Remember Those Stress Tests?

I’m curious to know how banks’ 2009 final results compare to the projections in the stress tests. My suspicion is that JPMorgan and Goldman did better than projected, but Citi may have done worse. Ideally you would compare both the new loan losses recognized over the year and the profits from current operations. But there are a couple of problems with doing this. One is that the stress test results were for 2009-2010 combined, without the separate years split out. The other problem is that it’s not immediately obvious how to map the line items from the stress test results to the line items on a bank’s income statement (or to changes on its balance sheet). I might be able to figure it out with a lot of study, but I might not.

Does anyone know of someone who has already done an analysis along these lines? Or does anyone know how to do the mapping correctly?

By James Kwak

16 responses to “Remember Those Stress Tests?

  1. Don’t worry. They all still pass.

  2. as a rough proxy, look at the estimated net chargeoffs for loans for each of the companies in the scap. you can make a proxy with a straight line estimate of what the quarterly loss should be. if the actual net chargeoffs come in below the quarterly estimates, then the banks should be in good shape since i would guess the worst should have occurred this past year. running the numbers, almost all of them had net chargeoffs well below the scap assumptions, with the worst being citi at around 75% of the assumptions.

  3. You forgot a third problem, which is really preliminary to the two you note: loan loss reserves are highly subjective and regulators who want to promote the appearance of soundness (as distinguished from actual soundness) too often turn their eyes the other way when loan loss reseves are calculated (or backed into, as is often the fact). I had the great pleasure to sue a number of folks for fraudulent understatement of loan loss reserves. But these are extremely difficult investigations that won’t go far without dedicated regulators. And we don’t have many of the latter in this day and age.

  4. I’d urge you remember Marty Whitman’s admonition that the central reality of accounting is not what the numbers are, but what they mean. This reality holds especially for financials whose “numbers” are rife with subjective judgements applied to liabilities and assets whose values and the accounting for which are all in unchartered territory.

  5. I might be able to do it. Do you have the full stress test you can send me? Pick a bank and I will take a look.

  6. Cops:
    http://www.structuredfinancenews.com/news/-202622-1.html

    Is that a part of the stress tests? Could this be a test case for bigger fish (or squid) that misled investors?

    “According to the SEC’s complaint and order, State Street sent investors a series of misleading communications starting in July 2007 about the effect the subprime market turmoil on the Limited Duration Bond Fund as well as other State Street funds that invested in it. But, at the same time, State Street provided particular investors with more complete information about the fund’s subprime concentration and other problems with the fund. These other investors included clients of State Street’s internal advisory groups, which provided advisory services to some investors in this fund and related funds.”

  7. James,

    At first, dd’s method sounds legit, but I think something has been forgotten here, namely, the relaxation of mark-to-market accounting. I thought the FASB voted to allow banks to only have to value the liquid portion of their toxic assets or rather, with each bank determining for itself whether the market for their assets is liquid/illiquid.

    Using dd’s method, of course the optics look good! The banks aren’t really trading the toxic assets, most of them are sitting on the books and probably aren’t getting valued. At the very least, the banks are carefully selecting which ones to charge-off now so that they don’t hammer their stock price given the past year’s meteoric rise out of the abyss.

  8. I think dd’s method should produce adequate results for loan losses. Jimmy J’s criticism is completely off-base here. All US banks are required to adopt, and adhere to, very strict chargeoff policies by their regulators. While many on this site may not believe this, regulators take at least this part of their job seriously. As such, chargeoffs generally reflect accurately losses on loans that have been non-performing for 90-180 days.

    The next critical piece of the equation is ascertaining realized *securities* losses or “Other Than Temporary Impairments” (OTTI). As a general matter, these have been de minimis in 2009, but you can go bank-by-bank and see who has taken OTTI or realized securities losses through a combination of the MD&A, notes, income statement, and cashflow statement. Of course, many have argued that UNREALIZED losses are a critical component, and are vastly understated. This may have been true in 2008, but with the dramatic improvements in the securities markets, is unlikely to be true AS A GENERAL MATTER now. Banks are actively selling many “toxic” assets at or near their marks. While this assertion is purely anecdotal, it is so widely supported by the reductions in unrealized losses across every financial institution of every type, that I tend to believe the anecdotes.

    This being said, I suspect that James’ question really had to do with identifying loan losses by category and comparing them to the SCAP. Some banks disclose this in their quarterly calls, and some don’t. Richard Ramsden at Goldman has some good data on this. That said, one thing you can do is compare NPA ratios across different loan categories to the SCAP-implied losses and make reasonable assumptions on severities. Thus, you can see how actual NPA’s are tracking vs. what SCAP would imply.

    What will you find? Every bank is so far tracking substantially better than the SCAP adverse scenario and generally in-line with the SCAP base scenario. It seems the stress tests were quite stressful indeed, and the banks are far more solvent than many give them credit for.

  9. Change the accounting rules at the governments behest to cloak lossess and enhance the banks standing, feed at the government trough to the tune of $14 TRILLION, and screw over the American people on deceptive and excessive fea’s and harsh loan requirements, funnel billions of dollars offshore carefully evading taxes, and…yeah…the TBTF banks look good on paper. But paper is paper. The realworld is not so easily manipulated, and in the realworld, – the TBTF banks are an insovlent den of vipers and thieves bruting PONZI schemes, and systemically lying to the socalled government, investors, and the American people.

    The next bubble is rapidly inflating. Nothing has changed. Fasten your seatbelts, and put your trays in the upright and fixed position, – because we are about to enter some turbulence!!!

  10. Some better, but, needless to say some worse. My contention is, however, had they been done under the old valuation methods for the toxic assets, no one would have passed then, and no one would pass now. Of this I am pretty well convinced. Just call it a hunch, but with the new accounting rules, they can call anything valuable, or just ignore it, their choice. Kind of like being in a poker game and being allowed to make up any hand you want, but then they are all “one-upping” Cool Hand Luke for the gambler of the century awards. Still!!!

  11. Insight into bank’s budgets for 2010 would help ascertain trends. Further, banks lending levels are somewhat reflexive to anticipated losses. Third, capital retention through zero dividend policy is indicator, as is non-repayment of Tarp. Finally, read between the lines if morose tone of certain CEO’s on last earnings call.

  12. JJ, I don’t have anything other than the document the Fed released on 5/7/09 (link in the post above). There is one page at the back per bank. I don’t know if more detailed results were every leaked–I never heard about them.

  13. Having done this for an S-CAP bank, and with proprietary access to the bank’s year by year projections and how to do the mapping, performance has been 30% or so better on PPNR (pre-provision net revenues) through 4Q09, and losses and provision expenses have also been 20% or lower on average for most Y9-C categories that the S-CAP tracked.

    The big issue is whether the timing/periodicity of the S-CAP particularly on losses is right. In other words, is there some bulge of losses that shows up in 2Q10? Don’t know yet – but looks unlikely absent additional shocks.

    You’ll have to trust me on this – I can’t tell you more than this.

  14. According to reported GAAP financials, Citi realized net credit losses of just over $30B in 2009. Based on NPAs and changes in NPAs, another $30B or so seems likely to be charged off. This would bring the two-year total to $60B, a little higher than the more adverse number for loan losses.

    For the other banks, it is difficult to see how charge-offs has progressed due to the impact of purchase accounting. Some banks provide disclosures which helps put the pieces together. Overall, loan losses look like they might be reasonably close or a little better than the more adverse scenario, at least based on my math. Are you (or is anyone else) seeing something different?

    I have not looked at securities in detail, but would imagine the situation is quite a bit better than the adverse scenario.

    I think that trading and counterparty losses have been minimal or zero (most of the BHCs have generated healthy trading profits), so these should be much better than the adverse scenario.

    Who knows what falls in the “other” bucket, but this might end up being too low if the big banks need to repurchase a lot of loans from MBS.

  15. Is the 20% for everything (including securities, trading, and other) or just loan losses?

    A key problem with the SCAP projections is that there was no attempt (at least to the best of my knowledge) to consider how things will look at the end of the two years. Using Citi as an example, there is a huge difference between (a) $59B of loan losses for 2009-2010 and $12B of non-accrual assets (50% less than at year-end 2008) and (b) $59B of charge-offs and $36B of NAAs (a 50% increase from 12/31/08).

  16. Several of the banks in the SCAP have published credit loss results compared to the SCAP assumptions. (Suntrust, Fifth Third, and Regions are three that come to mind…)

    This is the most extensive disclosure I’ve seen. Pages 12 and 13 of this link shows Fifth Third’s performance versus both the stress and baseline scenarios the bank submitted in May. (I haven’t seen any of the other banks disclose their baseline credit loss expectation, that would be very interesting.)

    Fifth Third’s charge-offs have been much better than the stress assumption (which drove their capital requirement) and also than their (much lower) May baseline as well.

    They provide updated stress and baseline scenarios for 2010 here as well.

    Looks like 2009 charge-offs were about 40% below the stress test and their base case scenario for 2010 charge-offs is less than half of the stress test.

    The presentation also focuses on pre provision earnings, reserves, and capital, which all look pretty good and must be better than stress test.

    http://ccbn.10kwizard.com/xml/download.php?repo=tenk&ipage=6746611&format=PDF