Observations on the Financial Crisis:

Financial Risk Propagation, Structural

Risks, and Policy Responses

Robert C. Merton
Harvard Business School

Central Bank of Argentina Money and

Banking Conference
August 31, 2009
 My remarks will touch on six issues
• Structure of credit-risk propagation. How can large risks build
up without recognition and then appear to explode? Why do
banks continue to report larger losses, even after booking no
new risks?
• On the need for integrated systemic risk policy: superposition
of benign processes can create substantial systemic risk.
• Systemic risk differences between a takeover versus
bankruptcy policy for financial institutions: LTCM vs. Lehman.
• Selected government policy recommendations: risk
measurement and risk management
management.
• Financial innovation and increasing the risk of crisis.
• Syste
Systematic
at c risks
s s from
o the
t e inevitable
e tab e incompleteness
co p ete ess ofo
models.
2
Copyright © 2009 by Robert C. Merton
 Functional Description of Being a Lender When There is
Risk of Default and of Writing a Guarantee of Debt
RISKY DEBT + GUARANTEE OF DEBT = RISK‐FREE DEBT 

RISKY DEBT =  RISK‐FREE DEBT ‐  GUARANTEE OF DEBT 

Corporation 

Operating Assets,  A
p g , Debt (face value B), D
( ),

Common Stock,  E 

A = D + E 
IN DEFAULT, THE HOLDER OF THE GUARANTEE RECEIVES PROMISED VALUE OF 
THE DEBT MINUS VALUE OF ASSETS RECOVERED FROM DEFAULTING ENTITY 
THE DEBT MINUS VALUE OF ASSETS RECOVERED FROM DEFAULTING ENTITY = 
MAX [0, B – A] 

VALUE OF GUARANTEE = PUT OPTION ON THE ASSETS OF BORROWER 

CREDIT DEFAULT SWAPS ARE GUARANTEES OF DEBT AND THEREFORE ARE PUT 
OPTIONS ON THE ASSETS OF THE BORROWER 
3
Copyright © 2009 by Robert C. Merton
 Non-linear Macro Risk Buildup
Firm/Mortgage
g g
C
Corporate/Household
t /H h ld Sector
S t
Debt

DC Liability
DC

D 'C
Corporate/Housing
Co po ate/ ous g Assets,
ssets, A C

Firm/Mortgage A 'C AC
Banking System Government
Debt Bank
G
Guarantee
t Liability Deposit Liability
Guarantee

GC GB

G 'C G 'B
GC GB

A 'C AC A 'B AB Bank Assets, AB

Corporate /Housing Assets, AC
Copyright © 2009 by Robert C. Merton 4
 Integrated Systemic Risk Policy: Refinancing
Ratchet Effect 1996
1996-2006
2006

• Trend #1: rising

g U.S. home pprices
• Trend #2: declining U.S. interest rates
• Trend #3: increasing efficiency of mortgage refinancing
• Each trend taken individually is beneficial or benign
• All three trends superimposed creates unintended
synchronization of homeowner leverage
• Result: residential mortgage market is six times more
vulnerable and estimated losses of $ $1.2 trillion between
June 2006 and December 2008

5
Copyright © 2009 by Robert C. Merton
 Systemic Risk Differences: Capital-Infusion-
and-Takeover
dT k versus B
Bankruptcy
k t
• LTCM (1998) versus Lehman (2008)
• Any default triggers cross-default provisions in securities
and contractual agreements
• Collateral seizures and sales or replacement of
contractual agreements cause a worsening of original
net position exposure
• Typically counterpart for long-side is not the same for
short-side of position
• With capital infusion
infusion, risk exposure is to net positions
• With bankruptcy, risk exposure is to gross positions
• Depending on character of position, gross risk can be up
to 40 times larger than net risk
6
Copyright © 2009 by Robert C. Merton
 Takeover
a eo e vs.
s Bankruptcy:
a uptcy Net
et versus
e sus G
Gross
oss Risk
s

Institution A
Bond #1 Bond #2
Counterpart B (chit from B) (chit to C) Counterpart C

Bond #1 Bond #1 Cash Loan

(collateral) (chit to A) Loan Loan ((collateral)) ((from A))
(to C) (from B)
Loan Bond #2
(to A) Equity (chit from A) Equity
Equity

Risk (default A) Risk (default A) Risk (default A)

Vol (Bond #1) Vol (Bond #1 – Bond #2) Vol (Bond #2)

7
Copyright © 2009 by Robert C. Merton
 Policy Measures: Risk Measurement

• Fair-value accounting g always

y required
q and considered
by regulator, whether or not capital-adequacy ratios and
other specific regulatory rules are based on it.
• Financial
Fi i l institutions
i tit ti provide
id prescribed
ib d risk
i kddata
t tto
central processing authority on confidential or coded
basis. Aggregate risk parameters to regulators and
public.
• Creation of National Capital-Market Safety Board as
independent agency to investigate financial failures and
make recommendations, but no regulatory authority.
(A.Lo, MIT). International coordination is critical but
single global body unrealistic.
Copyright © 2009 by Robert C. Merton
8
 Policy Measures: Risk Management
• OTC derivative contract positions between financial
institutions must have two
two-way
way mark
mark-to-market
to market collateral at
least equal to the contract liability value, independent of
credit rating.
• Central clearing for OTC contracts (above threshold
volume) or else position limits on non-cleared contracts.
• Require financial engineering knowledge among senior
management, board members, and regulators of financial
institutions, including central banks, BIS, and IMF.
• No financial product can be offered with either a fixed
redemption price/NAV or a fixed rate of return without an
explicit guarantor
guarantor.

9
Copyright © 2009 by Robert C. Merton
 On Mathematical
O e c Models
ode s in Finance
ce Practice
c ce

“ Even this brief discourse on the application to finance practice of

mathematical models in general and the options-pricing model in particular
would be negligently
g g y incomplete p without a strongg word of caution about
their use. At times we can lose sight of the ultimate purpose of the models
when their mathematics become too interesting. The mathematics of
financial models can be applied precisely, but the models are not at all
precise in their application to the complex real world. Their accuracy as a
useful approximation to that world varies significantly across time and
place. The models should be applied in practice only tentatively, with
careful
f l assessment off their
h i limitations
li i i in
i eachh application.”
li i ”

R.C. Merton, “Applications of Option-Pricing Theory: Twenty-Five Years

Later”, Nobel Lecture, 1997.
10
 Models are Always Abstractions from Complex Reality:
I li ti
Implications for
f Ratings
R ti Agencies
A i andd Regulators
R l t
Credit Evaluation: 1) Probability of Default

2) Expected Recovery Rate in Default

3) Degree of Procyclicality in Default

R ti
Ratings A
Agencies
i (S&P and
d Fitch)
Fit h)

1) Ratings based on Probability of Default only

Incomplete model for ratings induces bias in assets selected for structures

Behavior: Maximize value, subject to meeting ratings constraint

Minimize cost, subject to meeting ratings constraint

Prediction of bias in asset choices from good behavior

 Low Expected
p Recoveryy Rate in Default

 High Procyclicality (“Beta”) in Default

Copyright © 2009 by Robert C. Merton 11