Stress Test Your Portfolio — Not Yourself

Posted by Bill Hunt on Tue, Dec 10, 2013

stress testingA Dodd-Frank Act stress testing guide for senior bank executives whose institutions have $10B in assets or are growing quickly.

Think regulatory stress testing is only for the biggest banks? Well, maybe for right now, but this practice continues to be required of increasingly more banks, which means if you haven’t had to worry about stress testing yet, you might need to soon. Regulatory stress testing is designed to see if you have enough capital to survive a series of arbitrary shocks to the economy; it should not be confused with the types of analyses that support critical financial measures such as risk-adjusted returns or those that determine economic capital requirements.

Whether your bank is regulated by the Fed, FDIC, OCC, FHFA, or any other agency, and whether it will be subject to CCAR, DFAST, or another yet-to-be-defined testing protocol, it behooves you to be ready to submit official results before you're required to do so. Finding out whether your bank has a potential capital problem early on allows you to address it before you have to submit your report to your regulator. Why? Just ask the senior management at 25% of the largest stress-tested banks that were forced to address capital plan deficiencies after they submitted their stress testing results to their regulators, incurring the additional, substantial costs of remediation and having to explain to their stockholders why they weren’t prepared. On the other hand, your stress testing may reveal that you actually have sufficient capital, reducing your own stress level considerably.

Banks typically take one of two approaches for regulatory stress testing. One is a top-down analysis, which projects future losses based on historical loss experience in the aggregate and then incorporates shocks to produce stress losses. The other is a bottom-up approach, which applies the macroeconomic stress factors against each loan you hold to determine each loan’s potential losses and then aggregates up to arrive at a risk assessment for the entire portfolio.

For several very good reasons, it’s to your advantage to use the bottom-up approach. First, if your bank’s lending experience spans the Great Recession, which began in 2008, a top-down approach will include that period (unless somehow you are able to convince your regulator to let you exclude it — good luck with that).  This means the sins of the pre-recession past — old, looser underwriting practices, weak assets that you’ve already disposed of, and the like — are nevertheless going to be visited upon today’s portfolio.  Bottom line: a top-down stress test can actually significantly overstate the weaknesses in your current holdings because it can overweight credit crisis conditions while underweighting improvements in underwriting and market conditions.

By contrast, a bottom-up approach is based on what you currently hold in your portfolio: that is, better loans reflecting tighter underwriting standards, more conservative lending practices, and more aggressive tracking and monitoring. It starts with what you actually have today and projects those loans into the future. It does not commingle the weaker performance of the past with the more robust performance of today. 

But there’s another huge benefit to bottom-up approaches: the ability to forensically determine and rank the specific loans that are driving stress case losses and address them with more targeted capital remediation plans. This is simply not possible when you use a top-down approach.

Finally, a bottom-up approach sets the stage for far more productive conversations with regulators. Demonstrating the impact of macroeconomic scenarios directly on loan-level default and loss through a mathematically rigorous approach is very powerful, relatively straightforward, and defensible. By contrast, a top-down analysis done at an aggregate level requires many assumptions, including how macroeconomic factors will impact risk in the same way across portfolios that might be very diverse with respect to debt structures, collateral, and borrower quality, geography, and prior loan performance. Even in the best cases, such assumptions tend to be overly simplistic and take significant effort and time to defend.

With all these benefits, why isn’t every bank using bottom-up approaches to stress test its portfolios? One reason might be that the prospect of gathering, organizing, and applying sophisticated models to hundreds of thousands of loans may be too daunting to contemplate.

The good news is that banks don’t have to do this on their own. Market solutions are available today that can perform these sophisticated analyses, and they don’t require massive investments in IT, integration, and infrastructure to make it happen. 

Another issue — the upfront expense required for a comprehensive bottom-up solution versus the simplistic top-down analysis — might also be a hurdle for some. That is, until they consider the trade-off of solution costs versus potentially lower capital requirements, which overwhelmingly favors the bottom-up model.

In sum, there are many reasons to consider implementing a bottom-up approach. But there’s also one caveat: if you decide to investigate bottom-up stress testing solutions further, know that not all are created equal. Here are some critical guidelines and criteria for assessing them:

  • Know that the data elements required for the stress testing are no different than those captured in the original underwriting
  • Credit models should be well documented and independently validated
  • Regulatory stress scenario assumptions should be built in and updated continuously
  • Stress testing reporting analytics necessary to complete the process should be provided
  • Results should be available immediately with minimum of effort and be clearly presented; details should be available at both the portfolio and asset/loan level
  • There should be complete transparency as to how stress assumptions are converted into expected losses
  • All inputs, outputs, assumptions, and system metadata should be archived and accessible to provide a thorough audit trail

Want to learn more about stress testing analysis? Download our white paper, “Regulatory Stress Testing for Mortgage Loan Portfolios.”


Download Now


Topics: Big Data, Mobiuss