Wednesday, June 10, 2009


This from the Congressional oversight report regarding the Bank Stress Tests.  Please hold your chuckling until reading the entire quote...

And, as usual for value-adding governmental actions, we must now learn what "SCAP" (Supervisory Capital Assessment Program) and "ILR" (indicative loss range...what a strange concept) entail.
"... All the same, the stress tests should not be taken for more than they are.  As indicated above, they were conducted within the present supervisory context only, and they are a  temporary two-year projection of a one-time capital buffer that need not be rebuilt.  They do not  model BHC performance under “worst case” scenarios, and as a result they do not project the capital necessary to prevent banks from being stressed to near the breaking point.  Most important, for some observers, they do not address the question whether the values shown on bank balance sheets for certain classes of assets are too high; by restricting themselves to a two-year time frame, their conclusions thus do not take into account the possibility that the asset values assumed (particularly for so-called toxic assets) may undervalue bank liabilities to the extent that those liabilities result in losses after 2010. 

   "... From what we are able to discern about the specifics of the stress tests, the Fed’s approach appears to hybridize numerous canonical risk modeling approaches, and in broad strokes seems most consistent with a conditional loss approach. That is, the stress tests attempted to elicit information from the nineteen largest bank holding companies (BHCs) about likely losses that would be visited upon their asset portfolios over a two year time horizon under specified macro-economic conditions. The implementation of this approach ultimately boiled down to a four-step process. In the first stage, SCAP designers posited two macroeconomic hypothetical states – a “baseline” scenario and a “more adverse” scenario. Second, within each of these states, the Fed attempted to for-mulate expected Indicative Loss Rate (ILR) ranges within each asset class and across all institutions, which reflected estimates of both the frequency of default and losses given default under each scenario. In the third, the BHCs and the Fed applied a process that allowed each BHC to vary from the predetermined ILR ranges (above) into loss and resource estimates tailored at the firm level. In the fourth step, the banks reported their asset and exposure levels under each macro-economic scenario, which implied what (if any) additional common equity buffer was necessary at the BHC level.

 "... At the same time, SCAP’s design and implementation do leave some open questions in our minds. Perhaps the most significant of these questions concern the SCAP’s transparency and replicability. Each of the four stages outlined above evidently involved the combination of quantitative and qualitative measures. For example, in the initial setting of ILRs, the Fed evidently attempted to synthesize numerous alternative macro-economic models (which themselves produce noisy estimates of losses) with subjective judgments of experts in different asset classes. The precise mechanism for combining these various inputs, however, was left largely unspecified. In addition, the process by which the initial ILRs became tailored to each BHC in Stage 3 appeared analogously opaque. While such synthesis is sometimes a good way to deal with model uncertainty, data availability, confidentiality, and measurement error, it renders the results virtually incapable of replication (or even much detailed understanding) by an outsider. This lack of transparency and replicability is a potential cause for concern, and it ultimately confines our analysis to a general assessment of the program’s broad-brush approach.  In addition, we discuss a number of other concerns that we believe also to be material.  These include concerns that the SCAP was insufficiently sensitive to BHC ownership structure; that it neglected other sorts of micro- and macro-economic risks (such as interest rate, inflation, and cash flow / liquidity risks) that may be relevant in predicting loss ranges1 ; that the SCAP used a short time horizon (two years) that may have been insufficient relative to the maturity of the underlying illiquid assets; and that the Fed might have done additional robustness checks by varying the sizing of the cap or the measure of equity capital employed."

No comments: