Thursday, October 04, 2007

This is madness!

Spreads on everything except sub-prime stuff has been racing in. And its got me a little nervous that a true mania in credit might develop.

First the bad news: The ABX 2007-1 BBB- keeps hitting new lows...
Hard to hold out a lot of hope for Baa-rated sub-prime pools. I saw a pool offered by Goldman Sachs at a dollar price of $10 yesterday. It was originally rated Baa, now C. It was a 2006 vintage, so the loans are probably less than 2 years old. Just for fun I ran it through some modeling and found that if the current delinquency rate (15%) turns into a default rate, and the recovery is around 50% (generous), even at a $10 principal price you'd still wind up losing money. Bear in mind since the loans were probably less than 2 years old, there probably haven't been any resets yet! I didn't spend a lot of time on this particular piece so this is hardly complete analysis, but the point here is that there is a lot of shit out there that ain't coming back.

Meanwhile, asset-backed CP rates are improving dramatically. The following compares ABCP vs. 3-month LIBOR:

The ABCP rate is from the Fed's H15 release, and its only been tracked since 2006. But you can see that there was virtually no volatility in this rate until suddenly there was. This graph combined with the next graph on amount outstanding in ABCP:

ABCP outstanding looks a bit Beggar's Canyon back home eh? If you think about these two graphs together, my belief is that most of the bad actors in this space have been denied access to the CP market. So spreads are tighter on ABCP because the only issuers left are the stronger issuers. I'm hearing its mostly corporate guaranteed paper, i.e., a strong bank who has put a corporate guarantee on the SIV or whatever conduit has issued the CP. I think its important to realize that the concept of ABCP isn't a bad one. Hell, I think a lot of investors would prefer to have their CP backed directly by quality assets as opposed to being subject to corporate shenanigans. Currently the definition of "quality" has come into question, and the ABCP market is suffering for it, but in the long run, the ABCP market should be fine. That being said, I'm a little surprised its improved this much this fast. Its getting me just a little nervous

High yield corporate spreads have moved much tighter as well:

That's two years worth of data. Obviously this shows the big downward trend in spreads. I've said that I was constructive on high-yield, and I am still. But there is potential here for a mania phase to develop. Let me draw out a scenario...

  1. High Yield spreads "correct" from unusually tight spreads in response to problems with sub-prime mortgage loans. (already happened)
  2. HY spreads rapidly tighten back, not quite all the way to the previous tights, but maybe half way.
  3. People start justifying the move by claiming that the HY market never had anything to do with sub-prime anyway. So, they'll claim, the widening of HY spreads was not fundamentally justified.
  4. HY spreads keep tightening, blowing through all-time tight levels. This encourages more borrowers to come to market. LBO and covenant light deals re-emerge.
  5. This causes the same problem we had in sub-prime: too many deals. Marginal borrowers are always the weakest borrowers, be it consumer or commercial loans.
  6. Default levels rise significantly, and HY spreads gap out tremendously. See 2001.
According to Moody's, 32.7% of speculative-grade bonds default over a 10-year period. And the average corporate bond realizes 36% recovery after default. So a diversified portfolio of high-yield bonds has an expected loss rate of about 21% over 10-years. So to use simple math, if high-yield spreads fall to 210bps over Treasuries, the expected value of a high-yield portfolio is no better than a portfolio of Treasuries. Since obviously an investor in high-yield should expect some risk premium beyond just expected loss rates, there is some point above 210bps where value becomes very poor for high-yield.

So at 400+bps, there is value to high-yield. But be cautious. If a mania phase develops, don't fall for it. I'll be happy to unload my high-yield position at 300bps or so and watch the mania from the sidelines.


Anonymous said...

I'm a bit of a novice, but maybe this is confusing for some others as well - talking about the sub-prime pool offered by Goldman at $10, you said if you model a delinquency rate of 15% and 50% loss severity, you lose money? Is that for a pool that you buy at 10% of face value (is that what $10 on a dollar basis means?)

Are you also using a loss curve? or saying the ultimate cumulative loss will be 15%? I guess where I am confused is it seems to be like, even with no interest - if I pay 10% of face value, and even if 15% of the loans are a total loss, I should make decent money? I must be misunderstanding part of your analysts...

Accrued Interest said...

It was 15% default for the whole pool, but the tranche I was looking at isn't the whole pool. Now I don't remember exactly how big the tanche was and I don't have the CUSIP written down anywhere. But say the tranche in question was only 5% of the original deal, and it is the most junior piece.

All cash flow will go to pay off the senior tranches first. Only if there is cash flow left over will the junior class get paid. So the reason why my $10 tranche loses money is because there isn't quite enough to pay off all the senior tranches, leaving nothing for the junior tranche.

See this for more on how subordination works in CDO/ABS deals.

Anonymous said...

There should be a notation following " Spreads gap out tremendously. See 2001." The notation should read "possible buying opportunity".

Accrued Interest said...

Yeah I agree that there would be a huge buying opportunity should this play out. Basically I'm thinking that it will overshoot on both sides.

Anonymous said...

I'm a bit of a novice as well. You said.."and the recovery is around 50% (generous),".

I guess you mean 50% for the whole pool like you did in your 15% default? If so, are you saying that the underlying houses are only worth half of their loan value?

Thanks again for the blog. It's very educational.

Accrued Interest said...

I do mean 50% for the whole pool. And that doesn't mean the house is only worth 50%, it means that the sale price less fees, accrued interest, and payout to senior lien holders, there is 50% left. If the pool has HELOC's, you are probably a junior claimant.

Accrued Interest said...

I think you'd normally assume that the sale price of a foreclosure occurs at a discount to its "real" value, because the bank normally looks to make a quick sale.

Anonymous said...

John Maudlin had a great article on this back in August.

He said..."And it is not as if it should be a total surprise. Any investor can go to their Bloomberg and pull up a listing of subprime Residential Mortgage Backed Securities. There are 2,512 of them. If you sort by the ones with the most loans over 60 days past due, you find that the average RMBS has 12.39% of their mortgages over 60 days, and 2.39% have already been repossessed (REO in the next table), with almost 5% in foreclosure.

The table below shows the RMBS with the highest level of 60 day past due (or worse) mortgages in them. Yes, the worst two offenders are the 2006 vintage of RMBS. But notice that a lot are from 2000, 2001, 2003 and earlier, well before the supposedly lax standards of the past few years. The third listed RMBS, the INHEL 2001-B is selling at 18 cents on the dollar (you can't see this from the table), and has been dropping since 2003. Over 25% of the mortgages in that portfolio have already been repossessed or are in foreclosure, with another 25% past due for over 60 days. Can you say ugly?

But you can also find paper from 2001 that is not doing badly. It should be clear to anybody who did a little due diligence a few years ago that there were problems in the subprime RMBS markets. There was a great deal of difference in the quality of various offerings. So it paid you to do some homework. If you could not get transparency, then you were taking a gamble.""

Hope this helps.

Anonymous said...

off topic but wanted to share an idea i hadn't seen before with the author of this blog

Unknown said...

How does a 32.7% default rate with a 36% recovery rate turn into a 24% loss ( where .327 * .64 = .209)?

Accrued Interest said...

Paul: umm...

::making an edit::

I have no idea what you're talking about.

Accrued Interest said...

Dave M.: I've had some fun with that function as well. You'll find pools where like 88% is either delinquent or in foreclosure. Mmmm... maybe, just maybe there is something going on with that originator. Just a hunch.

Anon: Good article, interesting point. Here is my take.

Obviously a senior tranche of a CDO needs a economic disaster to default. I might argue that the same holds true for high quality corporate bonds. And the authors' point about writing out-of-the-money puts is fair as well. But I think it would also hold for high-quality corporate bond spreads.

Here is a problem I often have with academic work on where a bond "should" be priced. The authors almost always assume investors can do whatever they want. In reality, the bond market is very segmented. Buyers of one type of bond might not have the expertise or legal ability to buy another security.

AAA CDOs were bought mostly by SIVs set up by banks and insurance companies. In general, the SIV was set up to arbitrage the gap between their funding level and the CDO yield. Something like the stock market crashing would really be of no moment as long as their funding level stayed the same. And honestly, even though the funding level has risen, I'd say the flaw in the SIV game was more about misunderstanding the assets and less about assuming funding would never rise.

Anonymous said...

"The table below shows the RMBS with the highest level of 60 day past due (or worse) mortgages in them. Yes, the worst two offenders are the 2006 vintage of RMBS. But notice that a lot are from 2000, 2001, 2003 and earlier, well before the supposedly lax standards of the past few years"

Wouldn't you expect the earlier originations to have higher cumulative defaults since they've been around longer?

Accrued Interest said...


That's kind of the point. Normally we expect few defaults in a new vinatge, because under normal credit conditions, you'd expect most borrowers to make their first years worth of payments. But there were so many fraudulent loans made that many are going bust almost immediately.

Anonymous said...

ttdg-thanks for the quick edit; i was too embarassed to ask why I didn't get the same earlier figure as yours. :-)