Fannie and Freddie roiled the MBS market last week by announcing that they would buy “substantially all” (in Freddie's language) of the loans that are 120+ days delinquent out of MBS pools.  Freddie’s announcement indicated a one-time burst of prepayments that will hit the March prepayment tape (i.e., holders of record at the end of Feb).  Fannie’s was more ambiguous, but the effect will start in April (one month after Freddie) and last for “a few month[s]”.  Needless to say, the high coupon passthrough market has been all over the place (below).

Before today, neither GSE had a strict policy on the matter of when loans were bought out of pools, relying instead on ambiguous statements about the cost of guarantee payments vs. holding nonperforming loans in portfolio.  And in fact, neither GSE committed itself to continuing this policy over the longer haul.

Why Now?

Officially, both GSEs press releases cited adoption of SFAS 166 and 167.  And to be fair, these do affect the balance sheet treatment of troubled loans.  But who’s paying attention to the GSEs' balance sheets right now?  The reality is that the economics have long favored more prompt buyout of bad loans:  Simply put, Fannie and Freddie get to borrow 2-year money at approx 1% to call loans at par from pools with 5-7% or higher net coupons.  (Hint:  You don’t need an OAS model for this calculation.)  The real enabling factor was the administration’s year-end move to loosen the portfolio caps and provide more capital.

Short-Term Prepayment Impact

The table below (Exhibit 1) summarizes the delinquency stats released by the GSEs for major 30-year fixed rate cohorts, and turns those into estimates of the near-term prepayment impact.  A few points:  (1) As noted, Freddie’s hits the Mar prepayment tape; Fannie will spread out over “a few” months beginning Apr.  (2) The timing difference has been punishing for March Fannie high coupons; the ambiguity of Fannie’s plans will extend that pain into Apr-May TBAs.  (3) Fannie’s delinquency backlog relative to Freddie was even worse than we expected, with like-cohorts in 6.0s and above running as much as 79% higher in 120+.

Longer-Term Prepayment Impact

In the near term, we’re looking at a short-term prepayment burst.  After that – and assuming that the GSEs continue to hold to the “120 and out” policy – we think the run rate of speeds will be faster than before this incident.  For one thing, if the GSEs had been keeping up there wouldn’t be a backlog; this implies that the last year’s speeds have been understating the true run rate of involuntary prepayments.  (The good thing about “120 and out” is that involuntary prepayments naturally follow the 90-to-120 roll rate, which will be more organic and predictable over time.)

Also, the delinquency problem isn’t getting better by itself.  There are today’s D60s and D90s to prepay.  And if we extrapolate from non-agency Alt-A and prime, current-to-delinquent roll rates have deteriorated in recent months, with evidence that badly underwater borrowers may be getting more aggressive at exercising the Strategic Default option.  In other words, this is only the first round of a process that will play out for 2-3 years.

Fannie vs Freddie

We’ve felt for some time that the market has been underestimating the prepayment speed effects of the extent to which Fannie’s credit book is badly underperforming Freddie's.  As we noted on Monday, Fannie’s serious delinquencies have been 40% worse than Freddies for a couple of years.  If we continue to see bad credit performance from 2006-2008 premiums, we believe Fannie / Freddie speed differentials will be significantly higher going forward.  And even when we compare loans on a similar-credit variable basis (Exhibit 4 below, from the most recent 10-Qs), Fannie loans are performing worse than Freddies.

Strategy

Some initial thoughts:  (1) Even after this incident, we think it will continue to make sense to cull higher coupon holdings backed by weak-credit pools, both in passthroughs and CMOs.  (2) We believe the market may still be underestimating the future Fannie / Gold prepayment differential, creating opportunities in passthroughs, CMOs and Trust IO/PO (consider a Gold IO + Fannie PO combination).  (3) 10/20 fixed rate IOs will be a mess (from fast CPRs); sell.  (4) Short-reset hybrid ARMs will also be a mess; there may be some opportunities in “over-reaction” selling, but, as with fixed rates, be prepared for faster post-buyout prepayments.  (5) Ginnie credits are still much worse than either Fannies or Freddies, so it’s tough to have Ginnie 5.5s and above make sense at these dollar prices.  (6) Seasoning looks better than ever – 2005 and earlier origination, be careful with loans securitized recently, watch credit quality.

Fannie vs Freddie Credit Books

We’ve felt for some time that the market has been underestimating the potential prepayment speed effects of the extent to which Fannie’s credit book is badly underperforming Freddies.  As we noted previously, Fannie’s serious delinquencies have consistently been 40% worse than Freddies for a couple of years.

To the extent that we continue to see bad credit performance from 2006-2008 premiums – and the GSEs actually buy the loans out on a timely basis – we believe Fannie / Freddie speed differentials will be significantly higher going forward.

Even when we compare loans on a similar-credit variable basis (above, from the most recent 10-Qs), Fannie loans are performing consistently worse than Freddies.

As noted earlier, we believe the market may still be underestimating the future Fannie / Gold prepayment differential, creating opportunities in passthroughs, CMOs and Trust IO/PO (consider a Gold IO + Fannie PO combination).  Also, 10/20 fixed rate IOs are likely to be a mess (from fast speeds) for some time, as are short-reset hybrid ARMs.

Gold-Fannie Swap and Roll Analysis for Premium Coupons