3

CHAPTER

DETECTING SYSTEMIC RISK

Summary

T
he current crisis demonstrates the need for tools to detect systemic risks. Given that
there are many facets and causes of such risks, this chapter presents a range of mea-
sures that can be used to discern when events become systemic. The chapter first
reviews the standard financial soundness indicators’ ability to highlight those financial
institutions (FIs) that proved to be vulnerable in the current crisis. For the sample of global FIs
examined, leverage ratios and return-on-assets proved the most reliable indicators, while capital
asset ratios and nonperforming loan data lacked predictive power.
The chapter then proceeds to examine several techniques to analyze forward-looking market
data for groups of FIs in order to detect whether and when systemic risks became apparent.
Market-based measures that are able to capture tail risks seem to have given forward indica-
tions of impending stress for the overall financial system. Chapter 2 provides a slightly differ-
ent approach to systemic risk by examining interlinkages, both direct and indirect, between
selected FIs.
Finally, proxies for “market conditions” that influence (and reflect) the risks facing FIs are
examined to capture other key factors, such as investors’ risk appetite. The signaling capac-
ity of these indicators is examined by detecting whether and when they moved from low, to
medium, and to high volatility states, with the high state associated with systemic crisis. Several
measures signaled periods during which the financial system suffered a systemic crisis.
The various techniques clearly identify major stress events, such as those associated with the
merger of Bear Stearns and the failure of Lehman Brothers, as systemic. Some indicators, as
early as February 2007, also signaled rising systemic pressures. However, advance notice of
systemic stress was relatively brief and the extent to which some markets remained in high vola-
tility states was somewhat short-lived. Hence, the use of a number of market-based indicators
provides a more holistic picture.
Being able to identify systemic events at an early stage enhances policymakers’ ability to take
necessary exceptional steps to contain the crisis. In this regard, the chapter suggests enhancing
stress tests and capital requirements to take account of the buildup of systemic risks. Some of
the analysis presented could be a starting point to calibrate the risk contribution of FIs to over-
all systemic risk, thereby prompting additional regulatory capital and enhanced supervision to
discourage practices that increase systemic risk.
In sum, although systemic events are difficult to predict, and may only become apparent con-
currently in some cases, policymakers should monitor a wide range of market indicators tuned
to systemic risk, and have comprehensive crisis plans in place to be implemented quickly if
needed.

111

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

S
ystemic events are intrinsically difficult to to complement the more traditional macro-
anticipate, though once they have occurred oriented exercises attempting to predict finan-
it is easier to look back and agree that a cial crises. In particular, it focuses on the role of
disruption was, in fact, systemic. Because of the financial market signals as indicators of overall
severity and reach of the current crisis, renewed systemic risks.
attention on what constitutes a systemic crisis Specifically, the chapter seeks to answer the
and whether it can be uncovered, early or even following questions:
concurrently, has come to the fore. The task of • What were common factors among the
identifying warnings of impending systemic cri- financial institutions (FIs) that have required
ses has become increasingly complex as global public intervention? Did traditional financial
financial markets have become highly integrated soundness indicators (FSIs) provide meaning-
and hence systemic shocks can arise from and ful warnings?
extend beyond national borders. Analyzing • How can one determine which FIs are sys-
systemic risks is further hampered because temically important? Can one shed light on
there have been so few modern episodes of whether allowing Lehman Brothers to go
global systemic crises, particularly involving a bankrupt was or was not a policy “mistake” ex
core group of advanced economies. Even so, ante?
this chapter attempts to make inroads into this • What are early, or concurrent, indicators of
area by seeking to shed light on what constitute systemic risk? When might their reliability be
systemic events and by providing policymakers compromised?
with tools that can be used to recognize systemic • Can one determine when policymakers
risks. Instead of attempting to offer a single should enter and exit policies designed to
methodology, a range of empirical approaches is contain systemic risk?
examined in order to provide a more robust way The chapter presents a series of “modules” to
of detecting systemic risks.1 examine systemic risk from various perspectives.
The chapter focuses on measures of overall The chapter first looks at the “fundamental”
systemic risk derived from higher frequency characteristics of FIs based on the balance sheet
market data, rather than the identification of data that are typically used by supervisors and
underlying macroeconomic vulnerabilities based regulators. This analysis is further expanded to
on data at lower frequencies. While the latter review individual FIs from the markets’ perspec-
models are helpful in identifying the buildup of tive based on credit default swap (CDS) spreads
macroeconomic vulnerabilities, they are usu- and equity option prices. Then groups of institu-
ally not very successful in predicting the actual tions are analyzed jointly, building from simple
timing of crises or how they spill over across tools such as cluster analysis to more elaborate
global markets.2 Thus, this chapter is intended methods that look at the joint probability of vari-
ous outcomes. The role of global market condi-
Note: This chapter was written by a team comprised tions is then analyzed to shed light on whether
of Brenda González-Hermosillo (team leader), Christian certain factors, such as proxies for investors’
Capuano, Dale Gray, Heiko Hesse, Andreas Jobst, Paul
Mills, Miguel Segoviano, and Tao Sun. Yoon Sook Kim
risk appetite, affect the incidence of systemic
provided research support. The chapter also benefited
from comments from Andrew Lo and Kenneth Singleton.
1The use of multiple approaches is also present in

Chapter 2, where the perspective is to examine linkages

across institutions or groups of institutions. U.S. housing market was reached in mid-2005, the sub-
2Indeed, financial shocks (e.g., sudden stops in capital prime crisis was not revealed until 2007. Similarly, while
flows, the bursting of asset bubbles, etc.) often serve to many developing countries had sustained large current
reveal the unsustainability of macroeconomic imbalances. account deficits for several years, it was not until late 2008
Macroeconomic imbalances can last many years before that some of them began to face financing constraints
they result in crisis. For example, while the peak of the and dramatic pressures on their currencies.

112

©International Monetary Fund. Not for Redistribution

 WHAT CONSTITUTES “SYSTEMIC” RISK?

risk.3 Global market conditions are important guidance about when policymakers should use
in determining the market value of the FIs and the “systemic crisis” toolkit rather than policy
thus both influence and also echo the risks of tools meant to deal with individual institutions
individual FIs.4 or markets. Similarly, these techniques can be
Based on the sample of FIs examined, the used to determine when systemic risks subside,
results suggest that traditional balance sheet and thus provide guidance as to when to unwind
data are only partially able to detect, ex ante, guarantees and other supportive policies intro-
institutions at risk of failing. Although market- duced during the systemic phase.
based indicators are largely coincident with
events that have been deemed of systemic
importance, notably the collapse of Lehman What Constitutes “Systemic” Risk?
Brothers on September 15, 2008, some indica- “Systemic risk” is a term that is widely used,
tors are able to give some advanced signals of but is difficult to define and quantify. Indeed, it
risks. And although it would have been difficult is often viewed as a phenomenon that is there
to know ex ante that larger disruptions were “when we see it,” reflecting a sense of a broad-
coming, markets showed signs that a regime based breakdown in the functioning of the
change, a generalized breakdown of financial financial system, which is normally realized, ex
system functioning, occurred as early as late post, by a large number of failures of FIs (usu-
February 2007, when the price on the ABX ally banks). Similarly, a systemic episode may
(BBB) index began to decline and there was simply be seen as an extremely acute case of
a significant correction in the Shanghai stock financial instability, even though the degree and
market that reverberated across emerging severity of financial stress has proven difficult, if
markets.5,6 The various indicators examined not impossible, to measure.7 Systemic risk is also
suggest that letting Lehman collapse aggravated defined by the breadth of its reach across institu-
what appeared to be a global systemic financial tions, markets, and countries.
crisis already in the making because Lehman’s A natural starting point to begin to investi-
potential effects on other FIs were observable in gate systemic events is by examining individual
several indicators. FIs and their interlinkages (the latter is the
The techniques examined show some suc- focus of Chapter 2). However, during systemic
cess in revealing when the financial system is in events, channels over and above the normal
a systemically elevated regime, providing some fundamental mechanisms that link FIs and asset
markets during noncrisis periods can be impor-
3Other elements not directly considered in this chapter, tant sources of contagion.8 Contagious events,
such as the “shadow banking system” (e.g., hedge funds
and special-purpose vehicles) are also likely captured
by the various variables used to proxy for global market 7Some recent attempts to measure the degree of sever-

conditions. ity of financial stress in a given country include Illing

4For example, low interest rates reduce the default risk and Liu (2006). As well, Huang, Zhou, and Zhu (2008)
of loans. Similarly, the value of securities and other assets, develop a framework to assess the systemic risk of large
including derivatives, depend on market conditions such U.S. financial institutions. However, most empirical analy-
as overall volatility and global liquidity. ses of multi-country financial crises rely on a binomial
5The ABX (BBB) is an index based on credit default notion whereby the dependent variable takes the value
swaps written on investment-grade tranches of subprime of 1 during the known, ex post, crisis period or zero
mortgage-backed securities. otherwise with no information about the actual severity of
6Rosenblum and others (2008), Gorton (2008), and the crises (e.g., Kaminsky and Reinhart, 1999; Hardy and
González-Hermosillo (2008) also identify end-February Pazarbasioglu, 1999; Demirgüç-Kunt and Detragiache,
2007 as a period when early signs of stress began to 1998; Davis and Karim, 2008; and Weistroffer and Vallés,
emerge in global markets prior to the time when the 2008).
subprime crisis was clearly revealed in mid-2007. This cor- 8A body of literature on contagion examines these

rection reflected a reappraisal of market risks (see IMF, additional links. See, for example, Masson (1999);
2007, Box 1.5). Dornbusch, Park, and Claessens (2000); and Dungey and

113

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

which can result from asymmetric information Lastly, the observation that general “market
or uncertainty, generate changes in the normal conditions” matter for the existence and propa-
behavior of prices and thus in the distribution of gation of risks through the financial system is
returns used for trading and risk management used to examine periods of high vulnerability to
purposes, causing the distributions to be skewed shocks that may become systemic.
and “fat-tailed” (that is, exhibit more downside Since there are several concepts of systemic
than upside risk, the third moment or skewness; risk, it is natural to expect a collection of
and more “risk” generally, the fourth moment or measures rather than a single all-encompassing
kurtosis). Also important in identifying systemic index.12 Moreover, by examining systemic risk
events are the underlying “market conditions” with several complementary approaches, a
and the ability for events to further alter market more comprehensive and robust assessment can
conditions.9 For example, when the level of be made to guide policies, though not every
market uncertainty (measured by the implicit method can be expected to signal the same
volatility of assets) is high, then even a tempo- intensity or nature of systemic risk.
rary shock can lead to defaults and generate
significant aftershocks. Similarly when investors’
risk appetite is low or global liquidity is tight, “Fundamental” Characteristics of
then even relatively small shocks can have large Intervened and Nonintervened Financial
effects on global financial markets—and vice- Institutions
versa.10 Regulators and supervisors typically use a set
In this chapter, three basic concepts that of FSIs to assess the stability of their financial
underpin the measurement of systemic risk are system. Indeed, the International Monetary
used. First, several techniques apply the notion Fund (IMF) has promoted their construction
that interlinkages across institutions are impor- and collection over the last several years (see
tant—including identifying groups of similarly Annex 3.1).13 As a starting point for the analy-
exposed FIs and observing the effects of poten- sis of systemic risk, it is thus useful to examine
tial defaults of individual institutions on each whether traditional FSIs were able to discern
other and the financial system as a whole. institutions that would eventually require gov-
Second, changes in the return distributions
of FIs’ assets and equity are examined during
periods of stress to determine the additional
12Lo (2008), for example, considers that “systemic”
risks in the “tails” of such distributions and how
risk should be measured by leverage, liquidity, correla-
the “tails” of a multiple institution return distri- tion, concentration, sensitivities, and connectedness. The
bution can provide more accurate measures of Group of Ten (2001) extends systemic events to include
systemic risk.11 factors affecting the economy.
13Various studies have proposed early warning indica-

tors of impending turmoil in banking systems (e.g.,

others (2005, 2006, 2007). Dungey and others (forthcom- Demirgüç-Kunt and Detragiache, 1998, 1999, 2005; Hardy
ing) argue that the Long-Term Capital Management/ and Pazarbasioglu, 1999; González-Hermosillo, 1999;
Russian crisis in 1998 and the subprime crisis that began Hutchinson and McDill, 1999; Hutchinson; 2002; Rojas-
in mid-2007 have been the most contagious crises in the Suarez, 2001; and European Central Bank, 2005). The
past decade, based on a sample of advanced and emerg- IMF proposed sets of so-called “core” and “encouraged”
ing economies in which credit and equity market daily FSIs (Sundararajan and others, 2002), encapsulated in
data are modeled jointly across countries. the Compilation Guide on Financial Soundness Indicators
9For example, Brunnermeier and Pedersen (forthcom- (IMF, 2006), that have become essential for the macro-
ing) discuss liquidity spirals. prudential surveillance carried out by the IMF across
10Different measures of risk appetite are discussed in countries. However, recent studies suggest that FSIs may
European Central Bank (2007) and González-Hermosillo not fully capture risks (e.g., Cihák and Schaeck, 2007;
(2008). Poghosyan and Cihák, 2009; Bergo, 2002; and Sorge,
11These first two notions are also taken up in 2004), suggesting that FSIs need to be complemented by
Chapter 2. other indicators, including market data.

114

©International Monetary Fund. Not for Redistribution

 “FUNDAMENTAL” CHARACTERISTICS OF INTERVENED AND NONINTERVENED FINANCIAL INSTITUTIONS

Table 3.1. Selected Indicators on Fundamental Characteristics in Financial Institutions

Intervened U.S.
Nonintervened Banks Intervened Commercial Banks Investment Banks
1998:Q1– 2005:Q1– 1998:Q1– 2005:Q1– 1998:Q1– 2005:Q1–
2008:Q1 2007:Q2 2008:Q1 2007:Q2 2008:Q1 2007:Q2

Capital adequacy (in percent)

Capital/assets 14.5 19.4 17.9*** 20.3 17.3** 19.4
Common equity/assets 3.7 4.4 6.0*** 5.7*** 3.7 3.7**
Tier 1 capital/risk-weighted assets 4.9 10.8 8.1*** 9.0 ... ...
Tier 1 and 2 capital/risk-weighted assets 7.3 15.8 11.0*** 12.5 ... ...
Asset quality (in percent)
Nonperforming loan ratio 2.3 2.3 1.4*** 1.0** n.a. n.a.
Provisions for loan losses/loans 0.1 0.1 0.2*** 0.2*** n.a. n.a.
Leverage
Debt to common equity 7.5 7.6 8.1*** 9.0*** 13.3*** 13.7***
Short-term debt1 0.4 0.5 0.7*** 0.7*** 0.7*** 0.7***
Liquidity
Loans/deposits 1.1 1.3 1.2 1.3 n.a. n.a.
Loans/assets 0.6 0.5 0.5*** 0.5*** n.a. n.a.
Earning and profit (in percent)
Return on assets 1.2 1.2 1.9*** 1.6*** 3.9*** 4.3***
Return on equity 3.6 4.8 4.1 5.3 4.1 5.3
Stock market performance
Price/earnings ratio 15.5 12.6 16.8 12.0 15.6 13.1
Earnings per share 0.6 1.0 0.6 0.9 1.3*** 2.4***
Book value per share 14.8 21.7 14.1 18.3*** 34.0*** 50.5***
Sources: Thomson Reuters; and IMF staff estimates.
Note: A t-test is performed to determine whether two samples are likely to have come from the same two underlying populations that have
the same mean. The intervened commercial banks and the U.S. investment banks are compared to the nonintervened banks. *, **, and ***
represent the statistically significant differences at the 10, 5, and 1 percent levels, respectively.
1Short-term and other debt payable within one year.

ernment intervention from those that have not they are readily available and some are widely
from a small sample of major institutions.14 used by financial regulators. However, these indi-
The sample comprises 36 key commercial cators are also reported at low frequencies, are
and investment banks across the world (Annex generally static and backward-looking, and focus
3.2).15 The advantage of focusing on FSIs is that on an individual FI without much regard for
the spillovers from other institutions. Table 3.1
14In this chapter, intervened institutions are assumed to
divides the sample of FIs into nonintervened
be those that have gone bankrupt, or that have received
commercial banks, intervened commercial
government capital injections or loans, or that have had banks, and intervened investment banks during
assets purchased by government, or that have received 1998:Q1–2008:Q1 (before the wave of govern-
official loans to facilitate a merger or acquisition. Central
ment intervention) and 2005:Q1–2007:Q2
bank temporary liquidity injections are not considered
to be a type of intervention. Intervened institutions and (before the start of current cycle and the begin-
periods of intervention are detailed in Annex 3.3. ning of the subprime crisis).
15The insurance companies were excluded from the
The results in Table 3.1 show the following:
analysis given their different business lines. The rationale
for choosing these FIs is based on their systemic impor- • Capital adequacy ratios were unable to clearly
tance while keeping a balanced sample representative of identify institutions requiring intervention. In
the various regions around the world. Data constraints fact, contrary to the common belief that low
also played a role, as the sample chosen was limited to
FIs for which balance sheet and market-based data were
capital adequacy ratios would signal weak-
available. ness for a FI, all four capital adequacy ratios

115

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

examined for intervened commercial banks

were significantly higher than (or similar
to) the nonintervened commercial banks as
a whole (Figure 3.1). There are, of course,
regional differences among nonintervened
commercial banks. During 2005:Q1–2007:Q2,
Figure 3.1. Capital-to-Assets Ratio the capital-to-assets ratio for nonintervened
(In percent)
commercial banks in Asia and the euro area
35
were higher than for intervened commer-
Intervened U.S. 30 cial banks. However this was not the case for
investment banks
FIs in the noneuro area. This suggests that
25
regional differences can make direct compari-
20 sons problematic.16
• Several basic indicators of leverage appear to
15 be informative in identifying the differences
Intervened
Nonintervened banks in the institutions, although the reasons for
banks 10
this deserve further examination. The higher
5 ratios of debt to common equity, and short-
term debt to total debt in the intervened
0
1998 2000 02 04 06 08 commercial banks and intervened investment
banks, all indicate that these measures of
Sources: Thomson Reuters; and IMF staff estimates.
Note: The ratios of nonintervened banks, intervened banks, and intervened U.S.
leverage are especially informative about the
investment banks are the average of all institutions in each category. differences (Figure 3.2).17
• Traditional liquidity ratios are not very indica-
tive of the differences between intervened
Figure 3.2. Ratio of Short-Term Debt to Total Debt1
(In percent) and nonintervened institutions. In part, this is
100
because these liquidity ratios may not be able
Intervened U.S. to fully measure wholesale funding risks.
Intervened
investment banks banks • Asset quality indicators show a mixed picture.
80
Similar to the capital adequacy ratios, the
ratio of nonperforming loans (NPL) to total
60 loans for the intervened commercial banks
has been lower than for the nonintervened
40 commercial banks, indicating that NPL ratios
are not very reliable indicators of the dete-
20 rioration in asset quality. However, the lower
Nonintervened provisions for the loan-losses-to-total-loans
banks
ratio for the nonintervened commercial
0
1998 2000 02 04 06 08

Sources: Thomson Reuters; and IMF staff estimates.

Note: The ratios of nonintervened banks, intervened banks, and intervened U.S.
16The reasons that capital adequacy ratios are not
investment banks are the average of all institutions in each category.
1Short-term and other debt payable within one year.
always useful indicators of distress may reflect (1) dif-
ficulties in determining the actual riskiness of assets; (2)
deficiencies in mark-to-market accounting practices; and
(3) locating assets and contingent claims (e.g., deriva-
tives) in off-balance-sheet vehicles where they can receive
lower risk-weights.
17Short-term and other debt payable within one year.

116

©International Monetary Fund. Not for Redistribution

 “FUNDAMENTAL” CHARACTERISTICS OF INTERVENED AND NONINTERVENED FINANCIAL INSTITUTIONS

banks suggests that this is a better indicator

than the NPL ratio.
• The standard measures of earnings and
profits show a mixed picture. While return on
assets (ROA) for the intervened institutions is
much higher than that in the nonintervened
commercial banks, suggesting that elevated
risks are associated with higher returns,
return on equity (ROE) has not captured any
major differences between the FIs that were
intervened or not (Figure 3.3). This contrast
between the effectiveness in ROA and ROE
may reflect the high leverage ratio of inter-
vened FIs, which typically rely on higher levels
of debt to produce profits.
• Stock market indicators are able to capture Figure 3.3. Return on Assets
(In percent)
some differences. The price-to-earnings ratios,
7
earnings per share, and book value per share
of the intervened investment banks were 6
generally higher than those in the noninter-
5
vened commercial banks, which suggest that
the higher equity prices and earnings do not Intervened U.S.
investment banks 4
necessarily reflect healthier institutions, but
perhaps concomitant higher risks. 3

This section finds that (1) risk-weighted

Intervened banks 2
capital adequacy ratios have generally not been
informative in discerning financial firms that Nonintervened banks
1
eventually required intervention (in fact, the
0
intervened institutions sometimes had higher 1998 2000 02 04 06 08
capital adequacy ratios than the nonintervened
Sources: Thomson Reuters; and IMF staff estimates.
institutions); and (2) several indicators, such Note: The ratios of nonintervened banks, intervened banks, and
as the debt-to-common-equity ratio, short-term- intervened U.S. investment banks are the average of all institutions in each
category.
debt-to-total-debt ratio, ROA, and stock market
indicators, have been better at discerning the
differences between intervened and noninter-
vened institutions.
In conclusion, based on the sample of institu-
tions examined, which notably includes U.S.
investment banks, it would be useful to include
indicators on leverage and more on stock
market performance on the regulatory radar
screen, since they could provide a starting point
for a deeper analysis of vulnerable institutions.
Also, the center-stage focus on regulatory capital
adequacy ratios may need to be redefined,
especially if it can be shown that FIs were able
to shift risks to off-balance-sheet vehicles, which

117

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

receive lower risk weights, and thus the risks on simple measures using individual institutions
the balance sheet are underrepresenting those before moving to more sophisticated measures
of the FI. Though the analysis here has been that account for the interactions among a num-
partial and cursory, others have found similar ber of FIs.
issues with the application of FSIs, calling for
further improvements (see footnote 13). For less
sophisticated institutions and general financial Brief Taxonomy of Credit Risk and Tail-Risk
sector analysis, the FSIs can still be useful to Models
signal risks. The different tools to assess systemic risks by
examining FI risks, both individually and collec-
tively, are summarized in Table 3.2. One family
Market Perceptions of Risk of Financial of tools includes the contingent claims approach
Institutions (CCA), which explicitly accounts for the inher-
Financial soundness indicators, especially ent uncertainty in balance sheet components,
those based on accounting balance sheet data, and links the value of equity, assets, and debt
have certain limitations: they fail to anticipate in an integrated way.19 Generally, this set of
changes in market conditions and spillovers models takes the volatility of equity prices as the
from other FIs, and tend to be static and back- starting point and derives other risk measures
ward looking. In particular, investment positions from it.20 This approach has been widely applied
and bank loans that are apparently profitable at in the analysis of credit risk, as it permits the
a given time can turn into large losses if market estimation of asset values and asset volatility
conditions deteriorate going forward. More- (that are otherwise not directly observable),
over, in addition to general market conditions, which are used to provide an equity market-
asset prices may reflect how other FIs value based assessment of default risk (Box 3.1). The
similar assets. By contrast, these and other incorporation of uncertainty and asset volatility
issues, including business objectives and the are important elements in risk analysis since
management quality of firms, are continuously uncertain changes in future asset values rela-
monitored by markets and are reflected in their tive to promised payments on debt obligations
equity prices and CDS spreads, perhaps provid- ultimately drive default risk and credit spreads—
ing more sensitive assessments of the institu- important elements of credit risk analysis and,
tions’ future prospects and their interactions.18 further, systemic risk.
This section investigates how markets perceive Another set of tools uses equity options prices
FIs, attempting to discern whether such market- (or equivalently, their implied volatility) as start-
based measures gave any advanced knowledge ing points. Examining higher moments of equity
of the impending difficulties, or if they can be options is critical to account for nonlinearities of
used to determine when the disruptions become
systemic. The analysis that follows relies on mar-
19CCA is a generalization of the option pricing theory
ket perceptions of the FIs’ risk and starts with
pioneered by Black-Scholes (1973) and Merton (1974).
The approach is based on three principles: (1) the
values of liabilities are derived from the value of assets;
18These spreads are quoted as a spread over the equiva- (2) liabilities have different characteristics (i.e., senior
lent maturity U.S. treasury securities for U.S. institutions. and junior claims); and (3) the value of assets follows a
For institutions in various countries, they are a spread stochastic process.
over the comparable government security. Note that all 20These include risk exposures in risky debt, prob-

market-traded prices (CDS spreads, equity, and equity abilities of default, distance-to-distress, the present
options) also contain a liquidity risk component—the value of the expected loss (i.e., the value of the implicit
risk that an investor may or may not be able to trade at a put option), spreads on debt, and the sensitivity of the
price close to the last traded price. Such risks rise during implicit options to the change in the underlying asset and
periods of stress. other sensitivity measures.

118

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

Table 3.2. Taxonomy of Credit Risk Models

Univariate Measures Multivariate Measures
Merton Higher moments Time-varying multivariate
Accounting contingent claims CDS-based and multivariate density distress dependence
balance sheet approach model Moody’s KMV Option-iPoD1 PoD dependence2 and tail risk3
Calibrated using Accounting data Historical equity Historical equity Equity option CDS and Equity option Individual CDS-PoDs and/or
volatility4 volatility data recovery data stock prices5
rate
Outputs for (1) Financial (1) Implied asset EDF and EDF- (1) Univariate PoD n.a. n.a.
individual soundness distribution; implied CDS probability
institutions indicators; and (2) Implicit put density function;
(2) Other ratios option; and (2) PoD; and
(3) Credit (3) Probability
spreads of default
hitting leverage
threshold
Multiple n.a. n.a. n.a. n.a. n.a. (1) Recovers (1) Recovers multivariate
institutions multivariate density and thus common
density; and distress in the system:
(2) Dependence JPoD, bank stability index;
measures (2) Distress dependence
between matrix; and (3) Probability of
institutions cascade effects triggered by
particular financial institution.
Advantages Widely available Simple way to (1) Time-varying Accounts for Measures (1) Appears to (1) Able to use other PoDs;
measure and volatility; and deviations from map to lead CDS; and (2) Multiple outputs;
analyze credit (2) Provides log-normality disruptions (2) Generates (3) Includes linear and
risk EDFs that can and has model- in markets systemic risk nonlinear dependence; and
be mapped to determined measures (4) Endogenous time-varying
ratings default barrier distress dependence
Shortcomings (1) Static (1) Constant Assumed default Requires options Uncertain Potentially Drawbacks attached to the
backward asset volatility barrier quoted at a recovery affected by inputs (e.g., PoDs) would
looking; and unrealistic; and variety of strikes rate government affect the output
(2) Accounting (2) Assumed not directly capital injections
definitions can default barrier comparable with or dilution
differ across one-year default
countries probability
estimates
Estimated in “Fundamental” Box 3.1 Box 3.1 Box 3.2 n.a. Box 3.3 Box 3.4
this chapter Characteristics of
Intervened and
Nonintervened
Financial
Institutions
Source: IMF staff.
Note: CDS = credit default swap; EDF = expected default frequency; JPoD = joint probability of distress; option-iPoD = option-implied probability of default;
PoD = probability of default. The literature on credit-risk modeling is large; see Lando (2004) and Gray and Malone (2008), among others, for an overview of
popular models. The table describes the features of the models presented in the chapter. Enhanced contingent claims approach models include extensions of the
Merton model to include time-varying volatility (like MKMV) and other extensions. Some equity option-based credit risk models, such as Hull, Nelken, and White
(2004), explicitly use two or more equity options to calibrate higher moments of the underlying asset distribution. Other equity-option-based credit risk models,
such as Zou (2003), and option-iPoD, calibrate the entire probability density function of the underlying asset.
1Capuano (2008).
2Gray and Jobst (forthcoming).
3Segoviano and Goodhart (2009).
4Model can use implied volatility from options.
5Model can use PoDs estimated from alternative methods, not only CDS spreads.

changes of default risk and thus provides a tool (option-iPoD), featured below, uses equity option
to observe when FIs’ defaults may become sys- prices to infer default probabilities on individual
temic. The option-implied probability of default FIs, with the advantage that determining when

119

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Box 3.1. Modeling Risk-Adjusted Balance Sheets: The Contingent Claims Approach

Forward-looking equity market information

Implied Asset Distribution: Citigroup
can be combined with balance sheet informa-
tion to estimate risk-adjusted balance sheets that Crisis (February/March 2008)
provide useful and timely indicators of default Moderate distress (October 2007)
probability and credit risk. Calm (January 2007)
0.007
The contingent claims approach (CCA) is a Default barrier
risk-adjusted balance sheet framework where
0.006
equity and risky debt of a firm or financial
institution derive their value from assets, which 0.005
are uncertain. The total market value of assets Higher default
risk/spread

Probability
at any time is equal to the market value of the level 0.004

claims on the assets, which is represented by

0.003
equity, and risky debt maturing at time T:
Assets = Equity + Risky Debt 0.002

Asset values are uncertain and in the future

0.001
may decline below the point where debt pay-
ments on scheduled dates cannot be made. In 0
1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2
the CCA, the equity can be modeled and calcu-
Asset value (trillions of U.S. dollars)
lated as an implicit call option on the assets, with
an exercise price equal to the promised debt pay- Sources: Bloomberg L.P.; Moody’s KMV; and IMF staff
estimates.
ments, B, maturing in T–t periods. The risky debt Note: Implied asset distribution from equity option prices from
is equivalent in value to default-free debt minus a Bloomberg for three dates and the default barrier estimated by
Moody’s KMV.
guarantee against default. This guarantee can be
calculated as the value of a put on the assets with
an exercise price equal to B:
Risky Debt = Default-Free Debt − Debt Guarantee
In the CCA framework, the value of the equity $75, and T = 1 (one year), then the value of
can be computed as the value of an implicit the equity is $32.36, the value of risky debt is
call option and the value of the debt guarantee $67.63, and the credit spread is 534 basis points.
can be modeled as an implicit put option. The The Merton model has been extended in
balance sheet components can be calibrated many directions, including models where the
by using the value of market capitalization, asset volatility is not constant. For example,
the volatility of equity, and information from information from equity options can be used.
the balance sheet to define the “distress” or The figure shows the implied asset distribution
“default barrier.” Using two equations and two (in billions of dollars) for Citigroup in January
unknowns, the implied asset level and implied 2007 (calm period), October 2007 (moderate
asset volatility can be calculated. The credit risk distress period), and February/March 2008
indicators can be calculated, i.e., default prob- (crisis period). As can be seen, the left tail skew
abilities, spreads, distance-to-distress. Robert C. is very small in the calm period (credit default
Merton proposed the CCA framework and the swap [CDS] spread was 12 bps), but it increases
simple model is known as the Merton model, in the moderate distress period (CDS spread
where a constant volatility of assets is assumed. was 124 bps) and is even larger in the crisis
Example: Assuming that Assets = $100, volatil- period (CDS spread over 200 bps).
ity σ = 0.40 (40 percent), distress barrier B = Moody’s KMV is based on a CCA-type model.

Note: Dale Gray prepared this box.

120

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

the institution goes into default (the default (Table 2.8). The third output, probability of cas-
barrier) is also derived within the model in line cade effects whereby the distress of a particular
with the observation that the value of debt also FI affects another, is presented below.
moves with market conditions (Box 3.2). This is One disadvantage of using market data (CDS
an advance over other models in which a default spreads and equity options) to infer PoDs (or
barrier is assumed to be fixed. other tail behavior) in the current period is
Two general methods are then employed the recent extension of government financial
to examine FI interdependence and thus the guarantees on FI debt, as this can transfer risk to
incidence of systemic risk. The first uses higher the sovereign entity—thus sharing the credit risk
moments in equity and implied asset distribu- of FIs with the other debt holders. For example,
tions calibrated from equity options. Equity this alters the interpretation of CDS data for
option information can be used to calculate FIs.22
tail-risk indicators for individual institutions The use of several different tools and super-
as well as between institutions. These tail risks visory examinations to analyze similar FI risks is
encompass both the skewness and the kurtosis helpful because if the basic conclusions are the
and thus adjust to stressful conditions. More same, then policymakers will have more comfort
accurate indicators of interdependence of FIs in using the tools for their analysis of systemic
are obtained by “tail dependence” measures risks. Moreover, since some tools may not be
as compared to simple correlation measures appropriate under certain conditions (e.g.,
(Box 3.3).21 when government guarantees are in place or
The second method calculates a joint prob- when short-selling restrictions are imposed on
ability of distress (JPoD) among a group of FIs equities), it is useful to know which techniques
and then a banking stability index (BSI), which are still valid.
estimates the probability of default (PoD) of
other FIs if one institution defaults. Instead of
equity volatility or equity options, CDS spreads Measures of Risk Based on Individual Financial
are used to calculate the PoD for individual Institutions
institutions and as an input to the model,
though the general technique could be applied Conditional Correlations and Cluster Analysis
using equity prices (Box 3.4). Once the JPoDs A simple starting point for potential systemic
are estimated, there are three potential outputs: connections among FIs is to use conditional
the BSI; a matrix of (pairwise) distress depen- correlations and cluster analysis. Observing how
dencies; and the probability of one or more FIs (or whether) these measures change over time
becoming distressed if a specific FI becomes may provide supervisors with information about
distressed. Examples of the second application which institutions’ failures would affect others.
are discussed in Chapter 2, which presents a Based on a sample of 45 individual FIs, equity
matrix of distress dependencies before the crisis returns are used to investigate the conditional
and at different periods since the crisis began correlations and clusters among them during
various intervals beginning in January 2005.

21Although higher (Pearson) correlation coefficients

are commonly used to measure potential spillover effects 22In principle, one reason to choose either equity-based

and systemic risks, these conventional correlations are information or CDS spreads to deduce PoDs would be if
inaccurate measures of dependence in the presence there were a lead-lag relationship showing one as provid-
of skewed asset distributions and higher volatility. The ing default information earlier. Linear and nonlinear
standard correlation coefficient detects only linear depen- Granger causality tests suggest unidirectional Granger
dence between two variables, making it ill-suited for the causality from stock returns to CDS changes, although
examination of systemic risk when extreme events occur there are no clear-cut dynamics in all sample cases (Baek
jointly and in a nonlinear fashion. and Brock, 1992; and Hiemstra and Jones, 1994).

121

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

The conditional correlation matrices are based Table 3.3. Correlations Among 45 Financial
on residual equity returns, which are free from Institutions During Different Stress Periods
world and local market effects and volatility.23 Number of
Cluster analysis (also known as “look-alike Coefficients within
the Range
groups”) attempts to determine the natural
Correlation coefficient values 0.5–0.6 >0.6
grouping (a “class”) that captures similarity or
distance between observations. In particular, Post approval of the Troubled Assets Relief 23 10
Program (October 3, 2008–December 31,
the analysis is used to determine groups of FIs 2008)
where their residual equity returns behave in Lehman’s collapse to the approval of 87 68
the Troubled Assets Relief Program
similar ways. These companies can then be con- (September 15, 2008–October 2, 2008)
sidered to be “similar” institutions.24 The draw- Rescue of Bear Stearns to Lehman’s 73 52
back for both correlation and cluster analysis is collapse (March 17, 2008–September 12,
2008)
that even after controlling for world and local Bankruptcy of two hedge funds of Bear 41 19
market effects and volatility, the methodology Stearns to rescue of Bear Stearns (August
1, 2007–March 16, 2008)
may not fully capture nonlinear dependencies in
Shanghai stock market correction to the 16 2
the data.25 Despite this (important) caveat, the bankruptcy of two hedge funds of Bear
conditional correlation and cluster analysis show Stearns (February 27, 2005–July 31, 2007)
Before Shanghai stock market correction 17 8
a relatively higher degree of co-movements of (January 3, 2005–February 26, 2007)
most FIs during the stress periods than during
Sources: Bloomberg L.P.; and IMF staff estimates.
normal periods.
Specifically, a comparison between different
stress periods indicates the following: • The conditional correlations show that the
highest correlations among FIs occur in the
period between Lehman’s bankruptcy on
23To concentrate on the extra correlation among
September 15, 2008 and the approval of the
these 45 institutions, three steps are taken to get residual
returns. Specifically, first regress each institution’s equity
Troubled Assets Relief Program (TARP) on
return on the return on the world equity index and the October 2, 2008.26 The period between the
return on the relevant local equity index, respectively. rescue of Bear Stearns and Lehman’s collapse
Thus, the data is first purged by performing the following
ranks second in the context of high correla-
regression:
tions among institutions (Table 3.3).
ri = c + b1Wi + b2Li + resi ,
• The average variance in three clusters or
where the dependent variable r is the equity return for groupings of FIs rises from 1 in a normal
each of the institutions at time t, W represents the return
on the MSCI world equity index and L represents the period (before the Shanghai stock market
return on the relevant local equity MSCI index. Second, correction) to 2.7 in the stress period (after
GARCH(1,1) models are performed to account for excess the Lehman bankruptcy).
kurtosis and volatility clustering, resulting in new residual
returns. Third, conditional correlations are estimated
• The within-class variance in cluster 1, where
conditioned on negative MSCI world equity returns to most FIs are grouped together, is 86 percent
capture more directly systemic risks. higher during the stress period than during
24Though many types of cluster analysis exist, the
the normal period (Table 3.4).
agglomerative hierarchical cluster analysis is the most
popular. This approach combines FIs into groups of • The tree diagrams in Figure 3.4 for the
similar institutions. The algorithm initially views each groups of FIs show the greater extent of cross-
observation as a separate group (giving N groups each of border co-movement and interconnections
size 1). The closest two groups in terms of the Euclid-
ean distance are then combined (giving N–2 groups of
1, and one group of 2). This process continues until all
observations are combined into one group (of N financial
institutions). 26The TARP is the U.S. government program to
25As argued by Forbes and Rigobon (2002), correlation purchase assets and equity from financial institutions in
coefficient can be biased during periods of high volatility. order to strengthen the financial sector.

122

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

Figure 3.4. Dendrogram

(Euclidean distance)

Before Shanghai Stock Market Correction, January 3, 2005–February 26, 2007 600

All financial institutions

500

400

300

200
Asian and European Asian U.S.
Insurance financial institutions financial institutions
financial institutions
companies
European financial institutions 100

0
S
RU N
KN

A
V
UV

A
H
DA BN
K
Z
P

HS E
BA
G
K
S
K

CS N
GN

FO A
BA B
RC

HB U
LL S
OY

F
UF
P
C
GS

C
S
ER

K
M
C
G
C
B
M
I
K
BI
PM
SM
AL

AN

GL
DB

AX

ND

NS

BN

CB
UB
DB

G
R

RB

IS
BO

BS
M

IB

BA
W

AB
I

LE

UC

SA

PR

AI
JP

NO

M
SB

M
M

IN

Lehman Collapse to the Approval of the Troubled Assets Relief Program, September 15, 2008–October 2, 2008 600

All financial institutions

500

400

300
U.S., European, and Asian
financial institutions
U.S., European, 200
and Asian U.S., European, and Asian
financial financial institutions
European and Asian European and U.S.
institutions 100
financial institutions financial institutions

0
P
K
DA BN
K
BI

FO B
RB
IN
K
K
K
G
M

RU V
KN

S
S
C
A

HS C
BA
N
M
C

BA A
RC

U
ER

CS Y
GN
GA
GS

V
Z
S
F

MS
UF

H
C
G
P

HB E
OS

I
PM
SM
AL
AN

GL
IS
DB

NS

IB
AB
CB

DB
M
BO
ND
BA

BS
AX

RB

UB

BN
UC

SA

PR

LE

AI
NO

JP
M

SB

M
I

LL

IN

Sources: Bloomberg, L.P.; and IMF staff estimates.

Note: A dendrogram (tree diagram) is used to illustrate the arrangement of the clusters and determine groups of financial institutions whose residual equity returns behave in
similar ways. These companies are considered to be similar institutions. Sample of 45 institutions, see Annex 3.2.

among FIs during the stress period.27 During divided into U.S. investment banks in the mid-
the normal period, FIs are mainly clustered dle of the tree in magenta and U.S. commer-
based on geography and their primary line cial banks and insurance on the right-hand
of business, as indicated by obvious divisions side of the tree in blue) and a combination
between the U.S. FIs (which are further of the insurance, European-Asian FIs (on the
left-hand side of the tree in green). During
27The tree diagram (dendrogram) is used to illustrate the stress period, however, FIs are clustered
the arrangement of the clusters produced by a cluster- based completely on cross-border groupings.
ing algorithm. It is applied here to determine groups of In particular, the FIs cleanly divide into the
financial institutions where their residual returns (based
European-Asian group (in the middle of the
on the same data as the conditional correlation analysis)
behave in similar ways. tree in magenta), a smaller group of U.S.-

123

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Table 3.4. Cluster Analysis

Lehman’s Collapse to the Approval of the
Before Shanghai Stock Market Correction Troubled Assets Relief Program
(January 3, 2005–February 26, 2007) (September 15, 2008–October 2, 2008)
Cluster1 1 2 3 1 2 3
Number of institutions 31 4 10 27 10 8
Within-cluster variance of residual returns 1.31 0.77 0.89 2.45 3.04 2.60
Average variance across clusters 0.99 2.70
Sources: Bloomberg L.P.; and IMF staff estimates.
1Three clusters are determined automatically by the clustering algorithm.

European-Asian FIs (on the left-hand side of within the model of univariate probability
the tree in green) and a larger combination distributions.
of U.S.-European-Asian FIs (on the right-hand Applied to five institutions during the cur-
side of the tree in blue). In the latter group, rent crisis, the option-iPoD model would have
the bloc contains subgroups made up of provided some early warning signals of distress
U.S.-European institutions and U.S.-European- for some of the key FIs (Box 3.2). On several
Asian groups. occasions prior to their respective “default
In sum, although these techniques are fairly events,” the option-iPoD jumped by a multi-
basic and have a number of caveats, they can plicative factor for several of the institutions
be used to judge whether certain groups of that have required intervention.28 Ex post, the
institutions’ returns are perceived as being pattern of warning signals suggested that Bear
more similar during periods of stress, and thus Stearns, Merrill Lynch, and Wachovia were
to determine the prospects for spillovers to the perceived by markets as having a heightened
group in the case of a single institution’s dis- chance of default before their difficulties were
tress. Moreover, the tree diagrams can be used announced, although these signals were less
to provide a rough idea of which institutions severe for Lehman and Citigroup. Although the
are viewed by markets as having similar return model does not give definitive signals for all five
characteristics and can show how these relations institutions examined, an estimated leverage
may change over time. ratio from the model shows that it diverged
from the balance sheet measure of leverage well
Option-iPoD before each institution’s “default event.” This
As noted earlier, and despite their broad use, suggests that an estimate of the implied lever-
analyses based simply on correlations are less age may be one measure that better reflects the
than ideal when dealing with extreme downside risks being undertaken by the firm on a real-
movements, as fat tails tend to develop. Sev- time basis than other accounting-based ratios.
eral models provide a more general approach The models described above still suffer from
by looking at the characteristics of the entire the limitation that they focus on individual FIs
distribution of asset returns. A number of those without addressing how groups of FIs might be
models do this univariately (one firm at the related to one another—the key component for
time). As described in Table 3.2, an impor- systemic risks. The sections below relax those
tant shortcoming of these models is that they constraints by jointly examining groups of FIs.
require the modeler to assume a specific value
of debt, below which the institution will fail.
This assumption is relaxed in the option-iPoD
model as the default-barrier is determined 28Default events are listed in Annex 3.3.

124

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

Box 3.2. Option-iPoD Measures of Risk Across Financial Institutions

This box introduces two new risk indicators based interpreted as a forward-looking measure of
on the prices of equity-options.1 The option-iPoD capital-at-risk, and thus, together with option-
measures the probability of default, while the option- iPoD, might become a useful tool in the super-
leverage measures the likelihood that the leverage vision of financial institutions.
ratio will cross a prespecified threshold. In the current The added value of this methodology resides
crisis, these measures have performed well. in the relaxation of two key assumptions, typically
The methodology estimates the risk-neutral imposed in related structural credit-risk frame-
probability density function of the value of works: a prespecified probability density func-
the assets of an individual institution, which is tion of the value of the assets and a prespecified
used to obtain the probability of default, the default barrier, an assumed value below which
option-iPoD, and the expected development of the firm is expected to default. Following Kull-
balance sheet variables, such as assets, equity, back (1959) and Kullback and Leibler (1951), an
and leverage.2 optimization problem in which the current mar-
The probability density function allows one ket prices of equity-options represent the prob-
to compute the risk-neutral likelihood that lem’s constraints is solved. As a consequence, a
the ratio of the estimated market value of nonparametric density function is obtained that
assets to equity, the option-leverage, will cross a captures the well-documented deviations of asset
prespecified threshold. This likelihood can be prices from log-normality.3

Option-iPoD: An Indication of Impending Failure

(Percentage change with respect to the previous day)

3500 Lehman Brothers, September 15, 2008 Event 6000

Merrill Lynch, September 15, 2008
3000 Wachovia, September 29, 2008 5000
Citigroup, November 24, 2008
2500 Bear Stearns, March 14, 2008 (right scale)
4000
2000
3000
1500
2000
1000
1000
500

0
0

–500 –1000
–34 –32 –30 –28 –26 –24 –22 –20 –18 –16 –14 –12 –10 –8 –6 –4 –2 0 +2 +4
Business days to event

Sources: Bloomberg, L.P.; and IMF staff estimates.

Note: Option-iPoD is the probability of default implied by option prices.

3This type of optimization problem is known as a

minimum cross-entropy problem. Cover and Thomas

(2006) discuss the statistical properties of cross-
entropy, which, in intuitive terms, can be interpreted
Note: Christian Capuano prepared this box. as a measure of relative distance between two prob-
1The methodology is developed in Capuano
ability density functions. Buchen and Kelly (1996)
(2008). discuss a similar framework to extract a probability
2Capuano (2008) describes how to extend the
density function from equity options. Because of
methodology to obtain useful output for risk manage- put-call parity, a well-known no-arbitrage relationship,
ment, such as an estimated credit-spread and the researchers need to specify whether they want to use
so-called Greek letters. call or put prices (or a combination) as constraints.

125

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Box 3.2 (continued)

Option-Leverage: A Forward-Looking Measure of Distress

(LIkelihood option-Leverage > 30)

Lehman Brothers, September 15, 2008 Event 2.0

Merrill Lynch, September 15, 2008
Wachovia, September 29, 2008 1.8
Citigroup, November 24, 2008
1.6
Bear Stearns, March 14, 2008
1.4

1.2

1.0

0.8

0.6

0.4

0.2

0
–34 –32 –30 –28 –26 –24 –22 –20 –18 –16 –14 –12 –10 –8 –6 –4 –2 0 +2 +4

Business days to event

Sources: Bloomberg, L.P.; and IMF staff estimates.

Note: Option-leverage is the ratio of the estimated market value of assets to equity. Likelihood option-leverage > 40 for Bear Stearns and
Lehman Brothers.

The economic structure of the model fol- In order to investigate how this methodol-
lows Merton (1974).4 Most notably, instead of ogy has performed during the current finan-
prespecifying a value for the default barrier— cial crisis, a countdown to the event has been
which is calibrated, in general, to the current constructed—starting 35 business days prior to
value of on-balance-sheet liabilities—a key their collapse—for Bear Stearns, Lehman Broth-
improvement over existing methodologies is ers, Merrill Lynch, Wachovia, and Citigroup.5
to use the linear independence of the option- For this purpose, the PoD implied by the price
price constraints to treat the default barrier as of equity options is estimated by focusing on
a free parameter, and obtain a default barrier the contract whose expiration was the closest to
that is optimally estimated within the model. the day of the event. In addition, after optimally
Since financial institutions carry out exten- estimating the capital structure of the selected
sive off-balance-sheet activities, an optimally institutions, the likelihood that option-leverage
estimated default barrier is particularly attrac- would hit a prespecified threshold by the expira-
tive for financial stability purposes because it tion of the option contract is computed.6
allows one to estimate a market-implied capital
structure, which in times of distress might be 5A robustness check would need to be conducted

expected to significantly differ from the last with an extended sample, including institutions that
reported balance sheet. have not collapsed. In this sample, data availability on
specific option contracts prevents the countdown to
be further extended.
6While the selected thresholds cannot be directly
4In its simplest version, Merton (1974) postulates compared with the Federal Deposit Insurance Cor-
that the value of equity corresponds to the value of a poration Tier 1 leverage ratio, which is based on Tier
call option contract written on the assets of the institu- 1 capital, they nonetheless provide a useful insight
tion, with exercise (strike) price corresponding to the on the current capital structure as perceived by the
institution’s on-balance-sheet liabilities. equity options market.

126

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

In the selected episodes, option-iPoD has This appears particularly true for Bear Stearns
performed well (see figure). On several occa- and Lehman Brothers, suggesting that markets
sions prior to the event, and for all institutions, might have been aware of the significantly weaker
option-iPoD jumped up by a multiplicative factor. liability structure of these investment banks and
Ex post, the pattern of warning signals seems of the associated potential risks. Early during the
to have been particularly informative for Bear countdown, this divergence also became evident
Stearns, Merrill Lynch, and Wachovia, while less for Citigroup and Wachovia.
so for Lehman Brothers and Citigroup. In consideration of the forward-looking
The analysis of the likelihood that option- nature of this methodology, the proposed risk
leverage will cross a specific threshold pro- indicators appear to have been performing well
vides an economic interpretation of these during the current crisis, providing early warn-
events (see figure). During the countdown, ing signals of distress. When complemented
the divergence between the reported balance with other market and nonmarket information,
sheet and the estimated capital structure option-iPoD and option-leverage might become a
of the selected institutions became more useful tool for the daily surveillance of finan-
pronounced. cial and nonfinancial institutions.

Measures of Risk Based on Groupings of sures of dependence, particularly as financial

Financial Institutions markets become more integrated. In addition,
The analysis based on market perceptions standard correlations do not account for the
presented thus far, based on CDS and equity variation over time in the degree of depen-
prices, has been for individual FIs. The sec- dence, especially during episodes characterized
tions that follow address these issues from an by rising uncertainty about asset prices and
aggregate perspective by looking at measures illiquidity of overall financial markets. In times
based on CDS and equity prices for several of stress, illiquid markets sap diversification
groupings of global FIs. While a formal test of opportunities contributing to increased correla-
this dynamic relationship is not performed in tion, making accurate estimates of the impact
this chapter, and is reserved for future work, of higher volatility on asset prices difficult to
the subsections present snapshots of how vari- interpret. For these reasons, the examination of
ous potential measures of systemic risk appear tail dependencies is likely a better choice when
to have coincided during the current crisis. attempting to discern systemic risks.
Finally, the analysis is extended to include Since equity is the most junior contingent
risks in emerging markets, as these countries claim on the future asset performance of fi rms
were viewed by some as being “decoupled” (equity holders are paid last from the fi rm’s
during the earlier part of the crisis. profits), equity derivatives contain forward-
looking information of market participants’
Tail Risks of Financial Institutions Based on perceptions of downside risk. Moreover, the
Equity Options information content of prices has shifted
As noted above, the notion of systemic risk from price levels to higher moments such as
requires moving away from traditional mea- the variance, skewness, and kurtosis over the
sures of correlation between different financial course of the crisis as investors reposition
entities toward nonlinear, time-varying mea- themselves in response to uncertainty and

127

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

information asymmetries (Kim and Verrec-

chia, 1997). Thus, this section uses implied
volatilities from at-the-money equity options
to examine simultaneous co-movements in the
left-hand tails of the equity distribution as a
measure of “tail dependence” and the magni-
tude of systemic risk.29 Implied volatilities can,
in principle, be more revealing of information
pertinent to systemic risks than equity prices
alone. More specifically, the combined prob-
ability of the average co-movement as well as
very large negative shocks to several financial
Figure 3.5. U.S. and European Banks: Joint Tail Risk institutions can be estimated (Box 3.3).
of Implied Volatilities
The examination of multivariate dependence
U.S. banks (left scale) U.S. banks (CDS, right scale)
European banks (left scale) European banks (CDS, right scale) highlights two periods of high systemic risk
Start of U.S. Bear Lehman induced by large tail events—the buildup prior
subprime crisis Stearns Brothers to the subprime fallout (June 2007) and the
0.80 400
largely coincident period associated with the
0.75 350
Average CDS spread, in basis points

collapse of Lehman Brothers (September 2008).

Dependence parameter value

0.70 300 Extreme co-movements of equity prices (Fig-

0.65 250 ure 3.5) did not follow but preceded the bailout
of Bear Stearns. From a visual inspection,
0.60 200
the results also seem to indicate that higher
0.55 150 moments from equity price data may lead
price data on credit-sensitive assets and implied
0.50 100
default probabilities of CDS spreads, though
0.45 50 more thorough analysis will need to be done to
0.40 0 verify this claim.
2006 07 08 These indicators also show that systemic risk
Sources: Bloomberg, L.P.; and IMF staff estimates. has been increasing since February 2007. Aver-
Note: Sample period: 5/18/2005–12/31/2008 (946 obs.) of implied volatility
derived from at-the-money equity put options of three banks in each the United
age dependence among the global sample of
States and Europe. Rolling window (one year) estimation with bi-monthly updating. banks and insurance companies (Core 1 and
The line shows the estimated joint tail dependence (“asymptotic tail behavior”)
based on a nonparametric specification of a trivariate extreme value distribution Core 2) increased by almost 30 percent, while
(logistic model) with a convex dependence function whose upper/lower limits are
derived under complete dependence/independence. U.S. banks = Bank of America,
joint tail risk declined by about the same order
Citigroup, and JPMorgan Chase & Co. European banks = Deutsche Bank, Royal of magnitude (Figure 3.6), indicating that co-
Bank of Scotland, and UBS. CDS = credit default swap.
movements of large changes in equity volatility
occur more frequently. This means that extremes
(and aberrant swings in equity risk) have
become the norm rather than the exception
over the last year. As average dependence con-
tinues to increase above the historical trend, the

29Note that the use of implied volatilities from out-of-

the-money equity put options would be a superior input

variable for our approach. Due to the lack of continuous
prices on non-U.S. banks, we have chosen at-the-money
options instead.

128

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

recent surge of tail risk (from historic lows)—

together with the sharp increase in skewness
and kurtosis—represents elevated systemic risks. Figure 3.6. Higher Moments and Multivariate Dependence of
In sum, these indicators of systemic risk appear Implied Equity Volatility
to have detected rising, and now elevated, risk,
Core 1 Group: Dependence Measures Core 2 Group: Dependence Measures
potentially providing some advance notice for
Spillover risk: Systemic crisis: Spillover risk: Systemic crisis:
policymakers. Only extreme System enters into Only extreme System enters into
shocks translate historic tail area shocks translate historic tail area
into spillovers amid higher average into spillovers amid higher average
Common Distress in the System and Cascade co-movement co-movement
70 75
Effects

Probability of co-movement (in percent)

Probability of co-movement (in percent)

70
This section models the joint distress among
several specific groups of FIs using a slightly dif- Rising average
65
60 Rising average
co-movement,
ferent technical approach than the one above but increased co-movement, 60
differentiation but increased
(Segoviano and Goodhart, 2009). The joint of shocks differentiation
of shocks
55
statistical distribution of the implied asset values 50
50
of a group of FIs—the financial system multivari-
Start of U.S. Start of U.S.
ate density (FSMD)—implicitly characterizes both subprime crisis subprime crisis 45
the individual and joint asset value movements 40 40
of a chosen portfolio of FIs (see Box 3.4).30 The 2006 07 08 2006 07 08

FSMD thus captures interdependence among Entropy-based correlation (average co-movement)

Extreme value dependence (joint tail risk)
the FIs’ distress proxy variable (the probability of
default), which captures the FIs’ linear (correla- Core 1 Group: Higher Moments Core 2 Group: HIgher Moments
(Median values) (Median values)
tions) and nonlinear distress dependence and
their changes throughout the economic cycle, Start of U.S. Lehman Start of U.S. Lehman
subprime crisis collapse subprime crisis collapse
reflecting the fact that dependence increases
2.5 2.5
in periods of distress—a key technical improve-
ment over traditional risk models. Using the joint 2.0 2.0
(multivariate) distribution, other measures of
Parameter value

Parameter value
1.5 1.5
financial stability can be derived: (1) common
distress of the financial institutions in a system; 1.0 1.0
(2) distress between specific institutions; and (3)
distress in the system resulting from distress in a 0.5 0.5

specific institution.31 The three measures repre- 0 0

sent an advantage over the analysis of any single
one of them, since one can identify how risks –0.5 –0.5
2006 07 08 2006 07 08
Skewness
Kurtosis (log-scale)
30The FSMD is recovered using a particular technique, Gamma (tail shape)
the consistent information multivariate density optimiz-
ing (CIMDO) methodology (Segoviano, 2006), which is Sources: Bloomberg L.P.; Datastream; and IMF staff estimates.
Note: Estimates are based on implied volatility derived from at-the-money equity put options.
a nonparametric framework based on the cross-entropy Rolling window (one year) estimation with bi-monthly updating. The gamma parameter represents
approach (Kullback, 1959). the shape parameter of the generalized extreme value distribution, estimated via the linear ratio of
31The second measure—distress between specific spacings method. The higher the tail shape parameter (“gamma”), the greater the univariate tail
risk. The entropy-based correlation coefficient is based on the expected mutual information and
institutions—is analyzed in Chapter 2. These conditional
the joint distribution of individual entropies of each constituent time-series vector. It represents
probabilities, summarized in a distress dependence the nonparametric estimate of general multivariate dependence. In contrast, the nonparametric
matrix, should not only be seen as an indication of estimate of multivariate extreme value dependence represents the joint tail risk of ordered
bilateral stress among FIs, since the overall dependen- maxima. For Core 1 and Core 2 Groups, see Annex 3.2.
cies across the institutions in the sample are included
in the multivariate distribution from which the matrix is
constructed.

129

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Box 3.3. Higher Moments and Multivariate Dependence of Implied Volatilities from Equity Options
as Measures of Systemic Risk

This box describes the use of equity options to evalu- risks if distributions are skewed, it is important
ate the magnitude of systemic risk jointly posed by to use higher moments (derived from individual
financial institutions based on a measure for the joint firms’ equity options) to obtain nonlinear mea-
tail dependence across institutions and their average sures of dependence (Jobst, 2007a). Two models
co-movement. accounting for time-varying dependence are
If firms are leveraged, the seniority of credi- presented: (1) multivariate extreme value depen-
tors implied by the capital structure suggests dence (based on a limit law for joint asymptotic
that equity is the most sensitive contingent tail behavior); and (2) a dependence measure
claim on asset performance. Thus, we would based on “entropy,” which is a measure disper-
expect equity prices in cash and derivatives sion. While the former measures changes of
markets to reflect even small changes in expec- joint tail risk, the latter delivers a nonparametric
tations of default risk.1 This becomes even more estimate of general multivariate dependence.
important during times of stress, when the abil- First, a nonparametric measure of joint tail
ity to use options as forward looking measures dependence based on multivariate extreme
to hedge the downside risk of equity is more value theory is defined in order to quantify the
valuable (Gray and Jobst, forthcoming). possibility of common extreme shocks (Coles,
Recent research finds that if the volatility of Heffernan, and Tawn, 1999; Poon, Rockinger,
equity prices is negatively skewed (left-tailed), so and Tawn, 2004; Stephenson, 2003; and Jobst,
are the implied underlying asset distributions, 2007b). As an integral part of this approach,
which in turn are related to default risk (see this dependence structure links the univariate
Box 3.1). Thus, higher moments of equity price marginal distributions in a way that formally
dynamics better account for nonlinearities of captures joint asymptotic tail behavior. Using
changes in default risk if large risk exposures the empirical distribution avoids problems
become more frequent than suggested by the associated with modeling specific parameters
assumption of normal distributions. This means that may or may not fit these distributions
that accounting for higher moments of equity well—a problem potentially exacerbated dur-
options can deliver important insights about sig- ing stressful periods.2 This method of mea-
nificant changes in asset values of firms, which, suring “tail dependence” is better suited to
in the presence of fat tails, results in a higher analyzing extreme linkages of multiple entities
probability of default, and thus, higher spreads than the traditional (pairwise) correlation-
(Zou, 2003). Fat tails would indicate that market based approach.
perception of severe downside equity risk has Second, average dependence in the multi-
increased, and estimating economic capital variate case based on the concept of entropy is
based on volatility alone becomes unreliable,
upsetting the basic tenets of the risk-based regu-
latory framework. 2This approach is distinct from previous studies

Since the concept of conventional correlation of joint patterns of extreme behavior. For instance,
can give misleading information about systemic Longin (2000) derives point estimates of the extreme
marginal distribution of a portfolio of assets based on
the simple correlation between the series of individual
maxima and minima. However, in the absence of a
Note: Dale Gray and Andy Jobst prepared this box. principled standard definition of order in a high-
1Since the capital structure of firms establishes a dimensional vectorial space, the simple aggregation of
natural linkage between the cost of insuring against marginal extremes (without considering a depen-
default risk (via credit default swap spreads), on one dence structure) does not necessarily concur with the
hand, and claims on future earnings (via equity), on joint distribution of the extreme marginal distribu-
the other, changes in expectations of future firm per- tions. See also Embrechts, Lindskog, and McNeil
formance influence the market values of both. (2003) regarding this issue.

130

©International Monetary Fund. Not for Redistribution

 MARKET PERCEPTIONS OF RISK OF FINANCIAL INSTITUTIONS

investigated. Since the entropy of a set of vari- can then be computed based on the reciprocal
ables is maximized if observed data are uniformly of the marginal contribution of each univariate
distributed, minimizing joint entropy indicates entropy to the expected mutual information and
the maximum degree of dependence. In order analyzed. This method is suitable to extend the
to derive an overall measure of dependence concept of “average dependence” to the multi-
between several variables (called “expected variate case.
mutual information”), the effects of lower depen- In the chapter, both models are applied to
dences are eliminated from the sum of both the the implied volatilities of at-the-money equity
overall entropy and the individual entropy of put options of all financial institutions in our
each financial institution’s univariate marginal samples (Core 1 and 2). Our main findings
distribution by subtracting all joint entropies that confirm that both models yield complementary
do not include all variables (Preuss, 1980; and findings that provide comprehensive and timely
Theil, 1969). A scaled entropy-based measure information about the magnitude of systemic
of dependence (called “entropy correlation”) risk and possible developments going forward.

are evolving and which groups of institutions or The results indicate that distress in one FI is
a single institution may suffer from the distress associated with a high probability of distress else-
of another. This methodology can be flexibly where. Moreover, movements in the JPoD and
implemented, since the PoDs of individual FIs BSI coincide with events that were considered
represent the input variables, which can be by the markets to be particularly disruptive on
estimated using alternative approaches. Although specific dates (Figure 3.7). Risks also vary by the
in this exercise we used PoDs derived from CDS geographical location and business line of the
spreads, it would be straightforward to replace FI in the various groups (Figure 3.8). Distress
these input variables. This approach is also used dependence across FIs rises during times of
to analyze the joint risks across banks in advanced crisis, indicating that systemic risks, as implied
economies and emerging market sovereigns for by the JPoD and the BSI, can rise faster than
countries where such banks have large exposures idiosyncratic (individual) risks. Figure 3.9 shows
(see Annex 1.3 in Chapter 1). that this is the case—daily percentage changes
Common distress in the system: JPoD and BSI. of the JPoD are larger than daily percentage
Two variables are employed to analyze common changes of the average of individual PoDs. This
distress: the JPoD, and the BSI. These show larger empirical fact provides evidence that in times of
and nonlinear increases in distress for groups of distress, not only do individual PoDs increase,
FIs than for the individual component FIs.32 Esti- but so does distress dependence. Therefore,
mations of the JPoD and the BSI are performed measures of financial stability that are based on
from January 1, 2005 to December 31, 2008 and averages or indices could be misleading.
include major U.S., European, and Asian banks, Cascade effects. Another use of the joint prob-
which were grouped in alternative ways in Annex ability distribution is the probability of cascade
3.2. The JPoD variable measures the joint probabil- effects, which examines the likelihood that one
ity of distress of all the institutions in the sample, or more FIs in the system become distressed
and the BSI measures the expected number of given that a specific FI becomes distressed. It is a
other institutions that would fall into distress if a useful indicator to quantify the systemic impor-
specific institution were to default. tance of a specific FI, since it provides a direct
measure of its effect on the system as a whole.
32See Segoviano and Goodhart (2009) for definitions. As an example, the probability of cascade effects

131

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Figure 3.8. Joint Probability of Distress and

Banking Stability Index: By Geographic Region

Bank stability index (number of Fls, left scale)

Joint probability of distress (right scale)
2.5 Asia 0.030

0.025
2.0

0.020
1.5
0.015
1.0
0.010

0.5
0.005

0 0
Figure 3.7. Joint Probability of Distress (JPoD) and 2005 06 07 08
Banking Stability Index (BSI): Core 2 Group 3.0 Euro Area 0.20
3.5 1 2 34 0.025
2.5
0.15
3.0
0.020 2.0

2.5
1.5 0.10
0.015
2.0 1.0
BSI
0.05
1.5 (Number of FIs, left scale)
0.010 0.5

1.0 0 0
2005 06 07 08
JPoD 0.005
0.5 (Probability of default, percent, right scale) 3.0 0.35
Non-Euro Area

0 0 2.5 0.30
2005 06 07 08
0.25
2.0
Sources: Bloomberg L.P.; and IMF staff estimates.
Note: FIs = financial institutions. TARP = Troubled Assets Relief Program. For 0.20
Core 2 Group, see Annex 3.2. 1.5
Events: 0.15
1. Bear Stearns episode (3/11/08) 1.0
2. Lehman bankruptcy and AIG bailout (9/15-16/08) 0.10
3. TARP bill failure (9/30/08)
4. Global central bank intervention (10/8/08) 0.5 0.05

0 0
2005 06 07 08
3.0 United States 0.12

2.5 0.10

2.0 0.08

1.5 0.06

1.0 0.04

0.5 0.02

0 0
2005 06 07 08

Sources: Bloomberg L.P.; and IMF staff estimates.

Note: For financial institutions (FIs) in each region, see Annex 3.2.

132

©International Monetary Fund. Not for Redistribution

 IDENTIFYING SYSTEMIC RISKS THROUGH REGIME SHIFTS

is estimated given that Lehman or AIG became

distressed. These probabilities reached 97 and
95 percent, respectively, on September 12, 2008,
signaling a possible “domino” effect in the days
after Lehman’s collapse (Figure 3.10). Note that
the probability of cascade effects for both insti-
tutions had already increased by August 2007,
well before Lehman collapsed. Figure 3.9. Daily Percentage Change: Joint and
Average Probability of Distress, Core 2 Group
3.0
JPoD Core 2
Identifying Systemic Risks Through Average Core 2
2.5
Regime Shifts
The next objective is to examine when the 2.0

JPoD and the BSI, as aggregate measures of FIs’ 1.5

stability, switch from low- and medium-volatility
1.0
regimes into a high one, and vice-versa (Hesse
and Segoviano, forthcoming). Remaining in the 0.5

high-volatility regime could indicate that the 0

crisis has become systemic. From this perspec-
–0.5
tive, the BSI is of particular interest, in that it
measures the expected number of distressed –1.0
2005 06 07 08
institutions given that at least one institution
becomes distressed. Sources: Bloomberg L.P.; and IMF staff estimates.
The univariate Markov-Switching autoregres- Note: JPoD = Joint probability of distress. For Core 2 institutions, see Annex 3.2.

sive conditional heteroskedacticity (SWARCH)

Figure 3.10. Probability of Cascade Effects
model developed by Hamilton and Susmel
1.0
(1994) is used.33 The models are based on
daily data in first differences from January 1,
2006 to December 31, 2008. Figure 3.11 (first 0.8

panel) shows the SWARCH model using the BSI Lehman Brothers
measure for the Core 1 group of banks (United 0.6
States, Europe, and Asia) and the probability
of being in the high-volatility state. The results AIG
0.4
show the following:
• After the beginning of the subprime crisis, the
0.2
model only oscillates between the high and
medium states, while the precrisis period was
characterized by a low-volatility regime. 0
2007 08
• The model enters the high-volatility state in
late July 2007—the beginning of the subprime Sources: Bloomberg L.P.; and IMF staff estimates.

crisis—and the variations into and out of this

33This model allows for a time-varying variance and

state-dependent ARCH parameters—features that are

present in the types of financial data underpinning the
BSI and JPoD. Moreover, the technique allows the data to
determine the transition across the regimes rather than
the researcher making an ad hoc determination.

133

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Box 3.4. The Consistent Information Multivariate Density Optimizing Approach

This box provides details about how the assess their estimates of the PoD. Our analysis
financial system multivariate density (FSMD) shows that while no approach is free of issues,
is obtained from the data, demonstrating the the CDS-PoDs appear to be a good distress
advantages of the consistent information multi- signal. For this reason, the FSMD in this paper
variate density optimizing (CIMDO) technique uses CDS-PoDs. However, further statistical
relative to other more traditional ones. analysis to improve the estimation of individual
The FSMD embeds the banks’ distress PoDs is ongoing. Thus, if a better approach is
dependence structure, characterized by the found, it is straightforward to replace the cho-
CIMDO-copula function (Segoviano, forthcom- sen PoDs with another set.
ing), which captures linear (correlations) and The CIMDO starts with a formal, parameter-
nonlinear distress dependence among the finan- ized distribution of the financial institutions’
cial institutions in the system, and their changes input data (a prior) and then arrives at a
throughout the economic cycle, reflecting the final distribution (the posterior) by impos-
fact that dependence increases in periods of ing constraints that assure that the overall
distress. These are key technical improvements multivariate distribution contains marginal
over traditional risk models, which usually probability densities that satisfy the constraints
account only for linear dependence that is associated with the PoDs of each of the constitu-
assumed to remain constant over the cycle or a ent financial institutions. CIMDO-recovered
fixed period of time.1 distributions outperform the most commonly
Empirically, the CIMDO methodology is a used parametric multivariate densities in the
tool to recover the FSMD and hence to acquire modeling of portfolio risk under the probability
the joint relationships across the individual integral transformation criterion (a measure
financial institutions at the portfolio level. As of how well densities approximate the under-
such, it requires as inputs (exogenous vari- lying data). This is because when recovering
ables), measures of the probabilities of default multivariate distributions through the CIMDO
(PoDs) of individual financial institutions that approach, the available information embedded
represent the financial system, which can be in the constraints is used to adjust the “shape”
estimated using alternative approaches; for of the multivariate density. This appears to
example, the structural approach, option prices allow the distribution to more closely adapt to
and credit default swap (CDS) spreads. The the changes in entire distribution over time,
underlying data for use in the CIMDO approach but particularly in the tail of the distribution,
are important, as the results are a reflection relative to other approaches, which adjust the
of the input data. Athanasopoulou, Segoviano, “shape” of parametric distributions via fixed sets
and Tieman (forthcoming) present an extensive of parameters.
empirical analysis of different versions of the Once the CIMDO density is estimated, its
structural approach and the CDS approaches to copula function is recovered. Note that this is
an inverse approach to the standard copula
modeling, which first chooses and parameterizes
Note: Miguel Segoviano prepared this box. the copula function and then “couples” margin-
1Segoviano (forthcoming) shows that the structural
als to define multivariate densities. Indeed, the
approach produces, at times, estimates that appear
inconsistent with actual default probabilities due standard approach to model parametric copula
to problems related to lack of liquidity in certain functions is difficult to implement, since model-
markets and generalized risk aversion in times of ers have to deal with the choice, proper specifi-
distress. Credit default swaps-probabilities of default cation, and calibration of the copula functions.
also appeared to be affected by these problems, and
In contrast, the CIMDO methodology does not
at times they overshoot. However, although the magni-
tude of the moves may occasionally be unrealistic, the require the modeler to choose ex ante a copula
direction is usually a good distress signal. function to define distress dependence; that is,

134

©International Monetary Fund. Not for Redistribution

 ROLE OF GLOBAL MARKET CONDITIONS DURING EPISODES OF STRESS

the form of the copula function is defined by acterized by the CIMDO-copula appears to be
the data. Thus, the CIMDO-copula provides key more robust in the tail of the density, where
improvements and avoids drawbacks implied our main interest lies, that is, to characterize
by the use of standard parametric copulas as it tail risk dependence.
incorporates, endogenously, changes in distress By recovering the FSMD, which embeds
dependence and avoids the imposition of con- financial institutions’ distress dependence, Sego-
stant correlation parameters. viano and Goodhart (2009) can produce three
However, the CIMDO-copula maintains the measures that allow policymakers to examine
benefits of the copula approach to model different aspects of systemic risk. This permits
dependence: first, it describes linear and policymakers to identify not only how com-
nonlinear dependencies among the variables mon risks are evolving, but also where distress
described by the CIMDO-density; and second, might most easily develop and how distress in a
it characterizes the dependence structure specific institution can affect other institutions,
along the entire domain of the CIMDO-density. thus enabling them to make an assessment of
Nevertheless, the dependence structure char- the stability of the financial system.

state are mostly coincident with the periods in Bear Stearns rescue, and the Lehman episode
which there are large central bank interven- suggest that the financial system had entered a
tions and new policy initiatives, and unsurpris- systemic crisis, while until Lehman’s collapse,
ingly, the Lehman closure. many commentators thought the crisis was
• In two cases of the five variations exam- contained. Of course, this method should not
ined (Figure 3.11, panels 2 and 3) there be used in isolation but be complemented by
is a movement into the high-volatility state other systemic risk indicators. While the JPoD
in late February 2007. As discussed before, and BSI indicators measure different attributes
this corresponds to the sharp Shanghai of systemic risk, i.e., the joint probability of
stock market correction as well as the first distress versus the conditional expectation of
abrupt ABX (BBB) price decline of subprime distress probability, it is reassuring that the
mortgages.34 main crisis events are picked up by both data
• There are some differences in 2008 between series. For some of the events studied, notably
U.S. investment banks and European banks the February 2007 episode, the threshold of
(Figure 3.11, panels 4 and 5). The latter volatility only stays in the high mode for a short
appear to be in the high-volatility state most period of time, making it difficult, ex ante,
of the time, which could be explained by the to tell whether the financial system was going
higher variance of their BSI. to remain in this elevated volatility state and
Overall, the SWARCH models are useful whether it had thus entered a systemic crisis.
analytical tools to discern when aggregate mea-
sure of FIs’ stability (in this case, the BSI and
JPoD) switch volatility regimes. Persistent high- Role of Global Market Conditions During
volatility states such as the first months of the Episodes of Stress
subprime crisis, the months surrounding the This section examines how various proxies
for global market conditions can influence the
34These two events were roughly coincident. While it is
incidence of systemic risk.35 As noted above,
difficult to prove whether they were related events, they
appear be consistent with the rebalancing portfolios by
investors with high-yield positions. 35See González-Hermosillo and Hesse (forthcoming).

135

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Figure 3.11. Markov-Regime Switching ARCH Model: the value of assets on the books of FIs are
Joint Probability of Distress and Banking Stability Index highly dependent on the underlying financial
environment—such factors as the interest rate
BSI: Core 1 (percent change, left scale)
Probability of being in high-volatility state (right scale) environment (low or high) or the level of risk
15 1.0
appetite—and, as such, global market conditions
10 0.8 are thus important in determining their market
5 0.6 value and ultimately the strength or weakness
of financial institutions and the probability of a
0 0.4
systemic episode.
–5 0.2

–10 0
2006 07 08
Markov-Regime Switching Analysis
JPoD: Core 1 (percent change, left scale)
Probability of being in high-volatility state (right scale) 1.0 Markov-regime switching techniques take an
350
300 integrated approach to analyzing financial stress.
250 0.8
200 The SWARCH model of Hamilton and Susmel
0.6
150 (1994) is particularly well-suited for the purpose
100 0.4
50
since it differentiates between different volatil-
0 0.2 ity states (e.g., low, medium, and high), derived
–50 from the time-varying nature of volatility that
–100 0
2006 07 08 occurs in many high-frequency financial vari-
BSI: Core 2 (percent change, left scale)
Probability of being in high-volatility state (right scale)
ables, particularly during times of stress.36
15 1.0
A SWARCH model of the euro-U.S. dollar forex
10 0.8 swap reveals that the variable moves from a low- to
5
0.6 a medium-volatility regime in the beginning of
0
–5 0.4
August 2007 before entering the high-volatility
–10 state right after the Lehman collapse in September
0.2
–15 2008, remaining there until the end of November
–20
2006 07 08
0 2008 (Figure 3.12). Many non-U.S. banks, espe-
BSI: Europe (percent change, left scale) cially European ones, faced a shortage of U.S.
20
Probability of being in high-volatility state (right scale) 1.0 dollar funding for their conduits and structured
15 investment vehicles from the summer of 2007
0.8
10 onward. As the interbank market for dollar fund-
0.6
5 ing dried up due to heightened counterparty and
0 0.4 liquidity risks, these banks increasingly engaged
–5 in foreign exchange swap arrangements (Baba,
0.2
–10
Packer, and Nagano, 2008), leading to higher
–15 0
2006 07 08 volatility.37 The move of the forex swap into the
BSI: U.S. investment banks (percent change, left scale)
Probability of being in high-volatility state (right scale)
12 1.0
36Univariate SWARCH models are adopted here with

8 0.8 variables in first differences to account for the nonsta-

tionarity of the variables. The mean equation is an AR(1)
4 0.6
process and the variance is time-varying with the ARCH
0 0.4 parameters being state dependent.
37In particular, both euro and sterling were used as

–4 0.2 the funding currencies for the dollar foreign exchange

swaps. The spillovers from the interbank market to the
–8 0
2006 07 08 foreign exchange swap market led to a situation whereby
Sources: Bloomberg L.P.; and IMF staff estimates.
foreign exchange swap prices deviated from that implied
Note: JPoD = joint probability of distress; BSI = banking stability index. For Core by covered interest parity conditions. With the turbu-
1 and Core 2 groups, see Annex 3.2.

136

©International Monetary Fund. Not for Redistribution

 ROLE OF GLOBAL CONDITIONS DURING EPISODES OF STRESS

high-volatility state on September 15, 2008 reflects

the sharp increase in counterparty risk after the
Lehman failure, a sizable dollar shortage with mar-
gins and haircuts increasing across the board, and
Figure 3.12. Euro-Dollar Forex Swap
the breakdown of the LIBOR market.
Turning to the VIX, Figure 3.13 shows the Probability of being in low-volatility state (left scale)
Probability of being in medium-volatility state (left scale)
results of a daily SWARCH model from 1998 Probability of being in high-volatility state (left scale)
to end-2008.38 The probability of being in the Euro-U.S. dollar forex swap (basis point change, right scale)
1.0 150
high-volatility state varies considerably, spiking Beginning of Lehman’s failure
during previously identified episodes of instabil- subprime crisis
100
0.8
ity. Indeed, the findings show the switch to the
high-volatility regime in late February 2007 when 50
the Chinese stock market corrected sharply and 0.6

the first round of ABX (BBB) price declines 0

occurred, suggesting a potential warning sign 0.4
of systemic fragilities. The Lehman event then –50

triggered a rapid movement of the VIX into the 0.2

–100
high-volatility regime, where it remained until the
end of the sample period. Since the beginning of
0 –150
the subprime crisis, the VIX has only oscillated 2006 07 08
between the medium- and high-volatility regimes,
Sources: Bloomberg, L.P.; JPMorgan Chase & Co.; and IMF staff estimates.
in contrast to the predominantly low-volatility
regime predominant during 2003–07.
The SWARCH model is also estimated for the
Figure 3.13. Markov-Switching ARCH Model of VIX
three-month TED spread (Figure 3.14).39 This indi-
cator of short-term bank credit risk moved decid- Absolute change in VIX (left scale)
Probability of being in high-volatility state (right scale)
edly into a high-volatility regime during the summer Russian’s default WorldCom and Beginning of
of 2007 and persisted there for much of 2008. and LTCM Brazil’s election subprime crisis Lehman
25 1.0
Several of the measures examined (the VIX Turkey
20 crisis
index and the TED spread) also pick up other Bear Stearns 0.8
periods of stress in global financial markets, such 15

as Russia’s default and Long-Term Capital Man- 10

9/11
Shanghai
stock market 0.6
agement crisis in August/September 1998, the 5 correction

0
0.4
lence becoming more persistent, many non-U.S. financial –5
institutions also increasingly engaged in the longer-term Dot-com
foreign exchange swaps. This episode especially high- –10 bubble 0.2
burst
lighted the international interconnectedness of banks’ –15
funding requirements through foreign exchange swap
markets and their impaired liquidity. –20 0
38The VIX, the Chicago Board Options Exchange 1998 99 2000 01 02 03 04 05 06 07 08
volatility index, is a measure of the implied volatility
of S&P 500 index options over the next 30 days and Sources: Bloomberg, L.P.; and IMF staff estimates.
Note: ARCH = autoregressive conditional heteroskedasticity; LTCM = Long-Term
calculated from a weighted average of option prices. The Capital Management; VIX = Chicago Board Options Exchange volatility index.
model based on VIX is estimated in first differences due
to nonstationarity. This suggests that it may be useful to
examine higher than second moments in the probability
density function.
39The TED spread is the difference between the three-

month LIBOR and the three-month treasury bill rate.

137

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

liquidity shock of 9/11, and other episodes of

crisis in emerging markets as well as the dot-com
bubble and the WorldCom scandal.40 While the
Figure 3.14. Markov-Switching ARCH Model of TED recent persistence of the high-volatility period
Spread for the TED spread is unprecedented over the
past decade, that for the VIX is not, suggesting
Change in TED spread (basis points, left scale)
Probability of being in high-volatility state (right scale) a greater relative stress in credit markets during
Russian’s default Lehman this crisis episode.
100
and LTCM crisis Beginning of Brothers failure 1.0 The analysis is extended to include the
9/11
subprime crisis
80 interaction of risks with emerging markets
60 0.8 that, as discussed in Chapter 1, have been a
40
key link during the latter stages of the crisis.
In particular, the interconnection between
20 0.6
financial markets in advanced economies and
0
emerging markets is examined in Box 3.5. The
–20 0.4
results show that problems in advanced econo-
–40 mies readily spilled over into emerging markets
–60 0.2 as investors sought the safest and most liquid
–80 global assets. Similarly, an extension of the
–100 0 approach in Box 3.4 is used to examine cross-
1998 99 2000 01 02 03 04 05 06 07 08
country vulnerabilities between emerging mar-
Sources: Bloomberg, L.P.; and IMF staff estimates. ket sovereigns and specific banks in advanced
Note: ARCH = autoregressive conditional heteroskedasticity; LTCM = Long-Term economies with a large regional presence in
Capital Management; TED = the spread between the three-month LIBOR and
treasury bill rates. those countries (see Annex 1.3 in Chapter 1),
finding such spillovers increased dramatically
Figure 3.15. Markov-Switching ARCH Model of VIX, throughout the crisis.
TED Spread, and Core 2 Banking Stability Index While not integrated with the measures in
(Probability of being in a high-volatility state)
the sections above, the regime-shifting model
VIX TED Core 2 BSI can add to the assessment of systemic risks by
Shanghai stock Bear Stearns
1.0 overlaying the results to see if multiple mea-
market correction
sures demonstrate high levels of volatility simul-
Lehman taneously (Figure 3.15). The results show that
Beginning event 0.8
of the global market indicators examined here
subprime
crisis sometimes do not remain in the high-volatility
0.6
state for long, with some exceptions such as the
TED spread. This suggests they should be used
0.4 in combination with other tools to help policy-
makers detect systemic crises.
0.2

0
2007 2008

Sources: Bloomberg L.P.; and IMF staff estimates.

40Robustness tests were performed by estimating the
Note: BSI = banking stability index; VIX = Chicago Board Options Exchange
volatility index; TED = the spread between the three-month LIBOR and treasury bill model prior to the Lehman collapse. It also signaled a
rates. For Core 2 group, see Annex 3.2.
high probability of being in a high-volatility state over this
period. It is worth noting that several relevant data series
(such as CDS) did not exist prior to the early 2000s.

138

©International Monetary Fund. Not for Redistribution

 ROLE OF GLOBAL CONDITIONS DURING EPSIODES OF STRESS

Box 3.5. Spillovers to Emerging Markets: A Multivariate GARCH Analysis

This box examines the financial interlinkages between between these variables. Similarly, according to
advanced and emerging market countries during the the second panel of the figure, the relationship
financial crisis. between the S&P 500 and the EMBI+ regional
Although standard correlations are typically bond spreads encounters a potential break
flawed methods of examining spillovers and the during the Chinese episode, then correlations
potential for systemic risks to spread, a dynamic increase from the beginning of the subprime
conditional correlation (DCC) generalized crisis and reach their peak after the Lehman
autoregressive conditional heteroskedasticity failure. In terms of regional differences, it
(GARCH) model by Engle (2002) can be used to appears that the magnitude of co-movements
avoid many of the pitfalls.1 To examine the inter- between the S&P 500 and the EMBI spread for
linkages between advanced and emerging market Latin American countries dominates the other
countries, the model is applied for the sample regional spreads.
period 2003–08 (Frank and Hesse, forthcom- The third and fourth panels of the figure
ing). A few pertinent variables are used in order examine possible individual country interlink-
to analyze the co-movements: the three-month ages. The LIBOR spread is related to sovereign
U.S. LIBOR-OIS (overnight index swap) spread, bond and sovereign CDS spreads of the emerg-
proxying for funding liquidity and general stress ing market countries of Brazil, Russia, and Tur-
in the interbank market segment; the S&P 500 key. As before, the Chinese episode in February
as well as bond spreads; and stock market and 2007 is evident and so are the subprime and
credit default swap (CDS) measures for some the Lehman collapse in increasing correlation
selected emerging market countries or indices. magnitude order. The Bear Stearns rescue in
The findings suggest that implied correlations March 2008 also becomes visible, with co-move-
between the LIBOR spread and Emerging Mar- ments sharply reversing their downward trend
kets Bond Index Plus (EMBI+) bond spreads of prior to that.
Asian, European, and Latin American countries Overall, the findings from the DCC GARCH
sharply increase after the subprime crisis (see models indicate that the notion of possible
first panel of figure). In addition, the Chinese decoupling (in the financial markets) had been
stock market correction in February 2007 led to misplaced. It is true that emerging market stock
a temporary spike of the correlation measures markets reached their peak in November 2007
from 0.20 to almost 0.50. The Lehman collapse and later, but interlinkages between funding
caused the largest increase of co-movements stress and equity markets in advanced econo-
mies and emerging market financial indicators
Note: Heiko Hesse prepared this box. were highly correlated and have seen sharp
1The variables in the daily DCC multivariate
increases during specific crisis moments. Given
GARCH framework are in first differences to account
the interconnectedness of global financial mar-
for nonstationarity during the crisis period. In addi-
tion, the S&P 500 is included in order to account for kets, investors’ increase in global risk aversion
common shocks. The models are extended to account from problems in advanced economies rapidly
for explicit structural breaks using Capiello, Engle, spilled over into emerging market countries, as
and Sheppard (2006). Using the same methodology, investors sought to pull out from those coun-
Frank, González-Hermosillo, and Hesse (2008) exam-
ine the transmission of liquidity spillovers across asset
tries and only invest into the safest and most
markets in the United States during the subprime liquid assets in their home countries such as
crisis. government bonds.

139

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Implied Correlations from Dynamic Conditional Correlation Model

Spread between 3-month U.S. dollar LIBOR and overnight index swap (OIS) and Emerging Market Sovereign Debt (EMBI+)
Latin America 0.6
Asia
Europe 0.5

0.4

0.3

0.2

0.1

–0.1
2003 04 05 06 07 08
S&P 500 and Emerging Market Sovereign Debt (EMBI+)
Latin America 0
Asia
–0.1
Europe
–0.2

–0.3

–0.4

–0.5

–0.6

–0.7
2003 04 05 06 07 08

Spread between 3-month U.S. dollar LIBOR and OIS and Emerging Market Sovereign Debt (EMBI+)
Brazil 0.6
Russia 0.5
Turkey
0.4
0.3
0.2
0.1
0
–0.1

–0.2
2003 04 05 06 07 08

Spread between 3-month U.S. dollar LIBOR and OIS and Sovereign Credit Default Swaps
Brazil 0.6
Russia 0.5
Turkey
0.4
0.3
0.2
0.1
0
–0.1

–0.2
2003 04 05 06 07 08

Sources: Bloomberg, L.P.; and IMF staff estimates.

140

©International Monetary Fund. Not for Redistribution

 POLICY IMPLICATIONS

Policy Implications have been unanticipated due to off-balance-

sheet exposures and lenders’ dependence
For those responsible for safeguarding finan- on wholesale funding. Indeed, many “failed”
cial stability, monitoring measures of systemic institutions still met regulatory minimum
stress is now critical. This crisis has highlighted capital requirements. However, FSIs are still
the dangers of focusing supervisory practices helpful in assessing individual and systemic
and risk management simply on ensuring that vulnerabilities when reliable market data may
individual institutions are adequately capital- not be available—particularly in less-developed
ized and capable of surviving reasonable stress financial markets—as they can provide both an
events. The current crisis has demonstrated that indication of rising vulnerabilities and a check
a systemic approach is now urgently needed, when other information reveals weaknesses.
since complex financial systems can potentially For countries with more sophisticated sources
amplify the actions of single firms to a degree of information, FSIs could be usefully reevalu-
that can have damaging collective effects. ated, perhaps refocusing them on basic lever-
Indeed, a seemingly well-capitalized and liquid age ratios and ROA as a proxy for risk-taking.
institution can nevertheless become distressed Of course, FSIs should be complemented by
through the actions of its peers, a “run” by other measures and systemic stress tests, and be
wholesale creditors, or even contagious declines broadened to better capture off-balance-sheet
of equity values. exposures and liquidity mismatches.
The issue now facing authorities is not whether
to attempt to identify systemic risks, but how
best to do so in an interconnected global finan- Market-Based Indicators
cial system with incomplete information. This Low equity volatility and tight credit and CDS
chapter has reviewed and developed both bal- spreads were symptoms of, and contributors to,
ance sheet and market-based indicators to assess strong risk appetite prior to February 2007. As
the degree to which they gave some degree such, indicators derived from market data gener-
of forewarning of either a particular institu- ally provided coincident, rather than forward-
tion’s impending failure, or of severe knock- looking, indications of the break in sentiment
on effects. Some of the advanced techniques and transition to a systemic crisis. However,
presented here are new and therefore more some measures illustrated above (Table 3.5)
analysis is needed before a definitive judgment are successful in providing an indication of
as to the optimal set of measures can be made. how vulnerable a group of FIs is to the default
Indeed, given the complexity of the nature of of any one FI, and hence provide some signal
systemic risks, it would be prudent to use vari- of how “systemic” an individual default can be.
ous techniques and measures in order to arrive Such indicators complement those showing the
at robust results. A number of recommenda- degree of interconnectedness among FIs
tions flow from the results. (Chapter 2).
Moreover, some indicators, especially those
derived from implied volatility from equity
Financial Soundness Indicators options, seem to have given more reliable for-
Mixed results were found regarding the stan- ward signals of impending banking system and
dard FSIs’ ability to highlight those firms that individual institution stress (see Figure 3.10).
proved to be vulnerable. Basic leverage ratios Nevertheless, these signs of increasing implied
were most reliable, while capital-to-asset ratios volatility provided only a few months’ notice
(including risk-adjusted ratios) and nonper- that systemic risks were rising, and further work
forming loan data proved of little predictive is needed to confirm that such forewarnings
power. In the current crisis, key vulnerabilities were timelier than CDS spreads.

141

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Table 3.5. Summary of Various Methodologies: Limitations and Policy Implications

Weaknesses/Conditions When Measure
May Be Misleading Policy Implications
Accounting balance sheet When nonlinearity likely; feedback effects Should include indicators on leverage and stock
present; forward-looking requirements; market performance for individual financial
high-frequency; multiple-institutions. institutions.
Conditional correlation matrices When nonlinearity likely. Help policymakers gauge the co-movements and
and cluster analysis interconnections among financial institutions on a
frequent basis.
Option-iPoD When equity-options are not available; Help policymakers monitor default-risk and
subject to distortions from government the distance to specific leverage thresholds of
injections of capital. individual financial institutions at a daily frequency.
Can be used to perform stress tests.
Higher moments and Variations in data frequency and estimation Provide policymakers with an indication of both
multivariate dependence window might require adjustments to the nonlinear and time-varying linkages between
calibration algorithm of tail dependence financial institutions at different magnitudes of
when extremes are rare. common shocks.
Multivariate time-varying Depends on the inputs used in the Provide policymakers with information to identify
distress dependence methodology. If credit default swap used, not only how common risks are evolving, but
subject to distortions from government where spillovers might most easily develop and
guarantees. how distress in a specific institution can affect
other institutions.
Markov-regime switching Does not accommodate multivariate Provides useful information about status of
settings. systemic risk when certain variables (e.g., bank
stability indicators or global market variables),
change their volatility (or mean) states. The
techniques are readily available and could be
updated on a frequent basis.
DCC GARCH models Cannot make causal statements and does Can help policymakers to gauge the extent of
not elucidate feedback effects. co-movements between domestic and global
(foreign) market conditions in normal as well as
stressful periods.
Source: IMF staff.

Volatility Regime Indicators • Collect and publish more, relevant data. While
publicly available market indicators for FIs
There is also evidence that observing shifts in (equity and options prices, CDS spreads)
volatility regimes can be helpful in detecting the can yield useful indicators of systemic stress,
degree to which the financial system is suffering a alternative signals are probably being missed
systemic event. However, in some cases this signal because other relevant data are not being
proves to be relatively short-lived. Nonetheless, collected or published by supervisors in a
regime-switching indicators can show moves to systematic fashion. Most notably, bank FSIs
medium- and high-volatility states and hence can would become more useful with the inclusion
be used to assess the degree of current fragility of off-balance-sheet exposures in a standard-
and uncertainty. Such indicators may also be use- ized manner; the state of market liquidity
ful in establishing whether and when a systemic could be assessed more easily with the publica-
crisis is subsiding, particularly if the low-volatility tion of volumes and bid-ask spreads in credit
state persists, and thus when the withdrawal sup- markets; and systemic interconnections could
portive crisis measures can be safely considered. be properly assessed through the collection
and aggregation of individual cross-border
counterparty exposures. Overall levels of lever-
Policy Messages
age—potentially including for hedge funds—
The findings in this chapter point to a num- would provide information on the potential
ber of broad policy messages: vulnerability of a financial system to shocks.

142

©International Monetary Fund. Not for Redistribution

 CONCLUSIONS

• Diversify information sources and have a compre- • Charge for contributions to systemic risk through
hensive plan in place for systemic events. Some higher capital requirements. Some of the analysis
market-based indicators—using higher presented here allows for the calibration of
moments of FIs’ equity prices—did give a the contribution of individual institutions to
few months’ notice of rising systemic risks systemic risk, providing a starting point for
prior to July 2007. However, it would have additional regulatory capital to be required
been difficult to know at the time whether to penalize practices that add to systemic
these signals were prescient. In general, poli- risk giving due attention to potential procy-
cymakers should not depend on receiving clicality. In addition, indicators of distress
unambiguous signals of impending systemic could also be used to adduce the appropriate
crisis from market prices, and they should perimeter of regulation, or intensity of super-
be complemented with other indicators of vision, thereby allowing institutions whose
potential stress (including FSIs and macro- failure is unlikely to cause distress to others
economic vulnerabilities). Comprehensive to be less intensively supervised.
policies that are clearly communicated can
serve to reduce uncertainty and improve
overall market preparedness. The relatively Conclusions
short notice of systemic crisis, and high Although every measure of systemic risk
degree of noise in some signals, mean that has limitations to some degree, and indeed
policymakers should rely on a number of all models are by nature simplifications of
tools and measures to arrive at a robust the complexity of the real world, this chapter
assessment of when systemic risks are bound discusses various tools that can be used to shed
to materialize. In particular, stress tests that light on potential systemic events. Thus far,
take into account systemic effects and inter- financial sector regulation and supervision have
connections should be implemented. More- focused on the risk of failure of each financial
over, a comprehensive and coordinated crisis institution in isolation. The analysis presented
preparedness plan needs to be in place before here suggests that regulators should take into
systemic events are detected. account the risk of both individual and systemic
• Take care when interpreting market signals during failures. Indeed, some proposals have begun
the crisis. If supervisors and central bankers to surface on how to account for systemic risks
are planning to use market-based data to in prudential regulation (e.g., Acharya, 2009;
assess systemic risk, it is important that they and Pedersen and Roubini, 2009). Some rely
recognize that policy interventions them- on the assumption that correlation among FIs
selves may affect their informational content. is a good proxy for detecting systemic risks. As
For instance, prohibitions on short selling or discussed above, measures based solely on asset
other impediments to the free flow of infor- return correlations are constrained in their
mation into prices are likely to distort signals ability to detect (and address) systemic risks,
given by market prices. Similarly, the intro- since they fail to capture the “fat-tailed” nature
duction of government guarantees for bank and changes in the probability distribution of
debt can alter the informational content of asset returns of key FIs, which are characteristic
FIs’ CDS spreads and equity prices (Box 3.6). of systemic crises. This suggests that prudential
As such, market-based indicators may only norms based on simple return correlations will
contain relatively unbiased information about be insufficient to capture systemic risk, and
systemic risk in the early phases of a crisis, will need to be broadened. The results suggest
prior to policy actions. Further work on the that authorities need to diversify their sources
indicators to control for policy responses is of information and the tools used to detect
needed. systemic risk.

143

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Box 3.6. The Transformation of Bank Risk into Sovereign Risk—The Tale of Credit Default Swaps
In the fall of 2008, the introduction of govern-
ment guarantees on bank liabilities prompted Irish Banks and Sovereign Five-Year
a decline in bank credit default swap (CDS) Credit Default Swap Spreads
(In basis points)
spreads, making the spreads less informative
Allied Irish Bank 400
and increasing costs to the government. In
Bank of Ireland
several countries with large banking systems this Irish government 350
has also led to a convergence of sovereign and
300
bank CDS spreads, which can result in feedback
effects between sovereign and bank spreads. 250
In 2008–09, a number of developed-country
200
governments provided financial guarantees
on bank liabilities, which prompted a sharp 150
decline in bank CDS spreads, as default risk
100
was transferred to the sovereign. This has had
several consequences. 50
First, information from bank CDS on default
0
risk becomes less informative as government 2007 08 09
intervention distorts the interpretation of
credit market signals. Using information from Source: Bloomberg L.P.

equity markets in a contingent claims approach

(CCA) model may provide a more accurate
U.K. Banks and Sovereign Five-Year
view on whether bank risk is increasing or Credit Default Swap Spreads
subsiding. From a systemic point of view it may (In basis points)
be desirable to shift focus to the joint probabil-
600
HBOS
ity of banks falling below certain “minimum”
Barclays
capital or “prompt corrective action” thresholds HSBC 500
rather than a joint probability of default (since Royal Bank of Scotland
U.K. government
the government is insuring liability holders 400
against the costs of default).
Second, potential costs to the government of 300
the guarantees have led to a rise in sovereign
CDS spreads. This is particularly true where 200
the financial system is large compared with the
government’s balance sheet or GDP. The banks’ 100
credit spreads depend on (1) retained risk,
which is low given the application of government 0
guarantees and assurances of continuing support; 2007 08 09

and (2) the government sovereign credit spread,

Source: Bloomberg L.P.
since investors view the banks’ creditworthiness
as dependent on that of the sovereign guarantor.
(The CCA model assumes that the government’s remaining in the debt and deposits of the finan-
contingent liability—the value of the explicit or cial sector, as described in Gray, Merton, and
implicit sovereign guarantee—is a fraction α of Bodie, 2008.) Thus, bank credit spreads should
the total PF implied put option to the financial be equal to or greater than sovereign spreads.
sector. The remainder, (1–α)PF , is credit risk In Ireland, after financial guarantees were
granted to banks, their CDS spreads declined
Note: Dale Gray prepared this box. and converged toward that of the sovereign.

144

©International Monetary Fund. Not for Redistribution

 ANNEX 3.1. FINANCIAL SOUNDNESS INDICATORS

A similar pattern was evident in the United systemic financial and sovereign debt crisis.
Kingdom, after financial guarantees were intro- On the other hand, improvement in bank and
duced for new bank-issued debt (see figure). sovereign balance sheets can lead to a virtuous
This inter-relationship of spreads could lead cycle as bank and sovereign spreads decline.
to a destabilizing feedback process where both Countries in a currency union do not have the
bank and sovereign CDS spreads increase in option to use the exchange rate as an inde-
response to shocks to bank assets and/or to pendent policy tool to restore macroeconomic
the sovereign’s revenue potential. In some stability. In such circumstances, the potential
situations (as in Iceland), this vicious cycle for sovereign default needs to be contained
can escalate to a point where the inability of through measures to limit the downside risk of
the government to provide sufficient credible exposure to the banking system and fiscal mea-
guarantees to banks leads to a simultaneous sures to restore credibility.

Annex 3.1. Financial Soundness Indicators

Core Set
Deposit-taking institutions’ capital adequacy Regulatory capital to risk-weighted assets
Regulatory Tier 1 capital to risk-weighted assets
Asset quality Nonperforming loans to total gross loans
Nonperforming loans net of provisions to capital
Sectoral distribution of loans to total loans
Large exposures to capital
Earnings and profitability Return on assets
Return on equity
Interest margin to gross income
Noninterest expenses to gross income
Liquidity Liquid assets to total assets (liquid asset ratio)
Liquid assets to short-term liabilities
Sensitivity to market risk Duration of assets
Duration of liabilities
Net open position in foreign exchange to capital
Encouraged Set
Deposit-taking institutions Capital to assets
Geographical distribution of loans to total loans
Gross liability position in financial derivatives to capital
Trading income to total income
Personnel expenses to noninterest expenses
Spread between highest and lowest interbank rate
Customer deposits to total (noninterbank) loans
Foreign currency-denominated loans to total loans
Foreign currency-denominated liabilities to total liabilities
Net open position in equities to capital
Market liquidity Average bid-ask spread in the securities market
Average daily turnover ratio in the securities market
Nonbank financial institutions Assets to total financial system assets
Assets to GDP
Corporate sector Total debt to equity
Return on equity
Earnings to interest and principal expenses
Corporate net foreign exhange exposure to equity
Number of applications for protection from creditors
Households Household debt to GDP
Household debt service and principal payments to income
Real estate markets Real estate prices
Residential real estate loans to total loans
Commercial real estate loans to total loans
Source: Sundararajan and others (2002).
145

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

Annex 3.2. Groups of Selected Financial Institutions

Core Groups Regions
Insurance
Core 1 Core 2 Europe Asia/United States Companies
Australia & New Zealand AIG Euro area Asia AIG (AIG)
Banking Group Ambac Financial Intesa Sanpaolo (ISP) Australia & New Zealand Allianz (ALV)
Bank of America Bank of America BNP Paribas (BNP) Banking Group (ANZ) Ambac Financial
Bank of China Citigroup Commerzbank (CBK) Bank of China (BOC) (ABK)
Citigroup Deutsche Bank Deutsche Bank (DBK) DBS Group (DBS) AXA (AXA)
Deutsche Bank Goldman Sachs Fortis (FORB) ICICI Bank (IBN) MBIA (MBI)
Goldman Sachs HSBC ING Group (INGA) Industrial Bank of Korea Munich Re (MUV)
HSBC JPMorgan Chase & Co. Santander Hispano Group (IBK) PMI (PMI)
Industrial Bank of Korea Lehman Brothers (SAN) Mitsubishi UFJ Financial Prudential Plc (PRU)
JPMorgan Chase & Co. Merrill Lynch Société Géneralé (GLE) (MUF) Swiss Re (RUKN)
Lehman Brothers Morgan Stanley UniCredito (UCG) Nomura (NOM)
Merrill Lynch Royal Bank of Scotland State Bank of India (SBIN)
Mitsubishi UFJ Swiss Re Non-euro area Sumitomo Mitsui Financial
Morgan Stanley UBS Barclays (BARC) (SMF)
Royal Bank of Scotland Wachovia Credit Suisse (CSGN)
UBS Danske (DANSK) United States
Wachovia HBOS (HBOS) Bank of America (BAC)
HSBC (HSBA) Bear Stearns (BSC)
LloydsTSB (LLOY) Citigroup (C)
Nordea (NDA) Goldman Sachs (GS)
Royal Bank of Scotland JPMorgan Chase & Co. (JPM)
(RBS) Lehman Brothers (LEH)
UBS (UBS) Merrill Lynch (MER)
Morgan Stanley (MS)
Wachovia (WB)

Annex 3.3. List of Intervened Financial Institutions

Date(s) of Country Institution Date(s) of Country Institution
Intervention Intervention
Intervened institutions - banks Intervened institutions - investment banks
9/29/2008 United States Wachovia 3/14/2008 United States Bear Stearns
9/29/2008 Belgium/Netherlands/ Fortis 9/15/2008 United States Lehman
Luxembourg Brothers
10/3/2008 Belgium/Netherlands Fortis 9/15/2008 United States Merrill Lynch
10/13/2008 United Kingdom Royal Bank of 10/28/2008 United States Goldman
Scotland, Sachs
HBOS, 10/28/2008 United States Morgan
LloydsTSB Stanley
10/16/2008 Switzerland UBS Intervened institutions - insurance companies
10/19/2008 Netherlands ING Group 9/16/2008 United States AIG
10/28/2008 United States JPMorgan
Chase & Co.
10/28/2008 United States Bank of
America
11/24/2008 United States Citigroup
1/8/2009 Germany Commerzbank
1/19/2009 United Kingdom Royal Bank of
Scotland

146

©International Monetary Fund. Not for Redistribution

 REFERENCES

References Cover, Thomas, and Joy Thomas, 2006, Elements of

Information Theory, Wiley Series in Telecommunica-
Archarya, Viral, 2009, “A Theory of Systemic Risk
tions and Signal Processing (Hoboken, New Jersey:
and Design of Prudential Bank Regulation,” CEPR
John Wiley & Sons, 2nd ed.).
Discussion Paper No. 7164 (London: Centre for
Davis, E. Philip, and Dilruba Karim, 2008, “Could
Economic Policy Research).
Early Warning Systems Have Helped to Predict the
Athanasopoulou, Marialena, Miguel Segoviano, and
Sub-prime Crisis?” National Institute Economic Review,
Alexander Tieman, forthcoming, “Banks’ Prob-
ability of Default: Which Methodology, When, and Vol. 206, No. 1 pp. 35–47.
Why?” IMF Working Paper (Washington: Interna- Demirgüç-Kunt, Asli, and Enrica Detragiache, 1998,
tional Monetary Fund). “The Determinants of Banking Crises in Develop-
Baba, Naohiko, Frank Packer, and Teppei Nagano, ing and Developed Countries,” IMF Staff Papers, Vol.
2008, “The Spillover of Money Market Turbulence 45, No. 1, pp. 81–109.
to FX Swap and Cross-Currency Swap Markets,” BIS ———, 1999, “Monitoring Banking Sector Fragility: A
Quarterly Review (March), pp. 73–86 (Basel: Bank Multivariate Logit Approach,” IMF Working Paper
for International Settlements). 99/147 (Washington: International Monetary Fund).
Baek, Ehung, and William Brock, 1992, “A General ———, 2005, “Cross-Country Empirical Studies of
Test for Nonlinear Granger Causality: Bivariate Systemic Bank Distress: A Survey,” IMF Working
Model” (unpublished). Paper 05/96 (Washington: International Monetary
Bergo, Jarle, 2002, “Comment on ‘Using Financial Fund).
Soundness Indicators to Assess Risks to Financial Dornbusch, Rudiger, Yung Chul Park, and Stijn Claes-
Stability,’” paper presented at the IMF conference sens, 2000, “Contagion: Understanding How It
on “Challenges to Central Banking from Globalized Spreads,” World Bank Research Observer, Vol. 15, No.
Financial Systems,” Washington, September 16–17. 2, pp. 177–97.
Black, Fischer, and Myron Scholes, 1973, “The Pricing Dungey, Mardi, Renée Fry, Brenda González-
of Options and Corporate Liabilities,” Journal of Hermosillo, and Vance Martin, 2005, “Empirical
Political Economy, Vol. 81, No. 3, pp. 637–59. Modelling of Contagion: A Review of Methodolo-
Brunnermeier, Markus, and Lasse H. Pedersen, forth- gies,” Quantitative Finance, Vol. 5, No. 1, pp. 9–24.
coming, “Market Liquidity and Funding Liquidity,” ———, 2006, “Contagion in International Bond
Review of Financial Studies. Markets during the Russian and the LTCM Crises,”
Buchen, Peter, and Michael Kelly, 1996, “The Maxi-
Journal of Financial Stability, Vol. 2, No. 1, pp. 1–27.
mum Entropy Distribution of an Asset Inferred
———, 2007, “Contagion in Global Equity Markets in
from Option Prices,” Journal of Financial and Quanti-
1998: The Effects of the Russian and LTCM Crises,”
tative Analysis, Vol. 31, No. 1, pp. 143–59.
North American Journal of Finance and Economics, Vol.
Cappiello, Lorenzo, Robert F. Engle, and Kevin
18, No. 2, pp. 155–74.
Sheppard, 2006, “Asymmetric Dynamics in the
———, and Chrismin Tang, forthcoming, “Are Finan-
Correlations of Global Equity and Bond Returns,”
cial Crises Alike? From the 1998 Russian/LTCM
Journal of Financial Econometrics, Vol. 4, No. 4, pp.
Crisis to the 2007 Subprime Debacle and Liquidity
537–72.
Crisis,” IMF Working Paper (Washington: Interna-
Capuano, Christian, 2008, “The Option-iPoD. The
Probability of Default Implied by Option Prices tional Monetary Fund).
Based on Entropy,” IMF Working Paper 08/194 Embrechts, Paul, Filip Lindskog, and Alexander
(Washington: International Monetary Fund). McNeil, 2003, “Modelling Dependence with
Cihák, Martin, and Klaus Schaeck, 2007, “How Well Copulas and Applications to Risk Management,” in
Do Aggregate Bank Ratios Identify Banking Prob- Handbook of Heavy Tailed Distributions in Finance, ed.
lems?” IMF Working Paper 07/275 (Washington: by S.T. Rachev (Amsterdam: North-Holland).
International Monetary Fund). Engle, Robert, 2002, “Dynamic Conditional Correla-
Coles, Stuart G., Janet Heffernan, and Jonathan A. tion: A Simple Class of Multivariate Generalized
Tawn, 1999, “Dependence Measures for Extreme Autoregressive Conditional Heteroskedasticity Mod-
Value Analyses,” Extremes, Vol. 2 (December), pp. els,” Journal of Business & Economic Statistics, Vol. 20,
339–65. No. 3, pp. 339–50.

147

©International Monetary Fund. Not for Redistribution

 CHAPTER 3 DETECTING SYSTEMIC RISK

European Central Bank, 2005, Financial Stability Review Hamilton, James D., and Raul Susmel, 1994, “Autore-
(Frankfurt: European Central Bank). gressive Conditional Heteroskedasticity and
———, 2007, “Measuring Investors’ Risk Appetite,” Changes in Regime,” Journal of Econometrics, Vol. 64
Financial Stability Review (June), pp. 166–71. (September-October), pp. 307–33.
Forbes, Kristin, and Roberto Rigobon, 2002, “No Con- Hardy, Daniel C., and Ceyla Pazarbasioglu, 1999,
tagion, Only Interdependence: Measuring Stock “Determinants and Leading Indicators of Banking
Market Comovements,” Journal of Finance, Vol. 57, Crises: Further Evidence,” IMF Staff Papers, Vol. 46,
No. 5, pp. 2223–61. No. 3, pp. 247–58.
Frank, Nathaniel, Brenda González-Hermosillo, and Hesse, Heiko, and Miguel Segoviano, forthcom-
Heiko Hesse, 2008, “Transmission of Liquidity ing, “Distress Dependence, Tail Risk and Regime
Shocks: Evidence from the 2007 Subprime Crisis,” Changes,” IMF Working Paper (Washington: Inter-
IMF Working Paper 08/200 (Washington: Interna- national Monetary Fund).
tional Monetary Fund). Hiemstra, Craig, and Jonathan D. Jones, 1994, “Test-
Frank, Nathaniel, and Heiko Hesse, forthcoming, ing for Linear and Nonlinear Granger Causality in
“Financial Spillovers to Emerging Markets during the Stock Price-Volume Relation,” Journal of Finance,
the Global Financial Crisis,” IMF Working Paper Vol. 49, No. 5, pp. 1639–64.
(Washington: International Monetary Fund). Huang, Xin, Hao Zhou, and Haibin Zhu, 2008, “A
Framework for Assessing the Systemic Risk of Major
González-Hermosillo, Brenda, 1999, “Determinants of
Financial Institutions” (unpublished).
Ex Ante Banking System Distress: A Macro-Micro
Hull, John, Izzy Nelken, and Alan White, 2004,
Empirical Exploration of Some Recent Episodes,”
“Merton’s Model, Credit Risk and Volatility Skews,”
IMF Working Paper 99/33 (Washington: Interna-
Journal of Credit Risk, Vol. 1, No. 1, pp. 3–28.
tional Monetary Fund).
Hutchinson, Michael M., 2002, “European Banking
———, 2008, “Investors’ Risk Appetite and Global
Distress and EMU: Institutional and Macroeco-
Financial Crises: 1998–2007,” IMF Working Paper
nomic Risks,” Scandinavian Journal of Economics, Vol.
08/85 (Washington: International Monetary Fund).
104, No. 3, pp. 365–89.
———, and Heiko Hesse, forthcoming, “Global Mar-
———, and Kathleen McDill, 1999, “Are All Banking
ket Conditions and Systemic Risk,” IMF Working
Crises Alike? The Japanese Experience in Inter-
Paper (Washington: International Monetary Fund).
national Comparison,” Journal of the Japanese and
Gorton, Gary, 2008, “The Panic of 2007,” paper pre-
International Economies, Vol. 13, No. 3, pp. 155–80.
sented at the Federal Reserve Bank of Kansas City
Illing, Mark, and Ying Liu, 2006, “Measuring Financial
Jackson Hole Conference, “Maintaining Stability
Stress in a Developed Country: An Application to
in a Changing Financial System,” Jackson Hole,
Canada,” Journal of Financial Stability, Vol. 2, No. 3,
Wyoming, August 21–23. pp. 243–65.
Gray, Dale F., and Andreas A. Jobst, forthcoming, International Monetary Fund (IMF), 2006, Finan-
“Tail Dependence Measures of Systemic Risk Using cial Soundness Indicators: Compilation Guide
Equity Options Data—Implications for Financial (Washington).
Stability,” IMF Working Paper (Washington: Inter- ———, 2007, Global Financial Stability Report, World
national Monetary Fund). Economic and Financial Surveys (Washington,
Gray, Dale F., and Samuel Malone, 2008, Macrofinan- April).
cial Risk Analysis, Wiley Financial Series (Hoboken, Jobst, Andreas A., 2007a, “Operational Risk—The
New Jersey: John Wiley & Sons). Sting Is Still in the Tail But the Poison Depends on
Gray, Dale F., Robert C. Merton, and Zvi Bodie, 2008, the Dose,” Journal of Operational Risk, Vol. 2, No. 2,
“New Framework for Measuring and Managing pp. 1–56.
Macrofinancial Risk and Financial Stability,” Work- ———, 2007b, “Consistent Quantitative Operational
ing Paper No. 09-015 (Cambridge, Massachusetts: Risk Measurement and Regulation: Challenges
Harvard Business School, August). of Model Specification, Data Collection and Loss
Group of Ten, 2001, Report on Consolidation in the Reporting,” in Operational Risk 2.0: Driving Value
Financial Sector (Basel: Bank for International Creation in a Post-Basel II Era, ed. by Ellen Davis
Settlements). (London: Risk Books, Incisive Media Ltd.).

148

©International Monetary Fund. Not for Redistribution

 REFERENCES

Kaminsky, Graciela L., and Carmen M. Reinhart, Rojas-Suarez, Liliana, 2001, “Rating Banks in Emerg-
1999, “The Twin Crises: The Causes of Banking and ing Markets: What Credit Rating Agencies Should
Balance-of-Payments Problems,” American Economic Learn from Financial Indicators,” IIE Working
Review, Vol. 89, No. 3, pp. 473–500. Paper No. 01-6 (Washington: Petersen Institute for
Kim, Oliver, and Robert E. Verrecchia, 1997, “Pre- International Economics).
Announcement and Event-Period Private Informa- Rosenblum, Harvey, Danielle DiMartino, Jessica J.
tion,” Journal of Accounting and Economics, Vol. 24, Renier, and Richard Alm, 2008, “Fed Interven-
No. 3, pp. 395–419. tion: Managing Moral Hazard in Financial Crises,”
Kullback, Solomon, 1959, Information Theory and Statis- Economic Letter: Insights from the Federal Reserve Bank
tics (New York: John Wiley & Sons). of Dallas, Vol. 3, No. 10.
———, and R.A. Leibler, 1951, “On Information and Segoviano, Miguel, 2006, “Consistent Information
Sufficiency,” The Annals of Mathematical Statistics, Multivariate Density Optimizing Methodology,”
Vol. 22, No. 1, pp. 79–86. Financial Markets Group Discussion Paper No. 557
Lando, David, 2004, Credit Risk Modeling (Princeton, (London: London School of Economics).
New Jersey: Princeton University Press). ———, forthcoming, “The CIMDO-Copula. Robust
Lo, Andrew, 2008, “Hedge Funds, Systemic Risk, and Estimation of Default Dependence under Data
the Financial Crisis of 2007–2008,” testimony to the Restrictions,” IMF Working Paper (Washington:
U.S. House of Representatives Committee on Over- International Monetary Fund).
sight and Government Reform, November 13. ———, and Charles Goodhart, 2009, “Banking Stabil-
Longin, Francois, 2000, “From Value at Risk to Stress ity Measures,” IMF Working Paper 09/04 (Washing-
Testing: the Extreme Value Approach,” Journal of ton: International Monetary Fund).
Banking and Finance, Vol. 24, No. 7, pp. 1097–130. Sorge, Marco, 2004, “Stress-Testing Financial Systems:
Masson, Paul, 1999, “Contagion: Monsoonal Effects, An Overview of Current Methodologies,” BIS Work-
Spillovers, and Jumps Between Multiple Equilib- ing Paper No. 165 (Basel: Bank for International
ria,” in The Asian Financial Crisis: Causes, Contagion Settlements).
and Consequences, ed. by P.R. Agénor, M. Miller, D. Stephenson, Alec, 2003, “Simulating Multivariate
Vines, and A. Weber (Cambridge, United Kingdom: Extreme Value Distributions of Logistic Type,”
Cambridge University Press). Extremes, Vol. 6, No. 1, pp. 49–60.
Merton, Robert C., 1974, “On the Pricing of Corpo- Sundararajan, Vasudevan, Charles Enoch, Armida San
rate Debt: The Risk Structure of Interest Rates,” Jose, Paul Hilbers, Russell Krueger, Marina Moretti,
Journal of Finance, Vol. 29, No. 2, pp. 449–70. and Graham Slack, 2002, Financial Soundness Indica-
Pedersen, Lasse, and Nouriel Roubini, 2009, “A tors: Analytical Aspects and Country Practices, IMF
Proposal to Prevent Wholesale Financial Failure,” Occasional Paper No. 212 (Washington: Interna-
Financial Times, January 29. tional Monetary Fund).
Poghosyan, Tigran, and Martin Cihák, 2009, “Distress Theil, Henri, 1969, “On the Use of Information Theory
in European Banks: An Analysis Based on a New Concepts in the Analysis of Financial Statements,”
Data Set,” IMF Working Paper 09/09 (Washington: Management Science, Vol. 15, No. 9, pp. 459–80.
International Monetary Fund). Weistroffer, Christian, and Veronica Vallés, 2008,
Poon, Ser-Huang, Michael Rockinger, and Jonathan “Monitoring Banking Sector Risk: An Applied
Tawn, 2004, “Extreme Value Dependence in Finan- Approach,” Research Note No. 29 (Frankfurt:
cial Markets: Diagnostics, Models, and Financial Deutsche Bank, October 28). Available via the
Implications,” Review of Financial Studies, Vol. 17, Internet: [Link].
No. 2, pp. 581–610. Zou, J., 2003, “The Relationship Between Credit
Preuss, Lucien G., 1980, “A Class of Statistics Based on Default Probability and Equity Options Volatility
the Information Concept,” Communications in Statis- Surface,” presentation at the RISK USA conference,
tics—Theory and Methods, Vol. 9, No. 15, pp. 1563–85. Boston, June.

149

©International Monetary Fund. Not for Redistribution

 GLOSSARY

Asset-backed security (ABS) A security that is collateralized by the cash flows from a pool of
underlying assets, such as loans, leases, and receivables. Often, when
the cash flows are collateralized by real estate, an ABS is called a
mortgage-backed security.

Auction rate security Long-term debt or preferred stock for which the coupon or dividend
is regularly reset via Dutch auction.

Basel II An accord providing a comprehensive revision of the Basel capital

adequacy standards issued by the Basel Committee on Banking
Supervision. Pillar I of the accord covers the minimum capital
adequacy standards for banks, Pillar II focuses on enhancing
the supervisory review process, and Pillar III encourages market
discipline through increased disclosure of banks’ financial condition.

Book value per share The value of a company’s assets after deducting the value of its
liabilities, divided by the number of outstanding shares.

Common equity Shareholders’ total equity minus preferred equity.

Commercial mortgage-backed A series of indexes, each referencing 25 tranches of commercial

securities index (CMBX) mortgage-backed securities, with differing credit ratings.

Credit default swap (CDS) A credit derivative whose payout is triggered by a “credit event,” often
a default. CDS settlements can either be “physical”—whereby the
protection seller buys a defaulted reference asset from the protection
buyer at its face value—or in “cash”—whereby the protection seller pays
the protection buyer an amount equal to the difference between the
reference asset face value and the price of the defaulted asset.

Credit derivative A financial contract under which an agent buys or sells risk protection
against the credit risk associated with a specific reference entity (or
specified range of entities). For a periodic fee, the protection seller
agrees to make a contingent payment to the buyer on the occurrence
of a credit event (usually default in the case of a credit default swap).

Credit spread The spread between benchmark securities and other debt securities
that are comparable in all respects except for credit quality (e.g.,
the difference between yields on U.S. treasuries and those on single
A-rated corporate bonds of a certain term to maturity).

Derivative A financial contract whose value derives from underlying securities

prices, interest rates, foreign exchange rates, commodity prices, or
market or other indices.

150

©International Monetary Fund. Not for Redistribution

 GLOSSARY

EMBIG JPMorgan’s Emerging Market Bond Index Global, which tracks the
total returns for traded external debt instruments in 34 emerging
market economies with weights roughly proportional to the market
supply of debt.

Emerging markets Developing countries’ financial markets that are less than fully
developed, but are nonetheless broadly accessible to foreign investors.

Government-sponsored A financial institution that provides credit to specific groups or

enterprise (GSE) areas of the economy, such as farmers or housing. Most enterprises
maintain legal and/or financial ties to the government.

Hedge fund An investment pool, typically organized as a private partnership

and often resident offshore for tax and regulatory purposes. These
funds face few restrictions on their portfolios and transactions.
Consequently, they are free to use a variety of investment
techniques—including short positions, transactions in derivatives, and
leverage—to attempt to raise returns and manage risk.

Hedging Offsetting an existing risk exposure by taking an opposite position

in the same or a similar risk—for example, in related derivatives
contracts.

Hybrid security A broad group of securities that combine the elements of both debt
and equity. They pay a fixed or floating rate coupon or dividend
until a certain date, at which point the holder can have a number of
options, including converting the securities into the underlying share.
Therefore, unlike equity, the holder has a predetermined cash flow,
and, unlike a fixed-income security, the holder has the option to gain
when the issuer’s equity price rises. Hybrids are typically subordinate
to other debt obligations in the capital structure of the firm.

Implied volatility The expected volatility of a security’s price as implied by the price
of options or swaptions (options to enter into swaps) traded on that
security. Implied volatility is computed as the expected standard
deviation that must be imputed to satisfy risk neutral arbitrage
conditions, and is calculated with the use of an options pricing model
such as Black-Scholes.

Impulse response function An econometric technique typically used for vector autoregressions
that traces the impact to the variable in question over time from a
shock to another variable.

Institutional investor A bank, insurance company, pension fund, mutual fund, hedge fund,
brokerage, or other financial group that takes investments from
clients or invests on its own behalf.

151

©International Monetary Fund. Not for Redistribution

 GLOSSARY

Intermediation The process of transferring funds from the ultimate source to the
ultimate user. A financial institution, such as a bank, intermediates
when it obtains money from depositors or other lenders and onlends
to borrowers.

Investment-grade obligation A bond or loan is considered investment grade if it is assigned

a credit rating in the top four categories. S&P and Fitch classify
investment-grade obligations as BBB- or higher, and Moody’s classifies
investment-grade obligations as Baa3 or higher.

LCDX An index referencing credit default swaps on loans of 100 individual

companies that have unsecured debt trading in the secondary market.

Leverage The proportion of debt to equity (also assets to equity and assets to
capital). Leverage can be built up by borrowing (on-balance-sheet
leverage, commonly measured by debt-to-equity ratios) or by using
off-balance-sheet transactions.

Leveraged buyout (LBO) The acquisition of a company using a significant level of borrowing
(through bonds or loans) to meet the cost of acquisition. Usually, the
assets of the company being acquired are used as collateral for the
loans.

LIBOR The London Interbank Offered Rate is an index of the interest rates
at which banks offer to lend unsecured funds to other banks in the
London wholesale money market.

Mortgage-backed security A security that derives its cash flows from principal and interest
(MBS) payments on pooled mortgage loans. MBSs can be backed by
residential mortgage loans or loans on commercial properties.

Nonperforming loans Loans that the bank foresees it will have difficulty in collecting. They
include nonaccrual loans, reduced rate loans, renegotiated loans,
and loans past due 90 days or more. They exclude assets acquired in
foreclosures and repossessed personal property.

Originate-to-distribute model A business model for financial intermediation, under which financial
institutions originate loans such as mortgages, repackage them into
securitized products, and then sell them to investors.

Overnight index swap (OIS) An interest rate swap whereby the compounded overnight rate in the
specified currency is exchanged for some fixed interest rate over a
specified term.

Private equity Shares in privately held companies that are not listed on a public
stock exchange.

Private equity fund Pool of capital invested by a private equity partnership, typically
involving the purchase of majority stakes in companies and/or
entire business units to restructure the capital, management, and
organization.

152

©International Monetary Fund. Not for Redistribution

 GLOSSARY

Provision for loan loss Losses that the bank expects to take as a result of uncollectible
or troubled loans. Includes transfer to bad debt reserves and
amortization of loans.

Regulatory arbitrage Taking advantage of differences in regulatory treatment across

countries or different financial sectors, as well as differences
between the real (economic) risk and that as measured by regulatory
guidelines, to reduce regulatory capital requirements.

Repurchase agreement (repo) An agreement whereby the seller of securities agrees to buy them
back at a specified time and price. The transaction is a means of
borrowing cash collateralized by the securities “repo-ed” at an interest
rate implied by the forward repurchase price.

Risk aversion The degree to which an investor who, when faced with two
investments with the same expected return but different risk, prefers
the one with the lower risk. That is, it measures an investor’s aversion
to uncertain outcomes or payoffs.

Risk premium The extra expected return on an asset that investors demand in
exchange for accepting the higher risk associated with an asset.

ROA Return on assets, which equals (net income before preferred

dividends plus ((interest expense on debt-interest capitalized)
multiplied by (1 minus tax rate))) divided by last year’s total assets
multiplied by 100.

ROE Return on equity, which equals total income minus preferred

dividends divided by total common equity multiplied by 100.

Securitization The creation of securities from a pool of preexisting assets and

receivables that are placed under the legal control of investors
through a special intermediary created for this purpose (a “special
purpose vehicle” [SPV] or “special purpose entity” [SPE]). In the
case of “synthetic” securitizations, the securities are created from a
portfolio of derivative instruments.

Short-term debt and current The portion of debt payable within one year, including the current
portfolio long-term debt portion of long-term debt and sinking fund requirements of
preferred stock or debentures.

Spread See “credit spread” above. Other definitions include (1) the gap
between the market bid and ask price of a financial instrument; and
(2) the difference between the price at which an underwriter buys an
issue from the issuer and the price at which the underwriter sells it to
investors.

Structured credit product An instrument that pools and tranches credit risk exposure, including
mortgage-backed securities and collateralized debt obligations.

153

©International Monetary Fund. Not for Redistribution

 GLOSSARY

Structured investment vehicle A legal entity whose assets consist of asset-backed securities and
(SIV) various types of loans and receivables. An SIV’s funding liabilities
are usually tranched and include short- and medium-term debt; the
solvency of the SIV is put at risk if the value of the assets of the SIV
falls below the value of the maturing liabilities.

Subprime mortgage A mortgage loan to a borrower with an impaired or limited credit

history, and who typically has a low credit score.

Swap An agreement between counterparties to exchange periodic interest

payments based on different reference financial instruments on a
predetermined notional amount.

Tangible assets (TA) Total assets less intangible assets (such as goodwill and deferred tax
assets).

Tangible common equity Total balance sheet equity less preferred debt less intangible assets.
(TCE)

Tier 1 capital The core capital supporting the lending and deposit activities of a
bank. It consists primarily of common stock, retained earnings, and
perpetual preferred stock.

Tier 2 capital The supplemental capital supporting the lending and deposit
activities of a bank. It includes limited life preferred stock,
subordinated debt, and loan-loss reserves.

Total assets (banks) The sum of cash on hand and due from banks, total investments,
net loans, customer liability on acceptances, investment in
unconsolidated subsidiaries, real estate assets, net property, plant and
equipment, and other assets.

Total assets (insurance The sum of cash, total investments, premium balance receivables,
companies) investments in unconsolidated subsidiaries, net property, plant and
equipment, and other assets.

Total assets (other financial The sum of cash and equivalents, receivables, securities inventory,
companies) custody securities, total investments, net loans, net property, plant and
equipment, investments in unconsolidated subsidiaries, and other
assets.

Total capital The total investment in the company. It is the sum of common
equity, preferred stock, minority interests, long-term debt, nonequity
reserves, and deferred tax liability in untaxed reserves. For insurance
companies, policyholders’ equity is also included.

Total debt All interest-bearing and capitalized lease obligations.

Total deposits The value of money held by the bank or financial company on behalf
of its customers.

154

©International Monetary Fund. Not for Redistribution

 GLOSSARY

Total loans The total amount of money loaned to customers before reserves for
loan losses but after unearned income. It includes lease financing and
finance receivables.

Vector autoregression (VAR) An econometric time series technique that models the dynamic
interaction among the chosen set variables.

Yield curve The relationship between the interest rates (or yields) and time to
maturity for debt securities of equivalent credit risk.

155

©International Monetary Fund. Not for Redistribution