Uncategorized
A likely muted impact of COVID-19 on legacy RMBS
admin | May 8, 2020
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.
The potential impact of COVID-19 on borrower and servicer behavior in legacy RMBS suggests a muted impact on security performance even under fairly harsh stresses. Low mark-to-market LTVs, existing trends in making payments and a history of post-modification performance are likely tailwinds to continued relatively strong fundamental performance, potentially making legacy RMBS a more attractive way to express a view in mortgage credit than other sectors.
Given the relatively unique potential impact of COVID-19, stressing borrowers’ probability of making payments appears to be a reasonable approach to estimating both borrower and servicer behavior and ultimate defaults and losses across loans in legacy RMBS. The approach assumes a 20% decrease in payment probability for all loans for a period of six months. The COVID-19 overlay additionally assumes zero prepayments and no loans rolling to foreclosure, REO or liquidation for the same six months. Even with a 20% reduction in payment probability and other adjustments, the estimated impact on loan prices is quite modest, just $3.50 on average, representing a 4.3% price decrease between a base case run with no overlays and a stressed run across a universe of nearly $275 billion in collateral. (Exhibit 1)
Exhibit 1: A substantial slowdown in payment probability yields modest loan price changes
Note: Data shows the percent of priced loans showing each category of price change. Loans are priced using a 6% yield to discount projected back case cash flows, and repriced using a 6% yield for scenario cash flows. Estimates as of 5/4/2020. Source: Amherst Insight Labs, Amherst Pierpont Securities.
Not all loans respond to projected payment stress the same way. On average, loans backing subprime trusts fell 5.8% while prime, Alt-A and Option ARM loans all fell in the neighborhood of 3.0% (Exhibit 2). Subprime loans had the largest tail of loans with outsized price declines as roughly 13% of subprime collateral saw a greater than 10% price decline in the COVID-19 overlay, roughly double the amount of loans with large price declines in Alt-A and Option ARM credits. Liquidation rates on subprime loans with the largest price declines increased from 33% to 47% in the COVID-19 overlay, driving cumulative losses on the loans from 23% 32%. The outsized increase in defaults was likely driven by higher average mark-to-market LTVs and more recent modifications on the tail of subprime loans versus other cohorts.
Exhibit 2: Subprime collateral has the largest tail of loans with large price declines
Note: Data shows the percent of priced loans showing each category of price change. Loans are priced using a 6% yield to discount projected back case cash flows, and repriced using a 6% yield for scenario cash flows. Estimates as of 5/4/2020. Distribution is scaled to 100% of each discrete loan type. Source: Amherst Insight Labs, Amherst Pierpont Securities.
As a result of the COVID-19 related overlays, the percentage of legacy loans liquidated increased from 18% to 30% of the universe. This was largely driven by an increased amount of always performing loans being pushed to liquidation due to the slowdown in payment probability. Comparing the population of loans with the smallest change in liquidations to the cohort with the largest change shows the population of loans with the smallest increase in liquidations was comprised of just 1% always performing loans and 86% non-performing loans while the bucket with the largest increase was comprised of nearly 40% always performing loans and no non-performing ones. (Exhibit 3) This illustrates the impact of the slowdown in payment velocity as it will have a much more outsized impact on a loan that otherwise would have had a very high probability of remaining always current in the base case scenario.
Exhibit 3: Changes to payment velocity push more APLs to liquidation
Source: Amherst Insight Labs, Amherst Pierpont Securities as of 05/04/20
The percentage of increased liquidations were fairly evenly distributed across loan types as roughly the same percentage of prime collateral saw large spikes in liquidations as those seen in subprime loans. (Exhibit 4) Looking at the worst performing buckets of prime and subprime collateral show roughly 25% increases in the amount of loans liquidated in both cohorts. In both cohorts, the worst performing bucket contains no non-performing loans. Given the large concentration of always performing loans in the prime bucket, the increase in liquidations is much more pronounced on a percentage basis as liquidations jumped from 5% to 25%. Base case liquidations in the subprime cohort with a much smaller amount of always performing loans grew from 15% in the base case versus 40% with the COVID-19 overlays.
Exhibit 4: Increases in liquidations are evenly distributed across cohorts
Note: Data shows the percent of priced loans showing each category of price change. Loans are priced using a 6% yield to discount projected back case cash flows, and repriced using a 6% yield for scenario cash flows. Estimates as of 5/4/2020. Distribution is scaled to 100% of each discrete loan type. Source: Amherst Insight Labs, Amherst Pierpont Securities.
Even with large spikes in payment velocity related liquidations, projected cumulative losses are small. Cumulative losses rise by just 3% across the universe of legacy loans as a result of the reduction in payment velocity and other COVID-19 related overlays. This is largely a function of lower LTV loans pushed to liquidation as a result of the payment velocity overlay. Low cumulative losses are largely a result of lower LTV loans liquidating that otherwise would have not absent the slowdown in payment velocity. At the universe level, liquidated loans had an average mark-to-market LTV of 55. As a result of this, loss severities on average decline by six points from 42 to 36 between the base case and COVID-19 scenarios. The decline in severities is driven by the fact that many of the loans being pushed into liquidation in the COVID-19 scenario will ultimately liquidate without a loss, effectively looking more like a prepayment out of the non-performing loan pipeline or ‘no-loss liquidation.’
So despite what optically appear to be harsh stresses on borrower behavior, it appears that COVID-19 may be much more of an extension event in terms of modification than a default-and-loss event for legacy collateral. Capital structure leverage will still matter, and even a modest uptick in cumulative losses or decline in excess spread from rate modifications could have an outsized impact on thinner mezzanine bonds with no credit support and significant leverage to excess spread. So while COVID-19 may have a marginal impact on legacy collateral as a whole, its impact will not be zero and levered profiles may need to reprice accordingly.