Since early November, the ten-year Treasury has rallied down to around 2.90% from over 3.20%. Explanations include growth jitters, short-covering from over-enthusiastic shorts who were expecting inflation to continue picking up faster than it has, and asset allocation shifts due to falling equity markets. The 2s10s slope is now a mere 15 bps with the 2s5s portion of the Treasury curve fluctuating between flat and slightly inverted. The next Fed meeting in December promises to be quite interesting, perhaps more in terms of the after-party dialogue than in the rate decision itself. Meanwhile credit spreads, in both corporate and mortgage credit have continued widening out, with even agency MBS spreads a bit wider. This promises an interesting, highly data-dependent outlook for 2019, where employment, wage growth and inflation numbers will help determine inflation and term spreads, and quantitative tightening continues to pressure real rates.

With this outlook in mind, we have an interesting combination of articles to present in our final issue of 2018. Our first article, by Alex Levin, looks at the impact that the FHFA data update (which reflect slowing appreciation during our higher rate period) are having on our outlook for house price appreciation as well as the resultant impact on credit-sensitive bonds. Alex shows that HPA can be expected to slow, but that we do not see national HPI declines—an outcome perhaps more similar to what happened in 1994 than in 2008. He then goes on to examine the impact on a range of MBS collateral, including credit risk transfer.

Our second article, by Dan Szakallas, looks at the nexus of rates and house prices through the lens of refinancing behavior. Those following agency prepayments may have been puzzled by a surge in out of the money coupon prepayment speeds earlier this year (where both rates and prepayments rose). Dan identifies a particular segment of the CLTV distribution of borrowers that appears to have disproportionately contributed to this increase. His explanation that this suggests cash-out refinancing activity with a desire to do so before rates go even higher is a story that sounds accurate, though we will not be able to definitively confirm our hypothesis until the next loan-level data release from the agencies. The open question this leaves is whether this will result in “burn-out” in the cash-out component of refis after this wave, or if there may be additional CLTV cohorts that may surprise us when the next mortgage rate back-up seems to be upon us.

Finally, Sanj Chatterjee updates us with a full tuning update on non-agency LDM. While non-agency origination continues to be a low percent of an overall slowing origination market, the recent nomination for a new head of FHFA and some his prior statements have led to some speculation on the potential direction agency loan limits may take in the future. If reduced or even maintained flat, this has the potential to increase the relative size of the non-agency market in the future. In any case, Sanj shows that we continue to make sure that our non-agency model and its various transitions are being monitored and kept on track. The model has done an excellent job with relatively little tuning the last several years.

As always, we welcome comments, questions or suggestions for future articles. We’d like to wish all our readers a wonderful holiday season and a healthy and happy 2019

As we have heard from the media, the home price appreciation rate has slowed. At the end of November, the FHFA released quarterly (2018Q3) reports confirming this observation. Based on this information, we moderated our HPI outlook. As a result, we show that 5 to 10 percent of non-agency MBS and 19% of CRT tranches would be “downgraded” using AD&Co’s dynamic CPC (CreditProfile Category) metric. We project losses of conforming loans to increase in a range that extends from 7.4% (for base case) to 22% (for extreme stress)—relative to prior estimates.

It is important to emphasize that the updated HPI outlook points to slower future increases, not a fall, in HPI.

During the course of 2018, some interesting patterns emerged regarding prepayments for the FNMA and FHLMC security pool data. As mortgage rates began rising in January, prepayment speeds also began to rise, but at levels well above what the seasonal trend would normally bring. The increase was seen across the coupon stack, which was a bit puzzling because as rates rose, more and more of the coupon stack was becoming either par or out of the money. In this article, we are going to take a deeper look into the possible reasons for this behavior.

Andrew Davidson & Co., Inc. (AD&Co) has released new tunings for the Non-Agency LoanDynamics Model (NALDM). The past few years have seen improving economic conditions. Unemployment is at historically low levels, and home prices have increased. At the same time, however, rates have been creeping higher. Activity in the non-agency space is still limited to a handful of deals every year, and the legacy portfolios are slowly running off.

NALDM was last tuned about two years ago in several phases. This updated tuning set brings the model closer to actual performance over the last two years. In most cases, the model transitions needed only slight tuning adjustments. Also, the tunings have been stable over the past two years, meaning that the tuning values have remained stable over time. This is a promising outcome from a forecasting perspective, as it signifies that the base model is quantifying recent borrower behavior fairly well.

Tunings have been updated for all collateral types and for all NALDM model versions.