By the Numbers
Don’t be fooled by flatter S-Curves
Brian Landy, CFA | June 7, 2024
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. This material does not constitute research.
As 30-year mortgage rates dropped in the last two months of 2023, the small share of loans that became refinanceable showed a surprisingly modest response. Prepayments in those loans barely picked up, especially compared to similarly refinanceable loans a few years ago. Blame it on spread-at-origination, or SATO. Loans with high SATO usually respond slowly to emerging refinance incentives. But investors that generalize recent refinance behavior to loans now getting originated with low SATO could get fooled. Those low SATO loans should show the usual strong refinancing response that played out regularly in the first years of pandemic.
Prepayment speeds for refinanceable or in-the-money loans lately have lagged prepayments on comparable loans during the pandemic. That shows up clearly in S-curves—prepayment speeds plotted against rate incentive—for loans between $400,000 and $500,000 during the pandemic and after the pandemic (Exhibit 1). Rate incentive is calculated relative to Optimal Blue’s rate lock index for 720 to 740 FICO loans with LTV no more than 80%. Post-pandemic S-curves are slower than the pandemic S-curves when loans are in-the-money, and the gap grows as loans go further in-the-money. Speeds are also a little slower overall since housing turnover is not as robust as during the pandemic. But these pictures need more scrutiny.
Exhibit 1. S-curves for loans between $400,000 and $500,000.
Conventional 30-year fixed-rate loans, owner-occupied, 6 to 36 WALA, original FICO≥700, original LTV≤80. Low SATO is ≤25 bp SATO. Source: Fannie Mae, Freddie Mac, Santander US Capital Markets.
The collateral for these pandemic and post-pandemic S-curves is not comparable, even though the collateral has similar FICO scores, LTVs, WALAs, and loan sizes. The bar chart at the bottom plots the SATO of loans at different points across the S-curve, and there is a striking difference for in-the-money loans. The pandemic loans’ SATO was roughly 0 bp in-the-money, but the post-pandemic SATOs are very positive. For example, loans 100 bp in-the-money had, on average, 100 bp SATO. That means the speeds on the post-pandemic S-curve are unlikely to be representative of speeds on low SATO loans in a deeper rate rally.
The chart in the middle shows the S-curves for only low SATO loans. There is not a lot of data for the loans in-the-money, but the S-curve does appear to be steeper than the all-in S-curve. It has a similar shape to the pandemic S-curve, although is shifted to the right. The right-hand chart shows the S-curves for high SATO loans in both periods—speeds are notably slower in- and out-of-the-money and the elbow shift is evident, but the general shape of the S-curve is like the shape of the pandemic S-curve.
The difference is smaller at lower loan sizes (Exhibit 2). This shows loans between $175,000 and $200,000 original balance. The chart on the left shows a similar picture as the higher balance loans—the S-curve looks flatter in-the-money compared to the pandemic S-curve and it is dominated by high SATO loans. Controlling for SATO recovers much of the shape of the pandemic S-curve, although little slower from the elbow shift and slower housing turnover. However, these differences are smaller than they were for the higher loan sizes.
Exhibit 2. S-curves for loans between $175,000 and $200,000.
Conventional 30-year fixed-rate loans, owner-occupied, 6 to 36 WALA, original FICO≥700, original LTV≤80. Low SATO is ≤25 bp SATO. Source: Fannie Mae, Freddie Mac, Santander US Capital Markets.
At very small loan sizes the S-curves look very similar during and after the pandemic (Exhibit 3). This shows speeds for loans between $110,000 and $125,000 loan balance. The all-SATO S-curves on the left aren’t too different despite the bias to high SATO post-pandemic, although optically it looks like it is starting to flatten out as it approaches 100 bp in-the-money. And the pandemic and post-pandemic S-curves are even closer to each other after controlling for SATO.
Exhibit 3. S-curves for loans between $110,000 and $125,000.
Conventional 30-year fixed-rate loans, owner-occupied, 6 to 36 WALA, original FICO≥700, original LTV≤80. Low SATO is ≤25 bp SATO. Source: Fannie Mae, Freddie Mac, Santander US Capital Markets.
Investors should be cautious about assuming the S-curve is flattening and that speeds on recent vintages will not return to levels seen during the pandemic if rates dropped enough. Recent S-curves look flat because few loans have been in-the-money since late 2022, and most of those loans have high SATOs. But originators will be highly motived to restart the refinancing machine if the opportunity arises. And the general trend is likely towards making refinancing a little cheaper and easier as origination technology evolves, and the share of non-bank servicers keeps growing. Fast prepayment speeds caught the market off-guard in 2020, much of which was due to difficulty measuring the prevailing mortgage rate. Those speeds, and not the experience of the last few months, should guide views on speeds. If rates fall—don’t be fooled again!