Fannie’s Monday announcement giving details about the impending “catch-up” buyouts of delinquent loans may have been intended to calm the passthrough market.  It didn’t.  The disclosure, which projected the volume capacity and intended priority of the buyouts, did go a long way towards helping the market value current and future months pricing for premiums.  But we think significant price discrepancies still exist in front-month rolls and swaps; we also think Gold-Fannie swaps don't at all reflect longer-term differences in credit quality and involuntary prepayments.

Revised 30-Year Fannie Prepayment Projections

Below we give revised prepayment projections based on the new information from Fannie about how they're going to implement the buyouts.  In effect, Fannie has said that it will (1) do 150-200,000 loans in the first month; (2) start with the highest coupons; and (3) make fixed rate “CL” and “CI” pools a higher priority.  Unfortunately, the disclosure left a lot of blanks.  For example, Fannie failed to clarify whether that 150-200K volume would hold for future months; Fannie also didn’t specify whether coupon or loan type would be more important in their prioritization.

We estimate that approx 630K loans need to be bought out by Fannie.  Using 175K as the volume constraint (midpoint of 150-200K), it implies that this process will drag on through the July prepayment tape.  In drawing our estimates we devised two scenarios based on assumed “Efficiency Ratios" in pulling the loans out (described below).  The idea is that Fannie won’t be able to pull out 100% of their target loans on the first attempt; this implies a tail of higher coupon buyouts that will stray to the back months, and somewhat higher mid-coupon buyouts that get pulled to the front months.  We see major 6.5 cohorts hitting 77-93 CPR in the April tape; 6.0s should peak in May, at 52-75 CPR; 5.5s should peak in June at 40-66 CPR.  For nearly every coupon and sector, 2007 origination – the bottom of the credit cycle – should deliver the fastest peak speeds.

Bear in mind that a key factor in Gold/Fannie swap valuations was not included in this analysis, and does not appear to have been taken into account by the market:  adverse selection in TBA pool delivery.  Our valuations are all based on average performance.  But TBA has never been an average market – it’s a cheapest-to-deliver market.  We believe strongly that investors withpools and deliver worse-than-average credits into the TBA bid, wherever the market’s pricing average credit experience.  Also, as discussed below, today’s swaps should include an assumption about post-buyout prepayments and Gold-Fannie pricing.  We think the market hasn’t focused on this aspect of the problem yet, which makes Golds a very strong longer-term risk / reward proposition.

The Post-Buyout Outlook for Prepayments

Three observations stand out to us in this regard:  (1) Involuntary prepayments aren’t finished when the “catch-up” buyouts are done.  Credit’s going to continue to drive 2005-2008 origination high coupon speeds for the next 2-3 years.  (2) The new “run rate” of CPRs should be faster than we’ve experienced in the past year.  Simply put, if Fannie and Freddie had been buying enough loans out all along, they wouldn’t have built a backlog.  (3) Fannies should be significantly faster than Golds.  Fannie’s credit book continues to perform worse than Freddie’s, with no sign of letup.

With respect to the post-catch-up buyout run rate, we think Freddie’s Jan. 2010 release was helpful in bracketing the new base-case CPRs.  120+ day delinquencies increased by 0.2-0.7% for major 30-year fixed rate cohorts month-month.  To the extent that this growth in D120+ represents a typical “under-buying” of serious delinquencies, it translates into 3-7 CPR higher speeds for Freddie post-catch-up-buyout.  Fannies could easily be worse.

Regarding Fannies vs Golds (Exhibits 6-10), we compare the credit performance of the two single-family mortgage books.  Fannie’s serious delinquency rate has consistently been approx 40% higher than Freddie’s.  Fannie’s book from the bubble years has objectively worse credit characteristics, with noticeably higher exposure to loans with original LTVs >90% and moderately worse FICO distributions.  And as we’ve discussed for several months, even when we line up the Fannie and Freddie books on a “similar-characteristic” basis, Fannie almost always runs significantly higher delinquencies.  We should note that in the 2009 vintage, originated in a stringent credit environment, Fannie and Freddie are similar in both the observable credit characteristics and early delinquencies.  This implies similar performance in both defaults and traditional rate/term refi prepayments for new lower coupons.

We don’t think the market is being realistic in evaluating the differences in the credit quality of the two books and how these differences will affect prepayments and carry in higher coupon Golds and Fannies over the long term.  The table below takes a simple approach to valuing the swaps “post-catch-up-buyout” (i.e., in June or July).  Unless we see a miraculous turnaround in credit performance, Fannies will almost surely experience higher default rates – and therefore faster prepayments on 2005-2008 30-year 5.5s and above.  Even the most conservative approach to valuing the swaps – 10% faster Fannie prepayments, valued on an even-spread/N basis – results in swaps of +0-15, +0-12 and +0-07 for Gold-Fannie 6.5s, 6.0s and 5.5s, respectively.  Referring back to Exhibit 2, our estimates of “fair value” swaps in Mar assumed that the swaps revert to 8, 6 and 4 ticks, roughly half of the most conservative swap values.  And even with an ultra-conservative set of assumptions, Gold-Fannie swaps currently look cheap to our fair value estimates.  We strongly recommend emphasizing Golds at these price levels.

High Coupons after the Flood

The optimistic note for higher coupons (5.5s and above) is that the new policy – buying loans out automatically as they roll to 120 days delinquent – creates a higher-quality cash flow once the “catch-up” is complete.  Frankly, the “roll-to-120” policy removes a key source of uncertainty in the sector:  risk of dislocations from bureaucratic decisions.

A “V-shaped” recovery in home prices in the next 2-3 years is highly unlikely.  This means that prepayments on pre-2009 5.5s and above will continue to be driven primarily by defaults.  We expect Golds to settle in around 25-30 CPR, and Fannies in the low-mid 30s.  At current Gold 6.0 prices, a 30 CPR gives a 3.10-3.20% yield, with a spread of N+160s / Z+110s.  Even at 40 CPR, 6.0s yield > 2.00%, >100/N and >50/Z.  Given that there should be very little interest rate sensitivity to prepayments +/- 100-150 bps, and therefore limited negative convexity, these are attractive cash flows.

The main point of caution in this sector is that the market should eventually trade higher coupons to longer effective durations.  These are 2.5-3 year average life bonds whose cash flows have very limited interest rate sensitivity.  The market’s been trading the sector to near-zero effective durations.  These bonds should trade longer, and respond primarily to the 2-3-year part of the yield curve.

In terms of specified pools, we believe that investors should be cautious in embracing traditional spec pool strategies in 2005-2008 production.  Attributes such as low balance don’t protect against defaults, and “100% investor” is not what you want to see in 2007 Florida loans right now.  This by and large leaves seasoning – which also has positive characteristics for a bear market – and a trickle of late-2008 and later production that’s less likely to have defaults.  We recommend caution on clean new production, which will be exposed to traditional rate/term refi risk; that’s one area in which smaller loans are attractive.  Finally, keep in mind that with TBA prepaying to slower speeds, all spec pool payups should be lower than they were when generic average CPRs were in the 50-80 range.

Fannie vs Freddie Credit Books

Fannie's serious delinquency rate has been approx 40% higher than Freddie's for the past year-and-a-half.

Some of this difference could be attributable to the fact that Fannie's delinquency buyout backlog appears to be bigger.  However, there are observable differences in the credit quality of the two books from the bubble years.  And the two GSEs worked with different sets of larger originators, which may have contributed to differences in underwriting quality.

Fannie’s book from the bubble years has objectively worse credit characteristics, with noticeably higher exposure to loans with loans with original LTVs >90% and moderately worse FICO distributions.  Fannie also has slighlty higher exposure to the troubled "sand states" of CA, FL, AZ and NV.

It's worth noting that in the 2009 vintage, which was originated in a stringent credit environment, Fannie and Freddie are similar in both observable credit characteristics and early delinquencies.  This implies similar performance in defaults and traditional rate/term refi prepayments for new lower coupons.

Even when we compare loans on a "similar-credit characteristic" basis (at left, from the most recent 10-Ks), Fannie loans are performing consistently worse than Freddies.

We believe the market is still underestimating the future Fannie / Gold prepayment differential, creating opportunities in passthroughs and derivatives.

Gold-Fannie Swaps After the Delinquency Buyout Catch-Up Period

We don’t think the market is being realistic in valuing how differences in the credit quality of the two books will affect prepayments and carry in higher coupon Gold-Fannie swaps going forward.  The table above takes a simple approach to valuing the swaps “post-catch-up-buyout” (i.e., in June or July).  Unless we see a miraculous turnaround in credit performance, Fannies will almost surely experience higher default rates – and therefore faster prepayments on 2005-2008 30-year 5.5s and above.  Even the most conservative approach to valuing the swaps – 10% faster Fannie prepayments, valued on an even-spread/N basis – results in swaps of +0-15, +0-12 and +0-07 for Gold-Fannie 6.5s, 6.0s and 5.5s, respectively.

Referring back to pg 3, our estimates of “fair value” swaps in Mar assumed that the swaps revert to 8, 6 and 4 ticks, roughly half of the most conservative swap values.  And even with an ultra-conservative set of assumptions, Gold-Fannie swaps currently look cheap to our fair value estimates.  We strongly recommend emphasizing Golds at these price levels.

Paul Jacob is the Director of Research for Banc of Manhattan Capital.  For a hard copy of the report, please contact the author at This e-mail address is being protected from spambots. You need JavaScript enabled to view it .
Copyright 2010 Banc of Manhattan Capital.  All rights reserved.