While there were elevated levels of speculative-grade corporate rating downgrades in 2022 and 2023, ratings on broadly syndicated loan (BSL) collateralized loan obligation (CLO) fared better during the period, with more CLO ratings raised than lowered despite the economic environment. The difference in rating performance can be explained partly by CLO managers actively repositioning portfolios away from sectors likely to underperform, and the fact that BSL CLO portfolios typically have less exposure to 'CCC' and defaulted loans than the loan market as a whole.

The past year has been mostly benign from a CLO ratings perspective despite economic challenges for leveraged loan issuers and an increase in liability management transactions (LMTs), where stressed companies restructure their outstanding debt without going through a bankruptcy process. These could have threatened BSL CLOs with par loss and potential CLO ratings migration, but have only had a modest impact so far. In part, we think this is due to some CLO managers purchasing corporate bonds at a discount to par in the midst of a high rate environment, improving their CLOs' overcollateralization (O/C) ratios and collateral credit quality at the same time. The fundamentals of the CLO structure also worked to protect CLO ratings, with a large majority (97%) of BSL CLO downgrades from 2022 through third-quarter 2024 being on speculative-grade-rated CLO tranches.

Following two years of "higher-for-longer" interest rates and economic uncertainty, in September, the Federal Reserve provided a tailwind to corporate credit with a 50 basis point (bps) rate cut, reducing interest expenses for the companies that issue the loans backing BSL CLOs. Interest rate cuts should be broadly supportive of corporate credit going forward, benefitting CLO collateral. But as someone once said, every day that goes by brings us one day closer to the next downturn, whenever it may arrive. To that point, as we've done in previous years, we generated a series of stress scenarios to test the resiliency of our U.S. BSL CLO ratings if they were subjected to another downturn.

What's New In The Stress Scenarios For This Year

As we've done in previous years, we have generated a series of stress scenarios to see how our BSL CLO ratings would perform under different economic environments (see the Related Research section at the end of this article). For purposes of this year's exercise, we re-ran the four scenarios that we've published previously, allowing for comparisons of how BSL CLO ratings responded to the stresses over time. Each of the four scenarios envisions a proportion of corporate loan issuers experiencing a default, and then assumes that a proportion of the remaining (i.e., non-defaulted) obligors are rated in the 'CCC' range. Additionally, based on feedback from investors, we added a set of runs with a 30% recovery assumption for defaulted assets and present these results alongside the 45% recovery assumption modeling we've published in prior years. Our goal with these scenarios is to allow CLO market participants to take their forward view of prospective corporate loan defaults and CLO 'CCC' basket sizes (and now also recoveries) and assess what the impact on our BSL CLO ratings might be.

Beyond the stresses outlined above, we've also added two other new stresses (also based on investor feedback), where we take a different approach and simply notch the ratings on all obligors in BSL CLO collateral pools downward by one notch or two notches, and look at the impact on the CLO ratings. (Given the number of stresses and different ways to present the data, we couldn't fit results for all of these stresses into the tables shown in this article, but we present the key ones below, and results for all the scenarios are available in a spreadsheet that can be downloaded: Click here.)

The Stress Scenario Sample: 701 U.S. BSL CLOs

To produce our rating stress scenarios, we started with a sample of 701 U.S. BSL CLO transactions rated by S&P Global Ratings. We applied the various collateral default and downgrade stresses described below, and then generated cash flow and credit analysis similar to the quantitative analysis a surveillance committee might review when reviewing a CLO for potential downgrade. The results indicated are based solely on quantitative analytics and lack the qualitative input that a committee might choose to consider when making a rating decision.

Our full sample of 701 U.S. BSL CLOs has exposure to more than 3,100 loans from 1,700 companies. As of mid-2024, 254 of these companies were rated within the 'CCC' category ('CCC+', 'CCC', and 'CCC-'), and 60 had what we view as a nonperforming rating for purposes of our CLO analysis ('CC', 'SD', and 'D'). Further up the rating scale, the 701 CLO transactions in our sample had 25.5% of their collateral invested in 'B-' loans on average, down from around 30% in our study last year (for more discussion on the reduction of 'B-' companies in BSL CLO collateral pools, see "U.S. BSL CLO Rating Performance Under Four Hypothetical Stress Scenarios (2023 Update)," published July 18, 2023).

Table 1

The stresses were applied to the universe of obligors within our BSL CLO transactions, not at a CLO-by-CLO level. To achieve the target 'CCC' and default exposure for each of the scenarios above, we adjusted the ratings on as many CLO obligors as needed, starting with the weakest (based on rating and then loan price) across our sample of CLOs. This can produce CLOs with a range of exposures in the stress analysis. For example, in the "5/10" scenario, some CLOs end up with more than 5% exposure to defaulting loans because they start with weaker (lower rated and lower priced) collateral, and others less, but the average ends up at about 5% across the CLOs in the sample. In our view, applying the stresses this way produces more realistic results because it accounts for the starting credit quality of the individual BSL CLO collateral pools, rather than applying a uniform stress across each.

In addition to the 45% recovery we assumed in our published stresses in prior years, this year, based on investor feedback, we added stresses with a 30% recovery assumption. We view both of these as conservative assumptions given historical recovery rates on first-lien senior secured loans, which make up about 99% of BSL CLO collateral by par value held.

Chart 1

The 701 CLOs in the sample (and the chart above) includes both CLOs still within their reinvestment period and CLOs in their amortization phase. Given the difference in CLO credit metrics between newer and older transactions, within the tables linked above we provide results broken out between the 506 reinvesting CLOs and the 195 amortizing CLOs in the sample, as well as the combined (full sample) results as reflected in the graphic shown above.

Stress Scenario Results For "Core Four" Stresses Using 45% Recovery Assumption

We present the results of our stress scenarios with the 45% recovery assumption below. These four rating stress scenarios are identical to ones we've applied to our U.S. BSL CLO scenario analyses published each year since 2020:

• "5/10" scenario: 5% of loan issuers default and 10% of loan issuers lowered to 'CCC';
• "10/20" scenario: 10% of loan issuers default and 20% of loan issuers lowered to 'CCC';
• "15/30" scenario: 15% of loan issuers default and 30% of loan issuers lowered to 'CCC'; and
• "20/40" scenario: 20% of loan issuers default and 40% of loan issuers lowered to 'CCC'.

The core four stress scenarios have the benefit of being transparent and simple, allowing market participants to take their view of potential loan defaults and 'CCC' exposure amounts and assess what the potential CLO rating impact might be. Producing the same analysis on outstanding CLOs over time also provides insight into how the transactions are evolving and any changes in how they respond to the stresses. Overall results may have changed because of changes in individual transaction performance over the period, but also because of new issue CLOs being added to the sample of transactions being tested over time.

Table 2 below provides a summary of these four scenarios and their ratings impact using a 45% recovery assumption.

Table 2

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.

CLOs With Stronger Portfolio Credit Metrics Are Generally More Resilient

We find that the longer a CLO is outstanding, the more likely it will have weaker credit metrics (see table 3). To test this, we broke out our sample into three vintage-based cohorts: CLO originated before 2020, between 2020 and 2022, and in 2023 and the first half of 2024. The 2023 and 2024 vintage transactions are the newest, have closed in a high-rate environment, and have not experienced significant collateral deterioration yet. The 2020-2022 vintage transactions were mostly issued before the rate increases arrived and have reinvested through a period of rising rates; while the pre-2020 vintage transactions experienced the negative effects of the rate increases in 2022 and 2023 as well as the pandemic in 2020.

Table 3

Focusing on the "5/10" scenario with 45% recovery, although being the least punitive stress of the core four stresses, the tranches rated in the 'BB' rating category see material potential rating impact. Given the wide range of vintages included in the sample, there were also a wide range of outcomes to these 'BB' tranches under the stress, from affirmations to defaults. One quarter of the sample saw rating affirmations (no downgrades – the CLO tranche can withstand this scenario and still pass our cash flow stresses at its current rating), while the other three quarters of the sample experienced different levels of downgrade: 11.0% of the sample experience a downgrade into the 'CCC' category, while 5.7% saw their ratings lowered to 'CC' under the scenario, indicating a likely default under the stress. The difference in performance can be attributed to starting collateral quality; if we look at just the CLOs that saw 'BB' category tranche ratings affirmed under the "5/10" stress, we observe that these deals have the following characteristics:

• Stronger rating distribution: average S&P Global Ratings' weighted average rating factor (SPWARF) of 2566;
• Less exposure to assets rated 'CCC+' and below: average of 4.64%; and
• Larger 'BB' O/C cushion: average of 4.85%.

Metrics for the transactions that experience a 'BB' rating affirmation look very similar to the newer 2023/2024 vintage transactions summarized above. Meanwhile, the 'BB' tranches that experience larger negative rating movement under the stress scenario are more likely to have weaker CLO metrics even before the stress is applied. For example, the 11.0% and 5.7% of the 'BB' tranches that are at risk of downgrade into the 'CCC' category and default, respectively, under the "5/10" scenario with 45% recovery have CLO metrics that are closer to the average pre-2020 transaction in table 3 above.

When we review the list of 21 CLO 2.0 defaults that have actually occurred (see "Thirty Years Strong: U.S. CLO Tranche Defaults From 1994 Through Third-Quarter 2024," published Sept. 27, 2024), we find all are junior notes from transactions that originally closed in 2014 or earlier; and before defaulting, they all experienced two distinct periods of economic weakness: the commodities slowdown as well as the pandemic. In our prior scenario analysis in 2021, we find the effects of the pandemic in 2020 were somewhat similar to the "5/10" scenario, so by putting them through the scenario analysis, they are effectively experiencing two successive "5/10" scenarios, and it wouldn't be a surprise if some of these 'BB' tranches default under the hypothetical stress.

Table 4

Senior (Non-Deferrable) Tranches From Older CLOs Remain Resilient As They Age

Transactions from older CLO vintages tend to have weaker metrics, which, as noted above, tend to see greater CLO rating transitions in the scenario analyses. One exception is that senior non-deferrable tranches (originally rated 'AAA' and 'AA') from older CLOs can become more resilient as they get closer to (or further into) their amortization periods, despite their weaker CLO metrics. In our cash flow modelling, expected paydowns from amortization and coverage test cures generally benefit the model results of senior tranches. This generally does not apply for the deferrable mezzanine and junior CLO tranches (originally rated in the 'A' category and below), as they still have multiple classes of notes above them before they are next in line for paydowns. The deferrable tranches of the older transactions generally see greater average notch downgrades, as noted in table 5 below.

For example, the average impact on the non-deferrable (i.e., senior) tranches across the core four scenarios for the older pre-2020 transactions are:

• 0.89 notches down, on average, across the full sample;
• More resilient than the 2020-2022 vintage transactions, which saw 1.24 notches down, on average, despite the 2020-2022 vintage transactions having relatively better credit quality (as seen in table 3 above); and
• Only slightly less resilient than the newer 2023/2024 vintage transactions, which have much stronger credit metrics and saw 0.80 notches down, on average.

Meanwhile, as expected, the deferrable tranches of the older pre-2020 transactions with weaker credit metrics experience larger average notch downgrades, the deferrable tranches of the 2023/2024 vintage transactions with stronger credit metrics experience less average negative notch movement while average impact to deferrable notes of the 2022-2022 vintage transactions sit right in between.

Table 5

Same Core Scenarios With A Lower Recovery Assumption (30%)

Since 2020, for the core four scenarios outlined above, we have used a recovery assumption of 45% recovery for issuers assumed to be defaulted under the stress tests. This year, based on investor feedback, we have also generated these four stresses but at a lower recovery rate assumption (30%). The impact of the lower recovery assumption is an increase in par loss under each of the four scenarios. For example, under the "5/10" scenario above, we assume 5% of loans default with a 45% recovery, resulting in an average par loss of 2.78%. However, using a 30% recovery for that same stress, as we do below, increases the average par loss across the sample to 3.54%, an average increase in par loss of 76 bps.

Table 6

The increase in par loss from the reduced recovery assumption has a notable impact across the scenarios. For the reinvesting CLOs in the sample, under the very punitive "20/40" scenario (where we assume 20% of loans default with a 30% recovery and CLO 'CCC' baskets increase to 40%), the reduction in recovery assumption has a significant impact on the 'AAA' tranche results, with the average downgrade for these increasing to 2.6 notches from 1.3 notches under the 45% recovery assumption. However, even under this extraordinary stress nearly all of the 'AAA' tranches (99.65%) remain investment grade ('BBB-' and higher), and 96.60% of the 'AAA' tranches remain at 'A-' or higher after the stress is applied.

Two New Scenarios

In addition to the stress scenarios outlined above, based on feedback from a large global CLO investor, we ran two additional scenarios where we lowered ratings across all CLO obligors by one notch (in the first scenario) and two notches (second scenario), while assuming all ratings that ended up being lowered below 'CCC-' default and recover 30%.

In our core four stresses outlined in the previous sections, the proportion of assets assumed to default and obligors assumed to be in the 'CCC' range are, by definition, one-to-two ("5/10" scenario, "10/20" scenario, etc.). For these two new stresses, the proportions are quite different. The "downgrade everything by one notch" scenario results in something like a one-to-thirty two ratio, with 0.82% of CLO collateral assumed to default and 31.89% assumed to end up in the 'CCC' category. The "downgrade everything by two notches" scenario gives us a ratio of 2:58 (1.83% of CLO collateral assumed to default and 57.63% ending up in the 'CCC' basket). Relatively speaking, these two new scenarios are high downgrade/high 'CCC', but low default/par loss scenarios.

Table 7

The "15/30" stress from our core four scenarios above produces a 'CCC' basket similar in size to the one-notch downgrade scenario here. The ratings impact, however, is very different. The "15/30" scenario is much harsher, given the assumed 15% collateral defaults resulting in par loss of 10.61% (assuming a 30% recovery) compared to par loss of just 0.58% for the one-notch downgrade scenario under the same recovery assumption. The resulting ratings migration under the stress scenarios reflects this, as the average notch downgrade for 'AAA' tranches under the "15/30" scenario at 30% recovery was 1.14 notches, notably higher than the average notch movement of the one-notch scenario (0.72 notches) and even higher than that of the two-notch scenario (1.00 notches).

Chart 2

The new one-notch and two-notch scenarios, while appealing in their simplicity, are unlikely to play out in reality. The net ratings impact on CLO obligors during the GFC (between start of 2008 and middle of 2009) was just under a one-notch downgrade to CLO collateral ratings on average (or more precisely, 0.9 notch downgrade across all CLO 1.0 obligors). However, looking more closely, we see that those downgrades were distributed unevenly across obligors, with just under half of the U.S. CLO 1.0 obligors (44%) experiencing an average downgrade of 2.7 notches, while 40% experienced no rating action (0 notch) and 16% experienced a 2.0 notch upgrade on average. Given this, it's very unlikely every CLO obligor in a CLO would experience a one- or two-notch downgrade; but still, these studies can be helpful to highlight the different sensitivities across the ratings of senior and junior CLO notes to a systemic stress like these.

Comparing With Last Year's Reinvesting Transaction Results

When comparing the results of this year's "5/10" scenario with 45% recovery for reinvesting transactions to last year's study, the transitions this year show a somewhat greater magnitude of ratings migration for the mezzanine CLO tranches, especially at the 'BB' tranche level (see chart 3 below). This makes sense as many transactions within the sample are a bit weaker relative to last year given the deterioration in corporate credit quality over the past year, and junior CLO notes are the most sensitive to rating stress. Further up the capital stack, senior tranches in the sample have similar levels of resilience to the hypothetical stresses as they did last year.

Chart 3

Table 8

When we compare the average reinvesting CLO metrics in this year's study with last year, one change that may stand out is the decline in exposure to assets from 'B-' obligors, which is now 25% of total assets compared to 30% last year. This is a key driver to the lower average SPWARFs across the two published studies despite the average 'CCC' buckets looking similar. The big caveat is the change in the average portfolio par balances (as a percentage of target par), which are now lower by 22 bps. That may not seem like much, but par loss impacts the existing subordination levels for all tranches of the CLO and has a noticeable impact on our scenario results. In last year's scenario analysis, over two-thirds of the reinvesting CLOs were at or above target par, while in this year's study, only 45% of the reinvesting sample were at or above target par as most of the existing transactions have lost par since.

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

This year's stress tests have produced hypothetical transitions that are modestly more negative than the respective results from last year's study. For example, last year, 93% of 'AAA' CLO notes were within a one-notch downgrade under the "20/40" scenario (assuming a 45% recovery); this year, 83% of 'AAA' ratings were within one notch under the same stress. We find the difference is mostly attributed to the change in the sample used for this study. The bulk of the transactions are the same as the ones from last year, and they experienced additional credit deterioration and par loss over the past year given the economic (and corporate rating downgrade) environment.

We also explored the impact to our CLO ratings under the core four scenarios if we assume a lower recovery rate of 30% (in addition to the 45% recovery assumption we've used in the past). By changing the recovery assumption to 30% from 45%, the proportion of 'AAA' CLO notes remaining within a one-notch downgrade drops to 48% from 83% under the "20/40" scenario. But even under this very punitive assumption, CLO 'AAA' tranches showed impressive resilience. As in previous years, this year's study continue to show the fundamentals of the CLO structure protecting senior noteholders, with no 'AAA' CLO tranche defaults under any of the scenarios, and 99% of the non-deferrable CLO tranches ('AAA' and 'AA') paying off in full even under our most punitive scenario (20% loans defaults with a 30% recovery and 40% CLO 'CCC' baskets).

Many of the scenarios outlined in this article would be very unlikely to occur in the real world. For example, CLO collateral defaults and 'CCC' baskets reached about 7% and 11%, respectively, by mid-2009 after the GFC, and the history of the leveraged loan market has never seen defaults reach 20% or CLO 'CCC' get to 40% (or even halfway to 40%). However, looking at how our BSL CLO ratings might perform under such stresses gives us confidence in our analytical approach for rating CLO transactions, and provides transparency about how ratings might under scenarios ranging from realistic to those further out on the tail of the distribution.

Related Research

• Scenario Analysis: How Resilient Are Middle-Market CLO Ratings (2023 Update)? Oct. 16, 2023
• Scenario Analysis: U.S. BSL CLO Rating Performance Under Four Hypothetical Stress Scenarios (2023 Update), July 18, 2023
• Scenario Analysis: How Resilient Are Middle-Market CLO Ratings (2022 Update), Oct. 19, 2022
• Scenario Analysis: How The Next Downturn Could Affect U.S. BSL CLO Ratings (2022 Update), Aug. 4, 2022
• Scenario Analysis: How The Next Downturn Could Affect U.S. BSL CLO Ratings, June 17, 2021
• Scenario Analysis: How Resilient Are Middle-Market CLO Ratings? Feb. 26, 2021
• Scenario Analysis: How Credit Distress Due To COVID-19 Could Affect U.S. CLO Ratings, April 24, 2020