Thursday, June 19, 2008

In praise of fraudulent accounts

I wound up having a chat about Barclays with the head banking analyst from a major European broking house. He hit me with the “how is it possible” line about the assets being as miss-marked as I privately think. He talked about auditors and regulators and the like.

I thought that was shockingly naïve – because I am not even sure that miss marking is now the point.

Go back to my first substantive post on this blog. It was about scoping the US mortgage crisis. Seeking Alpha – who I originally sent the column to – changed the heading of this post to imply that the mortgage crisis is not as bad as it seems. (Did they read it?)

Anyway – in that post I note that it is quite common to be able to buy AAA strips of diversified non-standard (but not intentionally subprime) mortgage securitisations for 80c or less on the dollar.

This implies a shocking level of system losses.

In the old days a typical mortgage securitisation had 12 percent “protection” before you got to the AAA strip. For the AAA strip to default you needed something like 30 percent of the loans to default with a loss-given-default (or severity) of 40 percent. This was unthinkable – and so it was considered reasonable that the AAA strips covered 88 percent of the pool. I noted in the first post that for many Alt-A issuers I could not have ex-ante faulted that logic.

Well the unthinkable has happened. There are numerous pools where the defaults will be greater than 30 percent or the severity greater than 40 percent. But it is not obvious that the average non-GSE mortgage will look that bad.

However the market is trading quite a bit worse than that. It is very common to find the AAA strips trading at 80c in the dollar. To lose money on a strip trading at 80c in the dollar you need the losses to be fully 20 percent worse than the above 30 percent of 40 percent scenario.

If 50 percent of the pool defaults and you have a loss given default of 50 percent then you STILL make a profit buying the AAA strip at 80c in the dollar. [Ok – the technically minded will tell you that some of your loot could go to junior strips. But offsetting that you get a high yield in the early periods.]

The meaning of an 80c AAA strip is clear. The market is implying an absolute debacle – something beyond the scope of the really bearish (including me) to contemplate.

I do not think it will happen. I don’t know anyone else who does.

So why does the market keep pricing the AAA strips at 80c? Well I explain in the first post – but it really comes down to “you can’t borrow to buy this paper any more”. The deleveraging will stop when the assets are patently attractive to unlevered buyers – and they are not there yet.

Meanwhile the market seems to imply something that looks insane to me. The market is implying that more than 50 percent of diversified mortgages (not originally deliberately subprime – but not GSE suitable) will default at a severity greater than 50 percent.

So what is the implication of mark-to-market accounting?

Above I argue that the market is insane. It’s a dangerous thing to argue – but I argued the market was insane when subprime mortgages were trading at 101c in the dollar because they had high yields. And I will argue it now when securitisation paper is trading as if losses across the whole US will be greater than 50 percent of 50 percent.

It was insane (but technically correct) for investment banks to mark their book to market when the price of a subprime mortgage was 101. If you believed those accounts you would take on lots of risk and declare immediate profits. Anyone who managed their business as if that was a permanent state is now bankrupt.

It is similarly insane (but technically correct) for anyone subject to fair value accounting to mark a levered book to the current implied market default rate for mortgages.

The market was wrong then and it is wrong now. But fair value accounting required mark to insanity then and similarly requires it now.

If you do mark to the current insanity and you have a lot of mortgages your accounts will show you as breathtakingly insolvent.

So what do you do? Reclassify the assets as level three and mark them to model. It is technically fraud to do this when the assets can be priced – but no more wrong in my opinion than valuing the subprime mortgage at 101 in the first place.

The most obvious such example is Freddie Mac putting over 150 billion of its assets in the level three bucket. Almost all of those assets have a market. It is just that Freddie Mac (justifiably) doesn’t like the price and so marks-to-something-other-than-fair-value.

But it is not just Freddie Mac who is in this position. Everybody with a substantial list of stressed financial assets is. And everybody marks to model. And the models bear little resemblance to fair value as defined in accounting standards.

But if the model makes reasonable probability weighted estimates of the present value of what you will receive in cash on the asset then I can live with the mark-to-model (fraudulent as it is). In fact I will defend it – hence the title of this post.

Unfortunately there is a problem. Once you have decided to commit fraud (as I believe almost every financial institution has) then you might find yourself “in for a penny, in for a pound”. Having decided that you do not want to mark to “fair value” it is not far to decide that you want to mark to total myth. For instance you might (rationally) decide that it is not sensible to mark the mortgage to a 50 percent default and 50 percent severity. But that is no excuse to marking it to a 3 percent default and 15 percent severity. In for a penny…

The level three decision made by Freddie Mac is fraud. They can find a fair arms length value and they chose not to do so. But it is a reasonable fraud – and it is in the interests of Freddie Mac, its shareholders, probably its creditors and certainly its regulators. The accountants are happy to sign off on it too. I wish I knew what reality was and how far from reality Freddie’s models are. I don’t – and so I will not be buying Freddie Mac stock.

Freddie’s fraud is the same fraud committed by everyone who has a substantial amount of financial assets to mark to models.

So how to assess a financial institution?

Having decided that almost all financial institutions are fraudulent the question for a long or a short-seller is not is there fraud or not (as per my European banking analyst). The question is “how much fraud – and just how bad is the book really”.

This is a real problem. Whilst I do not think that the mortgages will have a worse than 50 of 50 outcome – I do believe that it will be worse than 30 of 40 in a wide range of circumstances.

Every one (reasonably) decided to ignore the illiquid markets. Here are a couple of places where they have degenerated into unreasonable myth:

  • Barclays manages to produce almost no losing trading days.
  • Royal Bank of Scotland has 10 billion dollars of second lien mortgages in the Midwest on which their provisions imply an almost zero loss rate and where the secondary market is way less than 40c in the dollar and where the house prices have fallen to zero.
  • Barclays had some private equity loans that they originally intended to originate-and-sell. They are stuck with them. They are stretched. The model prices remain in the high 90s where the companies are cash-stretched right now.

I could go on. These are by no means the atypical myths.

  • I crossed the bridge ages ago when I decided that financial institutions mark to things that don’t look like reality at all. (My UK banking analyst friend has not got there yet. That seems to me to be a willing suspension of disbelief.)
  • I crossed the bridge this year when I decided that the fraud was OK. I know my morals have slipped. But I will say it again. Some fraud is fair and reasonable.
  • I have also decided that you need to grade the fraud from “white lie” to “mark to hope” to “mark a myth so far from obvious reality it is comic”.

David Einhorn in his latest book demonstrates Allied Capital does a lot of marking to a myth so far from reality it is comic. Reading his book is great fun (though living it would have been somewhat rougher).

Unfortunately I think that mark to comic myth is becoming the norm for UK banks as well. The main problem is that the UK banks are much more levered than Allied Capital. Mark-to-myth at 40-50 times leverage is a recipe for tragedy.


Post note: In this post I say a few things that are dangerous - and which I do not believe except within the narrow context of other financial institutions (not me). For instance I say "some fraud is fair and reasoable". I wish never to be quoted out of context.

1 comment:

  1. John,

    Do a post of the top five reasons why people aren't buying this stuff. I don't fully believe the we-can't-get-financing argument, or at least I question how important it is here. What are the actual yields at 80 cents on the dollar? The i-banks are financing sales of this stuff, too.

    I think there is also fear about fraud, chaos (the servicers can't compile the true delinquency data) and political/legal fears connected to loan-mods etc.

    What think you?

    ReplyDelete