Transcript 6 Stylized steps for developing a macro stress testing framework

MACRO STRESS TESTING
Charles AugustineAbuka
Introduction
• One of the key issues in the recent financial crisis was that stress
testing was not an integral part of regulators’ decision-making
process.
• Stress testing exercises conducted by central banks in most of the cases
failed to reveal systemic risks nor led to any macroprudential actions.
• During the global financial crisis many of the central banks moved
toward an integrated approach and designed full scale macro scenario
based stress testing exercises that involved central banks, supervisory
agencies and market participants.
• Countries in the COMESA region need to develop integrated stress
testing frameworks for use in forward looking risk based financial
stability analysis.
1
Introduction
• Stand alone models are not sufficient for successful completion of
stress testing exercises;
• A comprehensive framework and cooperation among various
departments is a key prerequisite to success.
• Integrated stress testing frameworks allow staff to make better
judgment about capital adequacy of the whole banking system and of
individual banks in particular.
• The integrated stress testing frameworks integrate the whole stress
testing process, from scenario design to use of results for micro and
macroprudential purposes.
• The key benefit of integrated approach lies in its holistic view:
– Risk identification,
– Scenario design,
– Application in stress testing models and justification of policy responses.
2
MACRO STRESS TESTING FOR
COMESA COUNTRIES
3
Introduction
• Stress-testing as a framework for forward-looking
analysis of systemic risks has developed great
appeal among central banks across world.
• Why should Macro Stress testing be conducted?
– Assesses vulnerability of a financial system to
‘exceptional but plausible’ macroeconomic shocks,
– Part of the arsenal of forward looking macroprudential
tools.
4
• Experience of Macro Stress Testing in
COMESA
 Top down approach
 Bottom up approach
– The stress testing frame work: The five stage process
– Initial shock -> interaction of risk drivers-> mapping
defaults into losses, process of borrower defaults,
earnings of the financial sector-> impact on P&L and
capital of the financial sector <-> Feedback effects on
economy and markets.
6
Stylized steps for developing a macro stress testing
framework
• Step 1: Assessing pre-requisites,
• Step 2: Deciding the framework for calculations,
• Step 3: Determining the risks and the corresponding size of
shocks,
• Step 4: Expanding to assess inter-linkages including crossborder risks,
• Step 5: Forming a stress testing working group (STWG) ,
• Step 6: Deciding the frequency of conducting stress tests,
• Step 7: To develop a macro-financial stress testing model in
the long run.
8
Stylized steps for developing a macro stress
testing framework
• Step 1: Assessing pre-requisites
 Who will conduct stress-testing?
 Is data available?
 What is the frequency of data available?
 What are the data gaps?
 How are scenarios designed and assumptions
underlying them chosen?
 How are results interpreted and acted upon?
 Lack of data is a major hindrance to stress testing in
COMESA Countries
9
Stylized steps for developing a macro stress
testing framework
• Step 2: Deciding
calculations
the
framework
for
• Where data allows, econometric regressions can be run to estimate
elasticities of banking performance variables such as NPL ratios to the
changes in macroeconomic variables, primarily GDP.
• In developing countries with major structural upheavals in the recent
past or with a fledgling financial industry may not be able to ascertain
any robust relationships between macroeconomic and banking
variables.
• In these cases, it is beneficial to develop a longer time series of data
over the long term while focusing on micro stress testing efforts in the
short term.
10
Stylized steps for developing a macro stress
testing framework
• Step 3: Determining the risks and the
corresponding size of shocks,
• The sizes of shocks used for scenarios should be calibrated using
historical and/or ad-hoc assumptions.
• Shock parameters for market risk can be calibrated using historical
data. A multiple of the standard deviations are good ad hoc shocks to
interest rates and exchange rates.
• The size of shocks applied should be sufficiently severe yet plausible.
• Liquidity and credit risks can be combined into a single shock scenario
to make the simulated crisis more realistic.
11
Stylized steps for developing a macro stress
testing framework
• Step 4: Expanding to assess inter-linkages
including cross-border risks.
• A common COMESA stress testing exercise might be meaningful once
the level of financial integration and inter-linkages among financial
institutions substantially increases from the current level.
• Financial inter-linkages and cross-border exposure of banks within
each economy have to be taken into account. These are still very low.
Direct contagion risk in the region is low, albeit “psychological
contagion” is possible.
• Therefore, at this stage, COMESA economies should focus on
developing a framework for assessing risk on interconnection and
contagion within their own financial sectors, rather than regionally.
12
Stylized steps for developing a macro stress
testing framework
• Step 5: A Stress Testing Working Group
(STWG).
• A WG consisting of representatives of all the relevant departments in
the central bank and the other regulatory authorities may be
established with the following primary tasks:
– Development of methodology, scenarios and aggregation of results.
The WG prepares two scenarios: Baseline and Adverse.
– During the STWG’s meetings data are discussed and shared among
members.
– The STWG also decides about common risk factors for the region as
well as individual countries and proposes common adverse
scenario, designs common stress testing templates.
13
Stylized steps for developing a macro stress
testing framework
• Step 6: Deciding the
conducting stress tests.
frequency
of
It is advisable to conduct liquidity stress tests on a monthly basis and
credit stress tests on a semi-annual basis.
• Credit risk does not tend to change as quickly as liquidity risk.
Therefore, conducting credit risk stress testing on a monthly or
quarterly basis, as is currently the practice in some countries, may not
add much value.
• Instead, it absorbs resources that could be used in analyzing and
developing a more sophisticated and integrated stress testing
framework.
•
14
Stylized steps for developing a macro stress
testing framework
• Step 7: To develop a macro-financial stress
testing model in the long run.
• Countries need to develop a macro-financial stress testing models
using their national macro forecasting model projections as inputs into
it.
• Problems with quality of macro and absence of real estate, household
and corporate financial accounts data prevents the COMESA countries
from developing sophisticated macro financial stress testing models.
– Their results are usually not robust
– In many cases can only be used to explain past behaviour of NPLs but not to forecast
future developments.
• However macro financial stress testing models can be used for storytelling, i.e. describe shock transmission channels.
15
An Initial framework for Stress testing
banks
• To implement this framework staff in COMESA member central banks
need to have a broad understanding of their economies and their
financial sectors.
• They have to identify a number of vulnerabilities in their banking
systems.
• They should be able to construct a relevant set of trigger events to
probe the importance of these vulnerabilities.
• The framework employs a top-down approach to stress testing.
Balance sheet and income statement data should be available for the
largest banks in the country, which account for highest proportion of
total banking assets.
• Initially countries can focus on four main financial risk factors –
interest rates, exchange rates, credit quality, and liquidity.
16
An Initial framework for Stress testing
banks
• A set of macro shocks are translated automatically into changes to
bank balance sheets.
– Changes in interest rates and exchange rates have a direct effect on banks’ balance
sheets,
– The sign and size of these changes depend on the characteristics of banks’ portfolios.
• Changes in interest rates, exchange rates, GDP growth, and the
unemployment rate also have indirect effects on bank balance sheets as
a result of any change in credit quality.
• The results of the stress test can be summarized with graphs that:
– (i) Compare the post shock capital adequacy ratios (CAR) of various categories of
banks with their CARs under the pre-shock scenario;
– (ii) Present the amount of capital as a percent of GDP that will be necessary to
recapitalize banks;
– (iii) Assess the systemic importance of the test by illustrating the concentration of
“bad assets” in the banking system, by bank category
17
Mapping Macro Shocks to Changes in
Credit Quality
• Mapping changes in the macro environment into changes
in bank capital is a complex process. Why?
– Very detailed information on the asset and liability
positions of banks is necessary in order to understand
how changes in market interest rates and exchange
rates will affect bank portfolios.
– Additional information about borrowers (amount and
quality of collateral, hedging strategies, income and
wealth) is needed to understand their ability or
willingness to meet loan obligations.
– One also needs to understand how banks and bank
supervisors will react to adverse economic changes.
18
The Integrated Macro Stress Testing
Framework
• The integrated stress testing frameworks could consist of
the following elements :
A STWG
The definition of purpose and use of stress tests.
The frequency and coverage of the financial system.
The assumptions and scenario design.
The estimation of links between shocks and losses.
Calculation of capital adequacy.
Identification of the transmission channels and feedback effects on
financial system.
– The macroprudential policy implications.
– The communication policy.
–
–
–
–
–
–
–
19
The Integrated Macro Stress
Testing Framework
• The primary users of stress testing results are the
senior management of the macroprudential
authority that makes policy.
• Senior policy makers should be engaged in stress
testing process:
– They should rely on its results while making
macroprudential and Microprudential policy
decisions.
• Stress testing should be aimed at policymakers’
needs and provide support to their decisions.
20
The Integrated Macro Stress
Testing Framework
• Macro scenario design should focus on these
issues:
– Identification of structural weaknesses in
the economy,
– Identification of global market trends,
– Identification
of
weak
individual
institutions.
21
The Integrated Macro Stress
Testing Framework
• Elements of a credible scenario design include multiple
risk factors and assumptions about individual risk factors
pertinent for each institution included into exercise:
– Credit risk – the impact of macro shocks on loan losses
– Market risk – the impact of changes in the exchange
rate, interest rate and asset prices
– Liquidity risk – the impact of inflows and outflows of
assets
– Individual risks – large exposures, asset concentration,
balance sheet behaviour
22
The Integrated Macro Stress
Testing Framework
• Building Satellite models
– Linking banks losses and income to various
micro and macro factors.
– For example linking NPLs with macro factors
such as GDP growth and lending rates.
• Calculation of Capital Adequacy and
interpretation of results.
23
The Integrated Macro Stress
Testing Framework
• Further areas of improvement include:
– Development of capacity to implement Next Generation and other
Balance sheet based stress testing models:
– Better scenario design and use of satellite models to get stressed
NPLs;
– Implementing scenarios based on real estate price shocks.
Countries need to collect real estate prices,
– Building satellite models for profitability, interest income and
credit growth,
– Improving liquidity stress testing and design of new reporting
framework.
• Countries need to consider building joint credit and liquidity stress testing
scenarios.
24
The Integrated Macro Stress
Testing Framework
• The additional efforts that COMESA member states need to implement
integrated macro stress testing include:
– Formation of Stress Testing Working Groups,
– Strengthening the involvement of high level policy makers in
approval of scenarios and results,
– Gathering more data,
– Regular meetings between Financial Stability Units primary
sources of data,
– More effort required in mapping systemic vulnerabilities in the
financial system. Scenarios have to be designed incorporating
feedback effects and multiple sources of risk,
• Modelling efforts should be deepened to include profit and
credit growth equations;
• Liquidity and credit risk stress tests should be linked;
25
The Integrated Macro Stress
Testing Framework
– Member state central banks sould define
potential list of macroprudential policy tools
which can be applied once stress tests reveal
vulnerabilities in the financial system;
– Communication policy needs to be designed to
inform what goes into the Financial Stability
Reports.
26
References
Haldane, A., Hall, S. and S. Pezzini, 2007, “A New Approach
to Assessing Risks to Financial Stability”, Bank of England
Financial Stability Paper No. 2 (London: Bank of England).
Moretti, Stolz and Swinburne (2008) ‘Stress Testing at the
IMF’, IMF Working Paper
Oosterloo S. et al. Financial stability reviews: A first
empirical analysis. Journal of Financial Stability. Volume 2,
Issue 4, March 2007, Pages 337-355.
Ong. L, Maino. R, Duma. N, (2010) ‘Into the Great Unknown:
Stress Testing with Weak Data’ IMF Working Paper WP/10
27
References
Oosterloo, S., J. de Haan, and R. Jong-A-Pin, 2007, "Financial Stability
Reviews: A First Empirical Analysis", Journal of Financial Stability, No.
2, pp. 337–355.
Quagliariello, M. (edited by) (2009), ‘Stress-testing the Banking System,
Methodologies and Applications’, Cambridge University Press.
Sorge, Marco, and Kimmo Virolainen (2006). "A comparative analysis of
macro stress-testing methodologies with application to
Finland." Journal of financial stability2.2 (2006): 113-151.
Van den End, Hoeberichts and Tabbae (2006), ‘Modelling Scenario
Analysis and Macro Stress-testing’, Working Paper No. 119/November
2006, De Nederlandsche Bank
Wilson (1997), ‘Portfolio Credit Risk’, Risk, vol. 10, Issues 9-10.
28