Thursday, January 15, 2009

2009 Forecast: Agencies and MBS -- Back door huh?

For the rest of the 2009 Forecast series, click here.

I lump agency debt and MBS together because both are basically wards of the state at this point. They are also the best example of Quantitative Easing working that we have so far.

5-year bullet GSE debt (that is, non-callable) has declined in spread from about +150 in October and November to about +80bps. It will probably get down to about +50 before stalling there. Historically agency debt traded between +20 and +40bps, but I'd say that in a lower liquidity environment, agency paper won't get that tight.

80bps ain't nothin' these days. With the 5-year Treasury now below 1.40%, one could argue that Agencies are generating 57% more income than comparable Treasury bonds. That's going to continue to attract real money buyers looking for something that's both safe and liquid.

MBS suffer from a severe negative convexity problem. MBS investors have essentially sold short an interest rate option to the underlying mortgage borrowers. Those borrowers are now almost universally in the money. Many borrowers will have difficulty actually refinancing (more on that below) but regardless, the price for MBS securities will have difficulty rising above $104 or so with an embedded in-the-money option with a $100 strike.

To see what I mean, notice the price spread across the coupon stack (using Fannie Mae 30-year MBS for February settlement):

  • 4.5% Coupon: $101.938
  • 5%: $102.750
  • 5.5%: $103.203
  • 6%: $103.656
  • 6.5%: $104.500

This is the current dollar price for a mortgage security with the indicated coupon. Typically the underlying mortgages have a rate 0.5% higher than the coupon rate. Notice, for example, that the 6% bond at $103.656 is less than one point higher than the 5% at $102.75. To me that's telling you that if rates were to fall by 100bps from here (implying the 5-year Treasury is 0.40%), the 5% mortgage would only improve in price by 1 point.

All this is to say that those hoping for price appreciation out of MBS should look elsewhere. But those looking to just collect fat income may have found a home.

Say you buy a generic 6% Fannie Mae MBS at $103.656. According to Bloomberg, that bond will pay at approximately 51 CPR (if you aren't a bond person, just assume that means 51% of the loan will pay down each year, that's close enough). That produces a yield of 3.09% with an average life of 1.5 years.

That isn't half bad considering that Treasury bonds in 1.5 years are yielding about 0.40%. You have some reinvestment risk as the mortgage pays down, but by 1.5 years rates may be rising again, and so maybe that's a good time to be reinvesting anyway.

If you are willing to do a little more work, you can do much better. Say you can find a pool with mostly 2006 borrowers. Most of those borrowers have experienced negative home price appreciation to some degree (not as bad as you might think because we're only talking about conforming loans here, i.e., loans under $417,000). Anyway, couple this with a pool full of borrowers with 90+ LTV? Or mostly coastal geographics? Or relatively low credit scores? You might have a mortgage pool that will repay much slower than average.

Say you can pick a pool that pays at 40 CPR instead of 51 CPR. That increases your yield to 3.82%. 30 CPR? Now its 4.39% wih a 2.8 year average life. That compares very favorably with the 3-year Treasury trading at 1%!

Finding these kinds of pools is somewhat easier in hybrid-ARMs, which I love here on a pure relative value trade. But beware, the liquidity is much weaker in this sector. Otherwise I'd be a bigger overweight in hybrids than I already am. You can also get a lot of good high LTV loans in the GNMA space.

So if you are going to just buy and hold and don't care about keeping up with Treasuries in a rally, MBS are probably as good an investment as any. But I'm underweight MBS, focused entirely on the kinds of specialized pools that I described above. I've basically dumped all my 5.5% and lower bonds.

So to make a prediction about the sector, I'd say the the Fed's continued support will keep dollar prices relatively high, even if Treasury rates back off a bit. But overall I expect MBS spreads will perform extremely poorly should rates fall from here, and that has me cautious on the sector. I'd add heavily to MBS if the 5-year broke 1% or the 10-year broke 1.6%.

Callable agencies are kind of the worst of all worlds. There is no value-adding opportunities through pool selection akin to what can be done with MBS. And you have the same negative convexity problems, where if rates fall you get called away and if rates rise you get crushed. So I'd avoid most callable agency issues.

8 comments:

Ian said...

Nice analysis. In addition to the pool attributes you mentioned, I am also hanging out in low loan balance pre 2003 vintage cohorts. I figure that if a person in the 2000-2002 cohorts didn't refi during the 2003 refi wave, they are a burnout candidate and less likely to prepay going forward. Plus with cashout refis drying up, the accumulated equity is not an issue. Also I like 60 wala 15-yr paper, less option cost and convexity risk to hedge.

Agency CMO's are also extremely cheap compared to underlying collateral. I'd be a big fan of inverse IO's and seller of PO's here. Market is pricing in speeds that are way too fast. Even if speeds are fast, for many inverse IO's, the yields are still extremely fat according to the B-berg YT screen. Just my $.02 worth.

alex said...

There was an article today that Pimco has made a significant reduction in their MBS holdings.

I wonder if your explanation is the reason.

jswede said...

"If you are willing to do a little more work, you can do much better. Say you can find a pool with mostly 2006 borrowers. Most of those borrowers have experienced negative home price appreciation to some degree (not as bad as you might think because we're only talking about conforming loans here, i.e., loans under $417,000). Anyway, couple this with a pool full of borrowers with 90+ LTV? Or mostly coastal geographics? Or relatively low credit scores? You might have a mortgage pool that will repay much slower than average."

Slower, unless there is mass default and FNMA pays you back par, a lot quicker that you were ready for....

(assuming FNMA can -- they are supposed to anyway)

gingersue said...

What effect does this have on the stocks like Annaly, NLY, Agency,AGNC, and Hatteras, HTS? All seem to be in a sweet spot with low capital costs and owning all agency paper. Annaly is levered 7 times, last I checked.

jswede said...

another thought AGAINST these supposed coming pre-pays in MBS from re-fi's are that while the GSE loans look pretty good as far as LTV, let's not assume that that's the borrower's ONLY loan. He/she likely took a conforming loan and piggy-backed a HELOC so as to steer clear of PMI, and take advantage of GSE type interest rates.

So a $625k property bought with 20% down has a ~$415k conforming loan and a $85-100k HELOC-type 2nd loan. Even a conservative 5% drop in home value, and that 2nd loan should certainly keep him from re-fi'ing the first GSE wrapped loan...

right?

A lot of people did this btw -- it was 2 loans and 2 pay-days for the mortgage broker. This was standard from 2005 on, if not before.

Further, you had a ton of "cash-out" HELOCs that squeezed out the equity -- these are separate and not accounted for in the GSE loans' statistics, putting the total borrowers' LTV much higher than advertised in the DES pages of the Agency MBS security... and, ahem, a few folks did that also...

thought?

Accrued Interest said...

Ian:

IO's and PO's aren't for the feint of heart!

JSwede:
I've actually bought more high OCS, high LTV loans. My theory is that those loans are good borrowers with bad timing. Probably going to keep paying his bills.

Also you are right about other debts, but when looking at loan characteristics, you can't see that stuff. So I look for indicators that a borrower might have maxed out. Like if you see a pool full of EXACTLY 80 LTV puchase loans, the odds of a silent second are high.

Ginger:
I don't like NLY or HTS because they are still borrowing short-term. You don't know when that short-term funding is going to dry up on them.

"Mindles H. Dreck" said...

Isn't there still a possibility that they will waive re-appraisals (a la FHA loans), and your high LTV loans will be swept up in the intended refinancing boom? After all, the exposure doesn't change and the policymakers want the rate down.

Accrued Interest said...

Mindless:

That's a pretty big risk, for sure. I personally think that is a poor use of potential stimulus money. I'd rather see government cash focused on increasing the purchase of homes rather than refinancings.