After more than a year of interest rate increases and economic uncertainty, negative sentiment in the leveraged finance market continues. Many of our recent collateralized loan obligation (CLO) investor conversations have revolved around the increase in corporate rating downgrades and impact on U.S. broadly syndicated loan (BSL) CLO 'CCC' baskets, along with our outlook for loan defaults and recoveries. Much of the focus of these discussions has revolved around how CLO ratings might be affected in a downturn scenario.

As we've done in previous years, we recently generated a series of stress scenarios to test the resiliency of our U.S. BSL CLO ratings. Each scenario envisions a proportion of corporate loan issuers experiencing a default, then assumes that a proportion of the remaining (i.e., non-defaulted) obligors are rated in the 'CCC' range. For purposes of this current exercise, we re-ran four of the stresses we published in April 2020, June 2021, and August 2022, which should allow for performance comparisons across time. Our goal is to allow investors and other market participants to take their view of prospective corporate loan defaults and CLO 'CCC' basket sizes and see what the impact on our BSL CLO ratings might be.

For earlier versions of our published BSL CLO rating stress scenario articles, or for other U.S. CLO scenario analysis articles we've produced, see the Related Research section at the end of this article.

Since Our Last BSL CLO Rating Stress Test Was Published, Average BSL CLO Metrics Have Deteriorated Modestly

Over the past year, rising interest rates have affected lower-rated borrowers, and downgrades to speculative-grade corporate ratings have outpaced upgrades since May 2022, with the pace of downgrades accelerating in August. As a result, BSL CLO collateral metrics have deteriorated somewhat since we published our last U.S. BSL CLO rating scenario analysis in August 2022.

Between July 2022 and June 2023, reinvesting U.S. BSL CLO metrics saw the following average changes (see "CLO Insights 2023 U.S. BSL Index: Credit Metrics Mostly Stable; Exposure To 'B-' Assets Drops A Bit, Though O/C Cushions Turn Negative For A Few Pre-Pandemic CLOs," published June 29, 2023):

• The average U.S. BSL CLO 'CCC' basket increased to 6.47% from 3.60%;
• Exposure to nonperforming assets increased to 0.83% from 0.26%; and
• The average junior overcollateralization (O/C) ratio test cushion declined to 4.13% from 4.64%.

When looking at individual CLOs, metrics can differ considerably from the averages. Generally, CLOs originated in first-quarter 2020 or before (prior to the arrival of the COVID-19 pandemic downturn) have considerably weaker credit metrics than those originated after first-quarter 2020. Similarly, credit metrics for CLOs still within their reinvestment periods tend to be stronger than those of CLOs in their amortization period. In addition to the vintage effect, amortizing CLOs can also be affected by a negative selection bias in their portfolios where stronger obligors are more likely to pay down loans sooner, leaving (relatively) weaker obligors behind.

The average 'CCC' exposure of pre-pandemic CLOs continue to be higher than post-pandemic CLOs, but the difference has declined to less than 1% by the middle of 2023 from 2% a year ago: as of June 2023, 'CCC' exposure averaged 6.93% for pre-pandemic CLOs versus 6.05% for post-pandemic CLOs, both up from June of 2022 a year ago, where they averaged 4.66% and 2.66%, respectively.

The Stress Scenario Sample: 837 U.S. BSL CLOs

To produce our rating stress scenarios, we started with a sample of 837 U.S. BSL CLO transactions rated by S&P Global Ratings. We applied the collateral default and downgrade stresses, and then generated cash flow and credit analysis similar to the quantitative analysis that a surveillance committee would review when analyzing a CLO for potential downgrade. The results below are based on the quantitative analytics and lack the qualitative input that a surveillance committee might choose to consider when making a rating decision.

In previous rating stress articles, we focused only on CLOs that were within their reinvestment periods, but given the large proportion of U.S. BSL CLOs currently outside their reinvestment period--about 19% at the time of this publication, and expected to rise to over 36% by end of 2023 (and, if no further issuance, almost half of outstanding CLOs by end of 2024)--for this article, we include both CLOs that are still within their reinvestment periods and those that are in their amortization phases. Given the difference in performances between the two cohorts of CLOs, we provide results for the overall sample, but also break out the results between the 677 reinvesting CLOs and the 160 amortizing CLOs in our sample.

Across our full sample of 837 reinvesting U.S. BSL CLOs, there is exposure to over 3,000 loans from more than 1,800 obligors. As of May 2023, 256 of these obligors were rated within the 'CCC' category ('CCC+', 'CCC', and 'CCC-') and 58 obligors had a nonperforming rating ('CC', 'SD', and 'D'). Further up the ratings scale, the 837 CLO transactions also had exposure to $129.57 billion of loans from 493 'B-' rated issuers--more than 30% of total portfolio assets. This is more than double the proportion of assets from 'B-' rated obligors in U.S. BSL CLO collateral pools five years ago.

CLO issuance has been relatively modest since we published our last BSL CLO rating stress article in August 2022, but the proportion of post-pandemic CLOs included within the sample has increased: 55% of the overall sample in this study were issued in second-quarter 2020 or later, compared to 42% of the sample used for the 2022 study. Of the reinvesting CLOs in the sample, 45% were issued prior to the pandemic, compared with 97% of the amortizing CLOs.

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), on average, across our sample of CLOs. Note that 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, and others less, but the average ends up at about 5% across 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.

Finally, we assume a 45% recovery rate (or par loss given default of 55%) for the purposes of these four stresses. We think this is an appropriate, conservative assumption given the economic stresses implied by our scenarios.

Stress Scenario Results

We present the results of our stress scenarios below. These four rating stress scenarios are identical to the ones we applied in our U.S. BSL CLO scenario analysis articles published in April 2020, June 2021, and August 2022:

• "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'.

These scenarios have the benefit of being transparent and simple, and 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 (see below). Overall results may have changed because of changes in individual transaction performance over time, but also because of new issue CLOs being added to the sample of transactions being tested.

Chart 1 and table 2 below show a summary of these four scenarios and their ratings impact for the full sample of transactions, both reinvesting and amortizing. Table 4 in the Appendix I section provides results of reinvesting and amortizing CLOs separately.

Chart 1

Table 2

What The Scenario Analysis Results Show

The results of our current scenario analysis are in many ways similar to our prior studies. As expected, we see larger rating transitions among the more junior CLO tranches and a correspondingly lighter rating impact from the stresses on tranches further up the CLO capital stack. Also as expected, the average notch movement at a given CLO tranche rating level increases as the scenarios become more severe, although the impact across scenarios isn't linear.

Notable differences within the average notch movements across the 2020, 2021, 2022, and 2023 scenarios

This is our fourth installment of our scenario analysis study on U.S. BSL CLO ratings where we ran the same four scenarios described above for reinvesting U.S. BSL CLO ratings. Through the years, there have been notable differences within the average notch movements across the scenarios. In chart 2 below, we show the average notch movement for the "5/10" scenario across the scenario analyses conducted in 2020, 2021, 2022, and 2023.

Chart 2

In our April 2020 study, with results generated in March 2020, the CLOs tested had not yet seen the impact of the pandemic and associated corporate rating downgrades, so the reinvesting portfolios were still relatively clean. They had fewer 'CCC' assets and defaulted obligors in the collateral pools than CLOs would have after the wave of corporate rating downgrades in April 2020, and the 'CCC' and nonperforming exposures were 4.84% and 0.64%, respectively. Some of the earlier vintage CLOs at that time (pre-2016 vintage) did have slightly weaker portfolios, but most of these transactions were relatively clean in early 2020.

For the next rating stress analysis in mid-2021, a large majority of the transactions in the sample were reinvesting pre-pandemic transactions, and had on average experienced about 1% of par loss in 2020 due to the economic impact of the pandemic. Because a large majority of the sample had already experienced a loss in subordination, layering on the credit impact of a "5/10" scenario on top of a sample of mostly weakened transactions resulted in a large uptick in notch movement compared to the 2020 study.

In the subsequent 2022 and 2023 studies, as the pre-pandemic transactions recovered some of the par lost in 2020, and as new issue post-pandemic CLOs entered the sample, the average notch movement has reverted back towards the levels seen in the April 2020 study, when most of the sample was still relatively clean.

The results of this year's study

This year, we bifurcated the results of our study by CLO reinvestment status. Given the reset market has remained mostly dormant for more than a year, amortizing CLOs have begun to represent an increasing proportion of the U.S. BSL CLO universe, as noted above. Presenting the results this way has a similar effect to the vintage bifurcation we did last year, given that nearly all the amortizing CLOs this year (97%) were originated prior to the arrival of the pandemic in first-quarter 2020, while a majority of the reinvesting CLOs (55%) were originated after. We find that CLO vintage continues to be a significant determinant of rating transitions under the various stress scenarios. The differences between the pre- and post-pandemic reinvesting vintages continue to hold, where for most CLO ratings, the transitions are more severe for the pre-pandemic CLOs, largely due to their larger average exposure to 'CCC' and defaulted assets.

Amortizing CLOs under notable stress: stronger 'AAA' and 'AA' performance; weaker at 'A' and below

Across the amortizing CLOs in our sample, the average level of senior note paydowns was 80.6%, ranging from 0% (no paydown yet) to 100% (senior note completely paid down and no longer outstanding). Because many of the senior tranches in these CLOs have paid down significantly, the senior notes on average can withstand the harsher downgrades and defaults scenarios and still pass our cash flow stresses at their current ratings compared to the reinvesting CLOs). For example, within the "20/40" scenario section of table 2, we see a much larger proportion of amortizing 'AAA' and 'AA' ratings (43% and 33%, respectively) experience affirmations relative to the reinvesting 'AAA' and 'AA' ratings (4% and 1%, respectively).

In table 3 below (excerpted from table 4 in the Appendix I section), we show that under a less punitive "10/20" scenario, the average notch movement for the 'AAA' and 'AA' notes are slightly lower for the amortizing cohort (relative to the reinvesting cohort), while the average notch movement for the 'A' and below rated notes are higher. This divergence in senior versus junior note performance mirrors our experience in surveillance of amortizing CLOs, where the same transaction can simultaneously see tranche ratings raised and lowered (see "CLO Rating Movements: Same Credits, But Different Strokes For Different Notes," published Nov. 20, 2019).

Table 3

Senior note paydowns generally have a positive impact on the cash flow analysis results for the senior notes, all else equal. However, for the junior CLO notes, the impact of senior note paydowns is not always positive. We generally find there is a selection bias in prepayments where loans from higher-rated companies tend to pay down first, especially during periods of high refinancing activity across the BSL market. As a result, residual CLO portfolios can become increasingly concentrated with loans from weaker issuers with lower ratings, causing the S&P Global Ratings' weighted average rating factor (SPWARF) and 'CCC' buckets of many amortizing CLOs to worsen with time.

As In The Previous Year, Our Scenario Analysis Shows CLO Structure Protecting Senior Noteholders

Our update of our scenario analysis study this year continues to show the fundamentals of the CLO structure protecting senior noteholders, as over 93% of the 'AAA' CLO notes see at most a one-notch downgrade even in the harshest of our scenarios, where loan defaults and CLO 'CCC' buckets reach 20% and 40%, respectively. Further, no 'AAA' or 'AA' CLO tranches default under any of the scenarios, and no 'A' tranche from a reinvesting CLO defaults until you get to the "20/40" scenario.

S&P Global Ratings rated our first CLO back in the mid-1990s, nearly 30 years ago, and CLO ratings have performed well (see "U.S. CLO Tranche Defaults As Of April 1, 2023," published April 7, 2023). In both our hypothetical rating stress scenarios and during periods of actual economic stress, the mechanics of the CLO structure have worked to protect the senior CLO notes. When the ratio of assets (including haircuts to the carrying values of some assets) to CLO notes outstanding drops below preset levels, O/C ratio tests are triggered. Failure of these tests can redirect cash flows away from equity (and, if the failures are significant enough, from lower-rated CLO notes as well) to reduce the balance of the senior CLO notes to restore credit enhancement.

This is exactly what happened to many CLOs in the early months of the COVID-19 downturn: as ratings on corporate loans issuers saw downgrades into the 'CCC' range, and CLO collateral pools started to experience par losses, O/C tests were triggered and proceeds redirected to the 'AAA' notes. In all, at the peak, we saw about 25% of our reinvesting U.S. CLO transactions fail one or more of their O/C tests in 2020, leading them to pay down an average their 'AAA' notes by an average of about 1% of the total tranche balance before the tests were passing again. Of the U.S. BSL CLO ratings that saw downgrades in 2020, 37.3% were subsequently upgraded.

Appendix I: Rating Category Transitions For Amortizing And Reinvesting CLO Transactions

Table 4

Appendix II: Additional Scenario Testing Information

For the purposes of our scenario testing, we generated a quantitative analysis for our sample CLOs using the same tools we use when rating the transactions. Our CDO Evaluator credit model assesses the overall credit quality of a portfolio of assets based on the rating and maturity of each asset, as well as the correlation between assets, and produces expected default rates at different levels of economic stress associated with our ratings. Our S&P Cash Flow Evaluator model is used to assess the ability of the tranches in each CLO to withstand loss rates under various interest rate and default timing scenarios. While ratings are assigned by committee and our criteria encompass a variety of qualitative and quantitative components, looking at the output of these two models for a given CLO portfolio and structure should provide an indication of the ratings a surveillance committee might assign following a given stress scenario.

The broad approach to the stress testing has two steps: first, apply the collateral stresses of defaulting and downgrading loans in each CLO portfolio to hit the default and 'CCC' obligor exposure outlined for the stress. The collateral default and downgrade stresses are applied to the CLO portfolios at time zero, so don't consider changes that might occur if they occurred over time (manager actions, the healing effects of coverage test paydowns, etc.) that might mitigate some of the rating impact of the stresses. Second, generate automated quantitative results (CDO Evaluator and Cash Flow Evaluator runs) across the entire sample of CLOs and see what ratings the quantitative analysis points to assigning. Although this excludes qualitative discussions that might occur in the context of a surveillance committee, it gives a good idea for each CLO of what the CDO Evaluator/Cash Flow Evaluator results reviewed by the surveillance committee would look like under a given stress.

When reviewing the stress scenarios, it's worth taking note of a few things:

• The stresses we selected for the scenarios outlined above are hypothetical and not meant to be predictive or part of any outlook statement.
• The stresses we selected aren't meant to calibrate to any of the economic scenarios we associate with our ratings in our "S&P Global Ratings Definitions," Jun. 9, 2023, article.
• The results are based on the application of the models we use to rate CLOs; a rating committee applying the full breadth of our criteria and including qualitative factors might in some instances assign a different rating than the quantitative analysis would indicate.

Also, for each of the scenario results tables we include a column for the instances where our cash flow results point to a rating below 'CCC-'. In some of these instances, the CLO notes still have overcollateralization greater than 100% and a surveillance committee might not lower the ratings to a nonperforming rating below 'CCC-' based solely on this. In practice, the quantitative results are only part of the rationale behind assigning a nonperforming rating to a CLO note. Typically, CLO ratings are lowered to 'CC', when the notes are significantly undercollateralized and payment in full on the legal final date of the CLO note is highly unlikely.

Related Research

• 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 Credit Distress Due To COVID-19 Could Affect U.S. CLO Ratings, April 24, 2020