In Depth Analysis: CalculatedRisk Newsletter on Real Estate (Ad Free) Read it here.

Saturday, November 17, 2007

Fannie Mae's Credit Loss Ratio: Fuzzy Math or Fuzzy Reporter?

by Tanta on 11/17/2007 11:47:00 AM

This is going to be a long post. It is going to attempt to answer the question stated in the post title. It is also going to function as further proof of the old axiom that you can create quite a ruckus in 150 badly-chosen words, but it takes ten times that many words (at least) to return some sanity to the discussion. “Gotcha” reporters of course know this, which is why they do what they do. Most people don’t have the time or desire to wade through the high-attention span Nerd part to evaluate the reporter’s claim. That it’s a deadly serious business for anyone who owns shares in a publicly-traded company being compared to a criminal conspiracy on the basis of a misunderstanding of accounting rules doesn’t seem to bother writers who just want a “scoop.”

The ruckus started last week. On November 9, Fannie Mae released its 10-Q for the third quarter. This is the first time in years that Fannie has gotten a timely Q out; its 10-K for 2006 was released just in August, it never filed Qs for 2006, and the Qs for the first two quarters of 2007 were also just released in November. That little detail is important to this story. I assume everyone knows the long wretched saga of why Fannie Mae has been so far behind with its SEC filings. The point is that, in catching up, they have released a flurry of reports in a short period of time, which don’t have the same numbers on them (they shouldn’t; they are for different periods), and since they never reported quarterly numbers for 2006, we don’t have the by-quarter breakout that would provide details for some of the whole-year numbers reported in the 2006 K.

On November 15, Peter Eavis of Fortune Magazine published a breathless essay accusing Fannie Mae of having changed the method it uses to calculate its credit loss ratio in the Q3 filing. It is quite obvious that the presentation of this information has changed; the Q says so (page 54-55). Eavis, moving from a change in presentation to a change in calculation with intent to mislead at the speed of light, says “Uh oh. It’s Enron all over again.” Throughout the original article, he keeps referring to “bad loans” in such a way as to give the misleading impression that the metric in question, the credit loss ratio, is about reporting on delinquent loans, not on realized losses on defaulted loans in the current period. This allows him to accuse Fannie Mae of being “misleading” by not including fair-value write-downs on delinquent but not yet defaulted (not yet realized-loss producing) loans in the credit loss ratio.

Fannie Mae stock started to tank badly, and Fannie scheduled an analyst conference call for Friday morning to address this one very specific issue in one table in the Q. Fannie Mae explained, among other things, that the item excluded from the credit loss ratio calculation is, actually, included in net charge-offs on the consolidated financial statements. However, for the purpose of this specific metric, the credit loss ratio, fair value write-downs that have not yet produced an actual loss are excluded.

You can listen to the webcast here. Several analysts pointed out, quite nonconfrontationally, that they though Fannie could have provided more information to put this matter in context. Fannie agreed, and indicated that future disclosures would include more information. What’s amusing is that in two instances, analysts ended up asking whether in fact Fannie Mae wasn’t over-reserving for certain delinquent loans! Fannie Mae’s response was that they don’t think they’re under-reserving or over-reserving; they are simply applying GAAP rules for how fair-value write downs are taken. At no time in the conference call did anyone challenge this as a misapplication of GAAP rules.

So what did Eavis do, after the conference call and some delving into the credit loss ratio suggested that perhaps he merely misunderstood the math? He wrote a follow-up article on Friday, in which he continues to insist, even after Fannie Mae’s explanation of the issue, that the amount of exclusions from the credit loss ratio (that is, the amount of the fair-value write-downs on repurchased loans that are delinquent but not yet defaulted) is inexplicably large, and that these are forced repurchases, and that this is somehow sinister. This is after a conference call in which Fannie Mae explained, for those who have been living under a rock since last summer, why it is that write-downs on repurchased loans in Q3 can be many, many times the write-downs on loans repurchased earlier in the year, even if the total number of repurchased loans hasn’t grown all that much. There was this mere matter of a giant freeze in the mortgage secondary market, and spooked investors offering mere pennies on the dollar for delinquent mortgage paper, whether it was prime or subprime or something else. Eavis has to pretend to not remember that, I suspect, because his claim of “Enron all over again” is crumbling around his ears and he needs to keep kicking sand.

It is amazing to me that in light of all the real problems we have right now, we still want to go down expensive rabbit holes over “accounting tricks.” Nobody, least of all Fannie Mae, is trying to deny that there are severe problems in the housing and mortgage market, that large losses are being taken, and that this will hurt all over the place. I am not suggesting that Fannie’s Q is as clear as it could be, and I’m glad they indicated a willingness to report more color in future disclosures to make these numbers easier to evaluate. But writing a not completely helpful Q based on GAAP isn’t a crime in this country. I know a lot of people will argue that GAAP isn’t very helpful to the non-specialist. You get no argument from me about that, either. That still doesn’t make Fannie Mae Enron, and I for one would be livid if I were a Fannie Mae shareholder, and watched my money get flushed down the toilet for two days because, frankly, some reporter can neither read nor report. (I may well be a Fannie shareholder via indexed stock funds in one of my retirement accounts. But for disclosure purposes, I own no shares of FNM that I know about.)

Eavis could have gotten the same explanation from Fannie Mae that the analysts got in the call if he had asked for it, I am sure. No one forced him to write an article that makes accusations of willful dishonesty and implications of criminal behavior based on his inability to understand Table 26. He gave himself that assignment. And instead of apologizing for it, he continues to insist that the numbers don’t make sense.

Let me cut through the accounting archana (we can discuss that in the comments if we need to) to what I think is the real issue here. Fannie Mae has always had the option to repurchase seriously delinquent loans out of its MBS at par (100% of the unpaid principal balance) plus accrued interest to the payoff date. This returns principal to the investors, so they are made whole. If Fannie Mae can work with the servicer to cure these loans, they become performing loans in Fannie Mae’s portfolio. If they cannot be cured, they are foreclosed, and Fannie Mae shows the charge-off and foreclosure expense on its portfolio’s books (these are no longer on the MBS’s books, since the loan was bought out of the MBS pool).

Now, Fannie also sometimes has the obligation to buy loans out of an MBS pool. But we are—Fannie Mae made this clear both in the footnote to Table 26 of the Q and in the conference call—talking about optional repurchases. Why would Fannie Mae buy nonperforming loans it doesn’t have to buy? Because it has agreed to workout efforts on these loans, including but not necessarily limited to pursuing a modification. Under Fannie Mae MBS rules, worked out loans have to be removed from the pools (and the MBS has to receive par for them, even if their market value is much less than that).

There is, however, a little matter of accounting rules for booking these loans. Under GAAP, known internally to Fannie Mae as its SOP 03-3, the loan is taken onto the books at the lower of cost (par, in this case) or the fair market value of the loan at the time of repurchase. When the FV is lower than par, Fannie Mae has to charge-off the difference, at the time. This is not a true realized loss: it is a reflection of a mark to a real market value of a delinquent loan. You take the FV write-down at the time, even if you think that no loss, or a very much smaller loss, will actually end up occurring. That’s the rule. Anything else would be “mark to model” or “mark to myth.”

Now, anyone who hasn’t been living under a rock knows that bids for delinquent loans were either nonexistent or atrocious in July-September of this year. Certainly Fannie Mae knew that if it exercised its option to buy loans out of MBS in order to modify them, it, not the MBS, would have to take a nasty FV write-down. I don’t know about you, but I happen to think that a lot of private investors/servicers are refusing to do modifications of securitized loans precisely because they don’t want to have to buy them out of the pools and show that nasty write-down on their own books. They claim that it’s because securitization rules won’t let them modify loans, but I’ve never really bought that argument, nor have many regulators or the SEC.

The problem is that the market right now does not distinguish between a scratch & dent loan—one with a problem that could be cured with a modification—and defaulted nuclear waste that is facing 50% or more loss severity on imminent foreclosure. Whether it should be making that distinction or not—whether this is partly irrational panic or not—is not the point here. The point is that it just isn’t doing so, and so anyone who takes a loan to portfolio right now and uses a true market value instead of a fantasy is going to show the same huge write-down for the scratch & dent as for the nuclear waste. This will continue to be true until the market decides that everything isn’t nuclear waste any more.

So nobody wants outfits like Fannie Mae to mark to model; we want them to mark to market. We also want them to work out loans that can and should be worked out. Remember, we aren’t talking about horror subprime exploding ARMs here, we’re talking about troubled loans in typical Fannie Mae MBS. We’re talking about the kind of loans that would have taken a $60 million write-down in a past quarter, but that are taking a $600 million write-down in this quarter, solely because the market price of those loans has deteriorated so badly.

We also need to remember that Fannie has no intention of ever reselling these loans. If they can be cured, they will stay in Fannie’s portfolio. So establishing a market price is not about determining a loss Fannie would take if it cured the loan and then resold it. Establishing a market price is just a requirement of the accounting rules for a situation in which the price you must pay for a loan (par) is more than the market value of the loan. After all, what Fannie is doing here is making the MBS investor whole and taking the deteriorated asset onto its own books. The FV write-down means exactly that it is not hiding a loss this way.

Fannie Mae could avoid these write-downs by failing to exercise its option to buy the loans from the MBS. That would mean Fannie Mae refusing to work out loans with borrowers. Or, to put it another way, the price of Fannie agreeing to work with troubled borrowers is an out-sized hole in the current quarter’s charge-offs, not just because of the loan quality, but because of the total melt-down in the secondary market for nonperforming loans.

Fannie says that it does not expect most of these loans to default and produce large realized losses. You may or may not find that convincing. But Fannie does, because that’s why they bought them out of the pools. If they thought the stuff would go straight to the FC department, they’d have let the servicer take the loan out of the pool and foreclose, and let the losses hit the MBS guarantee fee income.

Fannie’s story is that in normal times, these FV write-downs of repurchased loans are included in total charge-offs, and therefore are reported in the credit loss ratio. In Q3 07, because of the enormous size of the FV write down, Fannie Mae backed out of the charge-offs the ones due to an up-front write down of delinquent but not defaulted loans. It added back any part of that write down that did, actually, become a realized loss, so that readers of the Q could get a true picture of how much the total charge-offs in Q3 were affected by repurchased loans.

Eavis is saying that this is inexplicable. Of course it’s explicable. You can and some analysts did on Friday ask Fannie Mae why it didn’t supply more information about what it was doing with those loans, and what its expected cure rate would be for these workouts. Fannie Mae acknowledged that such information would have helped and promised to provide more in future disclosures. But nobody on that conference call, as far as I remember, questioned the size of the Q3 07 FV adjustments. People who have been following the credit markets all year are not surprised here. Eavis is.

For the love of all that is holy, what does anyone think Fannie Mae (and Freddie Mac) are up to these days? The enormous pressure they are under by Congress and the public to modify as many loans as can possibly be saved has been so well-documented in the press that I’m sure they heard about it on Mars. It’s possible that Fannie is too optimistic about the cure rate of these loans. It’s possible that deep inside, they realize they are going to eat huge losses on all this stuff. But they were told in no uncertain terms to buy it out of the pools, take the FV write-down like a big kid, and start working out loans. Does anyone actually expect them to write a quarterly report that says “We think all this stuff will result in 100% losses in the next 90 days, but our regulators made us buy it anyway, so we’re reporting the worst possible credit loss ratio we can calculate, just to spite them”? On no planet, at no point in time, will that ever happen. You have to be willfully ignorant to think it would.

So there is, actually, a compelling story to be teased out of a couple of footnotes to a little table on page 55 of a 107-page quarterly report. It’s a story about political and market pressures and reactions; about the bottom-line impacts of workouts to investors like Fannie Mae; about the real-world effects on profitability numbers of things we see in fairly abstract forms, like those cliff-dives on the ABX and CDS charts or the dramatic ratings downgrades we post on regularly. There is a story that ties all of this together and shows how realized losses can be small compared to market losses, and how this ties into the debates over “mark to model” and other bad industry practices. There’s even possibly a story about how in certain unusual times, the old-fashioned GAAP rules fail to really tell investors what they need to know. These are fascinating and important subjects.

Eavis blew through all of those real issues to make a big deal about how Fannie projected losses in the 4-6 bps range for the year and might, actually, be at 7.5 bps as of Q3 if you calculate the number the way Eavis thinks you should. Think about this: he’s saying that it’s possible that credit losses on mortgage paper as of Q3 07 are worse than what was predicted at the end of 2006. OMG!!! No!! Really??? NEWS FLASH!! CALL THE POLICE!!!!! THE OBVIOUS!! IT IS BLINDING ME!!!

Sorry I let myself get out of control there, but come on. Anyone who has been reading USA Today on alternate weekends since February knows that losses are getting worse, and Fannie Mae did in fact explicitly report that losses were getting worse, even with the FV write-downs that were not realized losses backed out. What we are seeing is the whole problem with the “Another Enron” mentality: confusing the meaning of numbers with “accounting tricks,” and substituting some kind of gotcha for an honest attempt to understand the market mechanisms and economic reality that is creating those numbers. This mentality claims to be keeping companies honest, but it actually has the opposite effect, in my view, of inhibiting companies from presenting more detailed numbers, since the more you give people like Eavis the more they have to play gotcha with. And Eavis himself takes credit for his original article having lead to a 17% drop in Fannie Mae’s share price. I guess he’s proud of that. The other side of the Enron myth is the Famous Reporter Who Brought Down the Corporate Giant. It is worth not allowing oneself to get sucked into that sort of grandiose mythologizing.

I have to say I hate “blog triumphalism,” too. That’s the mindset too many internet writers have that us citizen journalists in our jammies are going to single-handedly bring down the Big Corrupt Media. I firmly believe in beating the press up a little when they do egregiously bad reporting, but that’s largely because I care about understanding what the real story is. And I hate being distracted by red herrings in my personal quest for understanding. Yesterday I spent over two hours rooting through SEC disclosures and listening to a 57-minute conference call trying to independently verify Eavis’s point; today I’ve spent a couple of hours writing this post. I am willing to believe that very few people have the time and the expertise to do what I just did. I therefore feel compelled to share my point of view with the rest of the world, in the interest of a worthwhile public discussion of financial and economic matters, which is the purpose of this blog. So I didn’t start out with the goal of catching Eavis being a lousy reporter; I started out with the goal of reading about Fannie Mae in a CNN Money article. But I believe that I did discover hyped, misleading, and ignorant reporting, and I believe it is fair to say so in public.