Fair Oaks Reflections – Issue 9
5 September 2019
CLOs have received positive attention in the press1 of late following the release of the Bank of England’s latest Financial Stability Report2 in July. As highlighted in the Bank of England’s report, current CLO structures (“CLO 2.0”, issued from 2011 onwards) are much more robust compared to pre-crisis structures (“CLO 1.0”, issued before 2009). As central banks warn of risks to economic growth globally, we weigh the structural protections in the CLO market against perceived deterioration in lending standards in the leveraged loan market.
CLO 1.0 performed well through the financial crisis
In addition to protection provided by low “loan-to-value” or high overcollateralisation (“OC”) at issuance, CLO debt tranches also benefit from Overcollateralisation Tests (“OC Tests”) designed to counteract par loss and collateral deterioration throughout the life of the deal. These mechanisms divert interest cashflows to pay down senior CLO debt or purchase additional collateral instead of making payments to the equity if the OC drops below a certain threshold. The tests are set at levels such that cash flow redirection kicks in early, well before asset coverage drops to 100% for any rated class and cash flow diversion stops once the OC has recovered to the test level. These tests are also pre-emptive, i.e. they may be triggered in scenarios where the exposure to CCC/Caa-rated loans increases, even if the number of actual defaults is low.
In 2009, over 70% of US CLOs breached at least one OC Test3, often driven by CCC/Caa balances, causing cashflows to be diverted to amortise CLO debt or acquire additional assets and effectively increasing OC and protecting CLO note holders. As OC levels increased, the proportion of CLO tranches failing OC Tests dropped below 10% within two years.
Ultimately, only 1.1% of CLO 1.0 tranches defaulted (58 out of 5,765 CLO 1.0 tranches rated by S&P)4. There have been no defaults of CLO 2.0 tranches in the US or Europe to date. It is also worth noting that despite temporary cash-flow diversions, 2007 CLO equity performed strongly5. Loans purchased to increase collateral test values were often purchased at a discount and generated gains for the equity as they were repaid at par.
Structural enhancements – CLO 1.0 versus CLO 2.0
When CLO issuance recommenced post crisis, deals were structured with higher overcollateralisation levels and stricter eligibility criteria on the collateral pool. These enhancements are still the market standard and have not been challenged by issuers even during periods of very high investor demand. For example, median AAA par OC of 155% for US CLOs today is 23% higher than for CLOs issued in 2006-2007 (132%)6. The median US CLO 2.0 BBB tranche benefits from OC of 114%, which is much closer to a US CLO 1.0 A tranche (115%) than US CLO 1.0 BBB levels (110%)6. Throughout the capital structure, OC levels have increased significantly, enabling structures to support higher loan loss rates.
Figure 1: Median EUR and US CLO 1.0 and 2.0 capital structures6 (data labels show par OC levels)
Aside from higher subordination levels, CLO 2.0 debt further benefits from stricter asset eligibility criteria. Structured finance assets and synthetic securities are no longer eligible, and limits on CCC/Caa rated assets and second lien loans are stricter. Compared to the 6-year reinvestment period featured in CLO 1.0 deals, EUR and US CLOs today are shorter with 4.5 to 5-year reinvestment periods.
Enhanced subordination levels and potentially lower loan recoveries
It is possible that future loan recovery rates might be lower than historical levels due to increasing leverage and weaker incurrence covenants in cov-lite loan documentation7. From 2003 to 2017, EUR and US loan recovery rates have averaged around 70%8. Assuming a significantly lower recovery of 60%9 and noting that over the last 15 years, average loan recoveries were at or below this level for only three years in Europe and one year in the US8, we compare breakeven default rates for CLO 2.0 tranches.
Figure 2: Yield of CLO 2.0 tranches in Europe and the US at different constant loan default rates using recovery assumptions of 60% and 70%10
Figure 2 shows yields for EUR and US CLO 2.0 BBB and BB tranches for given constant annual loan default rate assumptions10. For example, assuming a 70% loan recovery rate, the breakeven default rate (i.e. the default rate at which the yield drops to zero) for EUR CLO 2.0 BBB tranche is 21.5% p.a. When using a lower recovery rate of 60%, EUR BBB tranches are still able to withstand defaults of 14.5% p.a. A, AA and AAA rated CLO notes require even higher annualised defaults rates to breakeven.
When put into context of historical loan default rates, it is clear that the constant annual default rate required to generate a 0% IRR on investment grade CLO notes is a multiple of these values: the highest recorded five-year average loan default rate was 5.8% in the US (1999-2003) and 5.5% in Europe (2009-2013)11.
(1) Source: Bloomberg, “Senior CLO Debt Avoids Losses in Financial Crisis Scenario: BoE”, July 2019.
(3) Source: Wells Fargo, “The CLO Desktop Primer”, March 2017.
(4) Source: Standard & Poor’s, “Default, Transition, and Recovery: 2018 Annual Global Leveraged Loan CLO Default And Rating Transition Study”, June 2019. Study includes BSL CLO securities rated by S&P from 1996 to Dec 2018.5. Source: Intex, July 2019. CLO 1.0 figures based on median levels at origination for CLOs with closing date in 2006-2007. CLO 2.0 figures show median as at origination of currently outstanding tranches with factor of 0.9 or higher.
(5) Median IRR for CLO 1.0 equity issued in 2007 was 21.1% in the US and 8.4% in Europe assuming equity purchase at 90%. Source: Wells Fargo, as of Aug 2019.
(6) Source: Intex, July 2019. CLO 1.0 figures based on median levels at origination for CLOs with closing date in 2006-2007. CLO 2.0 figures show median as at origination of currently outstanding tranches with factor of 0.9 or higher.
(7) See Reflections #4 for a discussion on this topic.
(8) Moody’s “Annual default study: Corporate Default and Recovery Rates”: first-lien US loan recoveries averaged 70.4% between 1990 and 2018. Standard & Poor’s “2018 Annual Study of Corporate Recoveries in Europe: Little Change, For Now”: European first lien recoveries averaged 73.4% from 2003 to 2017.
(9) This is lower than Moody’s Investor Services’ recovery expectations for first-lien US loans of 61% as of Aug 2018.
(10) Breakeven default rate resulting in zero yield. Breakeven default rates for representative CLO 2.0 tranches matching the median capital structure for CLO 2.0 currently outstanding and current new issue CLO debt spreads. Where necessary, reinvestment periods were adjusted to match deals at origination to 5 years for US CLO 2.0 and 4.5 years for EUR CLO 2.0. Assumption of 15% prepayment rate and 9 months recovery lag, 350bp reinvestment spread at par, 12 months default holiday and LIBOR/EURIBOR forward curves as of 2nd September 2019.
(11) Source: LCD based on S&P European Leveraged Loan Index (ELLI) and S&P/LSTA Leveraged Loan Index (LLI).