Stress Testing Banks: DFAST/CCAR
Complying with the DFAST/CCAR requirements within an existing quantitative models and model risk management framework is one of the most daunting of the many recent challenges banks, Bank Holding Companies (BHC) and some Investment Holding Companies (IHC) currently face. The Dodd-Frank Act Stress Tests (DFAST) require all financial institutions with total assets above $10 billion to do stress tests on their portfolio and balance sheet. The Comprehensive Capital Analysis and Review (CCAR) is generally required to be completed once a bank’s total assets are above $50 billion. The objective of both exercises is to simulate a bank’s balance sheet performance and losses in a hypothetical severe economic downturn over the next nine quarters.1 Given this common objective, most risk managers consider and complete both exercises together.
With any DFAST/CCAR submission, affected financial institutions are not only required to complete the various mandatory balance sheet and capital templates, but also to submit the standard model risk documents for every model. Institutions cannot make any capital distributions (e.g., pay dividends) unless their regulator indicates in writing a non-objection to their submitted capital plan. Since all models are required to be validated before use, financial institutions are required to validate all their DFAST/CCAR models and include each model’s validation as a part of their submission package.
Below is a table summarizing an institution’s responsibility by total asset size and type of stress test:
Dodd-Frank Act Stress Tests (DFAST)
The Dodd-Frank Act Stress Tests (DFAST) are forward looking quantitative evaluations of the impact of stressed macroeconomic conditions on a bank’s capital. This involves stress testing a bank’s balance sheet over both pre-subscribed macroeconomic scenarios provided by the Office of the Comptroller of the Currency (OCC) and Federal Reserve and those required to be created internally. The Dodd-Frank Act, which was enacted in response to the 2008-09 global financial crisis, required the Fed and the OCC to conduct annual stress tests for institutions with more than $10 billion in total assets. Institutions above $50 billion in total assets are required to complete both the mid-cycle and supervisory stress tests. Previously, only institutions above $50 billion in total assets were required to conduct stress testing on internally developed macroeconomic scenarios. However, after the DFAST 2016 submission, all DFAST institutions will be required to conduct these internal stress tests.
DFAST submissions have historically been required to be completed by the end of Q1 or the beginning of Q2, with the DFAST 2016 submission deadline for CCAR banks set at April 5, 2017. The composition of a DFAST submission includes all capital and balance sheet components. A template is provided by the OCC for use in submitting an institution’s capital plan. Institutions are required to use a variety of macroeconomic scenarios, from both the regulator and the bank’s internal forecasts, to produce the DFAST capital figures.
Comprehensive Capital Analysis Review (CCAR)
The largest Bank Holding Companies (BHCs) are required to go one step further by completing both DFAST and the Comprehensive Capital Analysis and Review (CCAR) stress tests with the Federal Reserve. The CCAR stress test generally completes most of the DFAST requirements but adds additional quantitative and qualitative layers for the computation and submission of the BHC’s capital plan. Institutions must complete CCAR requirements if they meet the following conditions set forth by the Federal Reserve:
- Firms subject to the Fed’s Large Institutions Supervision Coordination Committee Framework (LISCC)
- Large and Complex Firms
- Total Assets in excess of $50 billion
CCAR has two different aspects which must be completed. The first is a quantitative capital plan which, like the DFAST capital plan, is a quantitative forecast of the bank’s capital on a variety of different macroeconomic scenarios. Also like DFAST, the capital plan has a presubscribed template which must be reviewed and approved by internal management and the board of directors before submission to the Fed. In both DFAST and CCAR, regulators can object to the bank’s capital plan, requiring the institution to revise and resubmit their capital plan.
The second component of CCAR reviews the qualitative aspects of the institution’s capital plan with the objective of ensuring that the submitted capital plan was created using strong risk management, good governance, and proper internal controls and policies. This qualitative review considers several key aspects of the institution, including the following:
- Stress testing method, including the generation of the internal BHC scenarios
- Internal controls to ensure an accurate and robust capital plan
- Model governance, including documentation and model validation
- Internal policies and procedures with clear and precise separation of responsibilities and approval processes
Firms are required to complete both the quantitative and qualitative CCAR requirements if they are either a LISCC firm or are large and complex. A Large Institution Supervision Coordination Committee BHC or those classified as large and complex by the Fed are those banks with any of the following characteristics:
- Have $250 billion or more in total consolidated assets
- Have an average total non-bank assets of $75 billion or more
- Are US global systemically important banks
Those firms which do not meet each of the three requirements listed above but are above $50 billion in total assets are determined to be large and non-complex firms. These firms are not subjected to CCAR’s qualitative review component but are still required to submit their quantitative capital plan for supervisory approval.
To populate the capital plan, institutions use a variety of stress scenarios to obtain stressed values for balance sheet line items. The sophistication of the chosen technique varies significantly, not only based on the size of the institution, but also on its stated objective. This approach can usually be split between models and non-models. A model in the DFAST/CCAR context has the same model definition used by supervisory guidance SR 11-7/OCC 2011-12. Models are typically used within the DFAST/CCAR framework to forecast out those assets and liabilities which the institution has determined there is high risk and a certain level of uncertainty. The use of a model or a tool to produce a DFAST/CCAR capital forecast is at the discretion of the institution with regulatory approval required.
The macroeconomic stress scenarios are divided into the following three categories based on severity and impact:
- Severely Adverse
The baseline stress scenario typically comprises the industry and regulatory consensus forecasts of the economy into the future and serves as the beginning benchmark of a bank’s performance against the stress scenarios. The adverse and severely adverse stress scenarios convey a what-if analysis of the economy to show slowing growth or a recession. The reasoning behind this approach traces back to the 2008-09 financial crisis where capital markets nearly collapsed, leaving institutions with little held capital vulnerable to potential liquidity and solvency issues. The adverse and severely adverse scenarios are run in this context to see how an institution’s balance sheet would perform through an event similar to or worse than the 2008-09 crisis.
In most cases, institutions leverage currently existing models within their inventory to produce the necessary DFAST/CCAR stress forecasts. A good example of this is the forecasting of an institution’s interest rate risk (IRR). Most institutions forecast IRR using a model to obtain projected values for both net interest income (NII) and/or the market value of portfolio equity (MVPE). Since this model already exists within an institution for this purpose, the existing model can be leveraged to apply the stressed macroeconomic scenarios and determine the behavior of IRR over the DFAST/CCAR forecast horizon.
A common misconception, due to its name, is that the severely adverse stress scenario should always have worse macroeconomic performance and a more significant impact on an institution’s capital than the adverse stress scenario. While the severely adverse scenario does generally give a more severe impact on a bank’s capital, there are cases where the adverse could have a much more severe impact. This is due to the nature of the scenario. For instance, the adverse scenario could focus on a 3% increase in the unemployment rate, while the severely adverse could decrease real GDP growth to -1% per annum. If the unemployment rate has a much greater impact on the institution’s balance sheet versus real GDP growth, then the adverse scenario will have a much larger impact than the severely adverse scenario.
The macroeconomic stress scenarios used in DFAST/CCAR are divided into two groups based on their source. The first group, which is required of all institutions required to complete DFAST/CCAR, contains the regulatory prescribed scenarios. These are macroeconomic scenarios supplied by the regulators themselves. These forecasts are quarterly and cover a wide range of macroeconomic variables, including real GDP growth, inflation and unemployment, among others. These scenarios are provided for both the domestic U.S. economy and overseas areas, including Europe and Developing Asia.
The second group contains the macroeconomic stress scenarios put together internally by the institution itself. Prior to DFAST 2016, only institutions with total assets exceeding $50 billion were required to produce their own scenarios. This distinction, however, was removed for DFAST 2016 and going forward. These internal scenarios are generally similar in both composition and effect to the regulatory scenarios but geared specifically for what the institution’s balance sheet is holding and where it operates. For instance, you would expect an institution whose footprint is limited to California to focus on the macroeconomic performance of California against the rest of the country in terms of forecasting potential macroeconomic recessions. Institutions usually generate their own macroeconomic forecasts using a combination of statistical modeling, industry best practice, and experience. These models and their construction will be the topic of a future post.
While the implementation of DFAST/CCAR into an institution can seem like a daunting and confusing task, understanding its objectives as well as how to leverage your existing model inventory can make this exercise rewarding in terms of risk management. Proper execution of the DFAST/CCAR process instills institutions with confidence that they have a sufficient capital reserve to ride out the next storm when it hits.
 Nine quarters is used since the last quarter of the year preceding the DFAST/CCAR submission is considered a part of the scenarios. For example, the 2015 DFAST/CCAR scenarios are provided from 2014 Q4 to 2016 Q4