By the Numbers
Exploring 15-year MBS returns
Brian Landy, CFA | March 12, 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 has generally underperformed other parts of fixed income over the last month as nominal MBS spreads have widened sharply. Spreads for par 15-year MBS have widened more than spreads for par 30-year MBS, but a steeper curve, higher volatility and tighter OASs has helped the shorter paper. Prices for 15-year pools could move even higher as investors seeking extension protection show more interest in shorter MBS and as a steeper curve helps slow prepayment speeds in premium 15-year pools.
Investors often track MBS performance against Treasury yields using the spread between the par MBS coupon and a fixed point on the curve. The average of the 5-year and 10-year Treasury yields is the typical reference for 30-year MBS, while the average of the 3-year and 5-year Treasury yields is typical for 15-year MBS. This helps account for the shorter duration of the 15-year product. The difference between these two spreads ranged from roughly 40 bp to 60 bp from 2016 through 2019 (Exhibit 1). However, the spread tightened in 2020 and has now collapsed to less than 10 bp.
Exhibit 1: The spread between par 30-year and 15-year MBS has collapsed.
The spread between the 30-year and 15-year current coupon OAS has also been tight throughout the pandemic, another indicator that 15-year MBS may offer good relative value to 30 year MBS (Exhibit 2). Before March 2020, the typical spread was 10 bp, with occasional widening to as high as 25 bp to 30 bp. During the pandemic, the typical spread has been 5 bp and touched 10 bp at its widest. However, the OAS spread has widened since late January and reached 10.8 bp on March 11.
Exhibit 2: OAS spreads have been tight throughout the pandemic
The nominal spread tightening would typically suggest that 15-year MBS underperformed 30-year MBS. However, a duration- and proceeds-neutral portfolio on February 1 that was long 15-year 1.5%s and 10-year Treasuries and short 30-year 2.0%s had a positive total return. An example of this portfolio would have been long roughly $900,000 FNCI 1.5% and roughly $100,000 10-year Treasuries, and short roughly $1,000,000 FNCL 2.0% (Exhibit 3). The portfolio is duration-neutral but has positive convexity, so it should outperform in large rate moves in either direction. There is also some exposure to the shape of the curve, as it has a negative 5-year key rate duration. This means it should increase in value if the 5-year rate increases.
Exhibit 3: A duration- and proceeds-neutral 15-year MBS portfolio.
The position earned $5,828 as of March 3 (Exhibit 4). The portfolio duration has shifted somewhat negative, which was to be expected given the positive convexity and move to higher interest rates. Most of this negative duration exposure is in the longest part of the curve. The portfolio is still positively convex, but less so than on February 1.
Exhibit 4: The position gained $5,828 from 2/1/2021 through 3/3/2021.
The long position’s return can be attributed to a combination of changes in interest rates, tighter OASs on the 15-year MBS, and higher volatility (Exhibit 5). Changes in interest rates account for roughly 40% of the total return. The curve steepened as rates moved higher, but the position has roughly half the exposure to 10-year yields than to 5-year yields. Since 10-year and 30-year rates moved very similarly, the long exposure can be approximated by adding the 10-year and 30-year key rate durations. The curve steepened roughly 9 bp, as 5-year yields increased 30 bp while the 10-year increased roughly 39 bp. The position earned money since the 10-year move was not at least double the 5-year move. There was also a boost to the return since the portfolio has positive convexity.
Exhibit 5: Curve, spread, and volatility all contributed to positive returns
Spreads also played a large role, contributing roughly 50% of the total return. Most of this was due to the 15-year position, which had its OAS tighten almost 6 bp. But the 30-year OAS also widened a little, which contributed some positive return.
Finally, volatility increased in February. The long position in 15-year paper and Treasury debt was less exposed to volatility than the short position, since the 15-year MBS has less optionality than the 30-year MBS and the Treasury note has none. This accounted for most of the remaining total return. There is also a small amount, labeled “unexplained.” This represents total return due to all three factors moving simultaneously, as opposed to changing them one at a time.
The 30-year position should outperform the 15-year position if markets do not move because the 30-year MBS offers roughly 1.3/32s additional carry each month. Markets need to move for the 15-year to outperform. Exhibit 4 shows that the hedged portfolio has negative rate, spread, and volatility durations so will benefit from higher rates, wider OASs, and higher volatility. The 15-year position will overcome the 30-year carry advantage if interest rates increase at least 6.9 bp, OAS tightens by at least 2.2 bp, or vol increases by 1.32%.
The position should also benefit from a steeper curve. The 10-year key rate duration is currently neutral but there is a lot of exposure to the 30-year rate. Assuming they move together the 2s10s curve slope duration can be approximated as the 30-year duration plus the 2-year duration, or 0.56 years. This implies a 7.2 bp steepener would also overcome the 30-year carry advantage.
Performance of 15-year MBS could improve further if the nominal spread relationship returns to historical levels. Investors could show more interest in 15-year MBS in the current environment, since the steeper curve should slow prepayments on premium 15-year MBS while the shorter term will attract interest from investors looking for some extension protection.