130

Quarterly Bulletin 2016 Q3

Stress testing of banks:

an introduction
By Kieran Dent and Ben Westwood of the Bank of England’s Stress Testing Strategy Division and
Miguel Segoviano of the International Monetary Fund’s Financial Sector Assessments and Policies Division.

• The usage and prominence of bank stress tests has risen substantially in the years following the
global financial crisis. They are now established as a key part of the bank regulation toolkit.
• Typically, bank stress tests measure the resilience of banks to hypothetical adverse scenarios like
severe recessions, with results used by central banks and regulators to measure risks and manage
them through the setting of prudential policy.
• Over time, to enhance their usefulness to policymakers, stress tests are likely to develop further,
for example by testing banks against a wider range of resilience metrics than capital, and further
exploring how stresses might be transmitted across the financial system (eg through contagion).

Overview

Stress testing involves putting a severe amount of pressure Prior to the global financial crisis, the use of stress tests by
on an object or system, to test how resilient it is under regulators was limited, particularly in the context of helping
extreme conditions. It is a tool used in a number of to set policy. But the crisis marked a step change, and
industries, from construction to cardiac health care. When several authorities have since started to develop and
applied to banks, stress testing involves analysing how these implement concurrent bank stress-testing frameworks.
institutions would cope with hypothetical adverse scenarios,
such as severe recessions or financial crises. The summary A concurrent bank stress test is a simultaneous stress test
figure provides a stylised depiction of a bank stress test. of several banks carried out under the direction of a
stress-testing authority, such as a central bank or banking
system regulator. These tests are designed to look at the
Summary figure Stylised depiction of a bank stress test
resilience of banks to potential future risks. They are
ultimately aimed at helping authorities to understand and
explain risks faced by banks, as well as, in some cases,
contributing to setting policy aimed at promoting resilience.
Economic Rising
recession unemployment
Stress tests generally start with the specification of
hypothetical stress scenarios. These scenarios tend to
incorporate paths for economic and financial market
Falling house
prices
Falling
share prices variables, which together are more severe than the
What does this
mean for stress-testing authority’s central expectations, and which
bank capital? might be expected to have an adverse impact on banks.
A variety of different modelling techniques are then used to
estimate the impact of the scenario(s) on banks’ profits and
CAPITAL RATIO
balance sheets.

Adequately Bank stress testing is still a relatively new field, and is likely
capitalised
to develop further over time in order to improve its
Inadequately
capitalised usefulness for policymakers, particularly with regards to
setting policy aimed at promoting the resilience of banking
Stress forecast horizon systems as a whole.
 Topical articles Stress testing of banks: an introduction 131

Introduction and are not run simultaneously. First, by evaluating banks

against a common scenario, concurrent exercises produce
In general, stress testing involves analysing how an object or results that are more comparable across banks. This helps
system copes under pressure. Doctors perform cardiac stress supervisory authorities responsible for promoting the safety
tests by getting patients to run on treadmills and monitoring and soundness of individual banks to conduct microprudential
their pulse and blood pressure. Engineers stress test policy in a more consistent manner.
construction materials by measuring their behaviour when
subjected to strain. Second, by assessing the impact across banks at the same
time, concurrent exercises allow policymakers to identify
Bank stress testing is designed to test the resilience of banks whether a particular shock is likely to affect many banks
to severe but plausible shocks. In practice, this typically or just a few. This is helpful in determining the likely
means modelling the impact of hypothetical adverse system-wide impact of the shock, and hence risks to the
macroeconomic and financial market scenarios on bank provision of financial services to households and businesses.
profitability and balance sheets.
Related to that, concurrent exercises also support attempts to
Adverse scenarios typically contain hypothetical future paths quantify the impact of the feedback and amplification
for a set of economic and financial market variables, which channels that operate within the banking sector, and between
together might be expected to stress bank business models banks, the broader financial system and the wider economy.
and to lead to losses. These scenarios are designed to be These mechanisms serve to exacerbate the impact of an initial
much worse than stress-testing authorities’ central shock, and spread its effects across a larger number of
expectations about how economic activity and financial institutions. For example, banks seeking to limit the impact of
market developments are likely to turn out. Examples might the stress may reduce their lending, thus contributing to an
include scenarios resembling severe recessions with falling even larger squeeze on economic activity and further raising
GDP, sharp contractions in house prices, and rising the likelihood that banks will make losses on their lending.
unemployment. Analysis of a scenario’s impact involves Incorporating feedback and amplification channels into a
modelling the way in which the scenario would be likely to stress test helps macroprudential authorities to quantify the
affect different aspects of participating banks’ businesses. system-wide impact of adverse events, and supports them in
For example, an increase in unemployment would reduce the designing policies that apply to all banks with the aim of
income of some households and may mean that more promoting and enhancing financial stability.
households default on their mortgages and other loans.
The first section of this article tracks the history of bank
From the perspective of central banks and bank regulators,
stress testing from its beginnings in the early 1990s to the
stress tests have the potential to support both risk
emergence of concurrent regulatory stress tests in response
measurement and risk management. In other words, as well
to the financial crisis. The following section then discusses the
as helping to measure the impact of potential future shocks on
main features of a concurrent stress-testing framework and
individual banks and the wider banking system, they can also
compares the different approaches taken to concurrent stress
be used to help set prudential policy aimed at making sure
testing internationally. The final section explores how
that individual banks (microprudential policy) and the banking
concurrent stress testing might develop, as regulators strive
system as a whole (macroprudential policy) are adequately
to improve their frameworks in support of their policymaking
resilient. For example, stress-test results help policymakers
objectives.
set capital requirements, which are in place to ensure banks
fund themselves with sufficient loss-absorbing capital to
reduce their likelihood of failure.(1) A brief history of bank solvency stress testing

This article is an introduction to public, concurrent stress tests Prior to the financial crisis, stress testing of banks was largely
of banks, which focus mainly on banks’ capital positions — conducted by banks themselves for internal risk management
otherwise known as solvency stress tests.(2) A concurrent purposes. While some regulatory authorities did conduct
stress test is defined as one carried out under the direction of stress tests before the financial crisis, these tended to be
an official body, such as a banking regulator, and in which the simple exercises with little direct impact on policy.(3)
entire balance sheets of several banks are simultaneously
subjected to the same adverse scenario. It is these stress tests
(1) See Farag, Harland and Nixon (2013).
that have received the most attention in the wake of the (2) Solvency-orientated stress tests focus on the implication of a stress for bank capital.
Stress tests may also focus on bank liquidity, though these are not the focus of this
global financial crisis. article. Some solvency stress tests may also incorporate liquidity stresses, as the two
are often connected.
Concurrent stress tests have two main advantages relative to (3) For example, simple stress tests were carried out on the French, UK and
Finnish banking systems. See De Bandt and Oung (2004), Hoggarth, Sorensen and
stress-testing exercises that do not impose common scenarios Zicchino (2005), Virolainen (2004).
132 Quarterly Bulletin 2016 Q3

Following the financial crisis, stress tests have taken on a much Table A Historical scenarios used by early stress-test practitioners
more prominent role within the regulatory toolkit. This Scenario Date Description
section sets out that history.
Black Monday 1987 International equity market crash, which
reduced the value of major international stock
Figure 1 provides a timeline of the key events that have markets by between 19% and 40% over the
month.
shaped the development of bank stress testing.
US interest rate 1994 A sharp, unexpected increase in US interest
shock rates significantly reducing the value of bond
Figure 1 Timeline of key events in the development of portfolios. The resulting shock spread into the
US equity market.
bank stress testing
Early 1990s Banks begin small-scale stress tests of their Mexican peso crisis 1994 A sudden, unexpected devaluation of the
trading activities. Mexican peso. This led investors to liquidate
their positions in Mexico and other developing
countries with contagion spreading
throughout financial markets in Asia and
Latin America.
1996 Market risk amendment to the Basel Capital
Accord.
Asian crisis 1997 A series of currency devaluations starting in
Thailand, following speculative attacks on the
1999 IMF and World Bank launch the Financial Sector baht, and subsequently spreading to other
Assessment Program (FSAP). Asian markets. As the crisis spread, most of
South East Asia and Japan saw significant
Early 2000s National central banks and supervisory currency depreciations and a collapse in the
authorities begin to develop their own value of stock markets and other asset prices.
independent bank stress tests.
Russian crisis 1998 A shock to the value of Russian stock, bond
and currency markets. This resulted from
2004 Basel II introduces requirement for credit risk falling investor sentiment after a period of
stress testing by banks. economic turmoil. In response, the Russian
government devalued the rouble, defaulted on
its domestic debt and suspended repayments
Federal Reserve begins the Supervisory Capital Assessment
on its foreign debt.
Feb. 2009
Program (SCAP).

May 2009 Committee of European Banking Supervisors (CEBS) while emerging market scenarios often focused on discrete,
begins inaugural EU-wide stress test.
disruptive events such as a government being unable to meet
its debt obligations.
2011 Federal Reserve begins Comprehensive Capital
Analysis and Review (CCAR) programme which
incorporates an annual bank stress test. The manner in which these early stress tests were used varied
greatly across banks. At some banks, they were aimed at
quantifying the maximum loss a bank might incur on a trading
2014 Bank of England begins annual stress-testing
programme. portfolio, while at other banks they were aimed at determining
trading limits, or quantifying the appropriate amount of
The emergence of stress testing within banks capital to fund a particular portfolio (Committee on the
Within banks, scenario-based stress testing first emerged as a Global Financial System (CGFS) (2000)).
discipline within banks’ trading activities in the early 1990s to
complement other statistical techniques used to evaluate risks The practice of using stress tests to evaluate trading
banks were running on their trading books (Blaschke et al portfolios was formalised in 1996 with an amendment to
(2001), McGee and Khaykin (2013)). Typically, trading desk the international regulatory capital regime for market risk
managers would test their portfolios against both historical (the risk of losses on positions associated with changes in
and hypothetical scenarios (Araten (2013)). market prices).(1) Following this amendment, banks seeking
to use their own internal models to quantify market risk for
As the name suggests, historical scenarios were based on past regulatory capital purposes were required to implement a
extreme market events and were used to evaluate the impact bank-wide stress-testing programme for market risk.
that a repeat of these events might have on current trading
portfolios. Table A describes some of the historical scenarios While firm-wide stress testing for market risk went on to
most commonly used by early stress-test practitioners. become standard practice at large international banks, the
development of stress tests for credit risk — the risk associated
While the severity of past events can provide a useful with a bank’s counterparties or borrowers failing to make
benchmark, past stress events have not tended to simply payments — significantly lagged those of market risk. In 1999
repeat themselves. Recognition of this fact led to demand for the Basel Committee on Banking Supervision (BCBS) found
scenarios that could test banks against potential future risks, that little progress had been made to develop techniques to
based on severe but plausible hypothetical events. In implement credit risk stress tests (BCBS (1999)). This was
advanced economies, these hypothetical scenarios were
often based on changes in economic growth prospects, (1) 1996 market risk amendment to the Basel Capital Accord.
 Topical articles Stress testing of banks: an introduction 133

despite the fact that credit risk was the most significant risk The use of stress tests within FSAPs helped encourage national
facing most banks. In 2005, the CGFS reported that this was central banks to develop their own, independent stress tests.
still the case, and highlighted the need to develop better stress These often began as updates of previous FSAP scenarios with
tests incorporating loan portfolios, as well as instituting central banks developing models that considered the banking
bank-wide stress tests aimed at capturing all of the risks banks system as a single entity. Over time, these approaches started
faced. to evolve into the concurrent stress-testing frameworks widely
used today.
The first steps towards addressing the lag between credit and
market risk stress tests were taken in a revision to the Prior to the financial crisis, the concurrent stress tests
international regulatory capital regime published in 2004, conducted by policymakers rarely had a direct impact on
known as Basel II.(1) This sought to make it a requirement for regulatory or broader financial policy. But their outputs were
banks using their own internal models to determine credit risk often incorporated into broader financial stability assessments,
for regulatory capital purposes to have in place a programme with the results sometimes published in central bank
of stress testing. Under these stress-testing programmes, publications such as Financial Stability Reports.(4)
banks would review the robustness of their model-based
assessments and the adequacy of capital buffers above the The development of stress testing since the global
regulatory minimum.(2) Upon implementation, all banks financial crisis
would be additionally required to subject their loan portfolios The global financial crisis highlighted substantial deficiencies
to stress tests regardless of whether or not they were using in risk measurement and management across the financial
their own models to determine credit risk capital. sector.(5) With respect to stress testing by banks, the scenarios
used prior to the financial crisis were revealed to be
Basel II had not been universally implemented by advanced
significantly more benign than the crisis itself, while the loss
economies prior to the onset of the financial crisis. And even
estimates these exercises generated were well below banks’
where it had, banks’ stress-testing models for credit risk and
actual loss experience (BCBS (2009)).
for capturing both credit and market risk were still at a
developmental stage (BCBS (2009)) and Schuermann (2013)).
As well as exposing the shortcomings of stress-testing
The use of stress testing by policymakers prior to the practices at banks, the financial crisis also brought with it a
global financial crisis step change in the use of stress testing within the regulatory
Unlike the stress tests conducted by banks, which focused on sphere. Regulatory stress tests moved from being small-scale,
the risks faced at the portfolio or individual institutional level, isolated exercises within the broader risk assessment
the stress tests conducted by policymakers prior to the programme, to large-scale, comprehensive risk-assessment
financial crisis sought to capture the impact of severe, but programmes in their own right leading directly to policy
plausible, shocks on the entire financial system or even the responses.
wider economy.
The first prominent example of this new wave of stress tests
Although policymakers had been considering the potential was the US Supervisory Capital Assessment Program (SCAP)
impact of adverse events on the financial system for some conducted by the Federal Reserve in early 2009. The SCAP
time, the use of stress tests as a tool for doing so only stress test assessed whether the largest US banks had
emerged towards the end of the 1990s. This was spurred on sufficient capital resources to absorb losses and continue to
by their use in the Financial Sector Assessment Program operate under a common stress scenario. By design, the
(FSAP) established by the International Monetary Fund (IMF) scenario was significantly more severe than the expected
and the World Bank. The FSAP, launched in 1999, recognised trajectory for the economy at the time (Bernanke (2010)).
the significant detrimental effects financial instability can have
on economic growth and the workings of financial markets, as In a marked departure from the past, the results of the
evidenced by the financial crises of the 1980s and 1990s.(3) SCAP were publicly disclosed on a bank-by-bank basis.
Those banks judged to need additional capital resources were
From its inauguration, stress tests have been a key component given six months to raise that capital, with the US Treasury
of the FSAP and they have been performed for every country
participating in the programme. The purpose of these tests is
(1) Basel II was the second of the Basel Accords, which comprise recommendations on
to provide a quantitative measure of the vulnerability of a banking laws and regulations issued by the BCBS. Following the financial crisis,
substantial changes were made to the international regulatory framework under
country’s financial system to different macrofinancial Basel III which was agreed upon by the BCBS in 2010–11.
scenarios and to complement the insights gathered from other (2) More specifically, banks using either the advanced or foundation internal
ratings-based (IRB) approaches for credit risk were required to conduct credit risk
components of the assessment. These include qualitative stress tests as part of Pillar II.
vulnerability assessments and a review of the regulatory and (3) See IMF (2014).
(4) See, for example, Bank of England (2006), Box 6 in Section 3.
crisis management frameworks in place in a country. (5) See Haldane (2009).
134 Quarterly Bulletin 2016 Q3

Department providing a backstop in the event that any bank markets; rather they represent unlikely severe outcomes
was unable to do so in private markets. In the event, almost chosen by stress testers because they could have a material
all of the banks were able to raise sufficient equity privately so detrimental impact on banks. Were such hypothetical events
as not to need Treasury support. The SCAP is widely regarded to materialise, they would likely lead to banks making losses
to have made a significant contribution to stabilising the and reduce the amount of capital available to absorb further
US financial system, and restoring broader market confidence losses.
with the Treasury backstop recognised as an important driver
of its success (Krugman (2014), Schuermann (2013) and Stress-test practitioners often supplement these adverse
Zhang (2013)). scenarios with a baseline scenario under which the
macroeconomic and financial environment evolves in line with
The success of the SCAP was followed by a proliferation their central expectations. Baseline projections can provide
of frameworks for regular concurrent stress testing across useful information about banks’ expected strategies for the
central banks and supervisory authorities. The first years ahead, as well as providing a benchmark against which
EU-wide concurrent stress test was conducted in late to analyse results under the hypothetical stress.
2009 under the direction of the Committee of European
Banking Supervisors (CEBS). This was followed by another Projections of banks’ capital positions conditional on the stress
exercise conducted under the direction of the CEBS in 2010, scenario tend to be the headline results of a stress test.
and a series of exercises conducted under the direction of the Results can be used for a number of purposes, with some
European Banking Authority (EBA) starting in 2011. In the authorities using them as a tool to highlight financial stability
United Kingdom, these EU-wide exercises initially served to risks, some using them as part of their approach to setting
complement the stress-test scenarios that were being individual bank capital requirements, and others using them to
provided to banks to run on a non-concurrent basis by help set macroprudential policy as well.
the former Financial Services Authority (FSA), before the
Figure 2 provides a stylised illustration of one type of
Bank of England launched its own concurrent stress-testing
concurrent stress test focused on individual banks. Other
programme in 2014.
stress tests may include different or additional features, such
This greater regulatory focus on stress testing has helped to as feedback loops from the projected behaviour of other
drive improvements in banks’ own stress-testing capabilities financial market participants in response to the adverse
and risk management practices, with sophisticated, bank-wide scenario.
stress testing now common practice at systemically important
The following three sections discuss scenario design, the
banks. And as the immediate turmoil that followed the crisis
production of stress-test results, and uses of stress-test results
has abated, the focus of regulatory stress-testing frameworks
for policy purposes in more detail.
has shifted away from the immediate need to recapitalise the
banking system towards an ongoing assessment of the
Scenario design
adequacy of banks’ capital resources and to informing broader
Typically, the first stage in putting together a stress test is
micro and macroprudential policy (the box on pages 138–39
designing a scenario. The two key elements of this step are:
considers the current use of stress tests by selected regulatory
(a) to select the types of risks to be explored by the test; and
authorities). The key features of these regulatory
(b) to calibrate the severity of shocks.
stress-testing frameworks are described in the following
section.
At one extreme, the test may simulate a severe broad-based
downturn affecting the real economy as well as impacting
How concurrent stress testing of banks works financial markets and other asset prices. This is the approach
at present taken by the majority of stress-testing authorities. An
advantage of this type of test is that it allows stress-testing
A high-level description of modern concurrent stress authorities to factor in the correlation of different risks faced
testing of banks by banks, and their impact across different parts of banks’
Concurrent stress testing involves analysing the impact of one balance sheets. For such scenarios, authorities tend to use
or more hypothetical stressful scenarios on the capital position economic models to help design stresses that are coherent
of a selected group of banks. — ie scenarios in which the combination of shocks makes
sense together.
The hypothetical stresses involved are designed to test the
resilience of banks against the risks they face, and tend to be Adverse scenarios of this type typically include projections for
adverse macroeconomic and financial market scenarios, like a economic variables such as unemployment, output growth
severe recession combined with financial market distress. and asset prices, which are broadly consistent with the paths
These are not central case projections for the economy and these variables might be expected to take during severe
 Topical articles Stress testing of banks: an introduction 135

Figure 2 An illustration of a concurrent stress test

Bank A:
Bank A: Policy response
Starting balance sheet

Stressed projections
Adverse scenario Modelling process of balance sheets, System-wide policy response
capital and profitability

Bank B:
Bank B: Policy response
Starting balance sheet

Stage 1: The adverse scenario is defined Stage 2: A combination of banks’ own Stage 3: At the end of the modelling Stage 4: The stress-test results are
and the cut-off date for banks’ starting models and those of the regulatory process, a series of stressed projections are used to inform both the system-wide
balance sheets is specified. authorities are used to model the impact of produced. These show the impact of the and individual-bank policy responses.
the adverse scenario on banks. stress on banks’ balance sheets, capital This could include changes to
positions and profitability. system-wide capital requirements, or
improvements to banks’ risk
management processes.

downturns. That means asset prices and output fall materially, meet or exceed over the life of the stress. Where authorities
while unemployment tends to rise substantially. set hurdle rates, different approaches are taken. For example,
in some cases minimum regulatory requirements are used,
Broad-based adverse macroeconomic scenarios can impact while in others capital requirements and buffers applying to
banks’ capital positions through a number of channels. For individual firms are incorporated.
example, higher unemployment might reduce the ability of
households to repay mortgages and unsecured loans. A fall in Figure 3 illustrates the projected impact of a stress scenario
house prices might lower the value of collateral held against on two different banks relative to a hurdle rate, with Bank A
mortgages and therefore increase banks’ losses when exceeding the hurdle rate and Bank B falling short.
households default on those mortgages. The stress also
increases the riskiness of banks’ portfolios of performing loans. Figure 3 An illustration of the impact of stress scenarios
This, in turn, increases the amount of capital banks must fund on different banks’ capital positions
themselves with, as stipulated under the regulatory capital Bank A Bank B
framework.(1)

Beyond changes in losses on lending and increasing the

riskiness of loan portfolios, adverse macroeconomic scenarios
Hurdle rate
also tend to have other impacts. For example, the profits
generated by banks’ lending and deposit-taking activity —
their net interest income — may fall if bank funding costs rise
as a result of the stress.

Some stress tests also include adverse traded risk scenarios,

where changes in financial market prices and conditions Starting Stressed Starting Stressed
capital ratio capital ratio capital ratio capital ratio
reduce the profitability of banks’ trading operations. Stress
tests may also contain elements less closely related to Given the range of options for calibrating stress scenarios,
economic conditions, such as an increase in redress payouts ultimately, the severity of a stress is likely to depend on a
relating to past episodes of misconduct. stress-testing authority’s risk appetite. Since banks provide
valuable services that support investment and economic
At the other extreme to broad-based tests, authorities may be growth, regulators try to ensure that bank failure is
interested in exploring a specific risk, and so run a narrower appropriately unlikely. The appropriate likelihood of bank
stress scenario.(2) Authorities may also choose to run more failure will be a judgement for the relevant authority.
than one adverse scenario with different focuses.
(1) The regulatory capital framework requires banks to fund themselves with a minimum
amount of capital. This increases banks’ ability to absorb losses that could otherwise
The overall severity of a stress test can be thought of as the threaten their solvency. For more details see Farag, Harland and Nixon (2013).
(2) For example, the Reserve Bank of New Zealand recently ran an exercise in which
combined severity of the individual shocks incorporated in the New Zealand’s five largest lenders to the dairy sector were stress tested against
test, together with any ‘hurdle rate’, which is the level of sustained low milk prices and sharp falls in dairy land values. For more details see
www.rbnz.govt.nz/research-and-publications/reserve-bank-bulletin/2016/
capital authorities stipulate that banks should be projected to rbb2016-79-05.
136 Quarterly Bulletin 2016 Q3

Based on evidence that adverse shocks are more likely when Meanwhile, the main benefit of employing top-down
risks associated with the state of the financial cycle are macro-models is that they are more likely to capture
elevated — that is at times when credit growth is strong and system-wide impacts of a stress, though the trade-off for this
asset prices appear overvalued — some authorities judge that might be more uncertainty about the projections for individual
it would be better if banks had larger capital buffers to fall banks. Top-down modelling is typically undertaken by
back on when cyclical risks are high. For authorities who wish stress-testing authorities. Banks are less able to factor in
to set system-wide capital in this way, another approach to system-wide impacts because they do not have access to
calibrating the severity of stress scenarios is therefore to factor detailed information on other firms in the system.
in the extent of these cyclical risks.(1)
A stress-testing authority’s choice of approach will be
There are several other facets of stress-test design not covered influenced by their objectives. These include:
in this article, which vary across concurrent stress-test
practitioners. For example, different authorities run their 1. Measuring risks affecting both individual banks and the
concurrent tests over different time horizons, and while some wider banking system.
authorities produce results based on the assumption that bank 2. Ensuring that individual banks are adequately capitalised
balance sheets do not change during the stress, others allow and operating appropriate risk management and
balance sheets to vary in an attempt to capture banks’ stress-testing policies.
managerial responses to the stress. These choices are all likely 3. Helping to set macroprudential rules — including
to be linked to the objectives of the test in question, and the minimum capital levels — for the system as a whole.
nature of the risks explored in the stress scenario. The
practices of different authorities are summarised in the box Different approaches to stress testing lend themselves to
on pages 138–39. achieving different objectives. For example, an approach that
places most emphasis on detailed modelling of the impact of
Producing projections of bank profitability and capital
the stress at a portfolio level within different institutions is
In general, producing stress-test results involves assessing the
likely to be more appropriate for authorities seeking to set
impact of both the baseline and stress scenarios on banks.
capital for individual banks. Stress testers seeking to better
This typically involves using models to produce projections of
understand systemic risks might be less interested in the
bank balance sheets, profitability and capital under each
insights to be gained from resource-intensive granular models,
scenario.
instead favouring macro-models that might better project the
There are a range of approaches to producing these systemic impact of a stress.
projections. For example, they may be produced either by
banks themselves, regulatory authorities, or by a mixture of Some authorities adopt a mixed approach whereby banks are
the two. Likewise, they may be produced using top-down responsible for producing their own projections, and
macro-models or bottom-up micro-models. Top-down authorities make adjustments to these based on a mixture
models aim to assess the impact of the stress at a of top-down and bottom-up analysis. Some of these
system-wide level, before considering its implications for adjustments may be intended to improve the consistency of
individual banks. Bottom-up micro-models aim to assess the projections across banks, whereas, in principle, others might be
impact of the stress at the individual-bank level, before aimed at accounting for some of the system-wide impacts of
considering its implications for the banking system as a whole. the stress that would not necessarily be picked up by an
individual bank’s models.
Bottom-up modelling may be undertaken by either the
stress-testing authority or participating banks. The main In addition to the quantitative results of a stress test, some
advantages associated with banks producing bottom-up authorities place considerable weight on a parallel qualitative
stressed projections are that they may have more detailed review of banks’ stress-testing capabilities. The findings of
data and better customised models than regulatory these reviews provide an important indicator of the adequacy
authorities, for whom building specific models for each of banks’ risk management practices. A qualitative review
participating bank is costly. Where banks themselves are typically involves an assessment of banks’ technical
responsible for providing projections, authorities generally stress-testing capabilities, and the governance processes
take on a quality assurance role, analysing banks’ various surrounding their stress-testing functions. Where deficiencies
approaches with the aim of making results as comparable as are identified, this may warrant remedial effort by the banks
possible across institutions. Authorities may also produce concerned, and potentially the addition of further safety
their own bottom-up projections. This has the benefit of buffers in the form of bank capital.
allowing greater control of the assumptions underlying
(1) For example, the Bank of England’s annual cyclical scenario is calibrated to reflect
projections, which helps to ensure comparability in results Bank of England policymakers’ judgement on the state of the financial cycle
across institutions. (Bank of England (2015)).
 Topical articles Stress testing of banks: an introduction 137

Using stress-test results — communication and the which further enhancements could improve their usefulness
policy reaction framework for policymakers.
The results of concurrent stress tests can be used in a number
of ways. That said, in general terms, all stress tests are tools Limitations of stress testing
for measuring and managing the risks banks face on a Within the broader regulatory capital framework, stress tests
forward-looking basis. For authorities concerned primarily focused on banks’ capital positions — otherwise known as
with risk measurement, the results give a quantitative solvency-orientated tests — are simply an analytical tool used
assessment of the scale of risks facing banks, while for those to assess banks against the requirements set out in that
stress-testing authorities more heavily engaged in risk framework. As such, stress tests are not a substitute for a
management, stress-test results are an input to their policy robust capital framework but a complement to it. Similarly,
decisions. For microprudential policymaking authorities, how stress tests cannot replace a robust supervisory regime that
banks’ capital positions fare under stress relative to specified ensures banks have adequate risk management and
hurdle rates or minimum capital requirements is one governance processes in place.
important yardstick to judge whether individual banks are
adequately capitalised, and how they might need to adjust Further, the results of stress tests are only as robust as the
their capital plans. In some cases, overall stress-test results data and methodologies used, and assumptions made, in
are also used to help macroprudential policymakers judge the producing them. While significant progress has been made to
appropriate level of system-wide bank capital buffers. develop these methodologies in recent years, stress-test
results remain subject to a high degree of uncertainty. It is for
Ensuring that the way authorities intend to use test results is this reason that stress-test results are used as just one input
well understood can improve the effectiveness of stress tests. into the policymaking process when determining the
The communication of results and the policy reaction appropriate level of bank capital.
framework around the results are the two key elements of this.
This section considers three main areas across which
policymakers may choose to focus their efforts in further
Clear publication of details about stress scenarios along with
developing stress tests to increase their usefulness for
stress-test results allows external observers to judge the
informing micro and macroprudential policy:
resilience of banks to the various risks incorporated, and
improves the accountability of the relevant stress-testing 1. Improving the ability of stress tests to assess the resilience
authority. It can also enhance the credibility of a stress test, of individual banks by exploring different types of, and a
provided markets judge that the test is adequately severe. broader range of, risks.
And it gives investors another source of information about the 2. Integrating amplification and feedback mechanisms and
risks facing banks, which should help them to make more incorporating behavioural responses into stress tests.
informed decisions, and improve market discipline. This may 3. Extending the scope of stress tests beyond the core
equally apply to tests carried out by those concurrent banking sector.
stress-test practitioners not responsible for setting capital.
Figure 4 illustrates how some of these developments might
For authorities responsible for setting capital requirements, in work.
general, publicising the policy reaction framework for a stress
test means being clear about what would prompt them to Improving the ability of stress tests to assess the
take policy action, as well as what sort of policy actions might resilience of individual banks
be taken in the event of different results. Again, a robust and By evaluating the impact of severe, but plausible hypothetical
well-publicised policy reaction framework can enhance the shocks on individual banks, stress tests provide useful
public credibility of the stress test. And the combination of a information to microprudential supervisors about the
clear communication strategy and a well-articulated policy resilience of regulated institutions. At present, the concurrent
reaction framework might positively influence bank behaviour. stress-testing frameworks run by many authorities are skewed
For example, banks may take pre-emptive action to strengthen towards assessing capital adequacy as opposed to other
their capital position ahead of the release of stress-test results. potentially important resilience metrics. Two closely related
metrics that enhanced concurrent stress-testing frameworks
The future of stress testing might seek to capture are liquidity and funding resilience.

Over the past 25 years, stress tests have moved from being an Liquidity resilience captures a bank’s ability to meet its
isolated risk management tool, used by banks to assess the short-term obligations as they fall due and cope with a
resilience of their trading portfolios, to become a core part of sudden, and unexpected, increase in withdrawals by its
the regulatory toolkit worldwide. But today’s stress tests are depositors and other creditors. It also measures the ease and
not without limitations and there are a number of areas across speed with which the bank’s assets can be converted into cash
138 Quarterly Bulletin 2016 Q3

Comparison of international concurrent The DFAST includes an adverse and severely adverse scenario
— with the results from the severely adverse scenario used for
stress-testing practices
the CCAR exercise. The severely adverse scenario is calibrated
Stress-testing practices vary widely internationally, with using outturns for macro variables observable during severe
several authorities still in the development stage of their recessions. It has countercyclical elements inasmuch as
stress-testing frameworks. This box summarises a selection of unemployment has to rise to at least 10% in the severely
concurrent stress-testing frameworks, though this group is by adverse stress. So the lower unemployment currently is, the
no means exhaustive. larger the shock to unemployment that will be required to hit
that minimum. The adverse scenario incorporates somewhat
Case study 1: the IMF less severe shocks to key macro variables, but incorporates
Stress tests performed by the IMF support financial stability different risks relative to the severely adverse scenario. The
assessments executed under the FSAP.(1) IMF stress tests US Federal Reserve uses a dynamic balance sheet approach to
normally include two or three adverse scenarios, constructed model the impact of these scenarios. This approach allows
around the macrofinancial risks judged to be most significant bank balance sheets to evolve through the forecast horizon in
for the economy concerned. Stress-testing approaches in line with banks’ corporate plans. Banks with large trading
FSAPs need to be adaptable to individual country operations take part in additional traded risk and counterparty
circumstances. For example, data availability, the types of default risk scenarios.
appropriate shocks and the level of development of the
relevant authorities’ own stress-testing framework are likely Case study 3: the EBA stress test
to vary. The EBA conducted its first concurrent stress test in 2011,
taking over responsibility for EU-wide stress testing from the
Prior to 2009, FSAP stress tests tended to focus mainly on CEBS. In 2016, 51 banks participated in the test from across
bank solvency risk. This has given way to broader risk the EU, representing a combined 70% of EU banking sector
coverage that now almost always includes market and assets.
liquidity risk. Stress tests have traditionally focused on the
banking sector, but some FSAP stress tests have been applied The EBA run a joined-up adverse macro scenario with a
to additional sectors, such as insurance and, more recently, three-year horizon. The EBA test is conducted on a static
money market funds.(2) balance sheet basis, meaning that bank balance sheets do not
FSAP analysis of interconnectedness among financial change through the forecast horizon. The adverse macro
institutions and of systemic risk continues to develop.(3) scenario is developed by the European Central Bank and
Analysis incorporates interconnectedness, factors related to European Systemic Risk Board (ESRB) and aims to capture the
direct and indirect linkages across entities, sectors (including systemic risks representing the most material threats to the
banks and non-banks) and borders. While systemic risk stability of the EU financial sector. This is accompanied by a
assessments complement stress tests of individual entities in traded risk stress incorporating an adverse scenario consistent
FSAPs, systemic risk amplifications have not yet been fully with the macro scenario, and another two based on historical
embedded into macroprudential stress-test frameworks. This stress episodes.
is a key development aim for the IMF and a number of other
authorities.(4) Through its FSAP-related stress-testing work, Case study 4: the Bank of England
the IMF also aims to help individual authorities to develop The Bank of England ran its first concurrent stress test in 2014,
their own stress-testing frameworks. though the 2016 test will be the first conducted under its new
framework.(5) That framework incorporates an annual cyclical
Case study 2: the US Federal Reserve scenario and a biennial exploratory scenario, which the Bank
Following on from the 2009 SCAP, in 2011 US authorities intends to run for the first time in 2017. In the Bank of
launched the Comprehensive Capital Analysis and Review England’s 2015 and 2016 stress tests, seven major UK banks
(CCAR). The CCAR incorporates concurrent stress testing took part. Together these institutions account for around
as well as the capital planning process for individual banks. 80% of the lending to the UK real economy.
CCAR takes into account results from concurrent stress tests,
required under the Dodd-Frank Act (DFAST), alongside a
qualitative assessment of banks’ risk management, their (1) From its inception, in the aftermath of the Asian financial crises of the 1990s, up to
end-2015, there have been almost 350 FSAP assessments across 170 jurisdictions.
capital plans and the results of stress tests banks carry out (2) For example, the integrated framework for solvency and liquidity stress testing
themselves on an individual basis. At the end of the process (Barnhill and Schumacher (2011); a primer for stress testing pension funds
(Impavido (2011)).
the authorities may object to banks’ capital plans on either (3) For example, the Systemic Risk and Interconnectedness (SyRIN) framework
quantitative or qualitative grounds, requiring firms to amend (Segoviano et al (2016), forthcoming) used in the 2015 US FSAP and 2016 UK FSAP.
(4) See Segoviano et al (2017), forthcoming.
them or submit entirely new plans. (5) For more details see Bank of England (2015).
 Topical articles Stress testing of banks: an introduction 139

The severity of the annual cyclical scenario, which is run over a financial conditions. The test results are not, however, used to
five-year horizon, is designed to vary with policymakers’ set capital for banks.
judgements on risks associated with the state of the financial
cycle.(1) This is judged using key indicators including credit The stress test incorporates data on more than 350 banks,
variables, financial market and other asset prices. The from large banks to very small local banks holding deposit
Bank of England uses a dynamic balance sheet approach, and accounts with the Bank of Japan. The framework is designed
the stress test also includes a traded risk scenario calibrated to to help the Bank of Japan to analyse feedbacks and spillovers
be consistent with the shocks in the macro scenario and an in a stress, between the banking and macroeconomic sectors,
additional misconduct stress. The hurdle rate framework for as well as across banks.
the annual cyclical scenario is set according to individual firms’
capital requirements and also includes buffers for systemically Table 1 summarises further detail on the stress-test
important banks. The hurdle rate is therefore greater than the frameworks described above. This is not an exhaustive list,
internationally agreed minima applying to all banks. however, as several other central banks and regulators also run
stress tests of various types. In many cases, these frameworks
Case study 5: the Bank of Japan are still developing.
The Bank of Japan runs a semi-annual top-down macro stress
test, used for risk identification and communication purposes,
the results of which appear in their Financial System Report.
(1) For details on the Bank of England’s inaugural annual cyclical scenario, see
The scenarios employed reflect the state of economic and Bank of England (2016).

Table 1 Summary of selected international stress-testing frameworks

Bank of England US Federal Reserve European Banking Authority IMF Bank of Japan

Design

Bank inclusion threshold £50 billion retail US$50 billion total assets At least €30 billion total Varies according to More than 350 banks are part of the
deposits assets country circumstances test, including 10 major banks.

Comprehensive macro √ √ √ √ √
scenario

Hurdle rate in excess of √ √ x Varies according to x

international minima country circumstances

Countercyclical √ Some elements x x √

Number of stress scenarios 1–2 2 1 macro Varies according to 2

country circumstances
Frequency Annual Annual Biennial With FSAP Semi-annual

Results production

Bottom-up (banks) x Collected but not typically √ x x

driver of results
Bottom-up (authorities) x √ x x x

Top-down (authorities) x Produced but not typically x x √

driver of results
Hybrid √ x x √ x

Explicitly incorporates √ (early stages of x x Sometimes √

amplification and feedback development)
channels

Results use

Microprudential policy √ √ √ Provides supporting x

setting(a) information

Macroprudential policy √ May feed in Via link to European Provides supporting x

setting(b) Systemic Risk Board information

Systemic risk identification √ √ Via link to European √ √

Systemic Risk Board

(a) For the purposes of this table, microprudential policy is defined as helping to set capital buffers for individual institutions and helping authorities judge whether institutions need to adjust their capital plans.
(b) For the purposes of this table, macroprudential policy is defined narrowly as helping to set system-wide capital buffers. The Bank of England has been explicit about the link between its stress tests and the setting of system-wide
buffers.
140 Quarterly Bulletin 2016 Q3

Figure 4 Illustrating potential developments in current stress testing

As stress testing develops further,
policymakers are likely to try to
explore risks emanating from a Bank A:
wider range of sources… Projected balance sheet
At the same time, stress tests can
Scenario 1: be expected to develop to better
Domestic recession capture the feedback and
amplification mechanisms that
Bank A: operate both within the banking
Projected profitability system…

Scenario 2:
Impact on Bank A Impact on banking system
Global financial crisis

Bank A:
Liquidity position
….to give a more complete picture
of the resilience of individual banks.

Scenario 3:
Funding market shutdown
Impact on
Impact on
Bank A: broader
wider
Access to funding financial
economy
system

….and measure the impact over a ….and between the banking system,
broader range of resilience the broader financial system, and the
metrics… wider economy.

to satisfy such withdrawals. Funding resilience, on the other time, authorities are likely to strive to make stress tests more
hand, measures the sustainability of a bank’s funding profile. systematic and automated to make the implementation of
In times of stress, a bank’s ability to roll over and raise funding multiple scenario approaches more feasible.
may become impaired such that it is unable to raise sufficient
funds, or can only do so at a much higher cost. Against these potential cost savings, stress-testing authorities
may have to weigh the risk of making tests easier for
Seeking to capture liquidity and funding resilience more participating banks to predict. This could focus banks’
completely within concurrent stress-testing frameworks attention away from conducting a holistic risk management
would give a more complete picture of banks’ resilience to the exercise towards simply passing stress tests. The qualitative
stress scenario. While some authorities already operate reviews of bank risk management practices run by many
liquidity and funding stress tests, they are generally less regulatory authorities in conjunction with their stress-testing
advanced (BCBS (2013)). And where regulatory authorities do exercises could, however, help to mitigate this potential
perform liquidity, funding and solvency stress tests, they tend downside of further automation.(2)
to operate them independently. Such an approach may miss
the interconnections that exist between these different Integrating amplification and feedback mechanisms
resilience metrics. For example, when a bank does not have and incorporating behavioural responses into stress
sufficient liquid assets or is unable to raise sufficient funds, it tests
may be forced into selling longer-term, illiquid assets at Stress tests have arguably the greatest potential to add value
discounted prices to raise cash. In extreme cases, the losses from a macroprudential policy perspective through illustrating
made on such sales may bring the bank’s solvency into how a stress could impact the financial system as a whole
question.(1) (Demekas (2015)). From this perspective, it is often feedback
and amplification channels that prove important in driving
Authorities are also likely to try to explore risks emanating contagion losses and exacerbating the impact of an initial
from a broader range of sources. At present, the concurrent shock (Constancio (2015)).
stress-testing frameworks of most regulatory authorities
incorporate a single or dual scenario approach. But since any For example, during the global financial crisis, significant losses
individual scenario is very unlikely to materialise, there is a at individual banks eroded their loss-absorbing capital
case for trying to increase the flexibility of stress tests such resources and brought into question banks’ ability to continue
that a greater number of scenarios can be explored in any to meet their regulatory capital requirements. Market
given year. A multiple scenario approach also offers the uncertainty over the solvency of different banks led to strains
flexibility to explore risks that may present new and emerging in bank funding markets, impairing banks’ ability to raise funds.
threats to financial stability, as well as testing bank resilience
against a more regular and well understood set of risks.

This more flexible approach could significantly increase the (1) See Farag, Harland and Nixon (2013) for further explanation of how solvency and
resource burden stress testing places on both participating liquidity can interact in this way.
(2) For more details on the Bank of England’s qualitative assessment of banks’ risk
banks and regulatory authorities. As a consequence, over management and planning capabilities, see Section 2.5 of Bank of England (2015).
 Topical articles Stress testing of banks: an introduction 141

In such circumstances, the behavioural responses of the While macroprudential authorities are already engaged in
banks themselves become another important feedback analysis of interconnections between different parts of the
channel. If all banks respond to funding market strains by financial system, no authority has yet undertaken a
looking to substitute wholesale funding for retail deposits, comprehensive system-wide stress test. Such a stress test
increased competition in retail deposit markets is likely to would seek to incorporate other financial institutions,
drive up interest rates. This, in turn, is likely to have an including central counterparties, hedge funds, insurers and
adverse impact on bank profitability. Gaining a better money market funds. The precise form of a system-wide
understanding of such feedback and amplification channels, stress test might well differ from bank stress tests. But it
and the role they play in driving contagion losses and could contribute to an improved understanding of how shocks
contributing to systemic risk is a key priority for policymakers. can propagate through the financial system giving rise to
systemic risk, which might also be reflected in banks’ losses.
Policymakers have a comparative advantage over individual
banks in this area because they are able to access projections Extending the reach of stress testing beyond the core banking
across stress-test participants. This enables them to take a sector would help in guarding against any perverse incentives
view of the broader market conditions that might prevail stress tests and broader bank regulation creates for
during the stress, and judge the feasibility of individual banks’ institutions to move activities outside the core banking sector
proposed responses in light of this. Several authorities have into the so-called ‘shadow banking sector’. It may also inform
already made efforts to incorporate behavioural responses policymaking to help ensure that banks and other regulated
into their stress-testing frameworks, through the use of entities are resilient to contagion risk emanating from
dynamic balance sheets and consideration of management unregulated financial institutions. To this end, the Financial
actions (see the box on pages 138–39). But there remains a Stability Board (FSB) recently recommended that regulatory
need for frameworks to assess the consistency of such actions, authorities should give consideration to system-wide stress
and analysis of other feedback and amplification channels is testing (FSB (2016)).
still at an early stage.
Conclusions
Extending the scope of stress tests beyond the core
banking sector Since the 1990s, stress tests have evolved from being a risk
As the financial crisis demonstrated, interconnections between management tool used by banks on specific portfolios, to
different parts of the financial system can serve to transmit being widely used by authorities as a regulatory tool, covering
stresses originating in a particular market segment across the large banking systems. Authorities engaged in concurrent
broader financial system, amplifying the effects of the initial stress testing have a wide range of differing objectives and to
shock. A stress test investigating interconnections between some extent this is reflected in the different stress-testing
banks and the wider financial system could explore both direct practices they have adopted. For example, some without
links (through financial transactions) and indirect links capital-setting powers use tests as a means to flag up financial
(through the behaviour of different financial institutions) that stability risks, while others use tests to help to set capital for
have the potential to transmit and amplify shocks. individual banks.

Financial transactions create direct links between both banks For stress testers using results to help set microprudential
and non-bank financial institutions. Repurchase agreements policy, a key area of potential future development is
(‘repos’) are one example of such transactions, and in times of incorporating a greater range of risks in concurrent tests. For
stress the haircuts — the discount applied to the asset used as example, better integrating liquidity and funding risks within
collateral — demanded by buyers in repo transactions tend to solvency-orientated stress tests should provide regulators with
increase, which can exacerbate funding difficulties faced by a more complete view of risks facing individual banks.
other financial institutions.
There are also several ways in which present stress-testing
Even in the absence of direct links, the behaviour of different practices could be augmented to make tests better suited to
financial institutions in a stress could propagate shocks across mitigating system-wide risks. These include integrating
the financial system. For example, some asset managers have amplification and feedback mechanisms into concurrent stress
mandates that prevent them from investing in assets with tests and incorporating behavioural responses, as well as
poor credit ratings. During the recent financial crisis, a extending the scope of stress tests beyond the core banking
significant number of financial assets had their credit ratings sector, to get a broader picture of the likely impact of a stress
downgraded. This led to asset sales on a large scale which on the financial system. A forthcoming IMF paper suggests a
significantly reduced the prevailing market prices for these way forward in that regard for authorities.(1)
assets and forced losses on other institutions holding these
same assets (Deb et al (2011)). (1) Segoviano et al (2017), forthcoming.
142 Quarterly Bulletin 2016 Q3

References

Araten, M (2013), ‘The advancement of stress testing at banks’, in Zhang, J (ed), CCAR and Beyond — Capital Assessment Stress Testing and
Applications, Risk Books.

Bank of England (2006), Financial Stability Report, July, available at

www.bankofengland.co.uk/publications/Documents/fsr/2006/fsrfull0606.pdf.

Bank of England (2015), ‘The Bank of England’s approach to stress testing the UK banking system’, available at
www.bankofengland.co.uk/financialstability/Documents/stresstesting/2015/approach.pdf.

Bank of England (2016), ‘Stress testing the UK banking system: key elements of the 2016 stress test’, available at
www.bankofengland.co.uk/financialstability/Documents/stresstesting/2016/keyelements.pdf.

Barnhill, T and Schumacher, L (2011), ‘Modeling correlated systemic liquidity and solvency risks in a financial environment with incomplete
information’, IMF Working Paper No. 11/263.

Basel Committee on Banking Supervision (1999), ‘Credit risk modelling: current practices and applications’, April.

Basel Committee on Banking Supervision (2009), ‘Principles for sound stress testing practices and supervision’, May.

Basel Committee on Banking Supervision (2013), ‘Liquidity stress testing: a survey of theory, empirics and current practices’,
BCBS Working Paper No. 24.

Bernanke, B (2010), ‘The Supervisory Capital Assessment Program — one year later’, available at
www.federalreserve.gov/newsevents/speech/bernanke20100506a.htm.

Blaschke, W, Jones, M, Majnoni, G and Peria, S (2001), ‘Stress testing of financial systems: an overview of issues, methodologies, and FSAP
experiences’, IMF Working Paper No. 01/88.

Committee on the Global Financial System (2000), ‘Stress testing by large financial institutions: current practice and aggregation issues’,
CGFS Paper No. 14.

Committee on the Global Financial System (2005), ‘Stress testing at major financial institutions: survey results and practice’,
CGFS Paper No. 24.

Constancio, V (2015), ‘The role of stress testing in supervision and macroprudential policy’, available at
www.ecb.europa.eu/press/key/date/2015/html/sp151029.en.html.

De Bandt, O and Oung, V (2004), ‘Assessment of stress tests conducted on the French banking system’, Banque de France Financial Stability
Review No. 5.

Deb, P, Manning, M, Murphy, G, Penalver, A and Toth, A (2011), ‘Whither the credit ratings industry?’, Bank of England Financial Stability
Paper No. 9, available at www.bankofengland.co.uk/financialstability/Documents/fpc/fspapers/fs_paper09.pdf.

Demekas, D (2015), ‘Designing effective macroprudential stress tests: progress so far and the way forward’, IMF Working Paper No. WP/15/146.

Farag, M, Harland, D and Nixon, D (2013), ‘Bank capital and liquidity’, Bank of England Quarterly Bulletin, Vol. 53, No. 3, pages 201–15,
available at www.bankofengland.co.uk/publications/Documents/quarterlybulletin/2013/qb130302.pdf.

Financial Stability Board (2016), ‘Proposed policy recommendations to address structural vulnerabilities from asset management activities’,
22 June 2016.

Haldane, A (2009), ‘Why banks failed the stress test’, available at www.bankofengland.co.uk/archive/documents/historicpubs/speeches/2009/
speech374.pdf.

Hoggarth, G, Sorensen, S and Zicchino, L (2005), ‘Stress tests of UK banks using a VAR approach’, Bank of England Working Paper No. 282,
available at www.bankofengland.co.uk/archive/Documents/historicpubs/workingpapers/2005/wp282.pdf.

Impavido, G (2011), ‘Stress tests for defined benefit pension plans — a primer’, IMF Working Paper No. 11/29.
 Topical articles Stress testing of banks: an introduction 143

International Monetary Fund (2014), ‘Review of the Financial Sector Assessment Program: further adaptation to the post crisis era’.

Krugman, P (2014), ‘Does he pass the test?’, The New York Review, July 10.

McGee, A and Khaykin, I (2013), ‘Financial institution perspectives on the evolving role of enterprise-wide stress testing’, in Zhang, J (ed),
CCAR and Beyond — Capital Assessment Stress Testing and Applications, Risk Books.

Schuermann, T (2013), Stress testing banks, Wharton Financial Institutions Center.

Segoviano, M, Malik, S, Lindner, P and Cortes, P (2016), ‘A comprehensive multi-sector framework for surveillance of systemic risk and
interconnectedness (SyRIN)’, IMF Working Paper, forthcoming.

Segoviano, M, Espinoza, R, Alla, Z and Li, Q (2017), ‘Systemic risk amplifiers in macroprudenital stress testing: introduction to contagion
effects’, IMF Working Paper, forthcoming.

Virolainen, K (2004), ‘Macro stress testing with a macroeconomic credit risk model for Finland’, Bank of Finland Discussion Paper No. 18/2004.

Zhang, J (ed) (2013), CCAR and Beyond — Capital Assessment Stress Testing and Applications, Risk Books.