The combined impact of COVID-19 and the collapse in oil prices present speculative-grade debt issuers (those rated 'BB+' and lower) with a challenge unlike anything seen in the history of the leveraged lending market. The deterioration of ratings assigned to corporate loan issuers has repercussions in the ratings mix of obligors found within U.S. CLO collateral pools. In the space of less than two months, since early March, U.S. CLOs have seen their average collateral credit quality drop to below 'B', and their average 'CCC' bucket increase to nearly 12% from about 4%. These changes are putting pressure on CLO ratings, especially for lower mezzanine and subordinate tranches. For U.S. CLOs, rating stress scenarios we have generated across our rated book suggest that even under dire circumstances, 'AAA' rated CLO notes are well protected and can withstand upward of 60% of their collateral loans downgraded (40% to 'CCC', and 20% to default at an average recovery rate of 45%) without suffering a loss. Under more plausible (but still harsh) stress scenarios, a large majority of senior U.S. CLO notes maintain current ratings, while tranches rated 'BBB' and lower bear the brunt of the downgrades and defaults.

Even before the current downturn, the ratings mix of obligors in U.S. CLO collateral pools had seen a downward migration in recent years, mirroring changes seen in the overall corporate loan market (see "To 'B-' Or Not To 'B-'? A CLO Scenario Analysis In Three Acts," published Nov. 26, 2019). Given the surge of recent corporate downgrades, we have already surpassed the stressed assumptions we used for the first scenario we presented in the November 2019 article.

Stress Scenario Breakdown Summary

For our current round of CLO stress testing, based on feedback from CLO market participants, we generated four stresses of increasing severity using relatively simple assumptions. To round these out, we also generated an additional scenario (Scenario Five) under a different set of assumptions:

• Scenario One (5/10 scenario): 5% of the assets default and 10% get lowered to 'CCC';
• Scenario Two (10/20 scenario): 10% of the assets default and 20% get lowered to 'CCC';
• Scenario Three (15/30 scenario): 15% of the assets default and 30% get lowered to 'CCC';
• Scenario Four (20/40 scenario): 20% of the assets default and 40% get lowered to 'CCC'; and
• Scenario Five ('CCC' default/one notch downgrade scenario): We lower every obligor rating by one notch and assume every obligor currently rated in the 'CCC' range defaults.

For all of the assets we default in the above stress scenarios, we assumed a recovery rate of 45%. This is significantly lower than historical first-lien senior secured loan recoveries over the past several decades (75%-80%) and the recovery rates inferred by our current mix of '1+' through '6' recovery ratings assigned to corporate loans (mid- to low-60% range). For assets that fell into the 'CCC' excess amount, we assumed a market value of 60% (for most broadly syndicated CLOs (BSL CLOs), loans from 'CCC' rated obligors above an allowable 7.5% bucket are carried at market value for purposes of calculating the CLO overcollateralization ratios.

None of the above scenarios are meant to be predictive of any specific outcome for the collateral within CLO transactions. But setting up the first four scenarios in the way we have, with a simple progression of increasingly severe assumptions, should allow CLO investors and others to take their own view among the four scenarios and gauge the potential CLO rating outcome. The fifth scenario stress was added to address questions we've received from market participants, and to test a simple benchmark level of stress and provide comparisons to future scenario analyses. (See chart.)

Broad Themes From The Scenario Analysis

CLO senior tranche ratings show resilience in the scenarios

Our analysis shows the fundamentals of the CLO structure protecting senior noteholders, with almost 99% of CLO 'AAA' tranche ratings either seeing affirmations or one-notch downgrades even under our most punitive scenario (20% of obligors default, 40% get lowered to 'CCC'). Outcomes for tranches further down the CLO capital stack depend upon the severity of the assumptions applied, but no tranche rated 'A' or higher defaulted under any of our scenarios. We should also note that the 'AAA' CLO tranche downgrade rate in the real world would likely be lower than indicated in the transition tables shown later in this report. This is because rating committees take into account qualitative considerations, our expectations for future senior tranche paydowns, and other factors beyond the results of the quantitative analysis when considering potential rating actions.

Same stress, different outcomes

We see that across the CLO transactions, there can be a wide range of rating outcomes even under the same stress. For example, after applying the 10/20 scenario to our sample of 434 CLOs, we see downgrade outcomes for the 'BBB' category CLO notes ranging from an affirmation (i.e., zero downgrade notches) to a seven-notch downgrade. This is evidence of the wide range of CLO portfolios and CLO structures across our sample. (See table 1.)

Table 1

As expected, some deals will experience greater impact within their portfolios from these stresses--in particular, the pre-2015 vintage portfolios, which have a higher average exposure to 'CCC' obligors to begin with (these deals will have a higher proportion of the issuers we will downgrade, especially in the 5/10 and 10/20 scenarios). If a deal has higher exposure to weaker ratings (in the 'CCC' category), they will have a higher likelihood of seeing a greater proportion transition to default. There are also other nuances (such as 'B-' exposure) that will affect the portfolios in different ways, which will translate into a divergence in the impact to CLO ratings per rating category.

Vintage matters, but only to a point

Many of these deals, in particular the 2015 and prior vintage CLOs (over one-fourth of our CLO sample) have already gone through the energy slowdown, during which they suffered par loss and credit deterioration in their portfolios. Many of these CLOs have been reset in recent years and are still reinvesting. As such, there was a pronounced vintage effect in the less punitive scenarios, with new-issue CLOs originated in 2018 and 2019 performing more strongly than earlier vintages (including CLOs that were originated in earlier years but reset or refinanced in 2018 or 2019). The earlier vintage portfolios tend to have a greater exposure to loans from 'CCC' rated companies, and it shows in the scenario analysis. Note that the vintage effect is much more noticeable in the less punitive scenarios; in the more stressful scenarios, the vintage effect is overwhelmed by the assumed level of corporate downgrades and defaults.

Some Assumptions And Caveats

For 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 rating levels. Our S&P Cash Flow Evaluator model is used to assess the ability of the tranches in a given 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 decide upon following a given stress scenario.

Also, it's worth highlighting a few caveats:

• The stresses we selected for the scenarios outlined below 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 "Understanding S&P Global Ratings' Rating Definitions," June 3, 2009.
• 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 arrive at a different rating outcome than the quantitative analysis would indicate.

Running The Stress Scenarios

To create our sample of CLO transactions, we focused on 434 U.S. BSL CLO transactions rated by S&P Global Ratings that were reinvesting as of early-March 2020. We excluded CLO X notes and combo notes. Due to a large volume of CLO refinancings and resets in recent years, there is a wide range of original vintages among our sample of reinvesting CLOs. The majority of our CLO sample were either newly issued in 2018/2019 or were reset in 2017/2018/2019. The starting point for the simulation was March 2020, ahead of the recent wave of corporate rating actions related to COVID-19 and the recent distress in the energy sector, at which point our sample of 434 CLOs collectively had exposure to 2,232 loans from 1,546 corporate issuers.

The S&P Global Ratings Weighted Average Rating Factor (SPWARF) of the portfolios across our CLO sample at that time was 2675, implying an average CLO portfolio rating that is somewhere between 'B+'and 'B' (closer to 'B'). SPWARF is a numeric representation of the rating quality of a CLO collateral pool, which a higher number indicating a lower average rating for a pool. Our 434 CLO transaction sample also had approximately 21.34% exposure to issuers rated 'B-', 4.84% to loans from obligors in the 'CCC' category, and 0.64% exposure to loans from nonperforming obligors. We should note that all of these metrics have been affected by the large number of corporate downgrades in recent weeks. (See table 2.)

Table 2

Given the early March ratings across the 1,500+ issuers, to apply the stress scenarios, we notched the ratings from the weakest on up to achieve the target 'CCC' and default exposures, on average, across our sample of CLOs. (See table 3.)

Table 3

After adjusting an increasing portion of the issuers under each scenario, again from the weaker ratings first, we can see the impact on the exposures across our CLO sample. For example, within the 5/10 scenario, we lowered 214 ratings such that the average 'CCC' and default exposure across the 434 CLOs was about 10% and 5%, respectively.

As noted earlier, Scenario Five is very different from the other four stresses. This scenario sees a greater magnitude of issuers being downgraded (over 96%--not 100% because 51 issuers already had nonperforming assets), but the impact is only one notch. A one-notch downgrade on a very large number of issuers will result in very different CLO rating outcomes, as we detail below.

The Results

Scenario One--the 5/10 scenario (5% of the assets default, 10% get lowered to 'CCC')

In this scenario, about 15% of assets held across CLOs experience downgrades; 214 issuers are downgraded by an average of 1.7 notches (118 issuers rated within the 'CCC' category default, 39 'CCC+' rated issuers experience a one-notch downgrade to 'CCC', and 57 'B-' rated issuers experience two-notch downgrades to 'CCC'). As a result, average 'CCC' buckets across our sample increased to 10.0% from 4.8% at the start of March, while average nonperform buckets across our sample increased to 4.9% from 0.6%, roughly achieving the 5/10 ratio between defaults and 'CCC's.

In this scenario, assuming a market price of 60 for excess 'CCC' haircuts and 45 for nonperforming exposures, just under half of the deals within our sample will fail one or more of their overcollateralization (OC) tests. We see vintage has a big impact on the OC tests within this scenario: A majority of 2015 and prior vintage deals fail, while, on the flip side, a majority of 2019 vintage deals pass (see table 4).

Table 4

After modeling this 5/10 scenario within our quantitative process, we captured the rating transitions (see table 5). Most of the downgrades are concentrated within the junior CLO notes; just over half of the 'B' category see downgrades, while about one-third of the 'BB' category see ratings lowered. A large majority of each of the investment-grade rated CLO notes (rated 'BBB- (sf)' and higher) do not experience a downgrade. Because 84% of the 'BBB' category CLO notes are at the 'BBB- (sf)' level, a one-notch downgrade resulted in a speculative grade rating of 'BB+ (sf)'.

Table 5

As described above, the wide distribution of 'CCC' and default exposures across the sample led to a wide distribution of notch movements (particularly for the speculative-grade ratings). Because we assumed the issuers with weaker ratings would experience downgrades, the transactions with higher 'CCC' and 'B-' exposure as of the start of March (i.e., the older vintage deals) were more likely to have exposure to more of the downgrades and defaults. For example, the 'BB' category notes of the 2015 and prior deals experienced a downgrade of about 1.7 notches on average, while the average notch down for 2019 vintage deals was just 0.2 notches (indicating most are affirmed). The 'B' category across vintage is not a great proxy since there were very few 'B's issued in 2019. (See table 6.)

Table 6

As we look up toward the senior part of the capital structure, we do see some 'A' downgrades, but none transitioned into the speculative-grade category in this scenario. There were some one-offs of 'AA' and 'AAA' ratings that saw downgrades. There was one 2019 vintage 'AAA' note that saw a one-notch downgrade under this scenario. This deal had a low 'CCC' bucket (about 2%); however, it had a very large exposure to 'B-' issuers (over one-third) at the start of March. The high 'B-' exposure resulted in a large increase in 'CCCs' under this scenario, even though defaults were low. The significant shift in overall credit quality of the portfolio caused the senior notes to transition down by one notch to 'AA+' (the 'AA' note that transitioned to 'AA-' was also from the same 2019 vintage deal).

There was a 2016 vintage that also saw its 'AA' note transition down to 'AA-' within this scenario. This deal had above-average exposure to 'CCC' and below-average subordination at the 'AA' level. As a result of this stress, several of the 'CCC' issuers wound up defaulting in this scenario, and with lower 'AA' subordination to begin with, this particular 'AA' note experienced a one-notch downgrade.

Scenario Two--the 10/20 scenario (10% of the assets default, 20% get lowered to 'CCC')

In this scenario, about 30% of assets held across CLOs experience downgrades; 544 issuers are downgraded by an average of 2.1 notches (all 157 issuers rated within the 'CCC' category default, along with 20 'B-' issuers; 359 'B-' and eight 'B' rated issuers experience two and three notch downgrades, respectively, to 'CCC'). As a result, average 'CCC' buckets across our sample increased to 20.1%, from 4.8% at the start of March, while average nonperform buckets across our sample increased to 10.0% from 0.6%, roughly achieving the 10/20 ratio between defaults and 'CCC's. (See table 7.)

Table 7

Given this 10/20 scenario and assuming a market price of 60 for excess 'CCC's and a recovery of 45% for nonperforming, we see that just about all OC tests will fail at this point, regardless of vintage (a large majority of the 'AA' OC tests fail, meaning all deferrable notes will likely be deferring in this scenario).

In this 10/20 scenario, the amount of exposure experiencing a downgrade has doubled, and the magnitude of the 544 downgrades has increased to 2.1 notches (from 1.7 notches in the 5/10 scenario). Assuming a 45% recovery for defaults, par loss increases to 5.5% on average across the sample.

We see a large majority of 'BB' and 'B' category CLO ratings experience multi-notch downgrades, many going into the 'CCC' category or lower (suggesting these CLO notes may be at risk of default at some point within their lives). About half of the 'BBB' category CLO notes experience a one-notch downgrade, and because most of these notes are rated 'BBB-(sf)', a large majority end up with a speculative-grade rating within this stress. About two-thirds of the 'A' category CLO notes see downgrades, though none lose their investment-grade rating. A large majority of the nondeferrable 'AAA' and 'AA' notes retain their current ratings. (See table 8.)

Table 8

As in the 5/10 scenario, there is a strong vintage effect within this stress as well. The junior and mezzanine CLO notes of the earlier vintages experience greater rating movement relative to the 2019 vintage. The 2019 vintage deals that saw downgrades within their senior notes were more likely to have had greater exposure to 'B-' rated issuers.

Scenario 3–the 15/30 scenario (15% of the assets default, 30% get lowered to 'CCC')

In this scenario, about 45% of assets held across CLOs experience downgrades; 625 issuers are downgraded by an average of 2.4 notches (all 157 issuers rated within the 'CCC' category and 56 'B-' issuers default, while 323 'B-' rated and 89 'B' rated issuers experience two- and three-notch downgrades, respectively, to 'CCC'). As a result, average 'CCC' buckets across our sample increased to 29.8% from 4.8% at the start of March, while average nonperform buckets increased to 14.9% from 0.6%, roughly achieving the 15/30 ratio between defaults and 'CCC's. (See table 9.)

Table 9

As with the 10/20 scenario, assuming a market price of 60 for excess 'CCC's and a recovery of 45 for nonperforming in this 15/30 scenario, just about all OC tests fail, regardless of vintage (all 'AA' OC tests fail, meaning all deferrable notes will likely be deferring in this scenario).

It's no surprise that as we dial up the stress, the average notch movement increases across all rating categories. In this scenario, the majority of 'BB' and 'B' category notes do not pass our 'CCC-' rating stresses, meaning they are at risk of default at some point in their lives. Almost all 'BBB' category notes are downgraded and a small portion (17%) of 'A' category notes are downgraded into speculative grade. Most AA category notes see multi-notch downgrades (average notch movement of 1.4), but all remain investment grade. About half of 'AAA' rated notes still do not see downgrades.

There were some deals with current 'BB', 'BBB', or 'A' category ratings that didn't transition in this 15/30 scenario. In this study, we used the current rating as the start point and applied the stress to measure the impact. These notes happened to have greater cushion at their current ratings before the stress. Rating cushion can originate at closing (transactions structured with less leverage/higher subordination, all else equal) or rating cushion can be built during the reinvestment period (for example, par build or credit improvement). We see that rating cushion can be a source of ratings stability even in a period of stress.

Scenario 4–20/40 scenario (20% of the assets default, 40% get lowered to 'CCC')

In this scenario, about 60% of assets held across CLOs experience downgrades; 902 issuers are downgraded by an average of 2.7 notches (all 157 issuers rated within the 'CCC' category and 114 'B-' issuers default, 265 'B-' and 366 'B' rated issuers experience two- and three-notch downgrades, respectively, to 'CCC'). As a result, average 'CCC' buckets across our sample increased to 40.0% from 4.8% at the start of March, while average nonperform buckets across our sample increased to 19.9% from 0.6%, roughly achieving the 20/40 ratio between defaults and 'CCC's in this scenario.

The significant credit transition causes the average SPWARF of the sample to increase to 4760 (between 'B-' and 'CCC+') from 2675 (between 'B+' and 'B').

In this scenario, all ratings experience significant downgrades. Just about all 'BB' and 'B' category CLO notes will be underwater in this stress. Most of the 'BBB' category notes will also transition to the 'CCC' category and may potentially default at some point in their lives. The 'A' category CLO notes mostly transition to speculative grade. Just about all 'AA' category notes remain investment grade (just one rating transitions to 'BB+', the same 2019 vintage deal with over one-third exposure to 'B-'). About 10% of the 'AAA' notes still pass under this stress, while about 90% of the 'AAA' ratings experience a one-notch downgrade; only a handful (about 1% of the 'AAA's) experience a greater than one-notch downgrade. (See table 10.)

Table 10

Scenario 5–the widespread one-notch downgrade and 'CCC' default scenario

In this scenario, all issuers rated 'B-' or higher experience a one-notch downgrade, while all 'CCC' category issuers default (all 157 issuers rated within the 'CCC' category default and 1,338 issuers rated 'B-' or higher are downgraded by one notch). As a result, average 'CCC' buckets across our sample increased to 21.3% from 4.8% at the start of March, while average nonperform buckets across our sample increased to 5.5% from 0.6%.

The significant credit transition causes the average SPWARF of the sample to increase to 3632 (around 'B-') from 2675 (between 'B+' and 'B'). The change in SPWARF of this stress and the 'CCC' buckets is similar to the 10/20 stress, where the SPWARF increased to 3567. Although the SPWARF may be similar, they got there very differently, with 30% of issuers experiencing an average downgrade of 2.2 notches in the 10/20 scenario versus just about all issuers experiencing one-notch downgrades in this scenario. This stress also yields very different 'CCC' to default ratio--almost 4:1 instead of 2:1 in the 10/20 stress.

The CLO rating transitions from this stress are very different from the 10/20 stress described above. Given the lower levels of default and par loss, the junior notes ('BBB', 'BB' and 'B' category) see a lower severity of downgrades. Meanwhile, due to the broader drop in overall credit quality (since about all issuers experience a downgrade as opposed to just 30% in the 10/20 stress), the higher increase in SPWARF of this stress has greater severity on the senior notes ('AAA', 'AA', and 'A' category). About 29% of the 'AAA' notes experience a one-notch downgrade under this stress, compared to just 8% in the 10/20 stress. (See table 11.)

Table 11

Where We Are Headed

The current round of corporate rating actions has affected CLO collateral pools more than any time since 2008-2009, and at a much more rapid pace. Corporate rating actions during the financial crisis played out over six or more quarters, while with COVID-19, we have seen a greater shift in CLO collateral pools over the past month and a half than during any other comparable period in the history of the CLO market.

The economic stress caused by COVID-19 and the related social distancing measures (not to mention the stress in the energy sector related to the precipitous decline in oil prices) can play out in many different ways. But in all of the wide range of stress scenarios outlined in this article, the fundamentals of the CLO structure work to protect the senior tranches. Projected rating changes at the 'AAA' CLO tranche level show nearly 99% of the ratings either being affirmed or downgraded by one notch (to 'AA+') even under our most punitive hypothetical scenario of 20% loan defaults and 40% CLO 'CCC' buckets. And no CLO tranche rated in the 'A' category or higher experienced a default under any of our hypothetical scenarios. Further down the capital stack, the results become more negative (as expected under stresses as significant as these), with 'B' and 'BB' rated CLO tranches showing the most adverse outcomes and 'BBB' and higher rated tranches showing successively better results.