A Bank Stress-Test Do-Over?

Tuesday morning, the head of the Congressional Oversight Panel (COP) of the TARP program, Elizabeth Warren, called for a “stress-test do-over.” The reason is that the original stress test might have been based on economic assumptions that were too optimistic.

This was a point that I and our banking analysts Eric Rothmann and Neena Mishra made repeatedly before the original test results were made public (and Zacks was not alone in that assessment either). I certainly agree, particularly the baseline assumptions. The more adverse scenario is what has been playing out so far, or at least the actual numbers are more in line with it than with the baseline.

There were three key sets of economic assumptions that went into each stress test scenario: real GDP, unemployment and housing prices. Relative to the scenarios, the economy has been doing best on the GDP front, with an actual decline of 5.7% in the first quarter. That is right near the middle of the two scenarios, with the baseline looking for a 5.0% decline and the more adverse scenario postulating a 6.9% decline.

As shown on the first graph (from http://www.calculatedriskblog.com/), on the unemployment front, even the more-adverse scenario has already proved to be too optimistic. In the first quarter, the actual unemployment rate averaged 8.1% rather than the 7.9% rate envisioned by the “worst case” scenario of the test. With two thirds of the second quarter under our belt, the unemployment rate is averaging 9.15%.

For the unemployment rate to actually average the 8.8% seen by the more adverse scenario, the June unemployment rate would have to drop to 8.1%. Well I guess if that happened it would sure end the debate over the effectiveness of the stimulus package, but it seems highly unlikely to me. The unemployment numbers on the baseline scenario are a joke — June’s unemployment rate would have to plunge to 6.6% for the quarter to average 8.3%.

Also note that May’s 9.4% rate already far exceeds the worst rate forecasted by the scenario for the whole cycle, and is higher than the third quarter average projected by the “more adverse” scenario. As I have pointed out repeatedly, the unemployment rate will continue to rise for a long time after the recession is officially over.

On the housing price issue, the second graph (also from http://www.calculatedriskblog.com/) shows that the actual data is much closer to the “more adverse” scenario.  While housing prices have returned to near historically normal levels based on relationships to rents and incomes, there is nothing that says that they will not overshoot to the downside just like they overshot so dramatically to the upside.

Also, the recession is causing incomes to fall. There is at least lots of anecdotal evidence of rents falling, even if it has not yet shown up in the owners equivalent rent or the “rent rent” parts of the CPI. Thus, those price to rent and price to income figures may be shooting at a moving (falling) target.

There is still a huge inventory glut of both new and used homes, particularly relative to the current sales rates, and distressed sales are making up almost half of all used house sales. There is a second huge wave of foreclosures coming — this time based on the Option ARM and “liar loan” mortgages.

Unlike the first sub-prime wave, these were mortgages that went to the “good neighborhoods” and gated communities, not to the “wrong side of the tracks,” the way the sub-prime loans did. Those “good neighborhood” houses are more likely to be move-up homes rather than starter homes. However the current used home sales that are distressed sales (mostly still sub-prime-oriented and at the low end) do not generate the chain reaction sales that normal low end sales provide. It is very easy for me to see home prices continuing to follow the more adverse path.

The good thing is that, with the exception of the unemployment rate, things have not been worse than the more adverse scenario, and the banks have been able to raise the capital needed to withstand that scenario. The release of the original stress test results focused almost exclusively on the more-adverse scenario. Thus, provided the economy does not turn out to be “more adverse than the more adverse,” none of the “too big to fail” 19 is likely to actually fail.

That, however, does not provide much of a margin of safety. It is sort of ironic that the chief overseer of the program is coming out with this call on the very day that 10 of the 19 have gotten approval to withdraw from the program. On the other hand, the stress test did provide pretty clear delineation between the strong banks like J.P. Morgan (JPM: 35.26 0.00 0.00%) and Goldman Sachs (GS: 149.31 0.00 0.00%), which are at the front of the line to get out from under the TARP, and the weaker institutions like Citibank (C: 3.41 0.00 0.00%) and GMAC (a.k.a. Ally Bank), which are a long ways away from being free of the need for government help.

Personally, I would like to see this process de-sensationalized. Rather than make it a big deal based on the current economic numbers, institute it as a regular part of the bank examination program.

One of the regulatory reform proposals that is working its way through Congress is the creation of a systemic risk regulator. Perhaps a key part of the duties of such a regulator should be the running of such stress tests on a regular basis — say quarterly or semiannually. The standard baseline for a routine stress test would be based on the “Blue Chip” forecast at the time, and a more adverse scenario based either on a set level of GDP growth below that level (say 2.0% less) and a set unemployment rate (say 2.0%) above the baseline, or developed from the average of the ten most pessimistic forecasters of each variable in the survey.

This would be just another part of the regulatory framework and the data just another economic release, rather than a big event. In good times it would just be part of the regular flow of data on the markets, but it might give us a heads-up about potential problems next time around.