Middle-market (MM) collateralized loan obligation (CLO) transactions were in the spotlight in 2024 as the investor base broadened, new managers entered the space and the topic (along with private credit more broadly) came up in almost every conversation. By the end of third-quarter 2024, per Pitchbook LCD, MM CLO new issuance volume had already surpassed full-year 2023 volume ($27.10 billion), which previously held the record for highest annual issuance. The volume of S&P Global Ratings-rated transactions grew in 2024 as well, and by the end of the third quarter, we had 256 MM CLOs outstanding containing 2,945 credit estimates (see slide 11 in "Private Credit And Middle-Market CLO Quarterly: The Times They Are A-Changin'," published Oct. 18, 2024).

Investor demand for MM CLOs has been driven by additional spread on offer compared to like-rated broadly syndicated loan (BSL) CLO tranches (although this has narrowed significantly over the past year), and also by MM CLOs' strong performance during the pandemic and since. MM CLO ratings have shown impressive resilience, with (at time of publication) only eight tranche ratings lowered since the start of 2020, including the pandemic--less than 1% of the total ratings outstanding. In part, this is due to structural features found in these CLOs, including higher average par subordination levels compared to BSL CLOs, and larger rating cushions seen in many MM CLOs. This leaves more room before performance deterioration causes a given MM CLO tranche to see its rating negatively affected.

However, higher-for-longer interest rates over the past couple of years have put pressure on leveraged companies in both the BSL and direct lending markets, and credit estimate downgrades on companies in MM CLO collateral pools have far outpaced upgrades over the period. While ratings on the MM CLOs themselves have proven resilient so far, they're not immune to negative rating actions arising from par loss and failing overcollateralization (O/C) ratio tests. While a large majority of our MM CLOs have stable ratings and we expect few downgrades in the coming year, this, combined with the overall growth in the space, makes it an interesting time to update our annual MM CLO rating stress scenarios.

Recap Of Some Differences Between Middle-Market And BSL CLOs

MM CLOs are a segment of the U.S. CLO market, representing just over 12% of the overall U.S. CLO market by par value of assets and slightly less by transaction count. While we use the same criteria to rate MM CLOs and BSL CLOs, they differ in some material ways.

Many of the companies that issue the loans in MM CLO collateral pools aren't publicly rated, and instead have credit estimates assigned for purposes of the CLO analysis (see "Anatomy of a Credit Estimate: What It Means And How We Do It," published Jan. 14, 2021). MM CLOs are backed primarily by senior secured loans to smaller companies, typically those with EBITDA of $100 million or lower (and often much lower; the median EBITDA for obligors in our credit estimate universe is about $35 million). Some MM CLO transactions also include loans from larger unrated companies and have been labelled as private credit CLOs by market participants, although for the purposes of this article, we include both under the MM CLO moniker. The proportion of loans from credit-estimated versus rated companies varies significantly from one MM CLO to another, but all MM CLOs have a much higher proportion of assets from credit-estimated obligors than BSL CLOs.

Because companies in the MM CLOs are smaller than those in BSL CLOs, MM CLOs have more exposure to loans from lower-rated (or credit estimated) obligors. As of September 2024, the average U.S. BSL CLO exposure to loans from companies rated 'B-' and in the 'CCC' range was about 25.2% and 6.5% of total assets, respectively; the proportion for MM CLOs was much higher at about 72.4% and 15.3% (note that the figures for MM CLOs include loans with either a credit estimate or a rating). To compensate for this, MM CLOs tend to have more par subordination for their rated tranches than like-rated BSL CLO tranches.

In addition to having a lower-quality credit profile, most MM CLO portfolios tend to have lower diversity, shorter average tenor of loan collateral, and higher loan spreads relative to BSL CLO portfolios. However, while there is less diversification within individual MM CLO collateral pools, there is much greater diversification across MM CLOs from different managers, as MM CLO loans are often unique to a single collateral manager or small club of managers.

For data on our universe of rated MM CLOs, including most of the items referenced above, see "Private Credit And Middle-Market CLO Quarterly: The Times They Are A-Changin'," published Oct. 18, 2024, which we update and publish quarterly.

What's New (And Not New) In The Stress Scenarios For This Year

As in previous years, we've set up four scenarios with a simple progression of increasingly severe collateral default assumptions (10%, 15%, 20% and 30%), with a 50% recovery assumption. By doing this, we hope to allow CLO investors and others to take their own view of potential collateral default rates and gauge the potential CLO rating impact. Using the same four scenarios allows comparison of the results over time. The results can change year-to-year based on collateral deterioration (or improvement) in existing transactions, and with the addition of newly rated MM CLOs to the sample of CLOs tested.

This year, based on feedback from investors, we've also added a set of runs with a 35% recovery assumption for defaulted assets, and we present these results alongside the 50% recovery assumption modeling we've published in prior years. This mirrors what we did with our recently published BSL CLO rating stress scenarios (see "Scenario Analysis: Stress Tests Show U.S. BSL CLO Ratings Able To Withstand Significant Loan Defaults And Downgrades (2024 Update)," published Oct. 10, 2024), where this year, we added stress scenarios with a 30% recovery assumption alongside the 45% assumption we'd used in prior years (see table 1).

Table 1

For BSL CLOs, we added the 30% recovery assumption to reflect shifting dynamics in the BSL market that are reflected in our recovery ratings on corporate loans, and the rise of liability management transactions (LMTs), where stressed companies sometimes restructure their debt outside the bankruptcy process (see "Demystifying Loan Liability Management Transactions And Their Impact On First-Lien Lenders," published Oct. 30, 2024). LMTs haven't been an issue among companies found in our MM CLO portfolios, but we added the 35% recovery assumption to the MM CLO rating stress scenarios to mirror what we did in the published BSL scenarios and to provide readers with multiple recovery assumptions alongside the range of collateral default assumptions. The differences between the BSL CLO recovery assumptions (30% and 45%) and MM CLO recovery assumptions (35% and 50%) reflect relatively tighter loan documents for MM CLO loans, along with the fact that a large majority of them have maintenance covenants (i.e., aren't covenant-lite), in contrast with the BSL market.

Generating The MM CLO Rating Stress Scenarios

Chart 1 shows a summary of our MM CLO stress scenarios and ratings impact.

Chart 1

The scenarios envision a proportion of loans in each CLO experiencing a given level of defaults, and then assess the potential ratings impact based on quantitative analysis generated for the CLOs, including cash flow modeling. The results lack the qualitative factors a surveillance committee might consider when deciding on whether to change ratings for a given CLO in the real world. For example, if a 'AAA' rated CLO tranche were failing its model results, a committee might look at expected paydowns from diverted cash flows on future payment dates and decide to wait before lowering the rating. As such, the stress scenario results below may modestly overstate the likelihood of CLO 'AAA' ratings transitioning under a given stress. Nonetheless, we think they provide insight into how MM CLO ratings might respond to varying stress levels.

We deep dive into the scenarios below.

The stress scenario sample: 184 U.S. MM CLOs

In this year's study, our sample includes 184 reinvesting MM CLO transactions rated by S&P Global Ratings. The full sample of U.S. MM CLOs tested has exposure to more than 5,424 loans from 2,991 companies. As of the start of Q4 2024, 486 of these companies were rated within the 'CCC' category ('CCC+', 'CCC', and 'CCC-'), and 45 had what we view as a nonperforming rating for purposes of our CLO analysis ('CC', 'SD', and 'D'). Further up the rating scale, the MM CLO transactions in our sample had on average 72.4% of their collateral invested in loans across 1,851 issuers rated or credit estimated 'B-' on average.

Differences between MM and BSL CLO stress scenarios

While the framework we use to produce the BSL and MM CLO stress scenarios is similar, there are some differences, including the way hypothetical defaults are applied to different CLOs under the scenarios.

The default rate across MM CLO portfolios in a significant downturn could be correlated by manager, in our view, in a way that would be less true for BSL CLOs. This is because many loans within MM CLO portfolios are originated by the manager in their capacity as a direct lender, so some of the loans are specific to a single manager or small group of managers in a way that's generally not true for BSL CLOs. For our stress runs, to maintain consistency across the sample we assumed each portfolio experiences the same default level. For example, under scenario two described below, we assume that 15% of the assets in each CLO within our entire sample experience a default.

This is different from the approach we use for our BSL CLO rating stress analysis, where we take a bottom-up approach and look at the entire set of companies across our BSL CLOs. We then sort the performing companies issuing the loans in the collateral pools from weakest to strongest (first by rating, then by loan price) and then hit the stress scenario default targets by starting at the bottom of the list and working our way up. In the BSL CLO rating stress scenarios we have published, the specific level of asset defaults within each CLO varies by their exposure to the companies we assumed would default which makes sense for since a large majority of BSL CLO collateral are held by multiple BSL CLO managers. This approach makes less sense for MM CLOs, in our view, because the collateral within MM CLOs are more likely to be unique to the MM CLO manager, or held across a small group of managers. Additionally, we currently do not have complete loan price coverage for MM CLO collateral, which is an important data point in creating the scenarios for the BSL study. More than 70% of the MM CLO collateral are rated or credit estimated at 'b-', many of which we do not have live pricing data.

Wide range of MM CLO structures

The structures of the MM CLOs we rate are much more diverse than those seen across our rated BSL CLO transactions, and the sample of MM CLOs we tested includes warehouse structures with revolving notes, delayed draw notes (some senior, some junior), turbo features on 'BBB' or 'BB' notes, and CLOs with large X note balances. Some transactions within our sample had a combination of these features mentioned (about 17% of our sample including one or more unique features). For purposes of the testing, we excluded revolving tranches that currently have an undrawn balance. Additionally, we also excluded X notes and combo notes from our sample.

Some MM CLOs have lower leverage capital structures that include only investment-grade tranches, while others have a junior note originally rated in the speculative grade range. The latter now makes up about half of the MM CLOs in our total sample. Additionally, the median senior subordination levels of the MM CLOs that have closed since 2023 tend to be slightly lower than the median senior subordination levels of the pre-2023 CLOs. The median 'AAA' tranche subordination of the CLOs in our sample this year is 42.0%, down from 42.5% last year.

Table 2

In reviewing the roughly half of our sample that issued a 'BB' tranche and the other half of our sample that only issue investment-grade tranches, we find various key fourth-quarter 2024 metrics within their reinvesting portfolios (see table 3). Aside from the differences highlighted above, we also find the transactions with a 'BB' tranche tend to have slightly less subordination across each rating category relative to the investment-grade-only transactions. In the coming months, we will provide further findings on how the ratings impact from our stress scenarios differed between these two cohorts of transactions.

Table 3

Stress Scenario Results

We present the results of our stress scenarios below.

"Core four" stresses using 50% recovery assumption

Table 4 below provides the results of our stress scenarios with the 50% recovery assumption. These four rating stress scenarios are identical to ones we've applied to our U.S. MM CLO scenario analyses published each year since 2021, with 10%, 15%, 20%, and 30% of CLO assets assumed to default producing net par losses of 5%, 7.5%, 10%, and 15%. These four stress scenarios have the benefit of being transparent and simple, allowing market participants to take their view of potential loan defaults and assess what the potential CLO rating impact might be.

Table 4

The results of our current scenario analysis are in many ways like those in our prior studies. As expected, we see larger rating transitions among junior CLO tranches and a correspondingly lighter rating impact on tranches further up the CLO capital stack. Unsurprisingly, the average notch movement at a given CLO tranche rating level increases as the scenarios become more severe, although the impact across the scenarios isn't linear. In all four of these stress scenarios under the 50% recovery assumption, the non-deferrable notes originally rated 'AA' are most likely to be affirmed amongst the other rating categories (even more than the senior tranches originally rated 'AAA'), mostly due to the fact these tranches were structured with more rating cushion at close, a trend we have noticed across our prior scenario analysis on MM CLOs.

"Core four" stresses using a lower recovery assumption (35%)

Since 2021, for the four scenarios outlined above, we have used a recovery assumption of 50% for issuers assumed to be defaulted under the stress tests. This year, based on investor feedback, we have also generated these four stresses using a lower recovery rate assumption (35%). The impact of the lower recovery assumption is an increase in par loss under each of the four scenarios (see table 5). For example, under the 10% default scenario above, we assume 10% of loans default with a 50% recovery, resulting in a par loss of 5%. Using a 35% recovery assumption for that same stress, as we do below, increases the par loss across the sample to 6.5%, having a material impact on the results.

Table 5

The increase in par loss from the reduced recovery assumption has a notable impact across the scenarios. Under our most punitive default scenario, which assumes a 30% default rate, reducing the recovery assumption to 35% from 50% increases the average rating notch impact on the 'AAA' tranche results to 1.77 notches from 0.77 notches from the same 30% default scenario with a 50% recovery. However, even under this extreme stress, all of the 'AAA' tranches in our sample retain investment-grade ('BBB-' and higher) ratings, and almost all (98.58%) of the 'AA' tranches remain investment grade as well. Further down the capital stack, the impact is more severe, with more than half of the tranches originally rated in the 'BB' and 'BBB' category defaulting).

Comparing this year's reinvesting transaction results with last year

Over the past year, some MM CLO transactions experienced par gain and stable credit, while others experienced par losses from defaulted assets or various haircuts to their overcollateralization (O/C) test numerator, resulting in large changes to their O/C test. The wide variations in CLO performance may be a result of the low manager overlap across MM CLO managers, the lower diversity levels of MM CLOs (when compared to BSL CLOs), and the wide range of CLO structures noted above. We also noted that some MM CLO managers have removed stressed assets from their portfolios and replaced them with stronger credits, sometimes at par. This has further contributed to the range in MM CLO performance.

When we compare the results of this year's 20% default scenario with 50% recovery for reinvesting transactions to the identical stress run from last year's study, the transactions this year show a slightly greater magnitude of negative ratings migration for all rating categories, especially at the 'BB' tranche level (note that only about half of our sample have a 'BB' CLO tranche) (see chart 2). This makes sense as several transactions that were part of our last study have reinvested since then, and on average, they lost about 0.18% of par (as a percentage of target par) while junior O/C cushions have declined by about 0.41% over the past year. Some reinvesting transactions have experienced declines of over 1% in par loss (see slide 22 and 23 from "Private Credit And Middle-Market CLO Quarterly: The Times They Are A-Changin'," published Oct. 18, 2024), and some have already exceeded their 'CCC' buckets or experienced an uptick in deferring assets, resulting in multiple percentage point declines in junior O/C cushion.

Chart 2

This Year's Stress Scenarios Still Show The CLO Structure Protecting Senior Noteholders

In this year's study, we also explored the impact to our CLO ratings under our scenarios if we assume a lower recovery rate of 35% (in addition to the 50% recovery assumption we've used in the past). By changing the recovery assumption to 35% from 50%, the proportion of 'AAA' CLO notes remaining within a one-notch downgrade drops to 54% from 92% under the 30% default scenario. But even under the most punitive scenario, CLO tranches showed impressive resilience. As in previous years, this year's study continues to show the fundamentals of the CLO structure protecting senior noteholders, with no 'AAA' CLO tranche downgraded below 'A-' under any of the scenarios, and 99% of the non-deferrable 'AA' CLO tranches remaining within investment grade even under our most punitive scenario (30% loans defaults with a 35% recovery assumption).

Related Research

• Demystifying Loan Liability Management Transactions And Their Impact On First-Lien Lenders, Oct. 30, 2024
• Private Credit And Middle-Market CLO Quarterly: The Times They Are A-Changin', Oct. 18, 2024
• Scenario Analysis: Stress Tests Show U.S. BSL CLO Ratings Able To Withstand Significant Loan Defaults And Downgrades (2024 Update), Oct. 10, 2024
• Anatomy of a Credit Estimate: What It Means And How We Do It, Jan. 14, 2021