Execution, Scope, and Coverage
The first question to consider in implementing a system-focused stress test is who crunches the numbers: the supervisory agency or central bank, or the institutions themselves? Ideally, individual institutions should be as heavily involved in the process as possible—regardless of whether a top-down or bottom-up approach is used—because individual institutions will typically have the best access to data and knowledge of their own portfolios. For institutions with sophisticated risk management systems or significant international operations, most will have systems and stress-testing procedures in place as part of their internal risk monitoring processes.5 For countries with financial institutions that have more rudimentary systems and have less expertise in modeling their portfolios, involvement in the process may be beneficial by expanding their knowledge. In those circumstances, it may be necessary for the central bank or supervisory agency to provide guidance or even to undertake parts of the empirical analysis, but this process should still involve individual institutions as much as possible. Having institutions cooperate in a stress-testing exercise may require some moral suasion or other incentives, including the ability to benchmark their own results against their peer groups or the ability to learn from other participants. At the same time, the supervisory agency or central bank needs to minimize conflicts of interest arising from the institutions’ participation in the exercise. In particular, it needs to minimize incentives of institutions to project an overly optimistic picture, which could compromise the quality of the test. The supervisory or central bank staff may need to confirm the validity of the tests, including confirmation by carrying out independent tests as needed.
Implementing a stress test also requires addressing this question: Which institutions should be included in the exercise? The coverage of the stress-testing exercise should be broad enough to represent a meaningful critical mass of the financial system, while keeping the number of institutions covered at a feasible level (e. g., fewer than 20). The total market share of the institutions involved (in terms of assets, deposits, or some other criteria such as importance in the payment system) can be used to determine a cutoff point, because the exercise may become unwieldy if too many institutions are involved. Depending on their interlinkages, both banks and nonbank financial institutions should be included in the analysis, although this involvement may present some difficulties if they are supervised by different entities or have different balance-sheet reporting dates or practices.6 In countries with a large number of small institutions, consideration could be given to either aggregating smaller institutions into a single balance sheet or taking a
representative sample of institutions, or even ignoring them if they are not systemically important.
Another important factor to consider in conducting a stress test is the data constraints. The availability and quality of data impose major constraints on the nature of stress tests that can be performed. Data limitations arise from the lack of basic data availability (especially in countries where information on balance-sheet exposures may not be available), difficulty in isolating specific exposures (especially in the case of large complex financial institutions, or financial institutions that are active in the derivative markets), lack of risk data (such as duration or default measures in countries where risk management systems are less sophisticated), and confidentiality issues (limitations on what supervisors are legally able to share with other parties).
If one is to overcome the data difficulties, it may be possible to work with the larger and more sophisticated institutions to get better data or to calibrate some parts of the exercise. For example, if the exposure of interest is the aggregate exposure to a specific borrower or sector, individual institutions may be able to produce information on that exposure from their internal risk monitoring systems, even if they do not report data to the authorities in that particular format. When confidentiality issues do arise, it may be possible for the institution with access to the data to conduct the stress testing while using agreed assumptions and methodologies and to share the results with the authorities in a form that is sufficiently informative of the risk exposures but that would not breach confidentiality laws or protocols.
The choice and implementation of stress-testing techniques in practice reflects the data quality and technical capacity available. In addition to the design of the stress-testing scenarios and the choice of top-down versus bottom-up approaches, an important part of the stress-testing process is the selection of technical tools to implement the stress-testing calculations. For each of the risk factors, there are several techniques or approaches to implementing the calculations. The techniques generally differ in the required volume of data and in their computational complexity. The choice, therefore, largely depends on the data availability (e. g., if no data are available on time to repricing of assets and liabilities, the interest rate risk can be assessed only by using very rough methods) and on the technical capacity available (e. g., software, staffing constraints, and time constraints).