Credit Risk

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Stress testing with incomplete data:

a practical guide

Stacia Howard 1

1. Introduction
With the ever-increasing diversification and inter-connectedness of financial systems,
regulators have been dedicating more resources to understanding the relationships within
their financial systems and investigating any inherent vulnerability. As such, a number of
methodologies have been developed to analyse the stability of the financial sector. One of
the more popular ways in which financial stability is assessed is through the use of stress
tests. Stress tests, in the commercial banking literature, refer to assessing the impact of a
rare but plausible shock to the financial system. Countries have to determine whether the
financial institutions to be included in the analysis would conduct their own individual stress
tests and then the regulator would aggregate the submitted data to arrive at a macro stress
test; or whether the regulators would collect the necessary data and conduct one stress test
based on the information received. Based on the stability reports published by various
regulatory bodies throughout the world, the preference seems to be to collect the data to
perform the stress test rather than rely on individual institutions to submit their results.
The first step is usually to determine what risks will be stressed and specifying the scenarios,
as this will assist in determining the methodology to be used and the data requirements. In
most instances historical data is employed to evaluate the sensitivity of commercial banks’
balance sheets to various shocks to macro fundamentals and then utilising the estimated
coefficients to simulate the impact on the financial system of possible stress scenarios in the
future. Three broad techniques have been used to implement the stress testing approach:
(1) time series analysis; (2) panel data regressions, and, (3) structural models (see Sorge
and Virolainen, 2006 for more details).
Time series models are perhaps the simplest technique to apply. Kalirai and Schicher (2002),
Hoggarth and Zicchino (2004) and Delgado and Saurina (2004) estimate models of the
determinants of loan write-offs or non-performing loans. The coefficients from these
regression equations were then employed to assess by how much one of the macro
fundamentals would have to change before the system experiences severe stress. Rather
than focus on one or two indicators of financial stress, Hanschel and Monnin, (2005) develop
a stress index using market price, aggregate balance sheet, non-public information and other
structural data. After estimating the stress index, the authors then try to forecast the index by
using macroeconomic imbalances.
One of the drawbacks of aggregate or time series models is that they aggregate the
microeconomic defaults that lead to financial stress. Panel data regressions can account for
bank-specific factors that may be highly correlated with financial stress at a particular
institution. Bangia et al. (2002) use credit migration matrices (which show the expected
changes in credit quality of borrowers) to provide the linkage between macroeconomic
conditions and asset quality. Using the credit ratings history of 7328 borrowers between 1981
and 1998, largely corporate institutions, the authors attempt to estimate the migration matrix,
or the probabilities of being in a particular debt-rating grade. The stress test is therefore done

1
Economist at the Central Bank of Barbados.

344 IFC Bulletin No 31


by analysing how the portfolio value distribution changes during expansions and recessions.
During recessions, Bangia et al. (2002) estimate that the level of economic capital needed is
30 percent higher than during an expansion year.
Panel regression approaches have also be been employed to evaluate the effects of the
introduction of new regulatory approaches. One of the main concerns regarding the Revised
Framework on bank’s capital agreed by the Basel Committee on Banking Supervision is that
it could potentially impact the level of lending to small- and medium-sized enterprises. Since
small and medium-sized enterprises have higher probabilities of default, banks would have
had to charge higher rates of interest of reduce the amount of lending to these firms in order
to comply with higher capital requirements. Using a large sample of Italian firms, Fabi, et al.
(2005) perform stress tests to compare the change in the overall risk of lending operations
using the new internal ratings based approach and previously utilised techniques. The
authors report that the two measures tend to be highly correlated, suggesting that the new
capital adequacy framework should not significantly influence lending to small- and medium-
sized enterprises. In contrast, Peura and Jokivuolle (2004), also using a stress testing
framework to evaluate the effects of the new rating-sensitive capital charges of the Revised
Framework on bank’s capital, argue that the new framework is indeed likely to have a
significant impact. The simulation-based approach conditions the rating transitions on the
business-cycle phase, and provides estimates of actual bank capital and minimum capital
requirements simultaneously. The joint dynamics of these determine the necessary capital
buffers, given a confidence level for regulatory capital adequacy chosen by bank
management. The numerical results indicate that the introduction of rating sensitive capital
requirements will oblige banks to hold higher bank capital buffers than are currently observed
due to the increased volatility of the minimum capital requirement.
de Bandt et al. (2008), in contrast to previous studies, apply the stress testing framework to
investigate the effects of macroeconomic shocks on the corporate segment of the debt
market. The study derives the equilibrium in the corporate debt market in terms of the
interest rate and the volume of debt of non-financial corporations by simultaneously
estimating a supply and demand schedule for debt. The model is estimated using data on
corporate firms in France, Germany, Italy and Spain between 1993 and 2005. Two scenarios
are considered: (1) a significant reduction in world demand and (2) a significant increase in
oil prices. The results from the stress test suggest that the equilibrium depends to a large
degree on the change in the default rate: in the second scenario the real debt burden is
about 25 basis points higher relative to the recession scenario, as higher oil prices lead to
higher short-term interest rates and second round effects on inflation trigger a reduction in
real GDP growth.
Structural models attempt to embed loan default models into structural macroeconometric
models. Basu et al. (2006), for example, outline a framework for augmenting the IMF’s
financial programming approach with a module to stress test the financial sector. The
proposed framework can also be used to conduct sensitivity analysis of the aggregated
financial sector to various types of risks, including foreign exchange, interest rate, and credit
risk. Similarly, Drehmann et al. (2005) develops the Bank of England’s Medium-Term
Macroeconometric Model to explain the relationship between the between write-off rates and
liquidation rates for the corporate and household sector based on various macroeconomic
fundamentals.
As a complement to the techniques described above, Čiják, (2007) proposes a framework
that allows the clear mapping of the macroeconomic shock onto the balance sheets of
financial institutions but could also stand on its own and rely more heavily on the judgement
of the analyst. In this way, if econometric techniques are difficult to implement, analysts are
still able to map the impact of a macroeconomic shock and arrive at reasonable results.
Furthermore, this framework permits contagion and feedback analysis, areas that are often
overlooked by other models.

IFC Bulletin No 31 345


It is clear based on this preceding summary that the conduct of stress tests, regardless of the
methodology chosen, requires a great deal of data to adequately examine the various risks
identified for a given financial system. Quite a bit of work has already been done on the
methodologies that may be employed to conduct stress tests of the financial system (see
Blaschke et. al. (2001) and Sorge (2004) for good surveys of existing methodologies). There
is, however, a dearth of work related to the data itself. Often what prevents a country from
including stress tests within their surveillance tools or, if included, prevents a comprehensive
stress test with useful results, is some form of data constraint. For example, the information
needed may not be collected, there are inconsistencies or there are concerns regarding the
accuracy of the data. Countries are often forced to compromise when it comes to data and
this can have a negative impact on the results. For those countries now introducing stress
tests to their supervision operations, there is limited advice available on how to work around
the data constraints. The main goals of this study are to address some of the data problems
encountered when conducting a stress test as well as suggest some possible solutions.
Following this introduction, section 2 of the paper outlines the minimum data requirements of
stress testing. Its seven sub-sections provide discussions of the most common macroeconomic
shocks considered as well as the various risks commonly analysed when conducting stress
tests (namely, interest rate, exchange rate, country, credit, interbank and liquidity risks).
Each sub-section offers a brief description of the methodology used, the resulting data
requirements and constraints and the solutions suggested (wherever possible). Section 3
uses the example of the country of Barbados to show how some of the common data
deficiencies were surmounted. The final section concludes by offering suggestions on the
direction of future work in the area.

2. Minimum data requirements

2.1 Macroeconomic Shocks


Before delving into the specifics, there is information that forms the basis of stress testing,
regardless of methodology, without which the entire process becomes next to impossible:
(1) macroeconomic data, and; (2) commercial bank financial statement data. In terms of the
macroeconomic data, the minimum the approach requires are estimates of aggregate
nominal or real GDP. Depending on the methodology to be employed, the explanatory
variables to be stressed and the level of preciseness required for the exercise, information
could also be collected on the following commonly used variables:
• National Accounts data by economic sector
• National Accounts by expenditure
• Inflation
• Nominal exchange Rates
• International reserves
• Unemployment
• Current account balance
• Nominal interest rates
• Foreign direct investment
• Money supply
The impact of these shocks should be transferred onto the financial statements of the
institutions included in the stress test exercise. As such, the analyst would also need to

346 IFC Bulletin No 31


collect financial statement data for the financial system; the level of disaggregation needed
would once again depend on the risks to be analysed. As a guide, the minimum data from
the financial statements should include:
• Economic Capital
• Capital Adequacy Ratio
• Net Income
The remainder of this paper assumes that at least either the economic capital, capital
adequacy ratio or net income figure is available, as any one of these could form the basis of
evaluating the impact of the shock on the viability of the financial institution. The preference
is to use either a capital adequacy ratio or economic capital because there are international
guidelines set forth by the Bank for International Settlements, and this allows for comparison
across countries.

2.2 Interest Rate Risk


Once the basic information requirements have been fulfilled, the main goal of stress testing is
assess the impact of various risks on the financial system. One of the most fundamental risks
to be assessed is interest rate risk. Stress testing for interest rate risk attempts to determine
the impact on financial stability of a change in interest rates through the consequent effect on
interest income/expenses as well as the interest sensitive components of the balance sheets
of financial institutions. Blaschke et. al. (1988) explains that interest rate risk develops as a
result of other risks, namely repricing risk, yield curve risk, basis risk and options risk. As
indicated in Čiják (2007), interest rate risk can be sub-divided into direct and indirect impacts,
with direct interest risk occurring when a change in rates results in a mismatch of interest
rate sensitive assets and liabilities (mainly the risks identified in Blaschke et al (1988)) and
indirect risk referring to the impact of the change in rates on the creditworthiness of
borrowers and their ability to repay. Given this aim, information must therefore be available
on interest rate sensitive assets and liabilities within either individual portfolios or on
aggregate basis as well as fairly detailed information on the clients of the financial
institutions.
In order to address direct risk, a number of frameworks exist ranging from Excel
spreadsheets (see Čiják (2007)) to econometric models (see Sorge, 2004). Nonetheless, the
following information, at a minimum, should be collected from the reporting institutions:
• interest rate sensitive assets, broken down into maturity buckets determined by time
to maturity (including loans, long- and short-term fixed income assets, interest
bearing deposits owned by the institution, etc.);
• interest rate sensitive liabilities, broken down into maturity buckets determined by
time to maturity (including outstanding loan obligations, long- and short-term fixed
income obligations, interest bearing deposits liabilities, etc.), and;
• the settlement date, maturity date, coupon, yield, frequency of coupon payment of
each fixed-income asset held by reporting institutions in order to determine the
average duration of the fixed income assets.
To successfully investigate indirect risk, econometric models would probably be necessary to
do justice to the exercise (see Basu et. al 2006) and further information would need to be
sought on:
• credit ratings for corporate borrowers (Standard & Poors, Moodys and other rating
agencies as well as reporting financial institutions);
• default rates for loan customers, broken down into useful categories (reporting
financial institutions);

IFC Bulletin No 31 347


• historical data on nonperforming loans (reporting financial institutions);
• income growth rates (national labour organisations, trade unions, etc.);
• measures of indebtedness (national surveys, consumer advocacy institutions, etc.);
• degree of collateralisation of loans (reporting financial institutions), and;
• real interest rates.

2.3 Exchange Rate and Country Risk


Exchange rate risk is defined by Papaioannou (2006) as “the possible direct loss (as a result
of unhedged exposure) or indirect loss in the firm’s cash flows, assets and liabilities, net
profit and, in turn, its stock market value from an exchange rate move”. The definition offered
by Blaschke et al (1988) differentiates between direct and indirect impact of exchange rate
risk – a direct impact would result from the institution holding a position in a foreign currency
whereas an indirect impact would result from the effect on the creditworthiness of the
institution’s borrowers or counter-parties – and explains that institutions could also be
impacted if they hold local currency assets that are indexed to foreign exchange rates.
The data requirements for exchange rate measurement are summarised below:
• For each currency
− all asset items, including accrued interest, denominated in the currency in
question
− all liability items, including accrued interest, denominated in the currency in
question
− all amounts to be received under forward exchange transactions, including
futures and the principal on currency swaps not included above
− all amounts to be paid under forward exchange transactions, including futures
and the principal on currency swaps not included above
− guarantees (and similar instruments) that are certain to be called and are likely
to be irrecoverable
− all other items representing a profit or loss in foreign currencies
• Exchange rates for each currency
Country risk is tied to exchange rate risk and is often not tested separately to exchange rate
risk. Nonetheless, it is worth noting because some countries issue debt in a currency other
than their national currency. An institution holding this debt on their books, therefore,
becomes exposed to both the risk associated with the country issuing the debt as well as the
risk emanating from the currency in which the debt is denominated.

2.4 Credit Risk


Credit risk is perhaps the most studied form of financial risk and a number of methodologies
have developed to measure an institution’s exposure. Credit risk can be considered as the
risk of default of the financial institutions’ debtors. An important distinction must be made
between expected and unexpected losses when measuring credit risk, as expected losses
(which are uncertain but occur on average) are covered by provisioning arrangements within
the institution and unexpected losses (which occur with unknown frequency) are not. The
main concern when stress testing, therefore, is the extent of the credit risk stemming from
unexpected losses and some frameworks (such as Čiják (2007)) consider evaluating the

348 IFC Bulletin No 31


provisioning arrangements for expected losses as creating the foundation for the remainder
of the analysis.
Assessing credit risk varies from country to country with the changing characteristics of the
financial system. Accordingly, the data requirements would be broader with highly developed
financial markets than with less developed markets. The variables listed below do not
attempt to address every possible scenario; rather, they represent only those variables likely
to be necessary even in the more underdeveloped financial systems.
• Non-performing loans, preferably by sector
• Performing loans, preferably by sector
• Values for top 5 (minimum) large exposures

2.5 Contagion Risk


Contagion risk refers to “the risk that an initial (bank) failure may spill over to the rest of the
(banking) industry and cause further (bank) failures” (see Schoenmaker). A number of
studies have been dedicated to measuring the extent of contagion risk in the banking sector.
Chan-Lau et al (2007) attempted to determine the likelihood of a large shock to one major
bank causing stress to another large bank using a binomial LOGIT model. They first
established the extent of an individual bank’s default/solvency risk and then calculated
percentage changes in this risk. Banks with corresponding changes were interpreted as
being interdependent and this measure was assumed to have incorporated all potential
channels of contagion. This methodology avoids the need to explicitly identify the various
interbank links and channels of contagion.
Other methods, however, are more data intensive, such as that proposed by Degryse and
Nguyen (2007), where they tested contagion risk using a (N × ( N + M )) matrix of interbank
bilateral exposures, X . Čiják (2007) also relies on quite granular data on interbank exposures
in order to simulate the likely impact of the failure of one institution. The information needed
for these types of studies include:
• Interbank exposures, by bank
• Interbank exposures, by instrument

2.6 Liquidity Risk


According to the Bank for International Settlements, “liquidity is ability of a bank to fund
increases in assets and meet obligations as they come due, without incurring unacceptable
losses”. 2 Given this definition, liquidity risk is the risk of not meeting obligations as they come
due without incurring unacceptable losses. From the point of view of an institution conducting
its own internal investigation into liquidity risk, all of the necessary information on obligations
as well as expected increases in asset categories would be available. From a supervisory
perspective, however, much of this detail is absent and would be time consuming to collect
and analyse on a frequent basis. As such, examiners often have to rely on the institution to
adequately account for liquidity risk, based on authority-determined definitions of liquid
assets. Assumptions are then made in order to deal with the varying methodologies that may
be employed by different institutions. It is difficult to streamline these definitions across
countries given the varying market structures; however, the International Monetary Fund

2
Principles for Sound Liquidity Risk, Basel Committee on Banking Supervision, BIS.

IFC Bulletin No 31 349


(IMF) considers liquid instruments as those that can be sold at, or close to, full market value
on short notice. 3 The information that should be collected would therefore depend on the
national definition of liquid assets/liabilities, but at a minimum would include:
• Cash
• Treasury bills
• Transferable deposits

3. A case study of Barbados

3.1 Introduction
Barbados is a small island economy in the Caribbean of about 274,000 inhabitants. It is an
open economy and therefore is fully exposed to the risks tied to international developments.
In addition, its membership of the CARICOM Single Market and Economy (CSME) exposes
the country to the risks common in high levels of regional integration. The existence of both
international and regional financial institutions in the country’s financial sector also adds
another level of global exposure. As with many countries in the world, the Barbadian financial
sector is heavily bank-based, with commercial banks accounting for 66 percent (2006) of
total financial system assets. In addition to the commercial banks, there are merchant banks,
trust and finance companies, mortgage finance companies, credit unions, pension funds,
insurance companies, asset management companies and a stock exchange. The activities of
these institutions are monitored by four regulatory bodies: the Central Bank of Barbados
(commercial banks, merchant banks, trust and finance companies and mortgage finance
companies), the Department of Cooperatives (credit unions), the Supervisor of Insurance
(insurance companies and pension funds), and, the Securities Commission of Barbados (the
stock exchange and associated brokerage firms).
The country of Barbados was chosen as the case study for a number of reasons. Firstly, it is
one of the highest ranked developing countries in the world (as per the Human Development
Index of the UNDP) and, as such, is dealing with some of the financial problems faced in
advanced economies while still hampered by the institutional framework of many developing
countries. In the opinion of the author, this offers a good example of what the average
developing country is facing, or could face in the near future, in their financial systems.
Secondly, it is a small open economy and is therefore exposed to international financial
developments with limited scope to impact global developments. Stress testing such an
economy requires a slightly different approach than commonly found in the existing literature
and often calls for additional data that may not be incorporated in the stress tests of
advanced economies. Thirdly, as part of a common market with goals of a monetary union –
the CARICOM Single Market and Economy (CSME) – it is part of a regional integration
movement similar to that of the European Union and this presents its own set of challenges,
especially with respect to the financial sector. Fourthly, the country is currently involved in an
effort to improve its monetary and financial statistics, with the assistance of the International
Monetary Fund, and the findings thus far may be of use to other countries.
As part of its surveillance of the financial systems, the CBB conducts stress tests in
conjunction with the IMF and World Bank during the Financial Sector Assessment
Programmes (FSAP) conducted by these international financial institutions. For the last
FSAP (in 2008), two different stress tests frameworks were utilised by the CBB: (1) a macro

3
Monetary and Financial Statistics Manual, International Monetary Fund.

350 IFC Bulletin No 31


framework using the top-down approach, and; (2) a more micro stress test using an
adaptation of the framework developed by Martin Čiják. The macro framework is linked to the
CBB’s financial programming model and so is dependent on the forecasts in the financial
programming model. This obviously allows for a consistent view of the future of the economy
as a whole, with the stress test model operating like a satellite model; but it also presented
some challenges because it becomes dependent on the forecasts. The micro framework has
the advantage that it could be linked to the financial programming model, but it could also
stand alone and still present solid results.
While the first framework was used by the CBB in previous FSAP missions, the second
framework was tested internally for the first time in 2008 and its results compared with the
first methodology to check for robustness. Both methodologies called for some of the same
data but the “micro” framework had additional data requirements. Although the more
information available the more all-encompassing the results are likely to be, regardless of
framework, a trade-off must be made between accuracy and computational difficulty and this
balance must be borne in mind by countries when designing their data collection process.
For the purpose of this paper, more attention will be paid to the micro framework given that it
had the greater data requirements.

3.2 Description of stress test framework

3.2.1 Introduction
The framework chosen is adapted from Čiják (2007) and is built in Microsoft Excel.
This basic analysis was augmented using a technique formalised in Worrell (2008). For each
macroeconomic shock, the variable was stressed until the system failed to ascertain how
long or how much stress it would take before systemic failure. This is a useful exercise
because it permits policymakers a further gauge of the resilience of the financial system as
well as contributing to the development of early warning systems.
The Barbadian banking sector has six banks, none of which are domestically owned. For
three of the institutions, there was no useful estimate of capital, with the largest bank
(measured by total assets) included. Given that the results of stress tests tend to be
expressed in terms of their impact on capital, the absence of this information for half of the
institutions and more than half of the assets of the banking sector presented a significant
challenge. In order to surmount this problem and still arrive at useful results, the results of
the stress tests were expressed in terms of their impact on the return on assets of the
institutions and as a percentage of nominal GDP. In this way, the macroeconomic as well as
profitability impacts of the various vulnerabilities could be easily identified, communicated
and understood. It, still however, does not allow for an assessment of whether the stress was
sufficient to cause the bank to fail. Therefore, comparisons were made to times in the past
when similar levels of loss or bank failures were experienced to give an idea of the severity
of the stress.

3.2.2 Interest rate risk


The tests for interest rate risk incorporated maturity gap analysis, net interest income impact
and repricing impact of various changes in the level of nominal interest rates. It required
information on:

Interest rate sensitive assets, broken down by time to maturity


The existing data collection forms utilised by the CBB only requested a breakdown of loans
by time to maturity. A breakdown of the other interest rate sensitive assets by time to
maturity was not available from the reporting institutions. The maturity dates of government
securities, however, were available from the department of the CBB that deals with

IFC Bulletin No 31 351


government debt (this information could also be sourced directly from government either
through statistical digests or a direct request). Other interest rate sensitive assets were not
available by time to maturity so, in order to be as conservative as possible, these assets
were originally assumed to fall into the most liquid category. This gave an unrealistically
pessimistic view of the interest rate risk of the banks, given the proportion of these assets to
total assets, and so adjustments were made following discussions with bank examiners to
give a more realistic approximation.

Interest rate sensitive liabilities, broken down by time to maturity


The existing data collection forms utilised by the CBB requested a breakdown of deposits by
original maturity and this was used as a proxy for deposits by time to maturity. A breakdown
of the other interest rate sensitive liabilities by time to maturity was not available from the
reporting institutions. Therefore, in order to be as conservative as possible, these assets
were assumed to fall into the most liquid category. Similar to the treatment of similar missing
asset data, discussions were held with the Bank Supervision Department and adjustments
were made to give a more realistic approximation.

The settlement date, maturity date, coupon, yield and frequency of coupon payment of each
fixed-income asset held by reporting institutions
As corporate bonds are rare in the Barbadian financial system, the data focused on
government paper as it represented the vast majority of fixed income assets of the banks
(based on conversations with commercial banks). As such, the information was sourced from
the department within the CBB that deals with government debt but this information could
also have been sourced directly from either government or the financial institutions included
in the sample. The settlement date, maturity date, coupon, yield and frequency of coupon
payment were collected for each fixed income asset and each bank. Based on this data, the
duration was calculated for each instrument, and then averaged for each bank.

3.2.3 Credit risk


The tests for credit risk incorporated tests for underprovisioning, broad-based increases in
non-performing loans, sectoral shocks to non-performing loans and default of large
exposures. It required information on:

Performing loans (pass loans and special mention loans) and non-performing loans
(substandard loans, doubtful loans and loss loans)
This data was already being collected by the CBB.

Provisioning rates for each category of loan


This data was already being collected by the CBB.

Sectoral distribution of total loans (eg agriculture, tourism, manufacturing etc.)


This data was already being collected by the CBB.

Sectoral distribution of non-performing loans (eg agriculture, tourism, manufacturing etc.)


A framework to collect this data had been distributed to the reporting institutions, but data
had not yet been received from all of the institutions. 4 However, information had been

4
The CBB requested reporting institutions to submit this information from the beginning of their financial year.
This date varied from one institution to another.

352 IFC Bulletin No 31


received from two institutions up to the time of conducting the stress test and was included.
The information for the remaining four institutions was estimated based on the figure
submitted for total non-performing loans and discussions with the bank examiners.

Values of top five large exposures


This data was already being collected by the CBB.

3.2.4 Liquidity Risk


The tests for liquidity risk incorporated tests for a liquidity run on all banks and a flight to
safety test. It required information on:

Total deposits, divided into demand (demand and savings deposits) and time deposits
This data was already being collected by the CBB.

Total liquid assets


In the case of Barbados, the CBB currently defines commercial banks’ liquid assets as cash,
assets due from the CBB, assets due from other local banks and government treasury bills.
All of this information was already being collected by the CBB. A framework to collect more
precise data on liquid assets (ie all assets with a maturity of less than 12 months) is being
designed within the CBB and would provide a more accurate estimate of liquid assets.

3.2.5 Exchange Rate Risk


The tests for exchange rate risk incorporated tests for both direct and indirect exchange rate
risk. It required information on:

Total foreign currency denominated assets, by currency


The CBB only collected information on assets by country. This was used as a proxy for
assets by currency since most of the assets by country are denominated in the currency of
the country. A framework has been developed to collect both a country as well as a currency
breakdown given that there are some assets that are denominated in a currency other than
the national currency of the country.

Total foreign currency denominated liabilities, by currency


The CBB collected information on total foreign currency deposits. Given the fixed exchange
rate regime of the country – the Barbados dollar is pegged 2:1 to the US dollar – the vast
majority of foreign currency deposits are held in US dollars. Therefore, all foreign currency
deposits were assumed to be held in US dollars as a breakdown was not available.
Information on total liabilities by country was available and this was used as a proxy for
liabilities by currency since most of the liabilities are denominated in the currency of the
country. Similar to foreign assets, a framework has been developed to collect both a country
as well as a currency breakdown given that there are some liabilities that are denominated in
a currency other than the national currency of the country.

Foreign Currency Loans


This data was already being collected by the CBB.

3.2.6 Contagion Risk


The tests for contagion risk incorporated tests for pure contagion risk (this would result from
selected institutions failing to meet their interbank obligations) and macro contagion risk

IFC Bulletin No 31 353


(which would result from the negative impact of a macroeconomic shock). It required
information on:

Interbank borrowing and lending by creditor and debtor


This data was already being collected by the CBB.

4. Conclusion
The data requirements of stress testing are one of the main hindrances for many countries
when trying to assess the vulnerabilities within their financial systems. This paper attempted
to outline the main variables for which data needed to be collected in order to measure the
most commonly analysed risks. In addition, it gave examples of how some data constraints
could be surmounted using Barbados as a case study. Further work, however, needs to be
done in the area of coming up with proxies for missing data, especially financial statement
data, as the implementation of new data collection forms is a complicated process that often
takes more time than is available. In addition, measurement of equity and commodity price
risks should be incorporated into future work.

References
Bank for International Settlements (2001) “A Survey of Stress Tests and Current Practice at
Major Financial Institutions” Report by a Task Force established by the Committee on the
Global Financial system of the central banks of the Group of Ten countries.
Bank for International Settlements (1999) “Credit Risk Modelling: Current Practices and
Applications” Basle Committee on Banking Supervision.
Bangia, A., Diebold, F.X., Kronimus, A., Schagen, C. and Schuermann, T. (2002) “Ratings
Migration and the Business Cycle, with Application to Credit Portfolio Stress Testing” Journal
of Banking and Finance 26: 445–474.
Basu, R., Choueiri, N. and Pascual, A. G. (2006) “Financial Sector Projections and Stress
Testing in Financial Programming: A New Framework” IMF Working Paper WP/06/33.
Basurto, M. A. S and Padilla, P. (2006) “Portfolio Credit Risk and Macroeconomic Shocks:
Applications to Stress Testing Under Data-Restricted Environments” IMF Working Paper
WP/06/283.
Blaschke, W., Jones, M.T., Majnoni, G. and Martinez Peria, S. (2001) “Stress Testing of
Financial Systems: An Overview of Issues, Methodologies and FSAP Experiences” IMF
Working Paper WP/01/88.
Boss, M., Krenn, G., Puhr, C. and Summer, M. (2006) “Systemic Risk Monitor: A Model for
Systemic Risk Analysis and Stress Testing of Banking Systems” Financial Stability Report 11,
Oesterreichische Nationalbank.
Central Bank of Barbados (2006) Annual Statistical Digest 2006.
Chan-Lau, J. A., Mitra, S. and Ong, Li Lian (2007) “Contagion Risk in the International
Banking System and Implications for London as a Global Financial Center” IMF Working
Paper WP/07/74.
Čiják, M. (2004) “Stress Testing: A Review of Key Concepts” CNB Internal Research and
Policy Note /2/ Czech National Bank.

354 IFC Bulletin No 31


______ (2006) “How Do Central Banks Write on Financial Stability” IMF Working Paper
WP/06/163 International Monetary Fund.
______ (2007) “Introduction to Applied Stress Testing” IMF Working Paper WP/07/59
International Monetary Fund.
Degryse, H. and Nguyen, G. (2007) “Interbank Exposures: An Empirical Examination of
Contagion Risk in the Belgian Banking System” International Journal of Central Banking,
June 2007, pp. 123–171.
Chan-Lau, J.A. and Santos, A. O. (2006) “Currency Mismatches and Corporate Default Risk:
Modelling, Measurement and Surveillance Applications” IMF Working Paper WP/06/269.
Fender, I., Gibson, M. S. and Mosser, P. C. (2001) “An International Survey of Stress Tests”
Current Issues in Economics and Finance, Vol. 7 No. 10 Federal Reserve Bank.
Haldane, A., Hall, S. and Pezzini, S. (2007) “A New Approach to Assessing Risks to
Financial Stability” Financial Stability Paper No. 2 – April 2007, Bank of England.
Harrison, R., Nikolov, K., Quinn, M., Ramsay, G., Scott, A. and Thomas, R. (2005) “The Bank
of England Quarterly Model” Bank of England.
International Monetary Fund (2000) “Monetary and Financial Statistics Manual” International
Monetary Fund.
Jenkinson, N. (2007) Comments on “Developing a Framework for Stress Testing of Financial
Stability Risks” at ECB High Level Conference on “Simulating Financial Instability”.
Jobst, A. (2007) “Consistent Quantitative Operational Risk Measurement and Regulation:
Challenges of Model Specification, Data Collection and Loss Reporting” IMF Working Paper
WP/07/254.
Jones, M. T., Hilbers P. and Slack, G. (2004) “Stress Testing Financial Systems: What to Do
When the Governor Calls” IMF Working Paper WP/04/127.
Papaioannou, M. (2006) “Exchange Rate Risk Measurement and Management: Issues and
Approaches for Firms” IMF Working Paper WP/06/255.
Schoenmaker, D. “Contagion Risk in Banking” www.imes.boj.or.jp/cbrc/cbrc-03.pdf.
Sorge, M. (2004) “Stress-testing Financial Systems: An Overview of Current Methodologies”
BIS Working Paper No. 165.
Worrell, D. (2001) “Price Volatility and Financial Instability” IMF Working Paper WP/01/60.
_______ (2004) “Quantitative Assessment of the Financial Sector: An Integrated Approach”
IMF Working Paper WP/04/153.
_______ (2008) “Stressing to Breaking Point: Interpreting Stress Test Results” IMF Working
Paper WP/08/148.

IFC Bulletin No 31 355

You might also like