by Calculated Risk on 5/09/2009 06:07:00 PM

The results of the stress test showed that the bulk of future bank losses will come from accrual loan portfolios, as opposed to exotic securities (most of the writedowns to date have come from these securities). Here is the quote from the stress test results:

"The bulk of the estimated losses – approximately $455 billion – come from losses on the BHCs’ accrual loan portfolios, particularly from residential mortgages and other consumer‐related loans. ... Estimated possible losses from trading‐related exposures and securities held in investment portfolios totaled $135 billion."
Federal Reserve, The Supervisory Capital Assessment Program: Overview of Results, May 7, 2009

Here is Professor Roubini's comment (from Ten Reasons Why the Stress Tests Are “Schmess” Tests and Why the Current Muddle-Through Approach to the Banking Crisis May Not Succeed):

Second, the capital/needs of these banks depend on a race between retained earnings before writedowns/provisioning that will be positive given a high net interest rate margin and the losses deriving from further writedowns. It appears that regulators have overestimated the amount of such retained earnings for 2009-2010. The IMF recently estimated that retained earnings (after taxes and dividends) for all US banks – not just these 19 ones – would be only $300 bn total over the 2009-2010 period. The stress tests – instead – assumed much higher retained earnings - $362 bn - for these 19 banks alone for the 2009-2010 period in the more adverse scenario. Since these 19 banks account for about half of US banks assets if one were to use the IMF estimate of net retained earnings for these 19 banks their net retained earnings for 2009-2010 would be $150 bn rather than the $362 bn assumed by the regulators. While the IMF may have been too conservative in its estimates of net retained earnings it appears that regulators may have been too generous to these 19 banks in forecasting their earnings in an adverse scenario. Thus, ex-post capital needs will be significantly higher if net retained earnings turn out to be lower than assumed in the stress tests.

And a more optimistic view from Jan Hatzius:

"The enormous volume of securities writedowns in 2008 basically overwhelmed the pre-provision earnings power of the banking industry. ... We think that most of the loss recognitions in securities has now occurred. U.S. banks will still need to recognize substantial losses, but this will be mainly the provisions for whole loans that are spread out over a somewhat longer period of time, and somewhat more predictable, and therefore won't be as violent as the securities write-downs, and as a result these remaining losses can be offset via pre-provision earnings to a much greater degree."
Jan Hatzius, Chief Economist, Goldman Sachs, Conference Call, May 8, 2009 (no link)

This brings up a key concept:

Imagine a bank holds a RMBS (Residential Mortgage Back Security). Forget about tranches - just imagine the security is based on 100 mortgage loans. All of the loans are current, but the security is actively traded, and the price falls to 50 cents on the dollar because investors believe that there are many default (and losses) coming. The bank has lost 50% immediately. The bank holds the security, not the loans - so it is the change in the value of the security that hits their income statement.

Perhaps the bank believes the most profitable thing to do is just keep the loans in its own portfolio (fair value accounting principle called "highest and best use"). Now the bank also has a portfolio of 100 loans with exactly the same characteristics as the RMBS. The Fair Value estimate for income producing loans for which there is no available market or counterparty will be based on the Income approach (discounted future cash flows). As before all of the loans are current, so the bank takes writedowns based on estimates of discounted future cash flows (they are being held to maturity). As the losses, both current and future, become estimable, the bank takes the writedown.

A large portion of Alt-A (like Option ARMs) is held on bank balance sheets. So the building Alt-A crisis will be written down as the losses are estimable. Tanta mentioned this last year in: Subprime and Alt-A: The End of One Crisis and the Beginning of Another

If the "subprime crisis" was about "exotic securities," the "Alt-A crisis" is going to be about bank balance sheets.

And this brings up a second key point:

The Fed estimated both future losses and future earnings. As Dr. Roubini noted, there will be a race between losses and earnings - and if the Fed overestimated earnings or underestimated losses, the banks will need additional capital. If Hatzius is correct, the banks will win (or draw) the race - if Roubini is correct, the banks will lose.

But whether the banks win or lose the race, the rapidly rising defaults for Alt-A (and HELOCs and Jumbo Prime) loans will impact house prices in neighborhoods where the loans are concentrated (mostly mid-to-high end areas).