But the president must have hit Larry and Tim with one of those visionary-looks-off-into-the-future of his, so stress-test it was. The banks submitted their data weeks ago, and were instructed by the regulators to keep mum on the process. The entire financial industry has been on tenterhooks ever since.  

As you have perhaps gathered by now, I can’t think of a single, positive thing to say about the stress test concept, the process by which it will be carried out, or outcome it will produce, no matter what the outcome is. Nothing good can come of this and, under certain, non-far-fetched scenarios, it might end up making the banking system's problems worse.

Let’s take a look at what’s so wrong about this misbegotten idea:

1. The underlying premise of the test is simply mistaken. From what I gather, certain economists reportedly convinced President Obama that the problem with the economy is that banks are refusing to lend, and are instead hoarding capital in order to shore up their wobbly balance sheets. The president seems to think, therefore, that if the banks are force-fed new capital, they can expand their balance sheets and resume lending, which would in turn help the economy begin to recover.
   
   There’s just one problem with this thread of logic: its basic premise is wrong. The problem with the economy isn’t that banks that aren’t lending, it’s that the non-banks that make up the “shadow banking system” have gone into forced hibernation. We know the administration knows this, since it has expanded the TALF program, which is designed to help jumpstart the securitization markets on which the non-banks depend for their funding.
   
   As for the banks, well, the chart below tells the tale. It was developed by my old partner, Robert Albertson, now the chief strategist at Sandler O’Neill, and it shows bank loan growth during the current recession compared to the four previous recessions. If anything, bank lending has stayed remarkably robust in the face of shrinking demand.
   
   
   
   Conclusion: the economy’s problems are not the result of banks’ unwillingness to lend. Banks are more than eager to make loans in areas they desire and understand.
   
   So the worst thing about the stress test is that the underlying reason for carrying it out in the first place, to give the banks new capital so they’ll re-start their lending, is simply wrong.
    
2. The name “stress test” was poorly chosen. Say the word “test,” and you imply that there are “right” and “wrong” answers and that, in turn, some banks will pass and others will fail. But these sorts of exercises shouldn’t work that way. Those of us on the outside of banks (and even some on the inside) have a hard enough time predicting a bank’s earnings and balance sheet changes one quarter ahead of time, let alone what might happen (under hypothetical conditions, no less!) over the next eight quarters.
   
   Rather than refer to the exercise as a stress test, Geithner should have, in my view, described it as a “sensitivity analysis,” since its output will have none of the precision that one typically associates with a test result.
   
   And in fact, many bank analysts (us included) perform such sensitivity analyses all the time. But if you compare one against another in any kind of detail, you won’t be impressed with their rigor, and will quickly be struck by the fact that, of necessity, conducting them requires the piling of one assumption on to another. The table below summarizes the cumulative loss forecast for large U.S. banks, as estimated by several analyses:
   
   
   
   
   
   At first glance, the broad range of estimated losses may not seem like a big deal. If you throw out the high number and throw out the low number, you’re left with an estimate that centers around 9% or so, and you may come away with a sense that these numbers might actually have real meaning. That would be a mistake. The similar bottom-line numbers mask huge differences among analysts as to loss rates for the basic building blocks of the analysis: loan type.
    
   The next table illustrates what I’m talking about. It breaks the cumulative loss estimates above down by loan type.
   
     
   
   Yikes! There are some huge differences. For instance, S.C. Bernstein sees cumulative losses in construction and development loans of 9%, while the private equity investor assumes 25%. That’s almost three times higher. Meanwhile, Deutsche Bank’s assumption of 28% cumulative losses in home equity is quadruple Bernstein’s 7% estimate.
   
   Who knows which are right? Nobody, that’s who. My point is that this sort of stress test/sensitivity analysis, at this point in the cycle, is a very, very crude tool. And what makes it even worse, none of these analyses adjust for unique characteristics of individual banks, such as their geographic footprint, historical record, or current performance. Neither, presumably, will the stress-testers in Washington.
   
   But if the Treasury Department chooses to modify the stress test results individual banks provide to conform to industrywide averages by loan category, they will be committing a huge analytical mistake.
   
   We think our approach is more rigorous. In particular, we adjust for the unemployment rate in the given bank’s footprint, the historical credit experience of the bank, and differences in loss timing by type of loan.
   
   However, even after we do all this we’re still left with, necessarily, a very crude predictor of future losses. The only reason we go through the process at all is that it can be helpful for relative ranking purposes and for comparison as quarterly results are reported.
   
   In any event, the name “stress test” implies a level of precision for predicting future losses and earnings that simply does not exist. I only hope the Treasury doesn’t overly adjust the banks’ own numbers and force unneeded capital raises based on the outcome.

3. The Treasury Department’s plan for disclosing the details of the stress test results is confused at best—and might be catastrophic. At this point, all we know is that regulators have told the banks not to talk about the stress test results on their quarterly conference calls. The Treasury and the regulators themselves seem to be confused about what they will eventually disclose.
   
   I don’t blame them. It’s currently illegal to disclose the results of a bank examination or a bank’s CAMELS rating—and for good reason. As former OCC head Gene Ludwig told the Wall Street Journal on Wednesday, “You can create a run on a bank pretty quickly," by making a bank’s test results public if they’re less than stellar. Even though, as we’ve seen, those results can be based on wildly speculative assumptions.
   
   I frankly don’t see why the results of individual bank stress tests should be disclosed, since those results figure to be even more arbitrary than the (already confidential) tests banks undergo as part of their annual examinations.
   
   Once a year, however, bank regulators do share the results of their Shared National Credit reviews in aggregate. I expect a disclosure along those lines, at the very least. As it is, the Treasury Department has painted itself into a corner with all its hullabaloo about the stress tests, and its emphasis that they are the cornerstone of the administration’s plan to help the banks and the economy. So the Treasury will have to disclose something. The more high-level and less bank-specific it is, the better. Otherwise, the government could conceivably be putting some perfectly healthy institutions at real risk.
   
   The stress test was a bad idea in all respects: its core premise, its misleading name, the promise of transparency, and the dilemma over the disclosure of results. My advice to the administration: disengage from the process as quickly and quietly as politically possible.

What do you think? Let me know!