Full text of Economic Policy Review (Federal Reserve Bank of New York) : Volume 13, Number 2 : New Directions for Understanding Systemic Risk

View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Preface
Guarding against systemic risk in the financial system is a key undertaking for
central banks. Defining this type of risk is difficult, but managing it with precision
is harder still. Complicating this task is the fact that institutional consolidation,
a broadening range of financial products, and greater connectivity among firms
have in recent decades materially changed the nature of systemic risk in the
financial system.
To stimulate fresh thinking on systemic risk, the Federal Reserve Bank of New York
and the National Academy of Sciences’ Board on Mathematical Sciences and Their
Applications cosponsored the conference “New Directions for Understanding
Systemic Risk” in May 2006. The main goal of the sessions was to explore parallels
between systemic risk in the financial sector and in selected domains of engineering,
ecology, and other fields of science. The event attracted more than 100 experts on
systemic risk from 22 countries, representing banks, regulators, investment firms,
U.S. national laboratories, government agencies, and universities. In addition to
bringing together many participants with backgrounds in banking, finance,
and economics, the conference broadened the discussions by including the
perspectives of mathematicians, statisticians, operations researchers, ecologists,
engineers, and physicists.
Although the topic of systemic risk may call to mind the possibility of deliberate
attacks, both cyber and terrorist, on the financial system, after careful consideration
the conference organizers decided against emphasizing this source of systemic risk.
They reasoned that such a focus would downplay the many ways in which systemic
risks can arise during the financial system’s normal operations. Analysis of the risks
of deliberate attacks might build on the concepts explored in the conference, but it
would require additional considerations and tools.
This volume was prepared to share some of the insight and excitement generated
by the conference and to encourage further cross-disciplinary conversations.
We hope you find it useful and informative.

—The Report Editors

Part 1: Introduction

he stability of the financial system and the potential for
systemic events to alter its functioning have long been
critical issues for central bankers and researchers. Developments such as securitization and greater tradability of financial
instruments, the rise in industry consolidation, growing crossborder financial activity, terrorist threats, and a higher
dependence on computer technologies underscore the
importance of this research area. Recent events, however, such
as the terrorist attacks of September 11, 2001, and the collapse
of the hedge fund Long-Term Capital Management (LTCM),
suggest that older models of systemic shocks in the financial
system may no longer fully capture the possible channels of
propagation and feedback arising from major disturbances.
Nor can existing models account entirely for the increasing
complexity of the financial system, the spectrum of financial
and information flows, or the endogenous behavior of different
agents in the system. Fresh thinking on systemic risk is
therefore required.
With that goal in mind, in May 2006 the National Academy
of Sciences and the Federal Reserve Bank of New York
convened a conference in New York to promote a better
understanding of systemic risk. The sessions brought together
a broad group of scientists, engineers, economists, and
financial market practitioners to engage in a cross-disciplinary
examination of systemic risk that could yield insights from
the natural and physical sciences useful to researchers in
economics and finance.1 Accordingly, presenters from the
natural and mathematical sciences and the engineering
disciplines provided examples of tools and techniques used
1

The conference program can be found in Appendix A.

to study systemic collapse in interactive systems in nature
and engineering. Similarly, research economists presented
studies of systemic risk in cross-border investments, liquidity
risk, and the payments system. To provide a context for
the discussions, risk managers at large finance institutions
described how systemic risk and shocks in the financial system
affect trading activities.

Transitioning from a Bank-Based
to a Market-Based Financial System
Financial market practitioners began the conference by
highlighting various aspects of systemic risk and systemic
events in the financial system. The topics of the presentations
ranged from historical systemic episodes, such as the liquidity
crisis of 1998 and the failure of LTCM, to risk assessment
techniques, such as value-at-risk (VaR) analysis and scenario
analysis. Charles Lucas of AIG (since retired), a member of
the National Academy’s Board on Mathematical Sciences and
Their Applications, introduced the first session by asking
the fundamental question: What is systemic risk?
According to Lucas, economists’ theoretical understanding
of systemic risk stemmed from the experience of the Great
Depression and specifically from John Maynard Keynes’s
interpretation of that experience in General Theory of
Employment, Interest, and Money. Keynes aimed the
formulation of his “general theory” at capturing the dynamics
that allowed an economy to transition to an inferior but stable

The views expressed in this summary do not necessarily reflect the position
of the Federal Reserve Bank of New York or the Federal Reserve System.

FRBNY Economic Policy Review / November 2007

equilibrium, in the process overturning the normal fullemployment equilibrium that defined classical models. During
the Great Depression, the economy underwent a shock that was
sustained by sympathetic movements throughout the financial
system—a sequence of events that has come to be called
“contagion.” Because of policy missteps and a feedback loop
with the financial system, the real economy settled into a
persistent state of underutilized resources and unemployment.
Despite structural changes since that time, the idea of a
feedback loop between the financial and real sectors of the
economy that leads to an inferior equilibrium with negative
consequences for the real economy remains pertinent to
current analysis of financial stability.
That system has changed dramatically since the Great
Depression, as described in the conference background
paper on the evolution of systemic risk.2 Though banks
still play a large role, many functions that defined their
traditional domain are increasingly performed by securities

As the relative importance of banks as
financial intermediaries has declined with
the growth of market-based financial
intermediation, market-based systemic
events such as the stock market crash of
1987 and the failure of [Long-Term Capital
Management] have shifted the emphasis
from funding liquidity to market liquidity.
markets and nonbank market participants. For example,
hedge funds, private equity groups, and other fund
managers now control larger shares of financial capital and
take active roles in asset and credit markets. Crises in this
more market-based financial system, such as the stock
market crash of October 1987 and the market liquidity crisis
of 1998, fit a general pattern of rapid decline in the price of
some asset or class of assets, leading to a drop in liquidity.
The result is contagion, in the form of further sympathetic
price declines and a shift in market conditions marked by
severely reduced financial market activity and potential
negative effects on the real economy. Likewise, although the
linkage mechanisms may have changed completely since the
Great Depression, and despite the skepticism Keynes’s
2

“Systemic Risk and the Financial System,” by Darryll Hendricks, John
Kambhu, and Patricia Mosser; the paper can be found in Appendix B of this
volume.

Introduction

theory received from the research community at the time,
this simple model captures the mechanisms underlying the
Depression. The endogenous shock in the United States that
led to the inferior equilibrium then was a stock market crash
followed by a wave of bank runs and loss of liquidity in a
feedback mechanism of self-fulfilling prophecy. As Lucas
suggested, this comparison offers a basic historical analogy
illuminating some of the modern phenomena of systemic
risks, such as sudden “regime shifts” in the financial system
and the role of feedback mechanisms.
The conference background paper by Hendricks, Kambhu,
and Mosser also describes the now well-studied phenomenon
of the “bank run.” In the classical model, a commercial bank
makes illiquid loans on the asset side of its balance sheet, and
takes demand deposits on the liabilities side that it is obligated
to pay back at any time. In a bank run, even though each
depositor would be willing to leave his or her funds on deposit,
the belief that other depositors are likely to withdraw theirs
causes all rational depositors to try to withdraw their funds as
quickly as possible. A run on the bank results, because the
bank’s loans cannot be liquidated immediately at their full
value, leaving the bank with no funds for the last depositors in
line. In such a scenario, a run can be triggered by concerns
about liquidity even if the bank is otherwise solvent.
Moreover, in this model, self-fulfilling prophecies can make
bank runs contagious: If depositors witness a run on one bank,
they may believe that runs are more likely to occur on others.
This scenario can be attributed to several factors. For example,
the issue that sparked a run on one bank, such as excessive loan
exposure to real estate or the oil industry, may be perceived to
affect other banks, or one or more other banks may have
significant interbank exposures to the affected institution.
As the Great Depression revealed, the withdrawal of
funding liquidity resulting from bank runs can accentuate
economic downturns and generally influence the real
economy as lending is curtailed to creditworthy entities.3
Thus, the primary policy approaches to managing financial
instability in a bank-oriented financial system—lender-oflast-resort facilities by the central bank, deposit insurance,
and banking supervision to ensure credit quality in loan
portfolios—were all aimed at preventing or mitigating
the effects of these potentially catastrophic withdrawals
of funding liquidity from the system. As the relative
importance of banks as financial intermediaries has declined
with the growth of market-based financial intermediation,
market-based systemic events such as the stock market crash
of 1987 and the failure of LTCM have shifted the emphasis
from funding liquidity to market liquidity. Moreover, as
3

See the conference background paper described in footnote 2.

Federal Reserve Board Governor Donald L. Kohn observed,
the Federal Reserve is midway through a long process of
adapting its policy tools to this new environment.4
As Hendricks, Kambhu, and Mosser describe in their
conference background paper (Appendix B of this volume),
the shift from a financial system dominated by banks to one
dominated by markets has as its hallmark a broadening of the
types of activities that banks and other financial intermediaries
engage in and the assets that they invest in. The large financial
institutions at the core of the system now intermediate the
movement of capital in many ways: They assist businesses in
the issuance of new stocks and bonds directly to the market
(investment banking), they intermediate secondary-market
trading of stocks and bonds after issuance on behalf of clients
(market making), they lend directly to households and
businesses (traditional commercial banking), and they manage
asset portfolios on behalf of individuals and institutions (asset
management). This latter example has led to the development
of new market entities that act as vehicles for household
savings, such as mutual funds and pension funds, as well as
more leveraged entities, such as hedge funds.
The conference background paper also explains how a
securities-market-based financial system works best when
capital markets are liquid. In this context, liquidity refers to
tradability: Markets are liquid when any individual trade is
unlikely to have a major effect on the asset price because large
numbers of willing traders are on the buy and sell sides of the
market. Liquidity normally rests on a number of foundations;
foremost are market making, trading, and arbitrage. Market
makers buy and sell securities out of inventory they maintain
to meet customer demand, thereby providing intertemporal
liquidity to smooth out short-run imbalances in supply and
demand. Traders contribute to market liquidity by trading on
bets that prices will converge to long-run fundamental levels.
This activity speeds the convergence of prices to fundamental
levels and provides stability to the market. Systemic shocks
occur when one of these foundations is compromised.
Market-oriented crises tend to begin with a large asset price
decline that becomes self-sustaining. Normally, when asset
prices drop sharply, investors step up to buy assets that have
declined sufficiently—an action that largely prevents market
stress from worsening. This type of stabilizing correction is
natural for a well-functioning, efficient asset market. In
systemic crises, however, investors and traders are either
unable or unwilling to step in, perhaps because their losses
have limited their risk-taking and market-making capacity or
because a structural failure in, say, the payments or settlement
system has made such a step difficult. As prices decline, more
market participants either sell from a change in their risk
4

Governor Kohn’s observations, as reported in this summary, are based on his
conference remarks, “The Evolving Nature of the Financial System: Financial
Crises and the Role of the Central Bank.”

appetite or are forced to sell by a tightening of financing
constraints, and prices are pushed down. Like the selffulfilling-prophecy aspect of the bank-run model, this
sequence of events can be self-reinforcing as market
participants retire to the sidelines.
Market-based systemic crises are often characterized
by a coordination failure: A wide cross-section of market
participants simultaneously decide to reduce risk taking and
effectively refrain from financial activities, such as trading

Market-based systemic crises are often
characterized by a coordination failure:
A wide cross-section of market
participants simultaneously decide to
reduce risk taking and effectively refrain
from financial activities.
stocks, issuing debt and equity, and lending. While no one
institution is necessarily insolvent or illiquid, each firm reduces
its activity and risk to protect capital. In aggregate, the firms’
actions combine to reduce financial market activity severely
as asset prices fall, possibly harming the real economy in the
process as the provision of financial services to otherwise
creditworthy entities is curtailed and declines in asset prices
impact firms’ balance sheets. As Governor Kohn explained, the
stock market crash of 1987 followed this pattern: simultaneous
efforts to reduce equity market exposures were followed by
a broad pullback in all risk taking.
In a market-based systemic crisis, as in the bank-run model,
the actions and beliefs of individual participants across the
financial system can combine to disrupt the entire system, even
though the great majority of institutions are not at risk of
collapse.5 When cast in these terms, the notion of systemic risk
in the financial system bears a strong resemblance to the
dynamics of many complex adaptive systems in the physical
world. Many of the features of complex systems described by
conference participants from the natural and mathematical
sciences are clearly present in the financial system. For
example, Simon Levin of Princeton University cited nonlinearities, multiple stable states, hysteresis, contagion, and
synchrony as features common to all complex adaptive
systems. These features are evident in models of financial
crises: 1) contagion is seen in the self-reinforcing character
of price declines and transmission of liquidity shocks across
institutions; 2) multiple stable states and hysteresis can
appear in the move to an inferior but stable equilibrium; and
5

See the conference background paper described in footnote 2.

FRBNY Economic Policy Review / November 2007

3) nonlinearities in expectations and investment decisions can
lead to sharp changes in the volatility and covariation of asset
prices in an apparent regime shift, as discussed by risk
managers later in this introduction. This commonality between
financial and other complex adaptive systems points to the
broad social importance of the study of systemic events.

Systemic Risk as a Generic Problem
In the world at large, complex systems abound. Accordingly,
the instability of these systems and their potential for large and
potentially catastrophic regime shifts are a dominant social
concern—and one of high importance to many environmental
and engineering sciences. For example, atmospheric scientists
examine such questions in the context of climate change, as do

In the world at large, complex systems
abound. Accordingly, the instability of
these systems and their potential for large
and potentially catastrophic regime shifts
are a dominant social concern—and one
of high importance to many environmental
and engineering sciences.
fishery managers concerned with the sudden collapse of certain
economically important fish stocks. As the presentations by
Massoud Amin of the University of Minnesota and Yacov
Haimes of the University of Virginia made clear, engineers
grapple with similar issues to prevent disruptions to the North
American power grid and to analyze for government entities
the wider economic effects of terrorist attacks.
The ubiquity of such problems across so many fields
suggests the possibility of finding common principles at work.
As George Sugihara of the University of California at San Diego
explained, engineers and public health professionals alike may
be interested in how actions to address high-frequency but
low-amplitude events, such as small floods or small outbreaks
of disease, might predispose a system to low-frequency but
very-high-amplitude disturbances. For instance, overuse of
antibiotics to combat small-scale outbreaks of disease can lead
to high-consequence outbreaks of antibiotic-resistant illnesses.
Or, in an example from recent experience, the construction of
levees in New Orleans to protect against intermediate-strength

Introduction

storm surges led to the higher consequence damage from
the lower probability Hurricane Katrina.
Recent studies that have identified many common
characteristics of nonlinear complex adaptive systems in the
physical world may point to a tentative vocabulary of systemic
risk. A key concept that can be used to describe the process of
adverse systemic change in both ecology and finance is the
tendency toward a rapid and large transition from one stable
state to another, possibly less favorable, state—what one might
call a regime shift. Levin cited this phenomenon in the
eutrophication of bodies of water, in which a shock, such as
excessive heat, can lead to overenrichment of the water with
nutrients, resulting in excessive growth of some organisms and
a depletion of oxygen that is damaging to other populations.
The new state of the body of water is a new stable equilibrium.
Somewhat analogously, in financial markets, an exogenous
change in the economic environment can lead to new profit
opportunities in certain assets that attract capital and, if
investment in the assets is excessive, to an asset price bubble
vulnerable to a change in investor confidence. If a shock
triggers a collapse of asset prices, there is a risk of a broader
contraction if the normal self-correcting features of markets
fail to work. Absent those self-corrections, the flight to quality
by investors seeking safe assets could become a self-sustaining
transition to a state with lower levels of credit and real
economic activity.
In Levin’s terminology, in both situations some shock leads
to coordinated behavior within the system, a process known as
“synchrony”—excessive growth of nutrients in the first
example and excessive investment in an asset price bubble in
the second. This synchronized behavior leads to reinforcing
feedbacks, causing the initial shock to spread and cause
contagion. Under the combined effect of the shock and
contagion, a system makes a transition, or regime shift, from a
stable state to an inferior stable state while shedding energy so
that it cannot readily recover its original state, a process known
as “hysteresis.” Levin explained that much of the research on
complex systems in the natural world has focused on the
properties of robustness and resilience to shocks that either can
prevent regime shifts and hysteresis from taking place, or can
lead to recovery if they occur.
The commonality of stability and resilience to shocks in
complex systems suggests that approaches to risk management
in natural and physical systems could be pertinent to financial
risk management. Amin and Haimes each illustrated some of
the methods for managing risk in engineering systems, such as
“multi-objective trade-off analysis,” in which Pareto-optimal
actions are derived by considering the subjective probabilities
and payoffs associated with different shocks. The methods
presented bore some semblance to those used in financial risk

analysis, and much of the subsequent conference discussion
centered around the prospect of adapting methods from
various engineering fields to financial risk management.
Adaptation is clearly necessary because the range of behavior
in financial markets is not mirrored in, say, the behaviors of
humans operating a complex engineered system. A risk analysis
of an engineered system can assume that the people involved
are attempting to fulfill their roles, which are relatively defined,
and share a common objective. In contrast, in the financial
system traders and investors operate in a competitive
environment and might change their roles and behaviors
opportunistically and creatively.

Systemic Risk in the Financial System
Systemic risk in the financial system is difficult to define
precisely. Although a literature on financial crises and systemic
risk exists, a range of views can be found on what constitutes
systemic risk.6
An adage among traders is that, in times of crisis, everything
is correlated. Though conference participants did not share a
consensus on the definition of systemic risk, the descriptions of
systemic events by risk managers at the conference reflected
this view. For example, Thomas Daula of Morgan Stanley
described systemic events as regime shifts in which periods of

Systemic risk in the financial system is
difficult to define precisely. Although a
literature on financial crises and systemic
risk exists, a range of views can be found
on what constitutes systemic risk.
extreme volatility combine with losses of liquidity to produce
solvency risk. These crisis periods, according to Daula, “are
characterized by very sharp increases in correlations and,
therefore, they look and feel a lot like a regime shift—and a
regime shift where you are moving from a normal regime,
where there are relatively low correlations amongst financial
markets, to a different regime, where you have extremely high
volatility and a sharp spike in correlation.”

For a central banking perspective, see Brimmer (1989); Bernanke and Gertler
(1990, pp. 87-114) describe links between financial distress and the real economy.
For a recent paper on systemic risk, see Chan et al. (2006).

Under such regime shifts, the normal assumptions culled
from historical experience that guide day-to-day trading break
down. As D. Wilson Ervin of Credit Suisse observed in regard
to the Russian default and the collapse of LTCM: “The most
memorable part of this episode were the questions around
fundamental issues that were normally unquestioned in dayto-day activities, about the reliability of your counterparties,
about how markets would work under various circumstances,
about whether liquidity would be there under certain
circumstances.” In the presence of such uncertainty and
market panic, traders can tend toward herd movement as they
attempt to avoid losses—what the literature refers to as “phase
locking”—and the normal mechanisms of price determination
can break down. According to Robert Litzenberger of Azimuth
Trust: “What happens is, in what we might refer to as crisis
periods or liquidation periods . . . prices are generated
internally by the market microstructure. Trades that were
previously uncorrelated become correlated because they are
being liquidated at the same time.”
In the tentative vocabulary of systemic risk suggested above,
the self-reinforcing uncertainty and market panic that can
characterize a systemic episode are a clear example of
contagion. The jump in correlations appearing at the onset of
a systemic event can in turn be seen as an example of selfreinforcing feedback and synchrony. Furthermore, the
transition from a normally functioning market to one in which
prices are generated by the internal market microstructure is
accompanied by widespread and simultaneous liquidations.
Financing constraints and the loss of liquidity make a return to
the pre-crisis state very difficult—an asymmetrical transition
and example of hysteresis. Thus, the notion of systemic risk,
which financial market participants are at least viscerally
acquainted with, can be worked into the framework of complex
systems research from other fields.
While conference participants from the financial industry
agreed on the “look” and “feel” of systemic episodes, there was
some diversity of opinion on the more academic question of
what actually constitutes systemic risk or a systemic event.
Darryll Hendricks of UBS compared systemic risk with the
Loch Ness monster: People claim that it exists or must exist, but
nobody can point to a definitive episode. As Hendricks noted,
most definitions of systemic events involve a transmission
of shocks from the financial sector to the real economy—
for instance, disruptions in credit provision as well as a
propagation mechanism such as self-reinforcing feedback.
Therefore, is a systemic event simply one that creates
externalities? Most would probably agree that this is too low
a threshold for classifying an event as systemic. Lucas put
forward the idea that a proper definition of systemic risk would
involve transition from a stable equilibrium to some inferior

FRBNY Economic Policy Review / November 2007

but stable equilibrium, as explained above. This idea accords
well with the regime-shift characterization used by financial
industry participants. However, questions remain about how
to characterize these equilibria. Was an event systemic because
a disturbance propagated across diverse actors through selfreinforcing feedback, or did some policy mistake form the
common cause of the disturbance to all actors, such as
insufficient liquidity provision during the Great Depression?

Systemic Risk and Regulation
With the range of opinions on the proper identification of
systemic risk, it is natural to wonder why the definition is so
important. The 1987 stock market crash and the 1998 nexus
of the Russian default, the failure of LTCM, and the resulting
liquidity crisis were episodes of systemic magnitude propagated and sustained by self-reinforcing mechanisms in the
financial sector, and these episodes had potential consequences
for the real economy. The definition is important, Daula
explained, because regulation to ameliorate systemic risk
constitutes a tax, and therefore a clear understanding of the
risks is needed for the most protection at the lowest potential
cost. Regulation is a tax in the sense that direct expenditures are
required to comply with regulatory directives, and potential
costs imposed in terms of efficiency losses in the allocation
of capital. For an example of the latter, consider capital

A detailed understanding of what
constitutes systemic risk is . . . important
to forming a regulatory regime that
balances costs and benefits.
requirements that vary by the nature of one’s business or the
assets on one’s balance sheet; they can create a wedge between
market prices absent regulation and actual market prices. One
could also wonder if a particular regulatory regime has a cost
effect on the banking sector’s industrial organization. For
example, do certain forms of capital requirements—or, more
generally, the costs of compliance with regulatory regimes—
encourage consolidation?
Given these considerations, the importance of a sound
method for identifying systemic risk becomes obvious.
Without it, policymakers face a strong incentive to build
expansive regulatory regimes capable of influencing practices
that may or may not truly reduce systemic risk, because the

Introduction

potentially disastrous consequences of a real systemic event
would justify the costs of such regulation. As Governor Kohn
stated: “The natural inclination is to take more intrusive
actions that minimize the risks of immediate disruption, and
this inclination is probably exacerbated by ignorance and
uncertainty.” He explained that too much regulation could
harm efficiency or generate moral hazard as market
participants begin to take regulators’ corrective measures for
granted and increase risk taking. For example, they may fail to
engage in adequate due diligence when extending credit or fail
to maintain adequate capital for the risks they undertake.
Further, to borrow from a theme raised by Levin, excessive
regulation could introduce rigidities that may limit the natural
flexibility of markets to respond to shocks.
A detailed understanding of what constitutes systemic risk
is therefore important to forming a regulatory regime that
balances costs and benefits. Indeed, in all the roles
policymakers fill in preventing systemic events and mitigating
systemic risk, a proper analytical framework is crucial for
defining the correct scope and mode of action. For central
bankers in particular, a clear method for identifying systemic
risk and the onset of systemic events is critical for decision
making on whether and how to intervene.

The Roles of Policymakers
The Federal Reserve’s role in setting monetary policy gives it the
ability to mitigate the consequences of systemic events by easing
access to liquidity. After the August 1998 financial market
turmoil associated with the Russian loan default and the
subsequent collapse of LTCM, for example, the Federal Open
Market Committee lowered the target federal funds rate to
soften the effects of “increasing weakness in foreign economies
and of less accommodative financial conditions domestically.”7
Other policymaking roles assumed by the Federal Reserve—
services provider, bank supervisor and regulator, and crisis
manager—also help to position it to mitigate systemic events.
As a financial services provider, the Federal Reserve, through
its Fedwire system, is the backbone of the interbank payments
system. The conference presentation on Fedwire, discussed in
part 4 of this report, highlighted how important this role is
in tempering the effects of a crisis. Referring to the hours after
the attacks of September 11, 2001, the study highlighted
how infrastructure disruptions and the resulting payments
miscoordination threatened to seriously disrupt the payments
system. In response, the Federal Reserve extended the operating
7

Federal Open Market Committee Statement, September 29, 1998 (<http://
www.federalreserve.gov/boarddocs/press/general/1998/19980929/>).

hours of Fedwire and increased liquidity provision by using the
discount window and open market operations, actions that
significantly reduced the impact of the disruption. This episode
exemplifies the Federal Reserve’s role as crisis manager.
Governor Kohn remarked that the Federal Reserve and
other regulatory agencies, as banking supervisors, can do much
to reduce systemic risk by maintaining a healthy banking
system. Collective efforts of regulators and the private sector
to enhance market discipline, improve risk management
practices, and strengthen the clearing and settlement systems
reduce the likelihood that a sharp change in asset prices or
questions about a major market participant will lead to a
systemic financial crisis. In today’s market-dominated
financial system, banks still have a large role to play in
financing traders’ securities positions and in clearing and
settling trades in their brokerage activities. In their role as
providers and conduits of liquidity, healthy banks can be
bulwarks against the propagation of financial turmoil.8
As a crisis manager, the Federal Reserve can avert many
problems by monitoring conditions and identifying risks as they
arise. Indeed, as Governor Kohn explained, a common element in
the Federal Reserve’s response to both the 1987 stock market crisis
and the 1998 liquidity crisis was its public acknowledgment that a
crisis was under way. In announcing a crisis and articulating its
response, the Federal Reserve reassured market participants that
it was working to mitigate the systemic effects of the crisis; such
reassurance can go a long way toward encouraging a return to
risk taking. In both episodes, the Federal Reserve also used open
market operations to ease reserve market conditions and the
stance of monetary policy, monitored the flow of credit through
the financial system, and worked with lenders to emphasize their
collective interest in avoiding a credit gridlock.
The Federal Reserve’s actions relied on an early determination of the potential systemic effects of the two events. Largely
as a result of these actions, neither the 1987 event nor the 1998
episode led to a disruption in real economic activity.

Systemic Risk in Historical
Perspective: The Events of 1998
Governor Kohn observed that the Federal Reserve has been
involved in a long process of adapting its tools to the marketdominated financial system that is still emerging today.
Accordingly, the 1987 stock market crash and the 1998
8

For analysis of banks as liquidity providers in a crisis, see Saidenberg and
Strahan (1999) and Gatev, Schuermann, and Strahan (2006).

liquidity crisis are natural case studies to examine when
defining systemic risk, as neither was triggered by the bank-run
phenomenon that characterized many systemic problems in
the nineteenth century.
The events of 1987 and 1998 had many common elements.
First, both began with sharp movements in asset prices that
were exacerbated by market conditions—portfolio insurance
in 1987 and the closing out of positions in 1998. Second,
market participants became highly uncertain about the

The 1987 stock market crash and the
1998 liquidity crisis are natural case
studies to examine when defining
systemic risk, as neither was triggered
by the bank-run phenomenon that
characterized many systemic problems
in the nineteenth century.
dynamics of the market, the true value of assets, and the future
movement of asset prices. In terms of the regime-shift scenario
described earlier, events outpaced the standard risk management systems, which had been based on historical data and
experience. Third, large and rapid price movements called into
question the creditworthiness of counterparties, which could
no longer be judged by now-obsolete financial statements.
Fourth, the decline in asset prices decreased wealth and raised
the cost of capital, developments that threatened to reduce
both consumption and investment in the real economy.
Although the 1987 and 1998 events shared many features,
conference participants tended to focus on the more recent
1998 episode because financial institutions, instruments, and
practices then were more similar to the way they are today.
The potential negative effects on the real economy and the
systemic character of this episode were highlighted by the
withdrawal of investors from the commercial paper market.
As noted by Ervin, the events of 1998 were catalyzed by the
Russian default. In 1998, Russia was in a precarious position
as a fledgling democracy attempting to transition to a marketbased economy. It had a high dependence on energy exports
at a time when the price of oil was dropping, a massive trade
deficit, an unsustainable pegged exchange rate, and a large
government budget deficit. It was also financed mainly by
short-term debt. Despite a large loan package in July 1998 from
the International Monetary Fund, a sustained reversal in
market sentiment led the Russian government to announce

FRBNY Economic Policy Review / November 2007

in August of that year that it would default on short-term
local-currency debt. The result was disastrous: Many Russian
counterparties failed, and liquidity in Russian instruments
dried up.9
Investor losses were estimated to be on the order of
$100 billion. As Ervin explained, every working assumption
about the Russian market came into doubt: the rules, the
participants, the prices, the functioning of markets, even the
legal system. This was surely a systemic crisis for Russia.
Moreover, it threatened to become a systemic crisis for the
international financial system when the market turmoil
affected a particular hedge fund, LTCM, and the liquidity
of core markets in the financial system.
In the mid-to-late 1990s, LTCM was a very large and
well-known hedge fund, both highly leveraged and highly
successful. Its primary investment strategy centered on finding
arbitrage opportunities or near-arbitrage opportunities in
which the market seemed to be out of line with long-term
economic fundamentals—a trading strategy based on the idea
that certain pricing gaps will close over some (potentially long)
period of time. As part of its trading strategy, Ervin explained,
LTCM would tend to buy older, illiquid Treasury bonds, then
short-sell current, on-the-run Treasury bonds and eke out a
small yield differential between the two. The goal was to
capture a small yield differential in relative asset prices,
allowing LTCM to earn steady returns as relative prices
converged to fair values while the fund avoided directional risk.
At the time at least, this strategy was considered somewhat
state-of-the-art, and many financial entities attempted to
emulate it, if not to mirror LTCM’s positions outright.
The primary problem with LTCM’s strategy was that, as
a relative-value trader, it was very exposed to liquidity shocks
and correlation assumptions, even if the fundamentals underlying its positions were correct. In the days following the
Russian default, there was a large flight to quality in developed
markets that caused credit spreads to widen sharply. Interestingly, this trend was not limited to U.S. corporate debt; interest
rate swap spreads—an indication of the credit conditions
of international banks—also widened sharply. As this shock
rippled through the financial system during August, it also
began to affect equity markets, and the Dow dropped
357 points on August 27 and a further 512 points on August 31.
Implied volatility in prices of equity options also increased
substantially, more than doubling its pre-crisis levels.
These events were preceded by the decision by Salomon
Brothers to close its bond arbitrage group in the spring and
summer of 1998. The departure of such a large bond trader
potentially left the market with less liquidity than it would have
9

This discussion draws heavily on Ervin’s conference presentation.

Introduction

had, because of both the liquidation of Salomon’s very large
positions in the months prior and the absence of a large player
whose trading otherwise would have contributed to market
liquidity. Vincent Reinhart of the Federal Reserve’s Board of
Governors and Litzenberger pointed to the Salomon decision as
creating the initial market stresses that escalated with the Russian
default and the emergence of LTCM’s financial problems.
The mechanisms that led to the subsequent fall of LTCM
highlight aspects of the nonlinearities and reinforcing
feedbacks cited in the conference’s discussion of a securitiesmarket-based financial system. LTCM had a strategy of
targeting the volatility of the Standard and Poor’s index as a
type of risk control mechanism, using it as a benchmark against

The mechanisms that led to the
subsequent fall of [Long-Term Capital
Management] highlight aspects of the
nonlinearities and reinforcing feedbacks
cited in the conference’s discussion of a
securities-market-based financial system.
which to assess the value-at-risk of its own positions. VaR
analysis is a widespread portfolio-management strategy that
calculates the maximum potential loss over a certain time
period, given a specified level of confidence. Historical
volatilities are used to form a VaR estimate, Litzenberger
explained. ARCH and GARCH methods (autoregressive
conditional heteroskedasticity and generalized autoregressive
conditional heteroskedasticity, respectively) are common tools
for obtaining a volatility estimate based on historical volatility
and covariance. The usual way of employing these estimation
techniques at the time was to consider historical volatilities,
not just volatilities during hypothetical crisis states.
The pressure on LTCM’s position caused by the liquidation
of Salomon’s bond arbitrage group was combined with
pressure from the widening of credit spreads following the
Russian default. As VaR models reflecting the spike in price
volatility indicated higher risk, market participants began to
liquidate their positions defensively. This reaction illustrates
the concept of reinforcing feedbacks: As volatility increased,
market participants reasoned that risk had also increased,
so they began to liquidate those positions, a step that in turn
led to further elevations in volatility and more decisions to
liquidate. The process also illustrates the importance of
linkages and nonlinearities in systemic events: Even though
Russian instruments were a small proportion of LTCM’s
overall portfolio, market participants began to question their

own rationale for holding other, non-Russian positions that
LTCM also held. Thus, they began liquidating those positions in
anticipation of liquidation spilling over into other markets, and
in this way a seemingly small disturbance propagated quickly.
As Litzenberger explained, in the 1997 run-up to LTCM’s
failure, the arbitrage market was marked by high liquidity
and low volatility. Under these conditions, to maintain a target
risk profile (for example, VaR) when volatility was low, traders
such as LTCM would leverage their positions. Recall that
LTCM maintained a strategy of targeting its risk taking on
the volatility of the Standard and Poor’s index; the fund’s
response to the situation in 1997 was essentially to add leverage
by returning a substantial portion of capital to its investors.
This strategy was consistent with attempting to maintain
profitability when trading opportunities were harder to find as
trades were mean-reverting faster. According to Litzenberger,
this would have been an entirely reasonable strategy if
conditions in 1997 had constituted a steady state. However,
the liquidation of Salomon’s bond arbitrage group and the
Russian default combined to disrupt this steady state and
cause a considerable rise in volatility. The subsequent apparent

increase in risk triggered widescale liquidations, as the assumptions underlying these positions came into serious doubt.
The resulting pressure on LTCM led it to send investors a letter
on September 2 asking for more capital. Just three weeks later,
the firm was taken over by its creditor banks to enable orderly
liquidation of the fund’s positions.
Ervin explained that, from the viewpoint of derivatives
trading desks, the events leading up to the collapse of LTCM
resembled a regime shift: At times, trading in U.S. dollar
interest rate swaps dried up completely; pricing for typical
bonds, such as investment-grade mortgages, widened to the
point that one could not get a price, or at least a real price.
“During this period, people simply didn’t have confidence
they understood what was going on. They weren’t sure they
understood the new rules of the game, who would survive,
and how they should play,” he said.

Beyond the breakdown of trading models based on
historical correlations, systemic events have a psychological
character, as all the rules seem to collapse and participants
enter into a state of high uncertainty about their counterparties. According to Governor Kohn, this points to a crucial
role for policymakers:
Heightened uncertainty is the key characteristic of
episodes of financial instability. The central bank may not
have any more information than market participants do.
In economic models, based on such uncertainties, it is the
central bank’s willingness to act in the face of uncertainty
that differentiates it from other market participants and
gives it a positive role to play during financial crises.
This role must be buttressed by a clearer understanding of
the fundamental dynamics underlying the securities-marketbased financial system; yet many obstacles to this ideal remain,
both for market participants seeking to insulate themselves
from the effects of crises and especially for regulators seeking
to prevent them. Among these obstacles are the difficulty in
simulating financial crises, the lack of historical episodes to
study, and—crucially for entities such as the Federal Reserve—
hindrances to the types of data sharing among market
participants and regulators that would allow central banks
to act with certainty during systemic crises.

Analytical Issues
In a bank-dominated financial system, Governor Kohn
observed, it is much easier to gather the information necessary
to regulate effectively against the possibility of systemic
disruptions. In such a context, the critical information would
come from fellow bank regulators with which the Federal
Reserve had been working and from banks the regulators had
been examining. However, in a more market-dominated
context, in which many financial institutions have a presence in
many cross-border business lines, obtaining the information
on counterparty exposure and risks necessary to develop
cogent analyses and to inform decision making in a possible
crisis requires widespread cooperation across disparate entities.
Moreover, in many instances, market participants may regard
this information as proprietary. Scant availability of data and
inadequate data sharing present challenges for regulators
and market participants alike.
Governor Kohn remarked that, as the prime source of
constraint on potential crisis-causing behavior, market discipline
through vigilance among private parties is always preferable to

FRBNY Economic Policy Review / November 2007

regulatory dictates. For market discipline to be effective, however,
counterparties must have a clear understanding of each other’s
risk profile. This often requires them to share proprietary
information, and confidentiality agreements between
counterparties may be necessary to ensure comfort.
He acknowledged that market participants may be wary
of sharing proprietary information. However, information
sharing can greatly increase the probability that credit will
continue to flow during systemic disruptions, resulting in a
lower probability of a sustained systemic disruption, a reduced
need for government intervention, and enhanced financial
stability without moral hazard.
Governor Kohn added that, in a market-based system,
sound risk management by all market participants is essential
to protect against the risk of a low-probability—or “tail”—
event causing a financial crisis. For example, the bringing
together of practitioners in risk management policy groups can

Sound risk management practices among
market participants rely heavily on
sophisticated analytical methods that
present challenges beyond limited data
availability and information sharing.
potentially lead to improved reporting of risk information to
counterparties and allow best practices to be transferred across
market participants with respect to valuation, exposure
measurement, limit setting, and internal checks and balances.
Indeed, a lesson drawn from the 1998 crisis by the President’s
Working Group on Financial Markets (1999) was how
weakness in risk management and counterparty credit
discipline enabled a firm to acquire large leveraged positions
of a size that could magnify the effects of negative events.
Governor Kohn described how financial regulators, through
supervision, can promote market discipline and sound risk
management. The regulatory capital framework proposed in
Basel II would: 1) require the largest internationally active
banking organizations to enhance measurement and
management of their credit and operational risks, 2) prescribe
a rigorous methodology for entities to assess overall capital
adequacy in relation to their risk profile, and 3) require entities
to disclose publicly information about their risk profile.
Sound risk management practices among market
participants rely heavily on sophisticated analytical methods
that present challenges beyond limited data availability
and information sharing. Discussions among economic

Introduction

researchers, financial market practitioners, and members of
the engineering and natural sciences fields pointed to the
considerable differences between the financial system and
other complex systems. Among these differences is the inability
to conduct or observe natural experiments on systemic crises
in the financial system because crisis occurrences are too
infrequent. Another difference is the role of human behavior in
the financial system and the nonlinearities and anticipatory
behavior it can introduce, a factor largely missing in studies
of complex systems in engineering or the physical sciences.
The presentations by financial market participants addressed
these issues in discussions of scenario analysis.
Scenario analysis, as Daula explained, is the primary tool
that market participants use to examine the risks posed by
systemic events. Aside from being what it implies, scenario
analysis was defined by Daula in economic terms: It starts with
a particular scenario about the economy and then defines a
general equilibrium, inferring the conditional expectations of
all the consequences of that scenario for markets around the
globe and their relative prices. He identified three ways of
specifying the scenario. The first is to look at historical
episodes. This approach has the advantage of being grounded
in an actual event; the drawback is that changes in market
structure since the chosen episode can lessen the predictive
power of the analysis. A second approach is to fashion a purely
hypothetical event. This has the advantage of allowing one to
match the scenario to the particular market structure at the
time; the obvious drawback is the difficulty knowing with
certainty whether the hypothetical event is at all likely or
whether the analysis performed accurately reflects how the
event would actually unfold. The third approach, which
addresses some of the pitfalls of the previous two, is to use a
hybrid, mixing in something that may have occurred in the past
in a slightly different context and analyzing how it may play
out in today’s context; conditional expectations for changes
in market structure are adjusted along the way.
It is difficult to choose the optimal scenario to analyze.
Daula suggested a method that pointed to some possible
interdisciplinary linkages between this type of financial
research and engineering approaches to systemic risk: Choose
a set of scenarios broad enough to span collectively the types
of market fluctuation likely to be encountered. If the scenarios
selected are sufficiently broad, common elements may emerge.
Daula emphasized, though, that this type of exercise may result
in unlikely scenarios. Providing an example, he noted that one
often-considered scenario is a monetary crisis in a reserve
currency such as the U.S. dollar, an event that arguably has not
occurred in thirty years. Incorporating extreme tail events such
as these would address Litzenberger’s concern that many

quantitative risk management approaches rely too heavily on
data from relatively benign periods and thus allow history to
grant a false sense of security.
The approach of collectively analyzing a broad range of
scenarios may allow for linkages with optimization
methodologies from engineering fields such as operations
research. Presenting one possible inroad, Haimes offered an
overview of the “partitioned multi-objective risk method,” in
which systems are analyzed both for the low-frequency but
high-cost events and for the high-frequency but low-cost
events. Drawing on such mathematical work from engineering
fields may also enable one to analyze how certain attempts to
increase a system’s resilience and robustness may actually

predispose the system to low-frequency but high-damage
events. Needless to say, any such analysis must be very careful
in its assumptions of probability distributions.10
Researchers and policymakers face many challenges in
arriving at a better understanding of systemic risk in our
evolving securities-market-dominated financial system.
Market participants and regulators face a dual problem:
They must determine the factors that can trigger contagion,
the prospect for sudden regime shifts, and the potential for
hysteresis; they must also craft policies that strengthen
resilience to the threat of systemic events in a way that neither
predisposes the system to even larger disruptions nor imposes
unjustifiable costs on market participants.

This topic is discussed in Goldenfeld and Kadanoff (1999).

FRBNY Economic Policy Review / November 2007

References

Bernanke, B., and M. Gertler. 1990. “Financial Fragility and Economic
Performance.” Quarterly Journal of Economics 105, no. 1
(February): 87-114.
Brimmer, A. F. 1989. “Distinguished Lecture on Economics in
Government: Central Banking and Systemic Risks in Capital
Markets.” Journal of Economic Perspectives 3, no. 2
(spring): 3-16.
Chan, N., M. Getmansky, S. M. Haas, and A. W. Lo. 2006. “Systemic Risk
and Hedge Funds.” In M. Carey and R. Stulz, eds., The Risks of
Financial Institutions. Chicago: University of Chicago Press.

Goldenfeld, N., and L. Kadanoff. 1999. “Simple Lessons from
Complexity.” Science 284, no. 5411 (April): 87-9.
Keynes, J. M. 1936. General Theory of Employment, Interest,
and Money. Harcourt, Brace and World/Harbinger (1964).
President’s Working Group on Financial Markets. 1999. “Hedge Funds,
Leverage, and the Lessons of Long-Term Capital Management.”
Washington, D.C.: U.S. Department of the Treasury, April.
Saidenberg, M. R., and P. E. Strahan. 1999. “Are Banks Still Important
for Financing Large Businesses?” Federal Reserve Bank of New York
Current Issues in Economics and Finance 5, no. 12 (August).

Gatev, E., T. Schuermann, and P. E. Strahan. 2006. “How Do Banks
Manage Liquidity Risk? Evidence from the Equity and Deposit
Markets in the Fall of 1998.” In M. Carey and R. Stulz, eds.,
The Risks of Financial Institutions, 105-30. Chicago:
University of Chicago Press.

The views expressed in this summary do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System.
14

Introduction

Part 2: Current Trends
in Economic Research
on Systemic Risk
A

conference session on current research directions featured
three papers examining market-based crises—crises in
which financial institutions are affected by shocks that
propagate through asset prices and market liquidity.1 In these
crisis models, shocks affect financial institutions through the
prices of securities that the institutions hold in common—not
through chains of connections between institutions, as in a
payments network.
While market-oriented models of financial crises differ
from the traditional bank-oriented models in the way
shocks are propagated, they share with bank models the
possibility of multiple equilibria and transitions driven by
positive feedback. Thus, a shock can cause a transition from
a normal state to a crisis state from which the system need
not recover endogenously. Indeed, the models often feature
path-dependent behavior in which the transition out of a
crisis state entails a path different from the one leading to
the crisis and may require some form of external intervention. These characteristics of market-based models—
and the dynamics of the models more generally—are the
subject of the three papers presented.

As the discussion that followed the presentations made
clear, the papers open some potentially productive new
avenues for research. More insight is needed into how
financial markets recover from crisis states and what
policies or regulatory regimes would speed that recovery
and contribute to a more robust financial system.2 A related
issue that merits further research is the trade-off in risk
management practices between the objective of limiting
risk ex ante and the effects of risk management constraints
in the midst of a crisis. For instance, mark-to-market
accounting is a risk management practice that makes trading
performance transparent and prevents managers and
traders from concealing losses while trying to gamble their
way out of losing positions.3 Further, marking to market the
value of trading positions, combined with risk management
loss limits that force a closeout of a losing position, can
prevent a loss from becoming large enough to bring down
a firm. (Some bank failures and catastrophic investment
fund losses are attributable to the failure to adhere to this
basic risk management discipline.) However, as the papers
presented suggest, the collective and mechanical exercise
of such discipline on a widespread scale after a large market
shock can create the type of liquidity spiral that leads to a
market crisis.

The large literature on systemic risk and financial crisis cannot be represented
in any set of three papers. The papers in this session of the conference were
selected to illustrate current thinking about financial crises that propagate
through securities markets (for example, the bond and stock markets). Further,
the conference organizers sought out analytical or theoretical papers that
would show the conceptual underpinning of the literature on financial crises;
empirical analyses of financial crises were not included.

For examples of research on these issues, see Allen and Gale (1994, 2005)
and Holmstrom and Tirole (1998).
3
Mark-to-market accounting requires that the value of an investment, which
might vary over the period for which it is held, be assigned the current market
price of such an investment.

The views expressed in this summary do not necessarily reflect the position
of the Federal Reserve Bank of New York or the Federal Reserve System.

FRBNY Economic Policy Review / November 2007

Wealth Transfers and Portfolio
Constraints
The first paper, by Anna Pavlova of the London Business
School and Roberto Rigobon of MIT (presented by Rigobon),
examined the transmission of shocks between countries with
cross-border trade and investment. Pavlova and Rigobon
(2006) began studying this issue after they uncovered a
divergence of views on a simple question: Would it be good for
the stock market in the United States if the dollar depreciated?
They found that the answer depended on whether the initial
shock was a supply or a demand shock and also on the effects
of wealth redistribution arising from the changes in the relative
prices of goods and financial assets. The presentation focused
on how a shock plays out in the real side of the economy and in
the financial system and how the two sectors interact through
the effects of wealth redistribution.
The paper highlights the ways in which financial market
imperfections and institutional features of the financial system
affect the transmission of shocks across countries. The model
presented has a center country and two peripheral countries;
significantly, it also includes a constraint on the center
country’s financial sector that can be interpreted as a risk
management constraint on that country’s investors—for
instance, a constraint against concentration risk. With this
model, Pavlova and Rigobon seek to understand how the
exchange rates, interest rates, and stock markets in the three

Would
Would it
it be
be good
good for
for the
the stock
stock market
market
in
the
United
States
if
the
dollar
in the United States if the dollar
depreciated?
depreciated?
countries evolve in response to shocks. Is there comovement in
asset prices of the peripheral countries and, if so, does it depend
on the tightness of the constraint? The analysis uses a general
equilibrium framework that illuminates the role of wealth
redistribution in the transmission of shocks.4
In the model, the constraint creates a common risk factor or
covariation in stock prices and terms of trade (the exchange
rate). In the presence of shocks, the portfolio constraint leads
to wealth transfers that create comovement among the terms of
trade and stock prices in the peripheral countries, while
reducing the comovement between the stock markets of the
center country and the peripheral countries. These results are
4

In a general equilibrium analysis, all decision makers behave optimally relative
to others (subject to constraints such as budget limitations), and supply and
demand in all markets are in balance at the equilibrium prices.

Current Trends in Economic Research on Systemic Risk

consistent with empirical findings documenting contagion
among the stock prices and exchange rates of countries
belonging to the same asset class (for example, emerging
markets). One of the model’s implications for policy is that
during a crisis, interventions that relax the portfolio constraint
in the center country’s financial system could be a more
effective response to a systemic crisis than providing assistance
to the country suffering the initial shock. The alleviation of the
constraint short-circuits the wealth transfers that transmit the
shock to others, reducing the likelihood of contagion.

Risk and Liquidity
in a System Context
Hyun Song Shin of Princeton University examined how
liquidity shocks can propagate through the linkages between
balance sheets of financial institutions and securities prices.
The starting point of Shin’s (2006) analysis is the fact that most
of the assets on the balance sheets of financial institutions are
claims against other parties. This fact leads to interesting and
possibly complex interrelationships in which asset prices can
fluctuate together. How creditworthy one party’s liabilities are
depends on the strength of the assets on its balance sheet, which
in turn depends on the creditworthiness of other parties’
liabilities, and so on.
In Shin’s analysis, the financial system is a system of interlinked balance sheets. An objective of the study is to analyze
fluctuations in apparent risk appetites that arise endogenously
from solvency constraints and financial institutions’ interlinked balance sheets. In the model, all assets are marked to
market, and economic agents are assumed to be risk neutral
so that the analyst can observe how asset prices respond to the
liquidity effects arising from market participants’ interlinked
balance sheets, rather than to changes in risk preferences or
risk aversion.
In the model, the market value of each firm’s debt depends
on the value of the firm’s assets. Since some of these assets are
the debt of other firms, linkages arise in the value of the debt
of all the firms. An equilibrium is a fixed point of these asset
value equations. With the addition to the model of a target
leverage ratio determined by, for instance, a risk management
constraint, financial institutions will shrink or expand their
balance sheets in response to shocks to their capital—actions
that will set off liquidity drains and lending booms. In this
model, supply and demand curves have counterintuitive
shapes, and a fall in prices can actually increase the supply of
assets. In such a case, a negative shock to bank capital raises a

bank’s leverage ratio above its target; to reduce leverage, the
bank must sell assets. These sales depress prices even more,
causing a further negative shock to all banks’ capital and
setting in motion additional asset sales and a downward spiral
in asset prices.
A policy-related implication of this analysis is the potential
for feedback effects to arise from mark-to-market accounting.
Now that a much wider range of assets can be marked to
market, will such an accounting convention enhance stability

A policy-related implication . . . is the
potential for feedback effects to arise
from mark-to-market accounting. Now
that a much wider range of assets can
be marked to market, will such an
accounting convention enhance stability
or undermine it?
or undermine it? Accounting is absolutely crucial for thinking
about incentive problems because gains and losses are
recognized on the balance sheet, and it is the unit of account
that drives decisions.
In thinking about systemic risk, Shin considers the
difference between domino effects and price effects. In domino
scenarios, shocks propagate between banks through the payments
system or through cascading defaults. Price effects, however, can
propagate shocks even when no balance sheet or payment linkages
exist. Further, price effects operate even in the absence of large
players. Price changes are a lightning rod that coordinates
expectations and actions and that affects the system through the
similarity of positions across firms regardless of firm size or the
lack of direct linkages between the firms.

Market Liquidity and Funding
Liquidity
In the session’s last paper, Markus Brunnermeier of Princeton
University (presenter) and Lasse Pedersen of New York University
explored the relationship between market liquidity and funding
liquidity, giving particular attention to how they interact through
risk management practices at financial institutions. Market
liquidity is the ease of trading an asset and is asset-specific, while
funding liquidity is the availability of funds and is agent- or
borrower-specific. Brunnermeier and Pedersen’s (2006) paper

links the two liquidity concepts by arguing that they are mutually
reinforcing: when funding liquidity is abundant, traders have the
resources to finance trading positions that smooth out price
shocks, and markets will be liquid. This process is self-reinforcing
because liquid markets are less volatile and assets become better
collateral—conditions that lead to a relaxation of funding
constraints on trading activity. This feedback loop is what
Brunnermeier and Pedersen set out to study.
They construct a model that would explain four stylized
facts about market liquidity. The first fact is the most
important one for the systemic risk question—the sudden loss
or fragility of liquidity. Second is the commonality of liquidity
and the way market liquidity comoves across different assets.
Third is the apparent correlation between liquidity and
volatility: whenever volatility is high, liquidity is low. The last is
the flight-to-quality phenomenon, whereby traders flock to
low-volatility securities when their capital is eroded, causing
the liquidity of riskier assets to deteriorate.
In the model, a market liquidity shock is defined as the price
deviation from the fair value of an asset. To examine
endogenous illiquidity effects, the researchers assume that
offsetting liquidity shocks exist: thus, in the initial period, a
liquidity shock causes the price to deviate from fair value and,
in the subsequent period, an offsetting shock occurs that
restores the price to its initial fair value.5 In addition to
liquidity shocks, a source of risk in the model is a fundamental
shock that changes the fair value of the asset. Traders in the
model buy and sell securities in an attempt to profit from the
liquidity shocks and, in so doing, provide liquidity to the
market. This liquidity provision is risky, however, because of
the fundamental shocks that change the fair value of the asset.
Traders are constrained by their net worth and need to finance
their trading positions subject to a margin or “haircut” on the
amount they can borrow, where the margin is a credit risk
mitigation device imposed by the lender and is determined by
the volatility of the fundamental value of the asset. The traders
face funding liquidity risk because a fall in their net worth or a
rise in the margin required for trading positions may deprive
them of funds needed for trading.
In this model, the relationship between the margin
requirement and the asset’s price and volatility will influence
whether equilibrium outcomes with fragile market liquidity
and illiquidity spirals occur. Trader losses from price shocks
can lead to self-perpetuating falls in market liquidity as trading
is endogenously curtailed because of the difficulty of funding
the margin required for trading positions.
5

Liquidity shocks are price shocks that are unrelated to fundamental value.
For example, an investor may sell bonds to meet a need for cash, placing downward pressure on the bond price; at a different moment, an investor who has
experienced a cash windfall may buy bonds, producing an opposite effect on
the price.

FRBNY Economic Policy Review / November 2007

Discussion
Herdlike Behavior and Incentives
for Contrarian Trading Strategies
The three papers presented in this conference session
highlighted the positive feedback effects that produce herdlike
behavior in markets, and the subsequent discussion focused
in part on means of encouraging heterogeneous investment
strategies to counter such behavior. Investors who sit on the
sidelines during boom times will not be weakened by the
inevitable downturn and will be well positioned to profit by
entering the market to buy assets at distressed prices. Such
contrarian investment behavior would mitigate the sort of

The three papers presented in this
conference session highlighted the positive
feedback effects that produce herdlike
behavior in markets, and the subsequent
discussion focused in part on means of
encouraging heterogeneous investment
strategies to counter such behavior.
systemic collapse that was analyzed in the papers presented.
A number of conference participants asked, what incentives for
this type of stabilizing behavior do fund managers have? Would
fund managers who were content to hold cash and low-yielding
liquid assets when the markets were flourishing be able to
convince their investors to stay with them when everyone else
was earning tremendous profits riding the upside of a bubble?
Which investors are willing to earn very little in anticipation of
realizing high returns by purchasing undervalued assets after a
market crash?
If it is costly to hold liquid assets in order to be a buyer and
to provide liquidity in a market crash, why would anyone
choose to do it? In an equilibrium analysis that accounts for the
incentives to sit on the sidelines in a boom, the market crash
must be big enough to assure liquidity providers that they will
earn sufficient profits buying at distressed prices to compensate
them for forgone profits. So, in the absence of government or
central bank intervention, the paradox is that the inducement
to adopt contrarian investment strategies is greater when the
severity of the crash is greater.6
6

Allen and Gale (1994, 2005) study these issues.

Current Trends in Economic Research on Systemic Risk

The conference participants discussed the role the central
bank or government might play in encouraging the sort of
contrarian behavior that would stabilize failing markets.
Collateralized lending by the central bank could be one way to
short-circuit the feedback in asset prices and distress-driven
selling of those assets; investors could acquire liquidity by
borrowing against assets instead of selling them.7 However, the
type of assets that investors might want to offer as collateral
could be different from the asset types normally used as
collateral when borrowing from the central bank—especially in
a situation in which investors’ best assets have already been
used in collateralized borrowing from the markets. Further,
there could also be incentive effects—such as moral hazard—
that change behavior in boom times in undesirable ways. If
investors anticipate that illiquidity would be mitigated in a
crash, they may have even more reason to ignore the risks in
an emerging price bubble.
Another policy option mentioned in the discussion in this
session would be to change reserve requirements and capital
requirements to counteract the positive feedback effects—that
is, to raise requirements in boom times and lower them in bad
times. Alternatively, when markets are prospering, banks could
be required to increase their liquid asset holdings so that they
can provide liquidity more effectively when markets fail. The
problem here, of course, is that these requirements act like a tax
on these institutions, and taxes are always unpopular and
would place the institutions at a disadvantage relative to other
market participants—at least in the good times.

The Range of Economic Models
in the Study of Systemic Risk
Participants in the session also discussed the types of models
used to study systemic risk and commented on the challenges
and trade-offs researchers face in developing their models. One
type of model is the falling domino model. When applied to
data on the linkages among banks through interbank loans and
exposures in the payments system, for example, the model is
used to study how cascading losses following the collapse of
a bank propagate through the banking or payments system.
In such an event, what would happen to other banks and how
would liquidity in the payments system be affected? Another
type of model takes into account the optimal behavior of
market participants in analyses of their response to shocks.
7

Examples of such liquidity provision are the discount window lending
facilities at central banks that provide emergency liquidity to banks, and the
repo options that the Federal Reserve made available to nonbanks to address
concerns about liquidity shocks associated with the Y2K vulnerability in
computing systems.

These models can be general equilibrium or game-theoretic
models: the former look at the interaction between financial
asset markets through, say, investors’ portfolio choices; the
latter examine strategic interaction between economic agents
in which agents act in anticipation of how others will behave.
In addition, the models can be either comparative static models
or dynamic models: the former analyze differences between the
pre-shock and post-shock equilibrium states of the financial
system, while the latter examine what occurs in the transition
from one equilibrium state to the other.
The work by Pavlova and Rigobon is representative of the
current literature on international crises involving exchange
rates and cross-border shocks to financial systems and
economic activity. The studies by Shin and by Brunnermeier
and Pedersen are illustrative of the models that look at feedback
effects to clarify the interactions between market prices and the
behavior of financial institutions. These papers highlight the
importance of the financial system’s institutional features—
mark-to-market accounting, margin requirements in trading,
and risk management constraints more generally—to an

Participants in the session also discussed
the types of models used to study
systemic risk and commented on the
challenges and trade-offs researchers
face in developing their models.

understanding of systemic risk. The papers are stripped-down
approaches examining the equilibrium of a system of price
determination equations to simplify the analysis of feedback
effects. Adding to the analysis a consideration of heterogeneity
among investment strategies, as in the discussion above,
increases the complexity of the effort considerably. For
instance, one could step back and ask how investors would
choose their initial portfolios if they anticipated the feedback
effects and linked sequences of events in possible future
scenarios. Or one could ask what incentives or compensation
arrangements would motivate an investor or fund manager to
act on that anticipation.
The challenge in these and other models is the trade-off
between analytical tractability and realism. Given the current

state of the art, significant simplification and abstraction are
required to build models that can be used to answer practical
questions. Yet the simplicity of a model by its nature means
that potentially important factors can be missed. Indeed, a key
goal of the conference was to determine whether there are
modeling techniques in other disciplines that can deal with
complexity yet still keep sight of the important features of the
system under study.

Adequacy of Buffers against Systemic
Shocks in the Financial System
A third discussion topic that drew considerable interest was
whether competitive pressures and risk management practices
are undermining the robustness of the financial system. More
sophisticated methods of assessing collateral and margin
requirements in the financing of trading positions may be
lowering the overall margin and collateral amounts held
against these exposures. For instance, the use of portfolio
margining allows the netting and offsetting of positions and
results in a lower margin on posted collateral. Certainly, the
technique has advantages: netting of margin across gaining and
losing positions in a portfolio can alleviate the liquidity shocks
from margin-driven selling of the losing position, reducing the
positive feedback effects analyzed above. At the same time,
however, portfolio margining reduces the amount of overall
margin, resulting in a smaller cushion if correlated shocks
occur simultaneously across the whole range of margined
investments.
A critical risk management issue here is the treatment of
correlation assumptions in determining margin amounts for a
portfolio of diverse assets. Correlations among asset prices can
change radically in a crisis. A conference participant observed
that truly sophisticated risk managers would set portfolio
margin requirements that take into account how those
correlations can change in a crisis, and not look myopically at
the average correlations of the last three years. Whether such an
approach would be rewarded, however, brings us back to the
earlier discussion of incentives and contrarian behavior: Do
risk managers have meaningful incentives to use conservative
portfolio margin requirements when their competitors are
basing their margins on optimistic assumptions about
correlations of margined positions?

FRBNY Economic Policy Review / November 2007

References

Allen, F., and D. Gale. 1994. “Limited Market Participation and
Volatility of Asset Prices.” American Economic Review 84,
no. 4 (September): 933-55.

Holmstrom, B., and J. Tirole. 1998. “Private and Public Supply of
Liquidity.” Journal of Political Economy 106, no. 1
(February): 1-40.

———. 2005. “From Cash-in-the-Market Pricing to Financial
Fragility.” Journal of the European Economic Association 3,
no. 2-3 (April-May): 535-46. Papers and Proceedings of the 19th
Annual Congress of the European Economic Association.

Pavlova, A., and R. Rigobon. 2006. “The Role of Portfolio Constraints
in the International Propagation of Shocks.” Unpublished paper,
MIT Sloan, December.

Brunnermeier, M. K., and L. H. Pedersen. 2006. “Market Liquidity and
Funding Liquidity.” Unpublished paper, Princeton University,
November.

Shin, H. S. 2006. “Risk and Liquidity in a System Context.” Bank for
International Settlements Working Paper no. 212, August.

The views expressed in this summary do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System.
22

Current Trends in Economic Research on Systemic Risk

Part 3: Systemic Risk
in Ecology and Engineering

everal fields of engineering and science share with
economics a keen concern with systemic risk. Systemic
risk is manifested in space shuttle accidents, airplane crashes,
the collapse of the New Orleans levees, electrical power
blackouts, and the failures of buildings, bridges, and many
other engineered systems. Because of these occasional system
failures, engineers have more relevant data for the study of
systemic risk than do economists. Using these data to conduct
retrospective analyses of system problems, engineers have been
able to identify and remove some sources of failure (for
example, in aircraft). Similarly, epidemiologists and public
health experts worry about disease outbreaks and spread,
which occasionally reach systemic levels, and they have learned
lessons in risk management by studying past epidemics.
And ecologists study changes in the state of ecosystems, which
may receive less press attention but clearly qualify as systemic
developments because they can result in a true regime shift
from one equilibrium to another.
There are two ways that one discipline can leverage the
experience of another. The first way is by adapting methodologies developed in one field to analyze structures and
phenomena in the other field. The examination of the Federal
Reserve’s Fedwire system in part 4 of this volume exemplifies
this mode of intellectual sharing: researchers adapt tools from
outside of economics—namely, network theory and graph
theory—to learn what insights can be gained by applying them
to a problem of systemic behavior in the area of payments. The
second way is by sharing insights that are particular to a given
field and that, by analogy, might apply to other fields. This is
the approach taken in this part of the volume.

Useful Concepts from Ecology
and Engineering
At the conference, ecologist Simon Levin of Princeton
University identified a range of concepts that have proved
helpful in understanding complex systems in ecology and that
might also apply to financial systems. One useful conceptual
model of an ecosystem is a “trophic web,” which represents
how species are interconnected. At a coarse level, a trophic web
in an ecosystem might be thought of as a set of predator-prey
relationships. In this case, sets of differential equations can be
successful in modeling the rise and fall of populations as the
ecosystem fluctuates around an equilibrium or becomes
unstable. More generally, however, “trophic” refers to the flow
of energy, so the trophic web for an ecosystem is a framework
for representing how the primary source of nutrition (say,
sunlight or geothermal vents) is transmitted between levels in
the food chain. This interpretation of the trophic web is more
applicable to financial systems, in which the interactions are
usually less extreme than those in predator-prey relationships;
we simply have to interpret “energy” as anything of value that
is transmitted through the system. Because of this analogy, it is
not surprising that we would find similar, if not identical,
phenomena in these two systems, and therefore similar insights
might be brought to bear in analyzing them. Complex systems
of any sort are characterized by nonlinearities, multiple stable
states, hysteresis, contagion, and synchrony, all of which have
relevance to the problem of systemic risk.
Nonlinear relationships are a key characteristic of virtually
any complex system. They can lead to multiple stable states,

The views expressed in this summary do not necessarily reflect the position
of the Federal Reserve Bank of New York or the Federal Reserve System.

FRBNY Economic Policy Review / November 2007

such that the system can exist in one configuration (basin
of attraction) for a period of time but then be knocked into
a different configuration by a perturbation or shock. This
transition can be accompanied by hysteresis, meaning that if
the system is to return to its original configuration, it must take
a different path. Often, pain and other costs are associated with
that recovery pathway.
Nonlinear feedbacks, which can be either positive or
negative, can drive a complex system away from a given
equilibrium state;1 the stability of any complex system is
determined by the nature of these feedbacks. Feedbacks can
result from the low-level processes in the system (for example,
the behaviors or individuals in a food chain, traders in a
market, or components of an engineered system), from an
explicit top-down control system, or from policies enforced

Complex systems of any sort are
characterized by nonlinearities, multiple
stable states, hysteresis, contagion, and
synchrony, all of which have relevance
to the problem of systemic risk.
by regulators. Positive feedbacks usually amplify the effect of
disturbances, thereby decreasing the stability of steady states.
In contrast, we usually think of negative feedbacks as
stabilizing. However, that is not always the case, as demonstrated by the suspension bridge over the Tacoma Narrows
known as “Galloping Gertie.” The bridge was subject to a
negative feedback (a damping) that overcompensated, with
the result that a certain wind condition led to escalating
oscillations and finally collapse.
Once a system is destabilized, it moves away from the linear
regime and can experience nonlinear behaviors such as path
dependence (meaning that the next state is dependent on the
sequence of events that led to it), sustained oscillations (such as
cyclicality in the financial sector), and regime shifts, by which a
system moves into an entirely different region of performance,
such as the less desirable equilibrium that characterized the
Great Depression. However, nonlinear behavior also means
that an effective remedy need not require a massive effort, just
a well-targeted one.
Another phenomenon common to complex systems is
contagion. In ecosystems, contagion is an important part of
ecological and epidemiological dynamics, as exemplified by the
mechanisms that spread forest fires and disease. In the financial
1

“State” is used here as a shorthand to mean either a single state or a set
of dynamically (possibly stochastically) related states in a common basin
of attraction, not something static.

Systemic Risk in Ecology and Engineering

sector, contagion manifests itself as cascading losses and
increased risk aversion, with the latter leading to herd behavior,
funding withdrawals, and a contraction of liquidity. Contagion
can be found in two forms in the electric power grid and other
complex networks such as road and communications systems.
A destabilizing form occurs when the failure of one node (for
example, a substation or a bridge) creates a buildup of load on
the rest of the system that in turn may lead to a cascade of other
failures. But when load switching and rebalancing can
effectively redistribute the load, contagion assumes a stabilizing
form: it spreads the stress and thereby reduces systemic risk.
Synchrony, another feature shared by some complex
systems, is evident when incentives or pressures lead individual
actors to fall into step and make similar choices. In nature, one
finds benign instances of this phenomenon: some species of
fireflies blink synchronously, and flocks of birds and schools
of fish can often turn almost as units. However, tight linkages
among individuals can also be a cause for concern because they
can induce systemic collapse. Conservation biologists have
shown considerable interest in the degree of synchrony in
species populations: In unsynchronized populations, some
individuals thrive while others are in decline; in synchronized
populations, a collapse in one place translates into a collapse
in all places. Like contagion, synchrony can lead to systemic
risk in the form of a system failure or a sudden jump to a less
desirable equilibrium.
Ecosystems, the financial system, and many other complex
systems are in fact complex adaptive systems, in which
collective behaviors emerge from individual actions. In
ecosystems, those collective behaviors include the flocking
of birds, herding of ruminants, and formation of fish schools.
In the world of finance, the Dow Jones Index reflects the
integrated effects of many individual decisions, making it an
emergent indicator. Many components of the financial system
pay attention to these emergent indicators, and what the
indicators imply about collective behaviors feeds back to affect
individual behavior, but on very different scales of organization
and time. Behavioral ecologists have developed some understanding of the principles of collective decision making among
animals.2
Complex adaptive systems consist of heterogeneous
collections of individual units that interact with one another
and thereby influence how the whole system evolves. Often
the phenomena that we are interested in are occurring on
different scales, and the systems essentially integrate
phenomena at multiple scales of space, time, and complexity.
The components of the electric power grid (transformers,
voltage regulators, generators, relay switches, and so forth),
for instance, are nonlinear and have different stochastic
2

See, for example, Couzin et al. (2005).

behaviors that might affect only a local neighborhood of the
grid, but they interact in ways that can lead to systemic shifts in
grid performance, or to failure. Moreover, the observed system
performance is actually the integrated result of the grid’s
behavior along with the behavior of layers of communication,
sensing, and control, the fuel supply, human behavior, and
the financial transactions that make it function. Clearly,
understanding and predicting the performance of a complex
adaptive system at that level is a major multiscale and
multidisciplinary endeavor.
The term “complex adaptive system” might leave the
impression that the system is adapting and adjusting itself to
beneficial effect. What it really means, however, is that some
components of the system are adapting and changing, not that
the system as a whole is changing in a coordinated way. The
adaptation might be in the influenza virus, and its ability to
become more effective is not necessarily good for the system
as a whole.
A critical attribute of complex adaptive systems that must be
properly modeled is path dependence. Imagine rolling a ball
down the side of a mountain range. Its path illustrates the
natural development of a system. The ball comes to certain
decision points where it enters one or another watershed. Once
it starts down one path, it is locked into that pathway unless a
major perturbation occurs. Thus, the future development of
the system is dependent on the path that has been taken—that
is, on the history of the system. If there is a major perturbation,
however, the system can jump into a new basin of attraction
that is conceptually and phenomenologically very different: the
system would move from one valley to another.3 This is a
regime shift, or system flip, which can be very disruptive. For
example, scientists studying ecological systems worry about
eutrophication, the over-enrichment of lakes. A system that
moves from a healthy oligotrophic lake to a eutrophic lake with
large quantities of algae is still a stable system, but the flip is
very detrimental for most of the species in the oligotrophic
lake. Analogously, a rich land can undergo desertification
and become a very different ecosystem.
On a larger scale, ocean circulation patterns can undergo
relatively sudden flips. Such flips have occurred in the past and
might be triggered again by climate change, but no one knows
the likelihood of their recurrence. A qualitative change in ocean
circulation patterns—one that altered the topology of the
flows—would have major impacts. It would be a regime shift,
a shift into a different domain of attraction. Economic markets
can go through crashes and recoveries that are also shifts in the
basins of attraction. Bank collapses can trigger chain reactions
that would represent the same type of shift as a phase transition
in physics.
3

In economic terms, each valley will have its own rates of saving, interest,
employment, productivity, and so forth.

As noted earlier, regime shifts can lead to hysteresis,
meaning that the behavior of the system in its recovery phase
may be quite different from its behavior in the destruction
phase. For example, in the ecological literature, there is
considerable interest in the spruce bugworm and other
defoliating insects that can completely denude forests of
spruce, balsam fir, and other species. After an outbreak of these
insects, the system recovers over time, but as the forest quality
increases, the bugworm population builds up enough to
re-emerge. Once this outbreak occurs, the quality of the forest
begins to decline until the system reaches a critical point and
collapses. Thus, the system goes through regular periods of
outbreak and collapse, each one representing what amounts to
a system shift. The fact that the pathway on the way down
differs from the pathway on the way up is a hysteresis effect.
Levin pointed out that, unlike systems designed for
robustness, complex adaptive systems are systems in which
whatever robustness exists has to emerge from the collective
properties of the individual units that make up the system;
there is no planner or manager whose decisions completely

Ecosystems, the financial system, and
many other complex systems are in fact
complex adaptive systems, in which
collective behaviors emerge from
individual actions.
control the system. Therefore, there are no guarantees that
things will work well. This leads us to the problem of the global
commons, in which we all engage in behaviors based on our
own agendas and interests; from these individual behaviors,
system properties emerge. For individual organisms, natural
selection encourages the development of robust physiological
properties. But an ecosystem, banking system, or economic
system has not been engineered for robustness.
Collapse in complex adaptive systems is the same as the loss
of robustness. If a system is working well, we think of it as
robust, whether it is an engineered system, a banking system, or
an ecosystem. In various literatures, the terms robustness,
resilience, rigidity, and resistance are often used to mean the
same thing, although they really describe different components
of the system’s capacity to function in the presence of internal
or external disturbances.
What leads to robustness in complex adaptive systems?
There are at least two ways in which a system can be robust in
the face of disturbances: by having a rigid design and reliable
components, or by having a flexible design that may also
include replaceable components. One can see these alternatives

FRBNY Economic Policy Review / November 2007

in a stressful marine environment with strong currents. Corals
resist the disturbances by being rigid, while kelp withstands the
disturbances by being flexible. These are two quite different
strategies for responding to the stress of strong currents, and we
see the same contrasting strategies in many other systems.
Rigidity—sticking with an existing design or decision (think of
the Polaroid company and its camera design)—might be the
best approach over short periods of time or if the environment
is relatively constant. But over longer periods of time or in
fluctuating environments, flexibility can prove a more robust
approach. In the camera industry, for example, Kodak has
continued to change its camera designs and products over the
years. Neither the Polaroid nor the Kodak strategy is “right”
per se, but each is right over a particular time horizon.
In changing environments, one needs flexibility, whether it
is in ecological systems or in banking systems. For example,
Levin noted that the flexibility of the influenza virus accounts
for its robustness. On the surface of the virus are proteins
called surface antigens, in particular haemagglutinin and
neuraminidase. The name of a flu strain—say, H5N1 flu—
refers to the particular forms of haemagglutinin and
neuraminidase associated with that strain, as those proteins
change over time. Once a person gets a particular strain of
influenza, he or she will never get it again. Individual variants
therefore are not very robust; they can be controlled or
eradicated by the human immune system if they return. But the
influenza virus itself has been around for centuries, maybe
millennia, so the virus seen more generally is very robust. It
survives because it is adaptive, continually changing its design
and its surface proteins.
Therefore, according to Levin, for a system to be robust it
must have diversity—analogous to the way the influenza virus
is really a family of viruses with variations in their surface
proteins—and it must have heterogeneity, so that there is scope
for adaptation in the system. For this reason, ecologists attach
great importance to biological diversity: even if they do not
know what particular species do, the presence of diversity
provides a form of insurance. When a system is too
homogeneous, it cannot adapt.
Modularity—the degree to which a system can be decoupled
into discrete components—also influences robustness. A basic
principle in the management of forest fires and epidemics is
that if systems are all connected, a perturbation will encounter
nothing to stop it from spreading. But when a system is
compartmentalized (when firebreaks exist or high-risk parts
of a population are vaccinated against an epidemic), then the
spread may be contained. Modularity can thus be an important
part of robustness if it ensures that an affected component will
be isolated from destabilizing feedbacks. However, modularity

Systemic Risk in Ecology and Engineering

often involves a trade-off between local and systemic risk.
Because the compartmentalized elements of a system will be
less able to withstand some shocks, modularity tends to
increase the risk that individual elements will be critically
damaged. Although the sacrifice of such elements is assumed to
decrease the risk of a calamitous systemic failure, the wrong

In changing environments, one needs
flexibility, whether it is in ecological
systems or in banking systems.
compartmentalization in financial markets could preclude
stabilizing feedbacks, such as mechanisms for replacing lost
liquidity, and so could actually increase systemic risk.
Robustness is not the same as stability, which refers to
the ability of a system to return to its equilibrium state.
It is interesting to note that ecologists have not been able to
agree on the relationship between biodiversity and stability.
In the 1950s, qualitative arguments led many to believe that
biodiversity and stability are positively correlated—for
instance, that biodiversity leads to robustness in some
macroscopic system properties such as nutrient cycling. But
theoretical arguments developed in the 1970s implied that as
system complexity or diversity increases, an equilibrium in the
relevant system of differential equations is less likely to be
asymptotically stable. Some argue that the instability of the
system dynamics (in the narrow sense of a stable equilibrium
of species densities) is what provides the adaptive capacity to
buffer the macroscopic properties: species replace one another,
or there are shifts in abundance, and these changes allow the
system to adapt to perturbations. Whether diversity increases
or decreases stability is an argument over the definition of
stability, and it is still being debated.4
The lesson that might be inferred is that understanding the
behavior of complex adaptive systems requires more than just
qualitative analysis and more than just theory. Ecologists have
applied alternative mathematical frameworks (for example,
interacting particle systems or systems of differential
equations), intensive simulations, data-driven analyses, and
even experiments in the effort to resolve this issue, and a similar
multifaceted effort might be needed to provide policymakers
with insights about the root sources of stability in financial
systems.

See National Research Council (2005, pp. 114-5) for a good discussion of this
debate. See also Levin (2000, chap. 7).

Methodologies for Prediction
and Management
In addition to providing useful concepts for the description
and analysis of systems in other disciplines, science and
engineering may provide some relevant methodologies for the
prediction and management of systemic risk. The rich scientific
literature on networks and graph theory, for example, may
have some bearing on the management of economic and
financial system risk. Networks influence the spread of
information, disease, and disturbances, and indeed the spread
of effects that can stabilize or destabilize a system. The topology
of the network is one of the key factors to study. For instance,
are there key nodes in the network whose removal would cause
the system to become decoupled? The potential for decoupling
might be seen as a vulnerability of the system because it could
impair the functioning of the network, but it can also suggest a
mode for limiting contagion in that it induces the modularity
that is important to robustness. Thus, to control the spread of
disease, scientists try to identify those who are super spreaders,
the individuals (say, prostitutes or hospital workers) who
connect different groups and make the system more likely
to exhibit undesirable synchronous effects. More generally,
researchers who study the topology of networks and the
relationship of that structure to network functionality will
consider how the properties of the network affect the spread
of money, disease, or information and propagate the spread of
disturbances that can cause systems to collapse.
Other scientific research relevant to the management of risk
is the literature on the modeling and control of forest fires, the
modeling and management of epidemics, and contagious
spread more broadly. The whole field of spatial stochastic
processes has focused largely on ecological and epidemiological
problems. As an example, Levin cited a National Institutes of
Health committee he recently chaired that oversaw several
agent-based simulations of the potential spread of pandemic
influenza in order to identify strategies for controlling that
spread. The models developed in this and other research efforts
are very computation-intensive. Levin indicated that
transferring the techniques from these models to the study
of financial systems would not be difficult, both because the
parallels were strong and because researchers in the financial
sector would be comfortable with the mathematical
techniques. The rich literature of epidemic theory, both
mathematical and computational, might then be applicable to
understanding runs on banks, as long as this approach was
properly augmented with knowledge of human behaviors that
contribute specifically to bank contagion. Levin suggested that
it might also be possible to transfer recent work on social
learning to the study of the financial sector.

George Sugihara of the University of California at San Diego
expanded on the possibility of rich analogies between
ecosystems and financial systems. Perfect parallelism is not
required if the goal is merely to stimulate fresh thinking that
generates productive hypotheses for research and even policy
formation related to financial systems, although empirical
corroboration of the analogy is, of course, one way to
strengthen its utility.
He pointed out that most ecosystems are innately robust
because they are survivors of extreme stress testing. Their
existence today sets them apart as the selected survivors of
many millions of years of upheaval and perturbation, having
withstood continental drift, meteor extinctions, climate
fluctuations, and the introduction or evolution of new
members. Those that survive show some remarkable constancy
in structure that may persist for hundreds of millions of years
(for example, the constancy of predator/prey ratios noted in
Baumbach, Knoll, and Sepkowski [2002]). Identifying the

The rich literature of epidemic theory,
both mathematical and computational,
might . . . be applicable to understanding
runs on banks, as long as this approach
was properly augmented with knowledge
of human behaviors that contribute
specifically to bank contagion.
common attributes of these diverse systems that have survived
rare systemic events could provide clues about which
characteristics of complex adaptive systems correlate with a
high degree of robustness. These attributes could then be
examined as candidate characteristics for lessening systemic
risk in other contexts, such as the financial sector. Because
experimental stress testing is not feasible in the financial sector,
examining such common structural properties of ecosystems
should be of interest, and it might help guide policy.
According to Sugihara, recent studies in nonlinear complex
systems show rapid and large transitions in state to be common
features of many “generic” interconnected dynamic (and
cybernetic) systems. Beyond the specific analogy between
ecology and economics, certain dynamical behaviors and
structural (topological) constraints are common to broad
classes of systems. Behaviors and network topologies that are
truly generic—as opposed to system-specific—can inform
many disciplines. For example, to understand the systemic risk
problem, it is useful to know the general properties of complex

FRBNY Economic Policy Review / November 2007

systems, particularly the structural ones that promote stability
or collapse.
As an example of scientific analysis that can readily be
applied to financial systems, Sugihara cited a recent paper in
Science (Bascompte, Jordano, and Olesen 2006) that examined
disassortative networks—networks in which nodes that are in
some sense “large” connect with many nodes that are “small,”
although the small nodes do not connect to many large nodes.
The paper, coming from the field of ecology, focused on the
network of pollinators and the plants that they pollinate, but
it also dealt more broadly with all networks that are positively
reinforcing. The paper showed that the disassortative nature of
the pollinator-plant network conveys a great deal of stability—
a result, Sugihara suggested, that generalizes to any type of
disassortative network, including the network linking U.S.
banks to the Fedwire system (see part 4 of this volume). In this
case, then, the theoretical analysis of a complex ecological
system highlights a characteristic of the financial system that
might be essential for stability and therefore worthy of
protection.

Risk Assessment of Extreme Events
Involving National Security
Yacov Haimes of the University of Virginia discussed his
work in modeling extreme events, especially those that affect
interconnected infrastructures and relate to national security.
It is generally impossible to build one single model to
represent any such complex system; there are too many
cross-cuts and too many ways to examine the processes and
effects of a complex system. The analysis of such a system
must instead be addressed from multiple perspectives,
perhaps hierarchically.
For his approach, Haimes has developed what he calls
“hierarchical holograph modeling” (HHM). This method is
hierarchical because it includes many different subtopics,
such as hardware, software, and organizational influences.
He emphasized that the last subtopic must be included in any
study of risk because many of the factors that contribute to risk,
or follow from extreme events, are organizational problems
and human problems. Risk analysis must consider such
matters as how well lines of communication function, how
much trust exists within a system, and who can share
information in a timely and effective way. And, of course,
the modern reliance on information technology means that
information assurance has also become critical. Haimes calls
his method “holographic” because it examines risk from many

Systemic Risk in Ecology and Engineering

different perspectives. For example, in a study conducted
for the President’s Commission on Critical Infrastructure
Protection, Haimes and his colleagues identified 300 major
sources of risk to the U.S. water supply. A good methodology is
necessary to structure an analysis encompassing that quantity
of information.
This approach to identifying and analyzing extreme events
in engineering differs from the approach often used in
modeling extreme events in economics and finance. The HHM

Many of the factors that contribute to
risk . . . are organizational problems and
human problems. Risk analysis must
consider such matters as how well lines
of communication function [and] how
much trust exists within a system.
method starts with an extreme outcome and provides a
methodology for exploring what factor or combination of
factors would produce that outcome. It is an inverse method
in that it works backward from an undesirable outcome to
infer what combinations of circumstances could give that result
and what the associated probabilities are. In contrast, systemic
risk analyses as conducted by financial economists or market
practitioners often project forward to infer the ramifications
of a hypothesized shock. The two approaches represent
different strategies for understanding what factors produce
extreme events.
In the study for the President’s Commission on Critical
Infrastructure Protection, Haimes and his colleagues used
HHM as the foundation of an adaptive multiplayer game.
Four teams, each with a very different perspective, were
assembled in 2005 to develop separate HHMs to learn about
the various sources of risk affecting Supervisory Control
and Data Acquisition (SCADA) systems. The red team
assumed the perspectives of attackers and hackers; the blue
team represented the perspectives of SCADA operators and
owners; a vendor team embodied the ideas of SCADA
developers and vendors; and a policymaker/stakeholder team
represented the interests of government and of industry
associations. About sixty experts participated in the four teams.
Interestingly, because of the teams’ differing perspectives, there
was less than 10 percent overlap in the specific risks identified.
For instance, several teams identified software and staff
training as key risks, but only the policymaker team identified
organizational decision making as a potential risk, and only
the operators/owners team identified the quality of electrical

infrastructure as a potential risk. This exercise underscores
the value of incorporating multiple views and perspectives in
efforts to identify sources of risks in complex systems. Team
approaches to generating input for risk analysis can be very
effective. The key to their success is the mechanism for
assimilating the information generated and for anchoring it to
concrete evidence. Uncertainty quantification plays a major
role in the degree of success of such efforts. The problem most
often encountered is that the results are not sufficiently
transparent to merit high confidence.
Another study of large-scale risk undertaken by Haimes and
his colleagues explored the regional and national economic
effects of an attack with a high-altitude electromagnetic pulse
(H-EMP).5 In an H-EMP attack, an enemy would use a nuclear
weapon to inflict systemic damage on the country’s electrical
and computing infrastructure. Specifically, an atomic bomb
would be exploded fifty kilometers above the United States,
and most of the damage would be to electronic systems, not
people or structures.
Using the inoperability input-output model (an adaptation
of Wassily Leontief ’s input-output model that puts more
emphasis on interdependencies), Haimes and his colleagues
estimated the percentages of dysfunctionality that would be
observed in 485 sectors of the regional economy as a result of
an H-EMP attack. These estimates were based on assumptions
about the impact of the H-EMP blast on the electrical and
computing infrastructure of each sector. As expected, the
predicted inoperability effects are not uniform across all
sectors, nor are the production losses, which would amount to
billions of dollars. By studying the heterogeneous effects of
such an event, Haimes explicitly avoids the spatial and sector
smoothing that is implicit in some analyses of risk, and draws
attention to the varied and localized nature of the economy’s
vulnerabilities. In this particular case, it was determined
that the major impacts sustained by some sectors would
nevertheless have a minor effect on the economy per se, and so
would not lead to systemic problems. This type of analysis
provides policymakers with valuable insights into priorities,
highlighting what resources should be protected first or most
securely. It can also help illuminate the trade-offs between
different recovery strategies, which can be striking.
Presenting another example of a complex analysis of
heterogeneous impacts, Haimes described his study of the
hypothetical economic impacts of a closure of the MonitorMerrimac and Hampton Roads bridge-tunnels in southeastern
Virginia. That area of Virginia contains a number of military
installations, including a major naval base. To understand the
5

This study was conducted for the Congressional Commission on H-EMP
Attacks on the United States.

economic effects, Haimes had to model the driving patterns of
many groups of workers and purchasers as they found alternate
routes, and the patterns emerging from those models
collectively created a picture of the overall system behavior.
If these tunnels were destroyed, it would take more than a year
to rebuild them, so they are very strategic for Virginia and for
national security more generally. This research provides the
foundation for choices that prepare us for extreme natural
hazards or terrorist attacks and for developing resilience in our
interdependent infrastructure and economic systems.
An analysis using an inoperability input-output model
revealed that the major sectors whose functioning would be
impaired by the closure of the bridge-tunnels would be
primary metal manufacturing and textile manufacturing.
All the other sectors would be minimally affected. As for the
overall economic loss, management services would be affected
most, followed by business services and retail trade, while

Team approaches to generating input for
risk analysis can be very effective. The key
to their success is the mechanism for
assimilating the information generated
and for anchoring it to concrete evidence.
Uncertainty quantification plays a major
role in the degree of success of such
efforts. The problem most often
encountered is that the results are not
sufficiently transparent to merit high
confidence.
the economic impact in many other sectors would be slight.
The analysis shows each sector from different perspectives,
producing a broader picture.
In all these risk analyses, Haimes and his colleagues assessed
the expected value of outcomes but supplemented that
assessment with other information because expected values
can be insufficient indicators of risk. Managers and decision
makers are often more concerned with the risk attaching to a
specific case than with the likelihood of an “average” adverse
outcome that may result from all similar risk situations. They
are also interested both in the low-frequency, high-damage
events—those with major, potentially regime-shifting
consequences—and in the more common risks, which
dominate the expected value.

FRBNY Economic Policy Review / November 2007

Haimes explained how he uses the partitioned
multiobjective risk method (PMRM)6 to measure and analyze
the risk of extreme and catastrophic events by partitioning the
probability into several sections, as shown in the following
equations:
β1

∫
(⋅) = E [X | X ≤ β ] =
∫
0

β1

p ( x)dx

Policy
option B

β1
β2
β1

∞

f 4 (⋅) = E [X | X > β 2

β1
∞
β2

∞

f 5 (⋅) =

∫
∫

xp( x)dx

∞

p( x)dx

(millions of
dollars)

Policy
option A

∫
]=
∫

Cost of risk
management

xp( x)dx

β2

f 3 (⋅) = E [X | β1 ≤ X ≤ β 2

The Trade-Off between the Cost of Risk Management
and Potential Losses

xp( x)dx
p ( x)dx

xp( x)dx

Policy
option C

f1( • ) versus f4 ( • )

Policy
option D

f1( • ) versus f5 ( • )
0

Loss of
capacity
100

(percent)

p( x)dx

∞

= ∫ xp( x)dx ⋅
0

associated cost for risk management and a corresponding loss
of functionality. For instance, option A consists of investing
significant resources in risk management in order to reduce the
likelihood of extreme events. The curve on the left shows the
expected value lost, while the curve on the right shows the
extreme loss. It is the more meaningful curve.

The probabilities displayed in these equations have the
following interpretations:
• f2(·) represents the risk with high probability of
exceedance7 and low damage, partitioned at β1 on
the damage axis.
• f3(·) represents the risk with median probability of
exceedance and medium damage, partitioned between
β1 and β2 on the damage axis.
• f4(·) represents the risk with low probability of
exceedance and high damage, partitioned between
β2 and ∞ on the damage axis.
• f5(·) represents the unconditional (conventional)
expected value.
The PMRM can be used to explore trade-offs between the
cost of risk management and the potential loss. The chart
presents a specific example in which the horizontal axis
represents a percentage of electric power capacity at risk and
the vertical axis, which is also f1(·), represents the cost of risk
management. Each of the policy options A through D has an
6

See Asbeck and Haimes (1984).
An exceedance probability (EP) curve specifies the probability that a certain
level of loss will be exceeded. If one views the loss as a random variable,
the EP is simply the complementary cumulative distribution of the loss.

Systemic Risk in Ecology and Engineering

Prediction and Management of
Systemic Failure in the Electric Grid
Massoud Amin of the University of Minnesota extended the
discussion of risk assessment, modeling, and prediction by
describing past and potential failures in the North American
electric power grid, another complex system. While this system
might not support multiple equilibria, as ecosystems and
financial systems can, it is certainly susceptible to nonlinear
amplification of instability, which leads to blackouts. The post
mortem analysis of major blackouts often shows the root cause
to be the failure of one or a few components (out of thousands
in the portion of the grid ultimately affected) that upsets an
equilibrium and leads to a cascade of failures. For example, on
August 10, 1996, North America experienced a major blackout
affecting more than 7 million customers in thirteen states or
provinces. It was later determined that the root cause was two
transmission faults in Oregon. Ultimately, that modest failure
led to power oscillations on the order of 500 megawatts,
overwhelming the system’s response mechanisms and leading
to the blackout.

Amin reported that some studies of the 1996 blackout
estimated that it could have been avoided if the grid had
intelligent controls and was able to reduce its load by
0.4 percent for thirty minutes. Such studies not only shed light
on how to prevent future failures, but also help to clarify what
recovery options exist if a similar failure does occur. Recovery
is an important part of risk management, and recovery options
can be identified by doing a scenario-based quantitative risk
assessment in advance. Of course, the technologies for
recognizing the incipient problem and tailoring a solution
are far from obvious.
Engineered systems such as the electric power grid or a
telecommunications network often include advanced control
systems that enable recovery. Amin reported on research
funded in the 1990s by the Electric Power Research Institute
(EPRI) that built on the technology used in control systems for
fighter planes. Because a power system includes substations
and generators that must operate at the same 60 hertz
frequency, controlling those elements in a coordinated fashion
is somewhat analogous to controlling planes that are flying
in formation. And responding to the loss of one or more
components is somewhat analogous to maintaining control
of an aircraft when a wing is damaged. Accordingly, EPRI’s
research was directed toward a control system that would have
some self-healing capability—a system, in other words, that
could anticipate disruptive events by detecting signals
indicating an important change, conduct a real-time
assessment of the changing state of the system, determine how
close the system is to some “edge” in performance, and remedy
or isolate the problem (isolation, sectionalization, and adaptive
islanding, which are discussed below). These same sorts of
capabilities would be desirable in a system designed to control
the financial system during disruptions.
Creating such a control capability for the electric grid
requires a mixture of tools from dynamical systems, statistical
physics, and information and communication science, as well
as research to reduce the computational complexity of the
algorithms so they can be scaled up to the large size of the
system being controlled.8 The electric grid poses a multiscale
challenge: troublesome signals must be detected within
milliseconds, with certain compensatory actions taken
automatically; some load balancing and frequency control on
the grid is handled on a timescale of seconds; and control
functions such as load forecasting and management or
generation scheduling take place on a timescale of hours or
days. Identifying at the atomic level what is amiss in a system
8

Working methods derived from the EPRI research program have been applied
in a variety of contexts, including the electricity infrastructure coupled with
telecommunications and the energy markets, cell phone networks on the
Internet, and some biological systems.

and then responding on a macro-scale requires multiresolution
modeling in both space and time.
To convey the complexity of modeling and controlling the
electric grid, Amin gave some basic facts. In North America,
there are more than 15,000 generators and 240,000 miles of
high-voltage lines. The overall grid is divided into several very
large interconnected regions, and modeling one of them
(which is necessary for understanding the systemic risks) might
entail a simulation with 50,000 lines and 3,000 generators. The
system is typically designed to withstand the loss of any one of
these elements. To determine whether the grid can attain that
design goal, we need to simulate the loss of each of the 53,000
elements and calculate the effects on each of the 50,000 lines,
leading to more than 2.6 billion cases. Although analysis of
these systemic risks is very challenging, the findings can help
researchers determine the best way to operate the system.
As an additional illustration of the level of detail that can be
successfully simulated, Amin presented a complex model that
predicts load and demand for DeKalb, Illinois, a sizable market

In any situation subject to rapid changes,
completely centralized control requires
multiple, high-data-rate, two-way
communication links, a powerful central
computing facility, and a sophisticated
operations control center. But all of these
features are vulnerable to disruption
precisely when they are most needed.

with a mixture of commercial and residential customers.
Deregulation of the electric system has reduced the correlation
between power flow and demand, thus introducing uncertainty
into the system, and so a number of researchers have sought
new ways to monitor and predict demand. The models and
algorithms are now sophisticated enough to simulate the
demand by customer type (residential, small commercial, large
commercial) on an hour-by-hour basis and attain 99.6 to
99.7 percent accuracy over the entire year. One benefit of these
predictions is that they enable power companies to dispatch
small generators to meet anticipated high demand.
More broadly, Amin argued that any critical national
infrastructure typically has many layers and many decisionmaking units and is vulnerable to various disturbances.
Effective, intelligent, and “distributed control” is required that
would enable parts of the constituent networks to remain

FRBNY Economic Policy Review / November 2007

operational or even to reconfigure automatically in the event
of local failures or threats of failure. In any situation subject
to rapid changes, completely centralized control requires
multiple, high-data-rate, two-way communication links,
a powerful central computing facility, and a sophisticated
operations control center. But all of these features are
vulnerable to disruption precisely when they are most needed
(that is, when the system is stressed by natural disasters,
purposeful attack, or unusually high demand).
When failures occur at various locations in such a network,
the whole system breaks into isolated “islands,” each of which
must then fend for itself. With the intelligence distributed, and
the components acting as independent agents, those in each
island have the ability to reorganize themselves and make
efficient use of the remaining local resources in order to
minimize the adverse impact on the overall network. Local
controllers will guide the isolated areas to operate independently while preparing them to rejoin the network, without
creating unacceptable local conditions either during or after
the transition. A network of local controllers can act as a
parallel, distributed computer, communicating via microwaves,
optical cables, or the power lines themselves, and limiting their
messages to only the information necessary to achieve global
optimization and facilitate recovery after failure.
If coordinated with the internal structure existing in a
complex infrastructure and with the physics specific to the
components they control, these agents promise to provide
effective local oversight and control without need of excessive
communications, supervision, or initial programming. Indeed,
they can be used even if human understanding of the complex
system in question is incomplete. These agents exist in every
local subsystem and perform programmed self-healing actions
that can avert a larger failure. Such simple agents are already
embedded in many systems today in the form of circuit
breakers and fuses as well as diagnostic routines. Echoing the
familiar tale of the kingdom that was lost for want of a
horseshoe nail, we might say that these agents are like the
missing nail: once restored, they can save an entire kingdom.
Another key insight relayed by Amin was drawn from the
analysis of forest fires. Researchers in one of the six EPRIfunded consortia found these fires to have “failure-cascade”
behavior similar to that of electric power grids. In a forest fire,
the transformation of a spark into a conflagration depends on
the proximity of the trees to one another. If just one tree in a
barren field is hit by lightning, it burns but no big blaze results.
But if there are many trees and they are close together, the
single lightning strike can result in a forest fire that burns until
it reaches a natural barrier such as a rocky ridge, river, or road.
If the barrier is narrow enough that a burning tree can fall
across it, or if it includes a burnable section, such as a wooden

Systemic Risk in Ecology and Engineering

bridge, the fire jumps the barrier and burns on. It is the role of
first-response wild-land firefighters such as smoke jumpers to
contain a small fire before it spreads by reinforcing an existing
barrier or scraping out a defensible fire line barrier around the
original blaze.
Similar outcomes can be observed for failures in electric
power grids. For power grids, the “one-tree” situation is one
in which every single electric socket has a dedicated wire
connecting it to a dedicated generator. A lightning strike on any
wire would take out that one circuit and no more. But the
efficient use of resources argues against such a system, and
instead favors one in which numerous sockets are served by a
single circuit and there are multiple circuits for each generator.
A failure anywhere in such a system causes additional failures
until a barrier—a surge protector or circuit breaker, say—is
reached. If the barrier does not function properly or is an
insufficient impediment, the failure bypasses it and continues
cascading across the system.
These findings suggest risk management approaches in
which the natural barriers in power grids may be made more
robust by simple design changes, or in which small failures
might be contained by active smoke-jumper-like controllers
before the failures grow into large problems. Other research

Echoing the familiar tale of the kingdom
that was lost for want of a horseshoe nail,
we might say that [independent] agents
are like the missing nail: once restored,
they can save an entire kingdom.
into the fundamental theory of complex interactive systems is
exploring methods of quickly detecting weak links and failures
within a system. Phased risk assessments have been very helpful
in this regard. That is, experience indicates the value of
performing “coarse-grained” risk assessments to identify
important contributors. Rather than considering fifty initiating
events for crisis scenarios, one might collapse them into five or
six key events, and then focus on what is most important.
According to Amin, work over the past nine years in this
area has led to a new vision for the integrated sensing,
communications, and control of the power grid. Some of
the pertinent issues are why and how to develop protection
and containment devices for centralized as opposed to
decentralized control and questions involving adaptive
operation and the resistance to various destabilizers. In
researching these issues, EPRI has refrained from conducting
“in vivo” societal tests, which can be disruptive, and has instead

performed extensive simulation testing (in silico) of devices
and policies in the context of the whole system. The EPRI
simulations have produced a greater understanding of how
policies, economic designs, and technology might fit into
the continental grid (while exposing some unintended
consequences of possible designs and policies), and provided
guidance on the effective deployment and operation of these
resources.
To mitigate the risk of systemic failure, the electric grid can
be engineered for robustness. Amin presented an example of
intelligent adaptive “islanding,” which is a method for blocking
contagion. His results were based on a simulation of a
hypothetical major blackout similar to the August 1996
blackout in the western United States. The simulation results
he displayed captured the steady decay in frequency from
60 hertz to less than 58 hertz, after which the system would
have deteriorated into a blackout. This simulation covered
3.5 seconds of simulated time. Then the simulation was re-run
with major power lines eliminated between Arizona and
Southern California to halt the contagion that led to the
simulated blackout. As a result, the Western Interconnect grid
was broken into two self-sustaining islands. Amin simulated
more than 12,000 cases to stress-test the islands, and found that
they consistently withstood the damaging contagion. With
intelligent islanding (isolation) shortly after a major system
disruption, the frequency recovered to close to 60 hertz before
a blackout could occur.
This example also illustrates the practice in some
engineering risk analysis of identifying undesirable outcomes
first and then developing the fault trees and associated
probabilities that could lead to those outcomes. The
engineering community extensively employs both inductive
reasoning (the event-tree thought process) and deductive
reasoning (the fault tree) in its risk assessments. The most
common approach is to use the event tree to structure the
scenarios and fault trees to quantify the split fractions of
the event-tree branch points.

Analogies in Economics and Finance
Vincent Reinhart of the Board of Governors of the Federal
Reserve System commented on three general forms of
nonlinearity that are important to systemic risk. First, he noted
that the consequences of events in the financial sector are likely
nonlinear. Therefore, in designing and enforcing laws and
regulations, the goal should not be to minimize the probability
of every adverse event, but to guard against those that have

more severe consequences: In other words, the risk
probabilities have to be weighted by some measure of the
welfare gain that would arise from the prevention of each
serious adverse event. That is the point of the partitioned
multiobjective risk method, which—as we saw earlier—is
designed to measure and analyze the risk of extreme and
catastrophic events.
In a second form of nonlinearity, some economic processes
are self-reinforcing. That is, in the run-up to a crisis, the size or
transmission of some events may be amplified. Margin calls

If intermediaries restrict the availability of
credit and therefore weaken spending,
that action becomes the “financial
accelerator.” These self-reinforcing effects
are similar to those that can occur in the
power grid when lightning strikes.

may cause selling that forces prices down more sharply, leading
to a “fire sale.” Concerns about collateral values or an uncertain
stock of capital may reduce arbitrage. If intermediaries restrict
the availability of credit and therefore weaken spending, that
action becomes the “financial accelerator.” 9 These selfreinforcing effects are similar to those that can occur in the
power grid when lightning strikes.
Trading activity can exhibit this second form of
nonlinearity. Consider a simple model in which two people go
to a market to trade. The amount of resources that one person
commits to trading depends on the amount that the other
person is expected to bring. This situation leads to collective
decision making, which can be a highly nonlinear process in
which small changes in cost bring about large changes in
overall market activity. Indeed, trading could dry up
altogether.
The third form of nonlinearity described by Reinhart
was the dependence of some economic processes on the
expectations of the players. This dependence can make
prediction very difficult and implies that there might be
multiple equilibria. How the market mechanism chooses
among these equilibrium outcomes may be unclear. As a result,
randomness and the sequence of events matter, suggesting that
the way policy decisions are communicated during the run-up
to a crisis can have an important influence on how the crisis
9

The term “financial accelerator” refers to how endogenous developments in
credit markets can amplify shocks in an economy. See Bernanke, Gertler,
and Gilchrist (1996).

FRBNY Economic Policy Review / November 2007

plays out. It also means that some techniques from the physical
sciences are not directly transferable to economic and financial
risk—the odds on a 100-year storm do not change because
people think that such a storm has become more likely.
Reinhart also noted that, in a simple economic model,
positive feedback can be destabilizing. But if one introduces
an asset that is priced in a forward-looking manner, positive
feedback is a mechanism for selecting a unique equilibrium.
In those same models, negative feedback introduces the
possibility of multiple equilibria—as was well known thirty
years ago.
Levin observed that, in contrast to management of the
electric power grid, there are only coarse or indirect options
for control of the financial system. The tools available to
policymakers—such as those used by central banks—are
designed to modify individual incentives and individual
behaviors in ways that will support the collective good. Such
top-down efforts to influence individual behaviors can often be
effective, but it is still difficult to control the spread of panic
behavior or to manage financial crises in an optimal way.
Within the financial system, robustness is something that
emerges; it cannot be engineered.
Levin also noted that the key determinants of robustness—
diversity and heterogeneity—are the same for biological,
engineered, and financial systems. The influenza virus is robust
because it takes on diverse forms; the analogue in the financial
sector is the variety of institutions and remediation
mechanisms, which makes the financial system more resistant
to large-scale failures. In both cases, the system is able to
adapt to change. But some redundancy—the ability of one
component to perform another’s function—is, of course, also
important in these systems. Otherwise, the chance loss of one
component could be catastrophic.

Discussion
Robert Oliver of the University of California at Berkeley noted
that both Haimes and Amin had an implicit taxonomy in their
risk analysis methodology: they first ran a risk assessment
and then explored risk management. Their talks gave some
guidelines for carrying out that linear process. However, those
talks did not illustrate how engineers also turn around risk
analyses to guide redesigns of system architectures and
topology and of the policies that are integral to system
performance. Since that process could be of value to central
bankers, Oliver asked for comments on how one might reach
new insights on those design and architectural questions.

Systemic Risk in Ecology and Engineering

Haimes suggested that a good way to proceed is to ask first
what can go wrong. Looking from many different perspectives
(as engineer, economist, social scientist, and so forth), one can
discover some things that have never been expected to go
wrong. To identify systemic risks, one has to look at everything.
Since no one can really capture all of the relevant perspectives,
systemic risks must be assessed through consultations with
multiple players, which ultimately converge on a picture of the
most important risks.
David Levermore of the University of Maryland pointed out
that large-scale, complex simulations as exemplified by the
work of Haimes and Amin are only part of the process of
analyzing systemic risk. In the physical and biological sciences,

To identify systemic risks, one has to look
at everything. Since no one can really
capture all of the relevant perspectives,
systemic risks must be assessed through
consultations with multiple players, which
ultimately converge on a picture of the
most important risks.
very tiny models, designed to build understanding, also play an
important role.10 These models are comparable in spirit to the
work described in part 2 of this volume, with one possible
distinction: in the physical and biological sciences, researchers
do not limit themselves to only those simple models that can be
solved analytically. The simple models might have only three or
four variables, or sometimes just one complicated nonlinear
variable, but still be complex enough to preclude analytic
solution. Thus, research in the physical and biological sciences
might rely more on computation than is the case in macroeconomics. Some of this research entails large-scale
computing, but one should also note that studies yielding
highly significant insights, such as the studies in dynamical
systems on the logistics map, have been undertaken on very
simple computers.
Douglas Gale of New York University observed that in
helping to identify speakers for the conference, he had looked
for those who would discuss the theoretical research being
done on financial stability. This emphasis may have given a
biased picture of current research in economics. In fact, Gale
noted, computational economics is a very large part of
10

See, for example, May (2004, pp. 790-3) and Keeling et al. (2003).
Both papers also illustrate the possible limitations of simple models.

economics, and economists typically make great use of data, a
point that was echoed by Reinhart. But an effort to model an
entire system, with the aim of learning how to control it better,
is a very large-scale project and one that academic economists
will not readily take on because of the way the profession is
organized and financed. They could follow such a path, but it
would require additional resources. Moreover, Gale expressed
some doubts about whether a large-scale computational
approach is the right way to look at a system. Instead, it might
be more fruitful to divide that system into understandable and
digestible pieces and then find ways of engineering the system
to ensure its robustness without a central control. Such an
undertaking would not require an ability to model the entire
system, still less an ability to control the full model.
Sugihara noted that the reliance on simple models,
abstracted from reality, can sometimes have misleading
consequences. For instance, the ideal gas laws, which are a
mainstay of the physical sciences, assume a certain kind of
functional form that often invites researchers to fit a scattering
of points to that form. But the reason for the scatter might be
quite important, and simplistic laws can lead researchers to

An effort to model an entire system, with
the aim of learning how to control it better,
is a very large-scale project and one that
academic economists will not readily take
on because of the way the profession
is organized and financed.
overlook it. In the study of fisheries, understanding the larger
systemic context of an individual species—the web as opposed
to the node—is very important. The presentation by Hyun
Song Shin of Princeton University, Sugihara noted, explicitly
addressed the web of claims and obligations. As researchers
and policymakers in finance and economics continue to
think in those larger terms, they are going to reach a fuller
understanding of the reality of the problem. Robert
Litzenberger of Azimuth Trust, however, pointed out the
value of abstraction in research, citing Milton Friedman’s
paper on positive economics, which assigns an important role
to assumptions and modeling. In Friedman’s view, assumption
allows the economist to abstract from the things that are
less important in order to focus on the key variables. The
economist’s model is not meant to offer realistic description,
which can fail to have predictive power. Simple models can
provide considerable insight and also produce very useful
predictions. The ultimate test of an assumption is its
predictive power.

Rather than choose between the extremes of simple and
complex models, several conference participants endorsed
the concept of nested hierarchical models. The collaboration
between Morten Bech of the Federal Reserve Bank of
New York, Walter Beyeler and Robert Glass of Sandia National
Laboratories, and Kimmo Soramäki of the European Central
Bank, described in part 4 of this volume, is a good example of
what could be accomplished in that direction. Pursuing the
notion of combining different types of models, Sugihara
suggested the following steps to build on the foundations
laid at the conference:
• Devise minimal (simple) models first to see how much
real variation in the data can be explained. Examples
might be Shin’s model of leverage, presented in part 2
of this volume, or agent-based models with simple sets
of rules. The latter would include models that can
reproduce certain statistical properties of aggregate price
series, such as the model proposed by Lux and Marchesi
(1999). The work in progress by Bech, Beyeler, Glass,
and Soramäki on an agent-based model for the Fedwire
payments network is a step in this direction. The
importance of empirical validation should not be
overlooked, and the meaning of the topological patterns
uncovered by Bech and his collaborators needs to be
understood. There is much to be learned from simple
models that can elucidate the systemic risk problem at
the most general level.
• Create more complex, mechanistic models to complement the simpler ones. This task aims for the ideal,
and it needs to be done carefully and in tandem with
the simple models. Nonlinearities in functional
relationships fix the scale of the model mechanisms
(aggregation problem) and can hinder the applicability
of those models across different market scales: firm,
industry, regional, national, and global. The difficulty of
developing complex models is exemplified by the early
efforts to develop ecosystem models. These models
appeared to be very complex because they incorporated
many variables. But the overall model behavior was
essentially simple logistic growth: much of the apparent
complexity did not add real insight. While the ecosystem
models provide a note of caution, it is nevertheless the
case that complex models can be built well.
Taking a broader view, Levermore noted that some
conference speakers seemed to focus on avoiding systemic
risk rather than managing the system. To evaluate risk
quantitatively (a first step toward avoiding it, if that is indeed
a realistic goal), one must be able to model the system to the
point where it can be plausibly simulated. An example was
Amin’s practice of testing various “islanding” schemes through
simulation. Once that level of simulation capability is achieved,
managing the system becomes easier. The primary benefit of

FRBNY Economic Policy Review / November 2007

this modeling and simulation capability, then, might not lie in
avoiding risk but in managing the economy more effectively.
For example, if the capability could help craft a regulatory tool
designed to manage risk, even if that tool could help the
economy run only a fraction of a percent more efficiently, the
benefit to society would be enormous, easily dwarfing any cost
in developing such a capability. If this capability also helped us
to avoid risk, it would be better still.
This discussion is not meant to imply that ecology and
engineering have overcome all the difficulties associated
with representing and analyzing complex adaptive systems.
Assessing the state of such systems is an ongoing challenge,

We do not know how to anticipate the
collapse of a system by looking at it and
recognizing that something is not right.
Are there ways to look at trends in the
stock market and know when a collapse is
coming? In the view of many observers,
complex systems produce signals that
will tell us when we are approaching
a precipice.
as is determining precisely what to measure. The validation of
models and verification of software remain major challenges.
Computational problems—including how to decouple models
into tractable components—are also a continuing source of
concern. Amin pointed out that self-similar systems can be
reduced, but complex systems such as the electric grid cannot.
Researchers can use approximations to decouple complex
systems, but it is difficult to analyze the errors thus introduced.
In this regard, Amin noted that if one can find parts of an
engineered system—and presumably parts of other systems—
that are weakly coupled in terms of the dynamics transferred
through the system, then one can approximate those portions
with standalone models. Such a strategy essentially reduces the
complexity by dividing and conquering. These component
models might assume a variety of forms: some might be
empirical models fit to data, others might be physics-based or
financial, and still others might include elements—such as
human behavior and performance—that cannot be modeled.
Haimes observed that, as an alternative strategy, one can
decouple the system at the lower level, model the lower level
components or subgrids, and then impose a higher level
coordination. In some cases, this can even be done with an

Systemic Risk in Ecology and Engineering

additional level of hierarchy. This type of decomposition is
a very effective way of addressing complex systems. In either
case, aggregating (composing) the outputs of these component
models into an overall picture is very challenging. To model the
electric grid, for example, researchers have parametrized some
of the component models so as to provide input to the next
level of modeling, using Bayesian analysis. Sensitivity analysis
is used to validate the resulting models.
Amin emphasized the difficulty of identifying meaningful
signals from complex systems. For example, when monitoring
a large fraction of the U.S. electric grid, how can we discern
whether a perturbation in the system is a natural fluctuation or
a sign of catastrophic failure? Is it a naturally caused
phenomenon, perhaps triggered by heat, high humidity, or
strong demand in one portion of the grid, or is it actually an
attack on the system or the precursor to a major disturbance?
How close is it to a regime shift or system flip? These questions
can be addressed only with detection systems that can call up all
the data and perform data mining, pattern recognition, and
statistical analysis to derive the probability that a catastrophic
failure is either developing or occurring now.
This system-monitoring problem is exacerbated if the
sharing of information is limited, as it is in the banking sector.
Charles Taylor of the Risk Management Association asked
Amin how one would monitor and control the reliability of the
electric grid under the assumption that companies did not
cooperate with each other but instead competed and did not
share information. Amin said that such a situation would lead
to a new control mechanism, and the logical question would be
whether that mechanism would stabilize or destabilize the
system. He pointed to a project undertaken by the Electric
Power Research Institute in the late 1990s—the Simulator for
Electric Power Industry Agents—which addressed such a case.
The analysis, applied to four large regions of the United States,
explored whether one could increase efficiency without
diminishing reliability.11 This preliminary analysis would need
to be carried out with more data and realism in order to reach
a definitive conclusion.
Levin identified particular challenges facing those who wish
to understand systemic risk more fully. For instance, we would
like to be able to develop structure-function relationships—
meaning that one could take a snapshot of a system and infer
something about its dynamic state. We do not know how to
anticipate the collapse of a system by looking at it and
recognizing that something is not right. Are there ways to look
at trends in the stock market and know when a collapse is
coming? In the view of many observers, complex systems
produce signals that will tell us when we are approaching a
11

See Amin (2002).

precipice. But the unfolding of market disruptions is affected
greatly by confidence, herding, and other behaviors that are not
mirrored in risk assessments for complex engineered systems.
Other questions include, How do we overcome the robustness
of undesirable configurations, so as to make it easier to move
out of them? How can we get systems out of potentially
problematic settings, and how can we achieve desirable
cooperative arrangements?
The tools are available to develop agent-based models
of banking systems—models in which one builds in rules
for the behavior of individual people or institutions. These
models help us understand how individual behaviors become
synchronized or integrated with one another and how they
spread through the financial sector. Of course, there are many
unknowns about these rules, and the gamesmanship and
proactive moves probably figure more importantly in the
financial sector than in ecology or engineered systems. This
is just one set of tools, but there are others. Sugihara has
developed an approach to nonlinear forecasting. John Doyle

of the California Institute of Technology and Jean Carlson of
the University of California at Santa Barbara have done work
on highly optimized tolerance in which they use a genetic
algorithm to evolve the properties of systems. They consider
a variety of systems with particular structures and feedback
properties, expose them to perturbations, observe their
recovery, and then—in the same way that one might “train”
a chess-playing program—modify these systems until they
become more tolerant of the disturbances to which they are
exposed. Doyle and Carlson’s strategy offers a way to improve
the structure of systems when the mathematics cannot be
solved. Nevertheless, as the authors themselves point out, their
approach does have a drawback: Systems that are engineered or
have evolved to be tolerant of a particular set of disturbances
often do so at the expense of their response to other classes of
disturbances. Such systems are at once robust and fragile—an
outcome that policymakers and researchers might wish to
guard against as they seek better ways to manage risk and avert
systemic failures.12

See, for example, Zhou, Carlson, and Doyle (2002) and Carlson and Doyle
(2002).

FRBNY Economic Policy Review / November 2007

References

Amin, M. 2002. “Restructuring the Electric Enterprise: Simulating the
Evolution of the Electric Power Industry with Intelligent Adaptive
Agents.” In A. Faruqui and K. Eakin, eds., Electricity Pricing
in Transition, 27-50. Boston: Kluwer Academic Publishers.

Couzin, I. D., J. Krause, N. R. Franks, and S. A. Levin. 2005. “Effective
Leadership and Decision-Making in Animal Groups on the Move.”
Nature 433, no. 7025 (February 3): 513-6.

Asbeck, E., and Y. Y. Haimes. 1984. “The Partitioned Multiobjective
Risk Method.” Large Scale Systems 6, no. 1: 13-38.

Keeling, M. J., M. E. J. Woolhouse, R. M. May, G. Davies, and
B. T. Grenfell. 2003. “Modelling Vaccination Strategies against
Foot-and-Mouth Disease.” Nature 421, no. 6919 (January 9):
136-42.

Bascompte, J., P. Jordano, and J. M. Olesen. 2006. “Asymmetric
Coevolutionary Networks Facilitate Biodiversity Maintenance.”
Science 312, no. 5772 (April 21): 431-3.

Levin, S. 2000. Fragile Dominion. New York: Perseus.

Baumbach, R., A. Knoll, and J. Sepkowski, Jr. 2002. “Anatomical and
Ecological Constraints on Phanerzoic Animal Diversity in the
Marine Realm.” Proceedings of the National Academy
of Sciences 99: 6854-9.
Bernanke, B., M. Gertler, and S. Gilchrist. 1996. “The Financial
Accelerator and the Flight to Quality.” Review of Economics
and Statistics 78, no. 1 (February): 1-15.
Carlson, J. M., and J. Doyle. 2002. “Complexity and Robustness.”
Proceedings of the National Academy of Sciences 99,
suppl. 1: 2538-45.

Lux, T., and M. Marchesi. 1999. “Scaling and Criticality in a Stochastic
Multi-Agent Model of a Financial Market.” Nature 397, no. 6719
(February 11): 498-500.
May, R. M. 2004. “Uses and Abuses of Mathematics in Biology.”
Science 303, no. 5659 (February 6): 790-3.
National Research Council. 2005. Mathematics and 21st Century
Biology. Washington, D.C.: National Academies Press.
Zhou, T., J. M. Carlson, and J. Doyle. 2002. “Mutation, Specialization,
and Hypersensitivity in Highly Optimized Tolerance.”
Proceedings of the National Academy of Sciences 99:
2049-54.

The views expressed in this summary do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System.
40

Systemic Risk in Ecology and Engineering

Part 4: The Payments System
and the Market for
Interbank Funds
T

he two papers in a session on systemic issues in the federal
funds market examined the interbank payments system
and the market for borrowing funds used to settle interbank
payments. Both analyses were based on data on payments made
through the Federal Reserve’s Fedwire system.
Fedwire—the nation’s primary interbank payments
system—is a real-time gross settlement system through which
payments are settled individually and with finality. More than
9,500 participants use Fedwire to send and receive time-critical
and/or large-value payments on behalf of corporate, financial,
and individual clients as well as to settle positions with other
participants stemming from payments in other systems, securities
settlement, and interbank loans. In the first half of 2006, an
average of 525,000 Fedwire payments were made each day, with
the daily value of funds transferred averaging $2.1 trillion.
Complementing the Fedwire payments system is the U.S.
federal funds market, where banks borrow funds to settle the
payments they make over Fedwire. Fed funds are the bank
reserves on deposit at the Federal Reserve used to settle
payments between banks. The fed funds market plays a critical
role in allocating liquidity in the financial system as well as in
supporting banks’ ability to finalize the settlement of interbank
payment obligations.
The first paper, by Adam Ashcraft of the Federal Reserve
Bank of New York and Darrell Duffie of Stanford University,
explored whether trading frictions in the fed funds market
affect the reallocation of excess reserves to banks requiring
funds to complete their payments. Ashcraft and Duffie found
that fed funds trading is driven partially by individual banks’
precautionary targeting of balances, and that this targeting

contributes to systemic stability. The banks’ trading of funds
mitigates the risk of overconcentration of reserves in some
banks, and contributes to the liquidity of the fed funds market
and keeps price volatility relatively low. The second paper,
by Morten Bech of the New York Fed, Walter Beyeler and
Robert Glass of Sandia National Laboratories, and Kimmo
Soramäki of the European Central Bank (who did not
present), analyzed the network structure of interbank
payments and developed a model for a payments system
within such a network. The paper concluded that a liquidity
market allows a payments system to achieve a specified level
of performance with much less liquidity than would
otherwise be required—a finding that sheds light on the
trade-offs between liquidity within the payments system
and friction within the liquidity market. The combination
of network topology and bank behavior within a model,
the study also found, is critical for analysis of systemic risk
in payments systems. According to the authors, it is not
sufficient just to understand the topology of the network;
knowledge of the processes operating in that topology,
such as bank behavior, is equally important.

Systemic Dynamics in the Federal
Funds Market
Ashcraft and Duffie analyzed how allocational frictions affect
trading in the federal funds market. They also considered

The views expressed in this summary do not necessarily reflect the position
of the Federal Reserve Bank of New York or the Federal Reserve System.

FRBNY Economic Policy Review / November 2007

whether these frictions could lead to systemic risk in the form
of “gridlock,” in which individual financial institutions fail to
transfer balances quickly to counterparties as they wait for the
counterparties to transfer balances to them. Gridlock creates a
self-fulfilling slowdown in the efficient reallocation of excess
balances.
Like every over-the-counter market, the fed funds market
is subject to allocational frictions because trading is executed
through bilateral negotiation.1 These frictions can be any
sources of transaction costs or delays in identifying suitable
counterparties, negotiating trades, or executing trades, and

Like every over-the-counter market, the
fed funds market is subject to allocational
frictions because trading is executed
through bilateral negotiation.
they can impact market efficiency. Existing theories of trading
dynamics in over-the-counter markets have focused on
“search” frictions, whereby traders locate each other with
delays, to some extent by trial and error, and negotiate prices
that depend in part on the difficulty of finding suitable
alternative counterparties. Prices also reflect the relative
benefits of making a trade immediately rather than later
(and with a newly found counterparty) for each of the two
counterparties. As frictions increase and the matching of
suitable pairs of counterparties becomes more difficult, a trader
with an urgent need to transact has relatively less “leverage”
during a bilateral negotiation, and this condition is reflected in
the contracted price. The efficiency of allocation and the effect
of search frictions on pricing are among the main concerns of
the various theoretical studies. Yet there is little empirical work
on these aspects of the microstructure of over-the-counter
markets.
Ashcraft and Duffie broaden this relatively small body of
empirical research as well as add a unique dimension to it.
Their analysis of allocational frictions uses transaction-level
data from Fedwire, and the majority of their work focuses on
the top 100 institutions by payment, or “send,” volume for
business days in 2005.2 This data set permits the construction
of real-time balances for each institution and allows for the
tracking of the sender and receiver of payments and loans for
1

In the fed funds market, brokers reduce but do not eliminate the allocational
frictions.
2
The researchers protected the confidentiality of the institutions by removing
firm-specific details and conducting all data-related work within the Federal
Reserve Bank of New York.

The Payments System and the Market for Interbank Funds

every minute of the day. The authors identified a particular
send as a loan, as opposed to another form of payment, by
analyzing the terms of payments in the reverse direction
between the same two counterparties on the next business
day. (Fed funds loans are for overnight repayment.)
Ashcraft and Duffie documented evidence of federal funds
trading being driven partially by individual banks’ precautionary targeting of balances. Banks are motivated to end each day
with non-negative balances in their reserve accounts at the
Federal Reserve because overnight overdrafts are not permitted
except in special circumstances.3 These institutions are also
motivated to end each day with relatively small balances, in
part because the Federal Reserve does not pay interest on
overnight balances and in part because the institutions have
other ways to meet their reserve requirements over their twoweek maintenance periods, such as by holding currency in large
ATM networks and by sweeping funds in reservable accounts
into nonreservable accounts.
By targeting its balances, a bank contributes to systemic
stability. When its balances are larger than they typically are
at a particular time of the day, a bank has an incentive to trade,
and especially to lend, so as to reduce the balances. Ashcraft
and Duffie showed that, empirically, banks indeed act consistently in response to this incentive. Likewise, when balances
are low, banks trade (in particular, borrow) on average so
as to raise them. This self-interested balance targeting at the
individual bank level promotes systemic stability.4 It mitigates
the risk of overconcentration of reserves in some banks and
underallocation in others. Balance targeting reduces the risk of
gridlock, and plays a role in keeping the federal funds market
liquid and funds rate volatility relatively low.
Drawing on their discussions with fed funds traders,
Ashcraft and Duffie reported that fed funds trading is
significantly more sensitive to reserve balances in the last hour
of the day. For example, at some large banks, fed funds traders
responsible for targeting a small, non-negative end-of-day
balance ask other profit centers of their institutions to avoid
large, unscheduled payments, such as settlement of currency
trades, near the end of the day. Once a fed funds trader has a
reasonable estimate of the extent of current and yet-to-beexecuted send and receive transactions, he can adjust pricing
and trading negotiations with other banks so as to push his
bank’s balances in the desired direction. Ashcraft and Duffie
uncovered empirical evidence of this behavior; furthermore,
3

The Federal Reserve’s discount window is available, but at terms that make it
preferable to achieve non-negative balances through fed funds trading with
other banks before the end of the day.
4
One may think in terms of the usual “eigenvalue” or “mean-reversion”
conditions for dynamic stability of a multivariate dynamic system. In this case,
the coordinate processes of the system are the current balances of each bank
in the system.

they found that such behavior is more pronounced following
increases in intraday rate volatility.
In addition, the authors raised the issue of, but did not
resolve, whether precautionary balance targeting by banks in
the fed funds market, coupled with a regime in which banks
forecast the targeting policies of other banks, could have
systemically destabilizing consequences. A potential systemic
problem could arise, for example, if several large institutions
during a day of extreme misallocation of reserves individually
“hoarded” reserves, given the heightened risk of other banks
doing the same or the institutions’ forecasts that other banks
are incapable of releasing excess reserves quickly to the market.
For instance, Ashcraft and Duffie reported traders’ accounts
of rumors that this type of behavior was initially feared on
September 11, 2001, with the communications disruptions that
day affecting the Bank of New York, a large clearing bank.5
Any such gridlock was in the end averted by energetic liquidity
provision by the Federal Reserve.
Without significant liquidity provision by a central bank
during such an event, “a run on reserves” could stress the
ability of the fixed intraday supply of reserves to be sufficiently
reallocated quickly to meet requirements. (The total amount
of reserves in the system is relatively small compared with
the total daily volume of transactions.) Even in an extreme
scenario, however, access to the discount window as well as
infusions of liquidity by the Federal Reserve and other central
banks would mitigate adverse systemic effects, as they did on
September 11.6

as to observe how those measures change during disruptions.
Bech et al. next presented a model with simple agents
interacting within a payments system network; the model
exhibited a transition from independent to highly interdependent behavior as liquidity was reduced. When the
authors applied the model to a liquidity market, they
demonstrated that a reduction in market frictions can reduce
this interdependence and thus the likelihood and size of
congestive liquidity cascades.

Network Topology of Interbank
Payment Flows
A payments system can be viewed as a specific type of complex
network.7 In recent years, many fields of science have sought
to characterize the structure of networked systems and the
relationship between network topologies and stability,
resiliency, and efficiency. From a communications or
information-technology perspective, Fedwire is a “star”
network, in which all participants are linked to a central hub:
the Federal Reserve. Because of its ability to wire funds,
Fedwire is a complete network, as all nodes (participants) can
communicate (send and receive payments) with all others.

From a communications or informationtechnology perspective, Fedwire is a
“star” network, in which all participants
are linked to a central hub: the Federal
Reserve.

Network-Based Modeling
of Systemic Risk in the Interbank
Payments System
The cross-disciplinary team of Bech, Beyeler, and Glass began
by discussing two new approaches for characterizing the
nonlocal, systemwide interconnections that may lead to
systemic risk, and then combined the approaches to analyze the
Fedwire interbank payments system. Their research first
examined the global structure of interconnections in Fedwire
by representing bilateral interbank relationships as a network.
This approach allows one to quantify the overall pattern of
interaction and interdependencies using well-defined
measures applied to other complex networks, as well
5

McAndrews and Potter (2002) describe how the disruption of the regular
timing of incoming payments made a bank’s liquidity management more
difficult, and for some banks the “increased uncertainty (regarding which
payments they might receive later in the day) led them to have higher
precautionary demand for liquid balances.”
6
For example, see McAndrews and Potter (2002) and Lacker (2004).

However, the actual behavior of participants, the flow of
liquidity in the system, and thereby the contagion channel for
financial disturbances are not captured by these network
representations.
The graph of actual payment flows over the Fedwire system
includes more than 6,600 nodes and more than 70,000 links,
but a subgraph, or core, of just 66 nodes and 181 links accounts
for 75 percent of the value transferred. A prominent feature
of Fedwire is that 25 nodes form a densely connected subgraph
to which all the remaining nodes connect. By itself, this core
is almost a complete graph, and this small number of banks
and the links between them process the large majority of all
payments sent over the network. Soramäki et al. (2006) show
that the network shares many characteristics with other
empirical complex networks. These characteristics include
7

See, for example, Newman (2003).

FRBNY Economic Policy Review / November 2007

a scale-free degree distribution, a high clustering coefficient,
and the “small-world” phenomenon. Apart from the core,
the network, like many other technological networks, is
disassortative. That is, large banks tend to connect to small
banks and vice versa.
Bech et al. showed that the topology of the network was
altered significantly by the attack on the World Trade Center
on September 11, 2001. First, the massive damage to property
and communications systems in Lower Manhattan made it
more difficult—and in some cases impossible—for certain
banks to execute payments to one another, as some nodes were
removed from the system or had their capacity reduced.
Second, the failure of some banks to make payments disrupted
the coordination upon which banks rely when they use
incoming payments to fund their own transfers to other banks.

A Payments System Model
In a paper by Bech et al. (2006), the authors use an elementary,
agent-based model in the spirit of models applied to understanding self-organized criticality.8 Physicists have used these
models to study cascading phenomena in a variety of systems
(for instance, Jensen [1998]), where models made of very
simple agents, interacting with neighboring agents, can
yield surprising insights about system-level behavior. In the
Bech et al. model, interbank payments occur only along the
links of a scale-free network based on the authors’ analysis
of Fedwire data; the model thus shows that only a very small
fraction of the possible interbank exchanges tend to be active.
Bank customers randomly instruct the institutions to make
payments, and banks are reflexively cooperative: they submit
payments if the balance in their payments system account
allows them to; otherwise, they queue the instruction for
settlement later.
If a bank receiving a payment has instructions in its queue,
the payment enables it to submit a payment in turn. If the bank
that receives the payment is also queuing instructions, it can
make a payment, and so on. In this way, a single initial payment
can cause many payments to be released from the queues of the
downstream receiving banks. This is an example of the cascade
processes typically studied in other models of self-organized
criticality.
In the Bech et al. model, a single parameter—overall
liquidity—controls the degree of interdependence of interbank payments. Abundant liquidity allows banks to operate
independently; reduced liquidity increases the likelihood that
a bank will exhaust its balance and begin queuing payments.
8

For example, see Bak, Tang, and Wiesenfeld (1987).

The Payments System and the Market for Interbank Funds

When liquidity is low, a bank’s ability to process payments
becomes coupled with the ability of other banks to process and
send payments. In this instance, the output of the payments
system as a whole is no longer determined by the overall
input of payment instructions but instead is dominated by
the internal dynamics of the system, and the correlation
between arriving instructions and submitted payments
degrades as liquidity is reduced. The model does not exhibit
a phase transition between completely noncongested and
completely congested states. The distribution of congestive
events, or liquidity cascades, is close to a geometric distribution, and it has an inherent length scale independent of
network size. This result differs from the finding obtained
in systems that exhibit self-organized criticality (such as in
Jensen [1998]); the result is directly attributable to differences in the underlying relaxation process that set the
simple payments system apart from systems that display
self-organized criticality.

The Market for Interbank Payment Funds
Going beyond the study of the payments system in isolation,
Bech et al. combined it with a simple model of a liquidity
market. Liquidity market transactions were represented as
a diffusive process, where liquidity flows are not confined to
the links of the payments network. In the model, each bank is
directly connected to a central node representing the market,
and this connection is characterized by a conductance

An understanding of how participants
interact and react when faced with
operational adversity will assist payments
system operators and regulators in
designing countermeasures, devising
policies, and providing emergency
assistance, if necessary.
parameter that reflects the cost or friction associated with
market transactions. The market creates a separate network
of liquidity flows that, having a star topology, operates parallel
to the network of payment flows.
The inclusion of a liquidity market weakened the coupling
between banks and reduced the size of settlement cascades.
Bech et al. identified trade-off functions that relate different
levels of system performance, in terms of cascade size and

queue size, to the value of initial system funding and market
conductance needed to achieve that performance. The liquidity
market is very effective in reducing cascade size and queue size.
For a given level of performance, the rate of liquidity flow
through the market relative to the rate of flow through the
payments system was surprisingly small. The performance of
the system can be greatly improved by the market even though
less than 2 percent of system through-put flows in the market.
A liquidity market allows a system to achieve a specified level
of performance with much less total liquidity than a system
without a market would require. Conversely, the performance
of such a system is highly dependent on the operation of the
market. Disruptions to the market would greatly increase
congestion and cascades unless they were mitigated, for
example, through the addition of liquidity.
According to the authors, if the combined payments and
liquidity market system is modeled with simple processes, the
boundary between noncongested and congested states can be
explained in terms of the relative values of three time constants:
a liquidity depletion time, which is governed by the initial
liquidity in the system; a net position return time, which
increases with total deposits in the system; and a liquidity
redistribution time through the market, which is associated
with the market conductance parameter. An understanding of
this boundary has significant practical applications because this
understanding allows for direct consideration of the trade-offs
between liquidity within the payments system and friction
within the liquidity market, both of which are modified by
central bank policies. While the Bech et al. model does not yet
include the behavioral feedbacks that likely factor into the
decision making of banks, those feedbacks can be included in
order to consider how the congestion boundary may move.
A goal of future studies will be to introduce into models this
complexity as well as variability in the size of bank payments.
Wide-scale disruptions may not only present operational
challenges for participants in a payments system, but they may
also induce participants to change the way they conduct
business, with the potential to either mitigate or exacerbate
adverse effects (Bech and Garratt 2006). An understanding
of how participants interact and react when faced with
operational adversity will assist payments system operators and
regulators in designing countermeasures, devising policies,
and providing emergency assistance, if necessary. Accordingly,
as Bech et al. argue, it is not sufficient just to understand the
topology of the network; knowledge of the processes operating
in that topology is as important.

Discussion
Robustness Issues
Much of the discussion following the presentations centered
on questions concerning the robustness of the fed funds market
and the interbank payments system. In particular, how much
shock can the system tolerate—and would a shock move
the system to a new, less desirable equilibrium? Moreover, if
there are multiple equilibria, which are stable and which are

Although the models presented in this
session suggested the resiliency of the
payments system when combined with
a liquidity market for interbank payment
funds, the models were not complete
enough to assess these issues definitively.

unstable? Although the models presented in this session
suggested the resiliency of the payments system when
combined with a liquidity market for interbank payment
funds, the models were not complete enough to assess these
issues definitively.
The models do not have multiple equilibria because they do
not yet include the anticipatory behavior that could produce
feedback effects leading to system gridlock. In the combined
payments system and liquidity market model, congestive
liquidity cascades within the payments system are mitigated
through transfers within the fed funds market. Each bank is
trying to mean-revert toward a target level of reserves, and the
more quickly they adjust toward the target, or the lower the
friction, the faster the system moves to a stable and stationary
equilibrium. If expectations and anticipatory behavior are
introduced, such behavior could produce a model with
explosive or unstable equilibria—the “gridlock scenario” that
so alarms payments system operators. While the model has not
been extended in that direction, it is conceptually feasible to
model the self-sustaining volatility that would prevent banks
from reaching a stable equilibrium. Doing so empirically,
however, would pose a challenge because the rarity of systemic
crises severely limits the historical data on the behavior being
modeled.

FRBNY Economic Policy Review / November 2007

Another challenge in modeling empirically the fed funds
market and the payments system is to account for the central
bank’s deep involvement in the system as well as its ability to
take corrective action to stabilize the system.9 Because of this
stabilizing role, no data exist on the extreme states of the system
where instability could be found. Those states never occur,
or are extremely rare, because of stabilizing intervention. This
consideration also complicates the formulation of the anticipatory behavior discussed above: Participants understand the
stabilizing role of the central bank and may come to rely on
it when forming their expectations—a point we consider later
in this summary.

Large-Scale Simulation Models
and Policy Decisions
Some of the discussion focused on what lessons might be
learned from simulation models and how the integration of
models of different scales might inform policy decisions. The
models presented analyzed liquidity in terms of the liquidity

At what level of policy can simulation
exercises be feasibly aimed—at a high
level to inform policymakers of trade-offs
and general principles, or at a more
granular level to emphasize specific steps
to take in a crisis?
properties of the network, such as the transition between
congested and noncongested states, and in terms of the
functioning of the marketplace for liquidity redistribution.
These models allow for simulation of the network’s responsiveness to liquidity injections by the central bank. What is
striking about them is their sensitivity to these injections.
A useful endeavor, suggested during the discussion, would
be to construct simulations that shed light on the amount
of liquidity needed by the system in different types of stress
scenarios. That is to say, how much liquidity should be
provided in response to different physical disruptions to the
network, and to which institutions should it be provided?
9

As an analogy, consider the electrical power distribution system: What if the
government instantaneously opened a new power line that was not there before
whenever a failure occurred in the grid?

The Payments System and the Market for Interbank Funds

For instance, while we tend to focus on the largest institutions—
perhaps on the assumption that they are the most connected—
a medium-size network participant might also be important
because of the high degree of its connectedness in the network.
Another issue raised was whether the simulation models’
degree of resolution would be comparable to the type of
information policymakers may actually have in a crisis. For
example, is it practical to build simulation models with a high
level of resolution? And at what level of policy can simulation
exercises be feasibly aimed—at a high level to inform policymakers of trade-offs and general principles, or at a more
granular level to emphasize specific steps to take in a crisis?
With regard to these questions, conference participants
would like to see closer integration between simulation models
at different scales—between the small-scale behavior of
individual decision makers and the large-scale aggregate
behavior of the system. Accordingly, a hierarchy of models with
different scales is needed. The general challenge here, common
to multiscale modeling in most domains, is building realistic
links between the small-scale behavior at the level of individual
agents and the large-scale aggregate behavior of the system.
A particular challenge in this application is likely to be the
range of behaviors and expectations at the micro level of
individual decision makers that can be feasibly modeled in
simulations. Thus, to what degree can anticipatory behavior
and expectations of economic agents be realistically
represented in simulation models, especially in simulations
of systemic crises that are not modeling “business as usual”?
While the issue has both conceptual and empirical dimensions,
the empirical basis for this behavioral component is apt to be
limited because systemic crises occur so infrequently.

Behavioral and Mutable Aspects
of Network Connections
John O’Brien of the Haas School of Business at Berkeley
commented on the application of the web-and-node model to
the events of 1987. As O’Brien explained it, the nodes represent
the individual investment banks and brokers, and they were all
under pressure to be more conservative. Therefore, each node
was trying to eliminate its assets and reduce its leverage. The
system overall would be harmed by that activity, however,
because as each node protected itself it would put pressure on
the others. At the system level, the Federal Reserve was trying
to make credit more readily available, but that change could
not be pushed down to the node level because the top priority
of the brokers in the investment banks was to meet margin

calls and reduce their leverage. Since the situation in 1987
seems to fit the web-and-node model, O’Brien observed, what
does the model suggest about policy if a similar situation were
to develop? Robert Litzenberger of Azimuth Trust explained
that the answer depends on knowing whether the shock will

Risk analysis of financial systems is more
dependent on human behavior than is risk
analysis of engineered systems.

dissipate or be self-reinforcing. Because many events will cause
only a mild market decline, we need to learn what types of
conditions will exacerbate problems. One relevant insight is
that economists sometimes partition the market into informed
traders and liquidity traders. If the shock is affecting mostly
liquidity traders, policymakers might decide to act, Litzenberger
said. However, if some of the selling is attributable to informed
traders, policymakers might instead opt to refrain from taking
action.
Douglas Gale of New York University added that in 1987,
some investors could not get in on the other side of the market
because dealers essentially would not answer their phones.
This problem illustrates one of the important differences
between contagion in a complex system that includes rigid
links between nodes (such as the electric grid), where the
collapse of a node has to be triggered by an event that hits it
in a physical sense, and the banking system, where links are
somewhat fluid depending on the perceptions and expectations
of the nodes. When the links are choices made by people
anticipating contagion, the links might start breaking before
the contagion reaches them. Gale’s comment points to the
central importance of human expectations and decisions in
the operation of financial markets. Risk analysis of financial
systems is more dependent on human behavior than is risk
analysis of engineered systems.

Evolution of the Payments System
The session concluded with a discussion of whether the
payments system could evolve into a less stable configuration.
Participants considered constraints that could be placed on the
evolution of the payments system to maintain its stability in the
face of ongoing changes in the financial system.

An important issue raised in this context was moral hazard.
Whenever the central bank intervenes to mitigate or reduce the
effects of a systemic shock, it can influence the future risktaking incentives of private sector decision makers by
weakening the perceived need to plan for the occurrence of
extreme shocks. This moral hazard issue can potentially
influence the state of the system and its vulnerability to shocks
through the risks taken by investors and financial institutions.
For researchers, this issue could be a factor when choosing the
behavioral features of their models. Consideration of such
behavioral issues, more generally, can add to the richness
of the results obtained from simulation models.
One notable issue that might be illuminated by the models
is the current tendency of payments in the system to migrate
toward the end of the day. This shift in payments timing would
appear to raise the likelihood of a congested state, as system
participants seem to hold back payments until late in the day.
Conference discussants noted that the Federal Reserve is
looking at how payments system policies affect use of the
liquidity pool. For example, which policies might be inducing
participants to shift their payments to later in the day, and
which might reverse that practice so participants can use the
available liquidity in the system more efficiently? Two changes
that might have played a role in the shift in payments timing
in recent years are the reduction in reserve balances, as banks
make more efficient use of reserves, and a higher demand on
the pool of liquidity attributable to increases in the volume
of obligations to be settled.
Another issue raised was whether the financial system is
moving toward increased homogeneity and greater
connectivity. Greater connectivity seems to be self-evident,
but in some ways we are actually seeing an increase in
heterogeneity. A shift has occurred from a bank—oriented
financial system to a securities-market–oriented system in
which a more diverse population of financial institutions
interact. The diversity of investors in today’s financial system—
such as pension funds, insurance companies, mutual funds,
private equity and hedge funds, investment banks, and
commercial banks—makes for a more heterogeneous set of
financial system participants than in the traditional bankoriented models of the financial system. In this sense, the
system has become more robust. That said, conference
participants observed that a handful of very large financial
firms play central roles in the financial system, and we need to
look carefully at ways to increase the robustness of the systems
and utilities that tie these firms together.

FRBNY Economic Policy Review / November 2007

References

Bak, P., C. Tang, and K. Wiesenfeld. 1987. “Self-Organized Criticality:
An Explanation of 1/f Noise.” Physical Review Letters 59, no. 4
(July): 381-4.

Lacker, J. M. 2004. “Payment System Disruptions and the Federal
Reserve following September 11, 2001.” Journal of Monetary
Economics 51, no. 5 (July): 935-65.

Bech, M., and R. Garratt. 2006. “Illiquidity in the Interbank Payment
System following Wide-Scale Disruptions.” Federal Reserve Bank
of New York Staff Reports, no. 239, March.

McAndrews, J. J., and S. M. Potter. 2002. “Liquidity Effects of the
Events of September 11, 2001.” Federal Reserve Bank of New York
Economic Policy Review 8, no. 2 (November): 59-79.

Beyeler, W., R. Glass, M. Bech, and K. Soramäki. 2006. “Congestion and
Cascades in Payment Systems.” Federal Reserve Bank of New York
Staff Reports, no. 259, September.

Newman, M. E. J. 2003. “The Structure and Function of Complex
Networks.” SIAM Review 45, no. 2: 167-256.

Jensen, H. J. 1998. Self-Organized Criticality: Emergent
Complex Behavior in Physical and Biological Systems.
Cambridge: Cambridge University Press.

Soramäki, K., M. L. Bech, J. Arnold, R. J. Glass, and W. E. Beyeler. 2006.
“The Topology of Interbank Payment Flows.” Federal Reserve
Bank of New York Staff Reports, no. 243, March.

The views expressed in this summary do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System.
50

The Payments System and the Market for Interbank Funds

Part 5: Concluding
Observations

omplex systems abound, and many different disciplines
are concerned with understanding catastrophic change
in such systems. People who study atmospheric science are
very interested in precipitous climate change, people in
ecology look extensively at so-called regime shifts and
precipitous ecological change, engineers design complex
systems so as to lessen the risk of catastrophic failures.
What opportunities exist to leverage this great interest from
across many fields for the benefit of the central banks and
financial authorities, the financial sector, and the nation’s
economy more generally? The conference explored this
question by focusing on three principal issues associated
with catastrophic events in complex systems: risk assessment,
modeling and prediction, and mitigation.

Risk Assessment
The economists, central bankers, market practitioners, and
scientists and engineers at the conference agreed in large part
on key mechanisms that produce instability in large systems.
Positive feedback—such as the portfolio insurance and
collateral and margin calls that may have played a role in
driving the stock market down so dramatically in October
1987—is one such mechanism. Another, synchrony, was
mentioned by Simon Levin of Princeton University as possible
in any complex adaptive system, sometimes with deleterious
consequences, and several conference participants pointed
to the increase in systemic vulnerability that can come about

when behaviors of various actors become too similar. Charles
Taylor of the Risk Management Association amplified this
idea in describing how banks’ decision making has changed:
A number of years ago, while there was a high level of
homogeneity in the mix of business taken on by banks, their
quantitative methods were less precise and more ad hoc—
with some variation in the speed of their responses to events.
The result was that individual banks would differ in how they
executed processes and how quickly they responded to changes
in conditions. Thus, there would be heterogeneity of response
to crisis. But now, as the banking system has become more
integrated and the time lags have been driven out by efficiency
measures, in Taylor’s view the system may be evolving in
a direction that makes it more fragile in some respects.
One area in which the approaches of financial economists
and market practitioners differ from those of engineers
such as Yacov Haimes of the University of Virginia and
Massoud Amin of the University of Minnesota is in identifying
extreme events. The conference background paper1 and the
keynote remarks of Governor Kohn discussed how potential
extreme events are identified through stress testing. This
procedure involves developing a model of an economic or
market process, applying extreme values from the distribution
of the drivers of the model, and examining the output. Those
who commented on stress testing acknowledged that a
limitation of this approach is its assumption that behavior
in the model does not change dramatically under extreme
conditions. This assumption conflicts with what market
participants in part 1 of this volume vividly described as the
1

The background paper can be found in Appendix B.

The views expressed in this summary do not necessarily reflect the position
of the Federal Reserve Bank of New York or the Federal Reserve System.

FRBNY Economic Policy Review / November 2007

feeling of regime shift during the events of 1997-98: the Asian
currency crisis, the Russian default, and the Long-Term Capital
Management collapse.
Part 3 of this volume explains the approaches followed by
Haimes and Amin for identifying possible extreme events—for
instance, a shutdown of the electric grid—and considering
what set of circumstances could produce the failure. Haimes
described a systematic process using small models and
arranging factors in a hierarchy that probes what failures,
mechanisms, and regime shifts in what combination might
lead to catastrophic failure. This paradigm of identifying a
range of possible bad outcomes (risks) and backtracking to
estimate their probabilities and identify options for reducing
their likelihood or lessening their impact is a common one in
engineering. It is in contrast to the paradigm in which a given
set of conditions is stipulated and then one explores, by means
of theory or simulation, how events might unfold in response
to a given stimulus. Taylor referred to the former paradigm
as “looking through the wrong end of the telescope.”
While Haimes’s process inevitably involves intuition and
judgment, the data-rich environment in which his methods
are applied grounds his modeling sufficiently so that one can
draw meaningful inferences, even if they are not susceptible
to classical statistical tests. For example, this method can be
used to refine estimates of unconditional and conditional
probabilities and correlations as well as the measurement of

One area in which the approaches
of financial economists and market
practitioners differ from those
of engineers . . . is in identifying
extreme events.

impacts. These estimates allow the analyst to make informed
judgments about factors that could trigger systemic collapse.
The stacking, if not necessarily nesting, of models in tiers also
allows the analyst to assess how behavioral changes during a
regime shift affect the potential for catastrophic failure.
Central banks over the last two decades have increasingly
devoted resources to research and analysis of financial stability.
A major purpose of these efforts is to identify potential triggers
of instability: events as well as market, policy, or institutional
mechanisms that can generate instability or propagate it once
the financial system is disrupted. The methodology used to
manage risk in engineering may provide insight into means
of identifying areas of potential financial instability more

Concluding Observations

systematically. Central banks may have an interest in
evaluating these methodologies.

Modeling, Prediction,
and Management
The conference generated lively discussion of differences in
the approach to research in economics, as illustrated in part 2
of this volume, and the research carried out in ecology and
engineering, as glimpsed in part 3. Economists were impressed
by the quantity and quality of data available to researchers in
the examples cited by Levin and by Haimes and Amin.

Research Culture and Directions
Douglas Gale of New York University suggested that the
conference brought out “a very striking contrast” between
some excellent theoretical research in economics and the
pragmatic, holistic modeling of risk in engineered systems.
The theoretical research was by young economists who are
coming up with new ideas and new concepts for understanding
very important phenomena. Although the panel of three talks
cannot represent the entire spectrum of economic research,
Gale felt it demonstrated the theoretical building blocks that
economists use when thinking about problems of financial
instability. The engineering research by Amin and Haimes
represents a very different approach. They engaged in very
large-scale projects—comprehensive, holistic modeling of
risk phenomena using real data—that aim at realism and at
prediction and control of particular systems rather than at
understanding general principles of a more generic system.
As a means to that end, Haimes stressed that these projects
integrate different models, using many different approaches
and techniques, rather than just focusing on one model.
In Gale’s view, the way economists select their research
projects reflects their incentives to pursue that course.
Economists certainly know about many of the techniques
described in the course of the conference—neural networks,
stochastic approximation, dynamical systems, optimal control,
and others—and they use them to the extent that they help to
accomplish their goals. One can readily imagine adapting the
kind of large-scale approaches undertaken by Haimes and
Amin to model the financial system. So, one logically asks
why academic economists have not pursued that line of
research—why they are not using such approaches to provide

a foundation for understanding systemic risks. The primary
reason is money: In academic economics, in Gale’s view,
no funding exists for that kind of large-scale research.
The relatively low level of funding for research in economics
has had a number of effects on how the discipline is organized.
It affects education, promotion and tenure, the publication
process, and so on. If, for example, academic economists
want to publish in a top journal, an achievement that is very
important for their professional recognition and advancement,
their papers must normally be about one model and focus on

Engineers as well as scientists in some
applied fields have more latitude [than
economists] in the types of research
they can pursue and the roles for which
they are rewarded, in part because a
wider array of funding sources exists.
economics rather than other issues. The papers typically must
include a methodological innovation. The prestigious journals
would not be interested in research that consists of applying
well-known techniques or models to some very practical
problem.
In contrast, engineers as well as scientists in some applied
fields have more latitude in the types of research they can
pursue and the roles for which they are rewarded, in part
because a wider array of funding sources exists. While some
engineering research is geared solely toward scholarly
publications, other work (even by the same individuals) might
consist of studies that inform very pragmatic decisions. The
premier honorary society for engineers in the United States—
the National Academy of Engineering—includes a mix of those
who have advanced the academic foundations of their field and
those who have advanced the profession in other ways, perhaps
as founders or managers of major enterprises. Economists,
operating in a very different culture, end up working in small
teams on what are to some extent theoretical, as opposed
to practical, problems. Even when conducting empirical
studies—as in applied economics—or when addressing issues
of regulation or optimal policy, economists generally do not
have incentives to produce work that can be immediately
applied. Economists are looking for insight, and that is a very
different kind of activity. Gale indicated that he could imagine
a role for research into systemic risk, one that would be
very exciting.
Some discussion centered on the level of resources devoted
to understanding systemic risk, with several conference

participants observing that the amount spent on studying
systemic risk is a miniscule fraction of the amount spent on
understanding and managing the risks of individual entities.
Gale noted that a prerequisite for significant change in the type
of research economists conduct is a large-scale shift in funding
for the discipline. The need is not just to provide money for
particular studies on the financial system or systemic risk,
but to change an entire discipline, which means changing
incentives across the field.
Vincent Reinhart of the Board of Governors of the Federal
Reserve System raised the possibility that change could occur
through revisiting scholarly work that had been overlooked
by the profession. In that connection, he quoted from work
by Levin (1992): “A popular fascination of theorists in all
disciplines, because of the potential for mechanistic understanding, has been with systems in which the dynamics at one
level can be understood as the collective behavior of aggregates
of similar units.” That is an appealing mechanism, if it were
true. But it is not true for the financial system or an economy
as a whole. The economy is a network of heterogeneous, not
similar, agents. Instead of transmission lines, transformers,
and switches, financial markets have market makers, brokers,
market utilities, beta providers,2 and individual investors with
different strategies. Economists have known for thirty years
that heterogeneity cannot be assumed away: In Micro Motives
and Macro Behavior, Nobel Laureate Thomas Schelling
provided many examples of how individual behaviors
produced clustering and self-organization. This conference
is evidence that the lure of a more mechanistic model
is waning.

The Role of Data
Reinhart suggested that the difference between the research
style of economists and that of engineers and physical scientists
(at least as demonstrated at the conference) might revolve
around data and computing power. As noted in part 3 of this
volume, there is more of a tradition of data sharing, and more
nonproprietary data with which to work, in engineering and
the physical and life sciences. As economists gain access to large
data sets—opening up the possibility of seeing redacted data on
individual transactions and individual behavior, as exemplified
by the Fedwire projects described in part 4—economists and
financial economists will be driven to cooperate more. To
the extent that economic researchers start developing more
complex models to represent the heterogeneity of economic
agents and combining them with large data sets—for instance,
2

Beta providers are investors whose trading drives the prices of related assets to
converge toward their normal relationships when prices diverge.

FRBNY Economic Policy Review / November 2007

of individual transactions in markets—their work will likely
become more computational, as has been the pattern in much
of the natural sciences.
In studies of systemic risk in the financial sector, key data
are transaction prices, transaction volumes and timing,
financial institution position and exposure measures, and
economic and other news. In centrally organized exchange
markets, such as the New York Stock Exchange (NYSE), good
data on prices, volumes, and timing are collected and could
be used in research. In over-the-counter markets, where
transactions are arranged between institutions and are not
recorded centrally, electronic quotation and trading systems
have improved the availability of price information. But a
preponderance of information required to study systemic
risk at some scale remains the proprietary information of
financial institutions.
The central bankers, regulators, and economists were
impressed by the cooperative arrangements in the electrical
power generating industry for sharing proprietary information
used in researching and managing systemic stability and the
insight gained from using detailed data. As risk management
and financial analysis have advanced over the last two decades,
financial institutions have developed large databases of
financial information. While financial firms are unlikely to
share very recent data, the proprietary value of information
in detailed financial institution data may decay fairly quickly,
given the rapidity with which market conditions and market
opportunities fade. If financial institutions share central
bankers and regulators’ interest in risk management tools,
the examples of data sharing from other industries might be
helpful in demonstrating the benefit of even a modest
information-sharing effort.

Potential Applications to Policy
The conference also compared sources of robustness in
financial and economic systems with those in ecological and
engineering systems and considered the implications for
mitigation. Several participants agreed that there is a need
for more research into robustness strategies in preventing
systemic events and for more analysis of the implications
for policy responses when such events occur.
The sessions revealed some important differences in
approaches to regime shift and hysteresis, with implications
for mitigation. Charles Lucas of AIG (since retired), a member
of the National Academy’s Board on Mathematical Sciences

Concluding Observations

and Their Applications opened the conference with a
discussion of the dramatic and deleterious regime shift that
occurred in the wake of the financial crisis of the late 1920s and
early 1930s: the shift from the booming, but troubled, 1920s to
the Great Depression. Economists considering systemic risk
have wrestled with the questions of when a financial disturbance can or will lead to macroeconomic effects, when those
macroeconomic impacts represent a new equilibrium for the
economy, whether the shift to a new and inferior equilibrium

While one response to financial crisis
might be to shut markets down, under
the implicit assumption that they are not
strong enough to withstand the shock,
financial economists and financial
authorities generally recommend that
markets remain open—a view based
on their trust in the flexibility of markets.
is the result of financial disturbance or policy errors, and what
sort of hysteresis—resistance to a return to the previous
equilibrium state—exists.
The effects of some financial disturbances are seen as
salutary by many economists and central bankers, leading to
improved risk management and a better long-term allocation
of resources, at least in some sectors. The banking problems
of the early 1990s and the failure of Barings in 1995 have been
widely cited as precipitating substantial innovation that
improved credit and counterparty risk management.
Many economists cite the resilience of financial markets in
handling disturbances, even long-run disturbances, principally
through the effectiveness of the price mechanism, but also
by creating new markets and contractual and institutional
arrangements. Even if prices fall very sharply, revaluation
of assets and liabilities, if allowed to occur, often results in
markets finding a new equilibrium after transactions resume.
That process may take weeks, as it did after the 1987 stock
market crash, when even though prices rebounded sharply
the next day, overall trading activity and international equity
capital flows took about ten weeks to recover to normal levels.
Or it may take longer, as it did after banks began writing down
their real estate loans and selling them off in the early 1990s.
Thus, the potential for regime shift and subsequent hysteresis
as a result of systemic events in financial markets is to some

extent offset by the flexibility and resilience of the markets
in assessing and responding to systemic shocks.
Consistent with Levin’s discussion of rigidity and
flexibility as strategies to create robustness in ecosystems,
financial systems appear to possess flexibility as a key
bulwark of robustness. One challenge for financial markets
is that the underlying infrastructure that manages the flow
of transactions may have some inherent rigidity because
of its legacy technologies and reliance on scale and network
economies; another question is whether the flexibility of
some activities is reduced by consolidation.
Rigidity and flexibility are opposite, but equally valid,
strategies to achieve robustness: a system can either be strong
enough to resist disturbances or it can be flexible enough to
“bend” to them. These two strategies also map to differing
perspectives on policymakers’ appropriate response to
financial disturbances. While one response to financial crisis
might be to shut markets down, under the implicit assumption that they are not strong enough to withstand the shock,
financial economists and financial authorities generally
recommend that markets remain open—a view based
on their trust in the flexibility of markets. There are
circumstances in which markets have been suspended: In the
immediate aftermath of the destruction of September 11,
2001, the equity markets remained closed for four days;
the NYSE instituted circuit breakers for trading after the
October 19, 1987, stock market crash; and banking holidays
are sometimes declared during major weather events.
In responding to systemic risk, monetary and financial
authorities need to think about the time frame over which
policy is expected to work. Reinhart speculated that the
presence of portfolio insurance and dynamic hedging in 1987
might have been a market mechanism that tended to amplify
the downtrend. It is not obvious what a central bank could
do in that event; the market was falling, and the central bank
could not just step into that process. It was able to remind
commercial banks that downstreaming funds to investment
banks would be a good thing, and it provided assurances
about the availability of liquidity. The markets were kept
open, trading resumed, and the markets rose subsequently;
the economy performed generally well despite the destruction
of wealth associated with the initial stock price decline.
Reinhart asserted that quick action is the right step to take,
but there is not nearly as much research available to inform

crisis management as there is to understand crisis propagation.
He thought it would be appropriate to apply the sophistication
of the work presented at the conference to crisis management
as well.
David Levermore of the University of Maryland suggested
that the ultimate benefit of the new directions suggested by the
conference might not apply so much to managing risk, which
is an important component of course, but to understanding
the economy better. Improved models of systemic risk can
incorporate and build on the theory and intuition of central
bankers and economists and refine them through additional
quantitative insight. For example, in redesigning a regulation
that currently affects all institutions of a certain type, future
policymakers might include gradations, such that perhaps
only large institutions are affected while smaller institutions
are relatively unencumbered because their health does not
constitute a systemic risk. Having that greater degree of latitude
will allow policymakers to be more creative and productive.
Reinhart noted that such a tiered system is already emerging as
a result of the Basel II Accord on bank capital requirements.
Taylor added that the public policy objective is to understand how systems can evolve so as to be more robust to tail
events. As Reinhart noted, though, we simply do not have
much data on tail events, by definition. Robert Litzenberger
of Azimuth Trust amplified that point. When we attempt to
implement risk models for catastrophic periods, we want
objective measures based in some way on historical data.
But if the data pertain to just one event, then that is a scenario
analysis, and there is no statistical reliability with respect to
its assessment. That is a major problem we face when we use
sophisticated empirical techniques with very limited data to
model the system fully. When we try to extend this thinking
beyond the Fedwire system, with its good data, to the broader
financial system, we run out of the data that would be needed
if the models are to make useful predictions. Litzenberger
compared the situation with that of econometric models of the
U.S. economy that he studied in graduate school. They were
impressive, but in truth they never predicted very well, and
many researchers eventually became disillusioned with some of
those models. To arrive at a better understanding of systemic
risk and to improve risk management tools and policies, the
discussion pointed to the immense potential value from
developing rich data sets of financial information, financial
asset prices, and institutions’ risks and earnings.

FRBNY Economic Policy Review / November 2007

References

Levin, S. A. 1992. “The Problem of Pattern and Scale in Ecology.”
Ecology 73, no. 6 (December): 1943-67.

The views expressed in this summary do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System.
58

Concluding Observations

Appendix A: Conference Program

New Directions for Understanding
Systemic Risk
Federal Reserve Bank of New York
33 Liberty Street, New York, New York
May 18-19, 2006

Thursday, May 18
8:30 a.m.-8:45 a.m. Welcome and Overview
of Conference Goals
Speakers: Christine M. Cumming, First Vice President and
Chief Operating Officer, Federal Reserve Bank of New York
Charles Lucas, Member, Board of Mathematical Sciences,
National Academy of Sciences
Industry consolidation, global networking, terrorist threats,
and heavy dependence on computing: these and other
trends introduce the possibility of new systemic risks,
perhaps with increased ramifications. Ms. Cumming and
Mr. Lucas will discuss the importance of understanding
and managing systemic risk and set the stage for the
conference as an opportunity for the central banking
community to take a fresh look at systemic risk, aided by
the insights of researchers who study similar risks in other
complex systems.
8:45 a.m.-9:45 a.m. Background on Systemic Risk
in the Financial Sector
Speakers: Darryll Hendricks, UBS
Thomas Daula, Morgan Stanley
D. Wilson Ervin, Credit Suisse 1
Three presentations will help to identify key issues relating
to systemic risk in financial markets and institutions,
describe the structure of financial markets, and give
historical examples of risks that concern central bankers
and market practitioners.

10:00 a.m.-12:30 p.m. Presentations on Current
Research Directions
Speakers: Roberto Rigobon, Massachusetts Institute
of Technology
Hyun Song Shin, Princeton University
Markus K. Brunnermeier, Princeton University
Chair: Franklin Allen, Wharton School, University
of Pennsylvania
A panel of experts in systemic risk in the financial sector will
present a cross-section of current work in this area. The
chair will raise discussion topics for the panel and serve as
moderator during a question-and-answer session with the
audience.
12:30 p.m.-1:45 p.m. Welcome and Keynote Speaker
Welcome and Introduction: Timothy Geithner, President,
Federal Reserve Bank of New York
Keynote Speaker: Donald L. Kohn, Governor, Board
of Governors of the Federal Reserve System
1:45 p.m.-4:00 p.m. Panel Discussion: Models of Systemic
Phenomena in Other Complex Interactive Situations
Panelists: Yacov Haimes, University of Virginia
Massoud Amin, University of Minnesota
Chair: Charles R. Taylor, Risk Management Association
Two researchers who model systemic phenomena in
nonfinancial systems will each make a presentation. The
first explores concepts and methods for analyzing risk in
complex engineered systems. The second outlines concepts
and methods for modeling systems, infrastructure
reliability, and catastrophic failures in complex networks
such as power grids. The speakers will address the treatment
of heavy-tailed events, the modeling of networks, and the
identification of vulnerabilities. The session chair will
compare and contrast the approaches with those typically
applied in studies of systemic risk in the financial sector
and then open the floor for discussion.

FRBNY Economic Policy Review / November 2007

Appendix A: Conference Program (Continued)

4:15 p.m.-5:30 p.m. Presentation on Systemic Dynamics
in the Federal Funds Market
Speakers: Darrell Duffie, Stanford University
Adam Ashcraft, Federal Reserve Bank of New York
The presenters will discuss preliminary results of a
simulation of the systemic risk arising from settlement
flows in the fed funds market.

Friday, May 19
8:00 a.m.-10:15 a.m. Panel Discussion: Models of Risks
Facing Complex Systems
Panelists: Simon Levin, Princeton University
Morten Bech, Federal Reserve Bank of New York
Walter E. Beyeler, Sandia National Laboratories
Robert J. Glass, Sandia National Laboratories
Chair: George Sugihara, University of California, San Diego
This session presents two talks about the risks facing
complex systems. The first speaker explores concepts and
methods for analyzing behaviors of ecosystems, especially

Conference Program

as they adapt to or approach precipitous changes. The
second talk, by a cross-disciplinary team of researchers,
presents a pilot attempt to analyze critical nodes in the
financial transaction system using tools and concepts that
are not in common use in the central banking community.
A discussion with conference participants will follow.
10:30 a.m.-12:00 p.m. Wrap-Up Panel Discussion:
What Has Been Learned?
Panelists: Douglas Gale, New York University
Robert Litzenberger, Azimuth Trust
George Sugihara, University of California, San Diego
Vincent Reinhart, Board of Governors of the Federal
Reserve System
Chair: Timothy Geithner, President, Federal Reserve Bank
of New York
Panelists from the fields of finance, economics, and science
will share observations on the conference findings and offer
thoughts for the road ahead. Conference participants will
be invited to respond.

Appendix B: Background Paper

Systemic Risk and the Financial System
Darryll Hendricks, John Kambhu, and Patricia Mosser

Introduction

his paper is intended as background material for a crossdisciplinary conference, sponsored by the Board on
Mathematical Sciences and Their Applications of the National
Academy of Sciences and the Federal Reserve Bank of New York,
on new approaches to evaluating systemic risks and managing
systemic events in the global financial system. A key objective
of the conference is to bring together a diverse group of leading
researchers who have developed analytical tools for the study
of complex systems in a range of fields of inquiry.
The stability of the financial system and the potential for
systemic risks to alter the functioning of that system have long
been important topics for central banks and for the related
research community. However, recent experiences, including
the market disruption following the attacks of September 11,
2001, suggest that existing models of systemic shocks in the
financial system may not adequately capture the propagation
of major disturbances. For example, current models do not
fully reflect the increasing complexity of the financial system’s
structure, the complete range of financial and information
flows, and the diverse nature of the endogenous behavior of
different agents in the system. Fresh thinking about systemic
risk is therefore desirable.
This paper describes the broad features of the global
financial system and the models with which researchers and
central bankers have typically approached the issues of
financial stability and systemic risk—information that can
serve as a shared reference for conference participants. The
conference itself will provide an opportunity for participants
to discuss related research in other fields and to draw out
potential connections to financial system mechanisms and
models, with the ultimate goal of stimulating new ways of
thinking about systemic risk in the financial system.

Darryll Hendricks was a senior vice president at the Federal Reserve Bank of
New York when this paper was prepared in May 2006; he is now a managing
director and the Global Head of Quantitative Risk Control at UBS Investment
Bank; John Kambhu is a vice president and Patricia Mosser a senior vice
president at the Federal Reserve Bank of New York.

Systemic risk is a difficult concept to define precisely.
A recent report by the Group of Ten (2001) on financial sector
consolidation defined systemic risk as “the risk that an event
will trigger a loss of economic value or confidence in, and
attendant increases in uncertainty about, a substantial portion
of the financial system that is serious enough to quite probably
have significant adverse effects on the real economy.” This
definition is broad enough to permit different views on
whether certain recent episodes within the financial system
constituted true systemic risk or only threatened to become
systemic if they had a significant adverse impact on the real
economy.
Some argue that even damage to the real economy is not
sufficient grounds to classify an episode as systemic; rather,
the key characteristic of systemic risk is the movement from
one stable (positive) equilibrium to another stable (negative)
equilibrium for the economy and financial system. According
to this view, research on systemic risk should focus on the
potential causes and propagation mechanisms for the “phase
transition” to a new but much less desirable equilibrium as well
as the “reinforcing feedbacks” that tend to keep the economy
and financial system trapped in that equilibrium.
While differences in the definition of systemic risk are
clearly important from a policymaking perspective, this paper
includes discussions of episodes that not everyone would agree
were systemic in nature. This is because our primary interest is
in stimulating further research on the types of propagation or
feedback mechanisms that might cause a small financial shock
to become a major disturbance, allow a financial shock to have
a material impact on the real economy, or mire the financial
system in a suboptimal equilibrium. In this regard, the
dynamics of nonsystemic episodes may still be very relevant to
the modeling of financial market behavior. Moreover, as noted
by a recent private sector report on risk management practices

The views expressed are those of the authors and do not necessarily reflect the
position of the Federal Reserve Bank of New York or the Federal Reserve
System.

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

(Counterparty Risk Management Policy Group II 2005),
“Unfortunately, in real time it is virtually impossible to draw
such distinctions.”
The remainder of this paper is divided into three sections.
The first describes the classical case of systemic risk in a
banking-dominated financial system and provides some
background information on the current workings of the
international banking system. The second section focuses on
issues that arise in market-oriented “panics” (for example, the
October 1987 stock market crash) and again seeks to provide
some relevant background information. The third section
discusses the challenges in understanding the full nature of
systemic risk posed by events of the last decade (for example,
the Asian currency crises and 9/11 payments system
disruptions) as well as key ongoing trends in the financial
markets generally. Significantly, this paper is meant primarily
to stimulate discussion of relevant issues at the conference;
it is not intended to provide a comprehensive overview of the
substantial economic literature on systemic risk and financial
instability.

Systemic Risk in Banking Systems
Banks have long been at the center of financial activity.
They remain so today, even though their share of financial
intermediation has been reduced by the growth of capital
markets and mutual funds and other developments of the last
few decades. The largest commercial banks have balance sheets
in the $1 trillion range, engage in extensive international
operations, and maintain a presence in a wide variety of retail
and wholesale financial business activities. These activities
include making loans to corporations and individuals;
underwriting debt and equity securities offerings; acting as
dealers in foreign exchange, securities, and derivatives markets;
providing asset management services; providing payments,
settlement, and custodial services; and taking deposits.
The classical model of a commercial bank is a firm that
makes loans on the asset side of its balance sheet and takes
demand deposits (checking and savings accounts) on the
liability side.1 The loans are typically perceived as being longterm “illiquid” assets in the sense that efforts to liquidate them
prior to maturity will yield a reduced value relative to their
1

The following discussion draws heavily on Diamond and Dybvig (1983).

Systemic Risk and the Financial System

intrinsic worth if held to maturity. However, the bank is
obligated to pay back demand deposits at any time the
depositor requests. Thus, banks are seen as providing a
fundamental maturity and liquidity transformation that is
both beneficial and inherently unstable.
This instability can be seen by considering a case in which
each depositor at a particular bank would be willing to leave his
or her funds on deposit, but believes that other depositors are
likely to withdraw their funds, thus making it necessary for the

Banks are seen as providing a fundamental
maturity and liquidity transformation that
is both beneficial and inherently unstable.

bank to call in its loans and suffer the associated losses. In this
case, all rational depositors will seek to withdraw their funds
as quickly as possible, producing a “run” on the bank. In this
simplified model, bank runs can be caused by concerns over
liquidity even if the bank’s assets are fundamentally sound
on a going-concern basis (that is, the bank is solvent). The
distinction between illiquidity and insolvency is one that
occurs repeatedly in discussions of systemic risk.
Moreover, in this model, bank runs can be contagious.
The contagion can arise simply as a result of a self-fulfilling
prophecy if depositors believe that other depositors will regard
a run at one bank as an indication that runs are now more likely
at other banks. Somewhat more concretely, contagion may be
more likely to occur if the issue that sparked the original run—
excessive loan exposure to real estate or the oil industry, for
example—is perceived potentially to affect other banks, or is
the result of concerns about significant interbank exposures
(that is, runs at banks seen as having large exposures to the
bank subject to the original run). Naturally, in this model, runs
are more likely at banks perceived to have a smaller equity
capital cushion to absorb declines in asset values and at banks
whose financial condition is difficult to assess in the first place.
Although the model just described is highly simplified, it
nevertheless captures the essence of past bank runs, which
occurred with some frequency before the 1930s. The primary
approaches to dealing with the risks inherent in banking
activity have included 1) controlling the relative risk of the
loans extended—for example, through regulation, 2) requiring
that bank balance sheets contain a larger share of equity capital

Appendix B: Background Paper (Continued)

and a smaller share of demand deposits, and 3) ensuring that
government provides a “lender of last resort” function and/or
deposit insurance.
The lender of last resort role is one of the most distinctive
functions of central banks. In this role, central banks such as
the Federal Reserve typically have the authority to provide
short-term loans to banks against collateral. For example,
a bank could pledge some of its loans to the central bank and
obtain cash on a short-term basis. In determining whether to
provide funds, the central bank must make a judgment about
the bank seeking funds. The conventional wisdom that
emerged in the nineteenth century was that central banks
should “lend freely at a penalty rate” when they believe that the
bank needing funds is illiquid but not insolvent, but should not
lend at all to a bank that is truly insolvent. Of course, there is
often substantial practical difficulty in distinguishing illiquidity
from insolvency.
The provision of deposit insurance in the United States
followed the bank runs of the early 1930s. Deposit insurance
aims to eliminate the threat of a bank run directly, by assuring
depositors that they will be paid regardless of whether the bank
ultimately fails. While clearly effective in discouraging bank
runs, deposit insurance further reinforced the need for bank

Although the overall importance of banks
within the financial system has declined
in the last few decades, the largest banks
in the key financial centers remain
sufficiently important that their failure
to function normally would raise questions
of a systemic nature.
regulation to limit the extent of banks’ risk taking. Economists
refer to the incentive problems created by the presence of
deposit insurance as an instance of “moral hazard.” That is,
bank managers will want to take on risk to increase their upside
potential, but insured depositors will have no incentive to
monitor or constrain their behavior. Thus, the bank runs of the
Great Depression served to shape the institutional framework
in which banks operate today—a framework that emphasizes
official regulation and supervision of banks.

In considering the systemic risk associated with banking
crises, one should also bear in mind the social costs of such
episodes. On balance, the economic literature on the Great
Depression in the United States concludes that much of the
social cost of this episode stemmed from the interruption to
credit allocation that occurred as a result of the bank runs
and contraction of the money supply. That is, the broader,
nonfinancial portion of the economy was seriously hurt by
the interruption in the financing of its activities and by the
reluctance of banks to extend new financing amid a series of
bank runs. Concern that financial sector crises may adversely
affect the functioning of many other parts of the economy
is a recurrent theme in discussions of systemic risk.
Although the overall importance of banks within the
financial system has declined in the last few decades, the largest
banks in the key financial centers remain sufficiently important
that their failure to function normally would raise questions
of a systemic nature. Significantly, these institutions exhibit
several of the characteristics discussed above.
• They are highly leveraged, with equity-to-total-asset
ratios ranging between 5 percent and 10 percent.
• While banks are less reliant on short-term-deposit
funding than the stylized model just outlined would
imply, such funding remains a material part of the
liability structure for the largest institutions.
• The scope and complexity of their activities and legal/
organizational structures make assessments of their true
financial condition by outsiders difficult, while also
posing significant management challenges for the banks
themselves.
• The largest banks typically have significant exposures to
one another, for example, through interbank deposit
markets, interdealer transactions in over-the-counter
derivatives, and wholesale payment and settlement
arrangements.
• According to some commentators, banks remain
particularly prone to cyclicality and myopia in their
credit processes, tending to forget the last cycle of bad
lending too rapidly when economic conditions brighten.
The old saying that “bad loans are made in good times”
captures the essence of this concern.
• Finally, the largest banks appear to be increasingly
subject to legal and regulatory risks stemming from
actions of their employees, risks that in some cases could
result in sudden adverse impacts.

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

Nevertheless, despite the vulnerabilities just outlined,
the financial system today does not seem highly prone to
contagious runs on very large banks. This reflects the relative
profitability and health of banks in many countries, their risk
management discipline, and the perception that the largest
banks would benefit from liquidity provision or other forms
of official assistance should runs appear imminent. In Japan,
for example, official intervention following the emergence of
significant banking sector problems in the 1990s largely
forestalled major bank runs. Interestingly, however, Japan’s
policies to prevent runs did not prevent economic weakness
associated with a banking sector too fragile to play a full and
vibrant role in financing broader economic activity. In the
United States, policymakers have indicated that large banks are
not “too big to fail” and have worked to ensure that such banks
maintain a strong financial condition and adopt rigorous risk
management policies and procedures.
This last point reflects a concern that inappropriate public
policy choices can serve to generate systemic risk. For example,
many observers note that the U.S. savings and loan crisis of the
late 1980s resulted in part from policies that paid insufficient
attention to “moral hazard” concerns. In addition, supervisors
failed to deal with insolvent firms promptly, creating strong
incentives for the management of such firms to invest in highrisk projects in an effort to restore solvency. The downside risks
of these (frequently suboptimal) investments ended up being
borne by society at large, both in the cost of government
bailouts of depositors and in the opportunity loss of numerous
investments yielding little or no return.
While the liquidity-based contagious run model of systemic
risk applies very directly to banks, it also has relevance to other
kinds of institutions. The largest securities firms rely on debt
rather than bank deposits as a significant funding source and
hold a greater share of their assets in the form of marketable
securities than do banks. Nevertheless, some analysts have
argued that securities firms may be vulnerable to contagious
runs because of their reliance on short-term funding sources
such as commercial paper, the complexity of their transactions
in less liquid securities markets, and their derivatives
businesses. As leveraged institutions, hedge funds that do not
effectively manage their liquidity risks could also be subject to
runs by their investors and creditors. Indeed, liquidity risk
management failures contributed to the problems experienced
by the Long-Term Capital Management (LTCM) hedge fund
in 1998. The case of LTCM is discussed further in this paper’s

Systemic Risk and the Financial System

final section, as it raises other issues about the sources and
propagation of systemic risk.
Significantly, a run on an individual firm alone might not
be enough to create systemic risk according to the definition
outlined above unless the liquidation of assets by the firm or an
associated reduction in the firm’s underwriting activities were
to have a material impact on economic growth. For example, in
2001, Enron suffered what amounted to a run on its short-term
liabilities in the period immediately preceding its bankruptcy
filing, but there appeared to be very limited systemic contagion
to other energy-trading firms and very little impact on the
broader economy.

Systemic Risk in Financial
Asset Markets
While the bank run model of systemic risk has been studied
fairly widely in the financial economics literature, more recent
examples of events in which concerns about systemic risk arose
have often been associated with disruptions to financial
markets, rather than runs on particular financial institutions.
For example, the 1987 stock market crash was not precipitated

A market-oriented systemic crisis typically
manifests itself as a breakdown in the
functioning of financial markets for traded
assets such as stocks and bonds, and
it may develop in response to a sharp
decline in the value of one particular
type of asset.
by concerns at an individual institution, nor was it the proximate cause of the failure of any large bank. Nevertheless, it was
clearly viewed—at the time and since—as an event with
potentially systemic consequences that warranted official
sector intervention.
A market-oriented systemic crisis typically manifests itself as
a breakdown in the functioning of financial markets for traded

Appendix B: Background Paper (Continued)

assets such as stocks and bonds, and it may develop in response
to a sharp decline in the value of one particular type of asset.
In addition to the 1987 stock market crash, examples of such
crises might include the widening of interest rate spreads and
decline in liquidity following the collapse of LTCM in 1998 and
the collapse of the junk bond market in 1989-90. In the more
distant past, the Dutch tulip mania of the 1630s and similar
episodes could, in their end-stage, be viewed as additional
examples of this type of crisis.
Consider first the characteristics of the 1987 stock market
crash. Two aspects of this systemic market episode are
particularly important to highlight. First, the episode suggests
that asset price declines can in some circumstances become
self-reinforcing and even feed into a reduced willingness on the
part of major financial institutions to bear risk across the full
range of their activities. Second, the episode underscores the
potential importance of not only the specific institutional
arrangements that are in place for clearance and settlement of
transactions but also the credit and liquidity exposures arising
from those arrangements.

Market-Based Financial Crises, Liquidity,
and Self-Reinforcing Price Movements
The shift of emphasis from bank runs to “market gridlock” as
a source of systemic risk has arisen from a number of factors,
not least the success of policies aimed at preventing bank runs
mentioned earlier. In addition, financial crises now manifest
themselves in markets rather than in institutions because
financial intermediation has moved into markets and away
from institutions. This “disintermediation” in financial activity
has been particularly pronounced in the United States in the
last thirty years. For example, in 1975, commercial banks and
thrifts held 56 percent of total credit to households and
businesses; by 2005, this figure had dropped to 33 percent.
A large fraction of financial assets—both equity and debt—
is sold directly by issuers/borrowers to investors, especially
institutional investors, via stock and bond markets, with
traditional banking effectively bypassed.
The shift from a bank-based to a market-based financial
system has expanded the types of activities that banks and other
financial intermediaries engage in and the assets that they
invest in. The large financial institutions at the core of the
system have expanded their activities to intermediate the
movement of capital among the various other participants in

multiple ways. They assist businesses in the issuance of new
stocks and bonds directly into the market (investment
banking), intermediate to buy and sell stocks and bonds
(market making) after issuance on behalf of clients (brokerdealers/trading desks), manage asset portfolios on behalf of
individuals and institutions (asset management), and lend
directly to households and businesses (traditional commercial
banking). A general trend toward consolidation of financial

Market-oriented crises tend to begin with
a large change—usually a decline—in the
price of a particular asset; the change
then becomes self-sustaining over time.
activity has led to the formation of large, complex institutions
at the core of the financial system. At the same time, however,
disintermediation has increased the importance of “end-user”
financial institutions that invest in securities on behalf of
households and firms. These include mostly unleveraged
institutional investors (mutual funds, pension funds) as
vehicles for household savings as well as more lightly regulated
and more leveraged risk-bearing entities (hedge funds).2
Market-based financial intermediation has a number of
advantages over a banking-oriented financial system. One
important advantage is that the investment risk in holding
securities is dispersed broadly among investors instead of being
concentrated in financial intermediaries. For example, debt
instruments issued by ultimate borrowers are held directly by
savers/investors to a greater degree than in a banking-oriented
financial system. Another feature of today’s financial system
that works to reduce systemic risk is the replacement of bank
deposits by mutual fund shares as an investment vehicle for
households. While the fixed face value of a bank deposit is
inherently fragile, the value of a mutual fund share fluctuates
with market prices daily. As a result, the mutual fund model is
better able to absorb and disperse shocks across a wide set of
investors.3
2

Note too that any large market participant itself consists of a very large
number of separate legal entities, with many different charters, incentive
structures, constraints, and regulations.
3
Money market mutual funds raise some of the same issues as bank deposits
because of their limited ability to bear credit losses; historically, parents of such
funds have absorbed impaired money market instruments rather than allowed
a credit loss to reduce the fund’s share value below $1.

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

Although superior to a banking-oriented financial system in
some respects, market-based financial intermediation carries
its own vulnerabilities. The capital markets work best when key
markets are liquid. In this context, the term liquidity refers to
tradability. When a market is liquid, any single trade to buy or
sell a particular asset is unlikely to have a major effect on the
price of the asset because of the large number of willing
transactors on both sides of the market. Market liquidity also
ensures that investors can buy and sell securities without undue
delay or loss in value from the price impact of the transaction.
Almost by definition, then, a market-gridlock systemic crisis is
a period when market liquidity is absent.
In normal circumstances, market liquidity rests on a
number of foundations. Foremost among them are market
making, trading, and arbitrage. Market makers buy and sell out
of inventory they maintain to meet customer demand. They
smooth out short-run imbalances in market supply and
demand, and profit from the bid/ask spread. Traders also
contribute to market liquidity by trading on bets that prices will
converge to long-run fundamental levels. These traders
typically take positions that they hold for potentially long
periods of time until prices converge to their long-run norms.
Traders play an important role in maintaining the stability of
markets and speeding up the convergence of prices to their
fundamental values.
Market-oriented crises tend to begin with a large change—
usually a decline—in the price of a particular asset; the change
then becomes self-sustaining over time. When asset prices drop
sharply, there are generally some participants willing to “swoop
in” and buy assets that have declined sufficiently in price—an
action that largely prevents the stress from becoming worse.
In systemic crises, however, investors and traders are either
unable or unwilling to step in, perhaps because their own losses
have limited their trading capacity or because an infrastructure
failure in, say, payments or settlement systems has made such
a step difficult. As prices decline, more and more market
participants sell, pushing prices lower. Eventually the price
declines become so large and persistent that no buyers emerge,
market liquidity dries up, market participants reduce their
intermediation activities and their risk taking, and market
gridlock takes hold. This sequence of events is in some measure
self-reinforcing: if price declines are sufficiently large to create
losses for traders and market makers, these participants may
cease providing liquidity to the market, thereby exacerbating
the price declines.

Systemic Risk and the Financial System

Market-based crises are often
characterized by a coordination failure in
which a wide cross-section of participants
in financial markets, including market
makers, simultaneously decide to reduce
risk taking and effectively pull back from
financing activities.
however, the firms’ actions combine to slow down or stop
financial market activity. In severe cases, the financial system
could become almost paralyzed and unable to perform its core
functions of channeling capital to investment opportunities.
The period immediately following the 1987 stock market crash
is an example of this type of coordination failure, although its
consequences were contained.
The potential for self-sustaining dynamics in financial price
movements has been studied extensively in the finance and
economics literature. Minsky (1977) and Kindleberger (1978)
have advanced theoretical explanations for many varieties of
financial crisis in which an exogenous change in the economic
environment leads to the creation of new profit opportunities
that attract capital fed by an expansion in credit.4 For a time,
these investments give rise to even more profit opportunities,
leading to a speculative euphoria that, by involving segments
of the population typically not involved in such ventures,
becomes a “mania” or a “bubble.” However, at a certain point,
knowledgeable insiders begin to take profits and sell out. Prices
begin to level off and some financial distress may ensue. A crisis
occurs when a specific event precipitates the equivalent of a run
on the asset class that was the subject of the speculative frenzy.
Aversion develops toward that asset class by banks and others
4

The discussion in this paragraph draws heavily on Kindleberger (1978).

Appendix B: Background Paper (Continued)

that had previously lent against it, and with this aversion arises
a desire to obtain liquidity at nearly any cost. The resulting
panic culminates when 1) asset prices fall so low that investors
are tempted back, 2) trading is cut off, perhaps by the closing of
the relevant exchange, or 3) a lender of last resort succeeds in
convincing the market that sufficient liquidity will be available
if necessary.
Although not all economists would subscribe to this broad
theory of speculative financial crises, the theory is useful to
keep in mind, especially in relation to those features of the
financial system that could make the system particularly
vulnerable to large, self-sustaining changes in asset prices,
and thus to market gridlock.
In modern financial systems, debt and leverage are
necessary and pervasive. Many market participants, including
the largest intermediaries, borrow funds in order to expand
their balance sheets and thus increase their ability to invest and
trade in financial assets. They adopt this strategy to increase
their return on equity capital invested (that is, by holding assets
expected to yield returns exceeding the cost of the funds
borrowed). As noted earlier, the largest banks are nearly all
leveraged more than ten to one, implying that such institutions
cannot afford to realize losses greater than 10 percent of the
value of their assets if they are to remain solvent.
The obvious implication of leverage is the need for financial
institutions to control their losses carefully and to take steps to
reduce their risk taking in the face of declining asset values. In
other words, leverage creates an incentive to sell assets whose
prices are declining, particularly if further price declines are
expected in the future. For example, if a firm is leveraged ten to
one, then even a 1 percent realized loss in asset value translates
to a 10 percent loss in the firm’s capital value. Collectively, of
course, widespread selling after an asset price decline will likely
push prices even lower and losses higher. This scenario raises
the obvious concern that such liquidations would further
amplify the underlying price movements.
Moreover, in some markets, liquidations after losses can be
automatic. For example, when an investor trades stocks on
margin accounts (by borrowing a percentage of the stock
value), a subsequent decline in the value of the stock requires
that the investor post (add) collateral—usually cash—in order
to bring the margin account back into compliance with the
margin rule of the stock exchange.
In the 1987 stock market crash, large margin calls required
investors to sell stock, thus putting further downward pressure

on stock prices. The sudden and large fall in stock prices
created large debits in the accounts of investors that had
purchased stock on margin at brokers or held long positions in
equity-linked derivatives contracts on futures exchanges. These
margin account debits created a need to transfer large sums of
cash that many investors were not able to provide within the
time frames required by brokers and the futures exchanges.
An additional feature of contemporary financial markets
that can create self-reinforcing asset price dynamics relates to
financial products that exhibit convexity in their price
behavior. Assets (or derivatives) with convexity are those that
become more or less sensitive to changes in an underlying asset
price (or other variable) as that price or variable changes.
The classic example of convexity is an option. The buyer of
a call (put) option has a right, but not an obligation, to buy (or
sell) a particular asset (for example, 100 shares of IBM) at a
particular price at some point in the future. Conversely, while
the option buyer has a right to exercise, the writer or seller of
the option has an obligation to perform. Those who have sold
options to others are exposed to what market participants call
negative convexity: as the underlying asset price moves against
the seller of the option, the value of the option position
becomes increasingly sensitive to further changes in the price of
the underlying asset. In the case of a put option, if the seller of

The obvious implication of leverage is the
need for financial institutions to control
their losses carefully and to take steps to
reduce their risk taking in the face of
declining asset values.
the option should try to compensate for this increased
sensitivity by selling the underlying asset as a hedge against
further price declines, it will put additional downward pressure
on the underlying asset price. But a further decline in the
underlying asset price simply increases the option sensitivity
again, prompting even more selling. Thus, what appears to be
a risk-mitigation strategy by the option seller locally is, in fact,
a strategy that can reinforce adverse asset price dynamics when
undertaken by a large number of sellers.
This phenomenon was evident in the 1987 stock market
crash. At that time, many institutional investors had purchased

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

portfolio insurance from intermediaries or were attempting to
replicate such insurance through dynamic hedging strategies.
Portfolio insurance is nothing more than a put option on the
underlying asset; it exhibits exactly the characteristics outlined
above. The seller of the insurance (or the firm trying to
replicate it) must hedge its position by selling in greater
amounts as prices decline, creating even further downward
price pressure. Although the extent to which such activity was
responsible for stock price declines in October 1987 is heavily
debated, there is little doubt that such strategies—if widespread—could create self-reinforcing market movements.

Importance of Clearance and Settlement
Arrangements
Clearance and settlement mechanisms contributed greatly to
the liquidity strains created by the large price declines across
cash, futures, and options markets, and the resulting margin
calls in the 1987 stock market crash. The different settlement
arrangements and time frames for different products (that is,
T+5 for stocks traded on the exchanges at that time in contrast
to same-day settlement for stock index futures) meant that
even investors that were hedged across the different markets
could face large cash demands during the interim period.
This sudden need for large cash transfers threatened to
create gridlock in the payments system and in the stock and
futures markets. Securities firms did not have the funds to
make margin payments at futures exchanges because their
customers had not made margin payments to them. The
futures exchanges’ credit risk management practices required
that positions be closed out when margin payments were not
made. This unwinding of futures positions would likely have
triggered further massive selling pressures in the stock market,
exacerbating what had already occurred. However, the
concentration of risk in the clearinghouses used to guarantee
settlement of both securities and futures transactions meant
that if positions were not closed out and markets fell further,
the integrity of the clearinghouses themselves could be
threatened. Since these clearinghouses form a core part of
the infrastructure supporting the relevant trading activities,
such an outcome could have significantly impaired market
functioning for a sustained period of time.
In the end, large commercial banks were persuaded of the
need to supply liquidity to those firms most heavily exposed
to equities (that is, by lending against the value of those

Systemic Risk and the Financial System

portfolios), and the most severe consequences were averted.
However, the banks’ action was due in part to official sector
appeals to their collective desire to avert a further deepening of
the crisis, as well as a stated willingness by the Federal Reserve
to make more liquidity generally available to the banking
sector.
This example illustrates the presence of systemic risk in
wholesale market payments, clearance, and settlement, owing
to the very sizable credit and liquidity exposures that typically

Clearance and settlement mechanisms
contributed greatly to the liquidity strains
created by the large price declines across
cash, futures, and options markets, and
the resulting margin calls in the 1987
stock market crash.

characterize such arrangements, particularly on an intraday
basis. In normal circumstances, the extension of such large
amounts of intraday credit and liquidity between the major
participants in these mechanisms facilitates more rapid
settlement of the transactions. During a crisis, however, the
reluctance of participants to continue doing business in this
fashion can potentially lead to gridlock.
A case in point is the 1974 failure of Herstatt Bank, a midsize
German bank that was closed down after it received the
deutsche mark leg of its deutsche mark-U.S. dollar currency
trades but before its pending U.S. dollar payments were
completed in the United States. This created a short-term
gridlock in the foreign exchange market that remained a source
of systemic concern until the mid-1990s, when central banks
made clear that the amounts of such “payment versus payment
mismatch” were too large to be tolerated indefinitely and the
large commercial banks invested in the CLS Bank, a system
for simultaneously settling both sides of foreign exchange
transactions.
Broadly speaking, central banks and others in the official
sector have been pushing for continuing improvements in
the robustness of payments, clearance, and settlement mechanisms. These improvements provide greater assurance to
investors that their transactions will settle, and that the
mechanisms for payment and settlement will not themselves

Appendix B: Background Paper (Continued)

become a channel for propagating systemic disturbances.
Nevertheless, these arrangements remain highly complex and
are increasingly concentrated. For example, most market
participants effectively outsource payments, clearance, and
custodial functions associated with their transactions to an
increasingly small number of global banks that specialize in
those activities.
In turn, these banks at the core of the financial system
interact with a relatively small number of specialized
organizations that actually provide the central settlement
functions for specific assets. For example, the Federal Reserve
provides settlement services for U.S. dollar wholesale payments
through its Fedwire service while the European Central Bank
does the same for euro-denominated payments. The Federal
Reserve is involved in settlement services for U.S. government
bonds through its book-entry and transfer services for those
securities, while the Depository Trust and Clearing

Broadly speaking, central banks and
others in the official sector have been
pushing for continuing improvements in
the robustness of payments, clearance,
and settlement mechanisms.

Corporation provides clearance and settlement services for a
wide range of securities, including all equities traded on U.S.
stock exchanges and corporate bonds. Significantly, all of these
systems continue in one form or another to provide large
amounts of intraday credit to their major participants.
Many financial markets (especially securities and futures
markets) have, in addition to the settlement mechanism, a
clearinghouse that provides further assurance that transactions
will settle by interposing itself as the legal counterparty to both
sides of the original transaction. The clearinghouse typically
imposes margin requirements or other controls on member
transactions while also maintaining its own financial resources
and/or the ability to call on its members’ resources. Although
clearinghouses have the ability to contain financial distress,
the concentration of settlement risk in an exchange has the
potential to focus it—as the 1987 stock market crash vividly
illustrates—if problems are severe enough to call the integrity
of the clearinghouse into question.

The Role of Central Banks
Central banks have historically played a key role in ensuring
that financial markets have sufficient liquidity to function
effectively. They have several tools that can be used in this
regard. First, they control the aggregate supply of bank
reserves—the ultimate unit of account. By increasing the
supply of reserves, central banks can increase the aggregate
amount of liquidity in the financial system. Second, central
banks function as the lender of last resort, a role that gives them
the ability to lend directly to individual commercial banks. In
extraordinary circumstances, the Federal Reserve System also
has the power to lend directly to any individual or corporation,
although this power has not been exercised since the 1930s.
Third, the central bank typically possesses sufficient influence
to persuade market participants that a collective decision to
make liquidity available in particular circumstances will
produce a better outcome than if individual market participants all seek to “free ride” on the actions of others. Largely
because these tools are so effective, the central bank can often
forestall liquidity pressures simply by announcing its
willingness to make liquidity available should the need arise.
Such announcements were made by the Federal Reserve
in the wake of the 1987 crash as well as after the events of
September 11, 2001. Elaborate assurances of this kind were
also given in advance of the Y2K rollover.
Of course, central bank actions to forestall financial crises
may themselves have a cost. In line with the moral hazard
argument discussed in the section on banking-oriented crises,
it is important that market participants not become so
complacent that they count on the central bank to defuse
any potential market-oriented financial distress and thus
underinvest in their own management of market and
liquidity risks.

New Sources of Systemic Risk
In the last ten to twenty years, financial markets have evolved
significantly. They are more global and involve a wider range
of more complex products than ever before. In some areas,
market activities have become increasingly concentrated in a
handful of very large firms. In other areas, the role of smaller,
more specialized entities has grown significantly. From a policy

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

perspective, there does not seem to be a clear consensus on
whether the financial system today is more or less vulnerable
to systemic disturbances than it was in, say, 1990.
Moreover, several of the most significant financial market
disturbances of the last decade manifested features that,
though present in earlier financial crises, have become more
prominent. As the “supply chain” has evolved from the
simplicity of a bank’s making and servicing a loan over its life
to the complexity of securitization (involving originators,

From a policy perspective, there does not
seem to be a clear consensus on whether
the financial system today is more or less
vulnerable to systemic disturbances than
it was in, say, 1990.

holders, servicers, trustees, and hedging markets), the focus
on core banks and securities firms and major markets must
expand to include other potential single points of failure. In
addition, the economic forces leading to consolidation have
included economies of scale in risk and liquidity management.
The liquidity needed in key market-risk-management markets
and in the processing of high-value dollar payments derives in
substantial part from the natural offsetting of risks or payments
when volumes are high. Finally, the global scale of large banks
and securities firms and some major investors has expanded
the channels that can transmit systemic risk.
These new features raise interesting questions about
whether the kinds of conceptual models outlined in the
preceding two sections fully capture the range of possible
causes and propagation channels for systemic risk. The
discussion below addresses two cases: 1) the events of 1997
and 1998 that involved currency crises in several Asian
countries, the Russian debt default, and the collapse of the
Long-Term Capital Management hedge fund, and 2) the
disturbances in payment and settlement arrangements
following operational disruptions resulting from the terrorist
attacks of September 11, 2001.

Systemic Risk and the Financial System

Asia, Russia, and LTCM
This sequence of events began in the summer of 1997 as certain
Asian countries faced a substantial change in market sentiment
that exposed relatively fragile macroeconomic conditions. In
particular, several countries had short-term foreign currency
debts that far exceeded their international reserves. The
countries were thus susceptible to a run on their currencies,
with generally negative consequences from a macroeconomic
point of view. While currency crises are an extremely wellstudied subset of economic crises, the Asian episode was
notable in several respects.
First, the Asian crisis was characterized by a significant
interplay between macroeconomic and financial sector factors.
This interplay reflected weakness in the banking sectors of
some countries that, while not the root cause of the crisis in all
cases, clearly affected how the crisis played out and how well
each country absorbed the macroeconomic impact of the crisis.
Second, consistent with the model of bank runs outlined
earlier, contagion figured very prominently in the Asian crisis.
Indeed, the events demonstrated a new mode of contagion.
Various trading and risk management strategies now
commonly used by market participants created linkages
between different assets and activities that may not have
previously existed, in some cases requiring positions in one
currency to be adjusted largely as a result of movements in
another. In some instances, a problem triggered by a currency
or maturity mismatch in one country or market would lead
global investors seeking to reduce risk to identify similar
vulnerabilities in other markets.
A year later, contagion figured in the relationships between
Russian debt and the debt of Brazil and other emerging
economies. Although the economies of Russia and Brazil are
not themselves closely integrated, the prices of their debt
fluctuated largely in tandem. In part, these parallel fluctuations
reflected the fact that many of the holders of this debt
specialized in holding the debt of emerging market countries,
regarded these countries as proxies for each other, and needed
to maintain some stability in their overall risk profile. Thus,
when Russian debt began to be perceived as increasingly risky
and to lose liquidity, some of these participants began to sell
their Brazilian debt to reduce their risk profile and to take
advantage of the Brazilian debt’s greater liquidity. Ultimately,
of course, the correlation between these two assets broke down
as Russia defaulted while Brazil did not.

Appendix B: Background Paper (Continued)

The Russian government default of August 1998 occurred
against the backdrop of the Asian crisis that had been playing
out over the preceding year, but otherwise took place in a
period that was characterized both by the strong macroeconomic performance of the United States and by the strong
financial condition of the major financial intermediaries.
Nevertheless, the Russian default set in motion a chain of
events that created significant fear among the leadership of
those same intermediaries and served to reduce liquidity across
most of the world’s capital markets for some months.
Long-Term Capital Management was a hedge fund that
conducted leveraged trades involving both securities and
derivatives on a large scale and used highly sophisticated
mathematical approaches to manage its risk. The firm suffered
a severe loss of capital when prices moved against its positions
following the Russian default. While LTCM’s uniquely high
leverage made it a fragile enterprise, it may not have been the
only leveraged investor to be vulnerable, and this broader
vulnerability may have played a role in amplifying the price
shocks that occurred in a number of markets following the
Russian default. For a year or two before the crisis, the liabilities
of financial intermediaries had increased substantially relative
to the liabilities of the nonfinancial sector, suggesting that
others besides LTCM had also taken on more debt and were
similarly vulnerable to price volatility and liquidity shocks. At
the onset of the crisis, however, signs of an abrupt scaling back
of leverage in trading activity emerged. For example, the
repurchase contracts that securities dealers use to finance their
own and customers’ trading positions showed a sharp and
unusually sustained decline in volume. An implication of the
deleveraging was that other traders that might have speculated
against the fall in asset prices and thereby stabilized the markets
were no longer a support in the markets.
As the ensuing market liquidity crisis unfolded during
August and September 1998, growing risk aversion made ever
larger numbers of investors seek out low-risk assets and retire
to the sidelines, and credit spreads widened sharply beyond
what had already occurred following the Russian default.
To avoid a disorderly default, and the potentially adverse
consequences of the further selling pressures it might have
incited, a consortium of LTCM’s trading counterparties
undertook a recapitalization of the hedge fund in what was
essentially an informal bankruptcy procedure conducted by the
creditors with the cooperation of the fund’s management.

Even after the LTCM recapitalization, however, spreads in
many markets continued to widen as participants showed an
ongoing aversion to risk. Other hedge funds in particular saw
dramatic changes in the willingness of major intermediaries to
finance their activities—a development that prompted further
selling and spread widening. By mid-October, reports had
grown that the situation was hindering the ability of
nonfinancial businesses to raise capital and that risk aversion
was beginning to manifest itself in payment and settlement
procedures. Only after the Federal Reserve surprised markets
with an intermeeting rate cut did the markets gradually return
to normal.
While analysts differ in their views on whether the disorderly
collapse of LTCM would have been a systemically significant
event, the episode nevertheless signals the need to think
broadly about the potential sources of systemic risk. In
particular, how has the growing emphasis on trading
activities—which are increasingly conducted through hedge
funds—affected the potential for systemic risk? Does this

The tremendous growth in the use of
financial derivatives reflects the increased
tradability of financial risk.

emphasis create mechanisms for propagation that did not exist
previously? Can these mechanisms be fully captured by the
classical models associated with bank runs or market gridlock,
or do they introduce fundamentally new elements?
Several recent trends in the financial markets bear on these
questions. One trend relates to the blurring of distinctions
between types of financial firms. Commercial banks that have
traditionally focused on making loans have increasingly
removed loans from their balance sheets through securitization
(pooling loans such as mortgages into securities sold to
investors) or outright trading of loans and securities; at the
same time, they have increased their investment banking
securities underwriting and trading activities. Conversely,
some of the largest investment banks and trading houses now
lend directly to businesses and households.
One result of this broadening of activities has been an
increased volume of trading in asset types that have in the past

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

been regarded as illiquid. Traditionally, financial assets have
been separated into liquid and illiquid assets: liquid assets
(such as stocks and government bonds) are priced and traded
regularly after issue on exchanges or in large interdealer
markets, while illiquid assets (such as bank loans) are held by
financial institutions, particularly commercial banks, over
long periods of time and are rarely traded or priced after
origination. In recent years, however, the sharp distinction
between liquid and illiquid assets has eroded, and liquidity, or
tradability, has become a continuum. While some types of
assets still trade very little after issuance, there is a trend toward
trading asset types that have traditionally been regarded as
illiquid—for example, bank loans, debt and equity of small
firms, and debt of bankrupt or distressed firms.
Moreover, financial institutions now securitize many
previously illiquid assets. Securitization involves pooling
together collections of illiquid assets such as mortgages, auto
loans, or credit card loans and creating a relatively standardized
security that pays investors the cash flows from these assets. As
a result of these changes, market participants today trade and
price a much wider array of risky assets—at least when markets
are functioning normally. During times of financial market
distress, however, the liquidity of many assets can drop sharply,
and differences in liquidity across asset types can widen
dramatically.
Similarly, the tremendous growth in the use of financial
derivatives reflects the increased tradability of financial risk.
A substantial amount of current financial market activity
involves the repackaging of claims on underlying assets and the
redistribution of the underlying risks. This last activity has
spawned enormous growth in the trading of derivatives, which
are contingent claims in which payoffs are conditioned on the
behavior of underlying variables such as interest rates or equity
prices. The institutions at the core of the financial markets not
only participate in these various activities, but also frequently
serve as market-making intermediaries.
Derivatives offer a number of advantages in the trading and
hedging of the price risks in underlying assets. First, because
they are equivalent to a leveraged trading position, derivatives
contracts can often be entered into with very little capital up
front. Thus, they are an ideal hedging instrument because the
underlying risk can be hedged without the cost of committing
a substantial amount of capital. At the same time, however,
the leveraged nature of derivatives contracts makes them risky
trading instruments, and traders that use these instruments to
speculate can lose large sums very quickly. Second, the ability

Systemic Risk and the Financial System

to structure and specify the particular underlying risk that a
derivatives contract is exposed to enables users to unbundle
a collection of risks embodied in an asset and trade the
components separately. This precision also makes derivatives
an ideal hedging and trading tool, since a hedger can choose
which risk to hedge and which to leave uncovered.
An important consequence of the widespread use of
derivatives contracts is the parsing and dispersal of the risks
embodied in underlying assets. Overall, this has provided a
net benefit to the economy, because risks that would have
remained locked up and concentrated in underlying assets are
now spread out and allocated to those more willing to bear
them. This ability to transfer unbundled risks through
derivatives contracts separately from the aggregates in
underlying assets enables investors to better select which risks
they are exposed to, providing two important benefits: lower
risk premia in asset prices because investors are no longer
locked into bearing unwanted risks, and the potential for a
better allocation of risks to those more able to bear them.
Accompanying the growth of trading in less liquid assets
and derivatives has been the general trend toward fair value
accounting for more types of instruments and positions. Fair
value, or mark-to-market, accounting imposes a discipline and

An important consequence of the
widespread use of derivatives contracts
is the parsing and dispersal of the risks
embodied in underlying assets.
transparency that can force institutions to take action to
address emerging problems that might not occur under
historical cost accounting. By contrast, historical cost
accounting is more likely to allow serious problems to go
undetected and unaddressed for longer periods of time.
A second significant trend, alluded to earlier, is the
increasing role played by a broader range of market
participants—not only hedge funds but also other forms
of specialized vehicles such as private equity firms and
collateralized debt obligation managers. These new agents for
risk bearing have the potential to alter the dynamics of how the
financial system as a whole manages risk. By allowing risk to be
spread more widely, they have the potential to help insulate the
financial system against external shocks. In the view of some

Appendix B: Background Paper (Continued)

analysts, however, a greater capacity for risk bearing may lead
the system to become even more inclined to cyclical behavior.
The extent to which these new entrants are stabilizing or
destabilizing depends in part on whether the extent of
aggregate leverage in the financial system is greater today than
in the past, since more highly leveraged institutions are more

Accompanying the growth of trading
in less liquid assets and derivatives has
been the general trend toward fair value
accounting for more types of instruments
and positions.

susceptible to large shocks that erode capital. Another critical
question relates to the linkages between these new entrants and
the traditional financial intermediaries. For example, in a
financial crisis, it may not be sufficient for banks to have
transferred risks to hedge funds if the ultimate source of
financing and liquidity for those hedge funds remains the
banks themselves. Again, the overall impact will likely depend
on whether the new arrangements increase or decrease the
amount of total equity capital at stake (including both bank
equity and hedge fund investors’ equity) relative to the size
of the risks being taken.
A third trend is the strong emphasis that leveraged
institutions—not only the large banking and securities
intermediaries but also the majority of hedge funds—put on
quantitative models for the pricing and risk management
of their activities. Risk management practices at such
organizations owe a significant debt to the efforts over the
past fifty years of many academics and practitioners to apply
statistical and mathematical techniques to the problem of
analyzing movements and comovements in market prices and
other relevant variables. Such analysis, leavened in most cases
by market experience, is used to help assess a firm’s ability to
operate safely with different combinations of assets and
leverage. Risk management strategies are also obviously critical
in influencing how financial market participants will react to
changes in market conditions. To the extent that there is
commonality in risk management models and strategies, there
is potential for a broad cross-section of market participants to
react similarly to changes in asset prices.

In valuing complex derivatives transactions, it is often
necessary to interpolate or extrapolate the fair value of such
instruments using mathematical models calibrated to the
observed market values of other, simpler instruments. In some
cases, these models are very difficult to test against an objective
reality beyond the fact that other participants are using similar
models. It is no accident that models are most commonly used
to price relatively illiquid assets; thus, during periods of
financial distress, actual prices are most likely to differ
substantially from modeled prices. A related issue is the degree
to which the positions and strategies of the diverse participants
in various markets are correlated. To the extent that many
participants are pursuing very similar strategies and will behave
very similarly in response to market shocks, the diversification
of the system as a whole may be less than it appears during
more benign periods.
All of these trends—a substantial emphasis on trading, risk
transfer, and derivatives; greater market involvement by hedge
funds; and a heavy reliance on quantitative risk management
models—were at work to some extent in the LTCM episode.
While the classical models of bank runs and market gridlock
were undoubtedly also relevant to LTCM, the episode
highlights the need to expand these models to incorporate
more fully the potential endogeneities and feedback effects
generated by the trends discussed here.

September 11, 2001, and the Reliance
on Critical Infrastructure
While the growth of hedge funds underscores how financial
market activities have expanded beyond the major commercial
and investment banks, the financial sector events following
9/11 emphasize the reliance of the financial sector on certain
core elements of infrastructure and on a relatively small
number of organizations. Two related aspects of the post–9/11
period merit discussion in this regard.
First, the terrorist attacks of that day did widespread damage
to both property and communications systems in Lower
Manhattan.5 Because many of the largest commercial banks
had operating facilities in this area (or had electronic
communications routed through hubs in the area), they were
unable to make payments as they normally would. Since most
5

This discussion draws heavily on McAndrews and Potter (2002).

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

large banks normally both send and receive a large volume of
Fedwire payments every day, relying heavily on incoming
payments for the liquidity to make their own payments, the
normal coordination of payments broke down and liquidity
shortages developed at many banks.
From a systemic perspective, the Federal Reserve attaches
extreme importance to keeping the Fedwire system open;
otherwise, this central aspect of the financial system
nationwide would not be able to function at all. Indeed, in
the wake of 9/11, the Federal Reserve extended the operating
hours of the system to help provide more time for banks to
execute their transactions. In addition, the Federal Reserve
made more liquidity available, both to individual banks
through its discount window operations and to the system
generally through open market operations. These measures,
along with the willingness of the Federal Reserve to permit
sizable intraday overdrafts, helped restore normal functioning
to the payments system.
A second set of issues arose in the market for U.S.
government securities. The clearance and settlement of these
securities (as well as a number of other fixed-income securities)
are concentrated in two commercial banks. These same two
banks provide the primary mechanism through which the
securities portfolios of the major securities firms are financed
on a daily basis (the “tri-party repo market”). The financing
itself is provided by money market mutual funds and pension
funds primarily, but the two banks provide the systems,
services, and intraday credit on which this nearly $1.5 trillion
market critically depends.
Following the 9/11 attacks in New York City, one of these
two clearing banks suffered very significant operational
disruptions, reflecting the proximity of its primary as well as
back-up operating sites to downtown Manhattan. Although
these disruptions did not completely obstruct the processing of
securities transactions, the processing slowed considerably.
Further, the destruction of brokers’ offices obstructed the
clearing and reconciliation of trades, and trade records were
not fully reconciled for several weeks. In the meantime, the
uncertainty arising from the disruptions contributed to a
significant increase in the number of trades that failed to settle.
This “fails” problem became so serious that the U.S. Treasury
conducted an unprecedented reopening of the auction for the
ten-year note in order to increase the supply of that security in
the marketplace.6
6

See Fleming and Garbade (2002).

Systemic Risk and the Financial System

Although the systemic financial consequences of the events
of 9/11 are probably best described as a “near miss,” they do
demonstrate the global financial system’s vulnerability.
Investigation of the possible outcomes of the attacks indicates
that if one of the two clearing banks had not, in fact, been
capable of operating for a sustained period of time, the task
of replicating such functionality elsewhere would have taken
considerable time, possibly as much as a year or more. In
the meantime, the underlying securities markets that are
supported by the financing activities that clear through these

Although the systemic financial
consequences of the events of 9/11 are
probably best described as a “near miss,”
they do demonstrate the global financial
system’s vulnerability.

banks would be disrupted. In particular, the U.S. government
securities market that forms the basis for the implementation
of U.S. monetary policy and the financing of U.S. government
activities and that is used as “riskless” collateral in countless
financial transactions worldwide could be impaired.
While this particular vulnerability was highlighted by the
events following the 9/11 attacks, it is almost certainly not the
only critical “choke point” in the global financial system today.
That is, the operational disruption of other relatively modest
organizations or physical facilities could significantly damage
the functioning of the overall financial system. Indeed, the last
decade has seen increased concentration in the provision of
critical infrastructure services such as payments, settlement,
and custody activities. Not surprisingly, the potential systemic
risk associated with threats to such critical infrastructure has
since 9/11 spurred a significant amount of effort by both the
public and the private sectors to increase the resiliency of that
infrastructure.
Clearly, traditional financial models of systemic risk cannot
readily capture the type of systemic risk that arises from the
potential for critical points of failure to lead to broader
disruptions in the system. For one, the proximate cause is more
operational than financial in nature. Nevertheless, the financial
aspects cannot be ignored. As the example of the breakdown in

Appendix B: Background Paper (Continued)

payments flows illustrates, even if the initial disruption stems
from physical damage or computer malfunction, the methods
of propagation may still be financial. Thus, there is a strong
need for models that are more capable of capturing the
complex interactions between operational infrastructure and
the financial flows that the infrastructure supports. Similar
models would be helpful in understanding the consequences
of a pandemic event that made it impossible for large numbers
of urban employees to work from their offices. Is the existing
financial system capable of a smooth transition to a temporarily reduced level of activity? Current models cannot readily
even frame such a question.

Implications for Systemic Risk
Three interesting themes emerge from the events discussed in
this section. First, the number of relevant points of failure has
increased with the growing complexity of the financial system.
Large financial institutions such as banks and major financial
markets such as the U.S. equity market continue to be focal
points in any assessment of systemic risk. But new sources of
risk have arisen with the growth of risk transfer through
securitization and derivatives as well as the increasing use of
central counterparties and other specialist financial institutions
that fill specific roles in the financial market infrastructure.
One further implication is that when individual institutions
have problems, the number of business relationships and
elements of risk has expanded dramatically.
Second, as the volume of transactions—payments,
derivatives, and secondary-market trading—has increased, the
apparently strong economies of scale in risk and liquidity
pooling have led to consolidation, typically into a subset of the
larger financial institutions. The high velocity of transactions
creates substantial efficiencies that are reflected in timing and
pricing. However, sharp slowdowns in transaction volume,
such as those occurring in the payments system after 9/11, can
reverse these efficiencies and potentially impair the perfor-

mance of the financial system when key parts of the system are
under stress. Similarly, a key institution’s loss of credit standing
can diminish the flow of business substantially and increase the
cost of managing its derivatives or payments books.
Third, in the information-rich global environment that has
emerged over the last few decades, the potential for contagion
has changed. That potential continues to include direct
linkages among large institutions through common credit and
market exposures or exposures to one another, although many
policy changes and enhancements to private risk management
have sought to reduce that potential. Now, however, the
potential for contagion has expanded to include associations
between risk dimensions created through common investors,
similarities in risk profiles and risk appetites, and common
exposures to macro-level risk factors such as geopolitical risk.
In periods of distress, such as the Asian currency crises, the
Russian debt default, or the LTCM collapse, such associations
may lead to the propagation of market disturbances in hard-topredict and probabilistic ways, and therefore make crises more
difficult to anticipate and manage.

Questions for Discussion
This background paper covers many different subjects at a
relatively high level. Some key questions that conference
participants might pursue are:
• What types of models of systemic failure or collapse have
proved useful in other disciplines? How applicable are
these models to the kinds of issues discussed above?
• Which aspects of the financial system seem most
important and/or challenging to capture in considering
the potential for systemic risk in the financial sector?
• What potential avenues for future cross-disciplinary
collaboration on systemic risk issues seem most
promising?

FRBNY Economic Policy Review / November 2007

Appendix B: Background Paper (Continued)

References

Kindleberger, C. P. 1978. Manias, Panics, and Crashes: A History
of Financial Crises. New York: John Wiley & Sons.

Counterparty Risk Management Policy Group II. 2005. Toward
Greater Financial Stability: A Private Sector Perspective.
Available at <http://www.crmpolicygroup.org>.
Diamond, D. W., and P. H. Dybvig. 1983. “Bank Runs, Deposit
Insurance, and Liquidity.” Journal of Political Economy 91,
no. 2 (June): 401-19.
Fleming, M. J., and K. D. Garbade. 2002. “When the Back Office
Moved to the Front Burner: Settlement Fails in the Treasury
Market after 9/11.” Federal Reserve Bank of New York
Economic Policy Review 8, no. 2 (November): 35-57.
Group of Ten. 2001. Consolidation in the Financial Sector.
Available at <http://www.bis.org/publ/gten05.html>.

McAndrews, J. J., and S. M. Potter. 2002. “Liquidity Effects of the
Events of September 11, 2001.” Federal Reserve Bank of New York
Economic Policy Review 8, no. 2 (November): 59-79.
Minsky, H. P. 1977. “A Theory of Systemic Fragility.” In E. I. Altman
and A. W. Sametz, eds., Financial Crises: Institutions
and Markets in a Fragile Environment. New York:
John Wiley & Sons.
———. 1982. “The Financial-Instability Hypothesis: Capitalist
Processes and the Behavior of the Economy.” In C. P. Kindleberger
and J.-P. Laffargue, eds., Financial Crises: Theory, History,
and Policy, 13-39. Cambridge: Cambridge University Press.

The views expressed are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York
or the Federal Reserve System.
80

Systemic Risk and the Financial System

Appendix C: About the Report Editors

John Kambhu is a vice president in the Financial Intermediation Function of the Research and Statistics Group of the
Federal Reserve Bank of New York, where he has worked since
1988. He focuses on issues relating to market liquidity, risk
management, financial derivatives, and public disclosure of
market and credit risks. Prior to joining the Bank, John was
an assistant professor of economics at Columbia University.
He received a Ph.D. in economics from New York University
in 1981.
Scott Weidman is the director of the National Research
Council’s Board on Mathematical Sciences and Their
Applications. He joined the NRC in 1989 with the Board
on Mathematical Sciences and moved to the Board on
Chemical Sciences and Technology in 1992. In 1996, Scott
established a new board to conduct annual peer reviews of the
Army Research Laboratory, which conducts a broad array of
science, engineering, and human factors research and analysis,

and he later directed a similar board that reviews the National
Institute of Standards and Technology. Scott has been with
the BMSA full-time since June 2004. During his NRC career,
he has staffed studies on a wide variety of topics related to
mathematical, chemical, and materials sciences, laboratory
assessment, risk analysis, computational science, and science
and technology policy. His current focus is on building up
the NRC’s capabilities and portfolio related to all areas of
mathematical analysis and computational science. Scott holds
bachelor’s degrees in mathematics and materials science from
Northwestern University and an M.S. and a Ph.D. in applied
mathematics from the University of Virginia.

Neel Krishnan received a B.A. in economics and
anthropology from Columbia University in 2005. He is
currently a research associate in the Capital Markets Function
of the Research and Statistics Group of the Federal Reserve
Bank of New York.

FRBNY Economic Policy Review / November 2007

Full text of Economic Policy Review (Federal Reserve Bank of New York) : Volume 13, Number 2 : New Directions for Understanding Systemic Risk

FRASER