Since its inception more than seven years ago, the non-qualified mortgage (non-QM) sector has grown in terms of issuance and outstanding, become more widely accepted as a viable asset subclass, and has increased credit availability to borrowers via various loan products. Nevertheless, the non-QM sector remains a small fraction of the overall mortgage originations in the U.S., which remain predominantly within the agency space. As the sudden shock to financial markets caused by the COVID-19 pandemic subsides, S&P Global Ratings is providing a snapshot of the non-QM sector in terms of performance and origination based on loans that are, or have been, in our rated pools. We are also providing our expectations for non-agency RMBS issuance.

Forbearance Levels Subsiding

Across the residential mortgage landscape, forbearance levels have fallen since last summer's peak. This coincides with the rollouts of vaccines and improving economic conditions across the U.S. While subsectors of the residential mortgage-backed securities (RMBS) market, such as credit risk transfer (CRT) and prime jumbo, see 60+ day delinquency/forbearance levels well below 5%, the non-QM sector is still hovering above the 5% mark. This is to be expected, however, given the weaker non-QM loan borrower attributes relative to those in conforming/jumbo prime securitizations. The June remittance data continue to demonstrate a decline in non-QM forbearance/delinquency percentages, whereas prime and CRT have largely flattened. Meanwhile, prepayment rates for the major subsectors appear to have converged, as non-QM speeds have generally increased between August 2020 and May 2021 (see chart 1). Recently, it appears as if the prepayment rates of prime/conforming, CRT, and non-QM are all starting to trend downwards.

Chart 1

Non-QM loans have exhibited a similar performance trend to that seen in the prime/conforming sector over the past 15 months. This includes the resolution of forbearance, in some cases based on deferrals and, in other cases, on a forbearance exit with full reinstatement or a full prepayment. The good news is that losses deriving from deferral activity have been relatively small. We are, however, keeping an eye on the remaining loans in forbearance to assess the resolutions that may arise. While three-month deferrals (assuming a fully amortizing 30-year fixed rate mortgage with a 5.5% coupon) result in securitization losses of 1.7% of the loan balance, an 18-month deferral plan would be roughly 10% of the loan balance. Under the scenario of 10% losses and a 5% forbearance level, assuming all loans end up deferring and were all 18-month plans, this corresponds to a loss rate of 0.50%, which remains inside the average non-QM 'B' credit enhancement level. For context, excess spread is roughly 2% annualized on average, which would cushion the 0.50% figure.

Prepayment Speeds Relatively High

On a credit-positive note, an uptick in prepayment speeds can be viewed as healthy from a deleveraging perspective (notwithstanding reduced excess interest for non-QM securitizations). The increase in non-QM prepayment speeds is likely a result of one or more of the following factors:

• A portion of non-QM refinance activity is made up of borrowers that stay within the non-QM space by taking out a new non-QM loan, presumably at a more favorable interest rate. With non-QM origination activity picking up, there are likely more refinancing options (assuming the agency channel can't be accessed).
• After the conforming 30-year fixed-rate mortgage hovered at or below 3% for more than a year, one expects a degree of refinance burnout on the agency side. This has caused lenders to refocus on non-QM refinancing.
• The combination of (1) seasoning away from forbearance plans by non-QM borrowers and (2) home-price appreciation leading to a drop in current loan-to-value (LTV) ratios has opened the agency refinance option for some non-QM borrowers.

Chart 1 indicates that prepayment speeds are relatively high across all three major non-agency sectors (prime/conforming, CRT, and non-QM). The recent downward trend could point to early stages of refinance burnout, especially in the prime/conforming and CRT sectors. However, it is more likely due to the inherent volatility of prepayment data when measured monthly. Clean-up calls of past non-QM transactions may also be contributing to recent drops in these aggregated prepayment rates in the non-QM sector.

Foreclosure Sale Exposure

While the delinquency/forbearance levels show a positive trend reverting to pre-pandemic levels, the question remains: What will happen to the population still in forbearance plans? We expect forthcoming forbearance resolutions to be generally more adverse than those that have already taken place. That said, strong home price gains across the country (driven by a variety of factors, including limited new home supply, de-urbanization, and low mortgage rates) suggest that some resolutions will result in natural (as opposed to forced) home sales with minimal losses. Loss exposure for the population that does go into foreclosure might also be a function of regional factors. In "Non-QM RMBS And COVID-19: Locking Down States' Exposure," published June 1, 2020, we analyzed the top geographic concentrations in our (currently or formerly) rated non-QM securitizations and determined that 70% of non-QM representation is in three states: California, New York, and Florida.

While these three states varied somewhat in terms of how their governments responded to COVID-19 and the related economic impacts over the past 15 months, they experienced comparably strong home price gains (14%, 12%, and 12%, respectively) between first-quarter 2020 and first quarter 2021, according to the Federal Housing Finance Authority (FHFA) seasonally adjusted purchase-only state house price index. Furthermore, our view regarding over- and under-valuation (as measured by affordability relative to historical norms) indicates that California and New York are undervalued, while Florida is slightly overvalued. However, population migration into Florida continues to be strong, thus creating demand for housing. The strong house-price gains in these states, along with relative under-valuation in California and New York and population migration-driven housing demand in Florida, could also temper losses due to foreclosures.

Current Non-QM Pool Characteristics

The average FICO score and the average LTV ratio in the non-QM sector hover near 700 and 70%, respectively. Recent securitizations indicate that overall weighted leverage has remained steady while credit scores have ticked up, pointing to some tightening in lending standards. To understand how non-QM loan origination borrower characteristics have changed over time, we stratified our (currently or formerly) rated non-QM loan portfolio by origination date and segmented the data into three vintages, as shown in the table below. The vintages are segmented by originations completed prior to the pandemic, from 2017 to February 2020 (two segments), and originations made during the pandemic, from March 2020 through June 2021.

Prior to the start of the pandemic in March 2020, loan documentation in the non-QM sector could be broken out (using origination balances) roughly as follows: 35% using full income documentation, 45% using alternate income documentation, and about 20% underwritten to the debt service coverage ratio (DSCR). Recent originations show that DSCR loans make up close to 30% of the population, while there has been a corresponding decrease in full income documentation loans, down to roughly 25%.

Surprisingly, alternative income documentation loans (such as bank/profit and loss [P&L] statement, CPA letter loans) have remained steady over time. One might have expected this share to decline because: (1) the economic impacts of the pandemic resulted in revenue pressure for self-employed borrowers, and (2) alternative income documentation loans exhibited more adverse performance in the early stages of the pandemic (see "Non-Qualified Mortgage Loans Summertime Blues Continue Despite Improved July Delinquencies," July 29, 2020). Instead, the early government stimulus packages helped consumers and economic conditions have improved.

While loan balances have crept up since March 2020, perhaps in line with rising house prices, the consumer debt burden has not risen. Borrower debt-to-income (DTI) ratios remain near 35%, likely kept at bay with the help of low mortgage rates, which now are 100-150 basis points (bps) lower than they were between 2017 and 2019. This is clear from the change in weighted average coupon (WAC) shown in the above table. Non-QM mortgage rates tend to be slower to respond to movements in broader market rates than those in the prime/conforming space. Nevertheless, we have observed that non-QM loans have sustained an approximate 250 bps spread over the corresponding Freddie Mac survey rate (see "Factors Affecting Non-QM Mortgage Interest Rate Spreads," Feb. 20, 2020).

The current low-rate environment has encouraged a shift toward the fixed-rate product. Whereas prior to March 2020, the non-QM sector was biased toward adjustable-rate mortgages, borrowers have since favored locking in a rate for the long term. We also noticed some tightening in credit standards with a significant reduction in cash-out refinance loans in the post-February 2020 originations. The geographic footprint has not changed much, as California, New York, and Florida continue to be the three states with most non-QM originations in our (currently or formerly) rated pools. We examined the subset of alternative income documentation loans and DSCR loans separately and noticed similar trends.

Impact Of The New QM Rule

The General QM final rule, which amends the original QM Rule, is in its optional adoption period, with mandatory adoption more than one year out. The new QM rule reflects two main changes to the current rule: (1) the move away from a DTI ratio threshold to a spread over the average prime offer rate (APOR), which designates a loan as QM (specifically, a spread less than 225 bps), and (2) the removal of Appendix Q, which is the current template for calculating income and debt. Under the new QM rule, loans that do not meet Appendix Q standards for income and liability assessment could potentially attain QM safe harbor or rebuttable presumption status. Roughly two-thirds of the loans in our (currently or formerly) rated securitizations that fall within a non-QM designation solely due to alternative income documentation, other Appendix-Q fall-outs, and/or a DTI ratio over 43% have spreads over APOR of less than 225 bps and, therefore, could potentially be classified as QM.

Although the mandatory adoption date of the new rule has been pushed out to Oct. 1, 2022, originators have the option to adopt and apply the new rule to underwrite loans. It is possible that loans to borrowers with income verified using alternate documentation (such as bank/P&L statements) could be considered QM loans. However, irrespective of the Truth In Lending Act designation and whether the loan remains non-QM or becomes QM, our assessment of loans underwritten using alternative income documentation indicates weaker performance compared to loans that rely on traditional income documentation verification, such as Appendix Q standards or the methods that would be applicable (e.g., Fannie Mae and Freddie Mac guidelines) for attaining QM status under the new rule. We would therefore continue to apply our adjustment factor for alternative income documentation, which increases the likelihood of default by 1.75x-2.25x, depending on the duration of the income documentation. (See "Credit FAQ: How The New Qualified Mortgage Rule Could Impact U.S. RMBS," March 1, 2020.)

Non-QM Securitizations Set To Gear Up

Once the pandemic-related economic disruption started last year, non-QM origination hit the brakes as credit became tighter. Although there was securitization of non-QM loan collateral, it mainly comprised pre-COVID originations and re-securitizations of clean-up calls. As a result, 2020 non-QM securitization issuance slowed and finished the year at $20 billion (roughly half of what we were projecting at the start of 2020) and origination activity in 2021 has also been tepid (see chart 2). However, we expect the economic recovery currently underway to result in a boost to non-QM originations in the second half of the year.

Chart 2

We examined the set of non-QM loans within our (currently or formerly) rated transactions and found that the number of unique loan originators selling into securitization grew rapidly between 2016 and 2019. However, it fell sharply in 2020 and has not recovered fully, which points either to consolidation in the industry or to the possibility that smaller originators continue to avoid originating non-QM loans.

The low interest rate environment has contributed to the smaller number of non-QM originators during the past 12 months, as originators have turned their attention to the refinancing boom playing out in the prime/conforming loan sector. As economic conditions improve, the refinancing boom abates, and strong demand for residential housing continues, we expect many originators to resume activity in the non-QM space and that non-QM volume will increase through the remainder of the year. We project non-QM issuance to be around $25 billion in 2021 and overall non-agency RMBS of approximately $150 billion for the year.