You are here

Stress Tests Still a Hurdle at Many Banks

First, in the interest of full disclosure, I should make it clear that I’m a fan of stress tests. Stress tests, such as capital adequacy testing and the Dodd-Frank Act Stress Test (DFAST), simply enable Basel governments—like the U.S. Federal Reserve—to treat the banks that they regulate as borrowers whenever a speculative bubble bursts, which now appears to be every 12 years or so. In fact, the stress-test process looks a lot like the due diligence that banks themselves perform on their own borrowers, including pro forma financial analyses based on dire economic projection scenarios. 

That said, it looks like the largest banks are still struggling with stress tests. This is interesting, as one would think that institutions skilled in performing due diligence on borrowers would be able to perform that task on themselves for the benefit of their own financiers of last resort. 

Yesterday, the Fed released results of the most recent annual stress-testing season. Here’s the quick synopsis:

  • “Qualitative concerns” caused the Fed to object to the plans of Citi, HSBC, RBS Citizens, and Santander. 
  • Capital inadequacy, on a projected basis, caused the Fed to object to Zions’ plan.
  • Capital adequacy has improved for the group of banks that pose systemic risk.  Their aggregate Tier-1 common equity ratio has improved from 5.5% in Q1 2009 to 11.6% in Q4 2013. In other words, the strength and financial resilience of non-government financial institutions capable of causing economic mayhem appears to be improving. 

As an expert on the topic of stress testing with recent surveying and reporting that mirrored very closely yesterday’s Fed announcement, here is what I think is important:

  • The incidence rate for objections on a “qualitative” basis is still high. Four out of 30 may not seem like a high rate, but the goal here is to prevent banks from turning to taxpayers as a last-resort capital source. A success rate in the high nines is a reasonable expectation. 
  • Although the Fed used the term “qualitative” nebulously to describe its objections, I’d bet the farm that data management and risk analysis were two primary issues.
  • Data management in stress testing probably hampered banks’ submissions because data sets are so heterogeneous, scattered across the enterprise, and difficult to aggregate. Even worse, when banks are able to aggregate risk-related data for stress testing seemingly innocuous changes, such as the combining of loan portfolios with different risk rating scales, data can cough badly, resulting in data-management fire drills and eroded trust in the data with regulators—banks’ second-most important audience (the first is the banks' own CROs and their teams).
  • Banks’ risk analysis is closely related to the data management problem and also a likely driver of the adverse “qualitative” findings. In addition to canned scenarios from the Fed, banks are required to perform “bank-specific down-case scenarios.” But it becomes extremely difficult to accurately craft such a scenario when a bank struggles to combine its risk-related data sets. This, I’m confident, underlies those “qualitative” concerns the Fed mentioned yesterday. 

Here’s what I think bankers should do:

  • Integrate and automate: The fewer manual processes and workarounds there are in the life cycle of a stress-test data point, the more the Fed will trust your data, as well as the resulting bank-specific down-case scenarios crafted by your stress-test team.
  • Monitor data aggressively: The life cycle of a stress-test data point is nasty, brutish, and long, and its endpoint involves an audience with the Fed. The more continuously your data cubes and Excel spreadsheets are monitored, and equipped with finely tuned sensors and alerts, the fewer credibility-crushing data brushfires you’ll have during the stress-test season and the more accurate your scenarios, models, and calculations will be. 
  • Know that the bell tolls for thee: Stress testing is now required of all banks with over US$10 billion in assets. The logic, demands, and complexity of these tests don’t necessarily scale downward with bank size. Smaller banks with limited analytics and compliance capabilities should begin their automation plans yesterday.