By the Numbers
The NAIC takes another step toward raising required CLO capital
Caroline Chen | October 20, 2023
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
The National Association of Insurance Commissioners lately has filled in some important details of its proposed approach to setting CLO capital requirements for insurers. The group proposes stressing CLOs against scenarios with significantly higher levels of loan defaults and lower recoveries than historic averages, although these scenarios will get low weights in the overall analysis. The scenarios nevertheless would likely raise capital required for CLO equity and some junior classes. Insurers holdings position in equity or junior debt would likely feel the impact, and the market for these securities would likely see wider primary and secondary market spreads.
Under the NAIC’s current risk-based capital rule, insurers holding every tranche of a CLO from ‘AAA’ down to equity have to hold significantly less regulatory capital than holding a pool of leveraged loans identical to the one backing the CLO. This is mostly because the ‘AAA’ tranche, typically accounting for 63% of a CLO, is only subject to 0.16% risk-based capital (RBC). That compares to 9.54% for a pool of mostly ‘B’ rated leveraged loans. However, the NAIC does not believe securitizing a pool of loans changes the economic risk and is revising its rules so the capital required for holding all tranches of a CLO is consistent with the capital required for holding the underlying pool.
The staff at the Securities Valuation Office in the NAIC has advocated fixing the current regulatory capital arbitrage loophole since last year. In an internal memo addressed to the Valuation of Securities Task Force in May 2022, the staff recommended starting the reevaluation using three modeling scenarios currently used in the NAIC’s stress test methodology and then expanding the scenarios to account for tail risk. In addition, the staff proposed adding two new RBC factors, 75% and 100%, under the NAIC rating category 6 typically used for equity (Exhibit 1).
Exhibit 1: Under current mapping, insurers’ CLO holdings have lower capital charges
Setting defaults and recoveries in expanded modeling scenarios
Under the NAIC current stress test methodology, the NAIC runs three scenarios—A, B and C—with increasing levels of stress on defaults and recoveries. Based on Moody’s study of US corporate defaults from 1970 to 2019, the NAIC groups corporate historical defaults by 40 10-year cohorts. The ratings of those corporates range from ‘Ba1’ down to ‘Ca-C’. For example, historic defaults in 1970-1979 are in the first 10-year cohort, while historical defaults in 2010-2019 are in the last cohort. Each rating category has its own 40 10-year cohorts. A weighted average default vector was then created using the number of issuers in each cohort as weights. The weighted average standard deviation of historical defaults in each rating category is also calculated in a similar fashion. Under this approach, the cumulative default rate for a ‘B3’ corporate issuer, for instance, ranges from 6.5% in Year 1 to 45.5% in Year 10, with one standard deviation rising from 5% in Year 1 to 11% in Year 10 (Exhibit 2). The complete NAIC default vector and standard deviation vector are here.
Exhibit 2: Historical defaults and standard deviation of ‘B3’ corps in the NAIC modeling
The NAIC also uses the same Moody’s study to model the recovery rates, with the historical recovery rates from first-lien bank loans to junior subordinated bonds being applied to scenario A. To stress the recoveries, the NAIC assumes debt will perform similarly to their next worst category. For example, leveraged loan recovery will be lowered to the historical recoveries of senior unsecured bonds. Stress recoveries are used in scenarios B and C.
Seven new NAIC scenarios
In the most recent NAIC meeting, the staff revealed seven new scenarios for modeling CLO risk, adding to the original A, B, and C scenarios (Exhibit 3). Among other things, the new scenarios expand the range of defaults from one standard deviation to two and put further stress on recoveries by reducing them to between 50% and 75% of historical levels. Each of the ten scenarios will have an assigned probability weight, which remains open for discussion.
Exhibit 3: A list of expanded scenarios proposed by the NAIC in October
Scenarios 7 to 10, with loan defaults being stressed by an additional one or two standard deviations above historical levels and with lower recovery assumptions, are likely to have a high cash flow impact on CLO tranches. These scenarios will likely result in principal and interest losses to tranches at the lower part of the capital stack. While the regulator expects no more than 2% probability weights to these tail scenarios, extreme losses to a CLO tranche may nevertheless have a meaningful impact on their risk-based capital charge.
Assessing CLO risk from their tranche level losses
To fix the regulatory arbitrage loophole, the NAIC will evaluate the risk of a CLO based on its tranche-level principal and interest losses under the proposed 10 scenarios. The staff at the NAIC has randomly selected six broadly syndicated loan (BSL) CLOs since they kicked off the analysis in April (Exhibit 4). All six deals are in their reinvestment periods, and none have reported any performance trigger breaches to date.
Exhibit 4: Six BSL CLOs are selected by the NAIC for analysis
The NAIC has run tranche-level cash flow for all six CLOs under scenarios A, B and C and shared the results with the public in May. Most ‘BB’ tranches reported principal and interest losses under scenarios B and C, while some mezzanine tranches were also hit with losses in a much-stressed scenario C (Exhibit 5).
Exhibit 5: P&I losses reached mezzanine tranches in stressed scenarios
At this stage, the NAIC hasn’t provided cash flow runs for the seven newly recommended scenarios, but they have disclosed the general RBC mapping approach (Exhibit 6). First, the principal and interest losses under each of the ten scenarios will be mapped to a mid-point of an NAIC table (Exhibit 7) to derive the RBC, and then a weighted average RBC will be calculated based on the probability weight of each scenario.
Exhibit 6: A hypothetical example for calculating a ‘BBB-‘ CLO risk-based capital
Exhibit 7: The NAIC table for mapping tranche P&I loss to a midpoint for RBC
A ‘BBB-‘ CLO under the current approach is mapped to the NAIC rating designation category of 2.C with an RBC factor of 2.17%, as shown in Exhibit 1. With the staff recommended scenarios, the same ‘BBB-‘ CLO in a hypothetical example may be mapped to the NAIC rating designation of 4.B with a much higher capital charge. The NAIC hopes to run every CLO on all state-regulated insurance companies’ books under their proposed 10 scenarios to assess the risk. While scenario probability weights remain uncertain and the modeling assumptions may still be subject to change, it is likely some mezzanine and junior CLOs may have a significant increase in their risk-based capital from the current level.
US life insurers may see a larger impact on their required capital level than their P&C peers. According to the regulatory filings, nearly 39% of life insurers’ $178 billion CLO investments as of 2022 were allocated to ‘A’ and lower credit quality CLOs; the CLO investments comprised 3.6% of life insurers’ total invested assets. By contrast, 22% of P&C insurers’ $49 billion CLO investments in 2022 were ‘A’ and lower credit quality CLOs, and the CLO investment represented 2.9% of their total invested assets.
The NAIC proposed RBC work may also impact mezzanine CLO investors other than insurance companies. US insurers’ $71.5 billion mezzanine CLO holdings as of 2022 are more than half of the $124 billion outstanding mezzanine CLOs today. A potential shrinking footprint of insurance companies in the mezzanine bonds due to the revised RBC may put pressure on secondary spreads. By contrast, ‘BB’ CLOs may also see a large rise in RBC, but less than 20% of the outstanding ‘BB’ CLOs are held by insurance companies (Exhibit 8).
Exhibit 8: Insurers have a large footprint in the outstanding mezzanine CLO market
With many details of the NAIC modeling work still under discussion, market participants and CLO investors need to carefully monitor developments. Fortunately, the NAIC, as promised, has always kept transparency during the process. At this stage, risk-based capital for ‘AAA’ and ‘AA’ CLOs seems the most unlikely to be impacted by the NAIC staff proposals, and a flight-to-quality allocation strategy may remain the best.