By the Numbers
Framing relative value in hybrid ARMs
Brian Landy, CFA | August 27, 2021
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
MBS investors looking for short duration may find value in hybrid ARM pools. The Fannie Mae and Freddie Mac transition to loans indexed to SOFR in late 2020 disrupted ARM supply, leading to a temporary drop in origination early this year. But production has rebounded lately, pushing ARM spreads wider. Hybrid ARM pools offer better yield and OAS than 15-year fixed-rate pools with similar duration. And hybrid ARMs currently offer better projected 1-year total return unless rates fall more than 100 bp.
Disruption in ARM supply
The transition from LIBOR- to SOFR-indexed hybrid ARMs disrupted supply for much of the past year. Both GSEs stopped acquiring LIBOR-indexed ARMs at the end of 2020 and would only acquire loans started before October 2020. This led to a sharp drop in supply from November 2020 through May 2021 (Exhibit 1). But production rebounded in June as originators began originating SOFR-indexed ARMs. Supply has topped $1 billion for each of the last three months. From 2019 through October 2020 supply averaged roughly $750 million a month and only exceeded $1 billion three times. Demand may also be lower for the new ARM pools since some investors may not have approval to buy pools indexed to SOFR.
Exhibit 1. Hybrid ARM supply has bounced back after plunging in late 2020
Source: Yield Book, Amherst Pierpont Securities
Spread and projected return relative to 15-year MBS
The sharp rebound in supply has pushed ARM spreads wider, and pools appear to offer relative value compared to 15-year MBS (Exhibit 2A). The exhibit compares two trades. The first is a 7/6 SOFR hybrid ARM—the loans pay a fixed rate for seven years, after which the loans reset every six months to SOFR + 2.37%. The weighted average coupon is currently 1.77%; although the loans’ coupons will not adjust for nearly seven years, the average coupon can change when loans prepay. The comparison trade is a duration- and proceeds-neutral combination of 15-year MBS and cash. The ARM pool offers higher yield and OAS than the 15-year position.
Exhibit 2a. A 7/6 SOFR ARM offers higher yield and OAS than 15-year 2% MBS
Note: All market levels as of 8/23/21 market close. Using Yield Book’s v21.7 prepayment model.
Source: Yield Book, Amherst Pierpont Securities
The ARM pool is somewhat more negatively convex than the 15-year trade. A 12-month total return profile sheds some light on this (Exhibit 2b). In the base case, the ARM is projected to return 0.35% more than the 15-year trade. The ARM does slightly better if rates increase but worsens if rates fall. However, the total return stays above the total return on the 15-year trade even if rates drop 100 bp.
Exhibit 2b. The ARM pool offers better one-year total returns in various rate scenarios
Note: All market levels as of 8/23/21 market close Using Yield Book’s v21.7 prepayment model. Interest rates shift linearly to the one-year horizon. Reinvestment at 1-month Treasury rates. Horizon repricing at constant OAS.
Source: Yield Book, Amherst Pierpont Securities
There are some caveats to consider. Investors that can roll their TBA positions may be able to take advantage of special dollar roll financing, which increases the value of the 15-year position. But not all investors can dollar roll. Another consideration is that the ARM market is less liquid than that for 15-year MBS, and ARMs are not included in the Bloomberg Barclays Index. Therefore, ARMs ought to trade at a spread concession to 15-year MBS. But there is little historical data available to measure the typical amount of that concession. Finally, the SOFR index, unlike LIBOR, does not incorporate a credit component. Investors should not expect the SOFR coupon to reset higher in a period of credit stress.
Another consideration is that the example pool is serviced by Quicken. However, a lot of recent production came from Quicken and loanDepot, and relatively less from Wells Fargo and Chase. But the two bank’s loans typically have initial note rates that are 15 bp to 20 bp higher than the non-bank loans, which adds some prepayment risk to those loans.
Value relative to PACs
It is also useful to compare the ARM pool to other mortgage products that do not have the same liquidity advantage as 15-year MBS. The first is a short PAC backed by jumbo 2.5% pools and pays a 1.5% coupon (Exhibits 3a and 3b). The static yield of the ARM is 10 bp lower than that of the CMO, but the ARM pool’s OAS is 22 bp higher. The ARM position is more negatively convex than the CMO, underperforming in a rally but outperforming in a sell-off. The choice may depend on an investor’s view of future interest rates.
Exhibit 3a. Value comparison of a 7/6 SOFR ARM to a short CMO PAC
Note: All market levels as of 8/23/21 market close. Using Yield Book’s v21.7 prepayment model.
Source: Yield Book, Amherst Pierpont Securities
Exhibit 3b. Total return comparison of a 7/6 SOFR ARM to a short CMO PAC
Note: All market levels as of 8/23/21 market close Using Yield Book’s v21.7 prepayment model. Interest rates shift linearly to the one-year horizon. Reinvestment at 1-month Treasury rates. Horizon repricing at constant OAS.
Source: Yield Book, Amherst Pierpont Securities
When evaluating the CMO investors should also consider that the PAC bands could narrow if interest rates drop and prepayment speeds increase. The CMO is likely to trade wider if that happens. However, the total return scenarios assume a constant OAS and therefore likely overstate the value of the CMO in a rally.
Value relative to agency CMBS
Fixed rate Freddie CMBS are another short-duration asset that investors use as an alternative to 15-year MBS and are a useful comparison to the ARM pool (Exhibits 4a and 4b). The results are mixed—the ARM pool offers a higher static yield but has more negative convexity and a lower OAS. The total return profile shows that the ARM outperforms if rates remain relatively range bound, but the CMBS outperforms in a rally or a larger sell-off. Once again, the choice may come down to an investor’s view of future interest rate risk.
Exhibit 4a. Value comparison of a 7/6 SOFR ARM to a short fixed-rate Freddie CMBS
Note: All market levels as of 8/23/21 market close. Using Yield Book’s v21.7 prepayment model.
Source: Yield Book, Amherst Pierpont Securities
Exhibit 4b. Total return comparison of a 7/6 SOFR ARM to a short fixed-rate Freddie CMBS
Note: All market levels as of 8/23/21 market close Using Yield Book’s v21.7 prepayment model. Interest rates shift linearly to the one-year horizon. Reinvestment at 1-month Treasury rates. Horizon repricing at constant OAS.
Source: Yield Book, Amherst Pierpont Securities