(Editor's note: We created interactive charts summarizing the results of the stress tests done in connection with this article. If viewing this article on CapIQ or as a pdf, to access the charts, click the following link: version of article with interactive charts embedded.

Following the cessation of LIBOR for new transactions at the start of 2022, U.S. collateralized loan obligations (CLOs) have been in a state of benchmark rate transition. In contrast to CLOs issued in and before 2021, all new CLO transactions originated after 2021 have had their notes indexed to the Secured Overnight Financing Rate (SOFR), even as most of the corporate loans in their collateral pools remain indexed to LIBOR. At the same time, a small number of legacy CLOs have also transitioned the rate on their notes, leaving a mix of LIBOR- and SOFR-indexed CLO transactions in the market. Meanwhile, on the asset side, benchmark transition is also underway, but the pace has been slower than initially expected due to lower issuance levels in the corporate loan market in the current economic environment. As a result, U.S. CLOs are seeing their assets and liabilities transition to SOFR from LIBOR at varying rates. At the time of this publication, the average U.S. broadly syndicated loan (BSL) CLO transaction has about 12.5% of its assets tied to SOFR, with a handful of them having exposure of 30% to SOFR-indexed assets.

This has left U.S. CLO transactions with varying combinations of LIBOR and SOFR notes and assets, potentially introducing basis risk and a reduction in excess spread in some CLOs if the two rates were to diverge significantly. Basis risk isn't new to CLO transactions, with managers long having to deal with varying rates between the three-month LIBOR on their CLO notes and one-month LIBOR for a large majority of their assets. But since the excess spread generated by a CLO is a source of credit enhancement for the notes (it's used to pay down the notes in the event a coverage test fails) and a component of the modeling we do as part of the rating process, we decided to perform various rating stress analysis to assess how our CLO ratings might behave under different scenarios of excess spread reduction. We review several hypothetical stress scenarios we performed and their impact on our U.S. CLO ratings in this publication.

Setting The Stage For LIBOR Transition

On July 27, 2017, the U.K.'s Financial Conduct Authority (FCA) announced that LIBOR would no longer be available after 2021. This announcement followed concerns that the FCA raised around the sustainability of the LIBOR benchmarks and the absence of active underlying markets to support them. Since this July 2017 announcement, the financial markets have gotten together to plan this transition away from LIBOR, including the following key dates and milestones:

• 2008-2009: Large banks were found to have manipulated LIBOR during financial crisis.
• 2012-2015: Regulatory actions were taken against banks, with more than $9 billion in fines and several bank CEOs removed.
• July 2017: the U.K.'s Financial Conduct Authority (FCA) announced that LIBOR would no longer be available after 2021.
• Nov. 30, 2020: The Federal Reserve Board Federal Deposit Insurance Corp. and Office of the Comptroller of the Currency issued supervisory guidance encouraging banks to stop entering into new contracts that use USD LIBOR as a reference rate as soon as practicable and in any event by Dec. 31, 2021, in order to facilitate an orderly transition.
• March 5, 2021: the FCA officially confirmed the dates for the future cessation of LIBOR benchmarks, December for 4 currencies and pushing out deadline for major settings of dollar LIBOR to June 2023.
• April 6, 2021: New York passed a law to assist "tough legacy" contracts with LIBOR.
• Jan. 1, 2022: New transactions are expected to use a rate other than LIBOR.
• June 30, 2023: Remaining transactions referencing LIBOR will be required to transition to a new benchmark.

The securitized market has historically widely referenced LIBOR with approximately $1.9 trillion of LIBOR-based contracts outstanding. The U.S. CLO market represents roughly $800 billion of that, making it the sector with the highest exposure to the transition.

Assessing The Potential Impact Of The Transition On Excess Spread And CLO Ratings

Like many other securitized products, excess spread is a common feature of CLOs. Excess spread is generally the surplus of asset spread over a CLO's liability funding costs, including fees or amounts payable senior in the priority of payments waterfalls. During stressful periods, a CLO may divert excess spread away from equity or junior noteholders if it fails one or more of its coverage tests, thus protecting the senior notes and making excess spread a source of credit enhancement. If coverage tests are all passing, excess spread is typically released to a CLO's equity holders.

During the pandemic-driven economic downturn in 2020, about one-quarter of our rated U.S. BSL CLO transactions failed one or more coverage tests and diverted excess spread to reduce the balance of the senior notes outstanding.

Typically, CLO coverage test failures occur when deals suffer par losses and/or when obligors in the portfolio see their ratings lowered into the 'CCC' range (above a deal's permitted allowance) or default, causing one or more of the overcollateralization ratios to drop below its minimum required threshold. Interest coverage tests, if breached, can also divert excess spread away from the subordinated note. All else being equal, the more excess spread a CLO has the greater the amount of cash it will be able to divert when a coverage test fails, something that is captured in our CLO rating analysis as a potential source of support for the CLO tranches.

Because the LIBOR-to-SOFR transition introduces a new source of basis risk and excess spread deterioration to CLO transactions (in addition to the one-month LIBOR versus three-month LIBOR basis discussed above), CLOs could potentially see a reduction in excess spread as the transition process continues. Given that CLOs represent the largest Structured Finance exposure to LIBOR (both on the corporate loan side and liabilities side), we will focus on and test a few hypothetical stress scenarios related to this and their impact on our U.S. CLO ratings below. But first, let's review the typical processes governing the transition of legacy (pre-2022) CLO transactions to a new benchmark based on the language in their transaction documents.

CLO Liability Fallbacks Are Generally Robust

Because the eventual cessation of LIBOR was signaled well in advance by the regulators, the CLO market has had time to prepare for the transition. Legacy (pre-2022) CLO documents often (although not always) include "fallback language" governing the transition of the CLO's liabilities to a new, non-LIBOR benchmark. We reviewed S&P Global Ratings-rated CLO transaction documents to inventory and assess their liability fallback language. Because of the very active CLO issuance in 2021, including significant refinancing and reset activity, many CLOs have updated their documents and incorporated robust fallback language. In many cases, these CLOs have more robust language than found in other structured finance sectors with legacy LIBOR transactions.

We believe CLO fallbacks can be classified across three broad categories:

• Alternative Reference Rate Committee (ARRC)-like language: These fallbacks are identical or nearly identical to language recommended by the ARRC. These provisions generally contain three key components: a trigger event, a waterfall of new rates with SOFR as the first in priority, and a credit spread adjustment (CSA) to account for the fact that SOFR, unlike LIBOR, does not contain a credit risk component. Combined, they provide an objective way to transition away from LIBOR.
• Manager-limited discretion language: In these documents, the fallback language gives the manager limited discretion to select a replacement rate, usually one widely used in the underlying loan collateral. While SOFR is the likely replacement rate, this is not guaranteed per the indenture language (as the language is not "hardwired"). These documents typically also include a credit spread adjustment to be used (although not always precisely defined). Given the safe harbor created by the March 2022 federal law on LIBOR transition when a SOFR-based rate is selected, we believe there are incentives for managers with discretion over replacement rates to select the fixed spread adjustments referenced in the ARRC language, including a CSA of 26 basis points (bps) for three-month (quarterly pay) notes.
• Weaker or no fallback language: This category reflects a mix of legacy fallback language that weren't necessarily designed for the permanent cessation of LIBOR. They are concentrated among older CLO transactions (issued before 2018). If LIBOR is not available, the fallbacks may lead to fixing the rate at the last quoted LIBOR or other approaches including bank polling. Most of these options may be overridden by federal LIBOR law to become a SOFR-based rate (details pending Federal Reserve rules), with 11 bps and 26 bps credit spread adjustments for one- and three-month maturities, respectively. Finally, a handful of transactions could transition to a "pre-defined rate" such as prime or other rates (absent of manager's intervention). That last category is typically found in esoteric transactions and/or bilateral facilities where the documents are easier to amend.

Starting Jan. 1, 2022, we have not seen any new CLO transaction pricing or closing with rates other than SOFR as the reference rate for the liabilities. The universe of LIBOR-based transactions is expected to gradually shrink until June 2023, when LIBOR would cease to be reported and any remaining transactions referencing LIBOR would be forced to transition.

As of June 30, 2022, we have seen a small number of pre-2022 CLO transactions transition from LIBOR to SOFR + the 26 bps adjustment recommended by ARCC (through supplemental indentures or through notices).

Liability fallbacks vary by vintage, segment, and manager

Fallback language has generally improved over time, as shown in chart 1 below. Vintages shown below include the latest time when a CLO indenture was updated through a refinancing or reset. For example, a transaction that initially closed in 2015 and was subsequently refinanced in 2017 and reset in 2019 would be categorized as a "2019" deal because it is likely that the transition language may have been updated at that time. As a result, the number of older transactions is relatively low (especially the 2019 transactions that have been widely refinanced and reset over the past couple of years).

Chart 1

Approximately 64% of outstanding S&P Global Ratings-rated U.S. CLOs contain ARRC-like language in the liabilities, another 33% contain manager-limited discretion language, and the remaining 3% contain weaker or no fallback language. This shows a significant shift towards the ARCC language since our last report from March 2021, when we observed only 26% of rated CLOs containing ARRC-like language. (see "SF Credit Brief: With LIBOR Transition Approaching In U.S. CLOs, Fallbacks Remove Some Uncertainty," published March 22, 2021).

Additional flexibility to select a replacement rate exists among middle-market CLOs, where we observed that 43% of this group contain manager-limited discretion language, vs. 31% among broadly syndicated CLOs.

Chart 2

We also noticed differences in fallback language across the top 10 CLO managers (by S&P Global Ratings-rated deal count). While the ARRC-like language remains the most dominant fallback language amongst the largest managers (by S&P Global Ratings deal count), we see that they tend to be granted a little bit more flexibility in their fallback languages, with 44% of their deals relying on the manager-limited discretion language (11 points higher from the 33% for the entire universe).

Chart 3

Chart 4

With this background in mind, we have decided to analyze the impact and reliance on excess spread of our rated CLO tranches.

How might LIBOR transition affect CLO ratings?

U.S. CLO liabilities and assets (corporate loans) are transitioning to SOFR at different rates. Typically, a CLO's assets will transition gradually, while the CLO notes will transition to SOFR from LIBOR in their entirety. This could introduce basis risk in the transactions and affect the amount of excess spread available for individual CLOs. It is also theoretically possible that some CLO assets or notes could transition to a rate other than SOFR.

At this point, the market consensus is that the majority of the corporates loans and CLO liabilities would transition to the same index (SOFR). There are, however, some debates around the CSA as CLO tranches typically refer to the ARCC recommended 26 bps adjustments for legacy deals while corporate loans can "bake in" the adjustment that they believe is "fair" or "appropriate". A different CSA on the asset side could lead to a reduction in excess spread.

How much excess spread do 'AAA' tranches actually need to support the current rating assigned? Can a 10 bps reduction in excess spread resulting from the LIBOR transition to SOFR cause CLO downgrades? How sensitive are non-investment grade ratings to excess spread? We performed hypothetical stress scenarios to address these questions and others below.

Scenario background: universe, methodology, and exclusions

LIBOR transition has brought excess spread and basis risk to the forefront, as they may shift on outstanding CLO transactions as rates change on assets and liabilities. There are different ways to test ratings sensitivity to these changes. Our approach assessed the quantitative impact of a weighted average spread (WAS) reduction ranging from -5 bps to -50 bps (throughout the life of the deal) in five bps increments, while keeping all other assumptions unchanged. After modeling these changes, we then compared the model-implied ratings generated under each scenario to the current ratings of the tranches tested.

Before we start, it's worth highlighting a few caveats:

• The stress 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 any of our ratings.
• 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.

None of the scenarios are meant to predict any specific outcome for the collateral within CLO transactions or CLO liabilities. However, setting up the scenarios in the way we have, with a simple progression of increasingly severe assumptions, should allow CLO investors and others to take their own views among the scenarios and gauge the potential CLO rating outcome and BDR movement.

Our sample of transactions evaluated included 686 S&P Global Ratings-rated CLOs with approximately 3,700 tranches. About 96% of these were BSL CLO transactions, and 4% were middle-market CLOs. Across both CLO types, about 95% of our sample consisted of CLOs still within their reinvestment period and 5% consisted of amortizing CLOs. We excluded atypical CLO transactions, including those with turbo tranches, combination notes, and low weighted average lives (WALs) remaining. In our analysis, we did not consider a CLO manager's intervention and assumed a spread reduction throughout the life of the CLO.

The charts below show:

• The average potential rating movement (downgrade) under each scenario as well as the average breakeven default rate (BDR) reduction. BDRs are the output of the cash flow analysis we do on CLOs, and represent the maximum amount of cumulative asset defaults a given CLO tranche can withstand in our modeling while still paying all of the interest and principle due. The average potential rating movement takes into account the number of tranches impacted as well as the severity of their movements. For example, an average tranche downgrade of 0.1 would mean that, on average, the tranches in the given rating category universe would decline by 0.1 notches. This could indicate that 10% of the universe would see its rating lowered by one notch, for example or that 5% of the tranches would be lowered by two notches.
• The average BDR for a given rating category before and after a given decline in excess spread.
• The proportion of ratings lowered under the scenario (based solely on the model-implied rating after a given excess spread decline scenario). This chart does not show the severity of the rating's movement.

Scenario 1: 25 bps reduction in excess spread

The ARCC-like transition language (which is the most common fallback language found in two-thirds of the deals) typically references SOFR as the new base rate to be used following a benchmark transition event (BTE). It also typically adds a CSA of 26 bps to the cost of the debt. A CLO tranche paying LIBOR + 1.20% would therefore transition to SOFR + 1.20 + 0.26 (or SOFR +1.46).

Of note, not all "BTEs" are equal and defined in the same way. Some transactions (especially more recent ones) added the concept of the "asset replacement trigger" in the definition of a "BTE" meaning that liabilities could switch when the majority of the loans held in the portfolio have transitioned to SOFR. We noted that roughly 70% of the ARCC-like fallback languages included the concept of the asset replacement trigger.

This scenario for CLOs envisions that the assets transition to SOFR with no credit spread adjustment or no pickup in margin (in an efficient market, the CSA would reflect the transition to a risk-free index from a credit sensitive one). At maximum, this would reduce the excess spread by 26 bps (we will round it to -25 bps for our analysis). The chart below illustrates the scenario. These are conservative assumptions, in our view, and are not reflective of the market consensus of what the transition is likely to bring.

Chart 5

Under these assumptions, we noticed an average impact on the BDRs of 1.0%-1.8% for investment grade (IG) and 2.0% impact for non-investment grade (NIG) reinvesting-BSL CLO tranches. Even under this conservative scenario, we do not expect many (if any) 'AAA' or 'AA' tranches to show a different model-implied rating under our quantitative analysis and the scenario would likely not lead to a rating action. However, some 'BBB' tranches (10%) and more 'BB' tranches (30%) could show a different model-implied rating. The average rating movement of 'BB' tranches would be around 0.3 notches, which is relatively modest even under this conservative scenario. While this is the output of our quantitative analysis and not necessarily reflective of rating actions that would be decided upon by a committee taking into account qualitative factors, it is likely that a portion of CLO 'BB' and below tranches would be negatively impacted.

U.S. BSL reinvesting: 25 bps reduction in excess spread 

Chart 7A

Chart 7B

Chart 7C

Of note, there is a correlation between the BDR impact and the reinvestment status and/or the numbers of years left in the reinvestment period. The closer the transaction is to the end of its reinvestment period, the lower the BDR impact tends to be. For example, the average 'BBB' BDR movement under this -25 bps scenario is at 1.80% for reinvesting deals vs. 1.60% for amortizing deals (see chart below).

However, amortizing transactions tend to have thinner cushions to begin with (cushions are the difference between breakeven default rates and scenario default rates (SDRs), and the SDR is an indicator of a CLO portfolio's credit risk derived in our analysis of CLOs), which explains the added ratings volatility displayed in the interactive charts. We typically see qualitative factors play an important role when reviewing amortizing deals.

U.S. BSL amortizing: 25 bps reduction in excess spread 

Chart 8A

Chart 8B

Chart 8C

Finally, a tranche's starting cushion plays a major role in the rating's movement analysis we are performing. Tranches with tighter rating cushions would have less room to absorb excess spread reduction.

The impact on middle-market (MM) transactions tends to be lower (see charts below) as they structurally have additional credit enhancement and therefore tend to rely less on excess spread. They also tend to have significantly larger cushions to begin with.

U.S. middle-market reinvesting: 25 bps reduction in excess spread 

Chart 9A

Chart 9B

Chart 9C

Scenario 2: 10 bps reduction in excess spread

In this scenario, we stress the ratings by reducing excess spread by 10 bps (assuming tranches transition to SOFR + 26 bps while assets only get a pickup of 16 bps, which is the basis between LIBOR and SOFR we observed in June 2022, leading to an -10 bps stress). The chart below illustrates the scenario.

Chart 6

Under this scenario, only a small portion of reinvesting BSL NIG tranche ratings would be potentially negatively impacted, and IG tranche ratings are expected to remain very stable. The average change in BDRs for this same universe is around 40-70 bps for IG tranches and 80 bps for NIG tranches.

Shifting to MM CLOs, we see even more rating-resiliency under this stress scenario given the extra credit enhancement these deals are issued with. The average BDR impact is also slightly lower for amortizing MM CLOs versus reinvesting.

U.S. BSL reinvesting: 10 bps reduction in excess spread 

Chart 10A

Chart 10B

Chart 10C

This study shows a clear (somewhat linear) correlation between excess spread and BDRs for reinvesting BSL CLO transactions, with an increasing impact on spread reduction as you move further down the CLO capital structure. This highlights the increasing reliance on excess spread as you go down the ratings spectrum. NIG tranches rely more on excess spread than IG tranches.

LIBOR Transition To Have Limited Impact On CLO Excess Spread And Ratings

In terms of our CLO ratings, we expect the transition away from LIBOR to have, at most, a limited impact on CLO excess spread, with the impact being primarily felt by unrated CLO equity. While there may be some hurdles to overcome on the operational side (as thousands of corporate loans and hundreds of CLO tranches will need to transition over the next few payment period cycles), we do not anticipate material CLO rating movements due to this upcoming transition. CLO managers are experienced and currently manage basis risk (with one- and three-month LIBOR) that has potential excess spread implications. While a spread reduction could occur, it is unlikely that it would last for the life of the transaction as managers have the ability to intervene (by either purchasing assets with higher margins and or reducing the WACD through a refinancing or a reset); and in any case, the transition should be complete as of July 1, 2023.

In summary, our stress test analysis highlights the fact that IG CLO tranche ratings primarily rely on subordination and recoveries. Even when significantly reducing the excess spread, we do not see IG ratings move significantly. NIG tranche ratings rely more on excess spread. However, under a moderate scenario assuming a 10 bps reduction of excess spread, we do not anticipate many NIG CLO tranches to be negatively impacted.