Stress Testing: Adding Value
Basel II focused attention upon stress testing. The UK’s Financial Services Authority (FSA) has also published its ‘Financial Outlook for 2007’, which further emphasises the importance of its use. But, away from the regulatory clamour, investment banks are discovering new and innovative uses for this risk management tool. Beyond simply illustrating a bank’s financial position following an extreme market event, stress testing has been embraced in a variety of areas, such as influencing a bank’s overall risk appetite, limit setting and economic capital allocation. With the correct processes, stress testing exemplifies the manner in which the market risk arm of a bank can make a constructive contribution to the enterprise as a whole. Through interaction with senior management, the economists and the analysts on the ground, stress tests can be designed and the results communicated in order to help drive the business forward. The flexibility of this instrument is such that it may be turned to any number of different scenarios, enabling it to form an integral pillar of a ‘living’ risk management process.
Stress tests can be divided into sensitivity and scenario-based tests. While sensitivity tests examine the effect of changing a single variable on a portfolio, scenario tests change an array of variables. Scenario tests can be further subdivided into historical and hypothetical categories. Hypothetical tests may be more useful in providing a forward-looking approach to potential stressors. However, the design of such tests consumes considerable resources. Historical tests suffer in that the results they produce may not necessarily be valid for the present market situation, but do have a wealth of information as regards the correlational relationship cross-asset class in a risk event. The market risk management team of a large institution will typically have sufficient means to adopt a multi-method approach. There is no specifically ‘correct’ method, stress testing is not simply an exact way of measuring a stress event, but a way in which senior management, and the firm as a whole, can be informed of particular potential risk issues. By combining stress testing with the other tools in the risk manager’s armoury, a clearer view of the organisation’s overall risk profile emerges and business strategy can be shaped accordingly.
The ideal method of constructing hypothetical scenarios will involve consultation between several areas within the organisation. Scenarios should be devised after considerable consultation with the firm’s senior economists, in order to accurately model macro economic factors. There also needs to be some input from staff on the ground, who will have a feel for the market. However, this input needs to be filtered effectively in order to avoid an overload of information. Traders are generally so focused on their specific area that they cannot appreciate a holistic view of the organisation. Their contribution should, therefore, be refined via the medium of more experienced individuals with responsibility over a broader area, for instance, a specific asset class. Further to this, the senior management who receive the stress test data should keep the business’ strategic goals in mind and feedback areas of interest to the risk management team. This iterative process should be continual, embracing the full spectrum of possibilities that the stress test provides in an evolving fashion.
Source: LepusUnfortunately, even the most artful scenario will not add value unless its results can be understood. A problem for all risk managers is that the more variables they build into any model, and the more accurately they intend to construct a scenario, the less easy the results are to interpret for key decision makers. Risk management cannot simply act in a vacuum, they must ensure that they contribute to the business as a whole. A way to ensure this is to present scalable scenarios, starting simply and then building in an increasingly complex array of factors. The risk manager then can present intricate, multi variable examples, but then reference back to more simple ones for ease of comparison and comprehension. This forms part of a scalable risk management system, where increasing strata of complexity can be applied to the testing process. This, in turn, should help to build the senior management’s understanding of stress testing results allowing easier interpretation of more complex tests.
One of the vital skills of the risk manager is the communication of their work, and clearly explaining to management that the level of abstraction involved is a necessity. The risk manager needs to succinctly illustrate to the business function the reasons for the complexity or otherwise of a specific test, contextualising the information they are providing. Once again the importance of discussion, communication and development of process is clearly evident.
Stress testing does not remove all problems associated with risk. It cannot show the probability of the event that it examines. Any attempt to ascribe a meaningful probability to a specific event endangers clouding the test in an even more complex set of numbers. This would further hinder the appreciation of test results by senior management. Ironically, these leading figures are among those arguing loudest for probability attribution.
There is a compromise solution to the problem of probability. A useful thought-tool for structuring discussion within risk management was to imagine a graph for events with axis of the degrees of plausibility and severity. They can use this to judge which scenarios to report to the board. Practitioners know that risk management is not an exact science, if it were there would be no need for hotshot traders. Judgments on certain issues will remain qualitative, a product of the risk manager’s experience and discussion throughout the firm.
The problem of interpretation also restricts the expansion of the use of stress testing. Many risk managers are comfortable utilising stress testing and would be happy to roll it out across other fields. However, there remains the worry that until a proper interpretative framework is set up such a move might, if anything, have a negative consequence. However, once the scenario development and presentation processes outlined above have been implemented, there is no reason for use of stress tests to be held back. It is malleable to the various requirements of different businesses and provides a definite opportunity for the risk team to add value to the organisation through active management.
The importance of stress testing has no doubt increased. For this to continue, there is a necessity to both recognise and express the limitations of the tool, but also remember the tremendous flexibility that this facilitates. All areas of business can contribute to the stress test; it can inform debate through both the creation of scenarios and the dissemination of their results, focusing the firm on potential risk issues. Stress testing is an organic process and should continue to evolve.