Sections 3 and 4 should not have left a doubt in the reader’s mind that every credit institution, and every other entity with major financial stakes, must have in place rigorous stress testing policies, procedures, and processes. These should be used in an able manner for the assessment of risk, as well as of capital ade- quacy. As part of its stress testing programme, a bank:
● Must measure its solvency target over the life of all trades, loans, and investments in its portfolio, and
● Compare results against the measure of limits to exposure set by the board, and by regulators under Pillar 2 of Basel II.
Here is the viewpoint of the Bank for International Settlements (BIS), expressed in its 75th Annual Report (2004–2005): ‘Mirroring the development of stress test- ing methodology at the level of individual firms, many central banks are now developing the infrastructure to perform robustness tests of the financial sector as a whole, relying on both micro and macro indicators. Such exercises often combine three elements:
● Macroeconomic models, built to guide monetary policy decisions,
● Models of the financial condition of households and the business sector, and
● Surveys of the potential impact of different scenarios on the performance of financial institutions and markets.’
BIS points out that in some jurisdictions, this infrastructure is used not only to carry out routine assessments of financial sector vulnerabilities for prudential reasons, but also to provide input into decisions concerning monetary policy.
Once in place, stress testing technology lends itself to ad hoc exercises that are more focused on the analysis of specific risks; for instance, risk arising from abrupt decline in asset prices.
Adding to this important reference to polyvalent use of stress test methodology, an article by Olivie Mahul, of the World Bank, published by the Geneva Association, emphasized the contribution stress tests provide in agricultural risk assessment.8 Catastrophe modelling, Mahul says, is an evolving science which assists policy-makers and other stakeholders in managing the risk from natural disasters.
The problem is that existing models, however, mainly focus on the impact of rapid onset disasters such as earthquakes, floods and hurricanes. While such emphasis is important, it is not sufficient; for instance, the assessment paradigm must also be used in connection to slow onset disasters like drought. The catas- trophe risk model Mahul built has four modules:
● A hazard module, defining the frequency and severity of a peril, at a spe- cific location, based on historical events.
● An exposure module, which values assets at risk (like crops, and live- stock), then computes the value for all types of exposures.
● A vulnerability module, quantifying the damage caused to each asset class by the intensity of a given risk event at a site, and
● A damage module, translating losses estimated in the vulnerability mod- ule into monetary loss – that is, the bottomline.
The latter module produces risk metrics like annual average loss and probable maximum loss, providing policy-makers and risk managers with essential infor- mation necessary to be in charge of their risks in future time periods.
Another noteworthy model mentioned in the same paper addresses itself to agri- cultural risk financing, a domain that has so far received much less attention than it deserves. This model deals with that part of risks which cannot be miti- gated with cost-effective preventive measures but, to the contrary, they would be financed through:
● Farmers’ self-retention
● Private financial markets, and
● Governments by means of an appropriate layering of risks.
Three layers are distinguished in this connection: top, mezzanine, and bottom.
The bottom layer of riskincludes high frequency but low impact (HF/LI) events, that affect farmers from a variety of mainly independent happenings. Losses relating to this layer are mainly caused by inappropriate management decisions, including adverse selection problems.
The mezzanine layer of risk includes less frequent but more severe types of exposure, affecting several farmers at the same time. Examples are hail, frost, floods, and drought. In this connection, the private insurance industry has shown its ability to cover resulting losses, but the insurers themselves may be exposed to fairly major aggregate insured losses.
In the top layer of riskare low frequency but high impact (LF/HI) events. These are essentially catastrophic risks that have not yet been properly studied and, even worse, have not been well documented. Yet, their probable maximum loss can be very large, and so is the corresponding insurance premium.
Farmers are usually unwilling to purchase the top layer risk insurance not only because of cost but also because they tend to underestimate their exposure to catastrophic risks. Or, alternatively, they rely on post-disaster emergency relief – which sometimes is unavailable and in other cases it comes too late. By concen- trating on the higher impact of lower frequency events, stress testing can look into mezzanine and LF/HI events:
● Through scenario analysis, or
● By means of simulation.
A banking industry example of hypothetical scenarios involving stress testing by using changes in portfolio value, looks at changes that would occur at end-of-day
positions given a certain level of volatility. Changes in volatility are expressed either through absolute asset values or by percentiles. Through scenarios we also model default or event risk, to achieve greater accuracy in estimates of exposure.
As an example, a simulation involving stress tests may address future changes in economic conditions that could unfavourably impact on the firm’s credit exposures.
Or, they may target an assessment of the bank’s ability to withstand changes such as:
● Economic or industry downturns
● Severe market events, or
● Market illiquidity conditions.
A credit institution must also stress test its gross and net collateral counterparty exposure, including jointly stressing market risk and credit risk factors. Stress tests of counterparty risk should consider concentration risk to a single counter- party, or groups of counterparties, as well as correlation risk across market risk(s) and credit risk(s).
As these examples demonstrate, stress tests are becoming increasingly more sophisticated. According to the Basel Committee, banks using the double-default framework – involving the likelihood of simultaneous failure of obligor and guar- antor – must consider as part of their stress testing framework the impact of a deterioration in the credit quality of protection providers. For instance,
● The impact of guarantors falling outside the eligibility criteria relating to an A-rating, and
● Consequent increase in risk and in capital requirements.
Additionally, stress tests should account for credit risk concentrations that have an adverse effect on the creditworthiness of each of the individual counterpar- ties making up the concentration. Notice that such concentrations are not addressed in Basel II’s Pillar 1 capital charge for credit risk, but they are becom- ing part of Pillar 2. And while credit risk concentrations may be reduced by the purchase of credit protection, banks are well advised to stress test whether the concentration remains because wrong-way risk is greater than that reflected in the calibration of a double default treatment.