Top-Down Stress Testing: The Poisonous Placebo

By Adam Mustafa, Invictus Group CEO


Community banks that take a “top-down” approach toward stress testing are taking the equivalent of a placebo when it comes to assessing how ready they are if the pandemic triggers a severe and protracted economic downturn.  Placebos are cheap and easy to make, harmless and yet can have a real psycho-physiological benefit. But unlike a placebo, a top-down stress test can do real harm. It gives a false sense of security and complacency. Even a cursory review of the process highlights the futility of depending on a top-down stress test, especially during the pandemic.

What is a Top-Down Stress Test?

A top-down approach toward stress testing generally involves applying an estimated loss rate to a type of loan and simply multiplying the loss rate against the loan balance outstanding for that loan type. Estimated loss rates are usually derived from historical periods of distress such as the Great Recession. The top-down approach has become popular because numbers can be crunched quickly and easily.

What’s the Problem?

A top-down approach is highly superficial. It ignores nearly all the most important variables that would separate and distinguish those banks that are far better positioned to absorb the impact of a severe and unique economic downturn triggered by the pandemic.

Most importantly, a top-down approach ignores the underwriting characteristics of the loans. News flash: Underwriting and credit quality matter. The likelihood and magnitude of stress occurring in the loan portfolio will primarily be a function of the underwriting characteristics of the loans.

Take two banks, each with the same concentration of hotel loans. A top-down stress test will assume that these two portfolios are homogeneous and will experience the same amount of losses. However, let’s assume the first bank’s hotel portfolio has an average age of 2.5 years and an average loan-to-value (LTV) of 70 percent. Now let’s assume the second bank has a more seasoned portfolio in which the average age of the portfolio is closer to five years and the average LTV is 60 percent. These differences are not minor. They will have a significant impact on the results of the stress test if it is done properly.

The point about LTV is probably obvious, but it’s not just how the loans were underwritten that matters, but when they were underwritten. Loans originated earlier in the cycle will not only have more seasoning, but they will also have more capacity to absorb losses because the underlying properties were not valued at the top of the market.

It’s also not just one or two underwriting characteristics that matter. Those loans that are most vulnerable will have a combination of multiple underwriting characteristics that have a cumulative effect. Frankly, by ignoring this information a top-down approach is an insult to the management of the bank.

Times Have Changed

There are several other fatal flaws with a top-down approach. For example, it assumes that the next downturn looks like the last one. We have already seen significant differences between the COVID-19 recession, which is still in the early innings, versus the 2008 Financial Crisis.  This is why the Federal Reserve did not rely on the CCAR stress test results as a proxy for a COVID-19 inspired downturn.

Also, the common approaches for estimating loss rates for different loan categories, often derived from the Great Recession, turns every bank into the lowest common denominator bank. Even if a bank uses its own loss history from the Great Recession as opposed to those from other banks, this is still inappropriate, if not unfair. Most banks have dramatically altered their underwriting philosophies since the Great Recession.

A comparison of a given bank’s loan portfolio at the end of 2019 with the same bank’s portfolio at the end of 2008 would be unrecognizable. In other words, a top-down approach fails to capture the concept of behavioral economics, which is especially important and very real.

The loans that comprise today’s balance sheet were originated under vastly different economic and interest rate conditions than their 2008 brethren. The conditions that exist on the day that a given loan is originated can dictate its ultimate risk profile more than anything else. This is true even if underwriting is neutral (which it has not been, as previously discussed in this article). There is little in common between loans on the balance sheet in March 2020 versus September 2008.

The Irony of It All and How to Wrap Your Brain Around It

You may be thinking that the concerns surrounding a top-down approach toward stress testing runs the risk of failing to detect stress when it exists (“the false negative” diagnosis). However, the opposite is also true. A top-down approach is just as likely to generate a false-positive result, leading to potential overstressing of a bank.

You may also be thinking that it’s good to err on the conservative side. And while there is some value in demonstrating how an overly conservative approach shows how a bank would remain sufficiently capitalized, its benefit is grossly over-rated.

A top-down approach transforms a critical strategic analytical process into a mindless pass/fail exercise. While precision is impossible in a world of uncertainty, direction matters, and degrees of direction matter more. We don’t need precise coordinates, but we need better than just “go left.”

Investors and regulators want more and deserve more. Stress test results are playing an increasing role in loan loss reserve accounting, under both the incurred loss methodology and CECL regimes. They also play an important role in assessing a bank’s ability to design and execute its strategic and capital plans – especially dividends. Degrees of direction matter here. A bank can still remain well-capitalized, but find itself in regulatory purgatory because of a spike in the classified asset ratio.

Stakeholders will be handicapping the bank’s future earnings based on stress test results. They cannot do that effectively with a top-down approach for all the reasons previously mentioned in this article.

Management should demand a bottom-up approach to stress testing. You need to be able to triage your loan portfolio based on the dollars you have at risk. This enables proactive risk management and workout strategies with borrowers to limit losses. This is not applicable with a top-down approach.

A bottom-up approach executed properly will still maintain all of the benefits of a top-down approach, such as the ability to incorporate the impact into a complete balance sheet, P&L, and capital adequacy schedule, which are pre-requisites to using stress testing for strategic and capital planning.

Wrap Up

A cottage industry of sorts has emerged, ranging from asset-liability modelers, software-as-a-service providers, “jack of all trades, master of none” local consulting firms, and even investment banks. All are throwing their hats in the ring and providing banks with top-down stress testing “solutions.” Some banks are using these approaches to get a ‘second opinion’ to compare with their bottom-up stress tests. While this makes sense at the surface, keep in mind the limitations of a top-down stress test — especially when the results of your top-down stress test appear more conservative than your bottom-up results, or when they approximate one another. At best, placebos make for a good comparison group.  But they should not be mistaken as the solution itself.