The global coronavirus outbreak has provoked questions and concerns from market participants surrounding the ultimate credit impact on European collateralized loan obligations (CLOs) (see "European CLOs: Assessing The Credit Effects Of COVID-19," published on March 25, 2020). In the first quarter of the year, downgrades among speculative-grade companies rose markedly, particularly in March. Following this increase in negative rating actions on nonfinancial corporates, we looked at their impact on CLO portfolio exposures (see "How COVID-19 Changed The European CLO Market In 60 Days," published on May 6, 2020).

Against this backdrop, S&P Global Ratings has taken a closer look at the recent evolution of European CLO overcollateralization (OC) ratios and what's behind their decline. We have also conducted a forward-looking scenario analysis to shed light on how European CLO OC ratios may evolve and when junior notes may start deferring interest.

Average OC Cushions Remain Positive, But One Deal Is Breaching The Junior OC Test

Charts 1 and 2 show the evolution of the senior and junior OC test cushions over the last six months for 131 reinvesting European CLOs that we rate and for which we have received the latest trustee reports as of June 30, 2020. Appendix II shows the evolution of OC test cushions across differently rated tranches (figures are calculated from reported trustee data).

OC cushions were stable until February 2020, with the average senior OC cushion at 9.4% and the average junior OC cushion at 4.2%. Since February, average senior OC cushions have fallen by about 1.3% to 8.2%, and junior OC cushions by 1.1% to 3.1%. Although average cushions remain healthy, OC cushions have dispersed more across transactions, with the minimum junior OC cushion falling by 2% to -0.1% from 1.9%, while the maximum junior OC cushion fell by 0.1% to 5.4% from 5.5%.

Our sample includes new transactions that have gone effective and for which we had received trustee reports from March. On average, these transactions had slightly higher OC cushions. Excluding such transactions, we observe even wider dispersion with the average junior OC cushion falling by 1.4% to 2.8%, the minimum junior OC cushion falling by 2%, and the maximum junior OC cushion falling by 0.5% to 5.0% from 5.5%.

Chart 1

Chart 2

To date, three European transactions (Accunia European CLO II, Barings Euro CLO 2018-1, and GLG Euro CLO IV DAC) for which we have received reports are failing their reinvestment OC tests (otherwise known as the interest diversion test). The proportion of transactions with an OC cushion of less than 2% increased to 28% in May from 3% in February (see chart 3). These tests are calculated in the same manner as the junior OC test, except that the threshold is set at a higher level. The interest diversion test differs from the OC test in that a percentage of interest proceeds is diverted to purchase additional collateral rather than to deleverage the transaction.

Nevertheless, failing reinvestment OC tests can indicate which transactions may start to fail their junior OC tests. In chart 4, we see that one transaction (GLG Euro CLO IV DAC) breached its most junior OC cushion and that the proportion of transactions with a cushion of less than 1% increased to 12% in May from 0% in February. The percentage of transactions with a cushion of between 1% to 2% also increased during the same period to 7% in May from 2% in February, bringing the total percentage of transactions with a cushion of less than 2% to 19% in May.

Chart 3

Chart 4

'CCC' Haircuts Are Not the Only Reason For The OC Numerator Decline

OC numerators are calculated using the transaction's starting collateral balance (which could change as a result of manager trading gains or losses) less haircuts applied for the following assets:

• 'CCC' excess obligations, which is the proportion of 'CCC' assets over a certain threshold (typically 7.5%), carried at market value (for more details on how an increase in 'CCC' assets affects coverage tests, see https://www.spglobal.com/ratings/en/research-insights/videos/20200324-clo-simplified-how-cccs-impact-coverage-tests);
• Defaulted obligations typically carried at the lower of the market value and the S&P recovery rate;
• Discount obligations typically carried at the purchase price; and
• Long-dated obligations, carried with a haircut to account for the market value risk of liquidation at maturity.

Given the restrictions in CLO documentation on managers' ability to purchase long-dated assets, and that we have yet to see assets being restructured to become long-dated assets, we have not seen long-dated asset haircuts in OC numerators.

Since February, downgrades of speculative-grade companies whose debts back European CLOs rose markedly, corresponding with an increased proportion of assets rated in the 'CCC' category ('CCC' assets). In June, the proportion of 'CCC' assets rose to an average of 9.8% from 1.9% in February (see chart 5). During this period, the proportion of defaulted assets also increased to an average of 0.32% from 0.04% (see chart 6).

Chart 5

Chart 6

To understand the reasons behind the OC cushion decline, we looked at a sample of 97 reinvesting European CLOs that we rate, for which we received both February and April trustee reports. From our sample of reinvesting European CLOs, 88% of the transactions experienced a negative OC numerator change, whereas 12% saw increased OC numerators due to the build-up of par with no haircuts.

While an increasing proportion of 'CCC' assets and the resulting haircuts have often been quoted as the reason for declining OC cushions, our analysis showed that an increase in 'CCC' assets was not the only reason. Of the transactions experiencing OC numerator declines, only 54% of them were affected by 'CCC' haircuts. Nearly 80% of deals experienced declines directly as a result of par loss due to manager trading, although 20% benefitted from trading gains. In addition, more than one-third of the transactions had haircuts due to defaulted assets and a quarter had some form of a discount obligation haircut.

Table 1

On Average 'CCC' Haircuts Contributed To 40% Of The OC Numerator Decline

The above only gives a snapshot of whether the OC numerator decline was due to trading activity, 'CCC' haircuts, defaulted obligation haircuts, or discount obligation haircuts but does not explain the extent of the impact of each reason for decline.

Charts 7 and 8 show the average magnitude of each factor contributing to the OC numerator change between February and April. For the average European CLO that had a negative numerator change, we attribute 40% of the decline to 'CCC' haircuts, 35% to trading losses, with 17% due to defaults and 8% to discount obligations. Nevertheless, the key reason for the OC numerator decline differed across transactions.

Ranking transactions with the highest decline in OC numerator to the lowest, European CLOs that fell in the first quartile and with the largest decline at 2.3% were most affected by 'CCC' haircuts, which contributed to 72% of the OC numerator decline, defaults 20% and trading 7%. On the other hand, for transactions that fell in the fourth quartile and with an average decline at 0.1%, trading losses were the key factor, with 'CCC' haircuts only totaling 19%.

Chart 7

Chart 8

What Could Cause More OC Test Failures?

To date we have seen one European CLO fail the most junior OC test. Although 'CCC' haircuts are not the sole driver of test failures, they do contribute significantly to OC numerator declines.

Based on our analysis of our rated universe of reinvesting European CLOs where we have assumed a market value of 70% for the 'CCC' assets, 25% would need to see an increase in 'CCC' assets between 0% and 8% to fail their 'B' OC test. Fifty percent of our universe would need to see an increase of between 8% and 14% to fail their 'B' OC test.

At more senior OC levels, we see more dispersion between transactions. One CLO would need to see a 1% increase in its 'CCC' assets to cause a breach of its 'BB' OC test, all else being equal; while one CLO would need to see a 27% increase from its current holdings (see chart 9).

Chart 9

We also reviewed a number of scenarios to see what percentage of our European CLO universe would start failing their OC tests. Table 2 shows that, all things being equal, a 7.5% increase in 'CCC' assets will result in the failure of a small percentage of some senior OC tests. A 10% increase in 'CCC' assets will lead to more than a quarter of transactions failing their 'B' and 'BB' OC tests, with a 15% increase in 'CCC' assets resulting in more than two-thirds failing.

While defaults have not to date been the key reason for average OC numerator declines, they nevertheless remain an important consideration. Of the three transactions that are currently failing their reinvestment OC tests, a combination of 'CCC' assets and defaults were the main contributors to the failures (the defaulted assets haircut contributed 68% of the OC numerator decline for GLG Euro CLO IV and 46% of the decline for the Barings Euro CLO 2018-1 transaction). From table 2, assuming a 35% recovery rate, a 5% increase in defaults would lead to half of the transactions failing their 'B' OC test and more than a quarter to fail their 'BB' OC test. With an 8% increase in defaults, this changes to 100% and 95% respectively.

Table 2

Excess Spread Provides Respite Before Deferrals

Does a breach in coverage tests matter and will it necessarily lead to deferred interest? What are the consequences of the breach of an OC test in a CLO transaction? The main purpose of OC tests is to protect the senior notes against defaults, losses, and deterioration of the credit quality of the portfolio (i.e., an increase in 'CCC' rated assets due to trading or downgrades, or purchase of discounted assets).

The breach of an OC test may not affect the transaction at all if the breach occurs in the middle of a period and is cured before the payment date (for example by trading gains, or by manager contributions).

If the OC test is still breaching on the payment date, interest proceeds would typically be used to amortize the notes sequentially, until the OC test, when recalculated, is no longer in breach. If insufficient interest proceeds are available to cure the tests, then principal proceeds may be used.

Because the curing of the OC tests stands in the waterfall before payment of interest on the deferrable notes, failure of OC tests may lead in the first instance to the equity no longer receiving payments, then the junior notes deferring.

As shown in table 3, the extent to which this may happen depends on two factors:

• The severity of the breach, i.e., what proportion of interest proceeds or principal proceeds will the issuer need to cure the breach; and
• The amount of excess spread, i.e., what proportion of proceeds will be left to pay interest on the deferrable notes and distribute to equity after the OC tests have been cured.

We have taken three transactions with different levels of excess spread and OC cushions (see table 3), and tested what percentage of 'CCC' assets each of these transactions would need to start deferring payments to junior notes and stop payments to the equity (all else being equal) (see tables 5, 6, and 7).

The OC cushions reflect trading gains or losses but do not include any discounted or 'CCC' haircuts. Each of these CLOs would start to fail their OC tests from the following levels of 'CCC' holdings.

Table 3

Table 4

Table 5 shows the breakeven amount of 'CCC' assets, everything remaining constant, from which the equity and junior notes in these CLOs would stop receiving payments. We have recorded the amount from which payments would be missed for one payment date. However, in most cases, payments resume on the next payment date, so payments would only stop for several periods if there were further OC deterioration. (For further details on our methodology for this scenario analysis, please refer to Appendix II.)

Table 5

We then varied the level of weighted-average spread (WAS) to test the impact on the risk of deferral.

Table 6

Although we considered the excess spread as the WAS minus the weighted-average cost of capital (WACC) to perform this analysis, the distribution of margins for each tranche, and whether the deal includes fixed rate notes, would also affect the results.

Even though a higher WAS allows the OC test to deteriorate further without affecting the payment of interest on the junior notes and the payments to equity, its impact is limited. The actual OC cushion tests provide a greater insight on when we think the junior notes will start to defer and when the equity will stop receiving payments (see table 7).

Table 7

Please see Appendix II for further scenario analysis.

Is A Breach In OC Test Triggers A Bad Thing?

European CLOs benefit from structural mitigants. Just as there are principal deficiency ledgers and interest deferral triggers in other asset classes such as residential mortgage-backed securities (RMBS), OC test triggers in CLOs protect noteholders by diverting interest to either purchase more collateral or toward the paydown of the transaction.

The impact of an OC test breach will be different for noteholders depending on their position in the capital structure, with the senior noteholders benefitting from the transaction's deleveraging, whereas depending on the transaction's available excess spread, equity noteholders may stop receiving equity distributions for that period and junior noteholders a deferral of interest.

In addition, not all OC tests are the same and the lack of an OC test breach may not necessarily mean that a transaction is outperforming one that has breached its OC test. First, not all transactions have junior OC tests or if they do, they may only apply after the end of the reinvestment period. Second, as seen in table 8, there is a range of OC triggers. A transaction that has an OC trigger set at a higher level, will breach quicker than another transaction with a lower trigger, even if its performance could be better than the other transaction.

Table 8

Appendix I

Chart 9

Chart 10

Chart 11

Chart 12

Chart 13

Chart 14

Chart 15

Chart 16

Chart 17

Chart 18

Chart 19

Chart 20

Appendix II

We have chosen transactions in their reinvestment period, and assumed no principal receipts before the end of their reinvestment period.

We have modeled senior trustee fees and administrative expenses at the level of each CLO's documented senior expenses cap.

We have assumed that in the numerator of the OC ratio, 'CCC' assets are carried at their market value once they exceed 7.5% of the portfolio. We have assumed a market value of 70%.

We have assumed 100% floating-rate assets.

We have calculated the WACC as the sum product of each tranche by its margin or coupon.

Table 9

Table 10

Related Criteria

• Global Methodology And Assumptions For CLOs And Corporate CDOs, June 21, 2019

Related Research

• European CLO Performance Index Report Q1 2020, June 30, 2020
• The European Speculative-Grade Corporate Default Rate Could Reach 8.5% By March 2021, June 8, 2020
• The U.S. Speculative-Grade Corporate Default Rate Is Likely To Reach 12.5% By March 2021, May 28, 2020
• The European Crisis Backstop Is Underpinning Corporate Funding Conditions, May 19, 2020
• How COVID-19 Changed The European CLO Market In 60 Days, May 6, 2020
• Ratings On 18 Classes From 14 European CLOs Placed On Watch Negative, April 27, 2020
• Credit Conditions In Europe Darken As Costs Of Lockdowns Add Up, April 27, 2020
• Redesigning The CLO Blueprint After COVID-19, April 21, 2020
• Europe Braces For A Deeper Recession In 2020, April 20, 2020
• First-Quarter 2020 CDO Monitor Benchmarks Reveal Relative Credit Quality And Diversity Of CLO Portfolios, April 10, 2020
• S&P Global Ratings May Add Additional Qualitative Factors When Rating CLO Tranches Due To Changing Credit Dynamics, April 9, 2020
• How Credit Distress Due To COVID-19 Could Affect European CLO Ratings, April 2, 2020
• European CLOs: Assessing The Credit Effects Of COVID-19, March 25, 2020