8
W^^\
MIT LIBRARIES
I
'
'
^
^
'
lllllil
3 9080 03317 5818
Massachusetts Institute of Technology
Department of Economics
Working Paper Sehes
The "Surprising" Origin and Nature of Financial
A Macroeconomic Policy Proposal
;
Ricardo
J.
Caballero
Pablo Kurlat
Working Paper 09-24
September 14,2009
RoomE52-251
50 Memorial Drive
Cambridge,
This
MA 02142
paper can be downloaded without charge from the
Paper Collection at
hitsp://ssrn.com/abstract=^ 473Q
Social Science Research Network
1
1
Crises:
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/surprisingoriginOOcaba
The "Surprising" Origin and Nature of Financial
A Macroeconomic Policy Proposal
Ricardo
J.
Crises:
Caballero and Pablo Kurlat^
September
14,
2009
Abstract
Severe financial crises
in
'"
developed economies are produced by
a
and
a slow/ policy
response.
We
propose
•--'
combination of three factors:
negative surprises that create uncertainty, concentration of macroeconomic
financial institutions
'-
a policy
risk in
leveraged
instrument. Tradable
Insurance Credits (TICs), designed to address crises stemming from these factors. TICs would be
issued by the central bank and give their holder the right to attach a central bank guarantee to
assets on
its
balance sheet, but only during a financial
required to keep a
monetary
policy
minimum
and
fits
crisis; financial
institutions w/ouid be
holding of TICs. TIC policy could be carried out
frameworks;
into existing institutional
have been used to address the 2007-2009 financial
crisis in a
in a
similar
we examine how
faster and
way
to
TICs could
more systematic way
than the ad-hoc measures undertaken.
JELcodes:G01, G28, E58
Keywords: financial
crises,
Knightian uncertainty,
macroeconomic
risk,
credit default swaps,
asset insurance
'
MIT and NBER, and MIT;
Macroeconomic
Giavazzi, Bengt
Stein
Policy,
respectively. Prepared for the Jackson Hole
August 20-22, 2009.
Homlstrom,
We
Anil Kashyap, Arvind Krishnamurthy,
and symposium's participants
for their
Symposium on
thank Tobias Adrian, Peter Diamond,
Financial Stability
Jeff
and
Fuhrer, Francesco
Juan Ocampo, Ken Rogoff, Alp Simsek, Jeremy
comments, and Matthew Huang
for outstanding research assistance.
Introduction
I.
The recent
financial crisis has
and
financial institutions,
can
we do
caused previously unimaginable wealth losses, the demise of
a global recession.
What
is
behind severe financial
to prevent a relapse or at least to reduce the
elite
and w/hat
crises,
economic costs of the next one?
Most professional (and amateur) economists and policymakers are currently seeking answers
we
these questions, and
are no exception. However,
rushing into conclusions and
we argue
and so on.
stopping there
is
We
is
converging too quickly on the standard post-crises themes of
is
insufficient regulation, real estate bubbles, excessive leverage
policy,
wisdom
that the conventional
to
and
monetary
capital flows, lax
do not disagree with the importance of some of these themes, but
bound
to lead either to fairly conventional policy
recommendations which
have already proved to be insufficient to prevent severe crises from recurring, or to asphyxiate
the
development
the
of
system
financial
through
excessive
requirements
capita!
and
deleveraging.
It is
important to avoid
falling into this intellectual
and policy
analysis on factors that are not part of the core of the conventional
wisdom.
In
highlight the importance of three key ingredients for severe financial crises
financial markets.
we mean
that
is
crisis
many
including
and most central ingredient
is
confusing, a (perhaps temporary) change
the surprise
mortgages were the
of
first
a significant
focus our
we
particular,
in
developed
negative surprise. By this
not a large negative shock of the kind economic agents can foresee, but something
new and
current
The
we
trap. For this reason
first
was not the decline
in
in
the paradigm.
real estate prices or
In
the context of the
the fact that subprime
to be affected by this decline. Rather, the surprise
was
in
parts of the financial system, even those very distant from the subprime
all
the distress
market
itself,
became
structured products, commercial paper, and interbank lending. Linkages
too complex and hard to understand, prime counterparties were no longer perceived as such,
and panic ensued.
In
many
of these instances, the data
wisdom-after-the-facts.'
potential
weak
links.
Reality
Ex-post,
it
available to recognize the problem, but this
immensely more complex than models, with
is
is
was
all
the
possible
linkages
irrelevant except during a severe crisis
paradigm, from irrelevant to
mostly
millions of
easy to highlight the one that blew up, but ex-ante
different matter. Each market participant and policymaker
understanding
is
critical
across these
when they
turn
knows
different
critical)
is
their
own
Vi/orlds
local
(which
is
a
world, but
are
mostly
too complex. This change
in
linkages, triggers massive uncertainty, indeed Knightian
See, e.g., Caplin and Leahy (1994) for a social learning
model of wisdom
after the facts.
unknowns
uncertainty (when the
shift
from known to unknown), and unleashes destructive
flights to quality.^
The second ingredient
in
is
for severe financial crises
is
the excessive concentration of aggregate
emphasis here
highly leveraged (systemically important) financial institutions. Note that our
on the concentration of aggregate
the problem
than
in
the current
in
risk
rather than on the
was not leverage per
crisis
se,
much hyped
risk
leverage.
our view,
In
which was high but not much higher
past recessions that did not turn into financial catastrophes, but the fact that banks had
held on to senior and super-senior tranches of structured asset backed securities which
more exposed
suggest. Thus,
to aggregate surprise shocks than their
when
when
rating,
were
misinterpreted, would
systemic confusion emerged, these complex financial instruments quickly
soured, compromised the balance sheets of their leveraged holders, and triggered asset
The
sales that ravaged balance sheets across financial institutions.
result
was
a vicious
fire
feedback
loop between assets exposed to aggregate conditions and leveraged balance sheets.
The
and
third
ingredient
last
we
highlight
is
response that
a policy
consequences of the previous ingredients. Almost every severe
phase, where
financial
it
too slow
addressing the
in
has an early bifurcation
crisis
can be contained by a decisive and systemic policy intervention supporting the
system or exacerbated by policy
unorthodox interventions
crisis,
is
The reasons
timidity.
because policymakers are often shell-shocked as
the current
is
for the latter are
well, but also
because the
many, not
political
simply too slow to catch up with a financial system
in
free
for
fall. In
Lehman
only turned systemic and aggressive enough after the
policies
least
tempo
debacle. Only then did politicians and policymakers (and most academic economists and
journalists)
from
seem
ill-timed
to have gotten the evidence they
to pick up the pace
and move away
moral hazard and distributional considerations. Unfortunately, by
task had already
grown
to
enormous proportions, and soon enough the
when
constraints built back again
implemented
needed
in late
political
this
time the
pressures and
faced with the exorbitant costs of an effective policy-package
stage.
For normal contractions, the institutional solution to the rigidity and biases inherent to the
process
political
is
to
remove monetary
policy
from
flexibility to react
flexibility
they have to deal with severe financial
is
most needed,
to deal for instance with
is
crises,
which
is
limited to lender of last resort functions,
See Caballero and Simsek (2009a, b) for
for a
central banks are
precisely
a
in this
model
of
paper
is
liability
crises.
a
rapid
insufficient
side of balance
to propose a policy that
endogenous complexity during
model of Knightian uncertainty during
when
which can be
problems originating on the asset rather than
sheets, or outside commercial banks. Our goal
and Krishnamurthy (2008)
Modern
with adequate speed to regular business cycles. Unfortunately, the
given the
response
this process.
crises,
is
a close
and Caballero
Other paradigms besides Knightian
uncertainty have similar implications; for instance hot/cold decision-making (Bernheim and Rangel, 2005).
analogue to monetary policy (and conventional capital-adequacy requirements)
management, but
In
that
is
The
surprises.
How/ever, there
responding to surprises stemming
a certain
order
in this
chaos.
In
is
since
first,
change
it
not understood
common
particular, a
implicit insurance. This observation
When
policy response: public insurance provision.
government must
quicl<ly
immediately hints
a
systemic
in
w/ould
the perceived
seem that
until after
in
the
it
is
happens.
it
pattern across
that the confusion triggers panics, and panics trigger spikes
is
and
explicit
discouraging at
is
something that keeps changing and
is
its
reducing the exposure of the leveraged financial institutions to these
in
paradigm and w/orking of the system,
episodes
in
centrality of surprises, defined as an (often temporary)
difficult to fight
terms of
targeted to systemic crises prevention and control.
addition to being flexible, the policy must be useful
from complexity and
in
all
these
demand
for
at the core of the required
uncertainty strikes,
crisis of
tlie
provide access to reasonably priced balance-sheet insurance to
financial institutions.
Drawing an analogy with cardiac arrest treatment, we view our policy proposal as the
equivalent to the strategic placement of defibrillators.
We
know
all
that a low-fat diet and
plenty of exercise and rest are important preventive factors, but two-thirds of the sudden
deaths from cardiac arrest
still
take place without any prior evidence of disease, and most of
these deaths could have been prevented with adequate treatment within the
minutes of the cardiac
arrest.
Under our proposal, the
TIC would entitle
during
a
systemic
set by the
CB
its
central bank (CB)
would
to hold
critical
a central
issue tradable insurance credits (TICs). Each
bank guarantee to assets on
The amount of TICs required to insure
fundamental
to adjust for differing
would be allowed
four
.
holder to attach
crisis.
first
and use
a
its
balance sheet
given type of security
would be
riskiness. All regulated financial institutions
and possibly hedge funds; private equity funds and
TICs,
corporations would be allowed to as well (especially under the extended reach of the
new
regulatory proposal). Prudential regulations, however, would require that during normal times,
regulated financial institutions hold a
minimum amount
of TICs as a proportion of risk-weighted
._..-
assets and systemic importance.
.
At a minimum, the TIC-policy would have the following key components:
A
convertibility rule.
The CB determines
a (necessarily noisy) convertibility-threshold level for
systemic panic, above which TICs can be attached to a bond or portfolio. An attached TIC
simply
a central
is
bank backed CDS.
See Caballero and Krishnamurthy (2008) for
and the optimal response of the central bank
in
a
model of
crises triggered by spikes in Knightian uncertainty
terms of insurance provision. To
provides the formal perspective underpinning the current paper and proposal.
a large extent, that
paper
Open market
sell
operations. During normal times, winen TICs are not convertible, the CB can buy or
TICs at a market price.
Minimum
hiolding
important institutions must preserve
How would
that
it
a
minimum
TIC/Assets ratio.
have worked during the current
TICs
approximately $2
normal times, highly leveraged and systemically
During
requirements.
trillion
(notional) worth of
TICs.
Suppose
crisis?
U.S.
banks had
The Fed could have responded
first
held
by hinting
might declare TICs convertible and then by making an increasing proportion of the
outstanding TICs convertible. Arguably, this would have required converting approximately the
amount
of ad-hoc insurance for specific institutions that
about $500
It
is
that an insurance policy exchanges
Rather, the TIC-policy
institutions will
is
is
(at least in
not a conventional insurance policy
is
a
fee,
the sense
ensures that financial
it
have access to insurance for their assets during systemic
as
in
fee during normal times for a cash injection during crises.
a
an "insurance-squared" policy: For
central, as a
economic agents behave
is
actually offered during the crisis,
billion.
important to notice that the TIC-policy
the policy
was
crises.
This aspect of
core feature of panics caused by Knightian uncertainty
if
the likelihood of a catastrophe were
the absence of the panic
economic agents are able to go back
higher than
it
that
actually
By providing insurance against these catastrophes,
itself).
to their
to ever have to deliver on this insurance (or
much
is
normal
is
activities
at least
much
while the government
less likely
is
unlikely
than panicked investors
estimate).^
Note also that TICs are equivalent to CDS during systemic
That
is,
TICs are contingent-CDS.
crises but
They become activated only when
not during normal times.
systemic
a
crisis arises.
By
targeting the event that needs protection, this contingent feature significantly lowers the cost
of insurance for financial institutions. This
calls for
In
is
an important advantage of this approach over the
higher capital adequacy ratios and deleveraging.
summary, the
TIC-policy
framework can
significantly stabilize the value of banks' assets
equity during systemic events and hence limit the panic-driven
that characterize severe financial crises. Through
fire sales
and
and market freezes
open market operations and announcements
on the convertibility threshold, not too different from the way central banks conduct monetary
Caballero and Krishnamurthy (2008) provide an example where
the event agents worry about
is
the average). The central bank cannot
providing a very low cost insurance.
collateral created by the
fall
the presence of Knightian uncertainty
into this fallacy of composition,
Woodford (1990) and Holmstrom and
government (and hence
funds for insurance provision.
in
individually plausible but collectively impossible (not
its
credibility),
everyone can do worse than
and hence solves the problem by
Tirole (1998) address the source of the
which stems from
its
ability to
pledge taxpayers'
policy,
they can manage the devastating consequences of
contemporaneous
financial sector without massive
much lower
The
rest of the
paper
is
two main
divided into
lining to financial
infrastructure which
in
how
it
parts and a conclusion.
we
crises
can then
fix
the future. The current financial
policy
mandates
to
more
detail.
In
the
In
first
part
the second part
we
we
of the alternatives.
Crises in Developed Economies?
that they reveal severe weaknesses of the financial
is
economy
to improve the resilience of the
crisis
symposiums and proposals devoted
monetary
in
compares with some
What Causes Severe Financial
silver
at a
cost to the private sector than large capital requirements.
describe the TIC-policy and
The
and
injections of public resources,
describe the three ingredients behind severe crises
U.
perceived uncertainty on the
rises in
is
no exception, and
to these
new
complete overhauls of
we have seen
to similar events
a large
number
of
These proposals range from new
fixes.
financial regulation, at both
domestic and
global levels.
To
cite just
two examples, the
first
Summit on Reshaping the Global
"...
world
leaders
will
internationally to agree
management by
two items
Financial
of the Executive
System
of the
2009 London
state:
how government and
consider
Summary
regulators
can
together
worl<
what further steps are needed to enhance corporate governance and
financial
institutions;
risk
agree on steps to strengthen prudential regulation,
including requiring banks to build buffers of resources
in
good
times...."
,,
and the Administration's recent reform plan includes:
"raising capital
and
requirements, with more stringent requirements for the largest and
liquidity
most interconnected
firms,
[...]
impos[ing] robust reporting requirements on the issuers of asset-
backed securities [and] harmonizing the regulation of futures and securities"
..
;
While the proposals that emerge from such meetings and policy committees are mostly good
and sound advice, they are hardly new, and more importantly, they run the serious
trivializing
progress
is
the nature of severe crises. Almost every
made, but
just as
one hole
themselves create new ones). This
is
plugged,
inability to
importance of the "newness"
factor, while
is
new ones
make
from the very nature of the dynamic process of
crisis
risk of
followed by similar proposals, and
arise (and
sometimes the
policies
the system completely crisis-proof stems
financial
development. Unfortunately, the
often mentioned,
is
minimized
at the
time of
proposing concrete policy solutions.
In
the remainder of this section
policy proposal put forth
in
we
Section
III.
describe the three features of crises that motivate our
Surprises
H.A.
There
human
extensive experimental evidence that
is
animals and faces
prevalent
in
in
clouds,
etc.
(pareidolia
beings crave patterns - sunbathers see
and cluster
illusion).^
time ago.
a long
it
also
is
common patterns
problem or we would have
the analysis of economic and financial crises. Surely there are
across the preludes of crises, but these cannot be the core of the
solved
This tendency
estate values to drop,
If all
crises
we would
were caused by underestimates
already have found a
way
to mitigate this particular risk.
from formal or informal case studies. Caprio and Honohan
Policy conclusions often derive
(2008) examine the factors that contributed to each of the
Woods
era.
involve
some combination
While each
crisis
of:
of the potential for real
is
many
crises of the post-Bretton
different, they argue that the contributing factors usually
fraud,
lax
mismanagement, excessive
internal controls,
risk
taking, financial liberalization, government-directed credit decisions, taxes or tax-like policies,
over-optimism, and herd effects. But case studies, by construction, suffer from selection bias:
The explanations and
appear too certain, as they ignore the many instances when
were present but nothing happened
similar factors
More
study).
culprits
the
generally,
certainty
in
these are not the subject of the case
(as
researchers'
and
policymakers'
authoritative
conclusions on the causes of crises contrasts sharply with the very limited predictive
models have
anticipating crises, even
in
when they
include the very
same
power
that
factors that are ex-
post considered as obvious parts of the explanation.
For example, so-called Early Warning System models, following Kaminsky, Lizondo and Reinhart
(1998), and Kaminsky and Reinhart (1999), attempt to predict banking and/or capital account
crises.
They do so by tracking the behavior of several macroeconomic and
and by issuing
a
financial indicators
warning signal when the estimated probability of an upcoming
crisis
crosses a
threshold. The KLR model issues warnings on a variable-by-variable basis and then combines
these individual warnings
(2004) survey the predictive
warnings are
their predictive
power
both
domestic credit growth and
in-sample
and
a
probit model. Berg et.
out-of-sample.
real interest rates,
account for
including
would
to
power. However, the magnitudes of the coefficients are not very
9%
more
without
a
warning.
It
is
models'
predictive
large.
warning
On
is
the
29%,
certainly possible that a different specification or
variables (real estate values, for instance, are not included
improve the
macroeconomic
Financial-sector
a large fraction of
basis of the KLR model, the conditional probability of a crisis being given a
compared
al.
models and conclude that the
of several variants of these
significant,
statistically
variables, such as
an index; other specifications use
in
power.
To
a
first
in
approximation,
the KLR model)
however,
the
factors that are often cited as the causes of crises explain about 20 percentage
points of increased likelihood. Residual uncertainty, including surprises and factors other than
SeeGilovich,
T. et. al,
(2002).
commonly analyzed macroeconomic
points.^
full
If
nothing else, this evidence suggests
we
should be humble about claiming to have
a
understanding of w/hat triggers financial crises and, consequently, about having policy
recommendations that we claim
banking
will
prevent them.
using an extended database, Reinhart and Rogoff (2008) study the incidence of
Similarly,
in
account for the remaining 80 percentage
variables,
They
crises.
on average every 14 years
find that crises are quite frequent, taking place
advanced economies and every
9 years
in
emerging economies. Capital account "bonanzas",
which are often mentioned as sources of unsustainable imbalances, do indeed increase the
The probability of
likelihood of crises:
after the
bonanza
significant
a
banking
18%, whereas the unconditional probability
is
and informative difference but not
occur or what
causes are. Furthermore, these figures
its
account
(a
government bonds.
account
deficit,
common
"sudden stop") tend to be more
times of
crisis,
crisis,
in
going to
change
in
haven
in
in
the
advanced
the perceived security of US
the U.S. did not have to rapidly reduce
which has gone from 6.1% of GDP
is
sample that includes both
emerging than
in
2006, to 5.3%
in
basket of currencies between the second quarter of 2007 and the
called global imbalances are certainly part of the
its
2007, to 4.9%
first
in
current
2008.
In
it,
as
some had
and incorporate the most obvious lessons of the
prevent the
ills
listed
crisis,
but
it is
not
feared.
It
would be unwise not to
past, such as the
importance of trying to
not our goal to discredit standard policy analysis and conclusions.
learn
quarter of 2009. So-
background of the current
the rapid unwinding of these imbalances which has caused
by Caprio and Honohan, into economic analysis and policy. Our point
simply that these lessons are only
impact of severe financial
it
a crisis
the value of the dollar increased approximately 6.3% against a broad trade-weighted
fact,
It is
a statistically
the U.S., capital flows have usually had a
in
as investors seek a safe
the current
In
a
when
crises associated with a sharp
economies (Mendoza and Terrones, 2008). Indeed
stabilizing role in
come from
window before and
13%. Again,
is
a decisive indication of
developed and emerging economies. Financial
capital
within a three year
crisis
a small part of
crises. Placing
too
what
is
needed
to prevent
is
and soften the
much weight on them may be counterproductive
if
translates into excessive straightjackets, a false sense of security, or simply an outdated and
hence wasteful precautionary measure. These concerns are
The values
for the
all
the
more
relevant given the
KLR model follow from the following table which includes both in-sample and out-of-sample
observations
False alarms are issued
in
26%
Warning
No warning
Crisis
333
232
No
803
crisis
of non-crises
2229
compared
to accurate alarms in
59%
of crises. The table cannot be
used directly to compute standard errors around the estimates of conditional probabilities because there
substantial serial correlation
in
the observations (Berg
et. al.
2004).
is
rapidly integrating global
according to
economy
that will undoubtedly bring back
many policymakers and commentators,
is
new
global
economy.^
important to build
In this
is
resilient to
not, these are part of the structure of
tautological since absent a surprise
(in
and
it
these accidents.
To some extent, the claim that surprises are
would be prepared
imbalances
context, unexpected accidents will continue to happen,
system that
a
of the elements that,
"trigger" crises, such as global
and high asset valuations. Whether we embrace them or
the
many
a
necessary condition for financial crises
is
most instances) economic agents and the government
for the negative event.
However, our claim
is
more
subtle than this. The
surprises that have the potential to trigger severe crises are not simply bad realizations within a
known
probabilistic environment. Rather, they are
"rare event" feature that holds the key as
it
changes
in
the environment
itself. It is this
has the potential to trigger sharp rises
uncertainty and flight to quality. Surprises of this kind destroy an
in
perceived
enormous amount
intangible capital built from an understanding of
how
management paradigms have
uncertainty-management ones, but the
is
not something
good
human
to be replaced for
of
the (previous) financial world works. Risk
latter
beings, let alone highly leveraged financial institutions, are particularly
at.
This perspective contains a
more general message
as well.
The continuous potential for
of-the-game changing surprises stems from the process of financial development
financial
itself.
rules-
New
instruments and practices emerge continuously and, by their very essence, are
untested with respect to major events and disruptions. Their
first
potential for large dislocations and confusion. For example, the
test invariably creates the
1970 default by the Penn
Central Railroad on $82 million of prime-rated commercial paper, a relatively
new product
at
the time, threw doubt onto the entire commercial paper market, causing investors to retreat.
Similarly, the collapse of
a
complex strategy to
Long Term Capital Management (LTCM), a giant hedge fund that used
profit
from small movements
and caused an immediate global
financial
panic.
in
In
security prices
contrast, the
was the
first
of
its
kind,
1997 Mercury Financing
commercial paper default, and the 2006 collapse of the Amaranth hedge fund caused
little
liquidity disruption.^
In
the current financial
crisis,
several turns of events, over and
beyond the
large decline in real
estate values, took market participants by surprise.
A comparatively slowly
evolving surprise
Many banks had
large holdings of senior
and super-senior
was the
crisis
of monoline insurers.
tranches of CDOs, insured by AAA-rated monolines. The insurance helped lower banks' capital
requirements and,
in
theory (and
in
financial reporting),
lowered their exposure to subprime
For a characterization of the macroeconomics of asset shortages, see (Caballero 2006, and Caballero et
2008a,b).
For a
more
detailed discussion of these and other examples, see (Caballero and Krishnamurthy, 2008).
al
assets.
was
However, monolines were themselves highly exposed to subprime
realized. In late 2007,
it
become
started to
clear that
assets,
subprime mortgages would indirectly
lead to huge losses for monolines, raising questions about their solvency. ^°
finally
downgraded
in
2008, this cascaded onto the over $2
including but not limited to subprime
investors and forced banks to write
assets. In retrospect
it
seems
CDOs."
down and
themselves. But
was not obvious
this
hold capital against large positions
was almost
risk
at the time, since
had been thought,
as reliable as
from spreading the
crisis into
on CDOs did not
"The troubles
at
most
swaps
When
was
it
decrease
realized that the
were
monolines
contributed to the overall uncertainty (aside
this
Lehman declared bankruptcy.
In
a leading
money market
testimony to Congress on
Chairman Bernanke stated that
Lehman had been
well
known
that the failure of the firm
-
really
of the instruments involved
for
some
time, and investors clearly recognized
as evidenced, for example, by the high cost of insuring Lehman's debt
default
the insured
the municipal bond market, yet another indirect linkage).
fund, "broke the buck" after
23, 2008,
in
perfectly correlated with the assets
A more sudden surprise took place when the Reserve Primary Fund,
September
they were
This triggered selloffs by ratings-constrained
untested with respect to major market disruptions.
were not
When
worth of assets they insured,
trillion
clear that monolines' insurance
banks' exposures, since the counterparty
more so than
was
-
the market for credit
in
Thus,
a significant possibility.
we
judged that
investors and counterparties had had time to take precautionary measures."
Chairman Bernanke was
its
ramifications
One
still
certainly right that
caught
of those ramifications
drop below one
about 1.2% of
dollar.
was
was not
that the Reserve Primary Fund
million in
Immediately after Lehman
massive run, with over $30
failure
billion in
entirely unpredicted. Yet
participants and policymakers by surprise.
The fund had invested $785
assets.
its
many market
Lehman's
filed for
saw
its
Lehman
Net Asset Value (NAV)
debt, which constituted
bankruptcy, the fund suffered
redemption requests (about
half of
its
a
total assets) before
a.m. the following day.
Money market
funds had been considered extremely safe, and had indeed benefited from the
flight to quality
it
stopped accepting redemption requests
at
$1 at 11
during the previous year, growing by about $850
Reserve Primary Fund's
NAV caused
There were net redemptions
for
billion
At the time,
rating;
if
it
lost
it
it,
in
the
investors to question the safety of the entire industry.
about $170
.billion
from prime funds towards funds investing exclusively
AAA
(34%) since mid 2007. The drop
during that week, as well as a large shift
in
government debt.
was estimated that MBIA would require about $4
the loss of market value of the assets
it
In
order to stem the
billion in additional capital to retain its
insured was estimated at around $200
billion
(Bloomberg, Decembers, 2007).
Ambac was
first
downgraded by
Fitch
on January
18;
MBIA was
first
downgraded on
April 4, also by Fitch.
10
on September 19 the
panic,
compensate investors
In
if
the
Treasury announced
U.S.
NAV
of participating funds
fell
guarantee program that would
a
below
$1.
retrospect, the consequences of Lehman's demise on the Primary Fund could have
predicted. Public filings
showed
large investments
in
Lehman
as early as
of a generally less conservative investment strategy (Investment
Anyone who took the trouble
it
inevitable that
is
might be further losses
quality. In the
some
dots
previously
in
Company
will
in
part
2009).
Institute,
what might
record of stability that had always
a track
unnecessary to inspect their holdings. Moreover,
system,
November 2007,
to connect the dots could, in principle, have foreseen
happen. However, money market funds had
been
made
complexly interconnected financial
a
remain unconnected. The realization that there
unexamined places
led investors to intensify their flight to
terminology of Gorton and Pennacchi (1990) and Holmstrom (2008), the losses
money market funds from
the Primary Fund suddenly transformed
assets to information-sensitive ones.
it
at
information-insensitive
Investors fled toward the few assets that
were
still
perceived as not requiring acquisition of information.
Surprises also played a large role
debts
in
August 1998
bonds had been
this
rising for
surprised the markets
exposed not
was
to
Russia defaulted on
including counterparty
basis points by the
which some key
LTCM was
began to
highly
on-the-run/off-the-run arbitrage
rise,
as illustrated
exposed to the surprise,
in
most
Figure
1.
for instance
in
July.
What
LTCM, were
As the troubles at
liquid assets, prices of
This soon
became
self-
making huge losses on
when spreads widened. The Fed-sponsored
creditors plus an unscheduled interest rate cut
end of
institutions, notably
investors fled towards the safest and
risk,
its
unexpected event. Spread on Russian
certainly not a completely
was the degree
When
just to Russia itself but to the reappraisal of risk that followed.
reinforcing since
its
previous crises, or near-crises.
months and had reached 1200
LTCM became known and
risk,
in
rescue of
LTCM
by
October 1998 persuaded investors that the
worst scenarios would not be reached and calm returned to (developed) financial markets
"^^
relatively quickly.
Emerging markets and high yields did experience
a crisis
but this did not compromise the core of the
global financial system.
11
Figure
Surprises during the 1998 flight to quality
1.
The focus on surprises does not mean that
conventional kind, especially
at his
in
impact of the subprime shock,
was not
to
miss
since this
it,
understand the impact of
all
is
difficult to
was
Warren
others.
is
a far
there were no excesses of the
Buffett publicly scolded bankers
"^^
However,
if
one looks
(relatively)
"small change,"
when
the entire securitization markets
illustrates the difference in the orders of
and the losses derived from
stems from
it.
We
computed the evolution
magnitude of the shock
May
4 (Bloomberg) - Berkshire Hathaway
regulators for being blind to the possibility
in half
home
Inc.
itself
of the market value (equity plus
long term debt) of the major U.S. banks since January 2007."^" From this
recession
it
itself.
The following calculation
"
at the direct
but rather their failure to
subtle and likely kind of mistake - as
more
impending
support these claims. The bankers' main mistake
the confusion that followed
to be questioned. This
complexity
it
crisis
for not anticipating the impact of the
many
the real estate market, as did
in
the current
the subprime market.
famous yearly meeting (Omaha 2009)
decline
began
in
we
obtained an
Chairman Warren Buffett lambasted bankers, insurers and
prices could
fall,
and said their shortcomings caused the worst
a century.
The procedure
for estimating this
was
as follows: For equity,
bank's market capitalization, excluding increases
in
we
simply tracked the evolution of each
the market cap due to issues of
new
shares. For debt,
we
estimated the duration of each bank's long term debt (including any preferred shares) from the maturity profiles
described
in
the 10-K statements as of December 2007, assuming the interest rate was equal to the rate on 10-
year Treasuries plus the spread on 5-year CDS for each bank, obtained from JPMorgan. Assuming an unchanged
maturity profile,
we
then tracked the changes
basis of the evolution of the
in
the implied market value of each bank's long term debt on the
CDS spread. The banks included
in
the calculation are the 19 banks that underwent
the "stress tests" plus Lehman, Bear Stearns, Merrill Lynch, Wachovia, and Washington Mutual.
12
estimate of total losses on the right hand side of these banl<s' balance sheets. Absent any
feedback
effects, this
should be equal to the losses suffered by the assets on the
of the balance sheet. However, as illustrated
in
Figure
2,
we
left
hand
side
on the right hand
find that losses
side are on the order of three times the IMF's (evolving) estimates of losses related to
mortgage assets accruing
to U.S. banks.
Bear Stearns, the overall
loss in
assets.
The market began to
"'^
Beginning
2008 and increasingly
in
after the
fall
of
market value becomes larger than the losses from subprime
price
in
from the
losses
overall disruption of financial markets, the
severe recession and losses on other types of assets which far exceeded the estimated losses
from the mortgage market
Iff
Figure
2.
07
MarD7
itself.
M^Q7
Sep 07
lul07
Nov 07
J3n OS
-
,,
,
Mat OS
May OB
Jul
AugOS
08
OclOS
DecOS
Mai 09
MoyO!
Losses from mortgage assets, total loss of market value and multiplier. Sources: IMF G!oba!
Financial Stability Reports, banlcs' fmancia! statements and IPMorgan.
The
losses for the overall
loss of
economy
are, of course,
an order of magnitude larger than simply the
market value of the banks. The cumulative
loss of
output from the recession over the
next five years, using projections from the Congressional Budget Office,
$5
trillion.
The stock market
lost
and March 2009, approximately
has since recovered about $3
stability of
approximately $9
six
trillion
between
its
will
be of the order of
peak
in
October 2007
times the estimated losses of the banking sector alone.
trillion
of the losses, partly
the financial system. This size of the recovery
due to increased confidence
is
again far larger that
in
It
the
what can be
attributed to the direct effect of market support policies.
The IMF uses
a
projection of
macroeconomic variables and default
market values to estimate losses on subprime-related
overreacted due to
fire sales,
securities.
rates to estimate losses
on loans and
To the extent that marl<et prices of securities
our procedure understates the multiplier.
13
The transmission mecinanisms by which the
large part has to
Figure 3
initial
losses
do with how negative surprises disrupted
show the
evolution, during the
summer
become
financial markets.
2007 and the
of
AAA
the LIBOR-OIS spread and the implied spread on the 2005-1
of insuring against default by
half-of-2006 vintage.
'^^
AAA
fall
of
The two panels
2008 respectively,
AAA
in
of
ABX, which measures the cost
tranches of subprime mortgage-backed securities of the
beginning of 2007,
At the
amplified are many, but a
first-
MBS were
tranches of subprime
perceived as very safe, with spreads around 25 basis points; the LIBOR-OIS spread was around
10 basis points. As bad news from the housing market continued to accumulate during the
half of the year, the
was
ABX
started to climb, reaching around 100 basis points by the end of July.
well understood that a decline
assets, although the
in
AAA
capital structure to affect
losses,
and the extent to which they might
might reach
AAA
it
is
this
second trimester of 2007, they
might have for interbank markets
markets more generally. The LIBOR-OIS spread remained low
shot up by 27 basis points after
BNP
Paribas
announced
three investment funds that were heavily invested
Central Bank provided €95 billion
soundness of the
A
Even as investors realized the
instructive.
securities during the
remained unaware of the ramifications that
financial
up the
trickle
tranches of structured products, was realized only gradually. The
contrast with the path of the LIBOR-OIS spread
and
It
housing prices would be bad news for subprime mortgage
magnitude of the
possibility that losses
first
emergency
in
until
in
particular,
August
9,
when
would freeze withdrawals from
it
subprime mortgages and the European
liquidity to banks.
Suddenly and unexpectedly, the
system became questioned.
financial
qualitatively similar but
more extreme pattern was repeated
in
September 2008. At
a
time
without any particularly bad news from the housing market, as evidenced by the high but stable
ABX, the bankruptcy of Lehman Brothers and the run against money markets were followed by
a
new jump
of the LIBOR-OIS spread,
different paths of these
shocks and what
we
two
call
considered an unlikely
which reached peaks of over 300 basis points. The
variables illustrate the distinction
"surprises".
possibility, as
The magnitude
of the shock to housing-related assets
evidenced by low spreads on
disruption of financial markets that resulted from this shock
even after large losses from housing assets had become
change
in
A
between large-yet-unsurprising
a
was
AAA
was
tranches. However, the
a true surprise,
unexpected
concrete possibility and involving
a
the understanding of the financial environment.
similar
comparison
is
made
by Gorton and Metrick (2009).
14
Figure
3.
Evolution of L!BOR-OiS spreads and ABX.
Surprises quickly trigger a chain of unexpected events following from the panic they engender.
One consequence
of the surprise
was the almost complete collapse
issues of asset-backed securities, as illustrated
of
new
issues of real-estate-related ABS,
during 2006 to virtually zero
in
two
thirds
in
fell
Figure 4/^ The
left
market for new
panel shows the evolution
which collapsed from about $200
billion
the second half of 2008. The right panel shows
ABS whose underlying loans are not
student loans, etc. These
in
of the
less
per quarter
new
issues of
directly related to real estate: auto loans, credit cards,
than mortgage ABS during 2007 but collapsed by more than
the third quarter of 2008. Securitization markets
became questioned
entirety, depriving issuers of a previously reliable source of funding
in
their
and feeding back into the
generalized scramble for liquidity.
Figure
4.
New Issuance of as.set backed
We thank Tobias Adrian
securities in previou.s three oionth.s. Source: |P
for sharing this data, also analyzed in Adrian
Morgan
Cha.se
and Shin (2009).
15
Il.B.
Aggregate Risk Concentration
The second important condition
in
Leveraged Financial Institutions
for a severe crisis to
develop
is
that the highly leveraged and
interconnected sector of the economy, typically the financial sector, be significantly exposed
(directly or indirectly) to a surprise of the kind discussed earlier.
In
general, the combination of leverage and exposure to aggregate risk
is
bad one, as
a
the potential to create severe downw/ard feedback-loops. IVIoreover, there
is
it
has
an extensive
literature describing different pecuniary externalities that lead private financial institutions to
take on too
much aggregate
any
it is
kind, so
amplified
runs (effectively,
Figure
5,
much
crisis
concern
is
these are susceptible to severe panic-driven
and the market for new issues (Figure
measured leverage
outset of the current
crises, since
larger aggregate shocks), as illustrated by the collapse
(Figure 3)
this
the aggregate exposure comes through nev^ financial instruments
been tested through
vuhich have not
ABS
the social optimum. "^^ Leverage magnifies risks of
reasonable to be concerned about excessive leverage. How/ever,
many times when
of existing
risk relative to
ratios,
except for foreign banks,
4).
in
both the values
Indeed, as illustrated
were only
slightly
in
higher at the
than at the beginning of the 2001 recession." However, investments
in
structured products exposed
in
the past.
financial institutions to
more aggregate
risk
and surprises than
'iME/BO/ZOOlj
Ike/bopootI
us primary dealers
See,
e.g.,
US bank hold ng companiel
(Geanakoplos and Polemacharkis, 1986; Caballero and Krishnamurtiiy (2001, 2006), and
Lorenzoni, 2008).
The use of off-balance sheet vehicles with either
sponsoring institutions means that true economic leverage
figure. During the crisis, leverage increased as losses
credit-line
may
vi/ell
or
reputational
dependence on the
have been higher than suggested by the
decreased the denominator.
16
Figure 5.Aggregate leverage ratios for commercial banks and broker/dealers. Source; Adrian
and Shin
(2009)
In
the current U.S.
banks were holding mostly senior tranches of
crisis,
ABS. According to an estimate by Lehman Brothers, as of April 2008,
securities
accounted for 85% of assets held
in
AAA
opposed
securitized (as
Giving rating agencies the benefit of the doubt, these
a large variety
may have been
or agency-backed
to whole-loan) form.
unconditionally
in
conditional on large aggregate shocks. They relied on protection by the junior tranches and
the law of large numbers
AAA
AAA
were not
the sense that the unconditional probability of default was very low. However, they
AAA
new
of
in
order to reduce the unconditional
The law of large numbers implies that losses on
rating.
risk of
takes place. Furthermore, the higher up the capital structure
tranches.
Losses
aggregate shocks, but
must be
enough
large
this
is
to
for
given security
a
what
AAA
tranche only occur
large surprises
do
(i.e.,
in
more than
misinterpreted, would suggest, and certainly
bonds. The latter are
still
factors play a larger role.
average
lost
downgrade
between
of
5
more systemic
affected by aggregate
and 6 notches. ^°
2001/2002 (30%
risk
of
situated, the
is
by the junior
states
of severe
AAA
tranches of
than their rating,
similarly rated "single
securities in the
when
name" corporate
macroeconomic conditions but
Downgraded structured finance
number
transform manageable
aggregate shocks into potentially devastating ones). Therefore holdings of
structured products exposed financial institutions to
to achieve
when an aggregate shock
to pierce the protection offered
it
affect the
exactly
enough
a pool with a sufficient
underlying assets, as was the case with most ABS, can only occur
larger the aggregate shock
default
idiosyncratic
2007/2008 period on
comparison, during the great corporate bond
In
of corporate
bonds were downgraded
in
that period), the
average notch-loss was only 1.8 (Benmelech and Diugosz, 2009).
Coval
et. al.
(2008) argue that the correlation between economic catastrophe and default by
highly rated structured products
ratings as a
went
sufficient statistic for
largely
pricing.
unappreciated by investors,
who seemed
to treat
Highly rated single-name CDSs and structured
product tranches traded at very similar spreads
(their data
is
for
September 2004
to
September
2007), despite the fact that on average the structured product tranche would likely default
much worse macroeconomic
in a
state.
Regardless of whether this correlation was underappreciated or not, the systemic consequence
of this risk
was that
highly leveraged and systematically important financial institutions
bearing more aggregate
risk
than would have been thought from simply observing the ratings
of their assets. Having the highly leveraged financial sector of the
respect to an aggregate surprise proved to be
In
most cases
this
were
downgrade implied
assets increased from 1.6% to either
4%
a
economy
holding the
risk
with
recipe for disaster.
that regulatory capital requirements for the banks that held the
or 8%.
17
II. C.
The
Behind-the-Ciirve Policy Response
develop
third ingredient for a severe crisis to
a policy
is
response which does not recognize
the nature of the rules-of-the-game changing surprise and
aspect of
a policy
response to prevent severe crises
of the policy debate
much
is
of the action
almost always
a
is
behind the curve. ^"^ A
is
the reaction to the
is
divided betvi/een ex-ante and ex-post policies.
happens
window
the policy response. That
in
when
of time
is,
during the early stages of the
There are many reasons
between the
many
crisis
why
tw/o.
the
Once
initial
We
critical
surprise.
Most
argue, how/ever, that
a significant surprise
takes place, there
can either be contained or exacerbated by
crisis
of the ex-ante aspects of a severe financial crisis
happen
rather than years earlier (as with cardiac arrests).
policy responses can be too slow,
but there are
two main
retardants: moral hazard concerns and political constraints.
Invariably, policies of supporting the financial
whether such intervention
will
so
weaken
system come together with lengthy debates of
incentives as to essentially cause the next
problem with the standard moral hazard view
disregards the incentive problems
that
is
generates within
it
is
it
typically rudimentary,
crises. In real
life,
unlike
crisis.
A
that
it
in
many
in
of our
models, crises are not an instant but a time-period. This time dimension creates ample
opportunity for
when and
if
all
types of strategic decisions within
knowing that
to let go of their assets,
a
a crisis. Distressed
miscalculation on the right timing can be
very costly. Speculators and strategic players have to decide
spiral,
and when to
stabilize
it.
intervening. Each of these agents
Governments have
is
in
it is
crisis
when
decide
to
of predicting
and speculators within the
the resolution of the ongoing
A standard
game
and the form
particular, the likelihood of a bailout
for both distressed firms
the
agents have to decide
to reinforce a
how
likely to do. In
expected to take change the incentives
These incentives are
crisis.
camp
advice stemming from the moral hazard
central, both to
crisis.
is
to
subject shareholders to
exemplary punishment (the words used by Secretary Paulson during the
is
time dimension,
shareholders were part of the
all
the bailout takes place the
to the
is
crisis.
crisis
is
boom
that preceded the
over; the next concern
Punishing shareholders
Bear Stearns
the absence of a time dimension within crises. With no
intervention). This
in
before
long to wait
what others are
and to the severity of the nexf
sound advice
downward
is
means punishing those
crisis
and
as
soon as
not to repeat the excesses that led
that led to the current
crisis,
and
it
better that they learn the lesson sooner rather than later, the righteous speech goes.
Of course the role of
this factor, as a
necessary condition, depends on the resources available to the
government. For example, emerging marl<ets embroiled
in
crises often
experience large capital flow reversals that
severely limit the government's options.
18
However,
this
advice can backfire w/hen
we add back
that shareholders will be exemplarily punished
inject
much needed
eager to
Some
policy
As
into
the
concrete example, sovereign wealth funds were
a
the
U.S.
financial
much
to
less
system after the Bear Stearns exemplary
(March 2008) than they were before the
in
Morgan Stanley
policy, as illustrated in Figure 6.
October 2008, only took place after the
in
investment would not be diluted
that the
Conversely, destabilizing speculators and shortsellers
by the policy of exemplary punishment.
memories
new
worsens delays investors' decision
crisis
of the capital injections that did take place after this, such as UFJ Mitsubishi's $9 billion
investment
them
equity
inject
punishment
capital.
if
the time dimension. Now, the expectation
of this intervention
equity
be
will
may
crisis
is
Treasury assured
government intervention.^^
saw the value
of their strategy reinforced
Moreover, from the point of view of future
hamper any chance
attempt to arbitrage the
less likely to
the within-crisis time dimension
current and future
also
'^^
future
a
in
U.S.
crises,
of a private sector resolution as
other words, once
initial fire sales. In
considered, the anti-moral hazard strategy
may morph
into a
enzyme.
20
i^H
15
^^^^
^^^_
i
II
5
111K0VHD7
Figure
5: Injections
Dec-07
Jan-OS
Feb-OB
Mar-OS
-r^
Mav-OB
Apf-OS
Jun-08
Jul-OB
of capital into U.S. financial system by Sovereign Wealth Funds. Excludes
injections to prevent dilution. Source:
RGE
Global Monitor.
http://dealbook.blogs.nytimes.com/2008/10/13/treasury-said-to-offer-mitsubishi-protection-on-morgandeal/
On June
selling,
warned
8th, 2009, Senator
that;
"...
concentrated downturn take
that opportunity to do
it
Kaufman, arguing the case
for
imposing new constraints on nal<ed short-
there are legions of hedge funds with capital ready to take action should another
place....
again
in
If
someone has made
the future."
a lot of
money
in
a particular endeavor, he will take
http://www.marketwatch.com/storv/senator5-push-sec-to-reign-in-
naked-short-selling
19
However, the biggest policy miscalculation during the current
moral
concerns
hazard
consequences of
letting
the current U.S. financial
one
as
Lehman
crisis
of
political
collapse
and
was
marked
it
was not so much
crisis
The
constraints.
anticipate
to
failure
the
management
order mistake during the
a first
a result of
of
a clear turning point for the worse.
Former Secretary Paulson has stated on multiple occasions that the main reason to not provide
support to Lehman was the lack of
why
legal authority to
did he not seek such authority, but
that question
is
bail
"Speaker of the House Nancy Pelosi
would have
of course
is
made
it
difficult
to obtain
out the firm, especially with the speed that would have
final
demise of Lehman, the WSJ reported
(D., Calif.) said
Thursday that Lehman's impact on the
been required. For instance, only days before the
credit markets
The obvious question
that he faced a political climate that would have
Congressional authorization to
that:
so.
not unreasonable to conjecture that the answer to
is
it
do
government moved to
to be evaluated before the federal
together a rescue package for the troubled investment bank."
^"^
pull
The noncommittal nature
of
her statement contrasts sharply with the urgency of the situation, which led to the bankruptcy
of
Lehman only
a
few days
later.
Governments may be
can be counterproductive, but the
tempo may
political
fully
aware that delaying during
require that a
full
blown
crisis
become
observable for bickering to be put aside. Former Assistant Secretary for Economic Policy
Swagel put
"[...]
massive intervention
were on the verge of
On the other
in
financial
markets could only be proposed
collapse. By then
it
if
in
system and economy
could well be too late."
LTCM
crisis,
containing what could have been a major panic trigger
markets."
in
played
a
very
U.S. financial
^
:
Setting the Stage for the Policy Discussion
U.D.
When
thinking about the future, there are
after a crisis,
behave
as
if
and the current one
is
two types
of policy-paradigms that typically
no exception. The
first,
and most dangerous,
the probabilistic statements of the early warning systems are
they really are, and come up with a long
list
of macro-policy
The argument of "moral hazard"
when
arises in
most cases
is
emerge
simply to
more accurate than
recommendations and prudential
http://online.wsi. com/article/SB122116292232524671.html?mod=hpp us
for failing to intervene
whats news
of successful interventions (and
is
used as an excuse
needed). Both the successful Mexican intervention during the mid 1990s and the
intervention during the late 1990s have been blamed for subsequent crises. However, "bailed out" investors
often experience
will
Secretary Paulson and
financial
side of the spectrum, the fast intervention during the
significant role
LTCM
Phillip
succinctly (Swagel 2009):
it
Chairman Bernanke went up to Congress and told them that the
^^
a crisis
enormous
losses in these scenarios.
It
seems farfetched
change future actions very much (beyond the impact of the
to argue that adding a
bit
to these losses
crisis itself).
20
regulations which, as
we argued
earlier,
can be both an excessive straightjacket and give the
system an unwarranted sense of security.
The second type
of policy response
simply forbid leverage,
which case
in
acknowledge the extreme unpredictability of crises and
to
is
doesn't
it
much matter what
financial multipliers are kept at bay. For example, Kotlikoff and
converting
since the
(2009) have proposed
banks into mutual funds, implying that the only way to issue something
all
equivalent to
may be
the shock
Goodman
demand
deposits would be with
100%
reserves.
Any other intermediation
activity
would be financed 100% by equity investors rather than depositors, an extreme form of narrow
A
banking.
serious problem with
between micro- and
even more
so)
recommendations of
management and
the
less
redistribution
capital than
requirement on aggregate
wasteful.
type
is
that they do not distinguish
macro-risk. The core of business for financial institutions
however, requires much
capital
this
risk
microeconomic
of
does managing aggregate
management
is
(and should be
risk.
This
activity,
Therefore, basing
risk.
be enormously
considerations could
^^
We
argue that
the
first
a
more
effective response
is
to
acknowledge our
inability to predict crises (unlike
response) and to design policies that have uncertainty spikes as a central concern,
without throwing away the baby with the bath water (unlike the second response). This
starting point of the
is
the
second part of the paper, which contains our policy proposal.
Economic Policy Proposal: Tradable insurance Credits (TICs)
Hi.
Given the three ingredients
we have
highlighted -surprises, concentration of surprise-risk on
the leveraged financial sector of the economy, and
deal with severe crises while they are
still
a political
manageable
-
tempo which
is
not fast enough to
the policy must be flexible to react
quickly to a truly unexpected event and focus on minimizing the impact of this surprise on the
highly leveraged sector of the
Note that what we seek
by
its
is
economy.
not to deal with the truly unexpected
nature can only be dealt with ex-post. Instead our goal
most devastating systemic consequences
flight to
emerge
quality.
in
is
itself,
for this
to address
is
something that
one of the main and
of such events: the widespread panic and associated
As Knightian uncertainty takes over and agents fear that exposures
unexpected places, prices of insurance spike, ravaging credit and
may
financial markets.
Of course under the conditions of the Modigliani-Miller theorem, there would be no cost to increasing
capital requirements, but this
is
not
a realistic
assumption for financial
instituions.
See Kashyap
et. al.
(2008) for a
discussion of the economic costs of bank capital based on agency theory.
21
Our
policy proposal begins with the simple observation that any debt involves a pure interest
component and
a credit risk
component. These two components are conceptually separate
and, either implicitly or explicitly, there
the promises on
a
debt contract
(a
is
bond)
a price for
is
one
each of them.
If
and the bond
dollar
the discounted value of
worth
is
B,
then
we must
have
P=
B +
where P
is
for credit default swaps,
their separate prices. Figure 7
shows the evolution
Insurance prices increased sharply
in
(1)
the price of insurance on the corresponding credit
Thanks to the growing market
again
1
in
increasingly simple to keep track of
it is
of
risl<.
CDS spreads
March 2008 around the time
September 2008 around the
fall
of
for selected industries.
of the
Lehman Brothers and
fall
AIG.
of Bear Sterns,
and
The increase was
greatest for directly-affected financial services and insurance companies but extended to
all
sectors of the economy.
Jan07
Figure
This point
was not
Mar07
7.
lost
Mav07
JulO?
Sep07
NavD7
Jan 08
UarOS
May08
JulOS
AugOS
DecOe
OctOa
Mar09
May 09
Industry CDS iodices for selected industries. Soui'ce: jPMorgan.
on central bankers during the current
crisis
where many saw the need
provide public insurance to substitute for the disappearing private insurance.
instance, offered banks an Asset Protection
Scheme.
In
exchange
for a fee, the
offered banks insurance against credit losses on assets affected by the
In his
described
some
^^
crisis,
The
to
U.K., for
UK treasury
after a first-loss
January 20, 2009 speech at the CBI Dinner, the Governor of the Bank of England, Mervyn King,
of these policies as "unconventional unconventional measures"...
22
borne by the banks. The
of
America
the entire
November 2008 and Bank
in
January 2009. The Public-Private Investment Program
in
elements as
similar
reached similar deals with Citigroup
U.S.
Banks can
well.
(instead of just the
risk
the U.S. contains
some
assets to newly-created investment funds, offloading
extreme downside, as
UK
the
in
Funds and the FDIC insures their
half of the equity for the
First losses (relative to
sell
in
case).
liabilities
The Treasury provides
(there
a
is
cap on leverage).
the purchase price, not the original contractual terms) are borne equally
by the government and private equity co-investors, and any subsequent losses are borne by the
FDIC.
These
initiatives
contributed to an easing of tensions
in financial
markets. The CDS spread on
the banking sector dropped by about 50 basis points during the
Citibank
was announced.
around $50
billion,
20%
of the
banks increased by $220
of course
it
may
due to the
other news
in
would seem
justified
probabilities)
and
greatly exceed
it
of the guarantee
90%
amount
fair actuarial
a
in
of the
and the market value of the main 20
of the guarantee.
Some
some
some may
the perceived probability of future bailouts, while
is
of
just
market that fluctuated widely. However, the effects are greater than
by the expected net transfer (using non-panic distorted catastrophe
seems
likely
that the reassurance they brought had multiplier effects that
expected cost.
its
tool during panics. First, the PPIP has
after a sharp recovery
it
of this increase
value of the insurance provided by the government,
Moreover, several recent developments support the view that insurance
cheap
asset insurance for
market value (equity plus long-term debt) increased by
amount
billion,
an increase
reflect
reflect
Citigroup's
week
in
financial markets,
it is
been scaled down
considered
significantly,
urgent than
less
an effective and
is
it
mostly because
once was. Second,
turns out that the Treasury and Bank of America never formalized the asset insurance
contract. The
stability
tried to
it
mere prospect
needed
of such a contract helped provide
to ride that panic
Bank of America with the
wave. After the panic subsided. Bank of America reportedly
renege on the agreement, since losses seemed
likely to
be even less than the insurance
fee.
These approaches were needed given the depth of the
available at the time, but they
U.S.,
Treasury
officials
eligible for insurance,
guideline of
during the
how
crisis
missing the early
were necessarily
had to bargain with banks on
will
is
always
a
and the limited instruments
In
both the U.K. and the
case-by-case basis over what assets were
of the first-loss, etc.
danger that
political
Without
a clear
considerations emerging
delay or prevent intervention, exacerbating uncertainty and possibly
window
of opportunity to contain the
Moreover, the effectiveness of intervention
example, the
a
what fee was payable, the amount
to intervene, there
crisis
reactive and ad-hoc.
U.K.'s Asset Protection
Scheme had
is
crisis.
'^^
Our proposal
is
to design a tool
also affected by these ex-post political considerations. For
less subscription
than seemed optimal at the time because
it
23
whereby
central banks can
manage the
effective availability of insurance at a systemic level,
analogous to the way they manage monetary
consequences of spikes
III.A.
in
premium on
risk-uncertainty
Under our proposal, the central bank would
crisis,
with the aim of preventing the devastating
financial institutions' balance sheets.
Essence
TlCs, the
systemic
policy,
each TIC would
issued and legacy securities.
entitle
All
issue tradable insurance credits (TICs). During a
holder to attach
its
bank guarantee to newly-
a central
regulated financial institutions would be allowed to hold and
use TICs, and possibly hedge funds, private equity funds, and corporations as well.
many
TICs could be used as a flexible and readily available substitute for
were created during the
crisis.
principle,
In
of the facilities that
The basic mechanism would consist of attaching them to assets,
but variants could include attaching them to
liabilities
particular needs of the distressed institutions
and markets, and they could also operate
collateral-enhancers for discount
point
starting
for
a
potentially
window borrowing. To
proposal,
useful
and even equity, depending on the
state the obvious,
than
rather
implementation issues need to be addressed before
as
a
we view
as
this as a
Many
complete one.
proposal can be transformed into
this
policy.
In
the current
crisis,
much
of the
uncertainty derived from the possibility of larger than
expected losses from loans and securities related to residential
real estate. Accordingly,
the ad-
hoc asset insurance programs that were undertaken concentrated on those kinds of assets. TICs
would have provided
a
systematic
way
same
of doing the
thing, by simply declaring
convertible into insurance for that type of security.
,
,,,,
.
.
them
Let us start from an extremely stylized formulation which illustrates the core idea. For this,
assume
that there
prevalent
P, rises
only one risky bond and that 9 captures the degree of systemic "fear"
the economy.
In
the absence of
a TIC-policy,
the CDS price associated to this bond,
as 9 increases, and, correspondingly, the price of the bond, B,
equation
extreme
in
is
1).
The concern of the Central Bank (CB
rises in 9
in
what
follows)
is
falls
one
for
one (see
to limit the impact of
on banks' balance sheets.
The TIC-policy has the following three key components:
A
convertibility rule.
The CB determines
attached to the bond. An attached J\C
similar to the
way
is
a
threshold level for
9,
above which TICs can be
simply a CB-backed CDS that makes a bond whole,
the public insurance wrapping
perfect precision and hence the reaction of the
GNMAs
CB
required very high insurance fees, probably set to optimize
is
do.
In
practice, 9
a probabilistic rather
political
is
not observed with
than
appeal rather than financial
a deterministic
stability. (Or,
more
precisely, to optimize financial stability subject to a tight political constraint.)
24
function. But the point that
convertibility
important to us
is
Open market operations. During normal
or
TICs at market price Q. This price
sell
is
that the probability of declaring the TICs
increasing with respect to the degree of panic
is
is
when
times,
necessarily
since the value of a TIC derives entirely from
in
the system.
TICs are not convertible, the
below the
CB can buy
option to be converted into a CDS
its
CDS
price of the corresponding
in
the near
future.
Minimum
During normal times, highly leveraged and systemically
holding requirements.
important institutions must preserve a
weighted
and
assets
The TIC-policy can achieve
limiting the
of TICs as a proportion
analogous
importance,
systemic
conventional capital-adequacy
minimum amount
consequences of
rising 9
We
on banks.
will
all
the effects
we
on the economy, but
as well as by
unable to influence
is
directly.
below derive from changing the impact of uncertainty spikes
highlight
it
itself,
focus on the conservative scenario
where, despite having an effective policy framework, the CB
Thus,
with,
ratios.
balance-sheet stabilization goal by reducing 6
its
risk-
complementary
and
to,
of
goes without saying that the indirect benefits from reducing 6 only
reinforce the virtues of the TiC-policy framework.
How does
the TIC-policy work? Let
example
composed
is
of
one
with market value Q, and
the value of a bank's assets, which for this
bond with value
risky
some
A denote
cash
C.
B, a
TIC that can be attached to that
Thus:
A^B-i-Q-^C
Using equation
to solve out B
(1)
from
A
In
=
(2)
this expression,
l-P
+
Q+
we
can write the assets of the bank as:
C
(3)
the absence of a TIC-policy, P rises with an increase
systemic fear
in
one. These patterns are illustrated by the thin-dotted lines
TIC-poiicy improves over this
is
is
the addition of
sufficiently
Q to
high.
compounding each
that
it
may
the balance sheet, which
other:
Q
in
for
rises in
this
is
9,
and A
one
falls
the two panels of Figure
outcome through two channels: The
The reason
bond
first
for
The
8.
and most direct channel
sharply increasing with respect to 9 once fear
"convex"
pattern
response to the increase
is
in
that there
are
two forces
the price of the underlying
CDS
eventually be converted to, and because the probability of this conversion rises with
systemic fear.^? The second channel
Note that
the whole point
is
in
is
that the policy reduces the sensitivity of the
the comparison the bank with the TIC
will
have
that these TICs are purchased during normal times,
less c since
when
6,
it
CDS
price to
used cash to buy the
TIC,
but
and hence Q, are very low.
25
increases
in 9.
The reason
converted into CDS
for this effect
is
and hence the
rises,
that as fear rises, the probability of having the TiCs
lil<elihood of a significant
supply of CDS also increases. The strength of
the CDS market. The solid lines
this
expansion
in
the effective
channel depends on the degree of
illiquidity of
Figure 8 illustrate the TIC-w/orld.
in
^TlCless
Figure
8. Effect
Consider now/ a TIC-OMO
cash at
a point in w/hich
systemic
result of such an operation
The
first
sheet
is
and most direct
is
is
risk
an increase
effect of this portfolio shift
the bank's assets to further rises
The reason
is
in
in
exchange
relatively low/ (point Go in Figure 9).
two
for
The
effects of this policy:
banks' protection since the representative balance
is
+
at 6o
and n
is
the
number
(l+n)Q+(C-nQo)
to rotate the
in
A
of
new
TICs
if
curve
(4)
in
the figure, reducing the exposure of
systemic fear.
is
a
market effect by which
that the expansion
expansion of the supply of CDS
work
still
in
:
The second effect of the TIC-OMO
TICs.
but
add TICs to the system
denotes the value
nov^ (w/here the subindex
A= l-P
and of
rising
is
to
illustrated in the figure. Again, there are
sold to the representative bank)
The
CB decides
w/hich the
in
bank asset values and TIC prkes
of systemic fear on
in
it
reduces the price of the CDS
the supply of TICs raises the expected
TICs become convertible. The change
the opposite direction on the value of the bank's assets, but
it
is
in
these two prices
easy to see that, on
26
net, these
changes are positive for the bank since
direct effect of the policy on P
The
total effect of the
must be
TIC-OMO
is
Q
is
larger than that
just an option value
on
its
illustrated by the thick lines in Figure 9.
the contemporaneous balance sheet of the banks and,
ensures
system
confidence.
the
becomes more
to
further
It
is
apparent that
more importantly,
deterioration
in
a
it
systemic
^°
Figure
Ill.B.
resilient
and the
P,
option.
TIC-OMO improves
that
on future
Open markel, operations with TICs
9.
Heterogeneity and Adverse Selection
The above example
is
concerned
heterogeneous assets and the
w/ith
A
as the Fed currently
follows from the discussion above that the
If
done
at
does
in its
AAA-rated bonds, 1.7 TICs per
9,
the policy has
little
effect
has the benefit of improving the resilience of the banks' portfolios to systemic spikes
(since the probability of convertibility
significant
amount, so again
all
is
If
haircut tables for TALF
maximum contemporaneous power
low levels of
(inexpensive) time for banks to stock up on TICs.
many
must specify how/ (and w/hether) they can be
liquidity facilities. For instance: 1 TIC per dollar of
place at intermediate levels of fear.
practice there are
large part of this heterogeneity can be addressed w/ith
much
differential conversion factors, very
It
single bond, w/hereas in
rules of TICs
attached to different kinds of assets.
and other
a
done when 9
is
TIC-OMO
takes
on current prices but
it still
of a
in fear. In fact, this is
already very high, then a TIC
is
almost
the right
like a
CDS
close to one), and the policy cannot affect the spread (P-Q) by any
the benefit comes from the direct effect.
27
dollar of AA, etc.
The
ratings should be recent to avoid the use of TICs for already-defaulting
assets and, obviously, the ratings should be set on the basis of the TIC-less bond. Moreover,
these relative conversion factors could be gauged from the pre-crisis CDS prices. A similar
principle can be used to avoid distorting the allocation of TICs
(w/here, other things being equal, they w/ould
towards long-duration assets
be more valuable). The conversion rule could
require x TICs per dollar of assets of a given rating per year of duration.
Still,
it
is
duration;
quite likely that there
it is
to be expected that TIC holders will attempt to
know
attaching TICs to assets that they
mitigate this
some degree
The
first is
may
of adverse selection
the most of the insurance by
to be the worst within a given rating.
likely to
is
In principle, this
in
One way
TIC contracts.
to
Still,
remain, which could bring two separate problems.
who might end up
simply a question of costs for the taxpayer,
in this
assets within a given rating and
make
be by including a Representations and Warranties clause
selected pool of assets.
engage
among
be heterogeneity
w/ill
need not be
a
insuring an adversely
great concern, since the opportunity to
kind of selection will be anticipated by TIC buyers and will be incorporated into
when
the price that TICs fetch
The second concern
is
financial institutions to
,-—
the Central Bank issues them.
potentially
more
serious.
remove uncertainty from
The objective of TICs
is
to provide a
their balance sheets at times
when
way
for
the private
sector makes uncertainty-insurance too expensive. For this to work, TIC holders must attach
TICs to assets that are greatly affected by market uncertainty.
TICs for mis-rated assets
is
If
gaming the system by using
too attractive, this might draw TICs away from the uncertainty-
affected assets where, from a social point of view, they would be most needed. This matters
most
if
mis-rated and uncertainty-affected assets are very different. However, debt contracts,
which TICs would be used for most, are characterized by having more
on the upside (Dang et
is
also likely to be
IILC.
2009). This
al
means
one whose true value
is
that a bond
risk
on the downside than
whose holder deems
to be over-rated
uncertain and where a TIC would be well allocated.
Discontinuity and Smooth Interventions
The above analysis assumes that
convertibility of TICs
an exact pre-specified value of
9.
In
smoother. One way
is
that instead of declaring
threshold
is
to
do
this
practice,
it
is
a
binary decision, which takes place at
might be worthwhile to make the policy
reached, convertibility can be de.clared for just
all
x%
of the outstanding TICs.^^
the percentage can be gradually increased until
continues to increase during
a crisis,
100%. This would give the CB
a
way
implemented
in
several ways. For instance, TICs could have serial
This can be
of dynamically fine-tuning the policy and
could extend to just those with serial numbers between
have TICs with diversified
serial
when
TICs convertible
and
x (prudential
it
it
the
If
9
reaches
would moderate
numbers and
convertibility
regulations could require that holders
numbers).
28
the extremely high stakes and consequent political pressures that would accompany an all-or
nothing convertibility decision.
UI.D. T!Cs for Equity, Liabilities,
and Debt-Equity Swaps
Invariably, during crises the leveraged sector's capital shrinks rapidly
and the need arises to
quickly recapitalize financial institutions to prevent runs and sharp loan contractions.
policy for these
however,
institutions'
may be
it
case asset-insurance
This policy can be
institutions.
guarantees
In
a
case
this
would operate
TICs
few years hence. During times
new
TICs also could be used
CB
is
new
equity issued by financial
Treasury)
(or
minimum
share-price
of systemic crises, financial institutions could
largest
domestic financial
more narrowly
new
performed by the
institutions.
to facilitate debt-for-equity
swaps during
crises.
That
is,
equity issued to the debt holders that are willing to
the exchange.
other circumstances, TICs might be more usefully attached to
a panic
in
private capital raised; the (price) level of the guarantee could be
TICs could be attached to the
In
as
to
result of a stress test along the lines of those recently
government on the
in
problems
insufficient.^^
is
determined from the
participate
direct policy to deal with capital shortage
implemented by extending the TIC program
attach TICs to any
U.S.
more
TIC-
reduce the need for rushed recapitalizations,
assets should
useful to have a
The
liabilities
affecting the funding side of financial sector balance sheets (as
case during the run on
their liabilities
(at
money
rather than assets.
was
If
for instance the
markets), the Central Bank could allow banks to attach TICs to
some conversion
guaranteed by the Central Bank,
a
rate).
This
would
measure that was
effect allow
in
them
to issue
debt
also tried during the current crisis (with
FDIC guarantees).
///.£.
Fees
The simplest TIC setup involves no payments by the holder during the
value of insurance
(perhaps measured
is
in
paid at the time the TIC
is
issued.
A
variant
basis points), possibly different for attached
may
life
of the TIC;
all
the
involve periodic fees
and unattached
TICs.
These
fees could be set so that holders of TIC guaranteed securities be willing to un-attach the TICs
if
the panic subsides and private insurance prices return to normal. Allowing the possibility of unattaching TICs and providing incentives to un-attach TICs can be a
involvement does not
last
way
to ensure that public
more than necessary.
For an equity guarantee proposal during
tlie
current
crisis,
see (Caballero, 2009a; and Caballero and
Kurlat2009).
29
These and other variants provide many degrees of freedom
in
designing a TIC program; actual
implementation requires deciding w/hich of these are w/orth taking advantage
TlCs and the Discount
III.F.
financial
Treasuries.
which
would
collateral value for discount
window borrowing.
Eligible
choose to pledge TIC guaranteed securities rather than naked
institutions could
collateral,
Window
enhance the
TICs could also be used to
of.
be
subject
to
lower
comparable to that of
perhaps
haircuts,
33
During the current
the Collateral
crisis
Management System (CMS)
handle large volumes of collateral tracking and
and frequency of
more than $5
par value and
engaged
in
a collateral
Report by the Federal Reserve Bank of Philadelphia).
central role, perhaps both
in
CMS was
value of $2.5
two
to be able to provide real time valuations within the next
shown
tracking assets with
trillion.
CMS
The
years. (See the
would seem that the
It
to be able to
an effort to improve the precision
end of 2008, the
collateral valuation. By the
trillion in original
is
has
CMS
expects
2008 Annual
should play
a
determining the TIC-conversion factors for different securities and
the haircuts for TIC guaranteed assets.
Orders of Magnitude
III.G.
'
I
..
:
;.....
•'
.
lUe ad-hoc insurance programs that were undertaken during the
JPMorgan/Bear Stearns, Citigroup, and Bank
billion,
provide
some
indication
magnitude the TIC program should have. The government-provided insurance
of the order of
for
crisis
approximately
6%
of America/Merrill Lynch totaled
about $500
of the respective institutions' assets
Maximum
total
First loss
%
borne
Exposure of
remainder
by insured party
assets
Net
maximum
exposure
Maiden Lane (Bear
Stearns)
30
Maiden Lane
1!
20
Maiden Lane
III
(AIG)
(AIG)
Citigroup
Bank of America
Total
Table
If
this
1,
order or magnitude
macroeconomic
risks,
capital requirements.
.
is
1
29
20
30
5
100%
25
306
29
90%
249
118
10
90%
504
TIC-like guarantees
97
421
for .specific institution.
.$
biUion
roughly the right size to insure the financial system against extreme
more
or less on par with regulatory
financial
system should be of the order
required holdings of TICs should be
The
100%
100%
total supply of TICs for the
Geanakoplos (2009) argues that the FED should
during crises. Our integration of TICs with the discount
US
find a
mechanism
window could be
to control the haircuts on collateral
a first step in achieving that goal.
30
of $1-2
worth
trillion.
It
is
only take place
if
there
is
recalling the $1-2 trillion
a crisis in
a notional
is
which TlCs are converted and the underlying assets default.
Importantly, one of the distinctive features of spikes
agents
exaggerate
greatly
the
amount. Actual outlays would
the
of
likelihood
Knightian uncertainty
in
default
event.
mean
the
In
that private
is
time,
the
government would obtain revenues from the sales of TICs.
An amount higher than $1-2
may
discussed above, TICs
might be required
trillion
provide a ready-made instrument to undertake different types of
insurance-like interventions, which could
We
as of
June 2009 was of $3
believe that TICs
inevitably
illustration of
trillion
how
worth of
Fed might
would have
a TIC
TICs.
initially
crisis.
effective policy tool to address the current crisis.
of
counterfactual
speculation,
program would be used. Suppose
When
the
first
have merely hinted that
it
but
it
in
the
also
to
sell
serve
summer
was considering making TICs
raised the perceived probability of convertibility
were more vulnerable. The Fed might
may
as
an
banks had held approximately $2
U.S.
surprises hit financial markets
were sound could have taken the opportunity
If
support
in
The outstanding balance of these
of 2007, the
convertible. This
and thus raised the price of TICs.
TIC holders that were not against prudential constraints and who believed
more
programs
would have been used and how the economy would have reacted
TICs
some degree
involves
trillion
of the myriad
(SIGTARP, 2009).
would have been an
how
Predicting exactly
encompass many
were undertaken during the
of the financial system that
programs
wider uses for TICs are envisioned. As
if
their asset portfolios
TICs at a profit to banks that believed they
have engaged
in
open market operations
to provide
TICs to the market, increasing the (contingent) supply of insurance.
fear nevertheless began to take over the markets (for instance, as the
lender and monolines began to surface
in
late
on
AAA
It
mortgage
billion TICs,
declaring
them convertible
asset-backed securities."^ This probably would have required a more
detailed schedule of convertibility factors as
were created equal.
of
2007 and early 2008), the Fed could have
declared convertibility, perhaps of around $50 to $100
into insurance
weakness
could also extend to
it
was becoming
non-AAA
clear by then that not
securities, at
some
all
AAAs
higher conversion
factor.
Presumably, troubled firms such as Bear Stearns would have been hoarding TICs
in
anticipation
of this possibility
and could now attach them to the assets on their balance sheet, reducing
their vulnerability
and enabling them to obtain repo financing against the TIC guaranteed assets
on much better terms. Note that banks that were not so exposed to aggregate surprises would
Recall that the original
huge
at
the time but pales
in
Maiden Lane program
for
comparison to what happened
Bear Stearns involved $29
billion, a figure
that
seemed
later.
31
have been able to
realize a significant profit by selling TICs, a
reward for prudence that most ad-
hoc insurance schemes did not provide.
The Fed's next steps would have depended on the market's
to which the increased supply of insurance helped to ease Knightian fears
necessary, convertibility could have extended to
Merrill
Lynch/Bank of America would probably have been among the
the possibility of attaching them
How
far the Fed
Citibank or
like
to avail
If
themselves of
to their assets to dispel doubts about their solvency.
would have needed to go
what other conventional measures such
conceivable that not
first
the market.
in
the outstanding TICs. Banks
all
on the degree
reaction, especially
much more than
is
certainly a matter of speculation.
window
as discount
the $500
billion that
were
lending
was
in
fact
Depending on
also used,
committed
it
is
asset
in
insurance for specific institutions would have been needed.
Complementary Insurance Proposals
in.H.
Private markets
and indexing
to
observable variables.
In
some
instances, there are elements of
the surprise that are predictable. For example, commodity producing economies are
to experience unexpected financial events
Also, large capital flow reversals to
with spikes
in
the VIX.
In
through the private sector.
Kashyap
et. al.
when
their
likely
terms of trade have declined sharply.
emerging markets more broadly, are negatively correlated
such cases,
may be
it
feasible to
•-•:
^^
more
implement insurance contracts
v-
•.
;
(2008) have proposed a policy for dealing with negative shocks that
is
based on
such private-sector insurance. Banks would be required to buy insurance policies that pay out
the event of
spirit
in
a
that
negative shock to the banking sector.
it
^^
A TIC framework
emphasizes insurance rather than higher
however, some key differences.
First,
TICs
involve
somewhat
similar
in
requirements. There are,
capital
one more
is
in
layer
of state-contingency
("insurance-squared"): Crises involve not just realized losses that necessitate increased capital
but also (due to Knightian uncertainty) an exaggerated fear of potential future losses, which
is
dealt with less expensively through insurance. Second, TICs do not require the private sector to
freeze capital to back up insurance promises, which would require huge amounts of resources.
The monolines and AIG were
partially in the business of providing
such insurance and they
proved undercapitalized to withstand extreme events. Third, TICs do not require an ex-ante
definition of
what the extreme event
incomplete and might prove useless
if
is.
Contracts between private parties are necessarily
the negative shock materializes
in
a
way
that
covered by the terms of the contract. The government can afford to be vaguer
For private markets based proposals to insure
macroeconomic
Caballero 2003; Caballero and Panageas 2007 and 2008; and Borensztein and
Gersbach (2008, 2009) analyzes the theoretical properties of
risk
in
Mauro
is
in
emerging markets, see,
not
its
e.g.,
(2004).
this sort of insurance.
32
what kinds
announcement
of
Hoimstrom and
Tirole (1998).
of adverse shocks
it
would respond
to, a
point also highlighted by
'.
'
,
,
Government insurance contracts and guarantees. There have been multiple proposals of
kind that vary on the type of asset or
several of
implemented
manifested
of America
approach
is
were offered guarantees on
more broadly (although the
a similar plan
them
socially optimal for
few
useful to deal with a
institutions but
the TICs system for the system as a whole, with
political frictions
policy,
that
we have seen
crisis
basis. ^^ In
participating
to be part of the arrangement). This
is
new
more cumbersome
entrants, etc., and
to
is
implement than
subject to strong
and backlash. Presumably, the TIC-policy could be used as the main systemic
which could be supplemented by customized insurance for specific new circumstances
may
arise.
'
the expansion on the
is
•
a policy
eligibility of
of the discount
is
window
it
that triggers the expansion
,
_
policy to the TIC-policy but not a substitute for
U.S. or U.K. that
window access
private markets), but
risk
access.
economies such as the
discount
was
that ought to be preserved on a contingent basis. Similarly to the TIC-policy, the
complementary
a
crisis
both institutions and assets to access the discount window.
Fed could determine states of the world considered of systemic
This
U.S.,
the latter case
political constraint in
Expanded discount window. A mechanism that was very important during the current
This
the
share of their assets, and the U.K
a
premium which discouraged many banks from
very high
itself in a
would have been
it
guaranteed, and during the
them implemented mostly on an ad-hoc rather than systemic
Citibank and Bank
when
liability
this
deals with
especially
it,
have large "shadow" financial systems. ^^ Moreover,
liability
problems
(it
is
a substitute for
new borrowing
does not help directly with perceived asset, and hence
in
capital, losses in
the short run. These losses trigger problems beyond lack of access to debt markets, which
what the discount window helps
in
is
to alleviate.
A dual currency economy. Brunnermeir
et
al
(2009) are currently working on
which the Fed controls two units of accounts, one for regular currency
debt. During normal times, the exchange rate
devalued during
while the debt-unit
is
burden of leveraged
institutions. This
is
between these two
crises, diluting
a
(dollars)
proposal
in
and one for
units of accounts
is
one,
debt-holders and lightening the debt
an interesting proposition but
holders decide to run against debt anticipating a devaluation.
Still, it
See, e.g., Caballero (2009a,b); Caballero and Kurlat (2009); Mehrling
it
could backfire
if
debt-
probably does make sense
and Milne (2008) and Milne (2009)
for proposals.
On
July 14, 2009, Trichet
argued that the main reason why the ECB had not expanded
purchase new and legacy securities nearly as much as the Fed and BoE,
centric that the U.S.
is
that the Euro zone
is
its facilities
to
much more bank
and the U.K.
33
to
add
second type of debt
this
system as
into the
mechanism
a
to substitute for out-of-
banl<ruptcy debt-for-equity swaps.
Overall,
and aside from the expanded-discount-window
policy over the
many
alternatives
and capital-requirements.
IV.
Final
operational similarity w/ith conventional monetary policy
up to practitioners to push
this
analogy even further.
Remarks
While there are many
shock
original
estate bubble or trouble
may come from
partially anticipated factors,
have seen
is
a
such as the burst of a real
rapidly growing derivatives market, but the real crisis arises
in a
produces an outcome that
often
is
at least temporarily, the perceived rules of the
the filtering of such shocks through the complex financial (and
economic agents and
problem
similarities across financial crises, the core of the
which suddenly changes,
significant surprise,
game. The
It is
is its
advantage of the TIC-
policy, a clear
highly unexpected.
All
of a sudden,
understand their
financial institutions to
political,
it
local
and
social)
when
network
no longer enough for
is
environment
since, as
we
the current crises, systemic events can seep through unexpected and distant
in
linkages.
When
that happens, the fundamental shock
is
compounded many times
human
turns risk into uncertainty, and the natural response of
institutions in particular,
into
many
is
the
trillions of
The main antidote
The
silver
lining
withdraw
way
few hundred
a
beings, and leveraged financial
into safe assets. This panic triggers asset fire sales
multipliers that cause
activates financial
markets. This
to
is
by panic. The surprise
enormous damage
to
balance sheets and credit
subprime losses
billions of real
and
in
the U.S. translated
output and wealth losses around the world.
to fear
of this
is
prime, government backed, insurance against what investors fear.
diagnosis
is
that providing such insurance
government, as once panic subsides the
real losses are
much
is
inexpensive for the
smaller than those
initially
feared
by investors.
The TIC-policy we propose
is
an "insurance-squared" policy: For
variety of financial institutions to issue
assets and
liabilities
when systemic
Note that the government does not
gives the right to a
their
panic destroys the value of otherwise worthy securities.
inject resources during the crisis, as
exaggerated (by panic) extreme outcomes.
more
it
government-backed insurance to protect some of
purchase or capital injection program, but rather,
costly but also
a fee,
If
it
it
would with an asset
only provides insurance against vastly
correctly designed, this insurance
efficient than capital injections in breaking the
is
not only less
downward feedback
loop
34
between the
losses
and the
financial
real sector that typically
due to macroeconomic shocks
As with our cardiac-arrest analogy
develops
(as the capital injection
when banks
are
to absorb
left
approach does).
the introduction, during severe financial crises
in
it
is
important to intervene early on. The TlC-policy has the virtue of offering a very
critically
expedient and flexible policy-tool to the central bank.
damaging
policy, but directly targeted at offsetting the
sheets and credit markets.
Having said
this,
antidote, as
it
the analog of conventional monetary
is
It
on balance
effect of uncertainty-spikes
____^
the benefit of a TIC-framework extends beyond the pure uncertainty-spike
provides an expedient channel to inject resources into a financial system
in
from other sources such as conventional runs or even fundamental-based
distress originating
problems. Although
in
the latter case the decision
may belong
to the Treasury rather
than to
the Fed.
One the longstanding debates regarding
must be
in
charge of deciding
the conduct of monetary policy, which has been
who
and what degree of discretion they should have. Tradeoffs
it
possible time-inconsistency) of the political process, democratic
involve the vagaries (and
control over policy decisions, policy stability and the flexibility to react to circumstances.
Modern arrangements
in
advanced economies have favored independent Central Banks, with
discretion over day-to-day decisions but
The current
crisis
has
shown
bound by
A necessary
far
is
embedded
the TIC-program
in
balance
settled, a fact that contributed to costly
carrying
level of
central
in
management and decision-making
is
policy, the institutional
it
out, especially given that
would be
it
committing public resources. This mandate should be subject to proper
debate ex-ante and some agreed-upon
But this
from
monetary
condition of a successful TIC program must be the clarification of the
boundaries of the Central Bank's authority
explicitly
reasonably clear mandate and principles.
that, unlike the case of
of authority to deal with financial crises
policy delays.
a
not a requirement, and
is
in
very
independence and discretion ex-post.
One
much what most
many
in
have
possibility
its
central banks already have in-house.
to
overcome
example, the Federal Reserves Act
limits the
instances
in
infrastructure created during the current crisis to
programs that were so central
We
banks because the infrastructure required for
specific institutional constraints. In the U.S., for
Fed's ability to "take risks."
political
it
will
this
depend on being able
case
is
to build
on the multiagency
implement the insurance and guarantee
preventing a repeat of the Great Depression of the 1930s. For
instance, the Central Bank might be given authority to declare TlCs convertible with respect to
certain types of securities or up to a certain
may
maximum amount;
conversions beyond that
limit
require notification and/or authorization from an oversight committee. Given that TICs are
designed to address surprising and confusing situations,
Central Bank to be
bound by very
it
is
unlikely that
one would want the
tight rules.
35
Finally,
it is
important to highlight that the TiC-policy
any insurance arrangement, supervision becomes
since TICs
would be
all
relatively inexpensive during
more protection per
unit of systemic risk that
requirements. Moreover, TICs could provide
requirements more sensitive to
not a substitute for supervision. As with
the
more important. But
normal times,
it
also true that
a regulator could require far
could ever require with expensive capital
it
a useful
cyclical factors.
is
complement
for proposals to
The weighting of assets
make
capital
for conventional capital
adequacy and TIC-adequacy should be
different, with macro-sensitive assets requiring
TICs (since the latter isolate systemic
risks)
and
less
more
more
macro-sensitive assets requiring
capital.
V.
References
Adrian, Tobias and
Hyun Song
Shin. 2009.
"Money,
Liquidity,
and Monetary
Policy."
American
Economic Review Papers and Proceedings, forthcoming.
Bernheim,
B.
Douglas and Rangel, Antonio. 2005. "Behavioral Public Economics: Welfare and
Policy Analysis with
Berg,
Non-Standard Decision-Makers," NBER Worl<ing Paper No. W11518.
Andrew, Eduardo Borensztein, and Catherine
Systems:
How Have They Worked
Boberski, David. 2009.
CDS
in
Practice?"
Pattillo.
2004. "Assessing Early Warning
IMF Working Paper 04/52.
Delivery Options. Bloomberg Press, NY.
-
.
;•
Borensztein, Eduardo and Mauro, Paolo, 2004. "The Case for GDP-lndexed Bonds." Economic
Policy, Vol. 19, No. 38, pp.
Caballero, Ricardo
J.,
165-216, April
2009a.
"
Dow
Boost and
a (Nearly) Private
Sector Solution to the
Crisis ."
VOX, February 22.
Caballero, Ricardo
J.,
2009b.
"
A Global Perspective on the Great
Financial
Insurance Run:
Causes, Consequences, and Solutions ." MIT mimeo, January 20.
Caballero, Ricardo
Ttie
J.,
Role of Money:
2006.
"
On
the Macroeconomics of Asset Shortages ."
Money and Monetary
i
In
i'
'
>
Policy in the Twenty-First Century
36
The Fourth European Central Banking Conference 9-10 Novembe, Andreas Beyer and Lucrezia
Reichlin, editors.
Pages 272-283.
Caballero, Ricardo
2003.
J.,
Proceeding, 93(2), 31-38,
Caballero, Ricardo
Commodity
"
The Future of the IMF ." American Economic Review, Papers and
IVlay.
•
Emmanuel
J.,
Farhi
and Pierre-Olivier Gourinchas, 2008a.
and Global Imbalances ." Brool<ings Papers on Economic
Prices,
.
.
"
Financial Crash,
Activity, Fall,
pp
1-
55.
Caballero, Ricardo
Model
Emmanuel
J.,
of "Global Imbalances"
Farhi
and Pierre-Olivier Gourinchas, 2008b.
and Low Interest Rates
"
An Equilibrium
American Economic Review, 98:1, pgs
."
358-393.
Ricardo
Caballero,
Fragility, "
"
and Arvind Krishnamurthy, 2009.
J.
Global
Imbalances and
Financial
American Economic Review, May.
Caballero, Ricardo
and Arvind Krishnamurthy, 2008.
J.
to Quality Episode ." Journal of Finance, Vol. 63, Issue
Caballero, Ricardo
5,
"
Collective Risk
in
a Flight
October.
"
and Arvind Krishnamurthy, 2006.
J.
Management
Bubbles and Capital Flow
Volatility:
Causes and Risk Management ." Journal Monetary of Economics, 53(1), pp. 35-53, January.
Caballero, Ricardo
Constraints
548,
in a
and Arvind Krishnamurthy, 2001.
J.
Model
of Emerging
Market
Crises
."
"
International and Domestic Collateral
Journal of Monetary Economics 48(3), 513-
December
Caballero, Ricardo
J.
and Pablo
Kurlat, 2009.
and Pablo
Kurlat, 2008.
"
Public-Private Partnerships for Liquidity Provison ."
MIT mimeo, March.
Caballero, Ricardo
a Literature
J.
"
Flight to Quality
and
Bailouts: Policy
Remarks and
Review ." MIT mimeo, October.
Caballero, Ricardo
and Stavros Panageas, 2008.
J.
"
Hedging Sudden Stops and Precautionary
Contractions ." Journal of Development Economics, 85 pp 28-57.
Caballero, Ricardo
and External
J.
and Stavros Panageas, 2007.
Liquidity
Caballero, Ricardo
J.
"
A Global
Equilibrium
Model of Sudden Stops
Management." MIT mimeo, September.
and Alp Simsek, 2009a.
"
Complexity and Financial Panics ." MIT mimeo,
June.
37
Caballero, Ricardo
J.
and Alp Simsek, 2009b. "Fire Sales
in a
Model
of Complexity ."
MIT mimeo,
July.
Caplin,
Andrew and John
Wisdom
Leahy, 2004. "Business as Usual, Market Crashes and
after
the Fact," American. Economic Review, pp. 548-565.
Caprio, Gerard and Patrick
Coval, Joshua D., Jakub
Honohan. 2008. "Banking
W. Jurek and
Erik Stafford.
Crises."
Discussion Paper No. 242.
///5
2008. "Economic Catastrophe Bonds."
HSS
Finance Worl<ing Paper No. 07-102.
Barry
Eichengreen,
and
Presumptions and Evidence",
in
Mario
Emerging Markets. Cambridge: MIT
John
Geanakoplos,
and
Heraklis
I.
"'Banking
2002.
Arteta.
Carlos
Blejer
and Marko Skreb,
Heller, R.
M.
Polemarchakis.
Honor of K.
Gilovich,
1986.
When
Arrow, Vol.
Ill,
Thomas, Dale W.
Cambridge University
Griffin
Psychology of Intuitive Judgment.
Emerging
Markets:
eds., Financial Policies in
"Existence,
Regularity,
the Asset Market
Starr and D. Starret (eds.) Uncertainty, Information,
J.
in
Press.
Constrained Suboptimality of Competitive Allocations
W.
Crises
and
Incomplete,"
in
and Communication: Essays
in
Is
Press.
and Daniel Kahneman, eds. 2002. Heuristics and Biases: The
New
Geithner, Timothy, and Lawrence
York:
Cambridge University
Summers. 2009. "A New
Press.
Financial Foundation."
Washington
Posf, June 15
Gersbach, Hans. 2008. "Banking with Contingent Contracts, Macroeconomic
Crises,"
Risks,
CER-ETH Working Paper 08/93.
and Banking
.',-;•
Gersbach, Hans. 2009. "Private Insurance Against Systemic Crises," CEPR Discussion Paper No.
7342.
•'
'
:-'
•
..
V
.-.:.,,
,-:
-
Gorton, Gary and Andrew Metrick. 2009, "The Run on Repo and the Panic of 2007-2008," Yale
working paper.
Gorton, Gary and George Pennacchi. 1990. "Financial Intermediation and Liquidity Creation."
T/ieiourno/o/f/nonce, Vol. 45, No.
1 (Mar.),
pp. 49-71.
.
.
Hoimstrom, Bengt. 2008. "Comment on: 'The Panic of 2007/ by Gary Gorton,"
in
Maintaining
Stability in a
Changing Financial System, Symposium sponsored by The Federal Reserve of
Kansas City
Jackson Hole, Wyoming.
at
Investment Company
Institute.
2009. Report of the
Money Market Working Group.
38
Kashyap, Anil
in
K.,
Raghuram
Maintaining Stability
in
G. Rajan
and Jeremy
C. Stein.
2008. "Rethinking Capital Regulation,"
a Clianging Financial System, Symposium sponsored by The Federal
Reserve of Kansas City at Jackson Hole, Wyoming. Kaminsky, Graciela, Saul Lizondo and Carmen
M. Reinhart. 1998. "Leading Indicators of Currency
Monetary Fund,
Crises."
Staff Papers
-
International
Vol. 45, No. 1 (Mar.), pp. 1-48.
Kaminsky, Graciela and Carmen M. Reinhart. 1999. "The Twin Crises: The Causes of Banking and
Balance-Of-Payments Problems." The American Economic Review,
Vol. 89, No. 3 (Jun.), pp.
473-
500.
Kotlikoff,
Laurence
J.
and John
C.
Goodman. 2009.
"Solving Our Nation's Financial Crisis w/ith
Limited Purpose Banking." Boston University.
Lehman
Brothers. 2008. "Residential Credit Losses
Mendoza, Enrique
G.
-
Going into Extra Innings?"
and Marco Terrones. 2008. "An Anatomy of Credit Booms: Evidence from
Macro Aggregates and Micro Data," NBER Working Paper 14049.
Mehriing, Perry and Aiistair Milne, 20008. "Government's role as credit insurer of
and how
it
can be
fulfilled,"
mimeo October
Milne, Aiistair. 2009. The Fall of the
and
New
last resort
House of Credit. Cambridge University
Press,
Cambridge
York.
Office of the Special Inspector General for the Troubled Asset Relief Program. 2009. Quarterly
Report to Congress,
Reinhart,
July.
Carmen M. and Kenneth
Rogoff.
2008. "Banking Crises: An
Equal Opportunity
Menace." NBER Working Paper No. wl4587.
Swagel,
Phillip.
2009. "The Financial
Crisis:
An
Inside View". Brookings Papers on
Economic
Activity, Spring.
UK
Treasury. 2009. The
Road
to the
London Summit, http://www.londonsummit.gov.uk
V\/oodford, Michael. 1990. "Public Debt as Private Liquidity," The
American Economic Review,
Vol. 80, No. 2 (May), pp. 382-88.
39