in a few lectures.
The lesson so far is that modern markets are a bit different than the
past.
The problem is nontraditional deposits or short term debt, often
held by other institutions, as opposed to traditional branch deposits.
But we could still apply the Diamond-Dybvig model, with some
relabeling:
Short term debt (of various forms) −→ Demand deposits.

Larger institutional investors −→ Small financiers/depositors

Applying DD in modern markets
So DD model could be thought of as applying in spirit, if not in

details.
Short term debt provides liquidity to the financier-institutions.
But it creates fragility for the borrowing bank (or bank-like
institution).
In fact, the problem is in a sense more severe because small deposits

are explicitly insured by the government (e.g., the FDIC), but short
term debt (as well as large deposits) are not.
This is why many public intellectuals and regulators thought the DD
model was a central feature of the crisis...

From Krugman’s NY Times column on October 10, 2011:
“One of the great things about Diamond-Dybvig is that it immediately
punctures any superficial notion that a bank can be defined by some
traditional appearance – that it basically has to be a marble building with
rows of tellers, i.e. a depository institution. Any arrangement that borrows
short and lends long, that offers investors claims that are liquid while using
their funds to make illiquid investments is a bank in an economic
sense– and is potentially subject to bank runs. Indeed, what we had in
2008 was mainly a run on shadow banks, on non-depository institutions.”
http://krugman.blogs.nytimes.com/2011/10/10/if-banks-are-
outlawed-only-outlaws-will-have-banks/
© The New York Times Company. All rights reserved. This content is excluded from our Creative

From Bernanke’s speech at the Kansas Fed Symposium at Jackson Hole,
September 2009, “Refiections on a Year of Crisis”:
“. . . the events of September and October [2008] also exhibited some
features of a classic panic, of the type described by Bagehot and many
others. A panic is a generalized run by providers of short-term funding to a
set of financial institutions, possibly resulting in the failure of one or more
of those institutions. The historically most familiar type of panic, which
involves runs on banks by retail depositors, has been made largely obsolete
by deposit insurance or guarantees and the associated government
supervision of banks. But a panic is possible in any situation in which
longer-term, illiquid assets are financed by short-term, liquid liabilities, and
in which suppliers of short-term funding either lose confidence in the
borrower or become worried that other short-term lenders may lose
confidence. Although, in a certain sense, a panic may be collectively
irrational, it may be entirely rational at the individual level, as each market
participant has a strong incentive to be among the first to the exit.”
Courtesy of the Federal Reserve Board. This content is in the public domain.

One case where DD forces could have been relevant is Bear Stearns.
SEC Chairman Cox said the following after the failure of Bear Stearns:
We will discuss this case next time. To prepare, finish reading the
HBS JP Morgan-Bear Stearns case (from page 8 onwards).
Courtesy of the U.S. Securities and Exchange Commission. This image is in the public domain.

Appendix: Are runs random events as in theory?
In the above model, runs are purely driven by a lack of coordination.

Theory doesn’t provide guidance about which equilibrium will be
played.
There could be panic without any bad news about the assets or the
economy. Panics are as likely in booms as in downturns.
This feature of the model is not consistent with data.
Gorton (1988) on the runs during the National Banking Era: “Every
time a variable predicting a recession reached a threshold level, a
panic occurred.”
Bordo (1990) makes the same point (see the earlier table).

Appendix: Equilibrium selection via global games
Economic theorists have come up with a partial fix.

By enriching the model’s information structure, we can obtain
additional predictions about the likelihood of each equilibrium.
The analysis is intuitive and suggests the coordination failures (bad

equilibrium) are more likely to obtain in bad times.
This is way outside our scope. To learn more, read Morris and Shin
(2001), “Rethinking Multiple Equilibria in Macroeconomic Modeling.”

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
63
Lecture 6: Collateralized Debt and Information Based
Panics
Alp Simsek

Revisiting runs: Is Diamond-Dybvig the whole story?
Diamond-Dybvig provides a plausible account of runs in history, and

after some relabeling, a contributing factor to the subprime crisis.
However, there is reason to think that it might not be the whole story.
Recently (and in history), much debt has been collateralized.
An example of collateralized debt is repo (sale-repurchase
agreement)...

Roadmap
1 Understanding the run on the collateralized debt
2 Information insensitivity of collateralized debt
3 Information-based panics and the leverage cycle
4 Revisiting the run(s) on Bear Stearns

What is repo?
Repo is effectively a collateralized loan.

B (the borrower/the bank), receives some money by temporarily
giving collateral such as treasuries, MBSs etc to F (the financier).
B pays back the loan with interest and reclaims the collateral.
The loan is often short term, e.g., one day, but could also be longer.
(The deal is technically structured as an initial sale of the collateral
and a right to repurchase with a prespecified price that refiects the
interest rate– hence the name.)

Repo haircuts
As we discussed earlier, the loan usually has a haircut or margin.

Recall B could borrow ρ from the Fs by using 1 dollar of the asset as
collateral. The difference, 1 − ρ, would be the REPO haircut.


Why is Repo important?
Repo is a huge market. In 2006 Q1, the average amount of outstanding

repo (and reverse) agreements were close to $5.67 trillion. (Source: Stigum
(2007), The Money Market)
Daily trading volume over this period averaged $1.6 trillion!
And this does not even capture all the repo deals out there.
Repo is typically used to finance treasuries and safer MBSs.
But it is also used to finance riskier MBSs (CDOs etc).
Gorton: There was a run on Repo, in the sense that the haircuts
on these types of riskier types of collateral increased...


Gorton: There was a run on the repo
Gorton (2010) on your reading list: “The figure is a picture of the banking
panic. We don’t know how much was withdrawn because we don’t know
the actual size of the repo market. But, to get a sense of the magnitudes,
suppose the repo market was $12 trillion and that repo haircuts rose from
zero to an average of 20 percent. Then the banking system would need to
come up with $2 trillion, an impossible task.”

A similar run on asset backed commercial paper
ABCP is different than Repo, but conceptually similar. Debt is

collateralized by assets in a conduit (set up by a sponsor bank).
Can we invoke DD to explain the run on repo and ABCP?
Traditional runs are not easy to apply to Repo
Collateralized debt is conceptually different than bank deposits.

The traditional run mechanism relies on multiple Fs having claims on
the same “collateral” (the banks’assets as a whole).
Many Fs are on the same boat, so a coordination failure is possible.
But in collateralized debt, e.g,. Repo, Fs money is backed by specific
collateral. Loosely speaking, F is on its own boat.
Thus, it is not very easy to apply a coordination mechanism.
So runs on collateralized debt might be driven by other forces.
Before we get there, let’s think about the economic rationale of
collateral...

Roadmap

From pawn shops to repo
• Repo logic is similar to a pawn shop.
• In a pawn shop, the financier (the lender of money) receives

a valuable commodity such as jewelry as collateral. The
financier returns the collateral back upon receipt of money
with interest.
• Pawn shop is an ancient institution. See Goetzmann and

Rouwenhorst (2005), The Origin of Wealth, for some
fascinating pictures of pawnshop loan records from China
circa 662-689 A.D. The records indicate that silk garments
were used as collateral at the time.

Why collateral?
The fact that collateral has been used for a long time, and under very
different institutions, suggests that it serves an important economic
function.
One rationale: Collateral helps to mitigate information frictions.
Once the loan is secured by collateral, F only needs to have a
rough sense that the value of collateral exceeds the amount lent.
Less need to know B’s financial health (less “adverse selection”). Or
what B will do with the borrowed money (less “moral hazard”).

Collateral helps to economize on information costs
In practice, producing information about borrowers or their actions

can be very costly (need credit registries, monitors, courts etc.)
So one view of collateral is that it allows the fiow of credit while
economizing on costly information acquisition.
This can also help to understand the form of the collateral contract.
It typically takes the form of debt as opposed to something else.
To formalize, suppose B promises to pay back s (x), where x denotes
the value of the collateral (on the due date)

The collateral contract could take this form

Or it could take this form

But instead, it typically takes this form
B pays back (fixed) D whenever x ≥ D but defaults when x < D.

Why not have the first two contracts?
There are actually some nice things about the first two contracts.
They have reasonable risk sharing properties: Fs share some risk in
the sense that Bs’liability is relatively low in low states.
If we had these types of contracts, crises could be less severe. Why?
But we do not have these contracts. Why might this be the case?

Debt contract is the least sensitive to information
Most likely because they are quite sensitive to changes in the value, x.
To price them, F would have to know quite a bit info about x.
This goes against the logic of using collateral to begin with!
In contrast, debt is largely insensitive to the value of collateral.
As long as x > D (fiat part), payoff doesn’t depend on x at all!
This further helps to economize on information/reduces frictions.

Collateralized debt economizes on information costs
In fact, in practice contract does not even make reference to the value
of the collateral, x. That reference is a construct of our formalism.
As long as x > D, B pays the debt and reclaims the collateral.
As long as there is mutual understanding that E [x] >> D, parties
can borrow and lend without discovering the exact value of x.
This is probably why even things like old jewelries or silk shirts
(presumably very diffi cult to price exactly) can serve as collateral.

How could they trade so much without knowing much?
Holmstrom (2015): MBSs could be analogues to old jewelries!

Remember that trillions of $s were being exchanged in repo markets.
MBSs were quite opaque and their pricing was not well understood even
by the experts on Wall Street.
In retrospect, this seemed puzzling to many observers. How could Wall

Street trade so much without knowing much? Was nontransparency of
these markets a cover for shady deals in the background?
Note, however, the valuation of jewelries in a pawn shop is not very

transparent either...

Because this is what collateralized debt markets are about
As Holmstrom notes, the Repo market is designed so that---thanks

to the information insensitivity of debt---people don't need to know
about the value of the underlying collateral.
Since these markets economize on information, opaqueness could

be one of their integral features.
Thus, nontransparency does not necessarily reflect an anomaly.

Moreover forcing these markets to become more transparent might
generate unintended adverse consequences (see Holmstrom, 2015,
for details).

Comparative statics of debt capacity/margins
Think of the information insensitive region as E [x] being at least a

few standard deviations above D so default is unlikely.
When this region is greater, all else equal (for a fixed E [x]) we can
have higher D while still ensuring overcollateralization.
This might be why safer collateral is associated with lower
margins/haircuts.
It also provides an explanation for the short horizon of

Repo/collateralized debt: Collateral is less risky over shorter horizons!


Roadmap

Crisis: Collateralized debt becomes information sensitive
• What happens if there is a severe shock that reduces the expected

value of collateral, E[x], into the information sensitive region?
• This is the Holmstrom and Gorton view of the crisis. Nontransparency

of the collateral makes sense in normal times and facilitates liquidity
(trade). However, once the collateral becomes information sensitive,
nontransparency exacerbates the crisis.
• The liquidity in short-term debt markets rely on overcollateralization

as opposed to information and price discovery about the underlying
collateral (which is very costly).
• In a crisis, investors are forced to discover information but in a

market that was not designed for this purpose (read Holmstrom
(2015) for a nice contrast with equity markets---which do facilitate
information discovery).

Crisis: Collateralized debt becomes information sensitive
Start of the crisis: Markets recognize risks in MBS. Valuations fall,

but more importantly they become more uncertain (both figures).
We are suddenly worried about the value of old jewelries, but in an
environment that was “designed” to obviate price discovery.
What would happen to debt capacities?



Crisis: “Run” on collateralized debt due to uncertainty
Consistent with theory, Fs started to run riskier forms of collateral (as

opposed to specific institutions, as in DD) starting mid-2007.
In fact, we had seen another person who made the same point using
slightly different language. Do you remember who he/she was?

© Dow Jones & Company, Inc. All rights reserved. This content is excluded from our Creative

Leverage cycle: Run on collateralized debt due to bad news
The bottom panels illustrate a measure of ρ (pledgeability in the

earlier lecture, or debt capacity in this lecture) for different assets.
Leverage ratios seem procyclical in the sense that they are high in
good times (high prices) but low in bad times (low prices).
Geanakoplos (2009), “The Leverage Cycle” proposed a theory of
procyclical leverage based on changes in uncertainty.
Bad times =⇒ Uncertainty =⇒ Nervous lenders=⇒ Less leverage
=⇒ Amplification....
Geanakoplos is essentially saying something very similar to
Gorton-Holmstrom!
(But he doesn’t explain why there are debt contracts to begin with).

Applying the theories of runs
So we now have two theories of runs:

1 DD: Run on specific institutions with unsecured short-term debt.
To lesser extent, run also on the secured debt, e.g., Repo, if there are
worries about collateral being seized in bankruptcy.
2 Holmstrom-Gorton-Geanakopos (HGG): Run on uncertain collateral.
The first run is quite sudden, the second run unfolds more gradually.
Let’s try to apply them in the context of the failure of Bear Stearns...

Roadmap

Revisiting the run(s) on Bear Stearns
By mid-2007, broker dealers such as Bear heavily on Repo financing.

Bear had the weakest liquidity buffers among all such dealers...

From the HBS case (Exhibit 2, Bear’s balance sheet).
Data on Q1-2008, 2007, 2006, 2005, 2004 ($million).
Exhibit 2 from "The Tip of the Iceberg: JP Morgan Chase and

Bear Stearns" removed due to copyright restrictions. Please visit
http://www.hbs.edu/faculty/Pages/item.aspx?num=36849.

Exhibit 7 from "The Tip of the Iceberg: JP Morgan Chase and Bear Stearns" removed due to
copyright restrictions. Please visit http://www.hbs.edu/faculty/Pages/item.aspx?num=36849.

By the end of 2007—early 2008, the run on Repo (for riskier assets)
had already started and gradually tightened liquidity at these
institutions.
Bear was the most exposed to this run, as the above table illustrates.
Combined with realized losses, this pushed Bear to the edge...



Sophisticated lenders had already run
Some money market funds like the Reserve Primary Fund continued
to lend to Bear against riskier forms of collateral.
(This fund later “broke the buck” and induced another layer of panic)
But the more sophisticated lenders like Goldman had already stopped
lending to Bear against riskier forms of collateral.
Probably because they knew Bear was facing serious risks.
This suggests that (somewhat different than in our model) the
riskiness of the borrower also plays a role in practice.
Imagine collateral + borrower reputation providing joint protection.

The case study on Bear Stears and JP Morgan Chase on your reading
list describes the events that lead to the Bear's failure in March 2008.
The final days (March 10-14) have all the fingerprints of a DD-type
bank run. Financiers (such as hedge funds that receive brokerage
services from Bear) start to question Bear's financial health and
withdraw their money, which further exacerbates Bear's financial
difficulties.
Interestingly, some Repo financiers also pulled out partly because

there were concerns that Repo collateral can be frozen in bankruptcy
courts.
So concerns about bankruptcy courts' dealing with collateral can make

Repo loans (temporarily) similar to an unsecured loan.

Bear Stearns' stock traded around $140 per share in July 2007.
The price fell to around $80 per share by February 2008.
The price was around $60 per share on March 11, 2008. It fell to
almost zero by March 14. This sudden drop is further indicative of a
DD-type run.
On March 16, Bear Stearns was purchased by JP Morgan Chase, but

only after the Fed agreed to backstop losses from the assets on
Bear's securities portfolio. So the government had to step in to
mitigate the panic (as it is usually the case with runs).
The initial offer by JP Morgan Chase was around $2 per share. The
offer was later raised to $10 per share (to avoid litigation by Bear
shareholders).

Lessons from the run(s) on Bear Stearns
My view: HGG mechanism was important to push Bear Stearns to the

cliff, creating the conditions that are ripe for a DD type panic.
(Recall that panics in practice typically happen in bad times.)
The DD type panic in turn did the final blow and pushed over the cliff.
So it seems that the two run mechanisms combined to create damage.
+

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
50
Lecture 7: Interconnections and Panics
Alp Simsek

Is DD and HGG the whole story of runs?
DD and HGG provide (complementary) perspectives on runs.

However, there might be yet another dimension to runs/crises.
This is elaborated by the Northern Rock case study by Shin (2009).
Understanding what happened to NR will also illustrate another key
feature of the system: interconnections, with other implications.

Roadmap
1 Credit crunch and runs
2 Counterparty risk
3 Complexity

Did Northern Rock experience a DD or HGG run?
“To turn the question around, the issue is why sophisticated lenders who
operate in the capital markets chose suddenly to deny lending to a bank
that had an apparently solid asset book and virtually no subprime lending.
Northern Rock was in the business of prime mortgage lending to U.K.
households. The asset quality of any mortgage bank is vulnerable to a
sharp decline in house prices and rising unemployment. However, 2007 was
the Indian summer of the housing boom in the U.K., and there were no
outward signs of seriously deteriorating loan quality. Thus, the sudden
refusal of lenders to fund Northern Rock needs an explanation. The
answers to the puzzle reveal much about the nature of banking in the age
of securitization and capital markets.”

Did Northern Rock experience a DD or HGG run?
Northern Rock did not hold subprime mortgages. So it seems unlikely

that they were subject to a (direct version of) HGG run.
We could try to apply DD after some relabeling– as we did with Bear.
But remember, generally panics happen after bad news about banks.
Losses erode buffers (capital&liquidity), and make a panic more likely.
So it is puzzling that Fs would panic and run out of the blue on
Northern Rock– didn’t hold subprime or make large losses.

Shin: Fs are also institutions/other “banks”
Shin notes that, unlike the basic DD model, many of the financiers
of Northern Rock are other banks (or sophisticated institutions).
So the financial network is more complicated than what we have

emphasized so far, with lending among banks.
In a financial crisis, the lending institutions might be forced to cut

their positions due to their own financial problems (through their
own net worth channel, runs etc).
This might look like a run from the perspective of the borrowing
institutions...

Shin describes an example as in this picture. Imagine B1 as the Northern
Rock and B2 as another bank that lends to Northern Rock...

What happened to Northern Rock?
- Shin writes: "Bank 2 has other assets (that is, loans it has made to other parties), as
well as its loans to Bank 1. Suppose that Bank 2 suffers credit losses on these other
loans, but that the creditworthiness of Bank 1 remains unchanged. The loss suffered by
Bank 2 depletes its equity capital. In the face of such a shock, a prudent course of action
by Bank 2 is to reduce its overall exposure, so that its asset book is trimmed to a size
that can be carried comfortably with the smaller equity capital."
"From the point of view of Bank 2, the imperative is to reduce its overall lending,
including its lending to Bank 1...However, from Bank 1’s perspective, the reduction of
lending by Bank 2 is a withdrawal of funding. If financial markets are deep and liquid,
Bank 1 will find alternative sources of funding at roughly the same price—after all,
nothing in Bank 1’s risk characteristics has changed, so it should be able to borrow just
as easily as it did before. But now imagine a situation where a combination of events
arises: i) the reduction in Bank 2’s lending is severe; ii) overall credit markets have seized
up in such a way that no one has access to funding, including Bank 1; and iii) Bank 1’s
assets are so illiquid that they can only be sold at fire-sale prices. Under these
circumstances, the prudent shedding of exposures from the point of view of Bank 2 will
feel like a run from the point of view of Bank 1. Arguably, this type of run is one element
of what happened to Northern Rock."

F F
B2 B1
Net worth channel Credit crunch, triggers
HGG-type run. an indirect run. Reduction
in (Secured or unsecured)
short-term debt.

Northern Rock: A run driven by a credit crunch
Recall that in mid-2007 the Repo and ABCP borrowing collapsed.

Banks also started to realize or expect losses from subprime.
These banks faced tighter borrowing constraints. They were forced to
(and perhaps also chose to) reduce their leverage and risks.
They would cut their positions across the board– not just in subprime.
In particular, they would stop reduce or renewing loans to other firms.
If these Bs only lent to the real sector, these actions would look like a
credit crunch (remember Lectures 2-3 and Chodorow-Reich).
When Bs also lend to one another, the credit crunch will also look like
a run on the Bs that lose financing (more severe externalities).

Lessons from the Northern Rock
Lesson: Initial problems can naturally spread to other Bs/collaterals.

Roadmap
2 Counterparty risk
3 Complexity

How about contagion?
The interconnections also suggests the possibility of contagion: A

bank failure can trigger problems elsewhere in the system.
Contagion is a big concern in policy discussions of bank bailouts.
One channel of contagion is the credit crunch as we discussed above.
Another channel is informational: The failure of a bank can send a
negative signal. Trigger or exacerbate HGG or DD type-runs.
But there is also the possibility of direct damage via counterparty
risk...

Ex-Fed Chairman Bernanke, in his testimony to the Senate on April 3,
2008 following the Fed’s Bear Stearns intervention, captures these
concerns as follows:
“Our financial system is extremely complex and interconnected, and Bear
Stearns participated extensively in a range of critical markets. The sudden
failure of Bear Stearns likely would have led to a chaotic unwinding of
positions in those markets and could have severely shaken confidence. The
company’s failure could also have cast doubt on the financial positions of
some of Bear Stearns’thousands of counterparties and perhaps of
companies with similar businesses.... Moreover, the adverse impact of a
default would not have been confined to the financial system but would
have been felt broadly in the real economy through its effects on asset
values and credit availability.”
Mention info contagion and credit crunch, but also counterparty risk.
Courtesy of The Federal Reserve Board. This material is in the public domain.

Counterparty risk
Bs are often exposed to one another through loans that are not fully
secured.
These exposures can emerge from explicit loans, as in the earlier
diagrams.
If B1 fails, B2 wouldn’t receive its loan back in full and suffer some
losses.
But these types of exposures could also emerge from “implicit
loans”– unsecured gains that accumulate in bilateral transactions.
In practice, implicit exposures are common in derivatives that are
traded over-the-counter...

Counterparty risk in OTC derivatives
Many types of derivatives are traded OTC: Forwards, swaps, interest
rate swaps, credit default swaps.
AIG case nicely illustrates the counterparty risk in these markets...

Counterparty risk in OTC derivatives
McDonald and Paulson (2015), “By construction, many derivatives

contracts have zero market value at inception; this is generally true for
futures, swaps, and credit default swaps. When a position has zero market
value, the two parties to a contract can, by mutual consent, exit the
contract without any obligation for either to make any further payment to
the other....As time passes and prices move, a contract initiated with zero
market value will generally not remain at zero market value: fair value will
be positive for one counterparty and negative by an exactly offsetting
amount for the other.”
The positive-value party is exposed to the negative-value party. If the
latter becomes bankrupt, the former loses a valuable asset!
So the positive-value party has implicitly made an unsecured loan.
As the exposures grow, the exposed can make a “margin call”...

Margin calls in OTC derivatives
“In such cases, it is common for the negative value party to make a
compensating payment to the positive value counterparty. Such a payment
is referred to as margin or collateral, in this context, the two terms mean
the same thing. Collateral can fiow back and forth as market values
change.”
Read page 93 for the details of how and when margin calls happen.
They happen when exposures exceed a pre-agreed positive threshold.
AIG faced margin calls before its failure and bailout in September
2008...

What is AIG?
McDonald and Paulson (2015): “The near-failure on September 16, 2018,

of American International Group (AIG) was an iconic moment of the
financial crisis. AIG, a global insurance and financial company with $1
trillion in assets, lost $99.3 billion during 2008 and was rescued with the
help of the Federal Reserve, the Federal Reserve Bank of New York, and
the US Treasury. The rescue played out over many months and involved
the extension of loans, the creation of special purpose vehicles, and equity
investments by the Treasury, with the government assistance available to
AIG ultimately totaling $182.3 billion. The decision to rescue AIG was
controversial at the time and remains so. AIG’s fate also provided an
important touchstone in discussions of financial reform. AIG motivated the
enactment of new rules governing nonbank financial institutions, as well as
rules about the treatment of financial derivatives.”


Margin calls can exacerbate the failure of the negative-value party.
If this happens, and there is no bailout, the positive-value party might
also suffer some losses (see above for margin shortfalls).
So unsecured exposures in OTC derivatives create counterparty risk.
These markets are large, so this type of risk is a real concern...

© Securities Finance Monitor. All rights reserved. This content is excluded from our Creative

© Securities Finance Monitor. All rights reserved. This content is excluded from our Creative
McDonald and Paulson calculate the (potential) losses that could

accrue to AIG’s counterparties absent a bailout...


Counterparty risk is non-trivial (although not too large)
“The rescue of AIG had many beneficiaries. The broader financial

system was spared the unpredictable consequences of a large and
complicated firm failing at a time when financial markets were very
fragile. Direct beneficiaries of the rescue included the life insurance
subsidiaries that received $20 billion in capital infusions, protecting
their policyholders. The counterparties to the credit fault swaps AIG
had sold on multisector credit default obligations (CDOs) were also
beneficiaries, although their direct benefit was the $17.7 billion in
collateral payments made after the rescue rather than much larger
figures that sometimes have been emphasized.”
So it seems that, absent a bailout, the bankruptcy of AIG would
trigger non-trvial (although not huge) losses on counterparties.

Academic research on counterparty risk
There is an interesting academic literature on counterparty risk.

Formalizes how counterparty risk can generate domino effects.
Also tries to empirically estimate the (potential) length of dominos.
Acemoglu, Ozdaglar, Tahbaz-Salehi (2013) on the reading list analyze
the resilience of different with different network structures and
different-size shocks.
Other papers try to identify the banks that are more systemic/central.
“Too Interconnected to Fail” as opposed to “Too Big to Fail”.
In Caballero-Simsek (2013), we use counterparty risk to emphasize

what we think is an even more pressing problem: complexity...

Roadmap
2 Counterparty risk
3 Complexity

Another problem during crises: Complexity
During crises, economic environment appears “complex”.

Holmstrom-Gorton: Need to figure out value of opaque collateral.
In addition, need to figure out whether– and how much– all the
amplification mechanisms we discussed in this course might shake the
system.
E.g., fire sales possible. Prices affected by “non-fundamental” factors.
Moreover, not much precedent (relatively rare). Unknown
unknowns.
Pricing models that Bs use in normal times might not be of much use.

Bank of England Financial Stability Report for June 2009:
Courtesy of the Governor and Company of the Bank of England. Used with permission.
Figure:
Counterparty risk is a source of complexity
Counterparty risk provides a particular and specific source of
complexity.
Haldane’s (2009) speech: “knowing your ultimate counter-party’s risk

becomes like solving a high-dimension Sudoku puzzle.”
Recall also Bernanke’s testimony from earlier: “Our financial system
is extremely complex and interconnected...The sudden failure of
Bear Stearns likely would have led to a chaotic unwinding of positions
in those markets and could have severely shaken confidence. The
company’s failure could also have cast doubt on the financial
positions of some of Bear Stearns’thousands of counterparties and
perhaps of companies with similar businesses....”
Courtesy of The Federal Reserve Board. This material is in the public domain.

A model of crises driven by counterparty risk and
complexity
In Caballero-Simsek (2013), we use a stylized model to illustrate how

complexity during a crisis can create considerable amplification.
Relatively small losses trigger informational regime change, “create
complexity,” and induce a disproportionately large credit crunch.
Government support is desirable, precisely because it reduces
complexity.
I will summarize the ingredients and results without getting into
details.

The basic ingredients of the model
We model the financial system as a network of cross-exposures.

Each bank starts with some legacy assets (e.g., mortgages) and some
cash.
Each bank makes essentially a single decision:
Sell (cautious action): Liquidate legacy assets and keep cash.
Buy (normal action): Keep assets, and make new investments.
The more banks sell (or cautious), the bigger is the credit crunch.
Key question: how banks’sell/buy decisions depend on complexity.



The key ingredients of the model
Banks learn that one bank, say, b 0 (e.g., Bear, Lehman, or another
bank) is hit by losses and will soon be bankrupt.
They realize losses will spill over to other banks via counterparty risk.
They can’t figure out if they will be caught up in a domino

cascade.
We formalize this ingredient as follows: Banks have only partial
knowledge of the network. They know their own counterparty
(understand their own exposures), but they do not know who the
counterparties of counterparties are, or who their counterparties are.
But let us first look at a benchmark scenario without this ingredient,
i.e., suppose banks have no uncertainty about the network...

No-uncertainty benchmark
Figure:
No-uncertainty benchmark: Relatively mild outcome
When banks know the network, there is a partial domino effect.

Losses bring down a few banks, but they gradually dissipate (since
each failing bank absorbs some), and the cascade eventually stops.
This would generate some credit crunch (especially with leverage etc).
But in a deep financial market such as the US, the damage would be
contained– other banks would step in (they all “buy” in our model.)
Enter our key ingredient: Uncertainty about the network....

.
Imagine these slots as placeholders: Each bank is assigned to a slot.

Complexity: Banks know that the network is a circle (for simplicity),
but they don’t know which bank is assigned to which slot.
They know their immediate counterparty, but not others’slots.
So they don’t know their distance from the troubled bank, b 0 .

Suppose also that Bs are risk averse (want to avoid bankruptcy if
possible).
For simplicity, suppose Bs act according to the worst case scenario.
This is not a bad assumption when there are unknown unknowns.
Economic agents tend to be more cautious when they face ambiguity
as opposed to quantifiable risk (supported by experiments).
How would Bs react in this case? For a small shock? Larger shock?

Equilibrium with small shocks: Same as no-uncertainty

Equilibrium with small shocks: Same as no-uncertainty
For small shocks, there is no cascade: The immediate neighbor

absorbs some losses, but the remaining banks are safe.
More importantly, those other banks know that they are safe:
They know they are not directly exposed to b 0 (know their
counterparty).
They can also rule out an indirect hit since the shock is small.
Since other banks face no bankruptcy risk, they buy. So the
damage/credit crunch is relatively small, as in the benchmark.
Note that there is still complexity (banks still don’t know the
network). But the complexity is dormant– not payoff relevant.

Equilibrium with larger shocks: Large credit crunch

Equilibrium with larger shocks: Large credit crunch
When the shock is larger, and a cascade becomes possible, local
information is no longer suffi cient to rule out an indirect hit.
Worst case scenario is one in which my counterparty is exposed to b 0 .

Since I cannot rule this out, and I am very risk averse, I act as if this
is the case.
Banks act as if they are closer to the distressed bank than they
actually are.
When the dust settles, relatively few banks will be bankrupt.
But during the crisis, they all take precaution: Large credit crunch!
So complexity creates amplification relative to the benchmark.

Broader idea: Crises increase complexity and uncertainty
The broader idea here is that the financial system is quite complex.
The system can work fine in normal times (or with small shocks),
with each “bank” understanding its local corner and markets.
But unusual events such as crises greatly increase the information
burden on the banks/make the complexity payoff relevant.
This induces banks/players to be much more cautious than normal.
The extreme caution exacerbates distress/panics (recall Shin).
Bailouts or support by the government can mitigate these effects,
precisely by reducing the (payoff-relevant) complexity.
The above analysis did not feature government, so it illustrates how
bad things could get if the government doesn’t step in...

Recall what Bernanke said about the potential failure of Bear Stearns.
Caballero-Simsek: “Unfortunately, Chairman Bernanke’s testimony
would prove prescient only a few months later during the Lehman
episode, when the demise of the investment bank wrecked havoc all
around the world.”
Our interpretation: Lehman bankruptcy changed the market
perceptions about bank failures and government support.
The market realized that the government might be unable or
unwilling to prevent the failure of a large bank. What happened?

When Lehman failed: Mother of all panics

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
47
Lecture 8: Lessons and Policy Responses
Alp Simsek

Lessons and policy respones
We have covered quite a bit of material on financial/banking crises.

Let’s do a recap and think about the ideal policy response.

First set of issues: Solvency, capital, leverage

Net worth channel, credit crunch, fire sales

Second set of issues: Liquidity, runs, panics

DD-type and HGG-type runs

Panics driven by interconnections
F F
B2 B1
Net worth channel Credit crunch, triggers
HGG-type run. an indirect run. Reduction
in (Secured or unsecured)
short-term debt.

We care because....

Evidence/theory: Government support mitigates panics

Evidence: Ultimate cause can be mistakes

Theory: Some discipline against moral hazard
@ Harvard Business School. All rights reserved. This content is excluded from our Creative

So what is the optimal crisis policy?
Based on our analysis, how should we think about crisis policy?
The question has two dimensions that might also interact:

1 What is the optimal ex-post policy to fight/mitigate crises?
2 What is the optimal ex-ante policy to prevent/rarefy crises?

Roadmap
1 Optimal policy during crises
2 Optimal policy to “prevent” crises

How to react to a developing crisis?
Caballero (2010) draws an analogy between a heart failure ("sudden

cardiac arrest") and a financial crisis ("sudden financial arrest").
The main treatment for a sudden cardiac arrest requires a defibrillator.

Moreover, time is of the essence: the defibrillator must be applied
within minutes of the arrest. Thus, it is important to have numerous and
easily accessible defibrillators.
Likewise, the main solution to a financial crisis is rapid and massive

government support to the financial system, either in the form of
bailouts (to mitigate the solvency problems) or guarantees (to mitigate
the liquidity problems). These are the financial defibrillators.

Moral hazard concerns are secondary during a crisis
However, many people resist the use of financial defibrillators, as they seem
to think this will generate moral hazard and exacerbate future crises.
As Caballero notes, this is similar to objecting the use of defibrillators out of

concern that, once people realize they are more likely to survive a heart
attack, they will eat more cheeseburgers and will become more likely to have
heart attacks.
The point is that the moral hazard concerns are secondary during a crisis.
Even if moral hazard is a major cause of the crisis (which itself is debatable--
see Lecture 4), it is best dealt with once the crisis is over and by applying
other policies. To continue the analogy, one could imagine taxing
cheeseburgers in normal times, as opposed to banning defibrillators during
crises.
Once the crisis hits, the pragmatic policymakers often realize that they must
apply the financial defibrillators. How did the policymakers in the US react to
the crisis?

Bernanke (January, 2009), “The Crisis and the Policy Response”
"Other than policies tied to current and expected future values of the
overnight interest rate, the Federal Reserve has-and indeed, has been
actively using-a range of policy tools to provide direct support to
credit markets and thus to the broader economy. As I will elaborate, I
find it useful to divide these tools into three groups. Although these
sets of tools differ in important respects,they have one aspect in
common: They all make use of the asset side of the Federal Reserve's
balance sheet. That is, each involves the Fed's authorities to extend
credit or purchase securities."

Traditional LLR to mitigate DD runs
“The first set of tools, which are closely tied to the central bank’s
traditional role as the lender of last resort, involve the provision of
short-term liquidity to sound financial institutions. Over the course of
the crisis, the Fed has taken a number of extraordinary actions to
ensure that financial institutions have adequate access to short-term
credit. These actions include creating new facilities for auctioning
credit and making primary securities dealers, as well as banks, eligible
to borrow at the Fed’s discount window... ”
“Liquidity provision by the central bank reduces systemic risk by
assuring market participants that, should short-term investors begin
to lose confidence, financial institutions will be able to meet the
resulting demands for cash without resorting to potentially
destabilizing fire sales of assets....”

Collateralized LLR to mitigate HGG runs
“On the other hand,....providing liquidity to financial institutions does
not address directly instability or declining credit availability in critical
nonbank markets, such as the commercial paper market or the market
for asset-backed securities, both of which normally play major roles in
the extension of credit in the United States.”
“To address these issues, the Federal Reserve has developed a second
set of policy tools, which involve the provision of liquidity directly to
borrowers and investors in key credit markets....In addition, the
Federal Reserve and the Treasury have jointly announced a facility
that will lend against AAA-rated asset-backed securities collateralized
by student loans, auto loans, credit card loans, and loans guaranteed
by the Small Business Administration. The Federal Reserve’s credit
risk exposure in the latter facility will be minimal, because the
collateral will be subject to a "haircut" and the Treasury is providing
$20 billion of capital as supplementary loss protection.”

QE1 to mitigate asset fire sales
“The Federal Reserve’s third set of policy tools for supporting the
functioning of credit markets involves the purchase of longer-term
securities for the Fed’s portfolio. For example, we recently announced
plans to purchase up to $100 billion in government-sponsored
enterprise (GSE) debt and up to $500 billion in GSE mortgage-backed
securities over the next few quarters. Notably, mortgage rates
dropped significantly on the announcement of this program and have
fallen further since it went into operation. Lower mortgage rates
should support the housing sector. The Committee is also evaluating
the possibility of purchasing longer-term Treasury securities.”

How about the treasury and the FDIC?
The Fed could mitigate liquidity issues, but it could not (by itself)
deal with all of the net worth/capital/solvency issues.
Treasury and the FDIC needed to step in.
Treasury needed authorization by the Congress.
FDIC had own resources and emergency authority to tap more funds.
Swagel (2015): “Constraints on the Crisis Policy Response”:

“The Troubled Asset Relief Program (TARP) was proposed on
September 18, 2008– the same week as the Lehman collapse and the
AIG bailout– and passed into law as part of the Emergency Economic
Stabilization Act on October 3, 2008. The TARP provided authority
for the Treasury to purchase or guarantee up to $700 billion of
troubled assets...”

TARP: From asset purchases to recapitalization
“While the intent of the TARP when it was proposed was to purchase
illiquid assets...It became clear to policymakers that a more rapid
approach was needed...The switch from asset purchases to capital
injections fit within the TARP’s legislative language, because shares
of banks that originated loans represented troubled assets related to
mortgages.”
Asset purchases is at best an indirect fix to net worth issues:
“Asset purchases would help cleanse bank balance sheets of illiquid
mortgages and contribute to price discovery but would raise firms’net
worth only if Treasury intentionally overpaid for assets (which was not
the plan) or if asset prices rose following the TARP purchases (a
possibility if the implementation of the reverse auctions lifted
confidence and thereby improved asset prices).”

TARP: Recapitalization to mitigate net worth channel
“The eight institutions ultimately receiving capital injections (after

Bank of America’s acquisition of Merrill Lynch) together accounted
for more than half of both the assets and deposits of the US banking
system. The existence of these mega-firms, while giving rise to
concerns over institutions that were too big to fail, also made it
possible to strengthen a broad swathe of the banking system rapidly.
Each firm received public capital equal to 3 percent of its
risk-weighted assets, for a total of about $125 billion. The remaining
thousands of US banks together would be eligible for another $125
billion in capital.”

TARP: Additional capital to particularly weak banks
“...Treasury decisions to provide additional assistance to Citigroup
and Bank of America in 2008 and 2009 beyond the initial capital
investment of $25 billion for each institution. These two banks (and
perhaps others) appeared to be insolvent at points during the crisis,
and were to require extraordinary assistance from the TARP, and yet
the government propped them up rather than invoking the usual bank
resolution authority of the FDIC. These decisions refiected several
factors. First, there was the concern that a government takeover of
Citigroup would lead to a renewed fiight from other still-fragile banks.
Second,...there was little confidence across the government in the
agency’s ability to run a mega-bank.”
“At the end...the firm did not fail. Meanwhile, bondholders and other
counterparties avoided losses entirely, which was in some ways less
than fully desirable, but did have the positive effect of limiting further
financial contagion.”

FDIC: Public guarantees to facilitate private borrowing
Swagel (2015): “...the Federal Deposit Insurance Commission
introduced the Temporary Liquidity Guarantee Program (TLGP),
under which it would insure senior debt issued by banks. The FDIC
further extended its deposit insurance to provide an unlimited
backstop on business transactional checking accounts that were
previously uninsured. The TLGP program was undertaken using the
FDIC’s emergency authority, which allowed the FDIC to put taxpayer
money...without the usual requirement to act in a manner that
ensured the least cost for taxpayers....”
“Veronesi and Zingales (2010) calculate that the guarantees from the
FDIC account for most of the benefits in terms of stabilization of the
financial system.
“...The Dodd—Frank legislation was later to prohibit a repeat of the
TLGP without explicit Congressional approval.”
(Can you see a problem/tension in the last two bullet points?)

Veronesi and Zingales (2010), “Paulson’s Gift”
“We calculate the costs and benefits of the largest ever U.S.
Government intervention in the financial sector announced the 2008
Columbus-day weekend. We estimate that this intervention increased
the value of banks’financial claims by $130 billion at a taxpayers’
cost of $21 -$44 billions with a net benefit between $86bn and
$109bn. By looking at the limited cross section we infer that this net
benefit arises from a reduction in the probability of bankruptcy, which
we estimate would destroy 22% of the enterprise value. The big
winners of the plan were the bondholders of the three former
investment banks and Citigroup, while the losers were JP Morgan
shareholders and the U.S. taxpayers.”
(The would-be destruction of the enterprise value is consistent with

DD type runs/ineffi cient liquidations on less liquid assets.)
Courtesy of Pietro Veronesi and Luigi Zingales. Used with permission.

Stress tests: Incentivize and back up private capital
Swagel (2015): “In 2009, TARP funds were again set to be used to
shore up the financial system, serving as the source of public capital
backstopping the so-called “stress tests,” in which bank balance
sheets were evaluated to see whether they could withstand an
additional period of financial stress. Banks that lacked the
appropriate capital as determined by the stress test would be given a
chance to raise additional capital from the private sector after which
they would be required by their regulator to accept it from the TARP
(on onerous terms meant to induce private capital-raising)....The
availability of TARP capital was essential to making the stress tests
credible in that public capital was available to be forced on firms that
could not (or would not) raise their own in response to the results of
the stress test.”

Stress tests: Incentivize and back up private capital
Hanson, Kashyap, and Stein (2011): “The penalty box in this case
was that any bank failing to raise the capital from the private markets
would be required to accept an equity injection from the Treasury,
which would have involved strict limits on executive compensation.
Remarkably, in the few weeks following the release of the SCAP
(stress test) results, the banks involved were able to raise nearly $60
billion in new common equity; by the end of 2009 this figure had risen
to over $125 billion.”
“Here is a case where a strong regulatory hand appears to have had
highly beneficial effects. Indeed, by being tough and giving banks no
choice, regulators probably made it easier for banks to do the capital
raising.”

How did the crisis end?
Stress tests are considered to be the turning point in the US crisis.

Europe also did stress tests in 2009 but much less successful.
Didn’t have credible public capital as backstop (political frictions).
Consistent with important role of government support during crises.
Caballero (2010): Make the government support more systematic....

Did the US follow Caballero’s advice to prepare for the next crisis?
Quote from page 9 from Caballero, Ricardo J. "Sudden financial

arrest." (PDF) IMF Economic Review 58, no. 1 (2010): 6-36.

Dodd-Frank Act: Glass is half empty
Swagel (2015): “What constraints will policymakers and regulators

face when the next financial crisis arrives? It seems safe to conclude,
based on political considerations, that there will not soon be another
TARP...Attacks on the bank bailouts in particular have become a
staple of political campaigns. Moreover, some emergency actions
taken during the crisis are no longer available to policymakers as a
result of provisions in the 2010 Dodd—Frank financial reform bill. The
Treasury is no longer permitted to use the Exchange Stabilization
Fund to guarantee money markets. The Federal Deposit Insurance
Corporation must now obtain Congressional approval to provide broad
debt guarantees. The Federal Reserve can no longer make emergency
loans to individual nonbank institutions but must instead devise
broad-based programs.”

Dodd-Frank Act: But it is also half full
“At the same time, the Dodd—Frank law provided important new
powers for government regulators to respond to a future financial
crisis. Title II of the Dodd—Frank law creates a nonbank resolution
authority under which the government can put taxpayer funds into a
failing institution to prevent a collapse. Government offi cials are
required to recoup taxpayer funds by imposing losses on shareholders,
bondholders, or other counterparties of the failing firm, and ultimately
through assessments on other financial sector participants if needed.
The FDIC is still developing the tools for such an intervention.”

Roadmap
1 Optimal policy during crises
2 Optimal policy to “prevent” crises

How about ex-ante policies to prevent crises?
The ideal ex-ante policies are much less well understood.

Moral hazard suggests that bank risk-taking should be restricted.
But mistakes also suggest the same thing in view of externalities.
Banks (managers) will tend to take excessive risks because they do
not internalize the effect that their failures would have on the rest of
the economy.
They will not think hard about actions and make excessive mistakes.
This is not moral hazard per se– it is simply maximizing own utility.
Regardless of MH or mistakes, it makes sense to regulate banks so as
to make them safer than they would be on their own...

Need: More capital and more liquidity

How much is a diffi cult question
There are two dimensions to bank safety: capital and liquidity.

Economic logic suggests increasing buffers on both margins.
There are also some costs to these policies since they might restrict
bank lending/investment in normal times.
The trade-off is diffi cult, especially if you take the mistakes view.
But it seems reasonable to increase capital requirements somewhat...

Does more capital restrict lending in normal times?
Hanson, Kashyap, Stein (2011): “But will these higher capital

requirements lead to increased costs for borrowers? In what follows,
we focus on the long-run steady-state consequences of higher capital
requirements, setting aside the transitional issues associated with
phase-in of a new regime. To preview, our reading of the theory and
relevant empirical evidence suggests that while increased capital
requirements might be expected to have some long-run impact on the
cost of loans, this effect is likely to be quite small.”
They do some back of the envelope calculations to assess costs.
They also look at the historical record to see trace of costs...


“...we have examined the behavior of various proxies for the markup
that banks charge on loans. In a variety
of regression specifications (not shown here), we found no reliable
time-series correlation between these markup variables and bank
capital ratios.”
“To illustrate the loose ties between loan costs and capital ratios,
Figure 2B plots capital ratios for the period 1920—2009 against two
markup proxies: 1) the net interest margin (net interest income over
earning assets); and 2) the yield on loans (interest income on loans
over gross loans) minus the rate paid on deposits (interest expense on
deposits over deposits). As can be seen, there is no apparent
correlation between capital ratios and either measure of markups.”

So how much more capital?
So evidence is limited but suggests reasonable increases in capital

requirements will not create havoc on the financial system.
HKS also do back-of-the envelope calculation to argue for capital
requirements (in good times) in the order of 15%.
“We have stressed the importance of requiring that financial firms
have both more capital, and, crucially, higher-quality capital. On this
score, the Basel III recommendations look quite good. They would
raise the minimum common equity requirement from 2 percent of
risk-weighted assets to 7 percent (this is inclusive of a “capital
conservation buffer”). While we have argued for a higher number,
this is a significant step in the right direction.”

How about liquidity?
For liquidity, the issues are even more complicated, because what
matters the most is liquidity in crisis/distress times.
Liquidity measures in good times can be misguiding (fire sales...).
There are also various complications (off-balance sheet positions...).
“Liquidity Mismatch Measurement”by Brunnermeier, Krishnamurthy,
Gorton (2013) discuss issues and propose an index.
Bai, Krishnamurthy, Weymuller (2015), “Measuring Liquidity
Mismatch in the Banking Sector”implement for the US banks...

“This paper implements a liquidity measure, “Liquidity Mismatch Index
(LMI),” to gauge the mismatch between the market liquidity of assets and
the funding liquidity of liabilities. We construct the LMIs for 2882 bank
holding companies during 2002 - 2014 and investigate the time-series and
cross-sectional patterns of banks’liquidity and liquidity risk. The
aggregate banking sector liquidity worsens from +$5 trillion before the
crisis to -$3 trillion in 2008, and reverses back to the pre-crisis level in
2009. We also show how a liquidity stress test can be conducted with the
LMI metric, and that such a stress test as an effective macroprudential
tool could have revealed the liquidity need of the banking system in the
late 2007. In the cross section, we find that banks with more ex-ante
liquidity mismatch have a higher crash probability and have a higher
chance of borrowing from the government during the financial crisis. Thus
the LMI measure is informative regarding both individual bank liquidity
risk as well as the liquidity risk of the entire banking system...”
Courtesy of Jennie Bai, Arvind Krishnamurthy, and Charles-Henri Weymuller. Used with permission.

Courtesy of Jennie Bai, Arvind Krishnamurthy, and Charles-Henri Weymuller. Used with permission.
The drop is consistent with the DD/HGG run theories we have seen.

So what causes crises and what to do about them?
Anonymous quote: “We have not succeeded in answering all our problems.
The answers we have found only serve to raise a whole set of new
questions. In some ways we feel we are as confused as ever, but we believe
we are confused on a higher level and about more important things.”
Thank you for your attention and interest!

MIT OpenCourseWare
http://ocw.mit.edu

January IAP 2016
45

Financial Crises

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Financial Crises

Uploaded by

Copyright:

Available Formats

Understanding Financial Crises: Lessons from

Alp Simsek (MIT) Lessons from the History of Crises 1

2 The subprime crisis of 2007-2009

3 Some lessons from the history of crises

Alp Simsek (MIT) Lessons from the History of Crises 2

Associate Professor of Economics

Alp Simsek (MIT) Lessons from the History of Crises 3

We will take an academic look at ﬁnancial-banking crises:

General mechanisms. Subprime crisis as case study/empirics

Alp Simsek (MIT) Lessons from the History of Crises 4

Alp Simsek (MIT) Lessons from the History of Crises 5

2 The subprime crisis of 2007-2009

3 Some lessons from the history of crises

Alp Simsek (MIT) Lessons from the History of Crises 6

Alp Simsek (MIT) Lessons from the History of Crises 7

Subprime mortgage: Borrowers with lower credit ratings.

Alp Simsek (MIT) Lessons from the History of Crises 8

Alp Simsek (MIT) Lessons from the History of Crises 9

Derivative security: Value derives from another security.

Financial innovation created new derivatives in recent years.

An interesting example is collateralized debt obligations (CDOs).

These are constructed in two steps:

Pool underlying securities (mortgages, but also corporate bonds,

Sell claims to parts of the cash ﬁows on the pool (“tranches”).

Alp Simsek (MIT) Lessons from the History of Crises 10

Consider a bond with promise (or face value) of $100.

Alp Simsek (MIT) Lessons from the History of Crises 11

Super Senior Tranche

Bond Portfolio’s Ability to Pay

Image by MIT OpenCourseWare.

Alp Simsek (MIT) Lessons from the History of Crises 12

Credit rating agencies rate bonds according to probability of

There is a scarcity of the bonds with the highest rating (AAA):

These bonds account for only about 5% of the supply of

bond is rated AAA.

The low-rated tranches can be sold to hedge funds and other

Alp Simsek (MIT) Lessons from the History of Crises 13

Courtesy of Efraim Benmelech and Jennifer Dlugosz. Used with permission.

Figure: From Benmelech and Dlugosz (2009).

Alp Simsek (MIT) Lessons from the History of Crises 14

In view of diversification, the risk of AAA tranches

Rating agencies are good at modeling idiosyncratic default risk.

Not so good at modeling aggregate shocks (and correlations).

Alp Simsek (MIT) Lessons from the History of Crises 15

Subprime Subprime Percent

2001 $2,215 $190 8.6% $95 50.4%

2002 $2,885 $231 8.0% $121 52.7%

2003 $3,945 $335 8.5% $202 60.5%

2004 $2,920 $540 18.5% $401 74.3%

2005 $3,120 $625 20.0% $507 81.2%

2006 $2,980 $600 20.1% $483 80.5%

Image by MIT OpenCourseWare.

This is vulnerable to a drop in nationwide house prices. Why?

Alp Simsek (MIT) Lessons from the History of Crises 16

Alp Simsek (MIT) Lessons from the History of Crises 17

Alp Simsek (MIT) Lessons from the History of Crises 18

Figure: From Brunnermeier (2009).

The spreads are calculated from CDS prices. They provide a

measure of the default probability for corresponding tranches.

Alp Simsek (MIT) Lessons from the History of Crises 19

A CDS is an insurance contract on the default of a particular

For example, suppose you own a corporate bond from company

XYZ with principal $1,000. If company XYZ defaults, you might

get back $500 instead of $1,000.