[House Hearing, 111 Congress]
[From the U.S. Government Publishing Office]
THE RISKS OF FINANCIAL MODELING:
VAR AND THE ECONOMIC MELTDOWN
=======================================================================
HEARING
BEFORE THE
SUBCOMMITTEE ON INVESTIGATIONS AND
OVERSIGHT
COMMITTEE ON SCIENCE AND TECHNOLOGY
HOUSE OF REPRESENTATIVES
ONE HUNDRED ELEVENTH CONGRESS
FIRST SESSION
__________
SEPTEMBER 10, 2009
__________
Serial No. 111-48
__________
Printed for the use of the Committee on Science and Technology
Available via the World Wide Web: http://www.science.house.gov
For Sale by the Superintendent of Documents, U.S. Government Printing Office
U.S. GOVERNMENT PRINTING OFFICE
51-925 WASHINGTON : 2010
-----------------------------------------------------------------------
Internet: bookstore.gpo.gov Phone: toll free (866) 512-1800; (202) 512�091800
Fax: (202) 512�092104 Mail: Stop IDCC, Washington, DC 20402�090001
______
COMMITTEE ON SCIENCE AND TECHNOLOGY
HON. BART GORDON, Tennessee, Chair
JERRY F. COSTELLO, Illinois RALPH M. HALL, Texas
EDDIE BERNICE JOHNSON, Texas F. JAMES SENSENBRENNER JR.,
LYNN C. WOOLSEY, California Wisconsin
DAVID WU, Oregon LAMAR S. SMITH, Texas
BRIAN BAIRD, Washington DANA ROHRABACHER, California
BRAD MILLER, North Carolina ROSCOE G. BARTLETT, Maryland
DANIEL LIPINSKI, Illinois VERNON J. EHLERS, Michigan
GABRIELLE GIFFORDS, Arizona FRANK D. LUCAS, Oklahoma
DONNA F. EDWARDS, Maryland JUDY BIGGERT, Illinois
MARCIA L. FUDGE, Ohio W. TODD AKIN, Missouri
BEN R. LUJAN, New Mexico RANDY NEUGEBAUER, Texas
PAUL D. TONKO, New York BOB INGLIS, South Carolina
PARKER GRIFFITH, Alabama MICHAEL T. MCCAUL, Texas
STEVEN R. ROTHMAN, New Jersey MARIO DIAZ-BALART, Florida
JIM MATHESON, Utah BRIAN P. BILBRAY, California
LINCOLN DAVIS, Tennessee ADRIAN SMITH, Nebraska
BEN CHANDLER, Kentucky PAUL C. BROUN, Georgia
RUSS CARNAHAN, Missouri PETE OLSON, Texas
BARON P. HILL, Indiana
HARRY E. MITCHELL, Arizona
CHARLES A. WILSON, Ohio
KATHLEEN DAHLKEMPER, Pennsylvania
ALAN GRAYSON, Florida
SUZANNE M. KOSMAS, Florida
GARY C. PETERS, Michigan
VACANCY
------
Subcommittee on Investigations and Oversight
HON. BRAD MILLER, North Carolina, Chair
STEVEN R. ROTHMAN, New Jersey PAUL C. BROUN, Georgia
LINCOLN DAVIS, Tennessee BRIAN P. BILBRAY, California
CHARLES A. WILSON, Ohio VACANCY
KATHY DAHLKEMPER, Pennsylvania
ALAN GRAYSON, Florida
BART GORDON, Tennessee RALPH M. HALL, Texas
DAN PEARSON Subcommittee Staff Director
EDITH HOLLEMAN Subcommittee Counsel
JAMES PAUL Democratic Professional Staff Member
DOUGLAS S. PASTERNAK Democratic Professional Staff Member
KEN JACOBSON Democratic Professional Staff Member
TOM HAMMOND Republican Professional Staff Member
MOLLY O'ROURKE Research Assistant
ALEX MATTHEWS Research Assistant
C O N T E N T S
September 10, 2009
Page
Witness List..................................................... 2
Hearing Charter.................................................. 3
Opening Statements
Statement by Representative Brad Miller, Chairman, Subcommittee
on Investigations and Oversight, Committee on Science and
Technology, U.S. House of Representatives...................... 6
Written Statement............................................ 8
Statement by Representative Paul C. Broun, Ranking Minority
Member, Subcommittee on Investigations and Oversight, Committee
on Science and Technology, U.S. House of Representatives....... 9
Written Statement............................................ 10
Panel I:
Dr. Nassim N. Taleb, Distinguished Professor of Risk Engineering,
Polytechnic Institute of New York University; Principal,
Universa Investments L.P.
Oral Statement............................................... 11
Written Statement............................................ 13
Biography.................................................... 56
Dr. Richard Bookstaber, Financial Author
Oral Statement............................................... 56
Written Statement............................................ 59
Biography.................................................... 67
Discussion
Can Economic Events Be Predicted?.............................. 67
Regulation of Financial Products............................... 69
`Too Big to Fail'?............................................. 70
Wall Street's Dependency on Government Bailouts................ 72
The Risks of Different Tupes of Institutions................... 74
Incentive Structures for Trades................................ 75
Holding Wall Street Accountable for Bonuses.................... 78
Malpractice in Risk Management................................. 79
Clawback Provisions............................................ 80
Credit Default Swaps........................................... 81
Were the Bailouts and Stimulus Funds Necessary?................ 82
Panel II:
Dr. Gregg E. Berman, Head of Risk Business, RiskMetrics Group
Oral Statement............................................... 84
Written Statement............................................ 86
Biography.................................................... 105
Mr. James G. Rickards, Senior Managing Director for Market
Intelligence, Omnis, Inc., McLean, VA
Oral Statement............................................... 106
Written Statement............................................ 108
Biography.................................................... 116
Mr. Christopher Whalen, Managing Director, Institutional Risk
Analytics
Oral Statement............................................... 117
Written Statement............................................ 118
Biography.................................................... 124
Dr. David Colander, Christian A. Johnson Distinguished Professor
of Economics, Middlebury College
Oral Statement............................................... 124
Written Statement............................................ 127
Biography.................................................... 141
Discussion
Appropriate Uses of Financial Models........................... 141
Proposals for Avoiding Recurrences of Financial Problems....... 144
Abuse of the VaR............................................... 145
Past Congressional Attempts to Regulate the Financial Industry. 145
Should a Government Agency Test Financial Products for
Usefulness?.................................................. 146
Identifying Firms That Are `Too Big to Fail'................... 148
Monitoring and Analyzing Hedge Fund Activity and Risk.......... 149
THE RISKS OF FINANCIAL MODELING: VAR AND THE ECONOMIC MELTDOWN
----------
THURSDAY, SEPTEMBER 10, 2009
House of Representatives,
Subcommittee on Investigations and Oversight,
Committee on Science and Technology,
Washington, DC.
The Subcommittee met, pursuant to call, at 10:04 a.m., in
Room 2318 of the Rayburn House Office Building, Hon. Brad
Miller [Chairman of the Subcommittee] presiding.
hearing charter
SUBCOMMITTEE ON INVESTIGATIONS AND OVERSIGHT
COMMITTEE ON SCIENCE AND TECHNOLOGY
U.S. HOUSE OF REPRESENTATIVES
The Risks of Financial Modeling:
VaR and the Economic Meltdown
thursday, september 10, 2009
10:00 a.m.-1:00 p.m.
2318 rayburn house office building
Purpose
The Subcommittee on Investigations and Oversight on Sept. 10, 2009
convenes the first Congressional hearing to examine the role of risk
modeling in the global financial meltdown. Risk models, and
specifically a method of risk measurement known as Value-at-Risk, or
VaR, are widely viewed as an important factor in the extreme risk-
taking that financial institutions engaged in leading to last year's
economic upheaval. That risk-taking has led to hundreds of billions of
dollars in losses to financial firms, and to a global recession with
trillions of dollars in direct and indirect costs imposed on U.S.
taxpayers and working families.
Given the central role of credit in the economy, the ability of
major financial institutions to operate without assuming undue risks
that gamble with the stability of the financial system, thereby
endangering the broader economy, is of the utmost importance to both
business and the public at large. The recent behavior by financial
firms that are deemed ``too big to fail'' suggests that the financial
system as currently structured and regulated creates a ``moral hazard''
because firms can expect that they will be bailed out if their risk-
taking fails to pay off. This is exactly what happened in the United
States in October of 2008 with great consequences to the taxpayers, who
have been called upon to shoulder much of the huge burden arising from
financial firms' underestimation of risk, poor judgment, and profligate
behavior. Relied on to guide the decisions of both financial firms and
federal regulators responsible for monitoring their soundness by
ensuring that they have sufficient capital, the VaR, whether it was
misused or not, was involved in inducing or allowing this situation to
arise.
Given this dual function, it is critical that the Subcommittee
examine: the role of the VaR and related risk-measurement methods in
the current world financial crisis; the strengths and weaknesses of,
and the limits to, the usefulness of the VaR; the degree to which the
VaR is understood, and may be manipulated, within the institutions
where it is in use; and the capabilities and needs of federal
supervisors who may be called upon to work with the VaR in carrying out
their regulatory duties. From a policy perspective, the most important
question is how regulators will use VaR numbers produced by firms and
whether it is an appropriate guide to setting capital reserve
requirements.
This is the second in a series of hearings on how economic thinking
and methods have been used by policy-makers both inside and outside of
government.
The VaR's Origins and Use
Risk assessment models in the financial industry are the product of
advances in economic and statistical methods developed in the social
sciences over the last fifty years. J.P. Morgan adopted these
techniques in developing the VaR in the 1980s as a tool to measure the
risk of loss to its traders' portfolios. The VaR could produce a single
number rating a trader's (or, in aggregate, the firm's cumulative) risk
of loss of portfolio value over a specific period of time at a given
level of confidence. The VaR provided managers a tool that appeared to
allow them to keep a handle on the risks they were taking as financial
instruments became more varied and complex and as assets became more
difficult to value. Morgan decided to give the methodology of the VaR
away, forming the now-independent RiskMetrics Group; this resulted in
the VaR rapidly becoming ``so popular that it was considered the risk-
model gold standard.'' \1\
---------------------------------------------------------------------------
\1\ ``Risk Management,'' by Joe Nocera, New York Times, Jan. 4,
2009. J.P. Morgan was not the only firm to look for statistical tools
to measure the risks of their portfolios, however Morgan's model became
the most widely used. The model can be tweaked in many, many ways to
meet the specific needs of a particular firm.
---------------------------------------------------------------------------
To put it very simply, the VaR captures the probability of outcomes
distributed along a curve-most commonly a ``bell'' or normal
distribution. It provides an answer to the question of, ``what is
likely to happen tomorrow to the value of an asset?'' by drawing from
historical performance data. The highest probability of tomorrow's
value is that it will be the same as today's value; the next highest
probability is for a very small movement in value up or down, and so
on. The more radical the movement in value, the lower the probability
of that occurring. A manager may ask for a projection of the potential
loss of an asset or portfolio at the 95 percent or even the 99 percent
confidence level. At those levels, a complete loss of value is
unlikely. The complete collapse of an asset or portfolio's value is not
a 1-in-100 event; such a collapse is more likely a 1-in-500 or 1-in-
10,000 or event. The VaR is unlikely to warn, then, of great shifts in
value. The danger to the financial firm or the community comes at the
extreme margins of the distribution curves produced by the VaR. As a
map to day-to-day behavior, the VaR is probably pretty accurate for
normal times, but for asset bubbles or other ``non-normal'' market
conditions, the VaR is likely to misrepresent risks and dangers.
While the VaR was originally designed for financial institutions'
use in-house, it has subsequently been given a key role in determining
capital requirements for large banks under a major multilateral
agreement, the Basel II Accord, published in 2004. That same year, the
U.S. Securities and Exchange Commission adopted a capital regime
applying Basel II standards to the Nation's largest investment
banks,\2\ a move that has been viewed as playing a role in those
institutions' subsequent over-leveraging and liquidity problems. Those
financial institutions assured regulators that the VaR was a way to see
the level of risk they were taking on and a low VaR justified lower
reserve requirements. (The terms of Basel II are currently being re-
evaluated in light of the global economic crisis.)
---------------------------------------------------------------------------
\2\ ``Alternative Net Capital Requirements for Broker-Dealers That
are Part of Consolidated Supervised Entities; Supervised Investment
Bank Holding Companies; Final Rules,'' Securities and Exchange
Commission, June 21, 2004, 69 FR 34428-72. (According to Aswath
Damodaran, Professor of Finance at the NYU Stern School of Business,
``The first regulatory measures that evoke Value-at-Risk, though, were
initiated in 1980, when the SEC tied the capital requirements of
financial service firms to the losses that would be incurred, with 95
percent confidence over a thirty-day interval, in different security
classes; historical returns were used to compute these potential
losses. Although the measures were described as haircuts and not as
Value or Capital at Risk, it was clear the SEC was requiring financial
service firms to embark on the process of estimating one month 95
percent VaRs and hold enough capital to cover the potential losses.''
Damodaran, ``Value-at-Risk (VAR),'' found at http://
pages.stern.nyu.edu/?adamodar/pdfiles/papers/VAR.pdf)
---------------------------------------------------------------------------
Along with extensive use, the VaR has come in for extensive
criticism. Although its merits were debated at least as far back as
1997,\3\ criticism of the VaR has mounted in the wake of last year's
collapse of such major financial institutions as Bear Stearns and
Lehman Brothers. Among the allegations: that the VaR is inadequate in
capturing risks of extreme magnitude but low probability, to which an
institution may be left vulnerable; that this shortcoming may open it
to manipulation by traders taking positions that seem profitable but
whose risks they know the VaR is unlikely to pick up, and that such
``gaming'' can increase extreme risk; and that use of the VaR, derided
for ``quantify[ing] the immeasurable with great precision,'' \4\
promotes an unfounded sense of security within financial institutions
creating an environment where firms take on more risk than they would
without the security-blanket of a VaR number.
---------------------------------------------------------------------------
\3\ ``The Jorion-Taleb Debate,'' DerivativesStrategy.com, April
1997, http://www.derivativesstrategy.com/magazine/archive/1997/
0497fea2.asp
\4\ ``Against VAR,'' by Nassim Taleb, in ``The Jorion-Taleb
Debate,'' ibid.
---------------------------------------------------------------------------
Those who advocate for the VaR argue that any misuse of the model
is not the model's fault and that it remains a useful management tool.
VaR defenders' argue that its purpose is ``not to describe the worst
possible outcomes;'' \5\ that it is essential to the ability of a
financial institution to arrive at an estimate of its overall risk; and
that in ``computing their VAR[, institutions] are forced to confront
their exposure to financial risks and to set up a proper risk
management function,'' so that ``the process of getting to VAR may be
as important as the number itself.'' \6\ Some also argue that the VaR
remains a useful tool for regulators to use as a baseline for
establishing reserve requirements for ``normal'' times.
---------------------------------------------------------------------------
\5\ ``In Defense of VAR,'' by Philippe Jorion, in ``The Jorion-
Taleb Debate,'' ibid.
\6\ Jorion, idem.
---------------------------------------------------------------------------
Witnesses
Panel I
Dr. Nassim Nicholas Taleb, Distinguished Professor of Risk Engineering,
Polytechnic Institute of New York University.
Dr. Richard Bookstaber, Financial Author
Panel II
Dr. Gregg Berman, Head of Risk Business, RiskMetrics Group
Mr. James G. Rickards, Senior Managing Director, Omnis Inc.
Mr. Christopher Whalen, Managing Director, Institutional Risk Analytics
Dr. David Colander, Christian A. Johnson Distinguished Professor of
Economics, Middlebury College
Chairman Miller. Good morning, and welcome to today's
hearing: ``The Risks of Financial Modeling: VaR and the
Economic Meltdown.''
Economics has not been known in the past for mathematical
precision. Harry Truman said he wanted a one-handed economist
because he was frustrated with economists who equivocated by
saying on the one hand, on the other hand. George Bernard Shaw
said that if all the world's economists were laid end to end,
they still wouldn't reach a conclusion. And apparently no one
is sure who first observed that economics was the only field in
which it was possible for two people to share a Nobel Prize for
reaching exactly the opposite conclusion about the same
question.
In the last 15 or 20 years, math and physics Ph.D.s from
academia and the laboratory have entered the financial sector.
Quantitative analysts, or `quants,' directed their mathematical
and statistical skills to financial forecasts at a time when
global financial markets were becoming more interdependent than
ever before.
The quants conceived such financial instruments as
collaterized debt obligations, or CDOs, and credit default
swaps, or CDSs, that would never have existed without them and
their computers. They developed strategies for trading those
instruments even in the absence of any underlying security or
any real market; for that matter, in the absence of anything at
all. They constructed risk models that convinced their less
scientifically and technologically adept bosses that their
instruments and strategies were infallibly safe. And their
bosses spread the faith in the quants' models to regulators,
who agreed to apply them to establish capital reserve
requirements that were supposed to guarantee the soundness of
financial institutions against adverse events. It almost seemed
like the economic models had brought the precision of the laws
of physics, the same kind of certainty about the movement of
the planets, to financial risk management. Engineering schools
even offered courses in ``financial engineering.''
The supposedly immutable laws underlying the quants'
models, however, didn't work out, and the complex models turned
out to have hidden risks rather than protecting against them,
all at a terrible cost. Those risks, concealed and maybe even
encouraged by the models, have led to hundreds of billions of
dollars in losses to investors and taxpayers, to a global
recession imposing trillions of dollars in losses to the world
economy and immeasurable monetary and human costs. People
around the world are losing their homes, their jobs, their
dignity and their hope.
Taxpayers here and around the world are shouldering the
burden arising from financial firms' miscalculation of risk,
poor judgment, excessive bonuses and general profligate
behavior. It is for this reason that the Subcommittee is
directing our attention today to the intersection of
quantitative analysis, economics and regulation. The Value-at-
Risk model, or VaR, stands squarely at the intersection of
quantitative analysis, economics and regulation. It is the most
prominent risk model used by major financial institutions. The
VaR is designed to provide an answer to the question, ``What is
the potential loss that could be faced within a limited,
specified time to the value of an asset?''
The highest probability is that tomorrow's value will be
the same as today's. The next highest probability is that there
will be a small movement in value up or down, and so on. The
more radical the movement in value, the lower the probability
that it will happen. In other words, the danger to a financial
firm or the community comes at the extreme margins of the VaR
distribution curve, in the tails of the distribution. As a map
to day-to-day behavior, the VaR is probably pretty accurate for
normal times, just as teams favored by odds makers usually win.
But just as long shots sometimes come home, just as underdogs
do sometimes win, asset bubbles or other non-normal market
conditions also occur, and the VaR is unlikely to capture the
risks and dangers. The VaR also cannot tell you when you have
moved into non-normal market conditions.
While the VaR was originally designed for financial
institutions' in-house use to evaluate short-term risk in their
trading books, it has been given a key role in determining
capital requirements for large banks under a major multilateral
agreement, the Basel II Accord, published in 2004. That same
year, the U.S. Securities and Exchange Commission, the SEC, at
the instigation of the five largest investment banks, adopted a
capital reserve regime, applying Basel II standards to the
Nation's largest investment banks--a decision that opened the
door to their over-leveraging and liquidity problems. Three of
the institutions that asked the SEC for this change in rules--
Bear Stearns, Merrill Lynch, Lehman Brothers--no longer exist.
At the time, those financial institutions assured regulators
that the VaR would reflect the level of risk they were taking
on, and that a low VaR justified lower capital requirements.
The result was exactly what the investment banks asked for:
lower capital requirements that allowed them to invest in even
more risky financial instruments all justified with risk models
that assured regulators that there was nothing to worry about.
What could possibly go wrong?
In light of the VaR's prominent role in the financial
crisis, this subcommittee is examining that role and the role
of related risk-measurement methods. From a policy perspective,
the most important immediate question is how regulators use VaR
numbers and other such models designed by regulated
institutions, and whether they are an appropriate guide to
setting capital reserve requirements. But, beyond that, we must
also ask whether the scientific and technical capabilities that
led us into the current crisis should be applied to prevent
future catastrophic events. Can mathematics, statistics and
economics produce longer-range models, more reliable models,
that could give us early warning when our financial system is
headed for trouble? Or are such models inevitably going to be
abused to hide risk-taking and encourage gambling by firms
whose failures can throw the whole world into a recession, as
they have in the last couple of years? If models cannot be a
useful guide for regulation, should we just abandon the
approach, or simply increase reserves, which will reduce
profits and perhaps reduce some useful economic conduct in the
short run, but protect taxpayers and the world economy in the
long run?
Those are big questions, but the stakes for taxpayers and
investors and the world economy justify some effort to get at
some answers.
I now recognize Dr. Broun for his opening statement.
[The prepared statement of Chairman Miller follows:]
Prepared Statement of Chairman Brad Miller
Economics has not been known in the past for mathematical
precision. Harry Truman said he wanted a one-handed economist because
he was frustrated with economists who equivocated by saying ``on the
one hand . . . on the other hand.'' George Bernard Shaw said that if
all the world's economists were laid end to end, they still wouldn't
reach a conclusion. And apparently no one knows who first observed that
economics was the only field in which two people can share a Nobel
Prize for reaching exactly the opposite conclusion.
But in the last 15 or 20 years, math and physics Ph.D.s from
academia and the laboratory have entered the financial sector.
Quantitative analysts, or ``quants,'' directed their mathematical and
statistical skills to financial forecasts at a time when global
financial markets were becoming more interdependent than ever before.
The quants conceived such financial instruments as collaterized
debt obligations, or ``CDOs,'' and credit default swaps, or ``CDSs,''
that would never have existed without them and their computers. They
developed strategies for trading those instruments even in the absence
of any underlying security or any real market. They constructed risk
models that convinced their less scientifically and technologically
adept bosses that their instruments and strategies were infallibly
safe. And their bosses spread faith in the quants' models to
regulators, who agreed to apply them to establish capital reserve
requirements that were supposed to guarantee the soundness of financial
institutions against adverse events. It almost seemed like economic
models had brought the precision of the laws of physics to financial
risk management. Engineering schools even offered courses in
``financial engineering.''
The supposedly immutable ``laws'' underlying the quants' models
didn't work, and the complex models turn out to have hidden risks
rather than protected against them, all at a terrible cost. Those
risks--concealed and maybe even encouraged by the models--have led to
hundreds of billions of dollars in losses to investors and the
taxpayers, to a global recession imposing trillions of dollars in
losses to the world economy and immeasurable monetary and human costs.
People around the world are losing their jobs, their homes, their
dignity and their hope.
Taxpayers here and around the world are shouldering the burden
arising from financial firms' miscalculation of risk, poor judgment,
excessive bonuses and profligate behavior. It is for this reason that
the Subcommittee has chosen to direct its attention today to that
intersection of quantitative analysis, economics, and regulation. The
``Value-at-Risk'' model, or ``VaR'' stands squarely at the center of
this intersection as the most prominent risk model used by major
financial institutions. The VaR is designed to provide an answer to the
question, ``What is the potential loss that could be faced within a
limited, specified time to the value of an asset?''
The highest probability is that tomorrow's value will be the same
as today's; the next highest probability is of a very small movement in
value up or down, and so on. The more radical the movement in value,
the lower the probability of its occurrence. In other words, the danger
to the financial firm or the community comes at the extreme margins of
the VaR distribution curve, in the ``tails'' of the distribution. As a
map to day-to-day behavior, the VaR is probably pretty accurate for
normal times, just as teams favored by odds makers usually win. But
just as long shots sometimes come home, asset bubbles or other ``non-
normal'' market conditions also occur, and the VaR is unlikely to
capture the risks and dangers. The VaR also cannot tell you when you
have moved into ``non-normal'' market conditions.
While the VaR was originally designed for financial institutions'
to use in-house to evaluate short-term risk in their trading books, it
was given a key role in determining capital requirements for large
banks under a major multilateral agreement, the Basel II Accord,
published in 2004. That same year, the U.S. Securities and Exchange
Commission, at the instigation of the five largest investment banks,
adopted a capital reserve regime applying Basel II standards to the
Nation's largest investment banks, a decision that opened the door to
their over-leveraging and liquidity problems. Three of the institutions
that asked the SEC for this change in rules--Bear Stearns, Merrill
Lynch, Lehman Brothers--no longer exist. At the time, those financial
institutions assured regulators that the VaR would reflect the level of
risk they were taking on, and that a low VaR justified lower reserve
requirements. The result was exactly what the investment banks asked
for; lower capital reserve requirements that allowed them to invest in
even more risky financial instruments all justified with risk models
that assured regulators that there was nothing to worry about.
In light of the VaR's prominent role in the financial crisis, this
Subcommittee is examining that role and the role of related risk-
measurement methods. From a policy perspective, the most important
immediate question is how regulators use VaR numbers and other such
models devised by regulated institutions and whether they are an
appropriate guide to setting capital reserve requirements. But, beyond
that, we must also ask whether the scientific and technical
capabilities that helped lead us into the current crisis should be
applied to prevent future catastrophic events. Can mathematics,
statistics, and economics produce longer-range models--models that
could give us early warning of when our complex financial system is
heading for trouble? Or are such models inevitably going to be abused
to hide risk-taking and encourage excessive gambling by firms whose
failures can throw the whole world into a recession? If models cannot
be a useful guide for regulation, should we just abandon this approach
and simply increase reserves, reducing profits and perhaps some useful
economic conduct in the short run, but protecting taxpayers and the
world economy in the long run?
These are big questions, but the stakes for taxpayers and investors
and the world economy justify the effort to get at some answers.
I now recognize Mr. Broun for his opening statement.
Mr. Broun. Thank you, Mr. Chairman. Let me welcome the
witnesses here today and thank them for appearing. Today's
hearing on financial modeling continues this committee's work
on the role of science in finance and economics.
As I pointed out in our previous hearing in May, for the
last several years Wall Street has increasingly leveraged
mathematics, physics and science to better inform their
decisions. Even before Value-at-Risk was developed to
characterize risk, bankers and economists were looking for a
silver bullet to help them to beat the market.
Despite the pursuit of a scientific panacea for financial
decisions, models are simply tools employed by decision-makers
and risk managers. They add another layer of insight but are
not crystal balls. Leveraging a position too heavily or
assuming future solvency based on modeling data alone is
hazardous, to say the least. Conversely, it stands to reason
that if we could accurately predict markets, then both losses
and profits would be limited since there would be very little
risk involved.
Modeling is a subject this committee has addressed several
times in the past, whether it is in regard to climate change,
chemical exposures, pandemics, determining spacecraft
survivability or attempting to value complex financial
instruments. Models are only as good as the data and
assumptions that go into them. Ultimately decisions have to be
made based on a number of variables which should include
scientific models but certainly not exclusively. As witnesses
in our previous hearing stated, ``Science describes, it does
not prescribe.'' No model will ever relieve a banker, trader or
risk manager of the responsibility to make difficult decisions
and hedge inevitable uncertainly.
This committee struggles enough with the complexities of
modeling, risk assessment and risk management regarding
physical sciences. Attempting to adapt those concepts to
economics and finance is even more complex. Appreciating this
complexity and understanding the limitations and intended
purpose of financial models is just as important as what the
models tell you.
We have two esteemed panels of witnesses here today who
will discuss appropriate roles and limitations of models such
as VaR. They will explain how these models are used and shed
some light on what role they may have played in the recent
economic crisis. I look forward to you all's testimony and I
yield back my time. Thank you, Mr. Chairman.
[The prepared statement of Mr. Broun follows:]
Prepared Statement of Representative Paul C. Broun
Thank you Mr. Chairman.
Let me welcome the witnesses here today and thank them for
appearing.
Today's hearing on Financial Modeling continues this committee's
work on the role of science in finance and economics.
As I pointed out at our previous hearing in May, over the last 30
years Wall Street has increasingly leveraged mathematics, physics, and
science to better inform their decisions. Even before Value-at-Risk
(VaR) was developed to characterize risk, bankers and economists were
looking for a silver bullet to help them beat the market.
Despite the pursuit of a scientific panacea for financial
decisions, models are simply tools employed by decision-makers and risk
managers. They add another layer of insight, but are not crystal balls.
Leveraging a position too heavily or assuming future solvency based on
modeling data alone is hazardous to say the least. Conversely, it
stands to reason that if we could accurately predict markets, then both
losses and profits would be limited since there would be very little
risk involved.
Modeling is a theme this committee has addressed several times in
the past. Whether it is in regard to climate change, chemical
exposures, pandemics, determining spacecraft survivability, or
attempting to value complex financial instruments, models are only as
good as the data and assumptions that go into them. Ultimately,
decisions have to be made based on a number of variables which should
include scientific models, but certainly not exclusively. As a witness
at a previous hearing stated, ``science describes, it does not
prescribe.'' No model will ever relieve a banker, trader, or risk
manager of the responsibility to make difficult decisions and hedge for
inevitable uncertainty.
This committee struggles enough with the complexities of modeling,
risk assessment, and risk management regarding physical sciences.
Attempting to adapt those concepts to economics and finance is even
more complex. Appreciating this complexity, and understanding the
limitations and intended purpose of financial models is just as
important as what the models tell you.
We have two esteemed panels of witnesses here today who will
discuss the appropriate roles and limitations of models such as VaR.
They will explain how these models are used and shed some light on what
role they may have played in the recent economic crisis. I look forward
to their testimony and yield back my time.
Thank you.
Chairman Miller. Thank you, Dr. Broun.
I now ask unanimous consent that all additional opening
statements submitted by Members be included in the record.
Without objection, that is so ordered.
Panel I:
We do have an outstanding group of witnesses today. I know
that Chairmen at hearings always say that but it is certainly
true. This time I mean it. On our first panel, we have two very
well known and respected authors whose books and other writings
warned against many of the practices of the financial industry
that resulted in the current economic meltdown. Both of them
have years of experience on Wall Street. Dr. Nassim Taleb is
the author of ``Fooled by Randomness'' and ``The Black Swan.''
After a career as a trader and fund manager, Dr. Taleb is now
the Distinguished Professor of Risk Engineering at the
Polytechnic Institute of New York University. And if you are
one of that slice of the American population for whom Bloomberg
and CNBC are your favorite TV channels, Dr. Taleb is a rock
star. Dr. Taleb is joined by another rock star, Dr. Richard
Bookstaber, who is the author of ``A Demon of Our Own Design:
Markets, Hedge Funds and the Risk of Financial Innovation.''
Dr. Bookstaber has worked as a risk manager for Salomon
Brothers, Morgan Stanley and Moore Capital Management. He also
runs equity funds and he began on Wall Street designing
derivative instruments. Does your mother know about that?
As our witnesses should know, you will each have five
minutes for your spoken testimony. Your written testimony will
be included in the record for the hearing. When you all have
completed your spoken testimony, we will begin with questions
and each Member will have five minutes to question the panel.
It is the practice of this subcommittee--it is an investigative
and oversight subcommittee--to receive testimony under oath. As
I pointed out to the panelists at our last hearing on economic
issues, to prosecute a case for perjury, the prosecutor, the
U.S. attorney would have to prove what the truth was, that you
knew the truth and that you consciously departed from it. I
think you can sleep easily without worrying about a prosecution
for perjury, but we will ask you to take an oath. Do either of
you have any objection to taking an oath? Okay. You also have
the right to be represented by counsel. Do either of you have
counsel here? If you would, please stand and raise your right
hand. Do you swear to tell the truth and nothing but the truth?
The record will reflect that both witnesses did take the
oath. We will begin with Dr. Taleb. Dr. Taleb.
STATEMENT OF DR. NASSIM N. TALEB, DISTINGUISHED PROFESSOR OF
RISK ENGINEERING, POLYTECHNIC INSTITUTE OF NEW YORK UNIVERSITY;
PRINCIPAL, UNIVERSA INVESTMENTS L.P.
Dr. Taleb. Mr. Chairman, Ranking Member, Members of the
Committee, thank you for giving me this opportunity to testify
on the risk measurement methods used by banks, particularly
those concerned with the risks of VaR events. You know, Value-
at-Risk is just a method. It is a very general method, not very
precise method, that measures the risks of VaR events. For
example, a standard daily Value-at-Risk tells you that if your
VaR is a million, daily VaR is a million, you have--it is at
one percent probability, that you have less than one percent
chance of losing a million or more on a given day. There are of
course a lot of variations around VaR. For me, they are equally
defective.
Thirteen years ago, I wrote that the VaR encourages
misdirected people to take risks with shareholders' and
ultimately taxpayers' money--that is, regular people's money. I
have been since begging for suspension of these measurements of
tail risks. We just don't understand tail events. And lot of
people say, oh, let's measure risks. My idea is very different.
Let's find what risks we can measure and any other risks we
should be taking instead of doing it the opposite way. We take
a lot of risks and then we try to find some scientists who can
confirm these methods, you know, that these risks we can
measure and that these methods are sound.
I have been begging, and actually I wrote that I would be
on the witness stand 13 years ago, and today I am here. The
banking system lost so far more than $4.3 trillion, according
to the International Monetary Fund--that is more than they ever
made in the history of banking--on tail risks, measurements of
rare events. Most of the losses of course were in the United
States, and I am not counting the economic consequences. But
this shouldn't have happened. Data shows that banks routinely
lose everything they made over a long period of time in one
single blow-up. It happened in 1982 because of multi-center
banks losing everything made in the history of multi-center
banking, one single event, loans to Latin America. The same
thing in variation happened in 1991, and of course now. And
every time society bails them out. Bank risk takers retain
their bonuses and say oh, one fluke, all right, and we start
again. This is an aberrant case of capitalism for the profit,
and socialism for the losses.
So I have five points associated with VaR that I will go
over very quickly, and I will give my conclusion. Number one:
these problems were obvious all along. This should not have
happened. We knew about the defects of the VaR when it was
introduced. A lot of traders, a lot of my friends, everyone--I
am not the only person ranting against VaR. A lot of people
were ranting against it before. Nobody heard us. Regulators did
not listen to anyone who knew what was going on, is my point
number one.
Point number two: VaR is ineffective. I guess I don't need
more evidence than the recent events to convince you.
Point number three, and that to me is crucial. You have a
graph that shows you the performance profile of someone making
steady earnings for a long time and then losing back
everything. You can see from that graph, figure one on page
four, that this is a strategy that is pretty much pursued by
the majority of people on Wall Street, by banks. They make
steady income for a long time, and when they blow up, they say,
well, you know, it was unexpected, it was a black swan. I wrote
a book called ``The Black Swan.'' Unfortunately, they used my
book backwards. Oh, and it was unexpected, highly unexpected.
They keep their bonuses. They go on vacation and here you have
a regular person working very hard, a taxpayer, a taxi driver,
a post office worker paying taxes to subsidize retrospectively,
all right, bonuses made. For example, a former government
official made $121 million in bonuses at Citibank. Okay. He
keeps his bonuses. We retrospectively are paying for that. That
I said 13 years ago, and it keeps happening, and now we are
still in the same situation.
So number four, and that is another crucial point. VaR has
side effects. It is not neutral. You give someone a number--it
has been shown and shown repeatedly, if you give someone a
number, he will act on that number even if you tell him that
that number is random. We humans cannot be trusted with
numbers. You don't give someone the map of the Alps if he is on
the Mount Ararat, all right, because he is going to act on that
map. Even nothing is alot better, if it doesn't work. This is
my central point, the side effects of numerical precision given
to people who do not need it.
Number five: VaR-style quantitative risk management was
behind leverage. We increased our leverage in society as we
thought we thought we could measure risk. If you think you can
measure your blow-up risk, you are going to borrow, you know.
You have more overconfidence, also, as a side effect of
measurement, and you are going to borrow. Instead of, you know,
taking equity from people, you borrow, so when you blow up, you
owe that money. And of course, as was discussed in my paper,
debt bubbles are very vicious. Equity bubbles are not very
vicious.
Conclusion: What should we be doing? Well, regulators
should understand that finance is a complex system and complex
systems have very clear characteristics, you know, and one of
them is low levels of predictability, particularly of tail
events. We have to worry--regulators should not encourage model
error. My idea is to build a society that is resistant to
expert mistakes. Regulators increased the dependence of society
on expert mistakes and other things also in the Value-at-Risk,
these AAA things. Okay. So we want to reduce that. We want to
build a society that can sustain shocks because we are moving
more and more into a world that delivers very large-scale
variables, and we know exactly how they affect us or we know
with some precision how they affect us, and we know how to
build shocks. So the job of regulators should be to lower the
impact of model error, and this is reminiscent of medicine. You
know, the FDA, they don't let you bring any medicine without
showing the side effects. Well, we should be doing the same
thing in economic life. Thank you very much for this
opportunity.
[The prepared statement of Dr. Taleb follows:]
Prepared Statement of Nassim N. Taleb
Report on the Risks of Financial Modeling,
VaR and the Economic Breakdown
INTRODUCTION
Mr. Chairman, Ranking Member, Members of the Committee, thank you
for giving me the opportunity to testify on the risk measurement
methods used by banks, particularly those concerned with blowup risk,
estimates of probabilities of losses from extreme events (``tail
risks''), generally bundled under VaR.\1\
---------------------------------------------------------------------------
\1\ The author thanks Daniel Kahneman, Pablo Triana, and Eric
Weinstein for helpful discussions.
---------------------------------------------------------------------------
What is the VaR? It is simply a model that is supposed to project
the expected extreme loss in an institution's portfolio that can occur
over a specific time frame at a specified level of confidence. Take an
example. A standard daily VaR of $1 million at a one percent
probability tells you that you have less than a one percent chance of
losing $1 million or more on a given day.\2\ There are many
modifications around VaR, ``conditional VaR,'' \3\ so my discussion
concerns all quantitative (and probabilistic) methods concerned with
losses associated with rare events. Simply, there are limitations to
our ability to measure the risks of extreme events.
---------------------------------------------------------------------------
\2\ Although such definition of VaR is often presented as a
``maximum'' loss, it is technically not so in an open-ended exposure:
since, conditional on losing more than $1 million, you may lose a lot
more, say $5 million.
\3\ Data shows that methods meant to improve the standard VaR, like
``expected shortfall'' or ``conditional VaR'' are equally defective
with economic variables--past losses do not predict future losses.
Stress testing is also suspicious because of the subjective nature of
``reasonable stress'' number--we tend to underestimate the magnitude of
outliers. ``Jumps'' are not predictable from past jumps. See Taleb,
N.N. (in press) ``Errors, robustness, and the fourth quadrant,''
International Journal of Forecasting (2009).
---------------------------------------------------------------------------
Thirteen years ago, I warned that ``VaR encourages misdirected
people to take risks with shareholders', and ultimately taxpayers'
money.'' I have since been begging for the suspension of these
measurements of tail risks. But this came a bit late. For the banking
system has lost so far, according to the International Monetary Fund,
in excess of four trillion dollars directly as a result of faulty risk
management. Most of the losses were in the U.S. and will be directly
borne by taxpayers. These losses do not include the other costs of the
economic crisis.
Data shows that banks routinely lose everything earned in their
past history in single blowups--this happened in 1982, 1991, and, of
course now. Every time society bails them out--while bank risk-takers
retain their past bonuses and start the game afresh. This is an
aberrant case of capitalism for the profits and socialism for the
losses.
MAIN PROBLEMS ASSOCIATED WITH VAR-STYLE RISK MEASUREMENT
1. These problems have been obvious all along
My first encounter with the VaR was as a derivatives trader in the
early 1990s when it was first introduced. I saw its underestimation of
the risks of a portfolio by a factor of 100--you set up your book to
lose no more than $100,000 and you take a $10,000,000 hit. Worse, there
was no way to get a handle on how much its underestimation could be.
Using VaR after the crash of 1987 proved strangely gullible. But
the fact that its use was not suspended after the many subsequent major
events, such as the Long-Term Capital Management blowup in 1998,
requires some explanation. Furthermore, regulators started promoting
VaR (Basel 2) just as evidence was mounting against it.\4\
---------------------------------------------------------------------------
\4\ My recollection is that the VaR was not initially taken
seriously by traders and managers. It took a long time for the practice
to spread--and it was only after regulators got involved that it became
widespread.
---------------------------------------------------------------------------
2. VaR is ineffective and lacks in robustness
Alas, we cannot ``measure'' the risk of future rare events like we
measure the temperature. By robustness, I mean that the measure does
not change much if you change the model, technique, or theory. Indeed
risk estimation has nothing to do with the notion of measure. And the
rarer the event, the harder it is to compute its probability--yet the
rarer the event, the larger the consequences.\5\
---------------------------------------------------------------------------
\5\ See Taleb, N.N. and Pilpel, A. (2007) Epistemology and Risk
Management, Risk and Regulation, 13.
---------------------------------------------------------------------------
Furthermore, the type of randomness we have with economic variables
does not have a well-tractable, well-known structure, and can deliver
vastly large events--and we are unable to get a handle on ``how
large.'' Conventional statistics, derived on a different class of
variables, fail us here.\6\,\7\G5,\8\
---------------------------------------------------------------------------
\6\ We are in the worst type of complex system characterized by
high interdependence, low predictability, and vulnerability to extreme
events. See N.N. Taleb, The Black Swan, Random House, 2007.
\7\ There are other problems. 1) VaR does not replicate out of
sample--the past almost never predicts subsequent blowups. (see data in
the Fourth Quadrant). 2) A decrease in VaR does not mean decrease in
risks; often quite the opposite holds, which allows the measure to be
gamed.
\8\ The roots of VaR come from modern financial theory (Markowitz,
Sharpe, Miller, Merton, Scholes) which, in spite of its patent lack of
scientific validity, continues to be taught in business schools. See
Taleb, N.N., (2000), The Black Swan: The Impact of the Highly
Improbable, Random House.
3. VaR encourages ``low volatility, high blowup'' risk taking which can
---------------------------------------------------------------------------
be gamed by the Wall Street bonus structure
Figure 1-A typical ``blow-up'' strategy with hidden risks:
appearance of low volatility, with a high risk of blowup. The trader
makes 11 bonuses, with no subsequent ``clawback'' as losses are borne
by shareholders, then taxpayers. This is the profile for banks (losses
in 1982,1991, and 2008) and many hedge funds. VaR encourages such types
of risk taking.
I have shown that operators like to engage in a ``blow-up''
strategy, (switching risks from visible to hidden), which consists in
producing steady profits for a long time, collecting bonuses, then
losing everything in a single blowup.\9\ Such trades pay extremely well
for the trader--but not for society. For instance, a member of
Citicorp's executive committee (and former government official)
collected $120 million of bonuses over the years of hidden risks before
the blowup; regular taxpayers are financing him retrospectively.
---------------------------------------------------------------------------
\9\ Taleb, N.N. (2004) ``Bleed or Blowup: What Does Empirical
Psychology Tell Us About the Preference For Negative Skewness?,''
Journal of Behavioral Finance, 5.
---------------------------------------------------------------------------
Blowup risks kept increasing over the past few years, while the
appearance of stability has increased.\10\
---------------------------------------------------------------------------
\10\ Even Chairman Bernanke was fooled by the apparent stability as
he pronounced it the ``great moderation.''
---------------------------------------------------------------------------
4. Var has severe side effects (anchoring)
Many people favor the adjunct application of VaR on grounds that it
is ``not harmful,'' using arguments like ``we are aware of its
defects.'' VaR has side effects of increasing risk-taking, even by
those who know that it is not reliable. We have ample evidence of so
called ``anchoring'' \11\ in the calibration of decisions. Information,
even when it is known to be sterile, increases overconfidence.
---------------------------------------------------------------------------
\11\ Numerous experiments provide evidence that professionals are
significantly influenced by numbers that they know to be irrelevant to
their decision, like writing down the last four digits of one's social
security number before making a numerical estimate of potential market
moves. German judges rolling dice before sentencing showed an increase
of 50 percent in the length of the sentence when the dice show a high
number, without being conscious of it. See Birte Englich and Thomas
Mussweiler, ``Sentencing under Uncertainty: Anchoring Effects in the
Courtroom,'' Journal of Applied Social Psychology, Vol. 31, No. 7
(2001), pp. 1535-1551; Birte Englich, Thomas Mussweiler, and Fritz
Strack, ``Playing Dice with Criminal Sentences: the Influence of
Irrelevant Anchors on Experts' Judicial Decision Making,'' Personality
and Social Psychology Bulletin, Vol. 32, No. 2 (Feb. 2006), pp. 188-
200.
5. VaR-style quantitative risk measurement is the engine behind
---------------------------------------------------------------------------
leverage, the main cause of the current crisis
Leverage\12\ is a direct result of underestimation of the risks of
extreme events--and the illusion that these risks are measurable.
Someone more careful (or realistic) would issue equity.
---------------------------------------------------------------------------
\12\ There is a large difference between equity and credit bubbles.
Equity bubbles are benign. We went through an equity bubble in 2000,
without major problems.
Some credit can be benign. Credit that facilitates trade and economic
transactions and finances conservative house-ownership does not have
the same risk properties as credit for speculative reasons resulting
from overconfidence.
April 28, 2004 was a very sad day, when the SEC, at the instigation
of the investment banks, initiated the abandonment of hard (i.e.,
robust) risk measures like leverage, in favor of more model-based
probabilistic, and fragile, ones.
CONCLUSION: WHAT REGULATORY STRUCTURE DO WE NEED?
Regulators should understand that financial markets are a complex
system and work on increasing the robustness in it, by preventing ``too
big to fail'' situations, favoring diversity in risk taking, allowing
entities to absorb large shocks, and reducing the effect of model error
(see ``Ten Points for a Black Swan Robust Society,'' in Appendix II).
This implies reliance on ``hard,'' non-probabilistic measures rather
than more error-prone ones. For instance ``leverage'' is a robust
measures (like the temperature, it does not change with your model),
while VaR is not.
Furthermore, we need to examine the toxicity of models; financial
regulators should have the same test as the Food and Drug
Administration does. The promoter of the probability model must be able
to show that no one will be harmed even if the event is rare. Alas, the
history of medicine shows translational gaps, the lag between the
discovery of harm and suspension of harmful practice, lasting up to 200
years in pre-modern medicine.\13\ Unfortunately, economics resemble
pre-modern medicine.\14\ But we cannot afford to wait 200 years to find
out that the medicine is far worse than the disease. We cannot afford
to wait even months.
---------------------------------------------------------------------------
\13\ ``When William Harvey demonstrated the mechanism of blood
circulation in the 1620s, humoral theory and its related practices
should have disappeared, because the anatomy and physiology on which it
relied was incompatible with this picture of the organism. In fact,
people continued to refer to spirits and humors, and doctors continued
to prescribe phlebotomies, enemas, and cataplasms, for centuries more--
even when it was established in the mid-1800, most notably by Louis
Pasteur, that germs were the cause of disease.'' Noga Arikha ``Just
Life in a Nutshell: Humours as common sense,'' in The Philosophical
Forum Quarterly, XXXIX, 3.
\14\ Most of the use of probabilistic methods lacking both
mathematical and empirical justification can be attributed to the
prestige given to modern finance by the various Nobel memorial prizes
in economics. See P. Triana, 2009, Lecturing Birds on Flying: Can
Mathematical Theories Destroy the Markets?, J. Wiley.
APPENDIX I:
AUTHOR'S WARNINGS, 1996-2007
1996-1997
VaR is charlatanism because it tries to estimate something that is
scientifically impossible to estimate, namely the risk of rare events.
It gives people a misleading sense of precision. (Derivatives Strategy,
citing from Dynamic Hedging)
VaR encourages misdirected people to take risks with shareholders',
and ultimately taxpayers' money. (Derivatives Strategy)
2003
Fannie Mae's models (for calibrating to the risks of rare events)
are pseudoscience. (New York Times--Alex Berenson's article on FNMA)
``What happened to LTCM will look like a picnic compared to what
should happen to you.'' (Lecture, Women in Hedge Funds Association,
cited in Hedge World)
2007
Fannie Mae, when I look at its risks, seems to be sitting on a
barrel of dynamite, vulnerable to the slightest hiccup. But not to
worry: their large staff of scientists deems these events ``unlikely.''
(The Black Swan)
Banks are now more vulnerable to the Black Swan than ever before
with ``scientists'' among their staff taking care of exposures. The
giant firm, J.P. Morgan, put the entire world at risk by introducing in
the nineties RiskMetrics, a phony method aiming at managing people's
risks. A related method called ``Value-at-Risk,'' which relies on the
quantitative measurement of risk, has been spreading. (The Black Swan)
APPENDIX II:
TEN PRINCIPLES FOR A BLACK SWAN
ROBUST WORLD
(FINANCIAL TIMES, APRIL 8, 2009)
1. What is fragile should break early while it is still small. Nothing
should ever become too big to fail. Evolution in economic life helps
those with the maximum amount of hidden risks--and hence the most
fragile--become the biggest.
2. No socialization of losses and privatization of gains. Whatever may
need to be bailed out should be nationalized; whatever does not need a
bail-out should be free, small and risk-bearing. We have managed to
combine the worst of capitalism and socialism. In France in the 1980s,
the socialists took over the banks. In the U.S. in the 2000s, the banks
took over the government. This is surreal.
3. People who were driving a school bus blindfolded (and crashed it)
should never be given a new bus. The economics establishment
(universities, regulators, central bankers, government officials,
various organizations staffed with economists) lost its legitimacy with
the failure of the system. It is irresponsible and foolish to put our
trust in the ability of such experts to get us out of this mess.
Instead, find the smart people whose hands are clean.
4. Do not let someone making an ``incentive'' bonus manage a nuclear
plant--or your financial risks. Odds are he would cut every corner on
safety to show ``profits'' while claiming to be ``conservative.''
Bonuses do not accommodate the hidden risks of blow-ups. It is the
asymmetry of the bonus system that got us here. No incentives without
disincentives: capitalism is about rewards and punishments, not just
rewards.
5. Counter-balance complexity with simplicity. Complexity from
globalization and highly networked economic life needs to be countered
by simplicity in financial products. The complex economy is already a
form of leverage: the leverage of efficiency. Such systems survive
thanks to slack and redundancy; adding debt produces wild and dangerous
gyrations and leaves no room for error. Capitalism cannot avoid fads
and bubbles: equity bubbles (as in 2000) have proved to be mild; debt
bubbles are vicious.
6. Do not give children sticks of dynamite, even if they come with a
warning. Complex derivatives need to be banned because nobody
understands them and few are rational enough to know it. Citizens must
be protected from themselves, from bankers selling them ``hedging''
products, and from gullible regulators who listen to economic
theorists.
7. Only Ponzi schemes should depend on confidence. Governments should
never need to ``restore confidence.'' Cascading rumors are a product of
complex systems. Governments cannot stop the rumors. Simply, we need to
be in a position to shrug off rumors, be robust in the face of them.
8. Do not give an addict more drugs if he has withdrawal pains. Using
leverage to cure the problems of too much leverage is not homeopathy,
it is denial. The debt crisis is not a temporary problem, it is a
structural one. We need rehab.
9. Citizens should not depend on financial assets or fallible
``expert'' advice for their retirement. Economic life should be
definancialized. We should learn not to use markets as storehouses of
value: they do not harbor the certainties that normal citizens require.
Citizens should experience anxiety about their own businesses (which
they control), not their investments (which they do not control).
10. Make an omelet with the broken eggs. Finally, this crisis cannot be
fixed with makeshift repairs, no more than a boat with a rotten hull
can be fixed with ad hoc patches. We need to rebuild the hull with new
(stronger) materials; we will have to remake the system before it does
so itself. Let us move voluntarily into Capitalism 2.0 by helping what
needs to be broken break on its own, converting debt into equity,
marginalizing the economics and business school establishments,
shutting down the ``Nobel'' in economics, banning leveraged buy-outs,
putting bankers where they belong, clawing back the bonuses of those
who got us here, and teaching people to navigate a world with fewer
certainties.
Then we will see an economic life closer to our biological
environment: smaller companies, richer ecology, no leverage. A world in
which entrepreneurs, not bankers, take the risks, and companies are
born and die every day without making the news.
In other words, a place more resistant to black swans.
Biography for Nassim N. Taleb
Nassim N. Taleb is currently Distinguished Professor in Risk
Engineering at New York University Polytechnic Institute and Principal
at Universa Investments. He spent close to 21 years as a senior trader
on Wall Street before becoming a full time scholar. He is a combination
of a scholar of risk and model error, literary essayist, and
derivatives trader. He is known for a multi-disciplinary approach to
the role of the high-impact rare event--across economics, philosophy,
finance, engineering, and history. He also runs experiments on human
errors in the assessment of probabilities of rare events as part of the
Decision Science Laboratory. His current program is to design ways to
live in a world we don't quite understand and help ``robustify'' the
world against the Black Swan.
Taleb is, among other books and research papers, the author of the
NYT Bestseller The Black Swan: The Impact of the Highly Improbable
which was according to The Times as one of the 12 most influential
books since WW-II. His books have close to two and a half million
copies in print in 31 languages.
Taleb has an MBA from Wharton and a Ph.D. from the University of
Paris.
Among other activities, he is currently on the King of Sweden
advisory committee for climate risks and modeling. The British Tory
opposition is using Black Swan thinking as part of their platform.
Chairman Miller. Thank you, Dr. Taleb.
Dr. Bookstaber for five minutes.
STATEMENT OF DR. RICHARD BOOKSTABER, FINANCIAL AUTHOR
Dr. Bookstaber. Mr. Chairman and Members of the Committee,
I thank you for the opportunity to testify today. My oral
testimony will begin with a discussion of the limitations of
VaR. I will then discuss the role of VaR in the recent market
meltdown and conclude with suggestions for filling the gap left
by the limitations of VaR.
The limitations of VaR are readily apparent by looking at
the critical assumptions behind it. For the standard
construction of VaR, these assumptions are, first, that all
portfolio positions are included; secondly, that the sample
history used in VaR is a reasonable representation of things
that are likely to occur going forward; and third, that the
normal distribution function that it uses is a reasonable
representation of the statistical distribution underlying the
returns. These assumptions are often violated, leading VaR
estimates to be misleading. So let me discuss each of these in
turn.
First of all, in terms of incomplete positions, obviously,
for risk to be measured, all the risky positions must be
included in the analysis, but for larger institutions, it is
commonplace for some positions to be excluded. This can happen
because the positions are held off a balance sheet beyond the
purview of those doing the risk analysis, because they are in
complex instruments that have not been sufficiently modeled, or
because they are in new so-called innovative products that have
yet to be added into the risk process. This provides a
compelling reason to have what I call `flight to simplicity' in
financial products, to move away from complex and customized
innovative products and towards standardization.
In terms of unrepresentative sample periods, VaR gives a
measure of risk that assumes tomorrow is drawn from the same
distribution as the sample data used to compute the VaR. If the
future does not look like the past--in particular, if a crisis
emerges, VaR will no longer be a good measure of risk, which is
to say that VaR is a good measure of risk except when it really
matters.
Third, in terms of fat tails and normal distribution,
largely because of crisis events, security returns tend to have
fatter tails than what is represented by a normal distribution.
That is, there tend to be more outliers and extreme events than
a normal distribution would imply. Now, one way to address this
well-known inaccuracy is to modify the distribution allowing
for fatter tails, but this adds complication to VaR analysis
while contributing little insight in terms of risk.
A better approach is to accept the limitations of VaR, and
then try to understand the market crises where VaR fails. If we
understand the dynamics of market crises, we may be able to
improve risk management to make it work when it is of the
greatest importance. A starting point for understanding
financial market crises is leverage and the crowding of trades.
These lead to the common crisis dynamic--what I call a
liquidity crisis cycle. Such a cycle begins when there is some
exogenous shock that causes a drop in a market that is crowded
with leveraged investors. The highly leveraged investors are
forced to sell to meet their margin requirements. Their selling
drops prices further, which in turn forces yet more selling,
resulting in a cascading cycle downward in prices. Now, the
investors that are under pressure discover there is no longer
any liquidity in the stressed market, so they start to
liquidate their positions in other markets to generate the
required margin. And if many investors that are in the first
market also have high exposure in a second one, the downward
spiral propagates to this second market.
This phenomenon explains why a crisis can spread in
surprising and unpredictable ways. The contagion is primarily
driven by what other securities are owned by the funds that
need to sell. For example, a simple example of this is what
happened with the silver bubble back in 1980. The silver market
became closely linked with the market for cattle. Why? Because
the Hunt family had margin calls on their silver position, and
so they sold whatever else they could, and what else they had
to sell happened to be cattle. So thus there was a contagion
based not on any economic linkage but based on who was under
pressure and what else they owned.
Now, this cycle evolves unrelated to historical
relationships, out of the reach of VaR-type models. But that
doesn't mean it is beyond analysis. But if we want to analyze
it, we need to know the leverage and the positions of the major
market participants. Gathering these critical data is the first
step in measuring and managing crisis risk, and should be the
role of a market regulator.
Now, let me talk specifically about the role of VaR in the
current crisis. Whatever the limitations of VaR models, they
were not the key culprits in the case of the multi-billion
dollar write-downs central to the current crisis. The large
bank inventories were there to be seen. You didn't need to have
any models or sophisticated detective or forensic work to see
them. Furthermore, it was clear that these inventories were
illiquid and that their market values were uncertain. It is
hard to understand how this elephant in the room was missed,
how a risk manager could see inventory grow from a few billion
dollars to 10 billion dollars and then to 30 or 40 billion
dollars, and not take action to bring that inventory down.
One has to look beyond VaR to sheer stupidity or collective
management failure. The risk managers missed the growing
inventory, or did not have the courage of their conviction to
insist on its being reduced, or the senior management was not
willing to heed their demands. Whatever the reason, VaR was not
central to the crisis. Focus would be better placed on failures
in risk governance than failures of risk models, whatever the
flaws of VaR are.
Now, in summary, let me first emphasize, I believe that VaR
does have value. If one were forced to pick a single number for
the risk of a portfolio in the near future, VaR would be a good
choice for the job. VaR illuminates most of the risk landscape,
but, unfortunately, the places its light fails to reach are the
canyons, crevices and cliffs.
So we can do two things to try to improve on and address
the limitations of VaR. One is to employ coarser measures of
risk, measures that have fewer assumptions and that are less
dependent on the future looking like the past. The use of the
leverage ratio mandated by U.S. regulators is an example of
such a measure. The leverage ratio does not overlay assumptions
about the correlation or the volatility of the assets, and does
not assume any mitigating effects from diversification. It
does, however, have its own limitations as a basis for capital
adequacy. The second is to add other risk methods that are
better at illuminating the areas VaR does not reach. So in
addition to measuring risk using a standard VaR approach,
develop scenarios for crises and test capital adequacy under
those scenarios. Critical, of course, to the success of this
approach is the ability to ferret out potential crises and
describe them adequately for risk purposes. We can go a long
way toward this goal by having regulators amass and aggregate
data on the positions and leverage of large financial
institutions. These data are critical because we cannot manage
what we cannot measure, and we cannot measure what we cannot
see. With these data, we will be better able to measure the
crowding and leverage that leads to crisis, and shed light on
risks that fail to be illuminated by VaR.
Let me close my oral comments by responding to comments by
both the Chairman and the Ranking Member. The analogy of VaR
and the models related to risk to models used in other
engineering and physical systems--I think there is a critical
distinction between financial systems and other engineering
systems, because financial systems are open to gaming. If I
discover a valve that is poorly designed in a nuclear power
plant and design a new valve to replace it, and install that
valve, the valve doesn't sit there and try to figure out if it
can fool me into thinking it is on when it is really off. But
in the financial markets, that is what happens. So any
engineering solution or any analogy to physical processes is
going to be flawed when they are applied to the financial
markets, because those in the financial markets can game
against the system to try to find ways around any regulation,
and to find other ways to do what they want to do. And I
believe that one of the key tools for this type of gaming are
sophisticated, innovative, complex products that can often
obfuscate what people are doing.
So, I think, parenthetical to the issues of VaR and other
models is, number one, the recognition that no model can work
completely in the financial markets the way they can in other
physical systems, and number two, that if we want to curb or
diminish the issues of gaming, we have to have more simplicity
and transparency in the financial instruments.
Thank you. I look forward to your questions.
[The prepared statement of Dr. Bookstaber follows:]
Prepared Statement of Richard Bookstaber
Mr. Chairman and Members of the Committee, I thank you for the
opportunity to testify today. My name is Richard Bookstaber. Over the
past decade I have worked as the risk manager in two of the world's
largest hedge funds, Moore Capital Management and, most recently,
Bridgewater Associates. In the 1990s I oversaw firm-wide risk at
Salomon Brothers, which at the time was the largest risk-taking firm in
the world, and before that was in charge of market risk at Morgan
Stanley.
I am the author of A Demon of Our Own Design--Markets, Hedge Funds,
and the Perils of Financial Innovation. Published in April, 2007, this
book warned of the potential for financial crisis resulting from the
growth of leverage and the proliferation of derivatives and other
innovative products.
Although I have extensive experience on both the buy-side and sell-
side, I left my position at Bridgewater Associates at the end of 2008,
and come before the Committee in an unaffiliated capacity, representing
no industry interests.
My testimony will discuss what VaR is, how it can be used and more
importantly, how it can be misused. I will focus on the limitations of
VaR in measuring crisis risk. I will then discuss the role of VaR in
the recent market meltdown, concluding with suggestions for ways to
fill the gaps left by the limitations of VaR.
What is VaR?
VaR, or Value-at-Risk, measures the risk of a portfolio of assets
by estimating the probability that a given loss might occur. For
example, the dollar VaR for a particular portfolio might be expressed
as ``there is a ten percent probability that this portfolio will lose
more than $VaR over the next day.''
Here is a simplified version of the steps in constructing a VaR
estimate for the potential loss at the ten percent level:
1. Identify all of the positions held by the portfolio.
2. Get the daily returns for each of these positions for the
past 250 trading days (about a one-year period).
3. Use those returns to construct the return to the overall
portfolio for each day over the last 250 trading days.
4. Order the returns for those days from the highest to the
lowest, and pick the return for the day that is the 25th worst
day's return. That will be a raw estimate of the daily VaR at
the ten percent level.
5. Smooth the results by fitting this set of returns to the
Normal distribution function.\1\
---------------------------------------------------------------------------
\1\ The risk for a Normal distribution is fully defined by the
standard deviation, and the results from Step 3 can be used to estimate
the standard deviation of the sample. If the estimated standard
deviation is, say, five percent, then the VaR at the ten percent level
will be a loss of eight percent. For a Normal distribution the ten
percent level is approximately 1.6 standard deviations.
Limitations of VaR
The critical assumptions behind the construction of VaR are made
clear by the process described above:
1. All of the portfolio positions are included.
2. The sample history is a reasonable representation of what
things will look like going forward.
3. The Normal distribution function is a reasonable
representation of the statistical distribution underlying the
returns.
The limitations to VaR boil down to issues with these three
assumptions, assumptions that are often violated, leading VaR estimates
to be misleading.
Incomplete positions
Obviously, risk cannot be fully represented if not all of the risky
positions are included in the analysis. But for larger institutions, it
is commonplace for this to occur. Positions might be excluded because
they are held off-balance sheet, beyond the purview of those doing the
risk analysis; they might be in complex instruments that have not been
sufficiently modeled or that are difficult to include in the position
database; or they might be in new products that have not yet been
included in the risk process. In the recent crisis, some banks failed
to include positions in collateralized debt obligations (CDOs) for all
three of these reasons.\2\ And that exclusion was not considered an
immediate concern because they were believed to be low risk, having
attained a AAA rating.
---------------------------------------------------------------------------
\2\ Regulatory capital on the trading assets that a bank does not
include in VaR--or for which the bank's VaR model does not pass
regulatory scrutiny--is computed using a risk-rating based approach.
However, the rating process itself suffers from many of the
difficulties associated with calculating VaR, as illustrated by the AAA
ratings assigned to many mortgage-backed CDOs and the consequent severe
underestimation of the capital required to support those assets.
---------------------------------------------------------------------------
The inability to include all of the positions in the VaR risk
analysis, the most rudimentary step for VaR to be useful, is pervasive
among the larger institutions in the industry. This provides a
compelling reason to have a `flight to simplicity' in financial
products, to move away from complex and customized innovative products
and toward standardization.\3\
---------------------------------------------------------------------------
\3\ I discuss the complexity and related risk issues surrounding
derivatives and related innovative products in Testimony of Richard
Bookstaber, Submitted to the Senate of the United States, Committee on
Agriculture, Nutrition, and Forestry for the Hearing: ``Regulatory
Reform and the Derivatives Markets,'' June 4, 2009.
Unrepresentative sample period
VaR gives a measure of risk that assumes tomorrow is drawn from the
same distribution as the sample data used to compute the VaR. If the
future does not look like the past, in particular if a crisis emerges,
then VaR will no longer be a good measure of risk.\4\ Which is to say
that VaR is a good measure of risk except when it really matters.\5\
---------------------------------------------------------------------------
\4\ One way to try to overcome the problem of relying on the past
is to use a very long time period in the VaR calculation, with the idea
that a longer period will include many different regimes, crises and
relationships. Such a view misses the way different regimes,
essentially different distributions, mix to lead to a final result. A
long time period gives muddied results. To see this, imagine the case
where in half of the past two assets were strongly positively
correlated and the other half they were strongly negatively correlated.
The mixing of the two would suggest the average of little correlation,
thus giving a risk posture that did not exist in either period, but
that also incorrectly suggests diversification opportunities.
\5\ As a corollary to this, one could also say that diversification
works except when it really matters.
---------------------------------------------------------------------------
It is well known that VaR cannot measure crisis risk. During
periods of crisis the relationship between securities changes in
strange and seemingly unpredictable ways. VaR, which depends critically
on a set structure for volatility and correlation, cannot provide
useful information in this situation. It contains no mechanism for
predicting the type of crisis that might occur, and does not consider
the dynamics of market crises. This is not to say that VaR has no value
or is hopelessly flawed. Most of the time it will provide a reasonable
measure of risk--indeed the vast majority of the time this will be the
case. If one were forced to pick a single number for the risk of a
portfolio in the near future, VaR would be a good choice for the job.
VaR illuminates most of the risk landscape. But unfortunately, the
places its light fails to reach are the canyons, crevices and cliffs.
Fat Tails and the Normal Distribution
Largely because of crisis events, security returns tend to have
fatter tails than what is represented by a Normal distribution. That
is, there tend to be more outliers and extreme events than what a
Normal distribution would predict. This leads to justifiable criticism
of VaR for its use of the Normal distribution. However, sometimes this
criticism is overzealous, suggesting that the professionals who assume
a Normal distribution in their analysis are poorly trained or worse.
Such criticism is unwarranted; the limitations of the Normal
distribution are well-known. I do not know of anyone working in
financial risk management, or indeed in quantitative finance generally,
who does not recognize that security returns may have fat tails. It is
even discussed in many investment textbooks, so it is a point that is
hard to miss.\6\
---------------------------------------------------------------------------
\6\ For example, Investments, by Bodie, Kane, and Marcus, 8th
edition (McGraw-Hill/Irwin), has a section (page 148) entitled
``Measurement of Risk with Non-normal Distributions.''
---------------------------------------------------------------------------
The issue is how this well-known inaccuracy of the Normal
distribution is addressed. One way is knowingly to misuse VaR, to
ignore the problem and act as if VaR can do what it cannot. Another is
to modify the distribution to allow for fatter tails.\7\ This adds
complication and obfuscation to the VaR analysis, because any approach
employing a fat-tailed distribution increases the number of parameters
to estimate, and this increases the chance that the distribution will
be mis-specified. And in any case, simply fattening up the tails of the
distribution provides little insight for risk management.
---------------------------------------------------------------------------
\7\ Extreme value theory is the bastion for techniques that employ
distributions with a higher probability of extreme events.
---------------------------------------------------------------------------
I remember a cartoon that showed a man sitting behind a desk with a
name plate that read `Risk Manager.' The man sitting in front of the
desk said, ``Be careful? That's all you can tell me, is to be
careful?'' Stopping with the observation that extreme events can occur
in the markets and redrawing the distribution accordingly is about as
useful as saying ``be careful.'' A better approach is to accept the
limitations of VaR, and then try to understand the nature of the
extreme events, the market crises where VaR fails. If we understand the
dynamics of market crisis, we may be able to improve risk management to
make it work when it is of the greatest importance.
Understanding the Dynamics of Market Crises
A starting point for understanding financial market crises is
leverage and the crowding of trades, both of which have effects that
lead to a common crisis dynamic, the liquidity crisis cycle.
Such a cycle begins when an exogenous shock causes a drop in a
market that is crowded with leveraged investors. The highly leveraged
investors are forced to sell to meet their margin requirements. Their
selling drops prices further, which in turn forces yet more selling,
resulting in a cascading cycle downward in prices. Those investors that
are under pressure discover there is no longer liquidity in the
stressed market, so they start to liquidate their positions in other
markets to generate the required margin. If many of the investors that
are in the first market also have high exposure in a second one, the
downward spiral propagates to this second market.\8\
---------------------------------------------------------------------------
\8\ The use of VaR-based capital can actually contribute to this
sort of cycle. VaR will increase because of the higher volatility--and
also possibly because of the higher correlations--leading potential
liquidity providers and lenders to pull back. This was a likely
exacerbating effect during the 1997 Asian crisis.
---------------------------------------------------------------------------
This phenomenon explains why a crisis can spread in surprising and
unpredictable ways. The contagion is driven primarily by what other
securities are owned by the funds that need to sell.\9\ For example,
when the silver bubble burst in 1980, the silver market became closely
linked to the market for cattle. Why? Because when the Hunt family had
to meet margin calls on their silver positions, they sold whatever else
they could. And they happened also to be invested in cattle. Thus there
is contagion based not on economic linkages, but based on who is under
pressure and what else they are holding.
---------------------------------------------------------------------------
\9\ As an illustration, the proximate cause of Long Term Capital
Management's (LTCM's) demise was the Russian default in August, 1998.
But LTCM was not highly exposed to Russia. A reasonable risk manager,
aware of the Russian risks, might not have viewed it as critical to
LTCM. But the Russian default hurt LTCM because many of those who did
have high leverage in Russia also had positions in other markets where
LTCM was leveraged. When the Russian debt markets failed and these
investors had to come up with capital, they sold their more liquid
positions in, among other things, Danish mortgage bonds. So the Danish
mortgage bond market and these other markets went into a tail spin, and
because LTCM was heavily exposed in these markets, the contagion took
LTCM with it.
---------------------------------------------------------------------------
This cycle evolves unrelated to historical relationships, out of
the reach of VaR-type models. But that does not mean it is beyond
analysis. Granted it is not easy to trace the risk of these potential
liquidity crisis cycles. To do so with accuracy, we need to know the
leverage and positions of the major market participants. No one firm,
knowing only its own positions, can have an accurate assessment of the
crisis risk. Indeed, each firm might be managing its risk prudently
given the information it has at its disposal, and not only miss the
risk that comes from crowding and leverage, but also unwittingly
contribute to this risk. Gathering these critical data is the first
step in measuring and managing crisis risk. This should be the role of
a market regulator.\10\
---------------------------------------------------------------------------
\10\ I discuss the need for firm-level position and leverage data
in crisis risk management in previous testimony before both the House
and the Senate. For example, Testimony of Richard Bookstaber, Submitted
to the Congress of the United States, House Financial Services
Committee, for the Hearing: ``Systemic Risk: Examining Regulators
Ability to Respond to Threats to the Financial System,'' October 2,
2007, and Testimony of Richard Bookstaber, Submitted to the Senate of
the United States, Senate Banking, Housing and Urban Affairs
Subcommittee on Securities, Insurance and Investment, for the Hearing:
``Risk Management and Its Implications for Systematic Risk,'' June 19,
2008.
The Role of VaR in the Current Crisis
The above discussion provides part of the answer to the question of
the role of VaR in the current market crisis: If VaR was used as the
source of risk measurement, and thus as the determinant of risk
capital, then it missed the potential for the current crisis for the
simple reason that VaR is not constructed to deal with crisis risk. And
if VaR was applied as if it actually reflected the potential for
crisis, that is, if it was forgotten that VaR is only useful insofar as
the future is drawn from the same distribution as the past, then this
led to the mis-measurement of risk. So if VaR was the sole means of
determining risk levels and risk capital coming into this crisis, it
was misused. But this does not present the full story.
Whatever the limitations of VaR models, they were not the key
culprits in the case of the multi-billion dollar write-downs during the
crisis. The large bank inventories were there to be seen; no models or
detective work were needed. Furthermore, it was clear the inventories
were illiquid and their market values uncertain.\11\ It is hard to
understand how this elephant in the room was missed, how a risk manager
could see inventory grow from a few billion dollars to ten billion
dollars and then to thirty or forty billion dollars and not react by
forcing that inventory to be brought down.
---------------------------------------------------------------------------
\11\ This is especially true when one considers the business of the
banks, which is to package the securities and sell them. The growth of
inventory was outside the normal business of the banks. That the
securities were not moving out the door should have been an immediate
indication they were not correctly priced.
---------------------------------------------------------------------------
Of course, if these inventories were not properly included in the
VaR analysis, the risk embodied by these positions would have been
missed, but one has to look beyond VaR, to culprits such as sheer
stupidity or collective management failure: The risk managers missed
the growing inventory, or did not have the courage of their conviction
to insist on its reduction, or the senior management was not willing to
heed their demands. Whichever the reason, VaR was not central to this
crisis.\12\ Focus would be better placed on failures in risk governance
than failures of risk models.
---------------------------------------------------------------------------
\12\ Indeed, in some important cases, VaR was not even employed in
the risk process. A case in point is the `super senior' mortgage CDO
positions which caused huge trading losses at a number of banks. There
is a common misconception that regulatory capital for trading assets is
automatically computed using VaR. In fact, trading assets are eligible
for VaR-based capital only if the bank can demonstrate to its
supervisor that its model is robust. Absent this, a coarser method is
applied. Many of the highly complex securities at the heart of the
recent crisis were not regarded as being suitable for VaR treatment,
and received a simpler ratings-based treatment, which proved to
severely underestimate the capital required to support the assets.
Summary: VaR and Crisis Risk
There are two approaches for moving away from over-reliance on VaR.
The first approach is to employ coarser measures of risk, measures
that have fewer assumptions and that are less dependent on the future
looking like the past.\13\ The use of the Leverage Ratio mandated by
U.S. regulators and championed by the FDIC is an example of such a
measure.\14\ The leverage ratio does not overlay assumptions about the
correlation or the volatility of the assets, and does not assume any
mitigating effect from diversification, although it has its own
limitations as a basis for capital adequacy.\15\
---------------------------------------------------------------------------
\13\ I believe coarse measures--measures that are not fine tuned to
be ideal in any one environment, but are robust across many
environments--are a key to good risk management.
\14\ The Leverage Ratio is the ratio of Tier 1 capital, principally
equity and retained earnings, to total assets.
\15\ The Leverage Ratio is inconsistent with Basel II because it is
not sensitive to the riskiness of balance sheet assets and it does not
capture off-balance sheet risks. By not taking the relative risk of
assets into account, it could lead to incentives for banks to hold
riskier assets, while on a relative basis penalizing those banks that
elect to hold a low-risk balance sheet. In terms of risk to a financial
institution, the time horizon of leverage is also important, which the
Leverage Ratio also misses. The problems with Bear Stearns and Lehman
was not only one of leverage per se, but of funding a sizable portion
of leverage in the short-term repo market. They thus were vulnerable to
funding drying up in the face of a crisis.
---------------------------------------------------------------------------
The second approach is to recognize that while VaR provides a guide
to risk in some situations, it must be enhanced with other measures
that are better at illuminating the areas it does not reach. For
example, Pillar II of Basel II has moved to include stress cases for
crises and defaults into its risk capital process. So in addition to
measuring risk using a standard VaR approach, firms must develop
scenarios for crises and test their capital adequacy under those
scenarios. Critical to the success of this approach is the ability to
ferret out potential crises and describe them adequately for risk
purposes.
This means that for crisis-related stress testing to be feasible,
we first must believe that it is indeed possible to model financial
crisis scenarios, i.e., that crises are not `black swans.' This is not
to say that surprises do not occur. Though recently popularized, the
recognition that we are beset by unanticipatable risk, by events that
seemingly come out of nowhere and catch us unawares, has a long history
in economics and finance, dating back to Frank Knight in the 1920s.\16\
The best defense against such risks is to maintain a coarse, simple and
robust financial structure. Rather than fine-tuning for the current
environments, we need risk measures and financial instruments which,
while perhaps not optimal for the world of today, will be able to
operate reasonably if the world changes in unexpected ways. VaR as
currently structured is not such a risk measure.
---------------------------------------------------------------------------
\16\ Knight makes the distinction between risks we can identify and
measure and those that are unanticipatable and therefore not measurable
in Risk, Uncertainty, and Profit. (1921), Boston, MA: Houghton Mifflin
Company.
---------------------------------------------------------------------------
However, although surprises do occur, crisis scenarios are not
wholly unanticipatable; they are not in the realm of Knightian
uncertainty. We have had ample experience with financial crises. We
know a thing or two about them.\17\ And we can further anticipate
crisis risk by amassing data on the positions and leverage of the large
investment firms. The regulator is best suited to take on this task,
because these are data that no one firm can or should fully see.\18\
With these critical data we will be better able to measure the crowding
and leverage that lead to liquidity crisis cycles and begin to shed
light on the areas of financial risk that fail to be illuminated by
VaR.\19\
---------------------------------------------------------------------------
\17\ For example, even beyond the insights to be gained from a
detailed knowledge of firm-by-firm leverage and market crowding, there
are some characteristics of market crisis that can be placed into a
general scenario. When a crisis occurs, equity prices drop, credit
spreads rise, and the volatility of asset returns increases. The yield
curve flattens and gold prices rise. Furthermore, the correlation
between individual equities rises, as does the correlation between
equities and corporate bonds. The riskier and less liquid assets fare
more poorly, so, for example, emerging markets take a differentially
bigger hit than their G-7 cousins. More broadly, anything that is risky
or less liquid becomes more common and negative in its return; the
subtleties of pricing between assets becomes overshadowed by the
assets' riskiness. However, short-term interest rates and commodity
prices are less predictable; in some cases, such as in the case of the
inflation-laden crisis of 1973-1974, they rise, while in other cases,
such as in the current crisis, they drop.
Each of these effects can occur with a ferocity far beyond what is seen
in normal times, so if these crisis events are overlaid on the
distribution coming out of the VaR model based on those normal times
one will come away saying the crisis is a 100-year flood event, a
twenty standard deviation event, a black swan. But it is none of these
things. It is a financial crisis, and such crises occur frequently
enough that to be understood without such shock and awe.
---------------------------------------------------------------------------
\18\ Financial firms will be justifiably reticent to have their
position and leverage information made public, so the collection and
analysis of the data will have to reside securely in the regulator.
\19\ With these data, the regulator is also in a position to run
risk analysis independent of the firms. Under Basel II, the regulator
still depends on the internal processes of the banks for the
measurement of risk and the resulting capital requirements.
---------------------------------------------------------------------------
Appendix
Related Blog Posts on VaR and Risk Management
The Fat-Tailed Straw Man
See http://rick.bookstaber.com/2009/03/fat-tailed-straw-man.html
My Time article about the quant meltdown of August, 2007 started
with ``Looks like Wall Street's mad scientists have blown up the lab
again.'' Articles on Wall Street's mad scientist blowing up the lab
seem to come out every month in one major publication or another. The
New York Times has a story along these lines today and had a similar
story in January.
There is a constant theme in these articles, invariably including a
quote from Nassim Taleb, that quants generally, and quantitative risk
managers specifically, missed the boat by thinking, despite all
evidence to the contrary, that security returns can be modeled by a
Normal distribution.
This is a straw man argument. It is an attack on something that no
one believes.
Is there anyone well trained in quantitative methods working on
Wall Street who does not know that security returns have fat tails? It
is discussed in most every investment text book. Fat tails are
apparent--even if we ignore periods of crisis--in daily return series.
And historically, every year there is some market or other that has
suffered a ten standard deviation move of the ``where did that come
from'' variety. I am firmly in the camp of those who understand there
are unanticipatable risks; as far back as an article I co-authored in
1985, I have argued for the need to recognize that we face uncertainty
from the unforeseeable. To get an idea of how far back the appreciation
of this sort of risk goes in economic thought, consider the fact that
it is sometimes referred to as Knightian uncertainty.
Is there any risk manager who does not understand that VaR will not
capture the risk of market crises and regime changes? The conventional
VaR methods are based on historical data, and so will only be an
accurate view of risk if tomorrow is drawn from the same population as
the sample it uses. VaR is not perfect, it cannot do everything. But if
we understand its flaws--and every professional risk manager does--then
it is a useful guide for day-to-day market risk. If you want to add fat
tails, fine. But as I will explain below, that is not the solution.
So, then, why is there so much currency given to a criticism of
something that no one believes in the first place?
It is because quant methods sometimes fail. We can quibble with
whether `sometimes' should be replaced with `often' or `frequently' or
`every now and again,' but we all know they are not perfect. We are
not, after all, talking about physics, about timeless and universal
laws of the universe when we deal with securities. Weird stuff happens.
And the place where the imperfection is most telling is in risk
management.
When the risk manager misses the equivalent of a force five
hurricane, we ask what is wrong with his methods. By definition, what
he missed was a ten or twenty standard deviation event, so we tell him
he ignored fat tails. There you have it, you failed because you did not
incorporate fat tails. This is tautological. If I miss a large risk--
which will occur on occasion even if I am fully competent; that is why
they are called risks--I will have failed to account for a fat tailed
event. I can tell you that ahead of time. I can tell you now--as can
everyone in risk management--that I will miss something. If after the
fact you want to castigate me for not incorporating sufficiently fat
tailed events, let the flogging begin.
I remember a cartoon that showed a man sitting behind a desk with a
name plate that read `risk manager.' The man sitting in front of the
desk said, ``Be careful? That's all you can tell me, is to be
careful?'' Observing that extreme events can occur in the markets is
about as useful as saying ``be careful.'' We all know they will occur.
And once they have occurred, we will all kick ourselves and our risk
managers and our models, and ask ``how could we have missed that?''
The flaw comes in the way we answer that question, a question that
can be stated more analytically as ``what are the dynamics of the
market that we failed to incorporate.'' If we answer by throwing our
hands into the air and saying, ``well, who knows, I guess that was one
of them there ten standard deviation events,'' or ``what do you expect;
that's fat tails for you,'' we will be in the same place when the next
crisis arrives. If instead we build our models with fatter and fatter
tailed distributions, so that after the event we can say, ``see, what
did I tell you, there was one of those fat tailed events that I
postulated in my model,'' or ``see, I told you to be careful,'' does
that count for progress?
So, to recap, we all know that there are fat tails; it doesn't do
any good to state the mantra over and over again that securities do not
follow a Normal distribution. Really, we all get it. We should be
constructive in trying to move risk management beyond the point of
simply noting that there are fat tails, beyond admonitions like ``hey,
you know, shit happens, so be careful.'' And that means understanding
the dynamics that create the fat tails, in particular, that lead to
market crisis and unexpected linkages between markets.
What are these dynamics?
One of them, which I have written about repeatedly, is the
liquidity crisis cycle. An exogenous shock occurs in a highly leveraged
market, and the resulting forced selling leads to a cascading cycle
downward in prices. This then propagates to other markets as those who
need to liquidate find the market that is under pressure no longer can
support their liquidity needs. Thus there is contagion based not on
economic linkages, but based on who is under pressure and what else
they are holding. This cycle evolves unrelated to historical
relationships, out of the reach of VaR-types of models, but that does
not mean it is beyond analysis.
Granted it is not easy to trace the risk of these potential
liquidity crisis cycles. To do so with accuracy, we need to know the
leverage and positions of the market participants. In my previous post,
``Mapping the Market Genome,'' I argued that this should be the role of
a market regulator. But even absent that level of detail, perhaps we
can get some information indirectly from looking at market flows.
No doubt there are other dynamics that lead to the fat tailed
events currently frustrating our efforts to manage risk in the face of
market crises. We need to move beyond the fat-tail critiques and the
`be careful' mantra to discover and analyze them.
The Myth of Non-correlation
See http://rick.bookstaber.com/2007/09/myth-of-noncorrelation.html
[This is a modified version of an article I wrote that appeared in
the September, 2007 issue of Institutional Investor.]
With the collapse of the U.S. sub-prime market and the after-shocks
that have been felt in credit and equity markets, there has been a lot
of talk about fat tails, 20 standard deviation moves and 100-year
event. We seem to hear such descriptions fairly frequently, which
suggests that maybe all the talk isn't really about 100-year events
after all. Maybe it is more a reflection of investors' market views
than it is of market reality.
No market veteran should be surprised to see periods when
securities prices move violently. The recent rise in credit spreads is
nothing compared to what happened in 1998 leading up to and following
the collapse of hedge fund Long-Term Capital Management or, for that
matter, during the junk bond crisis earlier that decade, when spreads
quadrupled.
What catches many investors off guard and leads them to make the
``100 year'' sort of comment is not the behavior of individual markets,
but the concurrent big and unexpected moves among markets. It's the
surprising linkages that suddenly appear between markets that should
not have much to do with one other and the failed linkages between
those that should march in tandem. That is, investors are not as
dumbfounded when volatility skyrockets as when correlations go awry.
This may be because investors depend on correlation for hedging and
diversifying. And nothing hurts more than to think you are well hedged
and then to discover you are not hedged at all.
Surprising Market Linkages
Correlations between markets, however, can shift wildly and in
unanticipated ways--and usually at the worst possible time, when there
is a crisis with volatility that is out of hand. To see this, think
back on some of the unexpected correlations that have haunted us in
earlier market crises:
The 1987 stock market crash. During the crash, Wall
Street junk bond trading desks that had been using Treasury
bonds as a hedge were surprised to find that their junk bonds
tanked while Treasuries strengthened. They had the double
whammy of losing on the junk bond inventory and on the hedge as
well. The reason for this is easy to see in retrospect:
Investors started to look at junk bonds more as stock-like risk
than as interest rate vehicles while Treasuries became a safe
haven during the flight to quality and so were bid up.
The 1997 Asian crisis. The financial crisis that
started in July 1997 with the collapse of the Thai baht sank
equity markets across Asia and ended up enveloping Brazil as
well. Emerging-markets fund managers who thought they had
diversified portfolios--and might have inched up their risk
accordingly--found themselves losing on all fronts. The reason
was not that these markets had suddenly become economically
linked with Brazil, but rather that the banks that were in the
middle of the crisis, and that were being forced to reduce
leverage, could not do so effectively in the illiquid Asian
markets, so they sold off other assets, including sizable
holdings in Brazil.
The fall of Long-Term Capital Management in 1998.
When the LTCM crisis hit, volatility shot up everywhere, as
would be expected. Everywhere, that is, but Germany. There, the
implied volatility dropped to near historical lows. Not
coincidentally, it was in Germany that LTCM and others had
sizable long volatility bets; as they closed out of those
positions, the derivatives they held dropped in price, and the
implied volatility thus dropped as well. Chalk one up for the
adage that markets move to inflict the most pain.
And now we get to the crazy markets of August 2007. Stresses in a
minor part of the mortgage market--so minor that Federal Reserve Board
Chairman Ben Bernanke testified before Congress in March that the
impact of the problem had been ``moderate''--break out not only to
affect other mortgages but also to widen credit spreads worldwide. And
from there, sub-prime somehow links to the equity markets. Stock market
volatility doubles, the major indexes tumble by 10 percent and, most
improbable of all, a host of quantitative equity hedge funds--which use
computer models to try scrupulously to be market neutral--are hit by a
``100-year'' event.
When we see this sort of thing happening, our not very helpful
reaction is to shake our heads as if we are looking over a fender
bender and point the finger at statistical anomalies like fat tails,
100-year events, black swans, or whatever. This doesn't add much to the
discourse or to our ultimate understanding. It is just more
sophisticated ways of saying we just lost a lot of money and were
caught by surprise. Instead of simply stating the obvious, that big and
unanticipated events occur, we need to try to understand the source of
these surprising events. I believe that the unexpected shifts in
correlation are caused by the same elements I point to in my book as
the major cause of market crises: complexity and tight coupling.
Complexity
Complexity means that an event can propagate in nonlinear and
unanticipated ways. An example of a complex system from the realm of
engineering is the operation of a nuclear power plant, where a minor
event like a clogged pressure-release valve (as occurred at Three Mile
Island) or a shift in the combination of steam production and fuel
temperature (as at Chernobyl) can cascade into a meltdown.
For financial markets, complexity is spelled d-e-r-i-v-a-t-i-v-e-s.
Many derivatives have nonlinear payoffs, so that a small move in the
market might lead to a small move in the price of the derivative in one
instance and to a much larger move in the price in another. Many
derivatives also lead to unexpected and sometimes unnatural linkages
between instruments and markets. Thanks to collateralized debt
obligations, this is what is at the root of the first leg of the
contagion we observed from the sub-prime market. Sub-primes were
included in various CDOs, as were other types of mortgages and
corporate bonds. Like a kid who brings his cold to a birthday party,
the sickly sub-prime mortgages mingled with these other instruments.
The result can be unexpected higher correlation. Investors that
have to reduce their derivatives exposure or hedge their exposure by
taking positions in the underlying bonds will look at them as part of a
CDO. It doesn't matter if one of the underlying bonds is issued by a
AA-rated energy company and another by a BB financial; the bonds in a
given package will move in lockstep. And although sub-prime happens to
be the culprit this time around, any one of the markets involved in the
CDO packaging could have started things off.
Tight Coupling
Tight coupling is a term I have borrowed from systems engineering.
A tightly coupled process progresses from one stage to the next with no
opportunity to intervene. If things are moving out of control, you
can't pull an emergency lever and stop the process while a committee
convenes to analyze the situation. Examples of tightly coupled
processes include a space shuttle launch, a nuclear power plant moving
toward criticality and even something as prosaic as bread baking.
In financial markets tight coupling comes from the feedback between
mechanistic trading, price changes and subsequent trading based on the
price changes. The mechanistic trading can result from a computer-based
program or contractual requirements to reduce leverage when things turn
bad.
In the '87 crash tight coupling arose from the computer-based
trading of those running portfolio insurance programs. On Monday,
October 19, in response to a nearly 10 percent drop in the U.S. market
the previous week, these programs triggered a flood of trades to sell
futures to increase the hedge. As those trades hit the market, prices
dropped, feeding back to the computers, which ordered yet more rounds
of trading.
More commonly, tight coupling comes from leverage. When things
start to go badly for a highly leveraged fund and its collateral drops
to the point that it no longer has enough assets to meet margin calls,
its manager has to start selling assets. This drops prices, so the
collateral declines further, forcing yet more sales. The resulting
downward cycle is exactly what we saw with the demise of LTCM.
And it gets worse. Just like complexity, the tight coupling born of
leverage can lead to surprising linkages between markets. High leverage
in one market can end up devastating another, unrelated, perfectly
healthy market. This happens when a market under stress becomes
illiquid and fund managers must look to other markets: If you can't
sell what you want to sell, you sell what you can. This puts pressure
on markets that have nothing to do with the original problem, other
than that they happened to be home to securities held by a fund in
trouble. Now other highly leveraged funds with similar exposure in
these markets are forced to sell, and the cycle continues. This may be
how the sub-prime mess expanded beyond mortgages and credit markets to
end up stressing quantitative equity hedge funds, funds that had
nothing to do with sub-prime mortgages.
All of this means that investors cannot put too much stock in
correlations. If you depend on diversification or hedges to keep risks
under control, then when it matters most it may not work.
Biography for Richard Bookstaber
Richard Bookstaber has worked in some of the largest buy-side and
sell-side firms, in capacities ranging from risk management to
portfolio management to derivatives research.
Over the past decade he has worked as a risk manager at Bridgewater
Associates in Westport, Connecticut, Moore Capital Management and Ziff
Brothers Investments. He also ran the FrontPoint Quantitative Fund, a
market neutral long/short equity fund, at FrontPoint Partners.
From 1994 through 1998, Mr. Bookstaber was the Managing Director in
charge of firm-wide risk management at Salomon Brothers. In this role
he oversaw both the client and proprietary risk-taking activities of
the firm, and served on that firm's powerful Risk Management Committee.
He remained in these positions at Salomon Smith Barney after the firm's
purchase by Traveler's and the merger that formed Citigroup.
Before joining Salomon, Mr. Bookstaber spent ten years at Morgan
Stanley in quantitative research and as a proprietary trader. He also
marketed and managed portfolio hedging programs as a fiduciary at
Morgan Stanley Asset Management. With the creation of Morgan Stanley's
risk management division, he was appointed as the Firm's first Director
of Market Risk Management.
He is the author of four books and scores of articles on finance
topics ranging from option theory to risk management. He has received
various awards for his research, including the Graham and Dodd Scroll
from the Financial Analysts Federation and the Roger F. Murray Award
from the Institute of Quantitative Research in Finance.
Mr. Bookstaber's most recent book is A Demon of Our Own Design--
Markets, Hedge Funds and the Perils of Financial Innovation (Wiley,
2007).
He received a Ph.D. in economics from MIT.
Discussion
Chairman Miller. Thank you very much. We will now have
rounds of questions of five minutes for each Member, and I will
begin by recognizing myself for five minutes.
Can Economic Events Be Predicted?
Dr. Bookstaber, what you just described, what I have heard
you describe as gaming, I have heard celebrated on the
Financial Services Committee, on which I also serve, as
innovation--that a lot of innovation seems to be simply a way
to evade existing regulations. And I think both of you got it--
Dr. Bookstaber, in that last bit of testimony you certainly got
at it, but the supporters of the VaR now they say want a do-
over, that the VaR model was perhaps flawed but it can be
fixed, and they can now develop a more reliable model that will
predict fat tail events, the unlikely events. Do you think that
it is a failure of that model, or do you think the failure is
in the idea that economic events can be predicted with the same
precision that the movement of the planets can be predicted? Do
you think that it is inherently flawed to think that we can
develop models that will be unfailingly reliable? Dr. Taleb.
Dr. Taleb. This is my life story. From the beginning--and I
heard, Dr. Bookstaber and I share a lot of opinions, you know,
on things like gaming, like the numbers that are going to be
gamed on the interaction between model and participants.
However, there are two things or three things that I heavily
disagree with, and the first one is, he said that we can use
different distribution to model tail events. Well, that is the
story of my life. This is why I provided this paper forthcoming
in which I look at 20 million pieces of data, every single
economic variable I could find, and tried to see if there is
regularity in the data helping to predict itself, you know,
outside that sample from which it was derived. Unfortunately,
it is impossible, and that is my first argument, that the more
remote the event, the less we can predict it, and that's my
first point. And the second one is, we know which variables are
more unpredictable than others, and it is very easy to protect
against that. And the third one is that I agree with Dr.
Bookstaber; if I were, you know, an omnipotent person seeing
all the leverage and everything in the system, and equipped
with heavy, you know, equations, I could probably figure it
out. However, this is Soviet-style thinking, that someone, some
regulator, some unit out there can see what is going on and be
able to model it, because unfortunately when we model in
complex systems, we have non-linearity. Even if I gave you all
the data and you missed something by $1 million, okay--your
probabilities will change markedly.
Chairman Miller. I will get to you, Dr. Bookstaber, but
your solution then is just higher liquidity requirements?
Dr. Taleb. No, my solution is figuring out--it is very
simple. I was a trader in the 1980s. There were some products
we could really risk manage on a napkin. Options, instruments,
futures, all these we could risk manage on a napkin. Once we
started having these toxic products--to me, the sole purpose of
these products is to create bonuses, like complex derivatives.
I was a complex derivatives trader. I have a textbook on
complex derivatives, and I tell you, these products, okay, can
hide massive amounts of tail risks. They are not needed for
anyone. A lot of these products should not be there. If you
eliminate some of the products, some of the exposure, it would
not change anything to economic life and it would make things a
lot more measurable. So my solution is to ban some products
that have a toxic exposure to tail events.
Chairman Miller. Dr. Bookstaber.
Dr. Bookstaber. Let me just correct one point. I do not
advocate trying to fix VaR by fattening the tails. I am simply
arguing that some people make that as a suggestion. I think VaR
is what it is, it does what it does, and the best thing to do
is recognize the limitations of VaR, which I stated, and use it
for what it is good for but not try to oversell it, not to
think that it represents all possible risk, because any
attempts to somehow make it more sophisticated are just going
to obfuscate it all the more. So you take VaR as one tool for
risk management, and then extend out from there.
The second point, just addressing what you are saying, is
that, number one, I don't think that you can use VaR and have a
`do-over' to try to expand it and have it solve these crisis-
type problems. I also don't think that we will ever be at the
point of being able to know all the risks. But I do think that
we can move somewhat in the direction of understanding crisis
risk more. But to do it, you need the data, and the data that
you really need to start with is: how highly leveraged are the
people in the market, and what are their positions--so that if
there is a shock in a particular market, will there be so much
leverage there that people will be forced to liquidate? What
other positions do they have, so how could that propagate out?
It is not a panacea. You can't have a silver bullet because of
the feedback and gaming capabilities but I think you can move
more in the direction of dealing with these crisis risks.
Regulation of Financial Products
Chairman Miller. My time has expired but I have a question
that is sort of in hot pursuit of what you both just said, and
I will be similarly indulgent to the other Members here.
Dr. Taleb, you said there should be something like a Food
and Drug Administration (FDA) to look at financial products, to
see if they actually do something useful, or if they simply
create additional risks that create short-term profits.
Apparently about 90 percent of derivatives--I was only half
kidding when I asked you if your mother knew you designed
derivatives. But in about 90 percent of derivatives, no party
to the transaction has any interest in the underlying, whatever
it was, that the derivative is derived from--credit default
swaps. Do you agree that some financial products should simply
be banned as having no readily discernible usefulness, utility
for society, for the economy--and creating a risk that we
cannot begin to understand? Should credit default swaps be
banned? Should they be limited to--have a requirement that is
equivalent to an insurable interest requirement in insurance
law? Dr. Taleb or Dr. Bookstaber?
Dr. Taleb. I cannot--I don't--I am not into regulation to
know whether we should be allowed to ban people based on uses
but--based on risk, okay, because society doesn't bear the
risk. I have here what I call the `fourth quadrant,' and we
should ban financial products--and when I call it the fourth,
it is a little technical, but it is a very simple rule of thumb
that takes minutes to check if a given financial product
belongs or doesn't belong to the fourth quadrant. In other
words, does it have any explosive toxic effects on either the
user or the issuer, or both, you know, so it is very easy. So
these products--and this is how I have my fourth quadrant--
these are the exposures we should not just compute, you know,
but eliminate. And there are a lot of things we can measure. I
mean, I may agree with Dr. Bookstaber, VaR may work for some
products, and we know which ones, but not for these products
that have open-ended, toxic, geometric--what I call geometric--
in other words, escalating payoffs.
Chairman Miller. Dr. Bookstaber.
Dr. Bookstaber. For reference, I refer the Committee to
testimony that I gave in June to the Agricultural Committee of
the Senate on the topic of derivatives, and there I pointed out
that, over time, derivatives have moved more and more towards
being used for gaming. In fact, I said that derivatives are the
weapon of choice for gaming the system. They are used to allow
you to hedge when you are not supposed to hedge, to avoid
taxes, to lever when you are not supposed to lever. There is
vested interest on both the sell and the buy side to have
derivatives that are complex and obfuscating, that are
customized. I believe, number one, that many derivative
instruments that exist today are used more for either gaming or
gambling purposes as opposed to having true economic function.
And I believe that there are many customized and complex
instruments that could easily be transformed into a set of
standardized instruments that would be easier to track, more
transparent, and possibly even put on an exchange. So I
certainly agree with the concept that derivatives is a point to
focus on, because it is one of the ways that we find risk
coming in these tail events in surprising ways.
Chairman Miller. Thank you, Dr. Bookstaber.
I now recognize Dr. Broun for nine minutes and 45 seconds.
`Too Big to Fail'?
Mr. Broun. Thank you, Mr. Chairman. I want to make a quick
statement. I believe, first thing, that there is no such thing
as an entity that is too big to fail, particularly when we look
at businesses, even large businesses such as the investment
banks, and I believe in holding people personally accountable
and responsible, and I believe that when you take away the
taxpayer safety net that people are utilizing to gamble away
other people's future, then people will be held more
accountable and will make better decisions. I think greed and
lust are two tremendous blinding factors when people start
making decisions.
Having said that, I also want to state that I think that
there were a lot of warning signs about this current economic
crisis that we found ourselves in, and many people sounded the
horn of warning saying that we needed to change federal law and
regulation to prevent what has happened, and those warnings
were unheeded by Congress and by people who were in the
decision-making process. Having said that, I am real concerned
too because investment banks took excessive risk based on these
models and commercial banks are also now forced to rein in
risk, even though they are not taking risky positions to begin
with, those commercial banks. What can we do to ensure that
small commercial banks around the country are not punished by
the risky behavior of large investment banks? Either or both,
who wants to go first?
Dr. Bookstaber. That is a difficult question, and I don't
know that I can illuminate it too much, but I can go in a
particular direction. You can correct me if I am going the
wrong way. I think there is a distinction between the larger
banks, which de facto actually are the investment banks, and
the smaller banks, because the larger banks end up quasi-market
makers in the sense that they take on positions of risk for
clients. They become market makers in the fixed-income market.
They issue and support derivatives. They also have proprietary
trading desks so they are also quasi-hedge funds. So I think
you can look at the various functions of banks, and look at
smaller banks, and they typically have a pure banking function.
Larger banks are not really just bigger versions of smaller
banks. They are actually institutions that take different types
of risk that smaller banks don't take, that can have some of
these tail events of their own creation--that are demons of
their own design--that they have created because they have
elected to go into the derivatives markets, or take market-
making functions.
So I think the question for a regulator is, do you have a
different set of regulations and requirements for the banks
that--it is not an issue of being too big to fail, but banks
that are taking on types of risk that make them distinct from
their smaller cousins.
Mr. Broun. Isn't it greed that drives that as far as the
large institutions, though?
Dr. Bookstaber. Well, you know, greed has a little bit of
spin to it. I mean, there are incentives, and people act based
on their incentives; and if we give somebody a set of
incentives that, as Dr. Taleb has mentioned, lead them to say,
`I want to take risks which might blow the bank up, with small
probability, but with very high probability will give me a
large bonus,' you are going to have people acting accordingly.
So I think the way to think of it is, not that they are acting
on the basis of greed, but they are acting on the basis of
incentives that lead to behavior that, for the market overall,
may be unduly risky.
Mr. Broun. Isn't it so particularly when you have somebody
else who is going to be held responsible for that decision-
making process?
Dr. Bookstaber. Right.
Mr. Broun. Like the taxpayer is going to be on the hook if
they make a bad decision.
Dr. Bookstaber. That is right. There is no doubt that
incentives have played a large role in what we have observed.
You know, had you had, for example, somebody like Mr. Prince
saying--apparently recognizing the riskiness of what they are
doing--and saying, well, as long as the music is playing, we
are going to keep dancing. Why is he going to keep dancing?
Because his incentive is based on next quarter's earnings, and
he can't walk away from that dance floor while his competitors
are still on it, because his incentives are structured to make
that incorrect decision.
Mr. Broun. And he has everything to gain and nothing to
lose in that process, correct?
Dr. Bookstaber. Yes.
Mr. Broun. Dr. Taleb.
Dr. Taleb. Yes. Well, I just wrote a paper with my
colleague (Charles Tapiero) in which we showed why--I don't
know if you have heard about the case of Societe Generale, the
French bank that lost $7 billion, $8 billion on a rogue trader,
and we showed that it came from too big a size. Size has effect
in compounding risk, and let me give you the intuition. If you
have a bank a tenth of the size of Societe Generale, and they
had a rogue trader that had a tenth of the size of the position
of that rogue trader, the losses would have been close to zero.
The fact that they had to liquidate, they discovered that that
rogue trader had 50 billion euros in hidden position and they
had to liquidate that, and liquidating 50 billion euros rapidly
costs a lot more than liquidating five billion euros. You
liquidate five billion euros at no transaction cost almost, or
a very small transaction cost, compared to liquidating 50
billion. So that would generalize to risks of unexpected events
tend to affect large size more.
And I have here another comment to make about banks. Banks,
of course, have done so far--I mean, we have evidence they have
done, so far, very little for society, except generate bonuses
for themselves, from the data, and that is not from recent
events that I am deriving that. When I wrote ``The Black Swan''
it was before these events. But look at hedge funds. Hedge
funds, I heard the number, 1,800 hedge fund failed in the last
episode. I don't know if many of them made the front page of
any Washington paper. So the hedge funds seem to be taking
risks without endangering society, or at least not taxpayers
directly. And this model of hedge fund corresponds to my norm,
okay--what is a complex system that is robust? The best one is
Mother Nature. Mother Nature has a lot of interdependence. We
have an ecosystem, a lot of interdependence. But if you went
and shot the largest mammal, a whale, or the largest land
mammal, an elephant, you would not destroy the ecosystem. If
you shot Lehman Brothers, well, you know what happened, okay.
You destroyed the system--too much interdependence means you
should not have large units. But hedge funds have shown us the
way to go. They are born and they die every day, literally
every day. Today I am sure that many hedge funds are born and
many hedge funds have died. So this is a model that replicates
how nature works with interdependence. But of course we have to
listen to Dr. Bookstaber's advice to make sure that they don't
all have the same positions you have to put the exclusionary
system, but they have a lot more diversity than banks.
Wall Street's Dependency on Government Bailouts
Mr. Broun. Isn't it though that the implied or even
outright safety net of the taxpayers picking up the pieces if
there is a failure, isn't that the thing that is driving the
derivatives and all these other complex financial instruments
that cause people to make these risky behavior judgments?
Dr. Taleb. Well, I am under oath and I will say exactly
something that I want to be on the record. I was a trader for
21 years, and every time I said what if we blow up, he said,
who cares, the government bails us out. And I heard that so
many times throughout my career, that, ``don't worry about
extreme risks, worry about down five percent, ten percent,
don't worry about extreme risks, they are not your problem
anymore, it is not our problem.'' I heard that so many times,
and here I am under oath and I say it.
Dr. Bookstaber. If I may add to that, there is the notion,
well known, of what is called the trader's option. The trader's
option is, I get X percent of the upside and limited or zero of
the downside, but that trader's option extends also in many
cases to the management of the firms. They get the upside and
so you would much rather, you know, construct a position that
makes a little, makes a little, makes a little and makes a
little and with small probability loses everything, because
that increases the chance that you have consistent earnings,
consistent bonuses, and in the extreme events, your downside is
limited because of the option characteristic of your
compensation.
Mr. Broun. So in the ten seconds I have left, I just want
to state that taking away the government safety net is going to
make people more responsible and they will make better
decisions on a real risk management basis, and I thank you all.
It is my opinion that that is what I am getting from you all,
correct?
Dr. Taleb. In my opinion as well.
Dr. Bookstaber. If I may, I would just say, it is not just
the safety net. If I am an individual in a firm, I don't care
about the safety net, I care about my own bonus, so with or
without the safety net for the firm overall, if my incentives
are, I make money if things go up, I get a new job if things
blow up, I don't know that the safety net matters to me
personally.
Dr. Taleb. May I respond to this point?
Chairman Miller. Dr. Taleb.
Dr. Taleb. I agree that if I am a trader, I don't care who
is going to bail me out. The problem is that the shareholders
don't care when society can bail them out because there is
unlimited liability, that shareholders are protected so society
bears the rest. So we have three layers: a trader, the
shareholder and thirdly, society. So in the end, the free
option comes from society.
Mr. Broun. Thank you, Mr. Chairman.
Chairman Miller. Thank you.
I think something like 90 percent of American households
have a household income of less than $105,000 a year, so for a
trader to make $100 million, $120 million does not seem like
make a little, lose a lot.
Ms. Dahlkemper for five minutes.
Ms. Dahlkemper. Thank you, Mr. Chairman.
I wanted to go back to your statement in terms of some--
that maybe some financial products should be banned, and there
are some that may argue that banning any financial product is
an excessive intrusion into the free market. So if you could
just give me your response to that claim.
Dr. Taleb. I believe in free markets but I do not believe
in state socialism, okay, and I don't believe--I believe the
situation we have had so far is not free markets. It is
socialism for losses and capitalism for profits. So if the
taxpayer is involved ultimately in bailing out, which the
taxpayer should be able to say, I want this product or that
product, the risk, okay? You know, my opinion, I am in favor of
free markets but that is not my definition of free markets,
okay, state-sponsored socialism for the losses and capitalism
for the profit--I mean, free market for the profit. That I
don't--as a taxpayer, and I am paying taxes.
Ms. Dahlkemper. Dr. Bookstaber.
Dr. Bookstaber. I think even in a capitalist system, the
argument that some products should not go forward or should be
banned is a reasonable one for the following reason: that if I
construct some new product, and let us say it is a fairly
complex product or has a fat tail and it can inflict problems
for society, there is a negative externality to that product
that is not priced--that is, I sell it, I create it, somebody
wants to buy it, but the negative externality is the increased
probability of crisis that it causes, and any time that you
have a non-price-negative externality is a time that I think
even a libertarian would argue you can have government
intervention.
The Risks of Different Tupes of Institutions
Ms. Dahlkemper. Thank you. I also wanted to go back a
little bit to the `too big to fail' subject in terms of the
institutions. When we look at the surviving large banks, they
are bigger than ever, so where do you know when an institution
is `too big to fail' and how do we restructure these firms?
Dr. Taleb. `Too big to fail,' you can see it. If anything
in nature is bigger than an elephant, it won't survive, and you
can see, I am sure anything bigger than a large hedge fund, to
me, is too big. But there is one thing here associated with the
problem. The reason we depend so much on banks is because the
economy has been over-financialized over the past 25 years,
over-financialized. The level of debt today in relation to GDP
is three times, according to some numbers, even more or less,
but three times the level of debt to GDP that we had in the
1980s. So that is rather worrisome. This is why we have `too
big' banks, all right, because it comes with the system. It is
a process, you know, that feeds on itself, that is a recursive
process. And if we definancialize the economy more, the debt
level will come down. Then the discussion about `too big to
fail,' about banks, will be less relevant. I mean, banks' role
is not so--you know, banks where I can withdraw money when, you
know, when I go to Atlanta and then there is a bank that is
used for letter of credit, very useful things for society. And
there are banks that trade for speculative reason, banks that
issue paper that nobody needs and there are banks, the banking
that corresponds to lending, you know, increased lending
because a lender makes a bonus based on the size of loans. So
if you brought this down, the size of banks would drop
dramatically. Particularly, the balance sheets would shrink
dramatically, and particularly if we moved the risk away from
banks. The banks are more of a utility in the end, and they are
hijacking us because a utility with a bonus structure, it
doesn't work. As I said here, don't give someone managing a
nuclear plant a bonus based on cost savings, okay, or based on
profitability. You don't, all right? So the problem is, they
are hijacking us because of the dual function of a utility that
we need them to have, a letter of credit or something as basic
as withdrawing cash, and at the same time they take risks with
bonuses. So if we brought down the level of banking, moved the
risks more and more to hedge funds, these people are adults,
they don't endanger anyone, just make sure they don't get big
and have Dr. Bookstaber's rules on, you know, leverage and
stuff like that well enforced . . . then the level of--then
that problem would disappear. So let us worry more about the
cancer rather than worry about the symptoms.
Ms. Dahlkemper. Dr. Bookstaber.
Dr. Bookstaber. You know, the Treasury came out with some
principles for regulation of capital on September 3, and one of
the key issues that they mentioned is dealing with `too big to
fail.' I think one of the difficulties is, I don't think we can
measure too big to fail. I don't think we know. It is not just
a matter of the capital that you have or the leverage that you
have. For example, LTCM was a hedge fund and it was a
relatively small firm and had $3 billion capital, yet in a
sense it was `too big to fail' because it almost brought down,
actually, Lehman along with it, and the Fed had to step in.
What matters is how what you are doing weaves in with what
other people, what other funds or firms are doing within the
economy. So you could have a `too big to fail' that is not
predicated on one institution and what that institution is
doing, but it could be based on some strategy or some new
instrument, where for anyone from that instrument that strategy
is relatively small, but if the exogenous shock occurs in the
market and it affects that strategy, it affects so many firms
in the same way that it has a substantial systemic effect. And
I get back to the point that we don't have the information to
even know right now what type of positions or leverage or
strategies might have that threading across different
institutions.
Ms. Dahlkemper. Thank you. My time is expired.
Chairman Miller. We are about to have 40 minutes of votes
shortly so I would like for both Mr. Wilson and Mr. Grayson to
have a chance to ask questions. I should just tell the panel
that this Charlie Wilson has never had a movie made about him.
So far as I know, he has never been in a hot tub. Mr. Wilson
for five minutes.
Incentive Structures for Trades
Mr. Wilson. Thank you, Mr. Chairman.
Gentlemen, good morning. I serve on the Financial Services
Committee also and I have to keep pinching myself that really I
am in a Science and Technology Subcommittee and so it is hard
to realize the conversations we are having. Dr. Taleb, if I
could say that, you know, what you said earlier in your
testimony about people not being concerned about the success or
failure of a firm because they knew there would be a public
bailout is frightening. That is certainly not the American way
or certainly not the way we want to do business. With those
things in mind, I have a couple questions I would like to ask
and maybe we can get some of your feeling as to how people
would get so far off track, that that would be the thought
process. That concerns me.
People have been outraged at the size of the bonuses and
especially when we were doing the voting for the bailout. Some
of the employees were bailed out, as you all know, with
government money, huge amounts of money to the Wall Street
firms. Much of the conversation was about firms being `too big
to fail,' and you say that in the bonuses, that is really the
motivator for everybody. I would hate to think that there was
no leadership that wouldn't try to keep people on the right
track rather than money being the only motivator, the true part
of it. So can you explain that? And I was going to address this
question, if I could, to Dr. Bookstaber if I could.
Dr. Taleb. You would like me to explain how people were
handling extreme risks.
Mr. Wilson. I did. That was confusing. I am sorry. I did
address that to you but I would be interested in Dr. Bookstaber
also. If you would go first, Dr. Bookstaber, please.
Dr. Bookstaber. Thank you. I don't think--I don't mean to
be cynical, but I don't think that leadership within a
financial firm can overcome the incentives that exist,
incentives not just including the trader's option, but to do
the bidding of the people who have put you in your position,
namely the shareholders whose interest is earnings and maybe
even earnings quarter by quarter. So I think the way that you
have to change things is through the incentive structure of the
people who are taking risk in ways that has been widely
discussed, and I think it is fairly clear that you don't
finally get paid until whatever trade you have put on, or
whatever position you put on, is off the books and has been
recorded. You can't basically put on positions and get paid
based on the flow of those positions until the trade is
realized, that is, until the book is closed on that trade. So
this is the notion of longer-term incentives. So if you have
longer-term incentives, if you have incentives where you can't
game the system by constructing trades or portfolios that again
make a little, make a little, maybe blow up, then people will
act based on those incentives. But the leadership of the firm
is always going to have the following statement, that our
responsibility is to the shareholders, we have to maximize
shareholder value, and then the shareholders, by the way,
although in theory they have a vote, in practice don't. And so,
you know, you have sort of this--the management pointing
towards the shareholders, the shareholders effectively being
silent partners within the corporation.
Mr. Wilson. Thank you.
Dr. Talber, am I saying that right?
Dr. Taleb. Taleb.
Mr. Wilson. Taleb. I am sorry.
Dr. Taleb. There are two problems, and I gave two names, a
name to each problem. The first one is called, the title of my
first book, fools of randomness, `Fooled by Randomness,' and
other people who believe their own story and actually don't
know that they are engaging in these huge amount of hidden
risks out of psychological, you know--as humans, we are not
good at seeing negative outcomes. We are victims of
overconfidence, so we make these mistakes whether or not there
is a bonus, is the psychological, the first one. And the second
one, I call them `crooks of randomness,' so there is `fools of
randomness' and `crooks of randomness,' and you always have the
presence of both ailments in a system. Like, for example, when
we had LTCM, Long-Term Capital Management, the problem, these
people had their money in it, so, visibly, they were not gaming
the system consciously, all right, they were just incapable of
understanding that securities can take large variations. So
there are these two problems. So the bonus, it is imperative to
fix the bonus structure, and as I said here, that I don't know
any place in society where people manage risk and get a bonus.
The military people, the police, they don't get a bonus. So
fix, make sure that he who bears risk for society doesn't have
a bonus. Fix the bonus structure that is not sufficient.
Mr. Wilson. One of the things that, you know, we have heard
a lot about since the money was invested in Wall Street was
that if the big bonuses didn't continue, the firms couldn't
necessarily keep the talent. Do you have any comment on that,
Dr. Taleb?
Dr. Taleb. I am laughing, sorry, because a lot of these
people--in my book there is a gentleman who had $10 million, a
big hotshot trader, and when he blew up, he couldn't even drive
a car. I mean, you can find better cab drivers. I don't know
what you could do with these Wall Street derivatives, high-
income people other than use them as drivers but even then, I
mean, you can use someone less reckless as a driver. So I don't
know what to use them for, honestly. I don't know what is the--
I was on Wall Street for 21 years and a lot of people I
wouldn't use for anything. I don't know if you have some
suggestions. So I don't know what you are competing against,
all right, and you have high unemployment on Wall Street, and
calling that `talent' is a real--it is a very strange use of
language, people who lost $4.3 trillion worldwide in the
profession, and then calling it `talent.' So there is talent in
generating bonuses, definitely, that you cannot deny. Other
than that, I don't know.
Dr. Bookstaber. There is--on this point, there are people
who are not merely talented, but gifted, in areas like medicine
and physics and other fields and they seem to get by on some
amount of money, $200,000, $500,000, $1,000,000. I don't know
that the talent in Wall Street is so stellar that it is worth
$50 million or $100 million versus the talent in these other
fields. The issue with the talent more is that the structure of
Wall Street somehow allows that level of compensation, so if
one firm does not allow it, people can move to another firm
that does. But if there is a leveling of the field overall so
that instead of $20 million people are making $1 million or $2
million, you know, then I think this issue of, you know, `we
will lose our talent' disappears. It has to be done in a
uniform way, as opposed to affecting just one firm versus
another.
Mr. Wilson. Thank you.
Chairman Miller. Dr. Taleb, do you want to respond?
Dr. Taleb. Yes, I have one comment. He is making a
socialistic argument to limit bonuses. I am making a
capitalistic argument to limit bonuses. I am saying if people
want to pay each other, they can pay whatever they want. I just
don't want society to subsidize bonuses. That is it. I am
making the opposite argument coming from--so this is an extreme
bipartisan conclusion here where----
Mr. Wilson. We have a few of those here.
Dr. Taleb. If people want to take risk, you know, and two
adults can hurt each other financially as much as they want.
The problem is, as a taxpayer, okay, I don't want these
bonuses.
Mr. Wilson. Thank you. Thank you both.
Mr. Chairman, just one comment if I could. It just seems
that we have to try to find a way to legislate maybe some
character to Wall Street.
Chairman Miller. Thank you. I misread the note that said
that we would shortly have 40 minutes of votes. We will have
votes at around 11:45 and they will last 40 minutes, so I am
delighted that we will be able to continue with this panel for
Mr. Grayson and for a second round of questioning. Mr. Grayson.
Holding Wall Street Accountable for Bonuses
Mr. Grayson. Thank you, Mr. Chairman.
We are talking today about what proper incentive structures
we should have on Wall Street, and I am wondering if we are
talking too much about carrots and not enough about sticks. In
fact, people on Wall Street did lose over $4 trillion of our
money, and I have seen almost no one punished for it. Don't you
think that it would be likely to deter bad behavior and an
overly fond view of risks if we actually punished people?
Dr. Taleb. I am not a legal scholar but there has got to be
a way to--there is something called malpractice, okay. There
has got to be a way where we can go after these people that I
haven't seen so far, because people are scared, because Wall
Street has `talents.' These people would run away and go to
Monte Carlo or something, so we are afraid of letting them, you
know, of them running away, but we should be doing it
immediately, find people who made [these losses]--like the
Chairman of an executive committee or the firm that we had to
support who made $120 million of bonuses, and supervised
unfettered risk taking and made sure that that gentleman got
returns of $120 million bonuses. The place where my idea was
most popular was Switzerland. The first event of a clawback in
any country took place in Switzerland, where the authorities
went to Mr. Marcel Ospel, head of UBS, after the events of
October and told him, listen, give us 12 million Swiss francs,
please, and it was voluntary and he gave back almost--a large
share of his--but he clawed back his bonuses.
Mr. Grayson. But it was voluntary only because the
government intervenes by limiting people's liability. The
concept of liability is determined by our law, not by the free
market. In fact, if we were to say that we will not give people
the right to hide behind corporate shields, wouldn't that have
a dramatic effect on holding people accountable for the bad
decisions that they make?
Dr. Taleb. To answer, okay, this is still the same problem,
fooled by randomness or not fooled by randomness. Some people I
have seen in Chicago trade their own money and lose huge
amounts of money, not knowing they could lose it, so someone
whose net worth is $2 million loses $2 million and had to go
burn his house to collect insurance money. So I have seen that.
It is not just--so people sometimes engage in crazy trades,
okay, where they have liability themselves. It may not be
sufficient, but it would be, for me, economically, a good way
to have a bonus compensated by malice because capitalism is not
just about incentives, it is about punishment.
Mr. Grayson. When you say it wouldn't be sufficient, all
you are really saying is that it wouldn't solve the problem for
all time, forever in every case, but it would certainly be a
step in the right direction.
Dr. Taleb. Oh, it would be imperative, not a step.
Mr. Grayson. Imperative?
Dr. Taleb. It is an imperative.
Mr. Grayson. Okay. Now, Dr. Bookstaber, I understand that
in Sweden, the bank managers have unlimited liability for the
mistakes that they make, but what happened in our system with
regard to blow-ups, with regard to crazy risks that people take
in order to pad their own pockets, what effect would that have
if we were to take that law and introduce it in America?
Malpractice in Risk Management
Dr. Bookstaber. You know, something along those lines that
I have advocated is to have the potential of penalties for the
risk managers within a firm similar to what are there for the
CFO of a firm. You know, if a CFO knowingly allows some
accounting statement to go out, where he knows it is incorrect,
he is on the hook not just from a civil but from a criminal
standpoint. If you had the risk managers have to sign on the
dotted line, that the risk--that they have executed their
function correctly, and all material risks have been duly
represented--I think that could go a long way towards solving
the problem, because they would then have an incentive to make
sure everything is right. And there are cases, I think, as we
go back to this last crisis, where the risk managers were in
some sense not up to the task, or possibly in bed with the
people involved in trading or with senior management, to where
they were willing to have their views overridden--because they
had no liability on the one side, and they didn't want to get
fired on the other.
Mr. Grayson. But don't we have to do more than that? Don't
we have to not only say to people, you have to fill out these
forms properly and you have to disclose, but we have to
actually hold people accountable for the mistakes that they
make, and hold them personally accountable? Isn't that what we
need to actually deter this kind of misconduct?
Dr. Bookstaber. I guess the question is what type of
mistake, because everybody makes certain types of mistakes. I
think that sort of mistake where you can hold people
accountable is where they--obviously if they knowingly
misrepresent--but where there is something material that they--
on the one hand it is a malpractice where you say, you know,
somebody doing this job in a reasonable way should have
discovered that.
Mr. Grayson. But let us talk about the specific problems we
have seen time and time again in the last few years. Let us
talk about, for instance, AIG. In AIG, the fundamental problem
is that the traders entered into literally billions upon
billions of dollars of heads, I win, tails, you lose bets, bets
that couldn't possibly be made good on by anybody but the U.S.
Government, and that wasn't a problem of not filling out the
form properly, not disclosing. Don't those people need to be
punished in order to deter that conduct in the future?
Dr. Bookstaber. Well, this gets to Dr. Taleb's point that
you would have to go into the mindset of the people. Was it, as
he is saying, you know----
Dr. Taleb. Crooks or fools.
Dr. Bookstaber. Yeah, were they crooks or fools. If you can
discern one from the other, then I agree with you, but what I
am saying is, you could also go one level higher to require,
which now is required, risk management oversight for those
functions where it is believed to be credible, and these were
supposed to be the people who know how to do their job, and
they have the responsibility to represent that this type of
event is not occurring.
Mr. Grayson. Dr. Taleb.
Dr. Taleb. Yes. Well, the problem I saw and I wrote about,
actually, in one of my writings not yet published, I say it is
easier to fool a million than fool a person and it is much
easier to fool people with billions than to fool them with
millions. Why? Because you have bandwagon effects, and you have
collective--something called diffusion of collective
responsibility, and I will tell you exactly why. If you have--
what risk managers are doing is to make sure they do exactly
what other risk managers do. If there is a mistake, it is a
mistake that they did not commit individually, but committed--
that had company on that. We call it `company on a trade.' It
is not like an individual doctor who is just incompetent. It is
collective incompetence. We had collective risk management
incompetence, but they were all doing what other people--the
hedge is to do what the other guy is doing and that, I don't
know if, you know----
Chairman Miller. Well, the note I got earlier was incorrect
and now it appears we are going to have votes at any moment, so
I will start a round of questions and we will try to keep it--I
know that everybody would like to ask questions of this panel.
Clawback Provisions
Just one--it is not clear to me whether you actually
supported a legal requirement that there be clawback provisions
in bonus contracts, that if a bonus is based upon a profit this
year, that if the very same transaction results in a loss in
two or three years there be requirement, a legal requirement
that that bonus be repaid. Dr. Taleb?
Dr. Taleb. Indeed.
Chairman Miller. You do----
Dr. Taleb. Indeed.
Chairman Miller. Dr. Bookstaber.
Dr. Bookstaber. I don't know that I would go to the extent
of having it be a legal requirement. Ideally, it should be
requirements placed on the corporation by the equity holders,
because it makes good economic sense. I think the issue of it
being a legal requirement gets into the question of, okay, if
we are ultimately the ones holding the bag if this fails, we
now have a societal obligation. But I think whether it is done
through the shareholders or if it is legislated, that type of
structure, incentive structure, clearly makes sense for
trading.
Chairman Miller. Dr. Taleb.
Dr. Taleb. There is an additional problem other than the
clawback. There is the fact that if in any given year, I take
$1 million from you, okay--say I win, I get my bonus, and I
lose, you keep all the losses, so that clawback situation
doesn't solve the free option problem. You are solving the
mistiming problem, you are not solving the free option problem.
So we have two problems with bonuses. The first one is
symmetry. In other words, I make, all right, either a big bonus
or nothing, whereas if he loses, I take his money, risk his
money. He loses or makes [money], all right, I just make
[money], I just earn. So that problem is not solved with the
clawback. For example, say the TARP money we gave Goldman, all
right--okay, let us forget about clawbacks. Had they lost
money, all right, it would have been--we would have eaten the
loss. If they made money, they kept the bonuses, okay, so that
idea of having just profits and never losses, net, net . . .
the clawback is about repaying previous bonuses, but it doesn't
address the vicious incentive of paying someone for the profits
and not charging him for the overall losses, and the clawback
doesn't solve that.
Chairman Miller. Are you suggesting that that should be
prohibited by law, or should people just have better sense than
to agree to that kind of compensation system?
Dr. Taleb. In other words, people should have skin in the
game. Net, net, net, if I fail, I should be penalized
personally some way or another. Don't have an option where I
only have the profits and none of the losses.
Chairman Miller. I am still not clear if you are suggesting
that that be a legal requirement or there simply should be a
change in the culture, that anyone who agrees to a hedge fund
compensation of 220 is a fool, and if people stopped agreeing
to it, the compensation system would change.
Dr. Taleb. No, to me, it should be only a legal requirement
wherever TARP or a possible society bailout is possible. If
there is no society--if someone signs no society bailout, then
no.
Credit Default Swaps
Chairman Miller. I asked the question earlier but I am not
sure I got a clear answer. Do you think credit default swaps
should be banned? If not, do you think they should be limited
to--they should have a requirement that would be comparable to
the requirement of an insurable interest in insurance law?
Dr. Bookstaber. I agree with the latter. I don't believe
that credit default swaps should be banned, because they do
have economic function in the sense that--if I have the debt of
a company and perhaps it is illegal, or for some reason it is
difficult for me to undo my risk by selling it, I can use the
swap to mitigate or hedge my risk. But I don't think that it
should turn into what it has turned into--basically, a gambling
parlor of side bets for people who have no economic interest at
all in the underlying firm. The point you mentioned, Mr.
Chairman, in your opening remarks, that the number of people
doing side bets far exceeds those who actually have an economic
reason to be taking that exposure.
Chairman Miller. Dr. Taleb.
Dr. Taleb. Mr. Chairman, these products are absurd. They
are class B products for me, for the simple reason that it is
like someone buying insurance on the Titanic from someone on
the Titanic. These credit default swaps, you buy them from a
bank, so they make no sense. And I have been writing about
these class B instruments that have absolutely no meaning and I
don't believe that they have economic justification other than
[to] generate bonuses.
Chairman Miller. The other analogy I have heard is buying
insurance against a nuclear holocaust; if you think you are
going to be around to file a claim, who do you think you are
going to file it with. I will give up my own time; Dr. Broun.
Were the Bailouts and Stimulus Funds Necessary?
Mr. Broun. Thank you, Mr. Chairman. Do you believe that
bailing out banks and transferring debt from private sources to
public sources is a responsible action?
Dr. Taleb. I mean, my opinion is, I am going to be very,
very, very honest--it is irresponsible because we have levels
of about $60 trillion, $70 trillion worldwide in excess debt
that is being slowly transformed into something for our
children. If a company goes bankrupt, that debt disappears the
old-fashioned way or it turns into equity. If government bails
out a company, it is a debt that our children and grandchildren
will have to bear. So it doesn't reduce debt in society, and
this is why I have been warning against the stimulus packages
and all of these. Transforming private debt into public debt is
vastly more vicious than just taking the pain of reducing the
level of debt.
Mr. Broun. Dr. Bookstaber.
Dr. Bookstaber. In the abstract, I don't think that makes
sense. In the current crisis, I think it was inevitable,
because we had to adjust for problems that got us to where we
are. So I would say we would want to construct a system with
regulatory safeguards, with adequate capital, with correct
incentives so that the event doesn't occur where we have to
move into the bailout mode that we had in the recent past. But
my sense is that if we hadn't taken this action, as distasteful
and costly as it may be, the end results for the economy may
have been far worse.
Mr. Broun. So you believe that stimulus spending and debt
accumulation and the bailouts are all necessary responses to
this economic crisis, is what I am gathering.
Dr. Bookstaber. Yes, I believe they were for this crisis. I
don't believe that as a general principle it is something that
we want to occur, and hopefully we can take steps so that it
doesn't occur again.
Mr. Broun. Dr. Taleb.
Dr. Taleb. I don't believe in deficit spending for the
following reason, and it comes from the very same mathematics
that I used to talk about tail risks. We live in a very
nonlinear world--as you know, the butterfly effects, a
butterfly in India causes a rainstorm in Washington. You know,
these small, little--we don't quite understand the link between
action and consequences in some areas, particularly monetary
policy. So if you have deficit spending, it is debt that
society has to repay someday, okay? You depend a lot more on
expert error and projections. I showed in ``The Black Swan,''
in my book, ``The Black Swan'' from 27,000 economic
projections, that an astrologist would do better than
economists, including, you know, some people here who are
economists making projections. So I don't want to rely on
expert projections to be able to issue a stimulus and say oh,
no, no, look what will happen by 2014, we will be paying it
back. These are more of the huge errors.
So what is the solution? The solution is going to be that
all this, all right, may lead to what governments have been
very good at doing for a long time--printing, okay. And we know
the consequences of printing; everybody would like to have a
little bit of inflation but you cannot. Because of non-
linearities, it is almost impossible to have the 3.1 percent
inflation everybody would love to have. You see, a little bit
of error could cause hyperinflation, or if you do a little
less, maybe it would be ineffective. So to me, deficit
spending, aside from the morality of transferring, you know,
private debt into my children's debt--okay, aside from that,
because someone has got to buy that bond, okay, the way it may
lead--you know, because of error in projection--[is] into
printing of money.
Mr. Broun. So from my previous questions as well as
others', I take it that both of you all would agree, looking in
the future, not only with this economic crisis but in the
future, to prevent other economic crises, the real solution is
to take away the taxpayer safety net which was implied and now
with Freddie and Fannie is express taxpayers being on the hook
for this mismanagement and their bad decisions. Would you both
agree, yes or no, that taking away that safety net will help
people be more responsible, and we will have more of the sticks
that my colleague was talking about and that they can within
their own company just to protect their own company's
viability, et cetera, will put in place more responsible risk
management and they will make better decisions. Would you both
agree with that statement?
Dr. Taleb. I agree with the statement, remove the safety
net.
Dr. Bookstaber. I don't know that I can say yes or no
because I have to envision what the future world looks like. If
we make no changes in terms of regulation and oversight, then I
wouldn't agree with the notion of taking away the safety net
because we have a flawed system where there is a notion of `too
big to fail' . . . where if certain institutions do fail, it
has severe adverse consequences for people on Main Street. I
think that we have to say, we want to get rid of the safety
net, and to do that we need to get the corrective incentive
structures, the correct level of oversight from regulators, the
right capital requirements. So as an end result, that is where
I believe we should go, but I don't think we can be there in
good conscience for the typical citizen without doing a better
job, you know, in the regulatory arena.
Dr. Taleb. I don't understand this logic because I don't
see how--in 1983, when banks were bailed out, and even one of
them was the First National Bank of Chicago. It set a bad
precedent. Every time I heard the same argument, you hear the
same argument, ``this is necessary, society can't function, but
in the future we'll make sure we don't do it again.'' I don't
understand this argument.
Mr. Broun. Thank you, Mr. Chairman.
Chairman Miller. Ms. Dahlkemper? Okay, Mr. Wilson?
Mr. Broun. I think we need to go vote.
Chairman Miller. We have been called to our votes. Thank
you very much to this panel. We will be gone for about 20
minutes, not 40 minutes as I earlier understood. But at that
point it does make sense to excuse this panel, but thank you
very much. It has been very helpful and even entertaining. And
then when we come back, when we return we will have the second
panel, although these are the last votes of the week so it is
possible some Members will not come back but go straight to the
airport. Thank you, and we will be at ease.
[Recess.]
Panel II:
Chairman Miller. Other Members may return or may not, but I
think we should begin the second panel, and I also mean it when
I say that this panel is unusually distinguished. Our witnesses
are leading experts in their respective fields. Dr. Gregg
Berman is the Head of Risk Business at RiskMetrics Group, which
is the present-day descendant of the group at J.P. Morgan that
created the Value-at-Risk methodology. He has worked with many
of the world's largest financial institutions on the
development of risk models. Mr. James Rickards is the Senior
Managing Director of the consulting firm Omnis Inc., is a
former risk manager and investment banker who has been involved
in the launch of several hedge funds. As general counsel of
Long-Term Capital Management during the 1998 crisis, he was the
firm's principal negotiator of a bailout plan that rescued it.
And Mr. Christopher Whalen is the Managing Director at
Institutional Risk Analytics, a provider of risk management
tools and consulting services. He volunteers as the Regional
Director of the Professional Risk Managers International
Association, and edits a weekly report on global financial
markets. And finally, Dr. David Colander, the Christian A.
Johnson Distinguished Professor of Economics at Middlebury
College, has written or edited over 40 books, more than 40
books, including a top-selling Principals of Economics textbook
and more than 150 articles on various aspects of economics,
including the sociology of the economics profession.
You will also have five minutes for your oral testimony,
your spoken testimony. Your written testimony will be included
in the record for the hearing. When you have completed your
spoken testimony, when all of you have, we will have rounds of
questions from the Members who are here, which may include me
repeatedly. It is the practice of this subcommittee, as you saw
earlier, to receive testimony under oath. Again, I don't think
any of you have to worry about perjury. That would require that
the prosecutor prove what the truth was, beyond a reasonable
doubt, and that you knew what the truth was beyond a reasonable
doubt. Do any of you have any objection to swearing an oath?
Okay, and I think you may sleep easy tonight without worrying
about perjury prosecution. You also have the right to be
represented by counsel. Do any of you have counsel here? And
all the witnesses said that they do not. If you would now
please stand and raise your right hand. Do you swear to tell
the truth and nothing but the truth?
The record will show that all the witnesses did take the
oath. We will begin with Dr. Berman.
STATEMENT OF DR. GREGG E. BERMAN, HEAD OF RISK BUSINESS,
RISKMETRICS GROUP
Dr. Berman. Thank you. I would like to begin by thanking
the Committee for this opportunity to present our thoughts on
Value-at-Risk and banking capital, especially in the context of
the present financial crisis.
My name is Gregg Berman and I am currently the Head of the
Risk Business at RiskMetrics Group. I joined as a founding
partner 11 years ago when we first spun off from J.P. Morgan,
and throughout that time I have had a number of roles, from
leading research and development to leading product design, but
mostly spending time with clients, and those clients include
some of the world's largest hedge funds, largest asset managers
and certainly the world's largest banks. During that time, and
even under oath I feel I can say this, I have not traded any
derivatives in any way, shape or form.
My comments today revolve around three essential points.
First, Value-at-Risk, or simply `VaR,' was created about 15
years ago to address issues faced by risk managers of large,
multi-asset, complex portfolios. The purpose of VaR was to
answer the question: how much can you lose? In this context, it
has actually enjoyed tremendous success, ranging from revealing
the hidden risks of complex strategies to communicating with
investors in a consistent and transparent fashion.
Second, VaR is a framework. It is not a prescriptive set of
rules. As such, it has been implemented in many different ways
across a wide variety of institutions. Criticisms of VaR that
focus on the use of normal distributions or poor historical
data must be taken in context. These issues are often the
results of specific VaR implementations that may not have kept
up with the best practices in the community.
Third, most VaR methodologies utilize recent market data to
estimate future short-term movements in order to allow risk
managers to make proactive decisions based on rapidly changing
market conditions. This is what VaR was designed to do.
Research shows that these estimates are indeed quite robust,
but they are not designed to predict long-term trends, and they
are not designed to operate when the markets themselves stop
functioning. Banks, on the other hand, must be protected
against adverse long-term trends and in situations where the
markets actually stop functioning. This, therefore, is not the
domain of Value-at-Risk.
So how do we tackle this problem? We start by noting that
the current crisis is driven by two primary factors: one, the
failure of market participants and of regulators to acknowledge
and prepare for large negative long-term trends, such as a
decline in home prices or buildup of leveraged credit, coupled
with, two, the failure of many institutions to accurately and
completely model how these negative long-term trends would
actually affect their financial holdings. In this context, I am
using the word ``model'' to mean a mathematical representation
of a security or derivative that shows how its value is driven
by one or more underlying market factors. Since both of these
issues were quite well known for quite long periods of time, it
is very hard to say that this crisis was unforeseeable,
unknowable or a fat-tailed event.
All market participants, including banks, must do a better
job at modeling complex securities and in understanding how
their strategies will fare under changing market conditions.
For example, if the holders of mortgage-backed bonds would have
known how sensitive these assets were to modest changes in
default rates, they may not have purchased them in the first
place. New rules, regulations and other types of policy changes
regarding better disclosure in data must be done in order to
address this critical issue.
But it is banks and regulators who must specifically focus
on preparing more for the negative long-term trends that lie
ahead and less on trying to predict things with probabilities.
Though current VaR methodologies are designed to estimate
short-term market movements under normal market conditions,
regulators nevertheless try to recast these models in order to
measure the probability of long-term losses under extended
market dislocations. We propose that it is not the model that
needs to be recast, but that regulators need to recast the
question itself.
VaR is about making dynamic decisions, constructing
portfolios, sizing bets and communicating risk. On the
contrary, banking capital is more like an insurance policy
designed to protect against worst-case events and their
consequences. Instead of having banks report probabilities of
short-term losses, banks should estimate the losses they would
expect to sustain under a set of adverse conditions chosen by
regulators. The question of `how much can you lose' is thus
changed to `how much would you lose.'
The conditions that banks are tested against should depend
on what type of events policy-makers in the public interest
believe that banks should be able to withstand. In this
fashion, models, probabilities, simulations and predictions are
left to those making ongoing risk-reward business decisions,
whereas the minimum levels of capital needed to ensure a bank's
survival are based on how regulators implement the broader
requirements of policy-makers. Perhaps one bank needs to
survive a 100-year flood whereas an orderly liquidation is all
that is required for a different bank. Perhaps all banks should
be able to weather a further ten percent downturn in housing
prices, but no bank is required to survive a 50 percent default
rate or a 40 percent unemployment rate--not because these
events are highly improbable, but because policy-makers decide
that this is too onerous a burden for a bank to bear.
In summary, VaR is an excellent framework for active risk
management by banks and other financial institutions and the
development of risk models must continue unabated. But banking
capital serves a different purpose, and a resetting of
expectations will allow for the development of much better
solutions driven by policy instead of by probability. Thank
you.
[The prepared statement of Dr. Berman follows:]
Prepared Statement of Gregg E. Berman
I'd like to begin by thanking the Committee for this opportunity to
present our thoughts on Value-at-Risk and banking capital in the
context of the present financial crisis. My name is Gregg Berman and I
am currently the head of the risk business at RiskMetrics Group, a
provider of risk and corporate governance services to the financial
community. I have been at RiskMetrics since its founding 11 years ago
and in the last decade have worked with many of the world's largest
financial institutions on the development of risk models, their use by
hedge funds, asset managers, and banks.
SIMPLE ROOTS OF A COMPLEX CRISIS
My comments today start with a rather bold assertion--the current
crisis was not unpredictable, unforeseeable, or unknowable. In that
sense I'm not sure it should be classified as a fat-tailed event.
Rather, it was caused by the coupling of two fundamental problems,
namely:
1. the inability of market participants to acknowledge and
prepare for the consequences of long-term trends, such as a
protracted downward spiral in home prices, or a leveraging of
the credit market through the use of CDS, and
2. the inability of market participants to recognize the
economic exposures they had to those trends through holdings
such as asset-backed securities and derivative contracts.
The fact that these issues went unchecked for many years led
directly to the creation of multiple, unsustainable market bubbles,
which when burst propelled us downwards into a full-blown crisis.
But if my assertion is correct and these events were foreseeable,
then what does that imply about all the financial models and risk
methodologies that were supposed to monitor and protect us from such a
crisis? It is the answer to this question that I'd like to explore.
THE INEVITABILITY OF VALUE-AT-RISK
In the early days of risk management size was used as a primary
measure of risk. After all, intuition tells us that $10,000,000 in Ford
bonds should be ten times riskier than $1,000,000 in Ford bonds. But
soon the market realized that $10,000,000 of Ford bonds is probably
riskier than a similar value of government bonds, but not as risky as
$10,000,000 of Internet start-up equity.\1\
---------------------------------------------------------------------------
\1\ The matter is further complicated by derivative contracts that
do not even have a well-defined measure of size. For example, what is
the size of a contract that pays the holder $1,000 for each penny-
increase in the average spread throughout September between the price
of natural gas for delivery in November and the price for delivery in
January? Technically the answer is zero since the holder owns no
natural gas, but the risk is certainly not zero.
---------------------------------------------------------------------------
To address these issues practitioners switched from asking ``how
large is your position'' to ``how much can you lose.'' But there is not
just one answer to that question since for any given security differing
amounts can be lost with different probabilities. One can estimate
these probabilities by polling traders, by building econometric models,
by relying on intuition, or by using variations of history to observe
relevant patterns of past losses. Each of these methods has their own
benefits and weaknesses. And unless we consider only one security at a
time, it will also be necessary to make estimates of how the movements
in each security are related to the movements of every other security
in a given portfolio.
These concepts are encapsulated by two well-known statistical
terms: volatility and correlation. If one could measure the volatility
and correlation of every security in a portfolio the question ``how
much can you lose'' could be meaningfully addressed. This process is
the basis of a popular risk methodology known as Value-at-Risk, or VaR.
HOW VaR IS COMPUTED AND HOW IT IS USED
Because security valuations are often driven by underlying market
factors, such as equity prices, spreads, interest rates, or housing
prices, VaR is usually calculated in a two-step process that mimics
this behavior. In the first step a model for the economic exposure of
each security is created that links its value to one or more of
underlying market factors. In the second step future trends of these
underlying factors are simulated using volatilities, correlations, and
other probabilistic methods. These two steps are then combined to
create a curve that plots potential profits-and-losses against the
probability of occurrence. For any given curve VaR is defined to be the
amount that can be lost at a specific level of probability. It is a way
of describing the entire profit-and-loss curve without having to list
every data point.\2\
---------------------------------------------------------------------------
\2\ Exhibit 1 on page 10 shows the potential one-day profit-and-
loss distribution of selling a short-term at-the-money put on the S&P
500. Out of 5,000 trials we see that about 50 of them have losses of
250 percent or worse. Thus VaR is 250 percent with a one percent
probability. Alternatively we can ask for the worst five out of 5,000
trials (a 0.1 percent probability) and observe these losses to be 400
percent or worse.
---------------------------------------------------------------------------
The accuracy of any VaR number depends on how well underlying
markets have been simulated, and how well each security has been
modeled. There unfortunately exists a tremendous variability in current
practices and different financial institutions perform each step with
varying levels of accuracy and diligence.\3\ Deficiencies in how VaR is
implemented at a particular firm should not be confused with
limitations of VaR itself.\4\
---------------------------------------------------------------------------
\3\ The marketplace is rife with common fallacies about VaR due to
poor implementations. When VaR first became popular in the mid-1990's
computing power limited how accurately instruments, especially
derivatives, could be modeled. Approximations that relied on the use of
so-called normal distributions (bell-shaped curves) were often
required. Also, the amount of market data that could be used, and the
frequency at which this data was updated, was limited by technical and
mathematical challenges resulting in further approximations. However,
by the early part of this decade many of these challenges were overcome
and today's simulation techniques do not rely on normal distributions
and are not restricted by limited data. Unfortunately many institutions
with older implementations still use somewhat outdated and approximate
methods that do a poor job in estimating the risk of multi-asset,
derivative-heavy portfolios.
\4\ One fundamental criticism of VaR is that it can be ``gamed'' or
manipulated since one number cannot by itself represent or reveal all
possible ``tail-loss'' events. This is easily rectified by simply
asking for VaR numbers at more than one level of probability, by
computing the average of all losses comprising a tail event (often
called conditional VaR or expected loss), or by examining the entire
distribution of estimated future losses and their corresponding
probabilities.
---------------------------------------------------------------------------
But indeed there are limitations. When computed according to
current best practices, VaR is most applicable for estimating short-
term market volatilities under ``normal'' market conditions. These
techniques are based on over a decade of well-tested research
demonstrating that in most circumstances recent market movements are
indeed a good predictor of future short-term volatility. VaR models
have seen tremendous success in a wide range of applications including
portfolio construction, multi-asset-class aggregation, revealing
unexpected bets, investor communication, the extension of margin, and
general transparency.
As such, VaR has become an essential part of risk management, and
when properly integrated into an overall investment process it provides
an excellent framework for deploying capital in areas that properly
balance risk and reward.
VAR AND BANKING CAPITAL
So why did this not foretell the current crisis? First and
foremost, many institutions and market participants did not perform
step one correctly--they failed to correctly model how their securities
would behave under changing market conditions. This failure is one of
the leading causes of current crisis.\5\
---------------------------------------------------------------------------
\5\ Many institutions and market participants did not recognize nor
understand how their portfolios and strategies would be affected by a
fall in housing prices or a widening of credit spreads. Regulators had
even less information on these effects and almost no information on how
they were linked across institutions.
It could be argued that if investors had understood the nature of the
mortgage-backed products they had purchased, many would not have
purchased them in the first place (which would have significantly
curtailed the formation of the bubble itself). If regulators had
understood how CDS contracts inherently lever the credit markets they
may not have allowed their unbridled expansion. And if insurance
companies understood that changes in the mark-to-market values of their
derivative contracts would require the posting of collateral to their
counter-parties many would not have entered into those deals. None of
these decisions involve predicting the future or modeling fat tails.
They do involve understanding the present, spending time on the details
---------------------------------------------------------------------------
of financial instruments, and being incented to care about their risk.
Tackling these significant shortcomings may require new regulations
regarding data availability, disclosure, and the analytical
capabilities of each market participant. Central oversight of the
markets themselves will be needed to monitor, and sometimes even limit,
actions that could trigger systemic risk and future liquidity crises.
The second issue is where banking capital comes in. Recall that our
crisis stems from long-term trends, not short-term volatility. And as
mentioned, most of today's VaR techniques are only applicable for
estimating potential short-term movements in well-functioning
markets.\6\ But it is long-term trends and non-functioning markets that
are the concerns of banking capital.
---------------------------------------------------------------------------
\6\ There is nothing endemic to VaR that limits its applicability
to short-term estimates or functioning markets. However, current
methodologies are optimized for those conditions and this is where most
parameters have been tested for proper use. Research into new models
that lengthen the prediction horizon and include factors like liquidity
to account for non-functioning markets is underway. As development of
these methodologies progresses we may see the domain of VaR extended
into more areas of risk.
---------------------------------------------------------------------------
Nevertheless regulators rely on VaR as the basis for many bank
capital calculations.\7\ And even today they continue to recast VaR-
like models in order to address VaR's perceived shortcomings.\8\ We
propose that it is not the model that needs to be recast but rather the
question that regulators want the model to address.
---------------------------------------------------------------------------
\7\ One technique employed to ``fix'' the short-term aspect of VaR
models is to utilize long-term historical data as the basis for
``better'' future estimates. This is a very common but dangerous
practice since it both invalidates any estimates of short-term
volatility (preventing proper use by risk managers trying to be
reactive to rapid changes to the market) and it doesn't actually
provide any better estimates of long-term trends. For a complete
discussion on this and other related topics see included reference by
Christopher Finger (RiskMetrics Research Monthly--April 2009) and
references therein (including a March 2008 report issued by the Senior
Supervisors Group on their study of how risk was implemented at a
variety of large banks).
\8\ See included reference by Christopher Finger (RiskMetrics
Research Monthly--February 2009) containing our comments on the Basel
committee's proposed Incremental Risk Charge--an extension that uses
VaR for additional types of capital charges.
POLICY-BASED BANKING CAPITAL
We believe that the foundation of banking capital is rooted in the
following two questions:
1) What are the adverse events that consumers, banks, and the
financial system as a whole, need to be protected against?
2) What is required from our banks when those events occur?
This is not the domain of VaR. On the contrary, banking capital is
more like an insurance policy designed to protect against worst-case
events and their consequences. Instead of having banks report
probabilities of short-term losses, banks should estimate the losses
they would expect to sustain under a set of adverse conditions chosen
by regulators. The question of ``how much can you lose'' is thus
changed to ``how much would you lose.''
The conditions that banks are tested against should depend on what
types of events policy-makers decide that, in the public interest,
banks should be able to withstand. In this fashion models,
probabilities, simulations, and predictions are left to those making
ongoing risk-reward business decisions whereas the minimum levels of
capital needed to ensure a bank's survival are based on how regulators
implement the broader requirements of policy-makers. Perhaps one bank
needs to survive a hundred-year flood whereas an orderly liquidation is
all that is required for a different bank. Perhaps all banks should be
able to weather a further 10 percent downturn in housing prices, but no
bank is required to survive a 50 percent default rate or a 40 percent
unemployment rate--not because these are highly improbable, but because
policy-makers decide that this is too onerous a burden to expect a bank
to bear.\9\
---------------------------------------------------------------------------
\9\ The recent stress-tests conducted on banks by the Federal
Reserve is an excellent example of how policy, as opposed to
probability, can help set capital requirements. This should not
diminish the role of simulations and the use of models to explore
possibilities and uncover unexpected relationships, but this should be
a guide of what the future may bring, not a prediction of what it will
(or will not) bring.
---------------------------------------------------------------------------
To summarize, we believe that key differences between the needs of
risk management and banking capital suggest different solutions are
required. And in doing so each field can separately develop to meet the
ever-expanding array of challenges we face today.
Biography for Gregg E. Berman
Gregg E. Berman, 43, is currently head of RiskMetrics Risk Business
covering institutional and wealth management offerings that serve Hedge
Funds, Asset Managers, Prime Brokers, Banks, Financial Advisors,
Insurance Companies, and Corporates. Mr. Berman joined RiskMetrics as a
founding member during the time of its spin-off from J.P. Morgan in
1998 and has held a number of roles from research to head of product
management, market risk, and of business management.
Prior to joining RiskMetrics Group, Mr. Berman co-managed a number
of multi-asset Hedge Funds within New York-based ED&F Man. His start in
the Hedge Fund space began in 1993, researching and developing multi-
asset trading strategies as part of Mint Investment Management
Corporation, a $1bn CTA based in New Jersey.
Mr. Berman is a physicist by training and holds degrees from
Princeton University (Ph.D. 1994, M.S. 1989), and the Massachusetts
Institute of Technology (B.S. 1987).
Chairman Miller. Thank you, Dr. Berman.
Mr. Rickards for five minutes.
STATEMENT OF MR. JAMES G. RICKARDS, SENIOR MANAGING DIRECTOR
FOR MARKET INTELLIGENCE, OMNIS, INC., MCLEAN, VA
Mr. Rickards. Mr. Chairman, my name is James Rickards and I
appreciate the opportunity to speak to you on a subject of the
utmost importance to global capital markets.
The world is two years into the worst financial crisis
since the Great Depression. The list of culprits is long,
including mortgage brokers, investment bankers and rating
agencies. The story sadly is, by now, well known. What is less
well known is that behind these actors were quantitative risk
models which said that all was well even as the bus was driving
over a cliff.
Unfortunately, we have been here before. In 1998, capital
markets came to the brink of collapse due to the failure of a
hedge fund, Long-Term Capital Management. The amounts involved
seem small compared to today's catastrophe. However, it did not
seem that way at the time. I know, I was general counsel of
LTCM. What is most striking to me now as I look back is how
nothing has changed and how no lessons were learned. The
lessons should have been obvious. LTCM used fatally flawed VaR
models, too much leverage, and the solutions were clear. Risk
models needed to be changed or abandoned, leverage needed to be
reduced, and regulatory oversight needed to be increased.
Amazingly, the United States Government did the opposite.
They repealed Glass-Steagall in 1999 and allowed banks to act
like hedge funds. The Commodity Futures Modernization Act of
2000 allowed more unregulated derivatives. SEC regulations in
2004 allowed increased leverage. It was as if the United States
had looked at the catastrophe of LTCM and decided to double
down. None of this would have happened without the assurance
and comfort provided to regulators and Wall Street by VaR
models. But all models are based on assumptions. If the
assumptions are flawed, no amount of mathematics will
compensate. Therefore, the root of our inquiry into VaR should
be an examination of the assumptions behind the models.
The key assumptions are the following: one, the efficient
market hypothesis, which assumes that investors behave
rationally; two, the random walk, which assumes that no
investor can beat the market consistently, because future
prices are independent of the past; three, normally distributed
risk. This says that since future price movements are random,
the relationship of the frequency and the severity of the
events will also be random, like a coin toss or roll of the
dice. The random distribution is represented as a bell curve.
Value-at-Risk would be a fine methodology but for the fact that
all three of these assumptions are wrong. Markets are not
efficient, future prices are not independent of the past, risk
is not normally distributed. As the saying goes, ``Besides
that, Mrs. Lincoln, how was the play?''
Behavioral economics has done a masterful job of showing
that investors do not behave rationally and are guided by
emotion. Similarly, prices do not move randomly but are
dependent on past prices. In effect, news may be ignored for
sustained periods of time until a kind of tipping point is
achieved, at which point investors will react en masse. The
normal distribution of risk has been known to be false since
the early 1960s, when studies showed price distributions to be
shaped in what is known as a power curve. A power curve has
fewer low-impact events than the bell curve but has far more
high-impact events. In short, a power curve corresponds to
market reality while a bell curve does not.
Power curves have low predictability but can offer other
valuable insights. One lesson is that as you increase the scale
of the system, the size of the largest possible catastrophe
grows exponentially. An example will illustrate the
relationship between the scale of the system and the greatest
catastrophe possible. Imagine a vessel with a large hold
divided into three sections, separated by watertight bulkheads.
If a hole is punched in one section and that section fills with
water, the vessel will still float. Now imagine the bulkheads
are removed and the same hole is punched into the vessel. The
entire hold will fill with water and the vessel will sink. In
this example, the hold can be thought of as the system. The
sinking of the vessel represents the catastrophic failure of
the system. When the bulkheads are in place, we have three
small systems. When the bulkheads are removed, we have one
large system. By removing the bulkheads, we increase the scale
of the system by a factor of three, but the likelihood of
failure did not increase by a factor of three. It went from
practically zero to practically 100 percent. The system size
tripled, but the risk of sinking went up exponentially.
If scale is the primary determinant of risk in complex
systems, it follows that descaling is the most effective way to
manage risk. This does not mean that the totality of the system
needs to shrink--merely that it be divided into subcomponents
with limited interaction. This has the same effect as
installing the watertight bulkheads referred to above. In this
manner, severe financial distress in one sector does not result
in contagion among all sectors.
This descaling can be accomplished with three reforms:
number one, the enactment of a modernized version of Glass-
Steagall with a separation between bank deposit taking on the
one hand, and market risk on the other; two, strict
requirements for all derivative products to be traded on
exchanges subject to margin position limits, price transparency
and netting; three, higher regulatory capital requirements and
reduced leverage for banks and brokers. Traditional ratios of
eight to one for banks and 15 to one for brokers seem adequate,
provided off-balance sheet positions are included.
Let us abandon VaR and the bell curve once and for all and
accelerate empirical research into the actual metrics of event
distributions. Even if predictive value is low, there is value
in knowing the limits of our knowledge. Understanding the way
risk metastasizes with scale might be lesson enough. It would
offer a proper dose of humility to those trying to supersize
banks and regulators.
Thank you for this opportunity to testify.
[The prepared statement of Mr. Rickards follows:]
Prepared Statement of James G. Rickards
The Risks of Financial Modeling:
VaR and the Economic Meltdown
Introduction
Mr. Chairman, Mr. Ranking Member and Members of this subcommittee,
my name is James Rickards, and I want to extend my deep appreciation
for the opportunity and the high honor to speak to you today on a
subject of the utmost importance in the management of global capital
markets and the global banking system. The Subcommittee on
Investigations and Oversight has a long and distinguished history of
examining technology and environmental matters which affect the health
and well-being of Americans. Today our financial health is in jeopardy
and I sincerely applaud your efforts to examine the flaws and misuse in
financial modeling which have contributed to the impairment of the
financial health of our citizens and the country as a whole.
As a brief biographical note, I am an economist, lawyer and author
and currently work at Omnis, Inc. in McLean, VA where I specialize in
the field of threat finance and market intelligence. My colleagues and
I provide expert analysis of global capital markets to members of the
national security community including military, intelligence and
diplomatic directorates. My writings and research have appeared in
numerous journals and I am an Op-Ed contributor to the Washington Post
and New York Times and a frequent commentator on CNBC, CNN, Fox and
Bloomberg. I was formerly General Counsel of Long-Term Capital
Management, the hedge fund at the center of the 1998 financial crisis,
where I was principal negotiator of the Wall Street rescue plan
sponsored by the Federal Reserve Bank of New York.
Summary: The Problem with VaR
The world is now two years into the worst financial crisis since
the Great Depression. The IMF has estimated that the total lost wealth
in this crisis so far exceeds $60 Trillion dollars, more than the cost
of all of the wars of the 20th century combined. The list of causes and
culprits is long including mortgage brokers making loans borrowers
could not afford, investment bankers selling securities while
anticipating their default, rating agencies granting triple-A ratings
to bonds which soon suffered catastrophic losses, managers and traders
focused on short-term profits and bonuses at the expense of their
institutions, regulators acting complacently in the face of growing
leverage and imprudence and consumers spending and borrowing at non-
sustainable rates based on a housing bubble which was certain to burst
at some point. This story, sadly, is by now well known.
What is less well-known is that behind all of these phenomena were
quantitative risk management models which told interested parties that
all was well even as the bus was driving over a cliff. Mortgage brokers
could not have made unscrupulous loans unless Wall Street was willing
to buy them. Wall Street would not have bought the loans unless they
could package them into securities which their risk models told them
had a low risk of loss. Investors would not have bought the securities
unless they had triple-A ratings. The rating agencies would not have
given those ratings unless their models told them the securities were
almost certain to perform as expected. Transaction volumes would not
have reached the levels they did without leverage in financial
institutions. Regulators would not have approved that leverage unless
they had confidence in the risk models being used by the regulated
entities. In short, the entire financial edifice, from borrower to
broker to banker to investor to rating agency to regulator, was
supported by a belief in the power and accuracy of quantitative
financial risk models. Therefore an investigation into the origins,
accuracy and performance of those models is not ancillary to the
financial crisis; it is not a footnote; it is the heart of the matter.
Nothing is more important to our understanding of this crisis and
nothing is more important to the task of avoiding a recurrence of the
crisis we are still living through.
Unfortunately, we have been here before. In 1998, western capital
markets came to the brink of collapse, owing to the failure of a hedge
fund, Long-Term Capital Management, and a trillion dollar web of
counter-party risk with all of the major banks and brokers at that
time. Then Fed Chairman Alan Greenspan and Treasury Secretary Robert
Rubin called it the worst financial crisis in over 50 years. The
amounts involved and the duration of the crisis both seem small
compared to today's catastrophe, however, it did not seem that way at
the time. Capital markets really did teeter on the brink of collapse; I
know, I was there. As General Counsel of Long-Term Capital Management,
I negotiated the bail out which averted an even greater disaster at
that time. What is most striking to me now as I look back is how
nothing changed and how no lessons were applied.
The lessons were obvious at the time. LTCM had used fatally flawed
VaR risk models. LTCM had used too much leverage. LTCM had transacted
in unregulated over-the-counter derivatives instead of exchange traded
derivatives. The solutions were obvious. Risk models needed to be
changed or abandoned. Leverage needed to be reduced. Derivatives needed
to be moved to exchanges and clearinghouses. Regulatory oversight
needed to be increased.
Amazingly the United States Government did the opposite. The repeal
of Glass-Steagall in 1999 allowed banks to act like hedge funds. The
Commodities Futures Modernization Act of 2000 allowed more unregulated
derivatives. The Basle II accords and SEC regulations in 2004 allowed
increased leverage. It was as if the United States had looked at the
near catastrophe of LTCM and decided to double-down.
What reason can we offer to explain this all-in approach to
financial risk? Certainly the power of Wall Street lobbyists and
special interests cannot be discounted. Alan Greenspan played a large
role through his belief that markets could self-regulate through the
intermediation of bank credit. In fairness, he was not alone in this
belief. But none of this could have prevailed in the aftermath of the
1998 collapse without the assurance and comfort provided by
quantitative risk models. These models, especially Value-at-Risk, cast
a hypnotic spell, as science often does, and assured bankers, investors
and regulators that all was well even as the ashes of LTCM were still
burning.
What are these models? What is the attraction that allows so much
faith to be placed in them? And what are the flaws which lead to
financial collapse time and time again?
The term ``Value-at-Risk'' or VaR is used in two senses. One
meaning refers to the assumptions, models and equations which
constitute the risk management systems most widely used in large
financial institutions today. The other meaning refers to the output of
those systems, as in, ``our VaR today is $200 million'' which refers to
the maximum amount the institution is expected to lose in a single day
within some range of probability or certainty usually expressed at the
99 percent level. For purposes of this testimony, we will focus on VaR
in the first sense. If the models are well founded then the output
should be of some value. If not, then the output will be unreliable.
Therefore the proper focus of our inquiry should be on the soundness of
the models themselves.
Furthermore, any risk management system is only as good as the
assumptions behind it. It seems fair to conclude that based on a
certain set of assumptions, the quantitative analysts and computer
developers are able within reason to express those assumptions in
equations and to program the equations as computer code. In other
words, if the assumptions are correct then it follows that the model
development and the output should be reasonably correct and useful as
well. Conversely, if the assumptions are flawed then no amount of
mathematical equation writing and computer development will compensate
for this deficiency and the output will always be misleading or worse.
Therefore, the root of our inquiry into models should be an examination
of the assumptions behind the models.
In broad terms, the key assumptions are the following:
The Efficient Market Hypothesis (EMH): This assumes that investors and
market participants behave rationally from the perspective of wealth
maximization and will respond in a rational manner to a variety of
inputs including price signals and news. It also assumes that markets
efficiently price in all inputs in real time and that prices move
continuously and smoothly from one level to another based on these new
inputs.
The Random Walk: This is a corollary to EMH and assumes that since
markets efficiently price in all information, no investor can beat the
market consistently because any information which an investor might
rely on to make an investment decision is already reflected in the
current market price. This means than future market prices are
independent of past market prices and will be based solely on future
events that are essentially unknowable and therefore random.
Normally Distributed Risk: This is also a corollary to EMH and says
that since future price movements are random, their degree distribution
(i.e., relationship of frequency to severity of events) will also be
random like a coin toss or roll of the dice. This random or normal
degree distribution is also referred to as Gaussian and is most
frequently represented as a bell curve in which the large majority of
outcomes are bunched in a region of low severity with progressively
fewer outcomes shown in the high severity region. Because the curve
tails off steeply, highly extreme events are so rare as to be almost
impossible.
Value-at-Risk would be a fine methodology but for the fact that all
three of these assumptions are wrong. Markets are not efficient. Future
prices are not independent of the past. Risk is not normally
distributed. As the saying goes, ``Besides that, Mrs. Lincoln, how was
the play?'' Let's take these points separately.
Behavioral economics has done a masterful job of showing
experimentally and empirically that investors do not behave rationally
and that markets are not rational but are prone to severe shocks or
mood swings. Examples are numerous but some of the best known are risk
aversion (i.e., investors put more weight on avoiding risk than seeking
gains), herd mentality (i.e., investors buy stocks when others are
buying and sell when others are selling leading to persistent losses)
and various seasonal effects. Prices do not smoothly and continuously
move from one price level to the next but have a tendency to gap up or
down in violent thrusts depriving investors of the chance to get out
before large losses are incurred.
Similarly, prices to not move randomly but are highly dependent on
past price movements. In effect, relevant news will be discounted or
ignored for sustained periods of time until a kind of tipping point is
achieved at which point investors will react en masse to what is mostly
old news mainly because other investors are doing likewise. This is why
markets exhibit periods of low and high volatility in succession, why
markets tend to overshoot in response to fundamental news and why
investors can profit consistently by momentum trading which exploits an
understanding of these dynamics.
Finally, the normal distribution of risk has been known to be false
at least since the early 1960's when published studies of time series
of prices showed price distributions to be shaped in what is known as a
power curve rather than a bell curve. This has been borne out by many
studies since. A power curve has fewer low impact events than the bell
curve but has far more high impact events. This corresponds exactly to
the actual market behavior we have seen including frequent extreme
events such as the stock market crash of 1987, the Russian-LTCM
collapse of 1998, the dot corn bubble collapse of 2000 and the housing
collapse of 2007. Statistically these events should happen once every
1,000 years or so in a bell curve distribution but are expected with
much greater frequency in a power curve distribution. In short, a power
curve corresponds to market reality while a bell curve does not.
How is it possible that our entire financial system has come to the
point that it is risk managed by a completely incorrect system?
The Nobelist, Daniel Kahneman, tells the story of a Swiss Army
patrol lost in the Alps in a blizzard for days. Finally the patrol
stumbles into camp, frostbitten but still alive. The Commander asks how
they survived and the patrol leader replies, ``We had a map.'' The
Commander looks at the map and says, ``This is a map of the Pyrenees;
you were in the Alps.'' ``Yes,'' comes the reply; ``but we had a map.''
The point is that sometimes bad guidance is better than no guidance; it
gives you confidence and an ability to function even though your system
is flawed.
So it is with risk management on Wall Street. The current system,
based on the idea that risk is distributed in the shape of a bell
curve, is flawed and practitioners know it. Practitioners treat extreme
events as outliers and develop mathematical fixes. They call extreme
events fat tails and model them separately from the rest of the bell
curve. They use stress tests to gauge the impact of extreme events. The
problem is they never abandon the bell curve. They are like medieval
astronomers who believe the sun revolves around the earth and are
furiously tweaking their geocentric math in the face of contrary
evidence. They will never get this right; they need their Copernicus.
But the right map exists. It's called a power curve. It says that
events of any size can happen and extreme events happen more frequently
than the bell curve predicts. There is no need to treat fat tails as a
special case; they occur naturally on power curves. And power curves
are well understood by scientists because they apply to extreme events
in many natural and man-made systems from power outages to earthquakes.
Power curve analysis is not new. The economist, Vilfredo Pareto,
observed in 1906 that wealth distributions in every society conform to
a power curve; in effect, there is one Bill Gates for every 100 million
average Americans. Benoit Mandelbrot pioneered empirical analysis in
the 1960's that showed market prices move in power curve patterns.
So why have we gone down the wrong path of random walks and normal
distributions for the past 50 years? The history of science is filled
with false paradigms that gained followers to the detriment of better
science. People really did believe the sun revolved around the earth
for 2,000 years and mathematicians had the equations to prove it. The
sociologist, Robert K. Merton, called this the Matthew Effect from a
New Testament verse that says, ``For to those who have, more will be
given . . .'' The idea is that once an intellectual concept attracts a
critical mass of supporters it becomes entrenched while other concepts
are crowded out of the marketplace of ideas.
Another reason is that practitioners of bell curve science became
infatuated with the elegance of their mathematical solutions. The
Black-Scholes options formula is based on bell curve type price
movements. The derivatives market is based on variations of Black-
Scholes. Wall Street has decided that the wrong map is better than no
map at all--as long as the math is neat.
Why haven't scientists done more work in applying power curves to
capital markets? Some excellent research has been done. But one answer
is that power curves have low predictive value. Researchers approach
this field to gain an edge in trading and once the edge fails to
materialize they move on. But the Richter Scale, a classic power curve,
also has low predictive value. That does not make earthquake science
worthless. We know that 8.0 earthquakes are possible and we build
cities accordingly even if we cannot know when the big one will strike.
We can use power curve analysis to make our financial system more
robust even if we cannot predict financial earthquakes. One lesson of
power curves is that as you increase the scale of the system, the risk
of a mega-earthquake goes up exponentially. If you increase the value
of derivatives by a factor of 10, you may be increasing risk by a
factor of 10,000 without even knowing it. This is not something that
Wall Street or Washington currently comprehend.
Let's abandon the bell curve once and for all and accelerate
empirical research into the proper risk metrics of event distributions.
Even if predictive value is low, there is value in knowing the limits
of our knowledge. Understanding the way risk metastasizes with scale
might be lesson enough. It would offer a proper dose of humility to
those trying to supersize banks and regulators.
Detailed Analysis--History of VaR Failures
The empirical failures of the Efficient Market Hypothesis and VaR
are well known. Consider the October 19, 1987 stock market crash in
which the market fell 22.6 percent in one day; the December 1994
Tequila Crisis in which the Mexican Peso fell 85 percent in one week;
the September 1998 Russian-LTCM crisis in which capital markets almost
ceased to function; the March 2000 dot corn collapse during which the
NASDAQ fell 80 percent over 30 months, and the 9-11 attacks in which
the NYSE first closed and then fell 14.3 percent in the week following
its reopening. Of course, to this list of extreme events must now be
added the financial crisis that began in July 2007. Events of this
extreme magnitude should, according to VaR, either not happen at all
because diversification will cause certain risks to cancel out and
because rational buyers will seek bargains once valuations deviate
beyond a certain magnitude, or happen perhaps once every 1,000 years
(because standard deviations of this degree lie extremely close to the
x-axis on the bell curve which corresponds to a value close to zero on
the y-axis, i.e., an extremely low frequency event). The fact that all
of these extreme events took place in just over 20 years is completely
at odds with the predictions of VaR in a normally distributed paradigm.
Practitioners treated these observations not as fatal flaws in VaR
but rather as anomalies to be explained away within the framework of
the paradigm. Thus was born the ``fat tail'' which is applied as an
embellishment on the bell curve such that after approaching the x-axis
(i.e., the extreme low frequency region), the curve flattens to
intersect data points representing a cluster of highly extreme but not
so highly rare events. No explanation is given for what causes such
events; it is simply a matter of fitting the curve to the data (or
ignoring the data) and moving on without disturbing the paradigm. This
process of pinning a fat tail on the bell curve reached its apotheosis
in the invention of generalized auto-regressive conditional
heteroskedasicity or GARCH and its ilk, which are analytical techniques
for modeling the section of the degree distribution curve containing
the extreme events as a separate case and feeding the results of this
modeling into a modified version of the curve. A better approach would
have been to ask the question: if a normal distribution has a fat tail,
is it really a normal distribution?
A Gaussian distribution is not the only possible degree
distribution. One of the most common distributions in nature, which
accurately describes many phenomena, is the power curve which shows
that the severity of an event is inversely proportional to its
frequency with the proportionality expressed as an exponent. When
graphed on a double logarithmic scale, the power law describing
financial markets risk is a straight line sloping downward from left to
right; the negative exponent is the slope of the line.
This difference is not merely academic. Gaussian and power curve
distributions describe two entirely different phenomena. Power curves
accurately describe a class of phenomena known as nonlinear dynamical
systems which exhibit scale invariance, i.e., patterns are repeated at
all scales.
The field of nonlinear dynamical systems was enriched in the 1990s
by the concept of self-organized criticality. The idea is that actions
propagate throughout systems in a critical chain reaction. In the
critical state, the probability that an action will propagate is
roughly balanced by the probability that the original action will
dissipate. In the subcritical state, the probability of extensive
effects from the initial action is low. In the super-critical state, a
single minor action can lead to a catastrophic collapse. Such states
have long been observed in physical systems, e.g., nuclear chain
reactions in uranium piles, where a small amount of uranium is
relatively harmless (subcritical) and larger amounts can either be
carefully controlled to produce desired energy (critical), or can be
shaped to produce atomic explosions (supercritical).
The theory of financial markets existing in a critical state cannot
be tested in a laboratory or particle accelerator in the same fashion
as theories of atomic physics. Instead, the conclusion that financial
markets are a nonlinear critical state system rests on two non-
experimental bases; one deductive, one inductive. The deductive basis
is the ubiquity of power curves as a description of the behavior of a
wide variety of complex systems in natural and social sciences, e.g.,
earthquakes, forest fires, sunspots, polarity, drought, epidemiology,
population dynamics, size of cities, wealth distribution, etc. This is
all part of a more general movement in many natural and social sciences
from 19th and early 20th century equilibrium models to non-equilibrium
models; this trend has now caught up with financial economics.
The inductive basis is the large variety of capital markets
behavior which has been empirically observed to fit well with the
nonlinear paradigm. It is certainly more robust than VaR when it comes
to explaining the extreme market movements described above. It is
consistent with the fact that extreme events are not necessarily
attributable to extreme causes but may arise spontaneously in the same
initial conditions from routine causes.
While extreme events occur with much greater than normal frequency
in nonlinear critical state systems, these events are nevertheless
limited by the scale of the system itself. If the financial system is a
self-organized critical system, as both empirical evidence and
deductive logic strongly suggest, the single most important question
from a risk management perspective is: what is the scale of the system?
Simply put, the larger the scale of the system, the greater the
potential collapse with correlative macroeconomic and other real world
effects.
The news on this front is daunting. There is no normalized scale
similar to the Richter Scale for measuring the size of markets or the
size of disruptive events that occur within them, however, a few
examples will make the point. According to recent estimates prepared by
the McKinsey Global Institute, the ratio of world financial assets to
world GDP grew from 100 percent in 1980 to 200 percent in 1993 to 316
percent in 2005. Over the same period, the absolute level of global
financial assets increased from $12 trillion to $140 trillion. The
drivers of this exponential increase in scale are globalization,
derivative products, and leverage.
Globalization in this context is the integration of capital markets
across national boundaries. Until recently there were specific laws and
practices that had the effect of fragmenting capital markets into local
or national venues with little interaction. Factors included
withholding taxes, capital controls, protectionism, non-convertible
currencies, licensing, regulatory and other restrictions that tilted
the playing field in favor of local champions and elites. All of these
impediments have been removed over the past 20 years to the point that
the largest stock exchanges in the United States and Europe (NYSE and
Euronext) now operate as a single entity.
Derivative products have exhibited even faster growth than the
growth in underlying financial assets. This stems from improved
technology in the structuring, pricing, and trading of such instruments
and the fact that the size of the derivatives market is not limited by
the physical supply of any stock or commodity but may theoretically
achieve any size since the underlying instrument is notional rather
than actual. The total notional value of all swaps increased from $106
trillion to $531 trillion between 2002 and 2006. The notional value of
equity derivatives increased from $2.5 trillion to $11.9 trillion over
the same period while the notional value of credit default swaps
increased from $2.2 trillion to $54.6 trillion.
Leverage is the third element supporting the massive scaling of
financial markets; margin debt of U.S. brokerage firms more than
doubled from $134.58 billion to $293.2 billion from 2002 to 2007 while
the amount of total assets per dollar of equity at major U.S. brokerage
firms increased from approximately $20 to $26 in the same period. In
addition, leveraged investors invest in other entities which use
leverage to make still further investments. This type of layered
leverage is impossible to unwind in a panic.
There can be no doubt that capital markets are larger and more
complex than ever before. In a dynamically complex critical system,
this means that the size of the maximum possible catastrophe is
exponentially greater than ever. Recalling that systems described by a
power curve allow events of all sizes and that such events can occur at
any time, particularly when the system is super-critical, the
conclusion is inescapable that progressively greater financial
catastrophes of the type we are experiencing today should be expected
frequently.
The more advanced risk practitioners have long recognized the
shortcomings of using VaR in a normally distributed paradigm to compute
risk measured in standard deviations from the norm. This is why they
have added stress testing as an alternative or blended factor in their
models. Such stress testing rests on historically extreme events such
as the market reaction to 9-11 or the stock market crash of 1987.
However, this methodology has its own flaws since the worst outcomes in
a dynamically complex critical State system are not bounded by history
but are only bounded by the scale of the system itself. Since the
system is larger than ever, there is nothing in historical experience
that provides a guide to the size of the largest catastrophe that can
arise today. The fact that the financial crisis which began in July
2007 has lasted longer, caused greater losses and been more widespread
both geographically and sectorally than most analysts predicted or can
explain is because of the vastly greater scale of the financial system
which produces an exponentially greater catastrophe than has ever
occurred before. This is why the past is not a guide and why the
current crisis may be expected to produce results as severe as the
Great Depression of 1929-1941.
Policy Approaches and Recommendations
A clear understanding of the structures and vulnerabilities of the
financial markets points the way to solutions and policy
recommendations. These recommendations fall into the categories of
limiting scale, controlling cascades, and securing informational
advantage.
To explain the concept of limiting scale, a simple example will
suffice. If the U.S. power grid east of the Mississippi River were at
no point connected to the power grid west of the Mississippi River, a
nationwide power failure would be an extremely low probability event.
Either the ``east system'' or the ``west system'' could fail
catastrophically in a cascading manner but both systems could not fail
simultaneously except for entirely independent reasons because there
are no nodes in common to facilitate propagation across systems. In a
financial context, governments should give consideration to preventing
mergers that lead to globalized stock and bond exchanges and universal
banks. The first order efficiencies of such mergers are outweighed by
the risks of large-scale failure especially if those risks are not
properly understood and taken into account.
Another example will help to illustrate the relationship between
the scale of a system and extent of the greatest catastrophe possible
in that system. Imagine a vessel with a large hold. The hold is divided
into three equal sections separated by watertight bulkheads. If a hole
is punched in one section and that section is completely filled with
water, the vessel will still float. Now imagine the watertight
bulkheads are removed and the same hole is punched into the vessel. In
this case, the entire hold will fill with water and the vessel will
sink. In this example, the area of the hold can be thought of as the
relevant dynamic system. The sinking of the vessel represents the
catastrophic failure of the system. When the bulkheads are in place we
have three small systems. When the bulkheads are removed we have one
large system. By removing the bulkheads we increased the scale of the
system by a factor of three. But the likelihood of failure did not
increase by a factor of three; it went from practically zero to
practically 100 percent. The system size tripled but the risk of
sinking went up exponentially. By removing the bulkheads we created
what engineers call a ``single point of failure,'' i.e., one hole is
now enough to sink the entire vessel.
Something similar happened to our financial system between 1999 and
2004. This began with the repeal of Glass-Steagall in 1999 which can be
thought of as removing the watertight bulkheads separating commercial
banks and investment banks. This was exacerbated by the Commodities
Futures Modernization Act of 2000 which removed the prohibition on many
kinds of derivatives. This allowed banks to increase the scale of the
system through off-balance sheet transactions. Finally, in 2004, the
SEC amended the broker-dealer net capital rule in such a way that
allowed brokers to go well-beyond the traditional 15:1 leverage ratio
and to use leverage of 30:1 or more. All three of these events
increased the scale of the system by allowing regulated financial
institutions to enter new markets, trade new products and use increased
leverage. Using a power curve analysis, we see that while the scale of
the system was increased in a linear way (by a factor of three, five,
ten or fifty depending on the product) the risk was increasing in a
nonlinear way (by a factor of 100, 1000, or 10,000 depending on the
slope of the power curve). VaR models based on normal distributions
were reporting that risk was under control and sounding the all clear
signal because so much of the risk was offsetting or seen to cancel out
in the models. However, a power curve model would have been flashing a
red alert sign because it does not depend on correlations, instead it
sees risk as an emergent property and an exponential function of scale.
The fact that government opened the door to instability does not
necessarily mean that the private sector had to rush through the door
to embrace the brave new world of leveraged risk. For that we needed
VaR. Without VaR models to tell bankers that risk was under control,
managers would not have taken so much risk even if government rules
allowed them to do so. Self-interest would have constrained them
somewhat as Greenspan expected. But with VaR models telling senior
management that risk was contained the new government rules became an
open invitation to pile on massive amounts of risk which bankers
promptly did.
Our financial system was relatively stable from 1934-1999 despite
occasional failures of institutions (such as Continental Illinois Bank)
and entire sectors (such as the S&L industry). This 65-year period can
be viewed as the golden age of compartmented banking and moderate
leverage under Glass-Steagall and the SEC's original net capital rule.
Derivatives themselves were highly constrained by the Commodity
Exchange Act. In 1999, 2000 and 2004 respectively, all three of these
watertight bulkheads were removed. By 2006 the system was poised for
the most catastrophic financial collapse in history. While subprime
mortgage failures provided the catalyst, it was the scale of the system
itself which caused the damage. The catalyst could just as well have
come from emerging markets, commercial real estate or credit default
swaps. In a dynamically critical system, the catalyst is always less
important than the chain reaction and the reaction in this case was a
massive collapse.
The idea of controlling cascades of failure is, in part, a matter
of circuit breakers and pre-rehearsed crisis management so that nascent
collapses do not spin into full systemic catastrophes before regulators
have the opportunity to prevent the spread. The combination of diffuse
credit and layered leverage makes it infeasible to assemble all of the
affected parties in a single room to discuss solutions. There simply is
not enough time or condensed information to respond in real time as a
crisis unfolds.
One significant circuit breaker which has been discussed for over a
decade but which has still not been fully implemented is a
clearinghouse for all over-the-counter derivatives. Experience with
clearinghouses and netting systems such as the Government Securities
Clearing Corporation shows that gross risk can be reduced 90 percent or
more when converted to net risk through the intermediation of a
clearinghouse. Bearing in mind that a parametric decrease in scale
produces an exponential decrease in risk in a nonlinear system, the
kind of risk reduction that arises in a clearinghouse can be the single
most important step in the direction of stabilizing the financial
system today; much more powerful than bail outs which do not reduce
risk but merely bury it temporarily.
A clearinghouse will also provide informational transparency that
will allow regulators to facilitate the failure of financial
institutions without producing contagion and systemic risk. Such
failure (what Joseph Schumpeter called ``creative destruction'') is
another necessary step on the road to financial recovery. Technical
objections to clearinghouse implementation based on the non-uniformity
of contracts can be overcome easily through consensual contractual
modification with price adjustments upon joining the clearinghouse
enforced by the understanding that those who refuse to join will be
outside the safety net. Only by eliminating zombie institutions and
creating breathing room for healthy institutions with sound balance
sheets can the financial sector hope to attract sufficient private
capital to replace government capital and thus re-start the credit
creation process needed to produce sound economic growth.
Recently a number of alternative paradigms have appeared which not
only do not rely on VaR but rather assume its opposite and build models
that are more robust to empirical evidence and market price patterns.
Several of these approaches are:
Behavioral Economics--This field relies on insights into human behavior
derived from social science and psychology, in particular, the
``irrational'' nature of human decision-making when faced with economic
choices. Insights include risk aversion, herding, the presence or
absence of cognitive diversity and network effects among others. While
not summarized in a general theory and while not always amendable to
quantitative modeling, the insights of behavioral economics are
powerful and should be considered in weighing reliance on VaR-style
models which do not make allowance for subjective influences captured
in this approach.
Imperfect Knowledge Economics--This discipline (under the abbreviation
IKE) attempts to deal with uncertainty inherent in capital markets by
using a combination of Bayesian networks, link analysis, causal
inference and probabilistic hypotheses to fill in unknowns using the
known. This method is heavily dependent on the proper construction of
paths and the proper weighing of probabilities in each hypothesis cell
or evidence cell, however, used properly it can guide decision-making
without applying the straitjacket of VaR.
Econoahysics--This is a branch of financial economics which uses
insights gained from physics to model capital markets behavior. These
insights include nonlinearity in dynamic critical state systems the
concept of phase transitions. Such systems exhibit an unpredictably
deterministic nonlinear relationship between inputs and outputs (the
so-called ``Butterfly Effect'') and scale invariance which accords well
with actual time series of capital markets prices. Importantly, this
field leads to a degree distribution characterized by the power curve
rather than the bell curve with implications for scaling metrics in the
management of systemic risk.
It may be the case that these risk management tools work best at
distinct scales. For example, behavioral economics seems to work well
at the level of individual decision-making but has less to offer at the
level of the system as a whole where complex feedback loops cloud its
efficacy. IKE may work best at the level of a single institution where
the hypothesis and evidence cells can be reasonably well defined and
populated. Econophysics may work best at the systemic level because it
goes the furthest in its ability to model highly complex dynamics. This
division of labor suggests that rather than replacing VaR with a one-
size-fits-all approach, it may be best to adopt a nested hierarchy of
risk management approaches resembling the following:
While all of these approaches and others not mentioned here require
more research to normalize metrics and build general theories, they are
efficacious and robust alternatives to EMH and VaR and their
development and use can serve a stabilizing function since they have a
strong empirical basis unlike EMH and VaR.
In summary, Wall Street's reigning risk management paradigm
consisting of VaR using a normally distributed model combined with
GARCH techniques applied to the non-normal region and stress testing to
account for outliers is a manifest failure. It should be replaced at
the systemic level with the empirically robust model based on nonlinear
complexity and critical state dynamics as described by the power curve.
This method also points the way to certain solutions, most importantly
the creation of an over-the-counter derivatives clearinghouse which
will de-scale the system and lead to an exponential decrease in actual
risk. Such a clearinghouse can also be used to improve transparency and
manage failure in ways that can leave the system far healthier while
avoiding systemic collapse.
Importantly, if scale is the primary determinant of risk, as
appears to be the case in complex systems such as the financial
markets, then it follows that de-scaling the system is the simplest and
most effective way to manage risk. This does not mean that the totality
of the system needs to shrink, merely that it be divided into sub-
components with limited interaction. This has the same effect as
installing the watertight bulkheads referred to in the example above.
In this manner, severe financial distress in one sector does not
automatically result in contagion among all sectors.
This effective de-scaling can be accomplished with three reforms:
1. The enactment of a modernized version of Glass-Steagall with a
strict separation between commercial banking and deposit taking on the
one hand and principal risk taking in capital markets on the other.
2. Strict requirements for all derivative products to be traded on
exchanges subject to credit tests for firm memberships, initial margin,
variation margin, position limits, price transparency and netting.
3. Higher regulatory capital requirements and reduced leverage for
banks and broker-dealers. Traditional ratios of 8:1 for banks and 15:1
for brokers seem adequate provided off-balance sheet positions (other
than exchange traded contracts for which adequate margin is posted) be
included for this purpose.
These rules can be implemented directly and do not depend on the
output of arcane and dangerous models such as VaR. Instead, they derive
from another proven model, the power curve, which teaches that risk is
an exponential function of scale. By de-scaling, we radically reduce
risk and restore stability to individual institutions and to the system
as a whole.
Biography for James G. Rickards
James G. Rickards is Senior Managing Director for Market
Intelligence at Omnis, Inc., a scientific consulting firm in McLean,
VA. He is also Principal of Global-I Advisors, LLC, an investment
banking firm specializing in capital markets and geopolitics. Mr.
Rickards is a seasoned counselor, investment banker and risk manager
with over thirty years experience in capital markets including all
aspects of portfolio management, risk management, product structure,
regulation and operations. Mr. Rickards's market experience is focused
in alternative investing and derivatives in global markets.
Mr. Rickards was a first hand participant in the formation and
growth of globalized capital markets and complex derivative trading
strategies. He held senior executive positions at sell side firms
(Citibank and RBS Greenwich Capital Markets) and buy side firms (Long-
Term Capital Management and Caxton Associates) and technology firms
(OptiMark and Omnis). Mr. Rickards has participated directly in many of
the most significant financial events over the past 30 years including
the release of U.S. hostages in Iran (1981), the Stock Market crash of
1987, the collapse of Drexel (1990), the Salomon Bros. bond trading
scandal (1991) and the LTCM financial crisis of 1998 (in which Mr.
Rickards was the principal negotiator of the government-sponsored
rescue). He has founded several hedge funds and fund-of-funds. His
advisory clients include private investment funds, investment banks and
government directorates. Since 2001, Mr. Rickards has applied his
financial expertise to missions for the benefit of the U.S. national
security community.
Mr. Rickards is licensed to practice law in New York and New Jersey
and the Federal Courts. Mr. Rickards has held all major financial
industry licenses including Series 3 (National Commodities Futures),
Series 7 (General Securities Representative), Series 24 (General
Securities Principal), Series 30 (Futures Manager) and Series 63.
Mr. Rickards has been a frequent speaker at conferences sponsored
by bar associations and industry groups in the fields of derivatives
and hedge funds and is active in the International Bar Association. He
has been the interviewed in The Wall Street Journal and on CNBC, Fox,
CNN, NPR and C-SPAN and is an OpEd contributor to the New York Times
and the Washington Post.
Mr. Rickards is a graduate school visiting lecturer in finance at
the Kellogg School and the School of Advanced International Studies. He
has delivered papers on econophysics at the Applied Physics Laboratory
and the Los Alamos National Laboratory. Mr. Rickards has written
articles published in academic and professional journals in the fields
of strategic studies, cognitive diversity, network science and risk
management. He is a member of the Business Advisory Board of Shariah
Capital, Inc., an advisory firm specializing in Islamic finance and is
a member of the International Business Practices Advisory Panel to the
Committee on Foreign Investment in the United States (CFIUS) Support
Group of the Director of National Intelligence.
Mr. Rickards holds the following degrees: LL.M. (Taxation) from the
New York University School of Law; J.D. from the University of
Pennsylvania Law School; M.A. in international economics from the
School of Advanced International Studies, Washington DC; and a B.A.
degree with honors from the School of Arts & Sciences of The Johns
Hopkins University, Baltimore, MD.
Chairman Miller. Thank you, Mr. Rickards. I did practice
repeatedly saying ``Taleb.'' I should have practiced
``Rickards'' as well.
Mr. Whalen.
STATEMENT OF MR. CHRISTOPHER WHALEN, MANAGING DIRECTOR,
INSTITUTIONAL RISK ANALYTICS
Mr. Whalen. Thank you, Mr. Chairman. I am going to just
summarize a couple points further to my written testimony. You
will notice in my comments I focused on the distinction between
subjectivity and objectivity, and I think this committee is
probably better placed to understand those distinctions than
most of the other panels in the Congress.
You know, we have seen over the last 100 years in this
country a shift in our financial world from focusing on current
performance of companies and financial institutions to focusing
on predicting the future. This is very well illustrated in the
Graham and Dodd volume, Securities Analysis, in chapter 38
where they talk about new era investing, and I urge you to
reread that if you have never done so before.
The bottom line to me as someone who has worked in the
industry as a supervisor and a trader and investment banker, is
that when you use assumptions and models, you have already
stepped off the deep edge, you know, the deep end of the pool,
and there's no water in the pool. You essentially are in the
world of speculation, and you have left the world of investing.
Why do I say this? Well, if we use the same rules that govern
the assumptions that go into most VaR models to design
airplanes and buildings and dams, all of these physical
structures would fail, because they violate the basic rules of
scientific method that the Members of this committee know very,
very well. I would submit to you that if we are going to allow
our financial system to design products that are based on
assumptions rather than hard data, than we are in big trouble.
My firm has over the last seven years shunned the quantitative
world. Our entire methodology is focused on benchmarking the
current performance of banks, and taking observations about
that current performance that may suggest what they are going
to do in the future. But we don't guess, we don't speculate. We
have almost 20,000 retail customers who use the bank monitor to
track the safety and soundness of their institution. It is an
entirely mechanical survey process. We stress-test every bank
in the United States the same way, whether it is J.P. Morgan or
Cullen/Frost Bank in Texas. We ask the same question, how did
you do this quarter, and we compare it to 1995, which was a
nice, boring year.
The second point I would like to make is that I think a big
part of the problem is that we allowed the economist profession
to escape from the world of social sciences, and enter into an
unholy union with commission-driven dealers in the securities
market. Your colleague, Mr. Broun, said earlier that economists
can't make up their mind. Well, yes, they can. When they are
working in the securities business they have no trouble making
up their mind. They offer opinions and hypotheses and `what if'
or `I want' in regards to the creation of a security. This is a
big problem. I wouldn't let most economists park my car, and
the problem is not that they are not smart people, not that
they are not interesting people, but they live in the world of
supposition rather than the world of fact, and again, their
methodologies are not governed by the iron rules that you find
in physics or chemistry or any of the other physical sciences
where you have to live by those rules. You can't come up with
some neat concept and say to your colleagues, hey, look at me,
or hey, look at this new CDO I designed, and then go out and
sell that security to the public.
I think it all comes down at the end of the day to what
kind of economy do we want. There is an old-fashioned American
concept called `fair dealing' that I spent a lot of time
talking in my testimony to the Senate Banking Committee earlier
this year, and it comes from the Greeks, the concept of
proportional requital. One person gives value, the other person
receives value. The problem with products like credit default
swaps, is that they are entirely speculative. There is no
visible underlying market for single-name credit default swaps
really. The corporate bonds that are supposedly the derivative
or the basis for the derivative are fairly liquid and not a
very good source of pricing information, so we use models and
we then sell these securities to anyone and everyone. I would
submit that that is unfair, and it goes against the basic grain
of American society that we are a fair and transparent nation.
So bottom line to me is, if you want to fix the problem, I
think we have got to reimpose not higher capital requirements
on banks that are out of control, and which take risks that no
one can really quantify. I think what we have to do is reimpose
restrictions on their risk taking and get them to the point
where an eight percent capital assets ratio makes sense again,
because it clearly doesn't now. Does anybody really think we
can get the private sector to double the capital of J.P. Morgan
when their equity returns are going to be falling for the next
couple of years? The only entity that would fund that
opportunity would be a government, so what we are really saying
is that these are GSEs. I think we have got to come back almost
to the Glass-Steagall-era draconian division between the
utility function of a bank and the transactional function of
hedge funds, broker dealers, whatever, and that latter group
can do whatever they want.
So let me stop there, and I look forward to your questions.
[The prepared statement of Mr. Whalen follows:]
Prepared Statement of Christopher Whalen
Chairman Miller, Congressman Broun, Members of the Committee, my
name is Christopher Whalen and I live in the State of New York. I work
in the financial community as an analyst and a principal of a firm that
rates the performance of commercial banks.\1\ Thank you for inviting my
comments today on this important subject.
---------------------------------------------------------------------------
\1\ Mr. Whalen is a co-founder of Institutional Risk Analytics, a
Los Angeles unit of Lord, Whalen LLC that publishes risk ratings and
provides customized financial analysis and valuation tools.
---------------------------------------------------------------------------
The Committee has asked witnesses to comment on the topic of ``The
Risks of Financial Modeling: VaR and the Economic Meltdown.'' The
comments below reflect my own views, as well as comments from my
colleague and business partner Dennis Santiago, and others in the
financial and risk management community.
By way of background, our firm provides ratings for assessing the
financial condition of U.S. banks and commercial companies. We build
the analytical tools that we use to support these rating activities and
produce reports for thousands of consumer and professional users.
We use mathematical tools such as models to explore the current
financial behavior of a given subject. In the course of our work, we
use these tools to make estimates, for example, as to the maximum
probable loss in a bank's loan portfolio through an economic cycle or
the required Economic Capital for a financial institution. Models help
us understand and illustrate how the financial condition of a bank or
other obliger have changed and possibly will change in the future.
But in all that we at Institutional Risk Analytics do in the world
of ratings and financial analysis, we do our best to separate objective
measures based upon empirical observations, and subjective analyses
that employ speculative assumptions and directives which are often
inserted into the very ground rules for the analysis process itself.
The difference between subjectivity and objectivity in finance has
significant implications for national policy when it comes to financial
markets and institutions.
I strongly suggest to the Committee that they bear the distinction
between objective and subjective measures in mind when discussing the
use of models in finance. Obtaining a better understanding of the role
of inserting subjectivity into models is critical for distinguishing
between useful deployments of modeling to manage risk and situations
where models are the primary failure pathway towards creating systemic
risk and thus affect economic stability and public policy.
Used as both a noun and a verb, the word ``model'' has become the
symbol for the latest financial crisis because of the use, or more
precisely, the misuse of such simulations to price unregistered,
illiquid securities such as sub-prime mortgage backed securities and
derivatives of such securities. The anecdotal cases where errant models
have led to mischief are many and are not limited to the world of
finance alone.
The Trouble with Models
The problem is not with models themselves. The trouble happens when
they are (a) improperly constructed and then (b) deliberately
misapplied by individuals working in the financial markets.
In the physical sciences, models can be very usefully employed to
help analysts understand complex systems such as disease, buildings and
aircraft. These models tend to use observable data as inputs, can be
scientifically validated and are codified in a manner that is
transparent to all involved in the process. Models used in the physical
world share one thing in common that financial models do not: they are
connected to and are confirmed or refuted by the physical world they
describe.
Financial models, on the other hand, are all intellectual
abstractions designed to manipulate arbitrarily chosen, human invented
concepts. The chief reason for this digression from the objective use
of models observed in the physical sciences is the injection of
economics into the world of finance. Whereas financial models were once
merely arithmetic expressions of expected cash flows, today in the
world of financial economics, models have become vehicles for rampant
speculation and outright fraud.\2\
---------------------------------------------------------------------------
\2\ See ``New Hope for Financial Economics: Interview with Bill
Janeway,'' The Institutional Risk Analyst, November 17, 2008.
---------------------------------------------------------------------------
In the world of finance, modeling has been an important part of the
decision-making toolkit of executives and analysts for centuries,
helping them to understand the various components in a company or a
market and thereby adjust to take advantage of the circumstances. These
decision analysis models seek to measure and report on key indicators
of actual performance and confirm the position of the entity with
respect to its' competitive environment. For instance, the arithmetic
calculation of cash flows adheres to the scientific method of
structures and dynamics, and is the foundation of modern finance as
embodied by the great theorists such as Benjamin Graham and David Dodd.
At our firm, we employ a ``measure and report'' model called The
IRA Bank Monitor to survey and stress test all FDIC insured banks each
quarter. By bench-marking the performance of banks with a consistent
set of tests, we are able to not only characterize the relative safety
and soundness of each institution, but can drawn reasonable inferences
about the bank's future performance.
But when the world of finance marries the world of outcome driven
economics--the world of ``what if'' and ``I want''--models cease to be
mechanistic tools for validating current outcomes with hard data and
assessing a reasonable range of possible future events. Instead models
become enablers for speculation, for the use of skillful canards and
legal subterfuge that ultimately cheat investors and cause hundreds of
billions of dollars in losses to private investors and insured
depository institutions.
Take the world of mortgage backed securities or MBS. For decades
the investment community had been using relatively simple models to
predict the cash flow of MBS in various interest rate scenarios. These
predictions have been relatively simple and are validated against the
monthly mortgage servicer data available to the analyst community. The
MBS securitization process was simple as well. A bank would sell
conforming loans to GNMA and FNMA, and sell inferior collateral to a
handful of investment banks on Wall Street to turn in the loans into
private MBS issues.
At the beginning of the 1990's, however, Wall Street's private MBS
secret sauce escaped. A firm named Drexel, Burnham, Lambert went
bankrupt and the bankruptcy court sold copies of Drexel's structured
finance software to anyone and everyone. It eventually wound up in the
hands of the mortgage issuers themselves. These banks and non-banks
naturally began to issue private MBS by themselves and discovered they
could use the mathematics of modeling to grow their mortgage conduit
businesses into massive cash flow machines. When brought to market,
these private MBS were frequently under-collateralized and could
therefore be described as a fraud.
Wall Street, in turn, created even more complex modeling systems to
squeeze even more profits from the original MBS template. The expanding
bubble of financial innovation caught the eye of policy-makers in the
Congress, who then created political models envisioning the possibility
that ``innovation'' could be used to make housing accessible to more
Americans.
Spurred on to chase the ``policy outcome'' of affordable housing,
an entire range of deliberately opaque and highly leveraged financial
instruments were born with the full support of Washington, the GSEs and
the Congress. Their purpose now was to use the alchemy of financial
modeling to create the appearance of mathematical safety out of
dangerous toxic ingredients. Wall Street firms paid the major rating
agencies to award ``AAA'' ratings to derivative assets that were
ultimately based on sub-prime mortgage debt. And the stage was set for
a future economic disaster.
In the case of sub-prime toxic waste, the models became so complex
that all transparency was lost. The dealers of unregulated,
unregistered complex structured assets used proprietary models to price
and sell deals, but since the ``underlying'' for these derivative
securities was invisible, none of the investment or independent ratings
community could model the security. There was no validation, no market
discipline. Buy Side customers were dependent upon the dealer who sold
them the toxic waste for valuation. The dealers that controlled the
model often time would not even make a market in the security.
Clearly we have now many examples where a model or the pretense of
a model was used as a vehicle for creating risk and hiding it. More
important, however, is the role of financial models for creating
opportunities for deliberate acts of securities fraud. These acts of
fraud have caused hundreds of billions of dollars in losses to
depository institutions and investors.
Whether you talk about toxic mortgage assets or credit default
swaps, the one common element that the misuse of models seems to
contain is a lack of a visible underlying market against which to judge
or ``mark'' the model. Indeed, the use of models in a subjective
context seems to include the simulation of a nonexistent market as the
primary role for the financial model.
In single-name credit default swaps or ``CDS'' for example, there
is often insufficient trading in the supposed underlying corporate debt
security to provide true price discovery. In the case of CDS on complex
structured assets, there is no underlying market to observe at all. The
subjective model becomes the market in terms of pricing the security.
In the spring of 2007, however, the fantasy land consensus that
allowed people to believe that a model is a market came undone. We have
been dealing with the consequences of the decisions that originally
built the house of cards since that time.
An Objective Basis for Finance and Regulation
The term ``model'' as it applies to finance can be a simulation of
reality in terms of predicting future financial outcomes. The author
Nassim Taleb, who is appearing at this hearing, says the term ``VaR''
or value at risk describes a statistical estimate of ``the expected
maximum loss (or worst loss) over a target horizon within a given
confidence interval.'' \3\
---------------------------------------------------------------------------
\3\ See Taleb, Nassim, ``Against Value-at-Risk: Nassim Taleb
Replies to Philippe Jorion,'' 1997.
---------------------------------------------------------------------------
VaR models and similar statistical methods pretend to estimate the
largest possible loss that an investor might experience over a given
period of time to a given degree of certainty. The use of VaR type
models, including the version embedded in the Basel II agreement,
involves a number of assumptions about risk and outcomes that are
speculative. More important, the widespread use of these statistical
models for risk management suggest that financial institutions are
subject to occasional ``Black Swans'' in the form of risk events that
cannot be anticipated.
We take a different view. We don't actually believe there is such a
thing as a ``Black Swan.'' Our observations tell us that a more likely
explanation is that leaders in finance and politics simply made the
mistake of, again, believing in what were in fact flawed models and
blinded themselves to what should have been plainly calculable
innovation risks destined to be unsustainable. Or worse, our leaders in
Washington and on Wall Street decided to be short sighted and not care
about the inevitable debacle.
We suggest that going forward our national interest needs to demand
a higher standard of tangible proof from ``outcome designers'' of
public policies. If financial markets and the models used to describe
them are limited to those instruments that can be verified objectively,
then we no longer need to fear from the ravages of Black Swans or
systemic risk. The source of systemic risk in the financial markets is
fear born from the complexity of opaque securities for which there is
no underlying basis. The pretext for issuing these ersatz securities
depends on subjectivity injected into a flawed model.
If we accept that the sudden change in market conditions or the
``Black Swan'' event that Taleb and other theorists have so elegantly
described arises from a breakdown in prudential regulation and basic
common sense, and not from some unknowable market mechanism, then we no
longer need to fear surprises or systemic risk. We need to simply
ensure that all of the financial instruments in our marketplace have an
objective basis, including a visible, cash basis market that is visible
to all market participants. If investors cannot price a security
without reference to subjective models, then the security should be
banned from the U.S. markets as a matter of law and regulation. To do
otherwise is to adopt deception as the public policy goal of the U.S.
when it comes to financial markets regulation.
As Graham and Dodd wrote nearly a century ago, the more speculative
the inputs the less the analysis matters. Models only have real value
to society when their workings are disciplined by the real world. When
investors, legislators and regulators all mistook models for markets,
and even accepted such speculations as a basis for regulating banks and
governing over-the-counter or OTC markets for all types of securities,
we as a nation were gambling with our patrimony. If the Committee and
the Congress want to bring an end to the financial crisis, we must
demand higher standards from our citizens who work in and regulate our
financial markets.
As we discussed in a commentary last month, ``Systemic Risk: Is it
Black Swans or Market Innovations?,'' published in The Institutional
Risk Analyst, ``were the failures of Bear Stearns, Lehman Brothers,
Washington Mutual or the other ``rare'' events really anomalous? Or are
we just making excuses for our collective failure to identify and
manage risk? A copy of our commentary follows this testimony. I look
forward to your questions.
Systemic Risk: Is it Black Swans
or Market Innovations?
August 18, 2009
``Whatever you think you know about the distribution changes
the distribution.''
Alex Pollock
American Enterprise Institute
In this week's issue of The IRA, our friend and colleague Richard
Alford, a former Fed of New York economist, and IRA founders Dennis
Santiago and Chris Whalen, ask us whether we really see Black Swans in
market crises or our own expectations. Of note, we will release our
preliminary Q2 Banking Stress Index ratings on Monday, August 24, 2009.
As with Q1, these figures represent about 90 percent of all FDIC
insured depositories, but exclude the largest money center banks (aka
the ``Stress Test Nineteen''), thus providing a look at the state of
the regional and community banks as of the quarter ended June 30, 2009.
Click here to register for The Institutional Risk Analyst.
Many popular explanations of recent financial crises cite ``Black
Swan'' events; extreme, unexpected, ``surprise'' price movements, as
the causes of the calamity. However, in looking at our crisis wracked
markets, we might consider that the Black Swan hypothesis doesn't fit
the facts as well an alternative explanation: namely that the
speculative outburst of financial innovation and the artificially low,
short-run interest rate environment pursued by the Federal Open Market
Committee, combined to change the underlying distribution of potential
price changes. This shift in the composition of the distribution made
likely outcomes that previously seemed impossible or remote. This shift
in possible outcomes, in turn, generated surprise in the markets and
arguably led to the emergence of ``systemic risk'' as a metaphor to
explain these apparent ``anomalies.''
But were the failures of Bear Stearns, Lehman Brothers, Washington
Mutual or the other ``rare'' events really anomalous? Or are we just
making excuses for our collective failure to identify and manage risk?
The choice of which hypothesis to ultimately accept in developing
the narrative description of the causation of the financial crisis has
strategic implications for understanding as well as reducing the
likelihood of future crisis, including the effect on the safety and
soundness of financial institutions. To us, the hard work is not trying
to specifically limit the range of possibilities with artificial
assumptions, but to model risk when you must assume as a hard rule,
like the rules which govern the physical sciences, that the event
distribution is in constant flux.
If we as financial and risk professional are serious in claims to
model risk proactively, then change, not static assumptions, must be
the rule in terms of the possible outcomes. Or ``paranoid and nimble''
in practical terms. After all, these modeling exercises ultimately
inform and support risk assumptions for decisions that are used in
value-at-risk (VaR) assessments for investors and for capital adequacy
bench-marking for financial institutions.
Even before the arrival of Benoit Mandelbrot in the 1960s,
researchers had observed that distributions of price changes in various
markets were not normally distributed. The observed distributions of
price changes had fatter tails than the normal distribution. Nassim
Nicolas Taleb, author of The Black Swan and Fooled by Randomness, and
others have dubbed significantly larger extreme price moves than those
predicted by a normal distribution as ``Black Swans.'' Indeed, Taleb
and others have linked Black Swan price change events to the recent
financial crisis, suggesting in effect that we all collectively
misunderstood on which side of the distribution of possible risk
outcomes we stood.
The argument is as follows: Current risk management and derivative
pricing regimes are based upon normal distributions. Price movements in
the recent financial crises were unpredictable/low probability events
that were also greater than predicted by normal distribution models.
Hence our collective failure to anticipate Black Swan events is
``responsible'' for the recent crises as mis-specified risk management
models failed due to fatter than normal tails.
The alternative explanation, however, links the extreme price
movements not to aberrations with respect to a stable, observable mean,
but instead to the activation of alternate stable means as a result of
jumping discontinuously through tipping points--much in the same way
particles jump quantum levels in energy states when subjected to the
cumulative effects of energy being added to or removed from their
environments. These tipping points are as predictable as the annual
migrations of ducks. Swans, alas, rarely migrate, preferring to stay in
their summer feeding grounds until the water freezes, then move only
far enough to find open water. Sound familiar?
Force feed a system with enough creative energy via permissive
public policies and the resulting herd behaviors, and the system will
change to align around these new norms, thereby erasing the advantages
of the innovators and creating unforeseen hazards. ``Advances'' such as
OTC derivatives and complex structured assets, and very accommodating
Fed interest rate policy, resulted in unprecedented leverage and
maturity mismatches by institutions and in markets that are the perfect
quantum fuel to brew such change.
While the exact timing of each tipping point and magnitude of the
crises remains somewhat inexact, the waves of change and the ultimate
crisis borne shift are broadly predictable. The probabilities attached
to extreme price moves are calculable as the cost of deleveraging an
accumulation of innovation risk that must be shed as the system
realigns. The ``Black Swan'' approach assumes a stable distribution of
price changes with fatter than ``normal'' tails. The alternative posits
that the distribution of possible price changes was altered by
innovation and the low cost of leverage. It also posits that the new
distributions allowed, indeed require, more extreme price movements.
Two examples will illustrate the alternative hypothesis.
Once upon a time, the convertible bond market was relatively quiet.
The buy side was dominated by real money (unleveraged) players who
sought the safety of bonds, but were willing to give up some return for
some upside risk (the embedded equity call option).
More recently the market has been dominated by leveraged hedge
funds doing convertible bond arbitrage. They bought the bonds, hedging
away the various risks. In response to the advent of the arbitrageurs,
the spread between otherwise similar conventional and convertible bonds
moved to more accurately reflect the value of the embedded option and
became less volatile.
When the financial crises hit, however, arbitrageurs were forced to
liquidate their positions as losses mounted and it became difficult to
fund the leveraged positions. Prices for convertible bonds declined and
for a period were below prices for similar conventional bonds--
something that had been both unheard of and considered impossible as
the value of an option cannot be negative.
Was this a Black Swan type event, or had the market for convertible
bonds and the underlying distribution of price changes, been altered?
The mean spread between otherwise similar conventional and convertible
bonds had changed. The volatility of the spread had changed. Forced
sales and the public perception of possible future forced sales
generated unprecedented behavior of the heretofore stable spread. The
emergence and then dominance of leveraged arbitrage positions altered
the market in fundamental ways. What had not been possible had become
possible.
Now consider bank exposures to commercial real estate. Numerous
financial institutions, hedge funds (e.g., at Bear Stearns), sellers of
CDS protection (e.g., AIG) and banks (many of them foreign as reflected
in the Fed swap lines with foreign central banks) suffered grievous
losses when the real estate bubble popped. Much of these losses remain
as yet unrealized.
As investors and regulators demanded asset-write downs and loss
realization, many of these institution expressed dismay. They had
stressed tested their portfolios, the large banks complained, often
with the support of regulators. The large banks thought their
geographically diversified portfolios of MBSs immunize them from falls
in real estate prices as the US had experienced regional, but never
(except for the 1930s) nationwide declines in housing prices. These
sophisticated banks incorporated that assumption into their stress test
even as they and the securitization process were nationalizing--that
is, changing--the previously regional and local mortgage markets.
Was the nationwide decline in housing prices an unpredictable Black
Swan event or the foreseeable result of lower lending standards, a
supportive interest rate environment, and financial innovation the led
to the temporary nationalization of the mortgage market? Risk
management regimes failed and banks have been left with unrealized
losses that still threaten the solvency of the entire system in Q3
2009.
However useful or necessary ``normal'' statistical measures such as
VaR might be, it will not be sufficient to insulate institutions or the
system from risk arising from rapidly evolving market structures and
practices. Furthermore, insofar as models such as VaR, which are now
enshrined in the bank regulatory matrix via Basel II, were the binding
constraint on risk taking, it acted perversely, allowing ever greater
leverage as leveraged trading acted to reduce measured volatility!
Remember, the convertible bond market at first looked placid as a lake
as leverage grew--but then imploded in a way few thought possible. Is
this a Black Swan event or a failure of the stated objectives of risk
management and prudential oversight?
We all know that risk management systems based solely on analysis
of past price moves will at some point fall if financial markets
continue to change. The problem with current risk management systems
cannot be fixed by fiddling with VaR or other statical models. Risk
management regimes must incorporate judgments about the evolution of
the underlying markets, distribution of possible price changes and
other dynamic sources of risk.
Indeed, as we discussed last week (``Are You Ready for the Next
Bank Stress Tests''), this is precisely why IRA employs quarterly
surveys of bank stress tests to benchmark the US banking industry.
Think of the banking industry as a school of fish, moving in generally
the same direction, but not uniformly or even consistently. There is
enormous variation in the past of each member of the school, even
though from a distance the group seems to move in unison.
Stepping back from the narrow confines of finance for a moment,
consider that the most dramatic changes in the world are arguably
attributable to asymmetric confluences of energy changing the direction
of human history. It's happened over and over again. The danger has and
always will be the immutable law of unintended consequences, which
always comes back to bite the arrogant few who believe they can control
the future outcome. And it is always the many of us who pay the price
for these reckless leaps of faith.
If the recent financial crises were truly highly infrequent random
events, then any set of policies that can continuously prevent their
reoccurrence seemingly will be very expensive in terms of idle capital
and presumably less efficient markets required to avoid them. If, on
the other hand, the crisis was the result of financial innovation and
the ability to get leveraged cheaply, then society need not
continuously bare all the costs associated with preventing market
events like the bursting of asset bubbles.
Policy-makers would like everyone to believe that the recent crises
were random unpredictable Black Swan events. How can they be blamed for
failing to anticipate a low probability, random, and unpredictable
event? If on the other hand, the crises had observable antecedents,
e.g., increased use of leverage, maturity mismatches, near zero default
rates, and spikes in housing price to rental rates and housing price to
income ratios, then one must ask: why policy-makers did not connect the
dots, attach significant higher than normal probabilities to the
occurrence of severe financial disturbances, and fashion policies
accordingly? Ultimately, that is a question that Ben Bernanke and the
rest of the federal financial regulatory community still have yet to
answer.
Questions? Comments? info@institutionalriskanalytics.com
Biography for Christopher Whalen
Christopher is co-founder of Institutional Risk Analytics, the Los
Angeles based provider of risk management tools and consulting services
for auditors, regulators and financial professionals. Christopher leads
IRA's risk advisory practice and consults for global companies on a
variety of financial and regulatory issues. He is a Fellow of the
Networks Financial Institute at Indiana State University. Christopher
volunteers as a regional director of Professional Risk Managers
International Association (www.prmia.org) and is a board adviser to I-
OnAsia Limited (www.ionasia.com.hk), a global business security and
risk consultancy based in Hong Kong. Christopher currently edits The
Institutional Risk Analyst, a weekly news report and commentary on
significant developments in and around the global financial markets.
Christopher has testified before the Congress and the SEC on a variety
of issues and contributes articles and commentaries to publications
such as The International Economy, American Banker and The Big Picture.
Chairman Miller. Thank you.
Dr. Colander.
STATEMENT OF DR. DAVID COLANDER, CHRISTIAN A. JOHNSON
DISTINGUISHED PROFESSOR OF ECONOMICS, MIDDLEBURY COLLEGE
Dr. Colander. Mr. Chairman, thanks for the opportunity to
testify. I am Dave Colander, the Christian A. Johnson
Distinguished Professor of Economics at Middlebury College. I
was invited here because I was one of the authors of the Dahlem
Report in which we chided the economics profession for its
failure to warn society about the impending financial crisis.
Some non-economists have blamed the financial crisis on
economists' highly technical models. My argument is that the
problem isn't the models, the problem is the way the economic
models are used, and I think a number of the other panelists
have made that point. Where I am going to lead or go with that
is that the issue goes much deeper than just with VaR and the
various models you are looking at, and it goes very much to the
general arguments about science and technology and the way in
which economists approach problems, and I think, you know, Mr.
Whalen had it directly right: we live in the world of
supposition. Why? Because that is what our incentives are. We
write articles. We advance through writing articles, we don't
advance by designing something positive. If we are working for
a business, we do, but within academics it is very much
directed towards, you know, sort of what can we publish, and so
I think Value-at-Risk models are part of a much broader
economic problem, you know, sort of in terms of what economists
accept and how they go about doing what they are doing.
An example I want to give is really about macroeconomics,
you know, sort of in the dominant model in macroeconomics,
which is the dynamic stochastic general equilibrium (DSGE)
model, which is a big model designed very much along the same
lines about efficient markets. It sort of took efficient
markets and said, what if we had efficient markets in the
entire economy? To get that model, you have to assume there is
one individual, because we can't solve it unless there is only
one individual. We have to assume that person is globally
rational, understands everything and he has complete knowledge
in looking into the infinite future, and then we can actually
solve it for a very small case.
By definition, this model rules out strategic coordination
problems. What would happen if somebody else did something
else? That is obviously the likely cause of the recent crisis,
but it was simply assumed away in the macroeconomic model and
that macroeconomic model has been dominant for the last 30
years and has been funded by NSF, the research, you need to be
looking into that.
If the DSGE model had been seen as an aid to common sense,
it could have been a useful model. It improved some of the
problems that some earlier models had. But for a variety of
sociological reasons that I don't have time to go into here, a
majority of macroeconomists started believing the DSGE model
was useful, not just as an aid to our understanding but as the
model of the macroeconomy. As that DSGE model became dominant,
really important research on the whole set of broader non-
linear and complex dynamic models that would have really served
some foundation for thinking about these issues just wasn't
done. It just wasn't allowed. You couldn't get anything
published on it in the main macro journals.
Similar developments occurred with the efficient market
finance models, which made assumptions very similar to the DSGE
model. And so, again, at first these served a useful purpose.
They led to technological advances in risk management and
financial markets. But as happened in macro, the users of these
financial models forgot that the models provide, at best, half-
truths. They stopped using models with common sense and
judgment. What that means is that warning labels should be put
on models, and that should be in bold print, `these models are
based on assumptions that do not fit the real world and thus
these models should be not relied on very heavily.' Those
warning labels haven't been there.
How did something so stupid like this happen in economics?
It didn't happen because economists are stupid, and I
appreciate the people before who said we are not. We are very
bright. It happened because of incentives within the economics
profession and those incentives lead researchers to dot i's and
cross t's of existing models. It is a lot easier to do that
than to design a whole new model that nobody else, a peer, can
really review. So they don't explore the wide range of
alternative models, and they don't focus their research on
interpreting and seeing that models are used in policy in a
common sense fashion.
So let me conclude with just two brief suggestions which
relate to issues under the jurisdiction of this committee that
might decrease the probability of such events happening in the
future, and these are far off but it has to do with, you know,
sort of the incentives for economists. The first is a proposal
that might add some common sense check on models. Such a check
is needed because currently there is a nature of the internal
to the sub-field peer review system, that works within NSF and
within the system, that allows for what can only be called an
incestuous mutual reinforcement of researchers' views with no
common sense filter on those views. My proposal is to include a
wider range of peers in the reviewing process for the National
Science Foundation grants in the social sciences. For example,
physicists, mathematicians, statisticians and even business and
government representatives could serve on reviewing those, and
it would serve as a useful common sense check, you know, about
what is going on.
The second is a proposal to increase the number of
researchers trained in interpreting models, rather than
developing models, by providing research grants to do precisely
that. In a sense, what I am suggesting is an applied science
division of the National Science Foundation, a social science
component. This division would fund work on the usefulness of
models and would be responsible for adding the warning labels
that should have been attached to those models.
The applied research would not be highly technical and
would involve a quite different set of skills than the standard
scientific research requires. It would require researchers to
have an intricate knowledge--consumer's knowledge of the
theory, but not a producer's knowledge of that theory. In
addition, it would require a knowledge of institutions,
methodology, previous literature and a sensibility of how the
system works. These are all skills that are not taught in
graduate economics today, but they are skills that underlie
judgment and common sense. By providing NSF grants for this
work, the NSF would encourage the development of a group of
economists who specialize in interpreting models and applying
models to the real world. The development of such a group would
go a long way toward placing the necessary warning labels on
models. Thank you.
[The prepared statement of Dr. Colander follows:]
Prepared Statement of David Colander
Mr. Chairman and Members of the Committee: I thank you for the
opportunity to testify. My name is David Colander. I am the Christian
A. Johnson Distinguished Professor of Economics at Middlebury College.
I have written or edited over forty books, including a top-selling
principles of economics textbook, and 150 articles on various aspects
of economics. I was invited to speak because I was one of the authors
of the Dahlem Report in which we chided the economics profession for
its failure to warn society about the impending financial crisis, and I
have been asked to expand on some of the themes that we discussed in
that report. (I attach that report as an appendix to this testimony.)
Introduction
One year ago, almost to the day, the U.S. economy had a financial
heart attack, from which it is still recovering. That heart attack,
like all heart attacks, was a shock, and it has caused much discussion
about who is to blame, and how can we avoid such heart attacks in the
future. In my view much of that discussion has been off point. To make
an analogy to a physical heart attack, the U.S. had a heart attack
because it is the equivalent of a 450-pound man with serious ailments
too numerous to list, who is trying to live as if he were still a 20-
year-old who can party 24-7. It doesn't take a rocket economist to know
that that will likely lead to trouble. The questions I address in my
testimony are: Why didn't rocket economists recognize that, and warn
society about it? And: What changes can be made to see that it doesn't
happen in the future?
Some non-economists have blamed the financial heart attack on
economist's highly technical models. In my view the problem is not the
models; the problem is the way economic models are used. All too often
models are used in lieu of educated common sense, when in fact models
should be used as an aid to educated common sense. When models replace
common sense, they are a hindrance rather than a help.
Modeling the Economy as a Complex System
Using models within economics or within any other social science,
is especially treacherous. That's because social science involves a
higher degree of complexity than the natural sciences. The reason why
social science is so complex is that the basic unit in social science,
which economists call agents, are strategic, whereas the basic unit of
the natural sciences are not. Economics can be thought of the physics
with strategic atoms, who keep trying to foil any efforts to understand
them and bring them under control. Strategic agents complicate modeling
enormously; they make it impossible to have a perfect model since they
increase the number of calculations one would have to make in order to
solve the model beyond the calculations the fastest computer one can
hypothesize could process in a finite amount of time.
Put simply, the formal study of complex systems is really, really,
hard. Inevitably, complex systems exhibit path dependence, nested
systems, multiple speed variables, sensitive dependence on initial
conditions, and other non-linear dynamical properties. This means that
at any moment in time, right when you thought you had a result, all
hell can break loose. Formally studying complex systems requires
rigorous training in the cutting edge of mathematics and statistics.
It's not for neophytes.
This recognition that the economy is complex is not a new
discovery. Earlier economists, such as John Stuart Mill, recognized the
economy's complexity and were very modest in their claims about the
usefulness of their models. They carefully presented their models as
aids to a broader informed common sense. They built this modesty into
their policy advice and told policy-makers that the most we can expect
from models is half-truths. To make sure that they did not claim too
much for their scientific models, they divided the field of economics
into two branches-one a scientific branch, which worked on formal
models, and the other political economy, which was the branch of
economics that addressed policy. Political economy was seen as an art
which did not have the backing of science, but instead relied on the
insights from models developed in the scientific branch supplemented by
educated common sense to guide policy prescriptions.
In the early 1900s that two-part division broke down, and
economists became a bit less modest in their claims for models, and
more aggressive in their application of models directly to policy
questions. The two branches were merged, and the result was a tragedy
for both the science of economics and for the applied policy branch of
economics.
It was a tragedy for the science of economics because it led
economists away from developing a wide variety of models that would
creatively explore the extraordinarily difficult questions that the
complexity of the economy raised, questions for which new analytic and
computational technology opened up new avenues of investigation.\1\
Instead, the economics profession spent much of its time dotting i's
and crossing t's on what was called a Walrasian general equilibrium
model which was more analytically tractable. As opposed to viewing the
supply/demand model and its macroeconomic counterpart, the Walrasian
general equilibrium model, as interesting models relevant for a few
limited phenomena, but at best a stepping stone for a formal
understanding of the economy, it enshrined both models, and acted as if
it explained everything. Complexities were just assumed away not
because it made sense to assume them away, but for tractability
reasons. The result was a set of models that would not even pass a
perfunctory common sense smell test being studied ad nauseam.
---------------------------------------------------------------------------
\1\ Some approaches working outside this Walrasian general
equilibrium framework that I see as promising includes approaches using
adaptive network analysis, agent based modeling, random graph theory,
ultrametrics, combinatorial stochastic processes, co-integrated vector
auto-regression, and the general study of non-linear dynamic models.
---------------------------------------------------------------------------
Initially macroeconomics stayed separate from this broader unitary
approach, and relied on a set of rough and ready models that had little
scientific foundation. But in the 1980s, macroeconomics and finance
fell into this ``single model'' approach. As that happened it caused
economists to lose sight of the larger lesson that complexity conveys--
that models in a complex system can be expected to continually break
down. This adoption by macroeconomists of a single-model approach is
one of the reasons why the economics profession failed to warn society
about the financial crisis, and some parts of the profession assured
society that such a crisis could not happen. Because they focused on
that single model, economists simply did not study and plan for the
inevitable breakdown of systems that one would expect in a complex
system, because they had become so enamored with their model that they
forgot to use it with common sense judgment.
Models and Macroeconomics
Let me be a bit more specific. The dominant model in macroeconomics
is the dynamic stochastic general equilibrium (DSGE) model. This is a
model that assumes there is a single globally rational representative
agent with complete knowledge who is maximizing over the infinite
future. In this model, by definition, there can be no strategic
coordination problem--the most likely cause of the recent crisis--such
problems are simply assumed away. Yet, this model has been the central
focus of macro economists' research for the last thirty years.
Had the DSGE model been seen as an aid to common sense, it could
have been a useful model. When early versions of this model first
developed back in the early 1980s, it served the useful purpose of
getting some inter-temporal issues straight that earlier macroeconomic
models had screwed up. But then, for a variety of sociological reasons
that I don't have time to go into here, a majority of macroeconomists
started believing that the DSGE model was useful not just as an aid to
our understanding, but as the model of the macroeconomy. That doesn't
say much for the common sense of rocket economists. As the DSGE model
became dominant, important research on broader non-linear dynamic
models of the economy that would have been more helpful in
understanding how an economy would be likely to crash, and what
government might do when faced with a crash, was not done.\2\
---------------------------------------------------------------------------
\2\ Among well known economists, Robert Solow stands out in having
warned about the use of DSGE models for policy. (See Solow, in
Colander, 2007, pg. 235.) He called them ``rhetorical swindles.'' Other
economists, such as Post Keynesians, and economic methodologists also
warned about the use of these models. For a discussion of alternative
approaches, see Colander, ed. (2007). So alternative approaches were
being considered, and concern about the model was aired, but those
voices were lost in the enthusiasm most of the macroeconomics community
showed for these models.
---------------------------------------------------------------------------
Similar developments occurred with efficient market finance models,
which make similar assumptions to DSGE models. When efficient market
models first developed, they were useful; they led to technological
advances in risk management and financial markets. But, as happened
with macro, the users of these financial models forgot that models
provide at best half truths; they stopped using models with common
sense and judgment. The modelers knew that there was uncertainty and
risk in these markets that when far beyond the risk assumed in the
models. Simplification is the nature of modeling. But simplification
means the models cannot be used directly, but must be used judgment and
common sense, with a knowledge of the limitations of use that the
simplifications require. Unfortunately, the warning labels on the
models that should have been there in bold print--these models are
based on assumptions that do not fit the real world, and thus the
models should not be relied on too heavily--were not there. They should
have been, which is why in the Dahlem Report we suggested that economic
researchers who develop these models be subject to a code of ethics
that requires them to warn society when economic models are being used
for purposes for which they were not designed.
How did something so stupid happen in economics? It did not happen
because economists are stupid; they are very bright. It happened
because of incentives in the academic profession to advance lead
researchers to dot i's and cross t's of existing models, rather than to
explore a wide range of alternative models, or to focus their research
on interpreting and seeing that models are used in policy with common
sense. Common sense does not advance one very far within the economics
profession. The over-reliance on a single model used without judgment
is a serious problem that is built into the institutional structure of
academia that produces economic researchers. That system trains show
dogs, when what we need are hunting dogs.
The incorrect training starts in graduate school, where in their
core courses students are primarily trained in analytic techniques
useful for developing models, but not in how to use models creatively,
or in how to use models with judgment to arrive at policy conclusions.
For the most part policy issues are not even discussed in the entire
core macroeconomics course. As students at a top graduate school said,
``Monetary and fiscal policy are not abstract enough to be a question
that would be answered in a macro course'' and ``We never talked about
monetary or fiscal policy, although it might have been slipped in as a
variable in one particular model.'' (Colander, 2007, pg. 169).
Suggestions
Let me conclude with a brief discussion of two suggestions, which
relate to issues under the jurisdiction of this committee, that might
decrease the probability of such events happening in the future.
Include a wider range of peers in peer review
The first is a proposal that might help add a common sense check on
models. Such a check is needed because, currently, the nature of
internal-to-the-subfield peer review allows for an almost incestuous
mutual reinforcement of researcher's views with no common sense filter
on those views. The proposal is to include a wider range of peers in
the reviewing process of NSF grants in the social sciences. For
example, physicists, mathematician, statisticians, and even business
and governmental representatives, could serve, along with economists,
on reviewing committees for economics proposals. Such a broader peer
review process would likely both encourage research on much wider range
of models and would also encourage more creative work.
Increase the number of researchers trained to interpret models
The second is a proposal to increase the number of researchers
trained in interpreting models rather than developing models by
providing research grants to do that. In a sense, what I am suggesting
is an applied science division of the National Science Foundation's
social science component. This division would fund work on the
usefulness of models, and would be responsible for adding the warning
labels that should have been attached to the models.
This applied research would not be highly technical and would
involve a quite different set of skills than the standard scientific
research would require. It would require researchers who had an
intricate consumer's knowledge of theory but not a producer's
knowledge. In addition it would require a knowledge of institutions,
methodology, previous literature, and a sensibility about how the
system works. These are all skills that are currently not taught in
graduate economics programs, but they are the skills that underlie
judgment and common sense. By providing NSF grants for this work, the
NSF would encourage the development of a group of economists who
specialized in interpreting models and applying models to the real
world. The development of such a group would go a long way toward
placing the necessary warning labels on models, and make it less likely
that fiascoes like a financial crisis would happen again.
Bibliography
Colander, David. 2006. (ed.) Post Walrasian Macroeconomics: Beyond the
Dynamic Stochastic General Equilibrium Model. Cambridge, UK.
Cambridge University Press.
Colander, David. 2007. The Making of an Economist Redux. Princeton, New
Jersey, Princeton University Press.
Solow, Robert. 2007. ``Reflections on the Survey'' in Colander (2007).
Appendix
1. Introduction
The global financial crisis has revealed the need to rethink
fundamentally how financial systems are regulated. It has also made
clear a systemic failure of the economics profession. Over the past
three decades, economists have largely developed and come to rely on
models that disregard key factors--including heterogeneity of decision
rules, revisions of forecasting strategies, and changes in the social
context--that drive outcomes in asset and other markets. It is obvious,
even to the casual observer that these models fail to account for the
actual evolution of the real-world economy. Moreover, the current
academic agenda has largely crowded out research on the inherent causes
of financial crises. There has also been little exploration of early
indicators of system crisis and potential ways to prevent this malady
from developing. In fact, if one browses through the academic
macroeconomics and finance literature, ``systemic crisis'' appears like
an otherworldly event that is absent from economic models. Most models,
by design, offer no immediate handle on how to think about or deal with
this recurring phenomenon.\3\ In our hour of greatest need, societies
around the world are left to grope in the dark without a theory. That,
to us, is a systemic failure of the economics profession.
---------------------------------------------------------------------------
\3\ Reinhart and Rogoff (2008) argue that the current financial
crisis differs little from a long chain of similar crises in developed
and developing countries. We certainly share their view. The problem is
that the received body of models in macro finance to which the above
authors have prominently contributed provides no room whatsoever for
such recurrent boom and bust cycles. The literature has, therefore,
been a major source of the illusory `this time it is different' view
that the authors themselves criticize.
---------------------------------------------------------------------------
The implicit view behind standard equilibrium models is that
markets and economies are inherently stable and that they only
temporarily get off track. The majority of economists thus failed to
warn policy-makers about the threatening system crisis and ignored the
work of those who did. Ironically, as the crisis has unfolded,
economists have had no choice but to abandon their standard models and
to produce hand-waving common sense remedies. Common sense advice,
although useful, is a poor substitute for an underlying model that can
provide much-needed guidance for developing policy and regulation. It
is not enough to put the existing model to one side, observing that one
needs, ``exceptional measures for exceptional times.'' What we need are
models capable of envisaging such ``exceptional times.''
The confinement of macroeconomics to models of stable states that
are perturbed by limited external shocks and that neglect the intrinsic
recurrent boom-and-bust dynamics of our economic system is remarkable.
After all, worldwide financial and economic crises are hardly new and
they have had a tremendous impact beyond the immediate economic
consequences of mass unemployment and hyper inflation. This is even
more surprising, given the long academic legacy of earlier economists'
study of crisis phenomena, which can be found in the work of Walter
Bagehot (1873), Axel Leijonhuvfud (2000), Charles Kindleberger (1989),
and Hyman Minsky (1986), to name a few prominent examples. This
tradition, however, has been neglected and even suppressed.
The most recent literature provides us with examples of blindness
against the upcoming storm that seem odd in retrospect. For example, in
their analysis of the risk management implications of CDOs, Krahnen
(2005) and Krahnen and Wilde (2006) mention the possibility of an
increase of `systemic risk.' But, they conclude that this aspect should
not be the concern of the banks engaged in the CDO market, because it
is the governments' responsibility to provide costless insurance
against a system-wide crash. We do not share this view. On the more
theoretical side, a recent and prominent strand of literature
essentially argues that consumers and investors are too risk averse
because of their memory of the (improbable) event of the Great
Depression (e.g., Cogley and Sargent, 2008). Much of the motivation for
economics as an academic discipline stems from the desire to explain
phenomena like unemployment, boom and bust cycles, and financial
crises, but dominant theoretical models exclude many of the aspects of
the economy that will likely lead to a crisis. Confining theoretical
models to `normal' times without consideration of such defects might
seem contradictory to the focus that the average taxpayer would expect
of the scientists on his payroll.
This failure has deep methodological roots. The often heard
definition of economics--that it is concerned with the `allocation of
scarce resources'--is short-sighted and misleading. It reduces
economics to the study of optimal decisions in well-specified choice
problems. Such research generally loses track of the inherent dynamics
of economic systems and the instability that accompanies its complex
dynamics. Without an adequate understanding of these processes, one is
likely to miss the major factors that influence the economic sphere of
our societies. This insufficient definition of economics often leads
researchers to disregard questions about the coordination of actors and
the possibility of coordination failures. Indeed, analysis of these
issues would require a different type of mathematics than that which is
generally used now by many prominent economic models.
Many of the financial economists who developed the theoretical
models upon which the modern financial structure is built were well
aware of the strong and highly unrealistic restrictions imposed on
their models to assure stability. Yet, financial economists gave little
warning to the public about the fragility of their models,\4\ even as
they saw individuals and businesses build a financial system based on
their work. There are a number of possible explanations for this
failure to warn the public. One is a ``lack of understanding''
explanation--the researchers did not know the models were fragile. We
find this explanation highly unlikely; financial engineers are
extremely bright, and it is almost inconceivable that such bright
individuals did not understand the limitations of the models. A second,
more likely explanation, is that they did not consider it their job to
warn the public. If that is the cause of their failure, we believe that
it involves a misunderstanding of the role of the economist, and
involves an ethical breakdown. In our view, economists, as with all
scientists, have an ethical responsibility to communicate the
limitations of their models and the potential misuses of their
research. Currently, there is no ethical code for professional economic
scientists. There should be one.
---------------------------------------------------------------------------
\4\ Indeed, few researchers explored the consequences of a
breakdown of their assumptions, even though this was rather likely.
---------------------------------------------------------------------------
In the following pages, we identify some major areas of concern in
theory and applied methodology and point out their connection to crisis
phenomena. We also highlight some promising avenues of study that may
provide guidance for future researchers.
2. Models (or the Use of Models) as a Source of Risk
The economic textbook models applied for allocation of scarce
resources are predominantly of the Robinson Crusoe (representative
agent) type. Financial market models are obtained by letting Robinson
manage his financial affairs as a sideline to his well-considered
utility maximization over his (finite or infinite) expected lifespan
taking into account with correct probabilities all potential future
happenings. This approach is mingled with insights from Walrasian
general equilibrium theory, in particular the finding of the Arrrow-
Debreu two-period model that all uncertainty can be eliminated if only
there are enough contingent claims (i.e., appropriate derivative
instruments). This theoretical result (a theorem in an extremely
stylized model) underlies the common belief that the introduction of
new classes of derivatives can only be welfare increasing (a view
obviously originally shared by former Fed Chairman Greenspan). It is
worth emphasizing that this view is not an empirically grounded belief
but an opinion derived from a benchmark model that is much too abstract
to be confronted with data.
On the practical side, mathematical portfolio and risk management
models have been the academic backbone of the tremendous increase of
trading volume and diversification of instruments in financial markets.
Typically, new derivative products achieve market penetration only if a
certain industry standard has been established for pricing and risk
management of these products. Mostly, pricing principles are derived
from a set of assumptions on an `appropriate' process for the
underlying asset, (i.e., the primary assets on which options or
forwards are written) together with an equilibrium criterion such as
arbitrage-free prices. With that mostly comes advice for hedging the
inherent risk of a derivative position by balancing it with other
assets that neutralize the risk exposure. The most prominent example is
certainly the development of a theory of option pricing by Black and
Scholes that eventually (in the eighties) could even be implemented on
pocket calculators. Simultaneously with Black-Scholes option pricing,
the same principles led to the widespread introduction of new
strategies under the heading of portfolio insurance and dynamic hedging
that just tried to implement a theoretically risk-free portfolio
composed of both assets and options and keep it risk-free by frequent
rebalancing after changes of its input data (e.g., asset prices). For
structured products for credit risk, the basic paradigm of derivative
pricing--perfect replication--is not applicable so that one has to rely
on a kind of rough-and-ready evaluation of these contracts on the base
of historical data. Unfortunately, historical data were hardly
available in most cases which meant that one had to rely on simulations
with relatively arbitrary assumptions on correlations between risks and
default probabilities. This makes the theoretical foundations of all
these products highly questionable--the equivalent to building a
building of cement of which you weren't sure of the components. The
dramatic recent rise of the markets for structured products (most
prominently collateralized debt obligations and credit default swaps--
CDOs and CDSs) was made possible by development of such simulation-
based pricing tools and the adoption of an industry-standard for these
under the lead of rating agencies. Barry Eichengreen (2008) rightly
points out that the ``development of mathematical methods designed to
quantify and hedge risk encouraged commercial banks, investment banks
and hedge funds to use more leverage'' as if the very use of the
mathematical methods diminished the underlying risk. He also notes that
the models were estimated on data from periods of low volatility and
thus could not deal with the arrival of major changes. Worse, it is our
contention that such major changes are endemic to the economy and
cannot be simply ignored.
What are the flaws of the new unregulated financial markets which
have emerged? As we have already pointed out in the introduction, the
possibility of systemic risk has not been entirely ignored but it has
been defined as lying outside the responsibility of market
participants. In this way, moral hazard concerning systemic risk has
been a necessary and built-in attribute of the system. The neglect of
the systemic part in the `normal mode of operation,' of course, implies
that external effects are not taken properly into account and that in
tendency, market participants will ignore the influence of their own
behavior on the stability of the system. The interesting aspect is more
that this was a known and accepted element of operations. Note that the
blame should not only fall on market participants, but also on the
deliberate ignoring of the systemic risk factors or the failure to at
least point them out to the public amounts to a sort of academic `moral
hazard.'
There are some additional aspects as well: asset-pricing and risk
management tools are developed from an individualistic perspective,
taking as given (ceteris paribus) the behavior of all other market
participants. However, popular models might be used by a large number
or even the majority of market participants. Similarly, a market
participant (e.g., the notorious Long-Term Capital Management) might
become so dominant in certain markets that the ceteris paribus
assumption becomes unrealistic. The simultaneous pursuit of identical
micro strategies leads to synchronous behavior and mechanic contagion.
This simultaneous application might generate an unexpected macro
outcome that actually jeopardizes the success of the underlying micro
strategies. A perfect illustration is the U.S. stock market crash of
October 1987. Triggered by a small decrease of prices, automated
hedging strategies produced an avalanche of sell orders that out of the
blue led to a fall in U.S. stock indices of about 20 percent within one
day. With the massive sales to rebalance their portfolios (along the
lines of Black and Scholes), the relevant actors could not realize
their attempted incremental adjustments, but rather suffered major
losses from the ensuing large macro effect.
A somewhat different aspect is the danger of a control illusion:
The mathematical rigor and numerical precision of risk management and
asset pricing tools has a tendency to conceal the weaknesses of models
and assumptions to those who have not developed them and do not know
the potential weakness of the assumptions and it is indeed this that
Eichengreen emphasizes. Naturally, models are only approximations to
the real world dynamics and partially built upon quite heroic
assumptions (most notoriously: Normality of asset price changes which
can be rejected at a confidence level of 99.9999 . . .. Anyone who has
attended a course in first-year statistics can do this within minutes).
Of course, considerable progress has been made by moving to more
refined models with, e.g., `fat-tailed' Levy processes as their driving
factors. However, while such models better capture the intrinsic
volatility of markets, their improved performance, taken at face value,
might again contribute to enhancing the control illusion of the naive
user.
The increased sophistication of extant models does, however, not
overcome the robustness problem and should not absolve the modelers
from explaining their limitations to the users in the financial
industry. As in nuclear physics, the tools provided by financial
engineering can be put to very different uses so that what is designed
as an instrument to hedge risk can become a weapon of `financial mass
destruction' (in the words of Warren Buffet) if used for increased
leverage. In fact, it appears that derivative positions have been built
up often in speculative ways to profit from high returns as long as the
downside risk does not materialize. Researchers who develop such models
can claim they are neutral academics--developing tools that people are
free to use or not. We do not find that view credible. Researchers have
an ethical responsibility to point out to the public when the tool that
they developed is misused. It is the responsibility of the researcher
to make clear from the outset the limitations and underlying
assumptions of his models and warn of the dangers of their mechanic
application.
What follows from our diagnosis? Market participants and regulators
have to become more sensitive towards the potential weaknesses of risk
management models. Since we do not know the `true' model, robustness
should be a key concern. Model uncertainty should be taken into account
by applying more than a single model. For example, one could rely on
probabilistic projections that cover a whole range of specific models
(cf., Follmer, 2008). The theory of robust control provides a toolbox
of techniques that could be applied for this purpose, and it is an
approach that should be considered.
3. Unrealistic Model Assumptions and Unrealistic Outcomes
Many economic models are built upon the twin assumptions of
`rational expectations' and a representative agent. ``Rational
expectations'' instructs an economist to specify individuals'
expectations to be fully consistent with the structure of his own
model. This concept can be thought of as merely a way to close a model.
A behavioral interpretation of rational expectations would imply that
individuals and the economist have a complete understanding of the
economic mechanisms governing the world. In this sense, rational
expectations models do not attempt to formalize individuals' actual
expectations: specifications are not based on empirical observation of
the expectations formation process of human actors. Thus, even when
applied economics research or psychology provide insights about how
individuals actually form expectations, they cannot be used within RE
models. Leaving no place for imperfect knowledge and adaptive
adjustments, rational expectations models are typically found to have
dynamics that are not smooth enough to fit economic data well.\5\
---------------------------------------------------------------------------
\5\ For a critique of rational expectations models on
epistemological grounds, see Frydman and Goldberg (2007, 2008) and
references therein.
---------------------------------------------------------------------------
Technically, rational expectations models are often framed as
dynamic programming problems in macroeconomics. But, dynamic
programming models have serious limitations. Specifically, to make them
analytically tractable, not more than one dynamically maximizing agent
can be considered, and consistent expectations have to be imposed.
Therefore, dynamic programming models are hardly imaginable without the
assumptions of a representative agent and rational expectations. This
has generated a vicious cycle by which the technical tools developed on
the base of the chosen assumptions prevent economists from moving
beyond these restricted settings and exploring more realistic
scenarios. Note that such settings also presume that there is a single
model of the economy, which is odd given that even economists are
divided in their views about the correct model of the economy. While
other currents of research do exist, economic policy advice,
particularly in financial economics, has far too often been based
(consciously or not) on a set of axioms and hypotheses derived
ultimately from a highly limited dynamic control model, using the
Robinson approach with `rational' expectations.
The major problem is that despite its many refinements, this is not
at all an approach based on, and confirmed by, empirical research.\6\
In fact, it stands in stark contrast to a broad set of regularities in
human behavior discovered both in psychology and what is called
behavioral and experimental economics. The corner stones of many models
in finance and macroeconomics are rather maintained despite all the
contradictory evidence discovered in empirical research. Much of this
literature shows that human subjects act in a way that bears no
resemblance to the rational expectations paradigm and also have
problems discovering `rational expectations equilibria' in repeated
experimental settings. Rather, agents display various forms of `bounded
rationality' using heuristic decision rules and displaying inertia in
their reaction to new information. They have also been shown in
financial markets to be strongly influenced by emotional and hormonal
reactions (see Lo et al., 2005, and Coates and Herbert, 2008). Economic
modeling has to take such findings seriously.
---------------------------------------------------------------------------
\6\ The historical emergence of the representative agent paradigm
is a mystery. Ironically, it appeared over the 70s after a period of
intense discussions on the problem of aggregation in economics (that
basically yielded negative results such as the impossibility to
demonstrated `nice' properties of aggregate demand or supply functions
without imposing extreme assumptions on individual behavior). The
representative agent appeared without methodological discussion. In the
words of Deirdre McCloskey: ``It became a rule in the conversation of
some economists because Tom and Bob said so.'' (personal
communication). Today, this convention has become so strong that many
young economists wouldn't know of an alternative way to approach
macroeconomic issues.
---------------------------------------------------------------------------
What we are arguing is that as a modeling requirement, internal
consistency must be complemented with external consistency: Economic
modeling has to be compatible with insights from other branches of
science on human behavior. It is highly problematic to insist on a
specific view of humans in economic settings that is irreconcilable
with evidence.
The `representative agent' aspect of many current models in
macroeconomics (including macro finance) means that modelers subscribe
to the most extreme form of conceptual reductionism (Lux and
Westerhoff, 2009): by assumption, all concepts applicable to the macro
sphere (i.e., the economy or its financial system) are fully reduced to
concepts and knowledge for the lower-level domain of the individual
agent. It is worth emphasizing that this is quite different from the
standard reductionist concept that has become widely accepted in
natural sciences. The more standard notion of reductionism amounts to
an approach to understanding the nature of complex phenomena by
reducing them to the interactions of their parts, allowing for new,
emergent phenomena at the higher hierarchical level (the concept of
`more is different,' cf. Anderson, 1972).
Quite to the contrary, the representative agent approach in
economics has simply set the macro sphere equal to the micro sphere in
all respects. One could, indeed, say that this concept negates the
existence of a macro sphere and the necessity of investigating
macroeconomic phenomena in that it views the entire economy as an
organism governed by a universal will.\7\ Any notion of ``systemic
risk'' or ``coordination failure'' is necessarily absent from, and
alien to, such a methodology.
---------------------------------------------------------------------------
\7\ The conceptual reductionist approach of the representative
agent is also remarkably different from the narrative of the `invisible
hand' which has more the flavor of `more is different'.
---------------------------------------------------------------------------
For natural scientists, the distinction between micro-level
phenomena and those originating on a macro, system-wide scale from the
interaction of microscopic units is well-known. In a dispersed system,
the current crisis would be seen as an involuntary emergent phenomenon
of the microeconomic activity. The conceptual reductionist paradigm,
however, blocks from the outset any understanding of the interplay
between the micro and macro levels. The differences between the overall
system and its parts remain simply incomprehensible from the viewpoint
of this approach.
In order to develop models that allow us to deduce macro events
from microeconomic regularities, economists have to rethink the concept
of micro foundations of macroeconomic models. Since economic activity
is of an essentially interactive nature, economists' micro foundations
should allow for the interactions of economic agents. Since interaction
depends on differences in information, motives, knowledge and
capabilities, this implies heterogeneity of agents. For instance, only
a sufficiently rich structure of connections between firms, households
and a dispersed banking sector will allow us to get a grasp on
``systemic risk,'' domino effects in the financial sector, and their
repercussions on consumption and investment. The dominance of the
extreme form of conceptual reductionism of the representative agent has
prevented economists from even attempting to model such all important
phenomena. It is the flawed methodology that is the ultimate reason for
the lack of applicability of the standard macro framework to current
events.
Since most of what is relevant and interesting in economic life has
to do with the interaction and coordination of ensembles of
heterogeneous economic actors, the methodological preference for single
actor models has extremely handicapped macroeconomic analysis and
prevented it from approaching vital topics. For example, the recent
surge of research in network theory has received relatively scarce
attention in economics. Given the established curriculum of economic
programs, an economist would find it much more tractable to study
adultery as a dynamic optimization problem of a representative husband,
and derive the optimal time path of marital infidelity (and publish his
exercise) rather than investigating financial flows in the banking
sector within a network theory framework. This is more than unfortunate
in view of the network aspects of interbank linkages that have become
apparent during the current crisis.
In our view, a change of focus is necessary that takes seriously
the regularities in expectation formation revealed by behavioral
research and, in fact, gives back an independent role to expectations
in economic models. It would also be fallacious to only replace the
current paradigm by a representative `non-rational' actor (as it is
sometimes done in recent literature). Rather, an appropriate micro
foundation is needed that considers interaction at a certain level of
complexity and extracts macro regularities (where they exist) from
microeconomic models with dispersed activity.
Once one acknowledges the importance of empirically based
behavioral micro foundations and the heterogeneity of actors, a rich
spectrum of new models becomes available. The dynamic co-evolution of
expectations and economic activity would allow one to study out-of-
equilibrium dynamics and adaptive adjustments. Such dynamics could
reveal the possibility of multiplicity and evolution of equilibria
(e.g., with high or low employment) depending on agents' expectations
or even on the propagation of positive or negative `moods' among the
population. This would capture the psychological component of the
business cycle which--though prominent in many policy-oriented
discussions--is never taken into consideration in contemporary
macroeconomic models.
It is worth noting that understanding the formation of such low-
level equilibria might be much more valuable in coping with major
`efficiency losses' by mass unemployment than the pursuit of small
`inefficiencies' due to societal decisions on norms such as shop
opening times. Models with interacting heterogeneous agents would also
open the door to the incorporation of results from other fields:
network theory has been mentioned as an obvious example (for models of
networks in finance see Allen and Babus, 2008). `Self-organized
criticality' theory is another area that seems to have some appeal for
explaining boom-and-bust cycles (cf. Scheinkman and Woodford, 1992).
Incorporating heterogeneous agents with imperfect knowledge would also
provide a better framework for the analysis of the use and
dissemination of information through market operations and more direct
links of communication. If one accepts that the dispersed economic
activity of many economic agents could be described by statistical
laws, one might even take stock of methods from statistical physics to
model dynamic economic systems (cf. Aoki and Yoshikawa, 2007; Lux,
2009, for examples).
4. Robustness and Data-Driven Empirical Research
Currently popular models (in particular: dynamic general
equilibrium models) do not only have weak micro foundations, their
empirical performance is far from satisfactory (Juselius and Franchi,
2007). Indeed, the relevant strand of empirical economics has more and
more avoided testing their models and has instead turned to calibration
without explicit consideration of goodness-of-fit.\8\ This calibration
is done using ``deep economic parameters'' such as parameters of
utility functions derived from microeconomic studies. However, at the
risk of being repetitive, it should be emphasized that micro parameters
cannot be used directly in the parameterization of a macroeconomic
model. The aggregation literature is full of examples that point out
the possible ``fallacies of composition.'' The ``deep parameters'' only
seem sensible if one considers the economy as a universal organism
without interactions. If interactions are important (as it seems to us
they are), the restriction of the parameter space imposed by using
micro parameters is inappropriate.
---------------------------------------------------------------------------
\8\ It is pretty obvious how the currently popular class of dynamic
general equilibrium models would have to `cope' with the current
financial crisis. It will be covered either by a dummy or it will have
to be interpreted as a very large negative stochastic shock to the
economy, i.e., as an event equivalent to a large asteroid strike.
---------------------------------------------------------------------------
Another concern is nonstationarity and structural shifts in the
underlying data. Macro models, unlike many financial models, are often
calibrated over long time horizons which include major changes in the
regulatory framework of the countries investigated. Cases in question
are the movements between different exchange rate regimes and the
deregulation of financial markets over the 70s and 80s. In summary, it
seems to us that much of contemporary empirical work in macroeconomics
and finance is driven by the pre-analytic belief in the validity of a
certain model. Rather than (mis)using statistics as a means to
illustrate these beliefs, the goal should be to put theoretical models
to scientific test (as the naive believer in positive science would
expect).
The current approach of using pre-selected models is problematic
and we recommend a more data-driven methodology. Instead of starting
out with an ad-hoc specification and questionable ceteris paribus
assumptions, the key features of the data should be explored via data-
analytical tools and specification tests. David Hendry provides a well-
established empirical methodology for such exploratory data analysis
(Hendry, 1995, 2009) as well as a general theory for model selection
(Hendry and Krolzig, 2005); clustering techniques such as projection
pursuit (e.g., Friedman, 1987) might provide alternatives for the
identification of key relationships and the reduction of complexity on
the way from empirical measurement to theoretical models. Co-integrated
VAR models could provide an avenue towards identification of robust
structures within a set of data (Juselius, 2006), for example, the
forces that move equilibria (pushing forces, which give rise to
stochastic trends) and forces that correct deviations from equilibrium
(pulling forces, which give rise to long-run relations). Interpreted in
this way, the `general-to-specific' empirical approach has a good
chance of nesting a multi-variate, path-dependent data-generating
process and relevant dynamic macroeconomic theories. Unlike approaches
in which data are silenced by prior restrictions, the Co-integrated VAR
model gives the data a rich context in which to speak freely (Hoover et
al., 2008).
A chain of specification tests and estimated statistical models for
simultaneous systems would provide a benchmark for the subsequent
development of tests of models based on economic behavior: significant
and robust relations within a simultaneous system would provide
empirical regularities that one would attempt to explain, while the
quality of fit of the statistical benchmark would offer a confidence
band for more ambitious models. Models that do not reproduce (even)
approximately the quality of the fit of statistical models would have
to be rejected (the majority of currently popular macroeconomic and
macro finance models would not pass this test). Again, we see here an
aspect of ethical responsibility of researchers: Economic policy models
should be theoretically and empirically sound. Economists should avoid
giving policy recommendations on the base of models with a weak
empirical grounding and should, to the extent possible, make clear to
the public how strong the support of the data is for their models and
the conclusions drawn from them.
5. A Research Agenda to Cope with Financial Fragility
The notion of financial fragility implies that a given system might
be more or less susceptible to produce crises. It seems clear that
financial innovations have made the system more fragile. Apparently,
the existing linkages within the worldwide, highly connected financial
markets have generated the spill-overs from the U.S. sub-prime problem
to other layers of the financial system. Many financial innovations had
the effect of creating links between formerly unconnected players. All
in all, the degree of connectivity of the system has probably increased
enormously over the last decades. As is well known from network theory
in natural sciences, a more highly connected system might be more
efficient in coping with certain tasks (maybe distributing risk
components), but will often also be more vulnerable to shocks and--
systemic failure! The systematic analysis of network vulnerability has
been undertaken in the computer science and operations research
literature (see e.g., Criado et al., 2005). Such aspects have, however,
been largely absent from discussions in financial economics. The
introduction of new derivatives was rather seen through the lens of
general equilibrium models: more contingent claims help to achieve
higher efficiency. Unfortunately, the claimed efficiency gains through
derivatives are merely a theoretical implication of a highly stylized
model and, therefore, have to count as a hypothesis. Since there is
hardly any supporting empirical evidence (or even analysis of this
question), the claimed real-world efficiency gains from derivatives are
not justified by true science. While the economic argument in favor of
ever new derivatives is more one of persuasion rather than evidence,
important negative effects have been neglected. The idea that the
system was made less risky with the development of more derivatives led
to financial actors taking positions with extreme degrees of leverage
and the danger of this has not been emphasized enough.
As we have mentioned, one neglected area is the degree of
connectivity and its interplay with the stability of the system (see
Boesch et al., 2006). We believe that it will be necessary for
supervisory authorities to develop a perspective on the network aspects
of the financial system, collect appropriate data, define measures of
connectivity and perform macro stress testing at the system level. In
this way, new measures of financial fragility would be obtained. This
would also require a new area of accompanying academic research that
looks at agent-based models of the financial system, performs scenario
analyses and develops aggregate risk measures. Network theory and the
theory of self-organized criticality of highly connected systems would
be appropriate starting points.
The danger of systemic risk means that regulation has to be
extended from individualistic (regulation of single institutions which
of course, is still crucial) to system wide regulation. In the sort of
system which is prone to systemic crisis, regulation also has to have a
systemic perspective. Academic researchers and supervisory authorities
thus have to look into connections within the financial sector and to
investigate the repercussions of problems within one institute on other
parts of the system (even across national borders). Certainly, before
deciding about the bail-out of a large bank, this implies an
understanding of the network. One should know whether its bankruptcy
would lead to widespread domino effects or whether contagion would be
limited. It seems to us that what regulators provide currently is far
from a reliable assessment of such after effects.
Such analysis has to be supported by more traditional approaches:
Leverage of financial institutions rose to unprecedented levels prior
to the crisis, partly by evading Basle II regulations through special
investment vehicles (SIVs). The hedge fund market is still entirely
unregulated. The interplay between leverage, connectivity and system
risk needs to be investigated at the aggregate level. It is highly
likely, that extreme leverage levels of interconnected institutions
will be found to impose unacceptable social risk on the public. Prudent
capital requirements would be necessary and would require a solid
scientific investigation of the above aspects rather than a pre-
analytic laissez-faire attitude.
We also have to re-investigate the informational role of financial
prices and financial contracts. While trading in stock markets is
usually interpreted as at least in part transmitting information, this
information transmission seems to have broken down in the case of
structured financial products. It seems that securitization has rather
led to a loss of information by anonymous intermediation (often
multiple) between borrowers and lenders. In this way, the informational
component has been outsourced to rating agencies and typically, the
buyer of CDO tranches would not have spent any effort himself on
information acquisition concerning his far away counterparts. However,
this centralized information processing instead of the dispersed one in
traditional credit relationships might lead to a severe loss of
information. As it turned out, standard loan default models failed
dramatically in recent years (Rajan et al., 2008). It should also be
noted that the price system itself can exacerbate the difficulties in
the financial market (see Hellwig, 2008). One of the reasons for the
sharp fall in the asset valuations of major banks was not only the loss
on the assets on which their derivatives were based, but also the
general reaction of the markets to these assets. As markets became
aware of the risk involved, all such assets were written down and it
was in this way that a small sector of the market ``contaminated'' the
rest. Large parts of the asset holdings of major banks abruptly lost
much of their value. Thus the price system itself can be destabilizing
as expectations change.
On the macroeconomic level, it would be desirable to develop early
warning schemes that indicate the formation of bubbles. Combinations of
indicators with time series techniques could be helpful in detecting
deviations of financial or other prices from their long-run averages.
Indication of structural change (particularly towards non-stationary
trajectories) would be a signature of changes of the behavior of market
participants of a bubble-type nature.
6. Conclusions
The current crisis might be characterized as an example of the
final stage of a well-known boom-and-bust pattern that has been
repeated so many times in the course of economic history. There are,
nevertheless, some aspects that make this crisis different from its
predecessors: First, the preceding boom had its origin--at least to a
large part--in the development of new financial products that opened up
new investment possibilities (while most previous crises were the
consequence of over-investment in new physical investment
possibilities). Second, the global dimension of the current crisis is
due to the increased connectivity of our already highly interconnected
financial system. Both aspects have been largely ignored by academic
economics. Research on the origin of instabilities, over-investment and
subsequent slumps has been considered as an exotic side track from the
academic research agenda (and the curriculum of most economics
programs).This, of course, was because it was incompatible with the
premise of the rational representative agent. This paradigm also made
economics blind with respect to the role of interactions and
connections between actors (such as the changes in the network
structure of the financial industry brought about by deregulation and
introduction of new structured products). Indeed, much of the work on
contagion and herding behavior (see Banerjee, 1992, and Chamley, 2002)
which is closely connected to the network structure of the economy has
not been incorporated into macroeconomic analysis.
We believe that economics has been trapped in a sub-optimal
equilibrium in which much of its research efforts are not directed
towards the most prevalent needs of society. Paradoxically self-
reinforcing feedback effects within the profession may have led to the
dominance of a paradigm that has no solid methodological basis and
whose empirical performance is, to say the least, modest. Defining away
the most prevalent economic problems of modern economies and failing to
communicate the limitations and assumptions of its popular models, the
economics profession bears some responsibility for the current crisis.
It has failed in its duty to society to provide as much insight as
possible into the workings of the economy and in providing warnings
about the tools it created. It has also been reluctant to emphasize the
limitations of its analysis. We believe that the failure to even
envisage the current problems of the worldwide financial system and the
inability of standard macro and finance models to provide any insight
into ongoing events make a strong case for a major reorientation in
these areas and a reconsideration of their basic premises.
References
Allen, F. and A. Babus, 2008, Networks in Finance. Wharton Financial
Institutions Center Working Paper No. 08-07. Available at SSRN:
http://ssrn.com/abstract=1094883
Anderson, P.W., 1972, More is different, Science 177, 393-396.
Aoki, M. and H. Yoshikawa, 2007, Reconstructing Macroeconomics: A
Perspective from Statistical Physics and Combinatorial
Stochastic Processes. Cambridge University Press: Cambridge and
New York.
Bagehot, W., 1873, Lombard Street: A Description of the Money Market.
Henry S. King and Co.: London.
Banerjee, A., 1992, A simple model of herd behaviour, Quarterly Journal
of Economics, 108, 797-817.
Boesch, F.T., F. Harary, and J.A. Kabell, 2006, Graphs as models of
communication network vulnerability: Connectivity and
persistence, Networks, 11, 57-63.
Brigandt, I. and A. Love, `Reductionism in Biology' in the Stanford
Encyclopedia of Philosophy. Available at http://
plato.stanford.edu/entries/reduction-biology/
Campos, J., N.R. Ericsson and D.F. Hendry, 2005, Editors' Introduction
to General to Specific Modelling, 1-81, Edward Elgar: London.
Chamley, C.P., 2002, Rational Herds: Economic Models of Social
Learning. Cambridge University Press: Cambridge.
Coates J.M. and J. Herbert, 2008, Endogenous steroids and financial
risk taking on a London trading floor, Proceedings of the
National Academy of Sciences, 6167-6172.
Cogley, T. and T. Sargent, 2008, The market price of risk and the
equity premium: A legacy of the Great Depression?, Journal of
Monetary Economics, 55, 454-476.
Criado, R., J. Flores, B. Hernandez-Bermejo, J. Pello, and M. Romance,
2005, Effective measurement of network vulnerability under
random and intentional attacks, Journal of Mathematical
Modelling and Algorithms, 4, 307-316.
Eichengreen, B., 2008, Origins and Responses to the Crisis, unpublished
manuscript, University of California, Berkeley.
Follmer, H., 2008, Financial uncertainty, risk measures and robust
preferences, in: Yor, M., ed., Aspects of Mathematical Finance,
Springer: Berlin.
Friedman, J., 1987, Exploratory projection pursuit, Journal of the
American Statistical Association, 82, 249-266.
Frydman, R. and M.D. Goldberg (2007), Imperfect Knowledge Economics:
Exchange Rates and Risk, Princeton, NJ: Princeton University
Press.
Frydman, R. and M.D. Goldberg (2008), Macroeconomic Theory for a World
of Imperfect Knowledge, Capitalism and Society, 3, Article 1.
Hellwig, M.F., 2008, Systemic Risk in the Financial Sector: An Analysis
of the Subprime-Mortgage Financial Crisis, MPI Collective Goods
Preprint, No. 2008/43.
Hendry, D., 2009, The Methodology of Empirical Econometric Modelling:
Applied Econometrics Through the Looking-Glass, forthcoming in
The Handbook of Empirical Econometrics, Palgrave.
Hendry, D.F., 1995. Dynamic Econometrics. Oxford University Press:
Oxford.
Hendry, D.F. and H.-M. Krolzig, 2005, The Properties of Automatic Gets
Modelling, Economic Journal, 115, C32-C61.
Hoover, K., S. Johansen, and K. Juselius, 2008, Allowing the data to
speak freely: The macroeconometrics of the cointegrated vector
autoregression. American Economic Review 98, 251-55.
Juselius, K., 2006, The cointegrated VAR model: Econometric Methodology
and Empirical Applications. Oxford University Press: Oxford.
Juselius, K. and M. Franchi, 2007, Taking a DSGE Model to the Data
Meaningfully, Economics--The Open-Access, Open-Assessment E-
Journal, 4.
Kindleberger, C.P., 1989, Manias, Panics, and Crashes: A History of
Financial Crises. MacMillan: London.
Krahnen, J.-P. and C. Wilde, 2006, Risk Transfer with CDOs and Systemic
Risk in Banking. Center for Financial Studies, WP 2006-04.
Frankfurt.
Krahnen, J.-P., 2005, Der Handel von Kreditrisiken: Eine neue Dimension
des Kapitalmarktes, Perspektiven der Wirtschaftspolitik 6, 499-
519.
Leijonhufvud, A., 2000, Macroeconomic Instability and Coordination:
Selected Essays, Edward Elgar: Cheltenham.
Lo, A., D.V. Repin and B.N. Steenbarger, Fear and Greed in Financial
Markets: A Clinical Study of Day-Traders, American Economic
Review 95, 352-359.
Lux, T. and F. Westerhoff, 2009, Economics crisis, Nature Physics 5, 2-
3.
Lux, T., 2009, Stochastic Behavioral Asset Pricing Models and the
Stylized Facts, chapter 3 in T. Hens and K. Schenk-Hoppe, eds.,
Handbook of Financial Markets: Dynamics and Evolution. North-
Holland: Amsterdam, 161-215.
Minsky, H.P., 1986, Stabilizing an Unstable Economy. Yale University
Press: New Haven.
Rajan, U., A. Seru and V. Vig, 2008, The Failure of Models that Predict
Failure: Distance, Incentives and Defaults. Chicago GSB
Research Paper No. 08-19.
Reinhart, C. and K. Rogoff, 2008, This Time is Different: A Panoramic
View of Eight Centuries of Financial Crises. Manuscript,
Harvard University and NBER.
Scheinkman, J. and M. Woodford, 1994, Self-Organized Criticality and
Economic Fluctuations, American Economic Review 84 (Papers and
Proceedings), 417-421.
Biography for David Colander
David Colander has been the Christian A. Johnson Distinguished
Professor of Economics at Middlebury College, Middlebury, Vermont since
1982. He has authored, co-authored, or edited over 40 books (including
a principles and intermediate macro text) and 150 articles on a wide
range of topics. His books have been, or are being, translated into a
number of different languages, including Chinese, Bulgarian, Polish,
Italian, and Spanish.
He received his Ph.D. from Columbia University and has taught at
Columbia University, Vassar College, the University of Miami as well as
Middlebury. He has also been a consultant to Time-Life Films, a
consultant to Congress, a Brookings Policy Fellow, and a Visiting
Scholar at Nuffield College, Oxford. In 2001-2002 he was the Kelly
Professor of Distinguished Teaching at Princeton University.
He is a former President of both the Eastern Economic Association
and History of Economic Thought Society and is, or has been, on the
editorial boards of the Journal of the History of Economic Thought,
Journal of Economic Methodology, Eastern Economic Journal, and The
Journal of Socioeconomics, and Journal of Economic Perspectives. He is
a member of the AEA Committee on Economic Education.
Discussion
Chairman Miller. I want to thank all of you.
Appropriate Uses of Financial Models
Let me begin this panel with a question of the earlier
panel. Some of those responsible, involved in developing
economic modeling now say that the fundamental problem was that
the model was wrong, there is more data. I don't think anyone
thinks that models should be prohibited or people should be
prohibited from acting on their models for their investment
decisions or whatever. The extent to which it can be used, it
should be used for regulation, safety and soundness regulation.
Do any of you--what do each of you think about whether the
models may be improved and will become reliable, sufficiently
reliable to base capital requirements on--or do you think that
it is so inherently unpredictable that economic forecasts will
never become like predicting the movements of the planets, that
it may be useful for seeing if a financial institution is
headed towards trouble, but not to say it is got nothing to
worry about? Any of you. Dr. Berman.
Dr. Berman. Thank you. I think models definitely have a
very significant role, not just in finance but in society in
general. The question is, what aspect of a model are you
looking to use. Certain models are designed to predict the
future. That is always very difficult to do. We can predict
where the planets are going to go but it is very difficult to
predict where the stock market is going to go today. That is a
very small portion of what financial modeling is about,
predicting the future. Unfortunately, that is what folks glean
onto when they start thinking about capital requirements. A
much larger portion of what modeling is about is understanding:
if something happens to X, what happens to Y? You don't have to
predict the future in order to do that, you just need to know
the relationships between two different things.
Let's take an excellent example. The world's largest
insurance company entered into massive amounts of credit
default swaps that ultimately were responsible for their
demise. The bet that they took might have turned out to be the
best bet that they ever could have made. We don't know because
those CDSs are probably still out there to a certain extent.
But they failed to account for the fact that, what would happen
if there was a small dip in the value of these, and my
counterparty asks for collateral? That's not a matter of
predicting the future, that's just understanding this is the
way that market works. When the value falls, your counterparty
asks for collateral. They missed that aspect of the model. That
had nothing to do with predicting the future but just in
understanding how that worked, and that ultimately led to the
demise. And you see that pervasive through many, many different
types of models throughout the system.
Chairman Miller. It does remind me of Yoga Berra's wisdom
that predictions are difficult, especially about the future.
Mr. Rickards?
Mr. Rickards. Mr. Chairman, I think it is interesting that
a number of Members and the witnesses today have referred to
planetary motion as an example of models that work, but I will
remind everyone that from 200 B.C. to 1500 A.D., the model of
the universe was a geocentric model in which the sun revolved
around the Earth. And it was obvious because you woke up in the
morning and the sun came up over here and went down over there,
and that was not just a religious belief, that was actually a
scientific belief, and many brilliant mathematicians worked for
centuries to write the equations. They weren't automated, of
course, but they wrote those models, and when people began
observing data from improved telescopes that didn't conform to
that model, they said well, we just need to tweak the model a
little bit. Instead of these cycles, they created epicycles.
They were little twirls within the big twirls, and they kept
going down that path. The model was completely wrong. Actually,
the model was right, the paradigm was wrong. The understanding
of how the world worked was wrong. The sun did not revolve
around the Earth; the Earth revolved around the sun.
That is my view of today. You can tweak it, you can improve
it, you can separate the so-called fat tail and zero in on that
tail, and there is a complex method called GARCH, Generalized
Autoregressive Condition Heteroskedasticity and variations on
that. They are all wrong because the paradigm is wrong, because
the risk is not normally distributed in the first place. So I
think these are fatally flawed.
If a hedge fund that is non-systemically important wants to
use this model, that is fine. They can use voodoo, as far as I
am concerned, but if you are talking about a bank or regulated
financial institution, they should be prohibited because they
don't work.
Chairman Miller. Mr. Whalen.
Mr. Whalen. I agree with them, and also I think Dr. Berman
made this point. When you are talking about safety and
soundness, you don't want to look at a tactical short-term loss
possibility, you want to look at the worst case, and we see
that now with the banking industry. By next year, I think we
are going to be looking at a double- or triple-digit deficit in
the insurance fund, and banks are going to have to pay that
back. No one anticipated that magnitude of loss. So what you
have is, on the one hand, a marketplace which is very short
term. They are working on today's earnings, next quarter's
earnings, what have you, and yet over time, since Glass-
Steagall, we have slowly eroded the limits on risk taking. So
the models--whether they worked or not is kind of irrelevant.
We slowly allowed banks to take more and more risk. So I think
what we have to do first is say, what risk do we want the
utility side of this industry, the depository, the lending, the
cash distribution part of banks, to take, and what part do we
want to force, for example, into the hedge fund community,
which is a perfect place for risk taking.
You know, we can't come up with the answer to your
question, Mr. Chairman, as to safety and soundness and capital
adequacy, unless we quantify the risks that the institutions
take. I will give you an example. Citigroup in 1991 peaked at
about three and a half percent charge-offs versus total loans.
I think they are going to get up to about six this time. Now,
can you imagine the public and market reaction when the large
money centers get up to something like two, maybe two and a
half times their 1990 loss rate? But that is how severe of a
skew we are seeing. In the Depression, we got up to five
percent losses on total loans, so we are closing in on the
1930s. I don't think it will be quite that bad, but we will see
how long we stay there. That is the other question, how long
will we see those losses? Will it be two quarters or four? This
is the kind of question you need to answer very precisely but
the only way you can answer your question about capital and
safety and soundness is if you first quantify the risk taking,
because otherwise I don't think you can get an answer.
And by the way, we wrote about this last week. I don't
think you can ask the markets to give more capital to banks. I
think the G-20 and Secretary Geithner are wrong. You have to
reduce the risk taking, because I don't think the markets would
let J.P. Morgan have 20 percent capital assets because the
returns will be too low. It would be low single digits at best
and on a risk-adjusted basis I think they would be negative.
This, by the way, is the context you ought to bear in mind.
Most of the large banks on a risk-adjusted basis really aren't
that profitable. It is only the super-normal returns that they
get from OTC derivatives, investment banking, proprietary
trading that helped the whole enterprise look profitable. If
you look at the retail side, the cash securities trading, it is
barely profitable, really, and that is why you have seen the
changes in the industry that you have.
Chairman Miller. Dr. Colander.
Dr. Colander. In answer to your question, in social science
you will never get the amount of exactness that you will get in
natural sciences, mainly because the agent in social science is
not an atom which sort of follows a set of rules, you know, it
is a human being, it is an agent who will try to do everything
he can to screw you every time you are trying to control him.
So the thought that you are going to be able to design any
model is pretty much impossible. That being said, I think
models can be used and have to be used. We all use models. How
can you not sort of picture what is going on? The question is,
what type of models, and how many different models do you have
in your mind, and how quickly can you jump from one model to
another and recognize we have really moved there, and that is
the issue that I think people are talking about.
Proposals for Avoiding Recurrences of Financial Problems
Chairman Miller. Interesting set of answers. Mr. Rickards
did mention at least three proposals for avoiding a catastrophe
like what we have had. Do the rest of you have specific
proposals as well of how we avoid this again? I think the
financial industry is already treating what happened last
September, October--we are still in it--as a hiccup, something
that was a fluke, will not happen again, we don't have to
change conduct very much. I assume all of you don't agree with
that, but what is it that we should do?
Dr. Berman. I think there are two courses of action. I
think most of the discussion on regulatory capital is trying to
solve a symptom as opposed to the cure itself. A good portion
of the funds come from investors who are feeding the big
bonuses, let's say, at large banks, so while there is lots of
talk about the restriction on bonuses and whether we should
hold people legally liable for clawbacks, et cetera, the fact
is that the fuel is there. That fuel causes crisis. The fuel is
done generally by greed, but mostly uninformed greed. Probably
the number one thing that regulators can enforce is better
transparency and better disclosure on finance itself. If more
people understood what they were actually buying, less people
would buy these things. Wall Street is a marketing arm as are
all commercial companies. Their practices came from the desires
of people to invest in those products, invest in those
services, and invest in the companies themselves. If we don't
like those practices, then we should make it clear what those
practices are and let investors choose whether or not they want
to engage in those. That would dampen further--well, certainly
it would help reverse this crisis a bit, and it would certainly
dampen the ability for the market to even create these very,
very large bubbles in the first place.
Mr. Whalen. One simple thing that I would add to Dr.
Berman's comment, and speaking as an investment banker, make
the lawyers your friend. What you want to do is, instead of
allowing banks to bring these structured assets and derivatives
in an unregistered forum, you force them to register with the
SEC, and what that does is two things. First off, the lawyers
of the deals will not allow more than a certain degree of
complexity, because once that deal is registered, it can be
purchased by all investors, and so they will force simplicity
onto their banks. Because otherwise they will get sued, and the
trial lawyers will enforce this, believe me. Remember, most of
the toxic waste, the complex structured assets, were all done
at private placements. You can't even get a copy of the
prospectus.
The second thing I would tell you is that, you know, in
terms of overall market structure, we've got to decide whether
or not, going back to my earlier comment, we are going to allow
people to contrive of any security for any investor that
doesn't have some rational basis, some objective basis in terms
of valuation, because that is really the key problem that we
have all faced over the last couple years, is valuation. When
the investors realized that they couldn't get a bid from the
dealer that sold them the CDO and they couldn't value it by
going to anybody in the cottage community, they just withdrew
from the market and we had a liquidity problem. If you force
these deals to be registered, guess what? Every month when the
servicer data for the underlying becomes available, they will
have to drop an 8K and then that data will be available to the
community for free. We won't have to spend hundreds of
thousands of dollars a year to buy servicer data so that we can
manually construct models to try and understand how a very
complicated mortgage security, for example, is going to
perform. You will open up the transparency so that the cottage
industry that currently supports valuation for simple
structures, which are very easy to value--credit card deals,
auto deals--there is really no problem with these and they are
coming back, by the way. You are starting to see volume come
back to that market. It is about disclosure. I think Dr. Berman
says it very well.
Abuse of the VaR
Chairman Miller. Dr. Colander? You don't have to speak on
every topic if you don't want to.
Dr. Berman, everyone agrees that the VaR can be abused, has
been abused, was certainly used foolishly in lowering capital
requirements for investment banks. Without revealing
proprietary information, can you give us some of the ways that
you have seen firms abuse the VaR, or try to abuse the VaR
apart from regulatory matters?
Dr. Berman. Sure. I don't think that VaR in itself was
purposefully or willfully abused. VaR is a model that requires
a significant number of assumptions. For example, if I buy a
product, such as an option, then I should assume that if the
value of the stock goes down, then the value of the option will
go down. If I write that option, so I sell it, then if the
stock goes up, I can lose a lot of money. If I don't have the
desire or the technology or the capability or the incentive to
bother being careful about that, then I will assume that, if
the stock goes up, I will make or lose a limited amount of
money. That is a very, very poor assumption, which I think we
have heard a lot today. If you take many of those poor
assumptions and you add them up, you wind up getting VaR
numbers, and not just VaR numbers but numbers of all sorts of
different models that wind up being all but meaningless because
of so many small poor assumptions that have added up into
something that is just wildly incorrect. But folks like to
believe their own numbers, especially when those numbers allow
them to do things that they weren't able to do before. So it
wasn't a willful misconduct as much as a carelessness, given
the incentive structures that are out there today.
Chairman Miller. Anyone else? Mr. Rickards.
Past Congressional Attempts to Regulate the Financial Industry
Mr. Rickards. Yes, Mr. Chairman, I just want to say that my
recommendations, if we are going back to something like Glass-
Steagall, there was more to that than just a walk down Memory
Lane. I am not saying, gee, the system today has obviously
failed, let us go back to what we had before. I actually
derived these from my own research into the power load
relationship that I talked about earlier, which is that scale--
as scale goes up, as you triple or quadruple or increase by
five or ten times the system, you are increasing risk by a
factor of 100, 1,000, 10,000. That is the non-linear
relationship that Dr. Taleb talked about earlier, and I very
quickly came to the conclusion--well, if that is the problem,
then descaling is the answer, and Glass-Steagall is an example
of that. There is a little bit, I think, of--you know, easy
with hindsight, but perhaps some arrogance in the 1998-2001
period where I think Members looked back at the Congress in the
1930s and said, you know, they were Neanderthals, they didn't
understand modern finance, they created this system. The
Members of Congress in the 1930s had actually lived through
something very similar to what we are living through now and
this was their solution. They actually had firsthand
experience.
Now, did a Member of Congress in 1934 understand fractal
mathematics? No, it was invented in the 1960s. But they had an
intuitive feel for the risks and I think their solution--we had
a system that worked from 1934 to 1999, for 65 years. When the
savings & loan (S&L) crisis happened in the early 1990s, it
didn't take hedge funds with it. When we had the banking crisis
in the mid-1980s, it didn't affect the S&L industry or it
didn't affect investment banking. We were compartmented, and
that is what saved the system. We have torn down all the walls.
Commercial banks look like hedge funds. Investment banks look
like hedge funds. Hedge funds originate commercial loans. It is
a big business for them. So when everyone else is in everyone
else's business, should it come as any surprise that if one
part fails, it all fails.
Chairman Miller. Thank you.
Should a Government Agency Test Financial Products for
Usefulness?
Dr. Taleb earlier suggested that there be something like
the FDA that approves--actually it was not clear to me in the
earlier panel to what extent they were calling for government
conduct or setting rules by government that would prohibit
things, or just people not doing them because they were stupid,
but assuming we are talking about rules that may be set by
government, Dr. Taleb suggested that the FDA reviews drugs to
see if they do any good, they don't allow--the FDA doesn't
allow patent medicines mixed up in a bathtub to be sold to cure
cancer anymore. You could do all that you wanted in the 1930s.
You can't do it now. And a great many of the financial
instruments that led to all this have no readily apparent
social utility and create enormous risk that is dimly
understood by even the people who are selling them, certainly
the CEOs and the boards of directors of their institutions.
Should we be reviewing financial instruments for whether they
have any useful purpose, and can you give examples of
instruments that have no apparent purpose and have done great
damage? Mr. Whalen.
Mr. Whalen. Well, I think the short answer is no. I am not
a big fan of regulation. I don't think the government has the
competency to analyze complex securities in the first place.
You would have to hire the people that do it. I think it is
better to let the market discipline this behavior. Large buy-
side investors, who I would remind you are probably the
survivors of this period, they are the ones with the money,
they tell the sell side what they want and they are going to
tell the rating agencies what they want to see as well, and if
you increase the liability to the issuers by forcing
disclosure, by forcing SEC registration, you are going to see
simplicity. Because otherwise my friends at the trial bar are
going to come over the hill like the barbarians, and they are
going to feast, and I think that is the way you do it. You
don't want to get the government into a role where they have to
make judgments about securities, because, frankly, who would
you ask? The folks at the Fed? I mean, the Fed is populated by
monetary economists who couldn't even work on Wall Street. I
mean, I love them dearly, I go fishing with a lot of these
people but I would not ever let them have any operational
responsibility because they just don't have the competency.
So I think we have to try and take a minimalist approach
that is effective, and the way you do that is by making the
issuer retain a portion of the deal that they bring so that
they have to own some of the risk. You make them make a market
in these securities too. They can't just abandon their clients
when they bring some complex deal and not even make a bid for
it. That is a big part of the problem. If you make the dealers
retain some risk and retain responsibility, then I think you
will see change.
Chairman Miller. Dr. Colander.
Dr. Colander. I wanted to expand a little bit on regulation
from a different perspective, again, agreeing very much with
what Mr. Whalen said, that there is a problem with government
regulation, and we can go back and think about Glass-Steagall.
You know, people responded to Glass-Steagall and said here is
the problem, you know, that we deregulated. The problem was,
during that time there was enormous technological change. We
had to change the regulations, and now you have to--regulation
isn't a one-time thing. It has got to be continually changed,
and here is the problem. My students, when we asked how many
were going on, you know, sort of--Paul Volker came up and spoke
and he said, you know, big audience, ``How many of you are
planning to go on and work for government?'' and I think two
people raised their hand. Then he said, ``How many people are
planning to go on to Wall Street?'' You know, you had all this
large number, and this was a number of years ago. When my
students coming out of Middlebury College as seniors can earn
$150,000 to $200,000 in the first or second year and somebody
coming into government can get, what, as a GS-8 or 9, you know,
sort of $34,000 or something. You know, where are you going to
go, how are you going to get the expertise to do it? And so
what happens is, you know, you have an unfair system there,
where no matter how much regulation you get, given the pay
structure, given what's there, the people who are having it
designed will be able to snow anybody who is trying to regulate
it, and that is why very much I think you have to design it,
not so we have to regulate it, but it is self-regulatory, and
that, I think, is what you are hearing from people, that you
have responsibility. If it's too big to fail, we have to
regulate it so therefore let us see that is not too big to fail
by making it smaller, that we structure it by the people who
know the institutional structure, so that here you figure why
you won't make that deal. But not for government to be coming
in mainly because government will get beat.
Chairman Miller. Mr. Rickards.
Mr. Rickards. Mr. Chairman, I think the idea that there
would be a government panel of some kind that would vet and
approve financial products in the manner that the FDA approves
drugs is probably not workable, probably beyond the ability of
government. But for example, credit default swaps: There is
actually a use for them. They are socially useful when they are
used to hedge a position in the underlying bond, but they
become a casino ultimately underwritten by the taxpayers when
they are used with no insurable interest. So it is hard enough
understanding what a credit default swap is, but to get that
distinction just right, when it may or may not be useful, would
be extremely difficult. But I do believe there should be a
quarantine in the sense that--let's have these products in
hedge funds, in long-run investors or maybe with mild leverage.
Let us keep them out of FDIC-insured banks and other
institutions that perform this utility function and are in
effect gambling with taxpayers' money.
I also endorse Dr. Colander's suggestion that, in the
National Science Foundation, in the peer review process, there
is a rule for looking at these things, perhaps not in the
regulatory sense of approving them but in the academic sense of
understanding them. And I believe what Dr. Colander is
referring to is what I call `cognitive diversity.' Let's just
not have a bunch of economists or, for that matter, a bunch of
physicists but let us have physicists, economists, behavioral
scientists, psychologists and others work together. I think it
is interesting that Dr. Kahneman at Princeton won the Nobel
Prize in economics a few years ago. He is the world's leading
behavioral psychologist. He wouldn't describe himself as an
economist, but he made very valuable contributions to
economics.
If you get 16 Ph.D.'s in a room and they all went to one of
four schools, let us say Chicago, MIT, Harvard and Stanford,
and they are all fine schools, you will actually improve the
decision-making if you ask two or three of them to leave and
invite in the first couple people who walk down the street. You
will lower the average IQ, but you will improve the overall
outcome because those people will know something that the
Ph.D.'s don't. So at the National Science Foundation level, to
encourage that kind of collaboration I think is very valuable.
I have actually--I am involved in a field called
econophysics which basically is understanding economics using
some physics tools, and I don't claim it is the answer to all
these things but it does make some valuable contributions. But
when I speak to--I have spoken at Los Alamos and the Applied
Physical Laboratory, and I get a very warm reception. The
physicists are very intrigued and they see the applications.
When you talk to economists, they have no interest. They are
like, what do physicists have to tell us. And I think more
collaboration would be helpful.
Identifying Firms That Are `Too Big to Fail'
Chairman Miller. Mr. Rickards, you know from personal
experience that it is not just depository institutions that are
systemically significant. How do we identify--I think you and
Dr. Colander both have spoken about the problem of scale. How
do we reduce the size of institutions? How do we identify the
ones that are systemically important, either because of their
size or their interconnectedness, as inappropriate for the kind
of risk taking . . . that if we assume that there are some
institutions, most hedge funds, that can be born and die
without any great consequence to the rest of the planet, and
that if they want to use voodoo, they can. How do we identify
those that we have different standards for? Mr. Whalen.
Mr. Whalen. Well, I think there is two simple answers.
First off, we have to revisit market share limits. You have
already seen this in process with the FDIC because they have
started to levy premiums against total assets, less capital,
instead of domestic deposits. I think that is a very healthy
change. But perhaps more important, we have to let institutions
fail, because if you convince investors that you are going to
put a Lehman Brothers or a Washington Mutual into bankruptcy,
they are going to change their behavior, and I think both of
those events were inevitable, by the way. I think it is
ridiculous to argue that Lehman could be saved. They were for
sale for almost a year. Nobody wanted to buy it. So, you know,
at the end of the day, if we don't allow failure, and we don't
inoculate our population against risk by letting them feel some
pain from time to time, then we will repeat the mistake.
Last point, we have got to get the Fed out of bank
supervision. Monetary economists like big banks. They love
them. I worked in the applications area of the Fed in New York.
I can't recall a merger, a large bank merger that they have
ever said no to. I worked on the ``Manny Hanny'' (Manufacturers
Hanover Trust) transaction, I worked on the Chemical Bank
merger, following that with Chase. In each case, you could make
a very strong case that those were bad mergers. They destroyed
value. And then look at Bank of America. They had to buy
Countrywide because they were the chief lender to Countrywide's
conduit. They had no choice. It was kind of like J.P. Morgan
buying Bear Stearns. There really was no choice. But then we
have the Fed slam Merrill Lynch into Bank of America to save a
primary dealer. These are the monetary economists saying oh,
dear, we want to have a few big primary dealers we can manage
and deal with, it is easier for us. Well, I think that is a
really skewed perspective. I would like to see another agency
responsible for approving mergers of financial institutions
that actually looks at it on an objective basis and says, is
this a good idea, because we have got a couple mergers, Wells,
Wachovia and Bank of America with Merrill Lynch that I am not
sure are going to work. I think both of those institutions may
have to be restructured and downsized significantly in the next
couple of years.
Chairman Miller. Mr. Rickards.
Monitoring and Analyzing Hedge Fund Activity and Risk
Mr. Rickards. Mr. Chairman, on the issue of what is a
systemically important hedge fund, at the end of the day there
will be some element of subjectivity in it--whether a $10
billion hedge fund is systemically important, but if you have
$9.8 billion, you are not. It will be a little bit arbitrary
and it can't be based solely on size. It has to be based on the
complexity. But the first step is transparency. You will never
be able to make any informed decisions like that without good
information, and every hedge fund manager--I have worked in
hedge funds banks and investment banks--they will resist that
for various reasons but I don't understand why the United
States Government couldn't create a facility that would keep
that information on a secure basis. We keep military secrets,
we keep intelligence secrets, we keep other information
confidential. You could have a firm like, you know, IBM Global
Services that would come in, build a facility. It could be
secure, get clear people running it, and then just say to all
hedge funds, look, you have to give us all of your information,
all of your positions, all of your factors in a standardized
format, in an automated format once a week, we will keep it in
a totally secure environment, it will not leak out, but we are
going to take that and load it into, you know, covariance
metrics. We are also going to do that for your firm, and we are
going to have an idea at that point when you are taking
systemic risks, and at that point there ought to be an ability
to intervene. And I agree with Mr. Whalen, it should not be the
Federal Reserve. They do a lousy job with their primary task of
preventing inflation, and I don't know why they have been given
all these other jobs. But there certainly would be expertise in
the government to do that much, and then to intervene when
necessary.
Dr. Berman. Adding that, taking all that data, bringing it
together----
Chairman Miller. Dr. Berman.
Dr. Berman.--and putting it into a large covariance matrix
sort of sounds like VaR. I mean, that is--so you come back to
those same questions all the time when you say how do we make
predictions? This may sound like I am answering the question
with the same exact question, but the best way to protect
against this is to ask the question to the bank: what would
happen if you failed? And then determine what the outcome to
society or to the economy would be. It is not based on the size
of the bank, it is based on, look at what would happen, not the
probability but if the bank fails, if a hedge fund fails, what
actually will wind up happening, what are the knock-on effects.
That requires an enormous amount of transparency but you don't
need to necessarily make the predictions about that, you just
need to follow that thread through.
Chairman Miller. Dr. Colander.
Dr. Colander. One of the principles, you know, sort of
within economics, is taxes have to have reasons and everything
else. And I think one of the things that thinking of the
economy as a complex system brings up is that bigness is, per
se, bad, you know, sort of an interconnection is, so we have
lost the sense that there can be a tax on `bigness' so that
people can decide but it is built within that. And to start
thinking that, here, if you have a complex system, you have got
to keep a whole number of different elements, and the only way
you are going to be able to do that--because there is enormous
pressure to grow--is to somehow design within the system a
counterweight to that, and so thinking along those lines, I
think is something that follows thinking of the economy as a
complex system.
Chairman Miller. Mr. Whalen.
Mr. Whalen. I will come back to something Mr. Broun said
about the community banks because I think it is very important,
and you all are going to be hearing about this a lot next year.
If you are going to tax institutions based on risk, and I think
that is sound, you start with the FDIC. The big banks should
pay more than the little banks, and when we see the size of the
hole that we have to fill in over the next, I don't know, 25
years from this crisis, I think that is going to become a very
compelling argument. The community bankers are going to be
living up here next year when they start seeing the estimates
for what they have to give up in revenue and income to fill in
this hole. Remember, we are still paying for the S&L crisis.
There is still debt out there that we are paying interest on.
We are going to be paying for this crisis for 100 years. That
is how big the numbers are. So think of that as a load on the
economy. That is kind of the cost of modeling run amuck, and,
you know, I am serious about this. We are going to be paying
for this, the banking industry, consumers, investors in banks
are going to be paying for this for many, many decades.
Chairman Miller. We are--Mr. Rickards.
Mr. Rickards. Just briefly. The inverse of complexity is
scale. You can have complexity at the small scale, a medium
scale or a large scale. Failure at the first two will not
destroy you. Failure at the third may, and so I am not against
complexity. There is going to be complexity. But, again, Dr.
Taleb's example, an elephant is a very complex organism, but if
it dies, the entire ecosystem doesn't crash. And so let us keep
these things in boxes and reduce the scale as the antidote to
complexity.
Chairman Miller. We are at the end of our time, but I want
to thank all of you for being here. Under the rules of the
Committee, the record will remain open for two weeks for
additional statements from Members and for answers to any
follow-up questions the Committee may have for the witnesses.
Again, I appreciate your willingness to come and testify, and
it will be useful to have all of you as resources for the
future as well. Thank you very much. The witnesses are excused
and the hearing is now adjourned.
[Whereupon, at 1:30 p.m., the Subcommittee was adjourned.]