Uncategorized
The rise of forebearance, the fall of prepayments in RPLs
admin | April 26, 2019
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.
Prepayment speeds have slowed noticeably in newer vintage private-label MBS backed by reperforming loans, and one significant likely cause is the growing amount of forbearance included in these transactions. The percentage of deal principal coming from forbearance has gone from 0% to 4% in the early years of RPLs to 8% to 13% lately. Given comparable WACs, RPLs with forbearance have prepaid significantly slower than those without. Speed differences across RPL vintages should not only effect yield and average life, but the pace of deleveraging, prospects for upgrades and total return, too.
More RPLs sourced from the GSEs
The amount of forbearance securitized in private-label RPL deals generally has been increasing, putting them more in-line with Freddie Mac’s RPL transactions. This looks due in large part to the rising share of RPLs in private deals sourced from the GSEs. Between 2014 and 2018, balances of modified 1-4 family RPLs on commercial bank balance sheets have fallen from nearly $70 billion to just over $30 billion. By comparison, the GSEs collectively held nearly $140 billion in performing RPLs as of the end of last year. If a growing share of loans going into private label RPL transactions are sourced from the GSEs, it’s reasonable to expect the disparity between the amount of forbearance in private and GSE RPL deals to converge.
Exhibit 1: Balances of performing RPLs across banks and the GSEs
Source: SNL Financial, Fannie Mae, Freddie Mac balances as of 12/31/2018.
A clear correlation between forbearance share and prepayment speed
GSEs deals, particularly earlier vintages, have significantly higher percentages of securitized forbearance and slower prepayment speeds than early private label RPL deals. However, as concentrations of loans with forbearance in private label RPL deals have risen over time, speeds have fallen (Exhibit 2). Deals with 0% to 4% reperforming principal have prepaid since early 2017 above 10 CRR while deals with 10% or more in reperforming principal have prepaid below 8 CRR.
Exhibit 2: Speeds on RPL deals tend to fall as forebearance balances rise
Note: Data show average voluntary CRR between January 2017 and March 2019. That window captures more months of prepayments on older deals than on newer ones. However, since RPLs typically come to market with significant seasoning, there’s usually no ramp or steady rise in speeds after issuance. The difference in number of months of prepayments should not bias the comparison. Source: Intex, Amherst Insight Labs, Amherst Pierpont Securities
While there appears to be a clear connection between the amount of forbearance in a deal and the level of prepayments, there may be a number of collateral attributes also contributing to slower speeds. Loan rate, amortization terms and levels of borrower equity will obviously all contribute to a borrowers’ propensity or ability to prepay. However, comparing loans with similar WAC but with and without forbearance suggests that loans with forbearance prepay significantly slower than loans without (Exhibit 3). It is not a surprising result. Forbearance often comes as a last resort, signaling a loan deeply underwater or a borrower with limited means to repay. More importantly, a loan with forbearance obviously pays 0% on part of the debt. Refinancing would result in the borrower paying interest on that interest-free balance. Even if the remaining balance carries a high rate, interest rates would have to drop significantly to result in a lower all-in monthly payment.
Exhibit 3: Loans with forebearance prepay significantly slower in RPL deals
Source: Amherst Insight Labs, Amherst Pierpont Sample is exclusive to TPMT shelf
Increasing amounts of balance modifications may not be the only factor driving slower prepayments. For instance, there has been a pronounced decline in average loan balances across the MCMLT shelf since issuance. Average loan sizes in 2015 issuance on the shelf were in excess of $300,000 and fell consistently below $200,000 in their four 2018 deals. And while likely less of a driver, original FICOs have fallen by as much as 50 points from 2015 to 2018 issuance. While a less consistent trend, average WACs after factoring in non-interest bearing balances across the TPMT shelf have trended lower over time. The slowest observed prepayments across the TPMT shelf are in their two 2017 Freddie Mac seller financed ‘SLST’ transactions, which had the highest amount of securitized forbearance across the shelf’s issuance at 19% and 18% in the FRE1 and FRE2 deals respectively. These deals have average WACs of roughly 3.25% after adjusting for securitized forbearance and are prepaying in the low single digits.
Some risk for credit
It’s important to note that the impact of forbearance in new RPL deals differs from the impact on legacy private-label MBS. In a majority of legacy private-label MBS, forbearance is taken as a loss to the trust when the balance modification takes place and offers upside to the remaining cash flow if it is partly or wholly recovered. In both new private-label and GSE RPL transactions, forbearance is securitized in the principal balance. Given this, a lack of recovery would result in a loss to the trust. Growing amounts of non-performing balances may not only be a prepayment concern but a credit concern as well.
Investment Implications
Increased forbearance and the correlated slowdown in prepayments could have meaningful implications for investors in private-label RPL transactions. Given the sequential nature of the structure, a significant source of potential return comes as the deals prepay and delever. With deleveraging, mezzanine and subordinate bonds will likely get upgraded and trade to tighter spread. Because of the sequential structures, these bonds have significant spread duration and price appreciation associated with deleveraging and upgrade can be significant especially in lower investment grade and non-investment grade classes. A material slowdown in prepayment speeds would slow the rate at which these deals delever, slowing any potential upgrades. Given this, total return investors may favor earlier vintage transactions with smaller amounts of forbearance than more recent transactions with greater amounts.