We couldn't help ourselves yesterday, so we sounded off on pundits who keep applying the same stress testing methodology used on the group of 19 (G19) largest financial institutions in the country to community banks. Doing so can be about as dangerous as getting into crocodile infested waters with bait around one's neck because there are so many things wrong with such an approach. Whatever happened to rational stress testing that is applicable to the institution and/or situation being stress tested?
We begin by pointing out that some components of the test applied to the G19 relied on modeling results from Nationally Recognized Statistical Rating Organizations (NRSRO) that project defaults. Unfortunately, as we have seen during this credit crisis, when you feed data into the mouth of an open model without paying attention to what you are doing, the results can unexpectedly clamp down on you if common sense is not applied. A prime case in point is how these models suggested that sub-prime mortgage borrowers who lied about their income (or whose income was never verified) were somehow AAA-rated just because they were grouped into a large pile of cashflows and fed into a model. As we have seen from the crisis, such modeling has been proven to be much less reliable than originally expected in quantifying default potential. In fact, the historical look-back of such instruments was unable to include a significant down-cycle because the instruments were so new. Here again, assumptions leveraging the historical payment experience arising from higher credit quality prime single family residential loans was somehow equivalent to non-investment grade credit (sub-prime and Alt-A loans). It is not clear how much the results from such models were used or adjusted during the most recent G19 testing, but community banks rely on traditional underwriting with local customers when placing credits on their balance sheets and always include verification of income. In short, modeling is great as a starting point, but no model can take the place of a solid underwriter that knows the local market and the customers.
Further differentiators separating the G19 from community banks are business model and distance. Community banking by its very design is based on lending within a local community. While that has its own risks, it also gives community bankers an edge when it comes to reacting faster and modifying practices as localized conditions warrant. The G19, by contrast, follow orders pumped out from a headquarters office that can be thousands of miles away. This can be further exacerbated if that bank holds assets in the form of securitizations or pools, where individual loan terms can not be modified in response to quickly changing conditions. Distance and business model differences mean the G19 have a delayed response, which mutes action and can end up slowing response to stresses that appear quickly in regional areas. Community banks do not have that issue.
While we like larger loans as a way to reduce cost of origination, they can limit liquidity. While this may sound counterintuitive, consider that within a stressed or consolidated marketplace, there are very few players that can trade very large credits. In these cases, when liquidity dries up, it does so in a hurry. As we have seen with the credit crisis, when liquidity and trust erode, the government may end up as the only outlet. Community banks in comparison almost always have other outlets for smaller loans. The sheer number of participants available on smaller credits to provide liquidity can significantly lessen the impact and magnitude of any potential loss.
Finally, applying the G19 stress test to community banks presumes those who are tested have a brain the size of a crocodile (about the size of a walnut). The test is a point in time and loss calculations assume individual banks will not or cannot proactively manage stresses appearing in their portfolio for a 2Y period. We can't think of a regulator anywhere that would let a community bank with a stressed loan portfolio just submerge under the water and reappear 2Ys later.
We conclude our discussion on the problems in applying the G19 stress testing to community banks with a final thought. The value of stress testing of any type is to identify areas of excessive risk and then taking action to mitigate that risk through planning and execution. Plans can include such factors as capital raising, adjusting balance sheet mix, altering underwriting practices or even restructuring performing credits. Otherwise, randomly applying stress testing without context can be as dangerous as buying into crocodile tears.