Tuesday, July 22, 2008

Agency MBS: You will be tempted by the yieldy side of the Force

Although agency mortgage-backed securities (MBS) have been beaten up the last couple days, avoid the temptation to jump in. Agency MBS spreads are not as attractive as they seem, and the technicals for MBS are horrible.

This has nothing to do with the financial condition of Fannie Mae or Freddie Mac. We'll have to see how the government bailout progresses. Perhaps just the act of allowing the GSEs access to the discount window will be enough to ensure liquidity. But by all indications, protecting the mortgage securitization market (i.e., keeping mortgage borrowing rates low) is a primary goal of any government action. It isn't credit quality which is behind this underweight call. Rather its plain old fashioned market conditions.

MBS analysis is more complex than for other investment-grade bonds, in that the yield on the security is highly depended on the pace of principal repayments. These payments primarily come from two sources: refinancings and housing turnover. Historically, refinancings were the primary driver of changes in mortgage payment speeds. Anytime interest rates would fall, borrowers would rush to refinance and thus pay off their old mortgage. Housing turnover was more consistent, as people tended to move from house to house based on life circumstances as opposed to macroeconomic events.

But times are anything but typical. Various conditions are coming together which will keep homeowners in their current residence far longer than historic norms. There is a large number of homeowners currently underwater on their mortgage, and an even larger number with less than 20% equity. Given that getting a mortgage with less than 20% down payment is difficult and very expensive right now, homeowners who currently have less than 20% equity would have to come up with a lot of cash in order to move to another home.

So the housing turnover element of mortgage principal payments is set to plummet. In addition, the same factors will prevent many refinancings. A borrower underwater on his current mortgage will not be able to refinance his loan just because rates fall 50bps.

This means that the average life of a mortgage is longer than is currently being assumed.

For example, a Fannie Mae 30-year 6% mortgage security currently has a nominal yield of 6.19% and an average life of 5 years. The average life is the median of a Bloomberg survey on prepayment estimates. That calculates to a nominal yield spread of 271bps.

Note that a 6% mortgage security is typically made up of borrowers with a 6.5% mortgage. Currently mortgage borrowing rates are 6.26%, according to Freddie Mac. Under normal conditions, one would assume that a 6.5% borrower is relatively close to a refinancing opportunity. Hence Wall Street prepayment models are assuming that this mortgage will pay principal slightly faster than this time last year.

More likely is that mortgages will prepay at historically slow rates. Cutting Wall Street's estimated prepayments in half, the mortgage's average life goes from 5 years to 9 years. Because the yield curve is so steep, that results in the yield spread falling to 219bps. If you cut Wall Street's estimate by a third, the spread falls to 202bps.

As investors come to terms with the extending average lives, prices are likely to fall rather than yield spreads contract. Holding the 271bps yield spread constant but extending the average life to 9 years causes the price to drop by over 3%.

Technicals for MBS remain ugly as well. Regional banks and credit unions were classically large buyers of agency MBS. But given the capital situation at banks, we are far more likely to see banks as net sellers of MBS over the next year. In addition, Fannie Mae and Freddie Mac will continue to dominate overall mortgage issuance, and both will be under political pressure to expand their guarantee business. This means more supply of agency MBS.

The best plays in MBS are securities where extension risk is limited. That's 15-year mortgages and hybrid-ARM securities. Both have a natural limit to how much interest rate risk can increase, given the shorter maturity/reset.


Gringcorp said...

S'interesting. Before the current crunch my most intimate knowledge of structured finance came from a bit of time following the European corporate CLO and whole business securitization market and Liar's Poker. Michael Lewis went on at length about prepayment rates being the most difficult thing for traders to get their heads round, but it's really taken a back seat to credit quality in the current market.

And did you notice the FSA review from Moody's? There's a bit of a mention about the investment and insurance portfolio, but also "material shifts in the demand function for financial guarantees, and the potential sensitivity of the company's franchise".

Which is their way of saying that the entire monoline business might not make sense after all.

I've spent much too long trying to say that the monolines' troubles have everything to do with a few lapses in underwriting discipline, and that their business has a sensible rationale. That rationale, however, presupposes the ratings agencies being both sensible and credible. They're neither right now.

nades said...

The best plays in MBS are securities where extension risk is limited.

What about interest only strips? Have they increased in value. If people are going to be stuck with the loans a lot longer it makes me think that IO faucet wont turn off as quickly as the models might predict....

Anonymous said...

In MBS land there is concept called OAS. Option adjusted spread. you would need to find out what it is. OAS is solved using monte-carlo simulations on term-structure of interest rates. So, you would also need a good prepayment model to change as term structure changes. Finding what the OAS will get you an idea of how high the option cost is? option cost to refinance/relocate, etc.
In your example, if 270bps is the spread. Your OAS could be 210bps. So, option cost is 60bps.

Accrued Interest said...

OAS! What?!?!? Never heard of it!

Actually, I've spent 10-years in the bond market, so I'm pretty familiar with OAS. And I could write a book on how much I hate OAS as a measure of MBS value.

My argument in the post would be even stronger if I made it on an OAS basis. Because extension would cause the option value to de facto decline, but since the OAS model is strictly interest rate based, its not going to understand that at all.

Accrued Interest said...

As for the IO, that sounds right to me. I don't follow IO trading very much, so I can't speak to what's priced in vs. what isn't.

Keith said...

One should of course understand with Mortgage backed IOs that if everyone prepayed tomorrow you would get absolutely no money back at all. Not likely to happen, But the risk of an unexpected refi wave can have major consequences.

For example: if the bank behind the IO pool were able to turn a bunch of the loans into these bailout FHA loans, presumably that would take them out of the pool and you will get nil from them.

I have heard plenty of buzz in mortgage modeling circles that prepay models are running too fast right now. Some may already be dialing them down and pricing in slower speeds.

Keith said...

P.S. I would personally love to read a rant about OAS pricing. I always thought it was pretty arbitrary myself, and dealer OASs are totally all over the map even on liquid products like TBAs.

Accrued Interest said...

Its been in the back of my mind to write a post about OAS for a long time.

My major problem is with the vol assumption. What should vol be? Who knows!?!?

My other problem is that the MBS prepayment decision is based on a number of economic factors. Not just interest rates. Right now is an extreme example, as home owners are more likely to just wait out this period rather than move to another house. But in most periods, there are non-interest rate events which play a very big role in home owners decisions.