Debt service coverage ratio (DSCR) loans are mortgages on investor properties that are underwritten primarily based on the cash flow from the expected rental income. This market segment makes up roughly half (by balance) of the mortgages backing the non-qualified mortgage (non-QM) securitizations rated by S&P Global Ratings between July 2022 and July 2024. DSCR loans have traditionally been associated with long-term (12-month) leases. However, the growing prevalence of Airbnb-style investments has led to increased origination and securitization of short-term rental (STR) DSCR loans, which we also see in non-QM pools, though they make up a relatively small portion (less than 10%).

STR DSCR loans have distinct characteristics that differentiate them from typical long-term rental (LTR) DSCR loans. STR debt service coverage (DSC), which is generally derived from annual revenue (adjusted for vacancies and certain haircuts), is used for underwriting nightly, weekly, or monthly rental properties. However, LTR DSC predominately depends on annual rental revenue. In addition to DSC-related underwriting differences, STR DSCR loans differ from LTR DSCR loans in terms of the number of originators/aggregators that finance the loans, and idiosyncratic factors that could affect how STR DSCR loans perform relative to LTR DSCR loans. This article examines these and other differences between LTR and STR loans. All analyses herein are based on transactions rated by S&P Global Ratings.

The Numbers Behind LTR And STR Loans

To better understand the differences between LTR and STR loans, we disaggregated our pools, bucketing and stratifying them by various loan characteristics. Focusing on securitizations for which both STR and LTR DSCR mortgages were identified in the pool, we assembled a sample of over 2,100 STR DSCR loans (from over 125 unique originators) and over 23,000 LTR DSCR loans (from over 550 unique originators) securing 37 non-QM transactions (issued between July 2022 and July 2024). For comparison we also considered over 19,000 full/alternate/other income documentation loans securing 41 non-QM transactions (issued between July 2022 and July 2024). Approximately 10% of the loans within these 41 non-QM pools are QM loans. Although STR loans are a relatively small share of the securitized DSCR universe in our sample, certain pools could have outsized exposures (see table 1).

Table 1 shows that relative to LTR loans, STR loans have higher loan-to-value (LTV) ratios and higher FICO scores at the pool level. The average STR loan balance across pools is almost 60% greater than that of LTR loans while STR property values are, on average, 50% higher than those of LTRs. Because the properties are primarily single-family homes or condos, the appraised values of the properties should reflect a sales-based comparable approach as opposed to an income/cap rate valuation. This is evident in both the underwriting and the predominant use of the standard residential appraisal Form 1004.

Table 1

In addition to having higher balances and LTV ratios on average, table 1 shows that the weighted average DSCR for STR loans (1.35x) is higher than that of LTR loans (1.11x). This means STR loans have higher rental yields (see "The Who And The How" section below).

Chart 1 shows the average fixed interest rates of LTR, STR, and other loans in our sample originated between September 2021 and June 2024. The chart also shows the conforming fixed rate (according to the weekly Freddie Mac survey). The earlier monthly data in our sample suggested a greater spread to the conforming rate for STR loans compared to LTR loans, which has since largely converged. This convergence might have been partly due to expansion of the STR market, which led to increased liquidity. Moreover, the growing use of securitizations to fund STR loans may have helped lower the spread by effectively widening the investor base. In any case, DSCR loans (as well as non-QM loans in general) appear to share a spread of roughly 100 basis points (bps)-150 bps over the conforming rate, down from over 200 bps at various points in the past.

Chart 1

Where STR Loans Are Concentrated

We'd generally expect STR property locations to overlap with tourist destinations. Frequent special events (concerts, festivals, sports, etc.) in a region could also boost the number of STR properties. That said, STR properties are often subject to local laws and regulations that restrict terms of use. These constraints are more nuanced than those governing standard LTR properties, which have been around for many decades. When examining the geographic concentrations of underlying properties backing the roughly 2,100 STR loans in our data set, we saw clustering in areas that don't typically make up large shares of standard non-QM or even LTR DSCR loan pools. Chart 2 ranks states by concentrations of STR and LTR loans.

Chart 2

Florida properties contribute the largest share of DSCR loans. While these include both LTR and STR properties, there's an outsized share of STRs, with Florida properties constituting roughly 26% of the STR loans in the sample, whereas Florida's LTR loans are less than 20% of all the LTR loans in the sample. Tennessee also has a relatively large share of STR loans. Almost 10% of the STRs are in Tennessee, whereas less than 2% of the LTRs are from that state.

For a more granular examination, we considered concentration of STRs within core-based statistical areas (CBSAs). The heat map in chart 3 shows that CBSAs in Tennessee have the highest density of STRs. On the other hand, New York CBSAs show a relatively low density of STRs despite the state's numerous vacation destinations. This low figure could stem partly from New York's short-term rental regulations, which may be more restrictive than those of other states.

Chart 3

Most states include at least a minimal presence of STRs, with only North Dakota, South Dakota, Alaska, and Nebraska without STR representation. However, our sample isn't without potential bias, as certain areas with elevated concentrations in STRs could be overrepresented due to specific originators being especially active in those areas. To put the geographic representation of STRs in chart 3 into perspective with LTRs, we ordered the top 10 CBSAs in our sample by percent of their respective STR and LTR cohorts (see tables 2A and 2B). We thought it was interesting that Sevierville, Tenn., which has the greatest density of STRs, doesn't even make the top-10 LTR list. Sevierville has been periodically highlighted as a vacation destination, with various natural attractions and music-related entertainment options. With its proximity by car to various surrounding areas (it sits in the southeast corridor of the Midwest), this is an ideal location for an STR property. Myrtle Beach, S.C., is another location that ranks high for STR presence but is absent from the LTR list.

Table 2A

Table 2B

The ability to attract visitors and tourists supports investment in STR properties. Indeed, interest from visitors outside the region is essential to this sub-industry and distinguishes it from LTR (traditional income properties). Therefore, in the event that demand for STR weakens, it's an interesting exercise to compare the average property prices in STR-dense areas to local incomes per capita. Tables 2A and 2B show that Myrtle Beach has a property price-to-income of about 17x and Destin, Fla., is almost 32x, while Sevierville, Tenn., leads the pack at 36x. Of course, these numbers are subject to small-sample bias, but STR property prices are consistently higher than those of LTRs. Note that rents and rental yields in tables 2A and 2B are derived from loan-level DSCR data, assuming a 2% combined rate of tax and insurance.

The Who And The How

Roughly 16% of DSCR originators originate/finance STRs, though underwriting requirements vary among this group. Some originators may require a 12-month trailing evidence of cash flow or a Form 1007 rent schedule, while others may rely on something like an AirDNA report. Consistent among these originators is that the income used in the DSCR calculation for STRs is subject to a greater number of adjustments (for expenses described below) than for LTRs. While the presence of STR DSCR loans in securitized portfolios has lagged that of LTR DSCR loans, it has become a more noticeable sub-product in recent years. Based on our STR population, the number of unique originators more than doubled from about 50 in 2021-2022 to over 100 originators in 2023-2024. The underwriting for STR and LTR is basically the same in terms of the mathematical approach used to calculate the ratio.

While there may be other limitations (such as no financing on more than four units/mixed-use or a haircut to the maximum permitted LTV ratio compared to LTR loans), the main difference between the two is the rental income used in the DSCR numerator. STR rental income is subject to a haircut of approximately 20% to account for recurring expenses--such as cleaning, marketing, and third-party fees (some originators require a higher haircut if the level of expenses are higher than usual). This practice is not commonly used to calculate LTR income.

Rental yields typically exhibit an inverse relationship to property values (see "U.S. RMBS: A Closer Look At DSCR Loans," April 6, 2023). This relationship is illustrated in chart 4, where we break out rental yields for STRs and LTRs by property value. To compute rent for the purposes of calculating the yield, we back-out the principal, interest, taxes, and insurance (PITI) and original loan balance components from the DSCR to solve for a dollar amount of monthly rent. This is then annualized and compared to the original appraised value. The rental yield for STR properties shows the same declining trend by property value as LTRs, but the yield for STRs is substantially higher than that of LTR properties.

Chart 4

It's not surprising that STR properties command a higher yield. They have greater recurring expenses relative to LTRs, greater occupancy volatility, and exposure to seasonality and weather-related events (for example, a serious hurricane affecting Florida could deter tourists for an entire season). Moreover, the STR to LTR rental yield spread also accounts for additional work and consistent sweat equity, regulatory risk of STR requirements/protocols within municipalities, and different insurance requirements for STR compared to LTR.

The rental income that we backed out of the DSCR embeds the average 20% haircut for DSCR we commonly observe. We also considered a hypothetical haircut of 40%, which turns out to be roughly the amount that is needed for the LTR and STR rental yields to coincide. However, one expects the STR rental yield to lie above that of LTR such that the spread represents a risk premium due to the risks inherent to STR. Therefore, chart 4 provides approximate bounds for the haircut (somewhere in the range of 20% to 40%).

There hasn't appeared to be any market consensus as to credit quality differentiation of STR and LTR DSCR loans regarding collateral backing a securitization. On one hand, STR cash-flow volatility could be greater; on the other, tenant default is less impactful, as it doesn't introduce blackout periods and expenses related to tenant eviction. Also, STR properties might tend to be better maintained so that they remain attractive to new clients. While over the long term, LTRs are less volatile from an occupancy perspective, supply and demand for vacation style rentals could be more impactful for STRs. As such, municipal-level provisions/laws related to short-term rentals are influential. Perhaps the question is whether the underwriting haircut on STR loans adequately captures associated expenses, resulting in the appropriate risk-adjusted rental yield (see chart 4). In any case, developing local regulations pertaining to STRs--and changes in consumer behavior and sentiment--will ultimately determine the course of STR DSCR and its representation in non-QM pools.

Related Research

• U.S. RMBS: A Closer Look At DSCR Loans, April 6, 2023
• Investor Property DSCR Loans: The Nonqualified Mortgage Exempt From Qualified Mortgage Rules, Aug. 27, 2019

We would like to thank Khilti Shah for her contributions to this report.