At its core, credit stress testing guards against unexpected loss. Banks, by the very nature of their business, expect some loss, usually covered by the loan-loss reserve account.
Unexpected loss, by contrast, has no such hedge and requires dipping into shareholder capital, disturbing the underpinnings of the bank's financial foundation. This makes it crucial for bank management to identify unexpected loss as far out in the future as possible. Stress testing helps banks anticipate losses given various potential scenarios.
Successful credit stress testing isn't just about the technology; like any bank initiative, it needs an underlying, coherent strategy. A set of stress tests, in conjunction with regular risk forecasts, should give management insight into the range of potential risk exposures.
In order to recognize that stress testing serves more than one purpose, it helps to break it down into various types.
Scenario-writing is written narrative, generally qualitative in nature, and giving an integrated vision of the bank's future state. The result is a scenario map that consists of one or more agreed-upon events that may occur in the future, even if the probability is small.
Sensitivity analysis isolates the effect of a given event on a chosen variable and also addresses one or more predefined moves in a particular risk factor.
Statistical inference involves tests under extreme conditions (for example, five, 10, or 15 standard deviations), which go well beyond sensitivities. This ranges from the analysis of spikes in time series to estimating outliers, stress conditions, and unexpected events.
Worst-case-scenario analysis looks at situations of very low probability but very high impact; it is concerned only with extreme events toward the end of a long leg of extreme events. The goal is to identify weak links and institute capital damage control procedures before a catastrophe hits.
These types of analysis contribute directly to the strategic nature of stress testing. Embracing the "types" concept and building it into the fabric of the bank becomes more appealing because it can help get the organization focused on the future.
Stress testing requires accuracy but not precision. Since it is a predictive activity, it is nearly impossible to be precise. Data sources need to be well defined and documented so deviations from the norm can be explained to shareholders, regulators, and employees. It is of paramount importance for a bank to do quality stress tests to achieve results that satisfy regulators, the board of directors, and management.
A large set of good, clean data covering a significant period within a stable organization provides a solid foundation of stress testing data.
Other tips to ensure the success of a stress test initiative include:
Senior management must be engaged. Since stress testing is a strategic initiative at its core, senior management must get on board early. The problems in the credit market were caused, in part, by senior managers who were too slow to react. Management passivity, even if it results from a fear of misinterpreting data, cannot be allowed.
Use the stress test's effectiveness fully as a communication conduit. If a formal plan is not in place to communicate stress test results back to senior management, one should be adopted immediately.
The initial design process must be rigorous. You can never be too meticulous; you do not want to keep changing the design once you start.
When designing data collection, go for consistency. Be sure that the data points and key indicators, once established, remain constant throughout the initiative. Data collection and compilation can be expensive, and there is no way to completely avoid it. However, it is easy to accumulate unnecessary expenses if the design keeps changing.
When implementing the model, do so gradually. Taking on a stress testing model is complicated enough, especially considering the variety of markets affecting your bank's credit portfolio. To make life simpler, roll out the initiative gradually.
Test regularly. Market conditions can change more rapidly and dramatically than ever, making it prudent to reassess your stress test methodology periodically. Do so at least once a year, if not more often. Make sure this testing is coordinated with those who will be using it.
Document the stress test. Your stress test methodology and its results will affect all areas of the bank, and those affected need a clear idea of how to interpret the results. This makes it important to document the test.