January 27, 2020 - As highlighted in Nomura’s Special Topics on January 24th, basis risk has been on CLO managers’ (and equity holders’) minds. While the focus generally has been on one-month/three-month LIBOR basis, basis questions also have emerged in the SOFR space. Why? Because there is potential that CLO assets (e.g., institutional term loans) might transition from LIBOR to Simple Daily SOFR in Arrears, while CLO liabilities might transition to SOFR Compounded in Arrears. We discuss all below. (Spoiler: The Simple/Compounded SOFR basis is very small and should have a marginal impact on CLO equity returns.)
As Figure 2 of the Nomura document demonstrates, one-month/three-month LIBOR basis has been in the 15 bps context since last October. Meanwhile, in 2018, the basis periodically topped 40 bps. This is problematic for CLOs because loan borrowers can choose shorter LIBOR interest periods (such as one-month LIBOR), while CLO liabilities generally are fixed at three-month LIBOR periods. When the LIBOR curve steepens and more loan borrowers choose shorter interest periods, this creates basis – and can hit CLO equity returns hard. As a (highly theoretical and overly simple) example, consider the following: One-month LIBOR is 20 bps below three-month LIBOR. All CLO loan assets are priced off one-month LIBOR, while all liabilities are priced off three-month LIBOR. For a 10x levered CLOs, this could theoretically mean that the 20 bps LIBOR basis would be increased 10-fold to 2% for the CLO equity. This could theoretically reduce the equity return by 2 percentage points per annum. That equity pain is not theoretical. As Fig. 1 demonstrates, the absolute value of one-month/three-month basis has averaged 14 bps in the past 20 years, and briefly topped 100 bps in 2008.
Thus, basis concerns are legitimate and should be discussed in the context of SOFR. However, the SOFR math should be reassuring. Using historical SOFR analogues, Fig. 2 shows the simple/compound SOFR basis for one-month and three-month SOFR going back to 1998. In fact, the average basis of one-month simple/compound SOFR basis was zero bps; for three-month simple/compound SOFR, basis averaged 1 bp. The biggest bases in the 20-year period were 2 bps and 5 bps, respectively.
Fig. 3 puts LIBOR and SOFR basis into perspective. Simple/compound SOFR basis is practically invisible when put on the same chart as LIBOR basis. The math demonstrates that SOFR basis is not remotely the same magnitude as LIBOR basis. Meanwhile, there are arguments for loans going to simple SOFR, at least initially. For products as flexible as loans (for instance, those that permit frequent intraperiod repayments) that also don’t trade with accrued interest, simple SOFR might fit better, might permit easier trading and will be more operationalizable. CLO practitioners should weigh the trade-offs between easier trading and settlement of loans using simple SOFR with the (small) basis risk that could emerge with CLO liabilities.