The impact of the COVID-19 pandemic and the related economic crisis on U.S. non-qualified mortgage (non-QM) loans is now clear. Since its inception a few years ago, the performance of non-QM securitizations had been relatively stable, with average 30-plus-day delinquencies in the 5%–10% range. However, the pandemic has led to a spike in forbearances, with 30-plus-day delinquencies rising and averaging 21% as of the July monthly distribution reports (see table 1).

The typical distribution and reporting method for non-agency mortgage securitizations is a lagging indicator based on the prior month. Although non-QM delinquencies declined 1.6 percentage points month over month in July (versus the 4.2 percentage points increase month over month in June), subsequent reporting periods will depend on the ongoing impact of the pandemic. Lenders and borrowers continue to face many pandemic-related hurdles, including the increasing cases of COVID-19 in the U.S., the economic impact of the containment measures, the end of 90-day forbearance plans, and the impact of the July 26 expiration of the extra $600 per week in unemployment compensation distributed under the CARES Act (although legislation is pending that may extend the benefit timeline at lower levels).

Table 1

Delinquencies Declined Slightly In July, But…

Although the pandemic started in the U.S. in March, its first major impact on month-over-month delinquency levels was reflected in the May remittance data (which mostly reflected April borrower activity) due to the lag in delinquency reporting. The rate of performance deterioration softened in June, with 30-plus-day delinquencies declining in July. However, things may not be all that rosy for non-QM loans: the average delinquency data shown in table 1 include some transactions for which deferrals were used in lieu of forbearance, some loans are reported as current even though payments are not being made, thus understating the delinquency figures. However, it seems that most of the reporting is based on whether the borrower has made the monthly payment, regardless of forbearance.

Non-QM borrowers were typically offered a 90-day temporary forbearance plan for COVID-19-related hardship starting in April, and many accepted. However, once these forbearance periods end, beginning in July, servicers will need to focus on loss mitigation efforts. Servicers will have to establish contact with the borrowers, assess their ability to resume making payments, and determine the appropriate solution to bring the loans current. Servicers are considering various options, including a lump-sum payment due at end of the forbearance period, repayment plans, and deferring amounts to the end of the loan term.

Delinquencies By Subsectors

The non-QM market is diverse in terms of underlying loan types (see "Non-QM's Meteoric Rise Is Leading The Private-Label RMBS Comeback," published Sept. 20, 2019). Broadly speaking, there are five main subsectors:

• Loans with prior credit events (PCEs),
• Loans with alternate income documentation (such as bank statements instead of the more conventional tax returns),
• Debt service cover ratio (DSCR) loans (loans underwritten to the cash flows of an investor property),
• Foreign nationals (FN) loans, and
• Interest-only loans and loans with debt-to-income ratios greater than 43% (IO/DTI loans over 43%).

To understand how these subsectors have performed since the start of the pandemic, we conducted a loan-level analysis (using June 2020 remittance data) of 78 non-QM transactions rated by S&P Global Ratings. We examined the five non-QM subsectors, including a range of subtypes (see chart 1). Given the impact of the pandemic on general mortgage performance, these subtypes offer a more granular comparison. We ranked the subtypes by delinquency rate from low to high. We also examined loan characteristics such as original loan-to-value (LTV) ratio and FICO score. Our analysis shows that the relatively worse performing loans are large-balance, alternate documentation, investor "no-ratio," and those that have interest-only features. Loans to borrowers with PCEs and loans with full income documentation performed relatively better.

Chart 1

Alternate income documentation loans

Alternate documentation loans are invariably made to self-employed borrowers. In March, we forecast that government mandated shutdowns due to the COVID-19 pandemic would affect self-employed borrowers disproportionately, causing the non-QM sector, which has significant exposure to alternate documentation loans to self-employed borrowers, to feel credit pressure (see "Credit FAQ: Assessing The Credit Effects Of COVID-19 On U.S.," published March 20, 2020). Indeed, the past two remittance-reporting periods have shown elevated delinquencies in the non-QM sector, particularly in the alternate documentation subsector.

Our analysis shows that alternate income documentation loan performance is worse than average among the non-QM loans we analyzed (see chart 1). We also note that loans with two or more years of documentation are performing worse than those with less than two years of documentation. While a number of factors could be at play, this is likely because loans with less than two years of documentation typically have better FICO scores and lower original LTV ratios--perhaps indicating stronger underwriting requirements to compensate for the shorter documentation history.

PCE loans

Loans to borrowers with a PCE (which typically makes them ineligible for conforming loans) account for less than 2% of the non-QM loans we analyzed. Interestingly, these loans have been showing stronger performance thus far in the pandemic relative to the overall pool of non-QM loans. While not entirely intuitive, especially when considering the borrowers' lower FICO scores, one reason for this could be that these borrowers had to go through a more stringent underwriting process, which is reflected in the better performance of these loans.

It could also be the case that these borrowers are less likely to take advantage of pandemic-related forbearance offers in order to refrain from further blemishing their payment histories (especially after having weathered the qualification process of their existing loan) in order to increase their chances of success at future attempts to refinance into a conforming loan product. We note that self-employed borrowers represent only a fraction of the PCE loan subsector and more than 80% are full income documentation loans. Moreover, the average balance of the PCE loans is lower than the average non-QM loan balance. These attributes could also be contributing to the better performance of PCE loans.

DSCR loans

DSCR loans make up about 20% of the outstanding non-QM loans we analyzed. These are broken into three groups in chart 1: loans with DSCR greater or equal to 1 (66% of the DSCR loans), loans with DSCR between 0 and 1 (29%), and loans underwritten to no DSCR (no-ratio loans; 5%). Not surprisingly, no-ratio loans are the worst performing of the three subgroups, with approximately 20% of those loans at 60 or more days delinquent. When assessing the default likelihood of no-ratio loans, we apply a 6x factor, given our view that they are relatively more risky.

Loans with a DSCR greater than 1, which fully covers principal, interest, taxes, insurance, and association dues (PITIA), have delinquencies in line with the overall pool of non-QM loans, with 16% of those loans 60 or more days delinquent. Relative to loans with DSCRs greater than or equal to 1, loans with DSCR less than 1 have substantially lower FICO scores. This suggests that FICO score may be a less meaningful underwriting threshold for loans underwritten to property cash flows. However, the combination of a lower average FICO score and a lower average DSCR indicates more overall credit pressure. Incidentally, DSCR loans may be less likely to receive full forbearance relief because of the very nature of the loan. For example, if the loan is a sizable cash-out loan or a recently originated business-purpose investor property loan, it may not fall within the same approval parameters for forbearance.

What Are FICO Score And LTV Ratios Telling Us?

Because FICO score and LTV ratio are typically regarded as the most predictive mortgage default modeling variables, we segmented the loans into several buckets based on these variables. Our analysis of 30-plus-day delinquency rates based on June remittance information suggests that performance may be more sensitive to FICO score than LTV ratio during this pandemic, with lower FICO scores and higher LTV ratios correlated with worse performance (see table 2). We also see that mortgage balance can be predictive of performance, with higher balances and lower FICO scores indicative of higher delinquencies (see table 3).

Higher balance loans typically include extra layers of underwriting (e.g., more than one appraisal) and suggest better credit performance. In this case, however, these loans signal harder hit regions and greater financial burdens. Higher loan balances are also associated with so-called jumbo fallouts, which are prevalent in states with expensive metro regions, such as New York City and the Bay Area, which have been seriously affected by the pandemic and the efforts to contain it. These loans also tend to have higher monthly payments and the pandemic-related unemployment relief may be less effective to support the borrowers' higher carrying costs.

It is worth noting that approximately 70% of borrowers with non-QM loans reside in either California (50%), Florida (12%), or New York (8%). These three states have been among those hardest hit by the pandemic and, as of June, two of them have meaningfully higher unemployment rates than the 11% national average at 15.7% in New York and 14.9% in California (the unemployment rate in Florida is 10.4%) (see "Non-QM RMBS And COVID-19: Locking Down States' Exposure," published June 1, 2020). This economic hardship adds to the strain on non-QM borrowers and impacts overall performance.

Table 2

Table 3

Proceeding With Caution

The slowing of the non-QM delinquency rate in July is encouraging news for the market, but risks remain due to the COVID-19 pandemic and its ongoing multifaceted impact across the country. S&P Global Ratings acknowledges a high degree of uncertainty about the evolution of the coronavirus pandemic. The consensus among health experts is that the pandemic may now be at, or near, its peak in some regions but will remain a threat until a vaccine or effective treatment is widely available, which may not occur until the second half of 2021. We are using this assumption in assessing the economic and credit implications associated with the pandemic (see our research here: www.spglobal.com/ratings). As the situation evolves, we will update our assumptions and estimates accordingly.