The non-qualified mortgage (non-QM) segment of the private label residential mortgage-backed securities (RMBS) space has evolved since its inception roughly a decade ago. One of the main developments has been the increase in debt service coverage ratio (DSCR) loans, for which the underwriting is based on expected rental cashflows (akin to a commercial mortgage). The alternative documentation (alt-doc) loan (e.g., bank statement instead of W-2/tax returns) used to be the more common non-QM loan type before DSCR grew in popularity. DSCR loans now make up over half the non-QM securitized loan population, with the remainder mostly comprising alt-doc loans.

Despite a relatively buoyed housing market, delinquency levels among securitized non-QM loans have risen for both the DSCR (including no-ratio zero DSCR) and non-DSCR (i.e., alt-doc and other non-QM loans) groups over the past two years. While the increase in 60+ day delinquencies is likely attributable to factors such as inflation, elevated mortgage rates, and other consumer affordability issues, not all borrowers are affected in the same way. In this paper, S&P Global Ratings examines the performance impact of several borrower attributes (including loan size, credit score, leverage, and geography) to better understand how this performance trend is playing out in the non-QM and DSCR markets.

Non-QM And DSCR Delinquencies Rise

DSCR loan performance has generally been consistent with that of the broader non-QM universe. While this may seem counterintuitive (as an investor may have only a limited attachment to a dwelling that is being rented to a tenant), housing fundamentals have supported non-owner-occupied homes in the single-family space with a sticky U.S. rental vacancy rate of about 6%. Also, DSCR loans typically have lower loan-to-value (LTV) ratios than alt-doc loans and comparable or higher FICO scores.

Chart 1 shows that non-QM loans, including those in the DSCR segment, have seen an approximate doubling of delinquency levels over the past two years. In our article, "2025 U.S. Residential Mortgage And Housing Outlook," published Dec. 16, 2024, we showed that certain loan characteristics tended to be overweight in delinquency buckets (e.g., cashout loan purpose and adjustable-rate-mortgage [ARM] loans). This is likely a result of ARM loan interest rates resetting in a higher rate environment. Higher delinquency rates for cash-out refinancings (some of which overlap with bank statement loans) in a higher interest rate environment could be the result of liquidity needs—either personal/housing expenses and/or self-employed business expenses.

Chart 1 also shows that delinquencies for prime/jumbo loans and credit risk transfer (CRT) pools (backed by agency collateral) have had generally stable performance over the same period, suggesting that, for now, the performance story is confined to non-QM and DSCR borrowers. This isn't unusual given that loan features differ from traditional loans associated with CRT and prime mortgages. To understand how certain loan characteristics may render a borrower more likely to become delinquent, we have analyzed our rated non-QM/DSCR securitizations (a population of over 150,000 loans with Truth in Lending Act designation of non-QM or DSCR) based primarily on January 2025 distribution data. While recent events, such as wildfires, could influence present impairment levels, our intent is to identify factors that may be contributing to the rise in impairments (i.e., 60+ days delinquent, foreclosure, and REO) over the last two years. (Note that while the DSCR designation typically falls under the broader non-QM grouping, unless indicated otherwise, we have separated them for the purpose of our analysis.)

Chart 1

DSCR And Non-QM Vintages Perform Similarly

It is instructive to consider how the different loan vintages have performed for both DSCR and non-DSCR segments. Different securitized vintages may show delinquency rates that vary due to declining pool factors and the extent of optional redemptions at the call dates. Chart 2 plots recent 60+ day delinquencies for quarterly vintage originations since early 2020 for various loan-level bins. (We excluded vintages prior to third-quarter 2020 due to lower loan counts.) Overall, non-QM and DSCR loans broadly have similar rates of 60+ day delinquencies, with some notable exceptions in some quarters.

Certain quarterly vintages between second-quarter 2022 and first-quarter 2023 appear to have higher delinquency rates, while the recently originated 2024 vintage lacks sufficient seasoning to show significant delinquencies. Chart 2 also shows that the weighted average original combined loan-to-value (OCLTV) ratio for both non-QM and DSCR loans has remained roughly constant over the past five years, while original weighted average coupon (WAC) started to rise in 2022 alongside the general rise in interest rates. However, the WAC spread between the DSCR and non-QM loan rates has decreased, perhaps driven by the market's general acceptance of the DSCR product. In terms of the actual DSCR, the weighted average coverage ratio of DSCR loans has declined slightly, which could be attributed to higher loan balances (driven by home-price appreciation) and higher interest rates. Although a portfolio may comprise a mixture of loans with lower/higher LTV ratios (a "barbelled" portfolio) non-QM and DSCR cohorts continue to demonstrate healthy origination LTV ratios, delinquencies notwithstanding, which appears to be offsetting losses at default (which we will cover further below).

Chart 2

The Coverage Ratio's Effect On DSCR Loan Performance

Except for underwriting parameters, which may include an assessment of the borrower's ability to manage a non-owner-occupied rental, DSCR loan underwriting relies heavily on the FICO score of the borrower, the LTV ratio, and the projected cash flow of the property. The more conventional non-owner occupied loan products instead assess the personal income and liabilities of the borrower, with positive or negative net rental income adding to the income or debt. For this reason, the DSCR metric might be expected to have a strong influence on borrower behavior because the cash flow stream from the property may be one of the main (or only) sources of income for loan repayment. On the other hand, cash flow from the property may contribute little to borrower performance if there exist additional sources of income, such as a crossed portfolio of other rental homes.

To assess the impact of the DSCR on the observed impairment levels for the cohort, we tabulated 60+ day delinquency rates for buckets of DSCR loans and varied the LTV ratio and, separately, the FICO score. Table 1 shows that, when broken out by LTV ratio, delinquencies are generally higher for loans with a DSCR less than 0.70 (note that no-ratio loans make up only 4.0% of the population of approximately 95,000 loans); however, above a DSCR of 0.70 there is no clear pattern. Meanwhile, there doesn't appear to be a strong relationship between observed impairments and LTV ratio.

When broken out by credit score, the relationship between DSCR and delinquency is not readily apparent within a given FICO score band. However, there is a clear stratification, with lower scores (especially under 700) showing greater impairment rates. This means that, despite DSCR loans being considered business purpose loans (and in some cases closing in the name of a company/entity as opposed to that of a borrower), it seems the borrower/guarantor's credit score is still influencing performance. This is particularly interesting in that reporting to credit bureaus on a loan/borrower/guarantor's underperformance may not be carried out to the same degree for DSCR loans as it is in the case of typical residential mortgages.

Table 1

Impairment Levels By Loan Size

To examine the impact of loan size on performance, we stratified our sample by FICO bucket for both non-QM and DSCR loans. Table 2 shows that not only do loans with lower FICO scores perform worse, but impairment levels tend to be higher for larger loan sizes for both non-QM and DSCR loans.

Table 2

Across all FICO buckets in the non-QM group, the $200,000-$300,000 balance bin had an impairment rate of 2.4%, while the $700,000-$1,000,000 bin had a rate of 3.7%. The corresponding numbers for the DSCR loans were 2.9% and 4.3%. Overall, the trend appears to exhibit monotonicity across these balance ranges and FICO buckets; however, this breaks down at the low and high ends of the spectrum. In the case of high balance loans, the decrease in delinquency rate for DSCR may be attributable to sophisticated borrowers and property management and/or sample size related variances. Increased rates of delinquency among borrowers of less than $100,000 may be due to idiosyncratic factors such as properties with lower values, which may be associated with cross-collateralized loans and/or multiple loans to a single borrower. In the aggregate, the 60+ day impairment rate was about 3.5% by loan balance, for both non-QM and DSCR loans, and 3.2% and 3.0% by loan count, respectively. (For bins in Table 2, loans with FICO scores less than 650 have weighted average scores of 613 and 621 for non-QM and DSCR, respectively. Correspondingly, loans with FICO scores greater than 775 have weighted average scores of 793 and 792 for non-QM and DSCR, respectively.)

Unlike prime/jumbo securitizations (which tend to comprise loans with large balances) or GSE CRT issuances (which are subject to the conforming loan limit), non-QM and DSCR loans exhibit a wide range of loan balances. As such, weaker performance among loans with higher balances naturally increases the delinquency percentage by pool balance compared to a simple loan count impairment rate. For the non-QM group, larger loan balances could be attributable to alt-doc borrowers that may not meet conforming loan requirements or qualify for a jumbo loan program. Our prior analysis (see "2025 U.S. Mortgage and Housing Outlook") showed that delinquent populations had a greater share of alt-doc loans than the "current" (non-delinquent) population. This is an indication that the alt-doc borrower (who would typically be self-employed) may be more dependent on non-conventional income, which can be more variable than a W-2 borrower with regular income or a traditional prime self-employed borrower qualifying with tax returns. Average balances of non-QM and DSCR loans are about $560,000 and $280,000, respectively, which could be affecting the impairment rates at the cohort level.

In terms of location, non-QM loan pools are geographically concentrated in California, New York, Texas, and Florida, as expected given the higher populations of these states. The correlation between performance and balance could suggest that regions with higher priced homes may have higher impairment rates. That said, while there is some regional variation, it wasn't apparent that higher-cost areas consistently have greater impairment levels than areas in the country with relatively modest property prices.

Impairments Are Rising, But Some Impaired Loans Are Cash-Flowing

To further compare performance of non-QM to DSCR loans and better assess how delinquent loans have historically transitioned, we tabulated transition rates over three six-month intervals (see table 3). We aimed to identify the portion of delinquent loans that are cash-flowing (possibly with a rolling delinquency) compared to the portion transitioning to a worse payment status. Our assessment examines three time horizons by marking the July remittance status of a loan present at that time (e.g., current, 30-day delinquent, etc.) and then looking at the loan status in the following January.

For loans identified as 30-days delinquent, roughly half were current or had prepaid six months later, about one-quarter of the subset remained 30-days delinquent (although they may have rolled in and out during the interval), and the remaining 25%-30% of loans rolled to a worse status. Although not a direct month-over-month transition assessment, this suggests that about one quarter of 30-day delinquent loans may be rolling, and the single delinquent payment has not cured. For 60-day impairments, the six-month transition to a better payment status or prepayment was comparable to the 30-day segment, but the rate of worsening was about 25%-40%. For the assessed population, we identified the following interesting observations.

• The impairment rate for loans moving from current to delinquent over a six-month period increased to near 3% in the second half of 2023 and 2024 from roughly 2% in second-half 2022. In this current-to-delinquent transition, there is little performance difference between non-QM and DSCR loans.
• The rate at which 30-day delinquent loans move to a worse delinquency status over a six-month period increased for the latter halves of 2023 and 2024 compared to second-half 2022. Since 2023, both segments had worsening rates close to 30%, compared to around 20% in 2022.
• The rate at which loans transitioned from 90-days delinquent to a better status over six months is higher for non-QM loans than DSCR loans for each period considered. The "better" rate from foreclosure is also considerably superior for non-QM loans for the 2023 and 2024 second-half vintages. We attribute some of these observations to loss mitigation and servicing approaches related to consumer residential mortgages compared to those applicable to business-purpose DSCR loans.

The transition table shows that a sizable fraction of loans prepay out of a 30- or 60-day delinquent status. This is probably because of built-up equity in a strong housing market, but, should home prices fall, borrowers might be more likely to transition into foreclosure and potentially suffer a liquidation loss. Chart 2 shows that LTV ratios continue to demonstrate sticky averages close to 70%; and correspondingly low cumulative loss rates are likely a result of such LTV ratios and home equity that reduces losses upon sale of the property following borrower default.

How Are Impairments Translating To Losses?

To further assess the extent of losses given default, and in turn loss severity, we identified (within the CoreLogic Loan Performance database) securitized pools labeled non-QM (which also contained DSCR pools) that were issued in 2019 and later and had loans with a payment status of 90+ days delinquent (including foreclosure and REO) at loan termination (i.e., when the balance went to zero). The data showed that most loans that terminated (i.e., the loan balance went to zero) from a 90+ day delinquency status (as last payment status before going to zero) had negligible or no loss severity, which is further corroborated by the low cumulative losses in the non-QM and DSCR cohorts.

The low observed loss rate is likely because the borrower (despite presenting a potential loss of income/cash flow to service the mortgage payment) was able to sell the property, extract equity, and satisfy the loan obligation. This hypothesis is supported by the relatively low original LTV ratios (close to 70%) characteristic of these loans, as well as the relatively strong pace of home-price appreciation. Table 4 shows (by loan liquidation year) statistics for loans in the population that had a last payment status before loan termination (zero balance) of 90+ days delinquent, foreclosure, and REO status loans. Balances less than $5,000 were excluded, as were loss amounts less than $5,000. Loss severities less than 1% were also assumed to be zero, and we excluded loans originated prior to 2014.

Table 4

While average observed loss severities have been low--particularly in the 90+ day delinquent and foreclosure groups--the share of terminations from foreclosure and REO have been increasing. In part, this is because a foreclosure/REO settlement takes time to complete (and the cohort has seasoned over the years). It may also indicate that less seasoned loans may not have the same embedded home price appreciation as older loans. However, the ultimate resolution depends on whether the original LTV ratios are sufficient to mitigate losses and whether the housing market holds up. In any case, housing stock nationally continues to be supply constrained (due in large part to mortgagors' reluctance to sell homes and give up historically low fixed rates), which has been a factor in preventing price declines at the national level. At the regional level, home prices could exhibit varying degrees of volatility due to local economic forces and housing market fundamentals.

Navigating The Credit Landscape

Despite a strong housing market that has supported mortgage performance for more than a decade, inflation and higher interest rates have resulted in affordability issues for many U.S. consumers. Even as existing homeowners see increasing wealth with rising home prices, the corresponding increase in property taxes and insurance premiums must be financially managed. The impacts are uneven, however, and some borrowers are affected more than others. Although spreads on non-QM mortgages (including DSCR loans) have contracted over the past decade, non-QM rates remain higher than those of conventional mortgages. This makes economic sense given that non-QM delinquency levels continue to exceed those in the conventional space, partly due to credit attributes that deviate from those of traditional prime mortgages despite similar FICO scores and LTV ratios.

Based on our credit analysis, non-traditional income verification and reliance primarily on rental income for DSCR loans results in higher loss projections relative to traditional mortgages. Indeed, this has been observed in that CRT and prime/jumbo pools have seen better and more stable performance, while the non-QM and DSCR cohorts have 60+ day delinquency rates at about twice the level of two years ago.

While the future of the economic landscape is uncertain, limited economic growth and rising inflation could have a strong influence on borrower behavior. As described in "Servicer Evaluation Spotlight Report: Servicers' Collection Skills Essential For Managing Delinquent Loans In A New Era," published Feb. 24, 2025, the current consumer woes are distinct from those during the pandemic. At that time, impairment/forbearance rates were directly tied to transient health-related factors, such that temporary forbearance and workouts were more systematically implementable. The current performance impairments appear to be more borrower specific. Moreover, since student loan forbearance has ended, many consumers may have additional monthly payment obligations.

While the unemployment rate currently sits at 4.2%, which is low by historical standards, our baseline forecast sees this number rise to 4.5% by the end of 2025 (see "Economic Research: Economic Outlook U.S. Q2 2025: Losing Steam Amid Shifting Policies," published March 25, 2025). Although a small increase, such an elevation in unemployment could contribute to an increase in delinquencies in both the non-DSCR space (directly, if a borrower loses a primary income stream) and in the DSCR space (indirectly, if a tenant loses a job and cannot make rental payments). Ultimately, the extent of impairments transitioning into losses will depend on the degree of home equity buildup, which has so far likely offset losses with an average severity of 1.5%. Assuming OLTV ratios remain steady (as they have been historically in the asset class) and home prices remain buoyed, losses should remain contained.