After the 2008 Great Recession and prior to the coronavirus pandemic, regulators turned to stress testing to establish bank capital adequacy levels. The Federal Reserve established a program called the Comprehensive Capital Analysis and Review (CCAR) for the largest banks, laying out economic scenarios and parameters for each stress test. Some community banks duplicated aspects of the CCAR approach, while others used loan-level stress testing focused only on the CRE portfolio in an attempt to prove to regulators that their banks would withstand a potential downturn. In all cases, the objective of the stress tests was to subject banks’ balance sheets to hypothetical adverse conditions and to evaluate the impact on bank capital under these scenarios.
The pandemic has changed everything. The historical data and credit fundamentals behind stress testing are no longer valid. Stress testing has moved from an exercise in evaluating the impact of future hypothetical economic changes to evaluating existing economic changes, the ongoing evolution of these changes and their delayed impact on the bank’s capital adequacy.
These changes are dramatic in both form and substance. The pre-pandemic stress scenarios relied on historical credit and collateral information to evaluate and quantify the degradation of the loan portfolio and its corresponding impact on the bank’s capital. There was enough historical data and traditional credit experience to create reasonably realistic scenarios. Statistical correlation techniques had reasonable validity under these scenarios.
In the pandemic and post-pandemic environments, statistical correlation is an absurdity. Stress testing companies and consultants staffed by "computer jocks" and statisticians with little or no direct credit experience are in no position to assist banks in their stress testing efforts. There are no correlating data points that would allow such an exercise.
I stress credit experience since any reliable stress test must rely on the bank’s credit process and its intertwined loan classification system. In a simplistic explanation, adverse economic circumstances cause loans to slide down this loan classification system, creating increasing demands on bank capital as credit degrades.
However, many community banks instead turned to oversimplified models entirely focused on shocking net operating income levels or black-box regression models that were highly dependent on metrics such as net operating income.
In the pre-pandemic world, there was sufficient data on collateral values and credit to allow fairly reasonable assumptions on the rate and nature of this degradation.
The post-pandemic credit environment is radically different. It is critical to understand the differences before even attempting a stress test. Each industry will react differently to the pandemic in regard to its rate, magnitude and timing of degradation. A reasonable knowledge of the credit/operating characteristics of companies within high risk industries such as hotels and restaurants is fundamental to creating a reasonable post-pandemic stress test. I can best explain this by using a simple example of a restaurant undergoing stress in a pre-and post-pandemic environment.
Example: Pre-pandemic restaurant under stress
A strong restaurant would have strong earnings and a strong debt-service coverage ratio (DSCR). A weak restaurant would have lower earnings and the weaker DSCR. As the market degrades, both restaurants would be in trouble, creating increased strain on the bank’s capital.
At the point of default of the weaker restaurant, the property could still be a going concern with potential sale value. Most likely inventories would be reduced with, hopefully, a corresponding decline in payables. The value of equity in the property could still be significant to the proprietors.
The strong restaurant has a higher probability of successfully absorbing a shock to revenues, in the magnitude of roughly 10-20 percent, which reduces NOI between 20-40 percent based upon the operating leverage of the property. Even with such a reduction to NOI, the DSCR remains strong enough to prevent the loan from defaulting.
Example: Post-pandemic restaurant under stress
Both strong and weak restaurants cease normal operations. Now we are talking about revenue shocks somewhere between 60 and 80 percent. Importance of earnings would be replaced by the reserve (generally off-balance-sheet) capital available to the respective proprietors. The property would, at least temporarily, become illiquid. Food inventories would degrade rapidly, and trade payables could reflect seniority to bank obligations. Proprietor’s capital would continue to erode under the pressure of existing leases and other fixed asset obligations.
Loan deferrals could delay defaults, but monthly costs would continue to drain the proprietor’s capital.
Historically strong operating performance and good classification levels would fall by the wayside, as the owner’s capital continues to be leached during the duration of the pandemic. The end of the pandemic would leave the restaurant (if it survives) facing a potential recession whose magnitude and duration is yet to be determined. Changes in consumer patterns further complicate the picture.
The credit degradation of a restaurant and its potential recovery in a general recession is completely different than in the conditions created by this pandemic. Historical data has little relevance and will be misleading. Weak small restaurants owned by wealthier individuals might recover much faster than larger and stronger restaurants where the principals have limited capital. In fact, in many cases it will be the capital held outside the restaurant entity that might ensure its reopening.
Other SIC codes, for example, hotels, will again have their own unique earnings issues. The all-important value of location might change permanently, which would in turn affect its breakeven occupancy rates and more.
Wrap Up
The same characteristics of a restaurant or hotel that make it a salvageable credit in 2008 will not be the same as in 2020. Proper credit analysis, uniquely applied to each relevant SIC code, is of paramount importance and must form the foundation of any post-pandemic stress test/strategic plan. The DSCR is a highly misleading metric because it ignores the strength/weakness of the obligor behind the underlying business. A basic stress test based on NOI shocks doesn’t work when revenue shortfalls are of the magnitude demonstrated by the pandemic. A stress testing process that does not work actively with the lending and credit departments to evaluate the new credit criteria necessary to properly classify loans in the post-pandemic environment is doomed to fail.
Stress testing in the pre-pandemic world was generally regarded as a regulatory exercise. In the post-pandemic world, stress testing must first be used as an essential tool for management in its strategic planning process. It can next be used as a way for directors to ensure shareholder protection and finally, to satisfy regulatory needs, but only after bank management teams gain a full understanding of how the results affect their ability to guide their banks through and beyond the crisis.
