By the Numbers
Loan size, occupancy and servicer effects in PLS speeds
Chris Helwig | October 1, 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.
As issuance increases across prime PLS, the profile of prepayment risk continues to change. The mix of PLS loans eligible for delivery to Fannie Mae and Freddie Mac has started to change, and that should shape prepayment risk in new deals. The servicer of loans in private-label trusts may influence prepayment rates as well.
Looking at most loans backing prime private-label trusts, investors can break down the universe into a relatively simple taxonomy:
- Owner-occupied loans with conforming jumbo balances
- Owner-occupied loans with non-conforming jumbo loans, and
- Loans backed by investment properties.
Some broad patterns apply. Given the same refinancing incentive, loans with agency-eligible balances prepay somewhat faster than those that cannot be delivered and that loans backed by investment properties are not as efficiently refinanced as owner-occupied loans or even second homes.
Breaking down the universe by loan size and occupancy
Historically, loans with higher balances have tended to prepay faster than those with lower ones. However, in recent years, conforming jumbo loans eligible for GSE delivery have prepaid faster than non-conforming ones. Loans eligible for the GSEs had lower frictions to refinancing, and that was especially true as liquidity from private-label conduit execution for non-conforming loans dried up last year. As the private-label market has rebounded, differences in prepayment rates between loans with jumbo conforming and non-conforming balances have narrowed. After controlling for any additional risk-based pricing on loans, conforming jumbo loans only pay 3 CRR faster than non-conforming ones given 50 bp of refinancing incentive and 5 CRR faster given 100 bp of incentive. (Exhibit 1)
And it appears that investors in private MBS will be exposed to growing amounts of larger loan balances going forward. Until the third quarter of this year, the percentage of conforming balance loans going into private-label trusts has been on a steady decline while the amount of larger loans has been growing. In the first quarter of 2019, roughly two-thirds of loans backing PLS trusts had conforming jumbo balances. That share dropped to just 23% of total issuance in the first quarter of this year. Conversely, loans with balances greater than $1 million made up just 10% of collateral backing private trusts in the first quarter of 2019 and jumped to 37% of collateral balances in the first quarter of this year. The combination of rising home prices and a substantial tightening of spreads of specified jumbo pool spreads relative to their TBA benchmarks suggests both that loan balances should continue to grow and that conforming jumbo loans may likely fetch better prices by being delivered to the enterprises especially in the wake of the expiry of the 50 bp adverse market delivery charge that was previously applied to all refinances delivered to the enterprises.
Exhibit 1: Prepay differences between conforming and non-conforming loans narrow
Source: CoreLogic, Intex, Amherst Pierpont – SATO adjusted, observations exclusive to fixed-rate loans
Turning to occupancy, loans backed by investment properties in private-label trusts continue to, by and large, prepay slower than those backed by owner occupied or second homes, although those differences vary somewhat materially after controlling for loan size. The most pronounced differences in prepayment rates given the same amount of SATO adjusted refinancing incentive exists in investor loans with non-conforming balance and that difference decreases as loan size decreases with the caveat that there is a small population of loans with non-jumbo conforming balances securitized in the outstanding universe of PLS trusts. Of loans with non-conforming balances and 100 bp of refinancing incentive, owner occupied loans have prepaid at 56 CRR while loans backed by investment properties have prepaid at roughly 32 CRR. In loans with jumbo conforming balances and the same 100 bp of refinancing incentive, owner occupied loans have paid at 62 CRR while investor loans have prepaid 20 CRR slower. While the difference between loans with conforming balances is just 3 CRR with owner occupied loans prepaying at 48 CRR and investor loans prepaying at 45 CRR.
Digging in on servicers
Controlling for loan size and looking first at servicers of non-conforming jumbo balance loans, which appear poised to make up a larger part of the private-label universe going forward, shows some pronounced differences in prepayment speeds for borrowers with the same amount of refinancing incentive. Simply controlling for loan size and incentive may not fully isolate the servicer effect as there are other factors away from loan size that may be contributing to a difference in prepayment rates. However, based on these simple controls, it does appear that certain servicers are exhibiting faster speeds. Given 100 bp of SATO adjusted incentive, loans serviced by Cenlar and Wells Fargo have prepaid the fastest while those serviced by First Republic and Nationstar have prepaid slower. (Exhibit 2)
Exhibit 2: Stacking up servicer speeds across non-conforming jumbo loans
Source: CoreLogic, Intex, Amherst Pierpont – SATO adjusted, observations exclusive to fixed-rate loans
Loans with jumbo conforming balances exhibit some differences in servicer speeds as well. Given 100 bp of SATO adjusted incentive, Loans serviced by Provident and United Shore have prepaid substantially faster than the majority of the cohort while loans serviced by First Republic have prepaid substantially slower than most of the cohort. (Exhibit 3)
Exhibit 3: Stacking up servicer speeds across conforming jumbo loans
Source: CoreLogic, Intex, Amherst Pierpont – SATO adjusted, observations exclusive to fixed-rate loans
Narrowing the scope of the analysis to loans backed by investment properties shows some differences as well. With 100 bp of incentive loans serviced by United Shore and Shellpoint have exhibited speeds faster than the broader cohort and substantially faster than those serviced by First Republic and Nationstar. Controlling the population for only conforming balance investor loans in private-label trusts shows an even more pronounced difference. (Exhibit 4)
Exhibit 4: Stacking up servicer speeds across conforming investor loans
Source: CoreLogic, Intex, Amherst Pierpont – SATO adjusted, observations exclusive to fixed-rate loans