This series uncovers the evolution of stress testing and the importance of institutions taking a proactive stress testing approach. This whitepaper details the many methods in which an institution can stress its portfolio, mitigate risk, and best practices to stay ahead of economic uncertainty.
Credit risk professionals hear the term “Stress Test” and it conjures up a detailed analysis of the bank using a deteriorating set of scenarios and an equally detailed submission to the Federal Reserve. Stress testing is more than just Comprehensive Capital Analysis and Review (“CCAR”) or Dodd-Frank Act Stress Test (“DFAST”). A stress test is, loosely defined, an analysis of portfolio performance using a set of negative economic scenarios. To understand what stress testing is and how it can effectively manage portfolio risk, it is important to understand the history of financial stress testing and the types of available stress testing.
Financial stress testing has included many types through the years, from borrower debt service coverage (“DSC”) testing to various reserve and loss calculations. Tests in the past tended to have limited inputs with equally limited calculations due the absence of any sophisticated technology beyond an adding machine, electrical or mechanical. Stress testing as a practice became more frequent and slightly more sophisticated in the 1980s and 1990s with the rise of mainframe systems, desktop computers and spreadsheet software. Firms were now able to develop testing programs with multiple scenarios, still largely using aggregate data. At that point, any stress testing was developed largely for internal risk management. U.S. bank examiners still considered stress testing practices a prudent risk management exercise but refrained from any form of mandate or even best practices.
Regulatory Stress Testing
In 1979, Federal examiners created the Uniform Financial Institutions Rating System (“UFIRS”) to measure performance and risk management practices at banks. The system ultimately became known as CAMEL and stood for capital, asset quality, management, earnings and liquidity. The term “sensitivity” was added in 1995 to make the term CAMELS and essentially became the first federally mandated stress test. This section within the CAMELS system was specifically added in response to a farming crisis and is a forward-looking view of commodity prices, interest rates and broader market risk on the balance sheet. Ultimately, this type of stress testing failed to predict the financial crisis of 2007. Following a worldwide recession in the early 1990s the first Basel Accord (there are now four) was created in 1996 by the central banks of ten countries (“Group of Ten” plus Spain and Luxembourg) for the purposes of standardizing bank regulation, prudent risk management principals and capital management.
The Basel accords introduced standards for credit risk, liquidity, and capital management. While technically not a stress test, the accords generally require the use of internal models to measure the impact of credit risk on capital. The financial crisis of 2007 had a tremendous negative impact on reserves and bank capital. In response, federal examiners developed the Supervisory Capital Assessment Program (“SCAP” / “Too Big to Fail”) which tested the effect of two macroeconomic scenarios on potential losses and the impact those losses would have on pre-provision net income, allowance and Tier 1 common capital. Nineteen banks and investment banks whose assets exceeded $100 billion were required to complete the test to determine if additional capital was needed. Ten of the banks required additional capital, solely based on the effect of the most adverse scenario even though they met minimum capital requirements before the SCAP test.
In July 2010, following the end of the recession, Congress passed the Dodd-Frank Act which introduced a host of banking regulations and consumer protections. The Act also introduced the concept of a forward-looking view of a bank’s balance sheet by creating what ultimately became the CCAR and DFAST tests for larger national and regional banks and their holding companies.
These stress tests replaced SCAP and generated impact to the capital structure and balance sheet by requiring firms to measure the impact by applying several predetermined economic scenarios. The tests require banks to consider a forward view of broader economic variables rather than the near-term review of delinquencies, bankruptcies, and non accruals that most institutions reviewed before the recession.
Scenario analysis is another form of stress testing that is quite distinct from regulatory stress testing. Where CCAR and DFAST are examiner initiatives that focus on modeling overall stability of large banks based on multiple economic scenarios, scenario analysis (frequently called stress testing as well)! is a risk management tool that develops certain outcomes, like expected loss or risk rating migration, under declining scenarios.
Scenario analysis is used by banks to strategically analyze all or part of the loan portfolio and measure the results on capital and key ratios, and to support strategic initiatives. Banks are not required to develop a scenario analysis program, but it is considered a best practice by examiners because it shows adherence to broader enterprise risk management principles. There are several types of scenario analysis methods with each having their own specific strength, and dependant on the portfolio being analyzed.Each method and its results, under mild, adverse, and severe scenarios, tells a story,especially when considered with a portfolio of other scenario analyses. It is important to not rely on one type of analysis over another. Banks should consider multiple methods as their results, that when taken together, should triangulate an expected outcome.
“A risk management tool that develops certain outcomes, like expected loss risk rating migration, under declining scenarios”
The most common test that banks run is a borrower or relationship test that models changes to net operating income, collateral value, or cap rate to determine impact on DSC and loan-to-value (“LTV”), both key risk rating drivers. Typically, these tests are performed for new money requests and upon loan renewal but are increasingly being incorporated into the loan review process. As a prudent risk management measure, strategic borrowers or large relationships should have a detailed test as a part of their annual loan review, given the potential for risk to the bank should that client default.
Bottom Up Testing
Using the logic of the borrower test, a bottom-up test takes the loan information for a specific segment of the portfolio and aggregates the information to determine the impact of expected losses or risk rating migration on the bank’s balance sheet and capital. This is typically a data-intensive exercise but develops meaningful results because it’s based on actual client data and reasonable assumptions. The two key parts to a meaningful bottom-up test are the segmentation and the accuracy of loan level information. Differing views of a segment could produce different results, sometimes dramatically. For instance, commercial real estate can be defined as a bank product, a call code, or a NAICS code. This context is important when considering the results. Bottom up tests focus on the aggregate impact of mild, adverse, and severe stress scenarios on borrower-level DSC and LTV. Stressing DSC and LTV will identify loans likely to default and charge off. Given the importance of DSC and LTV for a risk rating, a bottom-up test will also identify potential risk rating migration. Banks struggle with selecting proper stress scenarios so it is important to develop reasonable scenarios given the segment being tested. First, consider the metrics that are important to the segment and have a strong degree of correlation to default or loss. It might be the Commercial Real Estate (“CRE”) Index or RevPAR Index for commercial real estate or hotels, or maybe the Home Price Index (“HPI”) for residential real estate. Estimates for these indices should drive the range of scenarios.
Scenarios should also have some degree of logic and reasonableness. Alternatively, consider using key components from the Federal Reserve stress testing scenarios discussed above. The Fed scenarios provide an added layer of defensibility.
This method focuses on the observed risk rating migrations of a pool of loans over a defined period, applying that migration pattern to a current pool of loans and interpolating expected loss based on the loans that migrate to a substandard risk rating. As above, the defined periods would represent Mild, Adverse, and Severe scenarios and should be based on historical mild, adverse or severe migration patterns for that segment. This method is easily built and frequently used because it can be developed using simple loan-level data. All that is needed is risk rating data by count and by balance. From there, develop a table that compares two month-end files by risk rating and loan count. The table will show the number of loans (and percentage) whose rating did not change for that period and loans that are paid off or had a risk rating upgrade or downgrade. Next, identify historic monthly, quarterly or annual periods that best identify with mild, adverse and severe scenarios. Use the migration scenario percentages and apply them to current balance being tested to forecast risk rating migration over the next one year. For loans that migrated to a substandard risk rating, apply a specific loss given default (“LGD”) percentage against the balance to arrive at an expected loss.
Top-down testing typically uses segmented balances and applies a range of historic losses to identify expected loss. These losses may be based on observed historical losses by the bank or losses observed by peers or the industry in mild, adverse, and severe scenarios. Like a basic incurred loss allowance model, the model can be easily built using aggregate balances and loss percentages. Industry best practices have stated that this model should have a two-year loss horizon with first year losses subtracted from the balances and the loss rate applied to year two balances. This model is used quite frequently because it can provide a top-of-the-house view of a range of expected losses, including a break-the-bank view if necessary.
Stress testing is an important tool because it enables the bank to identify losses so they can be mitigated and accounted for. It is also important to understand that users tend to focus on pool-level or aggregate level results without always looking deeper. Stress tests return results based on their inputs, and nothing else. Running a test on CRE will only return results for CRE. Is it safe to assume results for CRE can be applied to an office or retail portfolio? What about within retail?
Is type or geography important? How about my largest borrowers? Understanding that sometimes you have more risk as a percentage within a smaller segment than in total is critical when managing a portfolio. A portfolio needs to be subdivided and tested at a granular level using multiple tests to make sure risk has been properly identified.
As an example, Simply Shoes is a fictitious borrower from the Retail sector.They currently have a $775,000 loan with a 2.90% interest rate. This is a Pass-rated loan with a 1.30 global DSC andan LTV of 65% – a safe loan, generally.It is important to understand that good loans can
Turn bad and stress testing can not only tell you what your loss might be,but it may also tell you who those loans are. Understanding the sum requires knowing the parts.
Next let us stress Simply Shoes using both reverse and bottom-up reverse stress tests at different levels of the portfolio.
First, here are the assumptions being used for Simply Shoes and the rest of the CRE pool, including the Retail segment:
A good reverse stress test to consider is based on the Federal Reserve forecast, given its acceptance by the market and the examiners. Using the baseline approach, including its unemployment assumptions, and a straight-line reversion, we can use a discounted cash flow model to determine that this loan has a decent chance of defaulting, likely due to its global DSC:
Retail and the Bank segments tell a different story. Using the Reverse stress test, both Retail and the Bank portfolio performed similarly but still much better than Simply Shoes did individually.This is likely due to stronger average DSC for the portfolio than Simply Shoes.
Stressing the portfolio using different tests can generate different results but knowing how to interpreting the results help give context and clarity when telling the story. Know that because each test will give results based on its inputs but since those inputs are applied aggregately, frequently you will see much different results the deeper you get into the portfolio.
Stress Test Focus
A strong stress testing program is an important part of a broader enterprise risk management plan because it quantifies credit risk and hopefully validates the risk appetite identified by the Board of Directors of the bank. Developing a broader enterprise risk management stress testing plan that incorporates interest rate risk and market risk testing should be a priority for all banks. Examiners have looked favorably on banks with an active stress testing program because it shows a proactive view of managing risk.
Other than the regulatory stress testing, there is little guidance on stress testing. The examiners really appreciate it when it is done but will not go on record as to document any types or best practices, leaving it to banks to develop a program that they feel best identifies risk. From community banks up to multinationals there are some best practices to consider:
- Understand the audience and know who will be reviewing the results. Is this for the Board or management? Examiners? If so, you will need to pre-screen your models and assumptions with key colleagues and incorporate the models into a broader model risk management program
- Present results logically using the range of expected losses to easily identify changes to the allowance, earnings, capital, and key ratios.
- When using bottom-up stress testing, get curious. As a risk manager, what part of the portfolio keeps you up at night? Identify where the real risk is by peeling back the layers of the portfolio to find loans or segments whose risk might be overlooked. Sometimes the sum of the parts might have more risk than the portfolio.
- Make scenario analysis a regular part of portfolio management. Consider testing more frequently than management requires.
- Whether managing credit risk or other types of enterprise risk, running stress tests is not the final step in managing risk, it is the first step. Developing a plan to mitigate the risk, however small or large, should be the final step.
- Using multiple stress test methods on the same portfolio will logically show similar results if built correctly, offering a degree of model validation and context.
- Present results in terms of Tier 1 or Tier 2 Capital as well as a percentage of pre-tax pre-provision income. You may also want to add a one-year look-back of current charge-offs for context.
- Use the results to improve your allowance process and risk rating guidelines.
There is so much to stress testing, but it is important to note that there is not one type or logic that can be applied across the industry. A stress test program should be unique to each bank and its portfolio.
About the Author
Rob Ashbaugh, Managing Consultant
Rob is a former banker and portfolio risk manager with more than 20 years of commercial banking and capital markets experience at both the large bank and community bank levels. He has managed the ALLL and portfolio stress testing process and been an invited speaker at many national and regional conferences. Rob is a past holder of the Series 7, 52, and 63 licenses and received his Bachelor’s degree in both economics and international business from Temple University. When Rob is not working, he enjoys skiing, fishing, and spending time on the Lake.