Note: This fall, banks with more than $10 billion in assets will be preparing stress test results for submission to regulators in early 2014, as required by the Dodd-Frank Act. This will be the first time stress testing is required of banks in the $10 to $50 billion range. While regulators are asking banks with less than $10 billion in assets about their stress testing plan, there is no current mandate subjecting them to the tests.
Bank directors working to implement a defensible stress testing framework are often challenged to strike a balance between rigor and flexibility. With the rules coming out of the Dodd-Frank Act subject to change, this is no easy task. Once the stress testing process is underway, bank teams can be quickly lulled into a false sense of security and fail to recognize seemingly innocent warning signs that stress testing is going to be a problem. We have identified five signs that suggest a bank is at risk to bungle its stress testing framework. If directors of large banks (with over $10 billion in assets) have not yet come across these topics in their stress testing process, it may be time to start asking questions. For smaller banks (under $10 billion in assets), there is no immediate cause for alarm, but starting to pose these questions will put the bank on firm footing when the time comes to begin running stress tests.
Red Flags to Watch For:
1. Stress testing is managed by analysts.
Be careful if stress testing is managed by analysts and the topic is not discussed in detail at board meetings from the beginning of the process. If management is not talking about the framework, cost of compliance, or the time involved, it is possible that the program has not been taken seriously or has not begun at all. For large banks, this would be troubling.
Some of the broader questions facing management include:
- Should we build internal models or outsource?
- Do we have enough data to construct and calibrate the model?
- How do we validate the models?
- Do we have the right people?
It should be noted that compliance with the stress testing mandates does not come cheaply. One large bank recently noted that it was spending $3 million per quarter on stress testing. While that kind of spending is not necessary for banks with assets below $20 billion, stress testing will still be a large line item. Making the correct decisions along the way can help to limit the amount on the check you need to write.
2. Stress testing is being conducted using an asset-liability management (ALM) model
(or by building a spreadsheet or application of historical industry levels).
Your models should utilize macro-economic variables, such as those supplied by regulators, to drive the results. If your bank is planning to use an existing system, an accounting database, and/or an ALM system, chances are that system has not been modernized to link macro-economic variables to probabilities of default across loan types, loan growth by business line, or net charge offs. The burden of proof that the bank’s assumptions are defensible, given the data, will be high.
Small and mid-sized banks do not have the staff in place to determine probabilities of default and loss given default across a spectrum of interest rate scenarios, unemployment levels, and GDP projections. As a result, retrofitting an existing model can’t be done easily. The biggest risk in going down this path is spending a lot of money up front only to learn that the regulators want something entirely different. Banks should look internally to determine what they are capable of, as well as at third-party vendors to attain the right model.
3. The model is built to handle only three defined scenarios.
The bank regulators have mandated that three separate scenarios be run by the banks: a baseline case, an adverse scenario, and a severely adverse scenario. The baseline case assumes an economy that is growing moderately. The adverse scenario is a currently a “stagflation” scenario—low growth, higher interest rates. The severely adverse scenario is a deep recession projection.
Banks that have implemented a model that meets these requirements may feel their work is done, but there is no obligation on the part of the regulators to maintain the same three scenarios. When the next set of regulatory macro-economic variables are released in mid-November, the Fed could opt for something entirely new, like a deflationary scenario. Accordingly, the models should be built to handle a wide array of inputs. Otherwise the bank will be re-building the model year after year and will be slow to see any benefits of cost tapering. Banks that get stress testing right may spend a lot up front but see costs taper over time. In order to see those benefits down the road, it is critical that a bank build the program on a solid foundation.
4. Regulators are brought into the process when results are submitted.
Banks want to ensure that they build a well-constructed model to accommodate future changes to regulatory requirements. Given that the regulations are still evolving, banks may find it difficult to follow a “measure twice and cut once” practice. The solution is to stay close to the regulators throughout the process. If your management team is not actively asking the regulators questions, whether it is about data sources, how they expect stress tests to look, how they feel about a top-down model versus a bottom-up model, or how to validate a model, the team is taking unnecessary risk. A bank does not want to submit results only to learn later that it had been going down the wrong path.
5. Stress testing is conducted as a separate mission from overall bank operations.
As many banks go through the stress testing process for the first time, they may set up their stress testing team as a separate unit from other divisions of the bank. Currently, that is probably a smart idea. The most important item right now for large banks is meeting the mandate. Along these lines, minimizing distractions and completing the task at hand is the focus for 2013 and 2014.
However, as stress testing evolves from a 100-yard dash to a more routine process, banks will be at a disadvantage if stress testing does not become part of a broader internal dialogue. Capital planning and forecasting should include stress testing results as a part of their ongoing processes, and management should challenge internal staff to consider alternative scenarios for risk management purposes. Those responsible for collecting the bank’s internal loan data should continually ask what data could be gathered to help generate more accurate projections. Over time, stress testing should develop to become an important consideration for all departments of a bank.
Most bank directors are worried about meeting the existing requirements under Dodd-Frank at the moment, but directors thinking about the next steps will avoid unnecessary pitfalls and costs. Given the amount of uncertainty that remains on this topic, board members should not hesitate to get more involved and ask questions, both of their teams and of the regulators. Hedging some of these common issues we have identified can cut costs both in the short- and long-term, minimize frustration when results are due, and help to develop an easily replicable process for the future.