December 16, 2019
There was increased demand for new-issue hybrid ARMs last week, which caused yield spreads to Treasurys to tighten a basis point or two. Hybrid ARMs outperformed shorter mortgage-related sectors with 15-year fixed-rate mortgage spreads unchanged. On the contrary, ARMs underperformed 30-year fixed-rate mortgages, which tightened 3 bps on the week. The broader bond market moved up in price, sending domestic yield slightly lower along the curve despite confirmation of a long-awaited “phase one” U.S. China trade deal.
Since the rally at the end of 2018, ARM pricing spreads have widened significantly, reacting strongly to each move lower in rates. For example, 5/1 ARMs have a 60 bp spread, almost 32 bps wider than they were in mid-February. Longer-reset 7/1s and 10/1s have a 64 and 73 bp spread, respectively, approximately 26 and 23 bps wider than levels in mid-February. Certainly, the environment for ARMs has changed dramatically over the years with the flattening yield curve, but today’s spreads are well wider than those seen during 2017 at lower dollar prices.
Factors such as diminished liquidity, lack of index sponsorship, and the small market size have increased their spread concessions to fixed rates. Despite the convergence, spreads are wider on 7/1s versus their 15-year fixed rate counterparts. Overall, we continue to see relative value in 7/1s due to appealing yields, shorter durations, and less negative convexity than comparable coupon 15-year fixed rate MBS. Investors concerned about potentially faster prepayments could focus on lower WAC new-issue pools or moderately seasoned paper.
The ARM origination cycle was light last week, with 125.8mm in new-issue ARM selling split amongst Fannie Mae (111.9mm), Freddie Mac (11.8mm), and Ginnie Mae (2.1mm). Supply continues to be focused in 7/1s with Fannie Mae issuing 62.6mm. Fannie Mae also contributed to longer-reset 10/1 issuance with 29.6mm. 3/1 ARM products continue to be largely abandoned by lenders and the GSEs with monthly issuance not exceeding 100mm since May 2018. ARM gross issuance remains at multi-year lows as it came under 1 billion for the seventh consecutive month in November.
Hybrid ARM issuance remains quite low. The monthly net supply of ARMs continues to run at a negative $2-3 billion pace, while fixed rates are expected to grow at $20-30 billion each month for the rest of the year. As of December, hybrid ARM issuance represents ~ 0.71% of overall MBS issuance.
Last week, ARM activity was spread across a variety of lists and primarily focused on the following:
- Moderately-seasoned Freddie 5/1s with lower yields (less than 2%) and less than 10 months-to-reset traded at a slight premium ($102+). With the bulk of the market still at a premium, prepayment risk is still a concern for many investors, and 5/1 borrower prepayment speeds tend to be more muted than 3/1s in most rate environments.
- New-issue Fannie 10/1s with ~2.3 yield and slightly less than 10 years-to-reset traded at a lower dollar price (~ $102). Compared to GNMA hybrids, agency hybrids have a more generous cap structure – typically 5/2/5 vs. 1/1/5 – which allows the coupon to move more if rates move up more dramatically.
- New-issue Fannie Mega 10/1s with ~2.2 yield and less than 10 years-to-reset traded at a lower dollar price (~ $102).
On July 11th, the Alternative Reference Rates Committee (ARRC) released a white paper detailing how an average of the Secured Overnight Financing Rate (SOFR) can be used in newly-issued ARMs in a structure that is comparable to today’s existing ARM loans. The white paper shows how SOFR can be used to develop products that are built on a robust reference rate that is grounded in market transaction. Here’s an overview of the ARRCs proposed models of SOFR ARMs:
- Most aspects of a SOFR-based ARM would use the same conventions that currently exist in US Dollar LIBOR-based ARMs, including the range of fixed-rate periods available, the timing of payment determinations, and the structure of caps on the amount that mortgage payments can rise at the end of the fixed-rate period and over the life of the mortgage.
- However, a few key components would differ:
- These ARMs would be based on a 30- or 90-day average of SOFR, rather than 1-year LIBOR. Because SOFR tends to be lower than 1-year LIBOR, the margin for newly originated SOFR-indexed ARMs would likely be adjusted upward so borrowers’ overall floating-rate payments are comparable to existing LIBOR-based ARMs.
- In order to ensure that these SOFR-indexed ARMs can be offered at rates consistent with other competitive rates in the market, under the proposed models SOFR-indexed ARM, the borrower’s monthly floating-rate payment would adjust following the fixed-rate period once every six months rather than once every year, as is currently the market standard for LIBOR-based ARMs in the United States.
- To safeguard against unexpected payment increases to the borrower, the proposed models would cut the periodic adjustment cap to 1%. As a result, the borrower’s payment, even in a period of rapidly rising interest rates, would not change by more than 2% over a 12-month period, in accord with the current market convention for LIBOR-based ARMs.
The desk continues to look to bid odd-lot positions for both conventionals and Ginnies for clean-up. The disposition of odd-lot positions can result in enhanced transactional liquidity and higher earnings. Also, this is an opportunistic time to consider eliminating smaller line items that are subject to standard safekeeping and accounting fees that are more palatable for larger block sizes.
Ricky Brillard, CPA
Senior Vice President, Investment Strategies
Vining Sparks IBG, LP