Uncategorized
FICO and LTV signal the early delinquencies in non-QM
admin | May 1, 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.
Non-QM underwriting appears poised to get its first big test in the coming months as the effects of COVID-19 begin to stress borrowers. Loans 30-days delinquent doubled from March to April, and early indications show the strongest links to FICO and LTV. But it is still early. Occupancy, self-employment and documentation may signal risk if unemployment has an outsized impact on renters and small business owners.
The population of non-QM loans 30-days delinquent doubled from March to April to a monthly rate of nearly 3% (Exhibit 1). The increase now puts non-QM delinquency rates nominally in line with those of securitized RPLs, with the caveat that RPL transition rates were already somewhat elevated before COVID-19. In contrast, 30-day delinquency rates on post-crisis prime jumbo collateral jumped just 8 bp from March to April, representing a roughly 25% increase.
Exhibit 1: Non-QM 30-day delinquency rates double between March and April
Note: Transition rates are shown as month over month change on a monthly basis. Source: Amherst Insight Labs, Amherst Pierpont Securities
The first available data also show FICO closely linked to rising delinquencies. Loans with original FICO scores below 660 saw 30-day delinquencies jump to 6% or more in April, while loans with original FICOs greater than 720 transitioned at 2% or lower. Percentage increases were generally higher in lower FICO loans, as those with scores between 620-660 saw a roughly 130% increase month-over-month. However, the most pronounced increase on a percentage basis were loans with FICO scores between 740 and 760, which increased 225% month-over-month. This may very well be a 1-month anomaly, but certainly one worth monitoring as even more creditworthy borrowers may ultimately not be immune from the economic effects of the virus.
Exhibit 2: Low FICO non-QM loans saw much higher delinquency rates in April
Source: Amherst Insight Labs, Amherst Pierpont Securities
LTV mattered, too. Delinquencies on higher LTV loans were elevated relative to loans with lower LTVs in April. LTV appeared to have less of an impact than FICO as the difference between delinquency rates on lower and higher LTV loans was roughly 1%. Loans with original LTVs greater than 60 were roughly in line with the broader universe at approximately 3% or slightly higher in April. Lower LTV loans saw more modest rates.
Other loan attributes showed interesting patterns. Based on one month of performance after COVID-19, early delinquency rates on owner-occupied and investor loans came in roughly the same. In fact, rates on owner-occupied loans were modestly higher than investor loans in April at 3.1% and 2.7% respectively. So too, somewhat surprisingly were rates on full- and limited-documentation loans at 3.3% and 3.1%.
While it is impossible to glean from current unemployment data, there is a reasonable possibility that the impact of COVID-19 may have an outsized impact on renters relative to homeowners. This potentially puts investor loans, especially those underwritten using rental rather than borrower income, at higher risk of delinquency and ultimately default than owner-occupied ones. As of the April remittance, 30-day delinquency rates on DSCR investor loans were only 10 bp higher than those of fully underwritten investor loans at 2.8%. However, if the virus does disproportionately impact renters versus owners, DSCR loans could underperform relative to other types of non-QM loans.
Additionally, while early-stage delinquencies are comparable across both full- and limited-documentation loans, limited-documentation loans may ultimately carry more risk. To the extent that underwriting of limited documentation loans vary somewhat significantly across originators of non-QM loans and that the Ability-to- Repay standard remains largely untested, limited documentation loans carry significantly more risk of ATR challenges than fully documented ones. A successful ATR challenge by a borrower would preclude the originator from being able to foreclose on that borrower. Absent the ability to foreclose, originators are left only with the option to modify the borrower. Borrowers who bring successful ATR challenges could have significant leverage over terms of loan modifications. If a pool is subject to a series of successful ATR challenges, it will likely experience more high-touch rate and principal forbearance modifications which could reduce excess spread and extend the duration of the underlying collateral pool.
Additionally, as it pertains to loan modifications, it seems plausible that shelves where the sponsor has a captive originator and servicer like Angel Oak’s AOMT program and Caliber’s COLT program should allow for greater control over special servicing of delinquent loans than other aggregator models, where the sponsor may have been buying pools of loans where the servicing was retained by the originator. So while FICO and LTV have been the better predictor of early stage delinquencies in non-QM pools, occupancy, underwriting and the nature of the sponsor may play a much more meaningful role going forward.