At root, all financial institutions do is manage risk. It's a fact of banking that not all lent funds finds their way back. But in a crisis scenario that cost becomes an existential threat. So it's incumbent upon lenders to develop models that predict how and when a crisis might occur, an exercise that can become a sort of philosophical exercise in determining what is knowable.
Something that apparently wasn't knowable: the global financial crisis, which made mincemeat of these models. So among Dodd-Frank's requirement was that big banks undergo regular Comprehensive Capital Analysis and Reviews – i.e., stress tests – in which the Federal Reserve forces the banks to give their risk models a whirl in a hypothetical crisis.
That seems like a reasonable ask, given how poorly big bank models fared during the crisis. But, as a new Government Accountability Office review finds, the same uncertainty around models that plagues the banks also bedevils the Fed. How can we know that its models for judging the models are sound?
The GAO recommends, among other things, that the Fed should “apply its model development principles to the combined system of models used in the supervisory stress tests.” In other words, stress-test the Stress Test.
If you're going to make banks subject their own models to rigorous scrutiny, you're going to need a model to test them – which the Fed has. But who is to say that that model is sound? Someone is going to have to test it.
“The Federal Reserve’s organizational structure for the stress tests does not include a formal process through which model development or risk management at the aggregate—or system-of-models—level is implemented,” the GAO found.
But you can't really get any more central than the central bank. Eventually this becomes a epistemological problem. Who is the ultimate authority on what can be known in banking? Can there really be a model of all models?
Of course, there's a practical takeaway here: The Fed's models determine how much capital banks need to sock away and how much they can hand out to shareholders, so there's certainly some dollars-and-cents motivation out there to make sure the Fed's models are calibrated favorably. But again, who's to say?
The GAO made a bunch of other recommendations in the review, which was spurred by friend of finance Rep. Jeb Hensarling. Echoing the complaints of banks, the watchdog recommended that the Fed make its guidelines more transparent and explain its decisions with greater clarity. As the Donald prepares his reign, some are expecting a much leaner stress test process.
But first, the Fed has to get its metaphysical house in order.