As structured products gain a growing share in today’s institutional portfolios, investors might feel a false sense of security in their risk/return profiles. True resilience in portfolio construction demands a framework that consistently analyzes both “normal” risk (well captured by static models) and "uncertainty" (tail risk).
Alla Gil
There are, of course, historical precedents for the uncertainty threat presented by collateralized loan obligations (CLOs) during times of market stress. For example, as some readers may recall, nearly half of senior tranches in CLOs and collateralized debt obligations (CDOs) were downgraded to junk amid the global financial crisis.
Today’s CLOs are less risky, because of all the safety features protecting senior CLO tranches – including improved subordination, higher-quality collateral and shorter reinvestment periods. However, running a handful of stress scenarios – combined with, say, conservative, static correlation assumptions – still might be insufficient for assessing current CLO risks.
Traditional modeling approaches try to capture tail risk separately (as in extreme value theory) or through a few stand-alone scenarios (with subjective probability assessment). These approaches, however, miss the key feature of financial crises: increased volatility in the bell-shaped part of risk distribution impacts the probability of arrival of tail events.
The point is that extreme tail risks are not independent from the rest of the distribution and must be analyzed consistently. So-called “black swans” (like the COVID-19 pandemic) are among the most impactful tail risks – but even those events can be captured in advance by dynamic, forward-looking models that employ full-range scenario analysis. These advanced models project realistic shifts in macro and market environments and their respective impact on the portfolio, including CLOs and other structured and private investments.
Without robust, forward-looking diagnostics, investors risk sleepwalking into a repeat of the past – or worse, being unprepared for a more complex crisis (e.g., the regional bank crisis of 2023) whose precursors could have been understood in advance. This is especially true with potential new additions to collateral pools – i.e., the growing asset class of private credit.
While CLOs have been the best performing fixed income asset class in the past decade, they are still more complex than other fixed income products and must be thoroughly analyzed. During the periods of stable market behavior, CLO tranches are fairly liquid and don’t exhibit much of mark-to-market (MTM) volatility, because the underlying loans have floating risk-free interest rates. But during credit crunches, all the nice features of modern CLOs – like cash flow waterfalls and senior tranches repayment – might not help.
CLOs are mainly driven by credit dynamics of the underlying collateral pool, and depend a lot on copula correlations between defaults and downgrades in the underlying loans. On the one hand, the credit quality of individual loans can be thoroughly analyzed by research analysts and credit experts. On the other hand, the impact of increased correlations within the pool can be underestimated, even while these correlations potentially trigger multi-notch downgrades in senior tranches.
To illustrate, consider an AA-rated CLO tranche backed by 300 commercial and industrial loans. The total potential loss across the entire pool – assuming every loan defaults – is simply the sum of each loan’s probability of default (PD), multiplied by its loss given default (LGD). This aggregate loss does not depend on correlations within the pool. However, how these losses are distributed across the CLO’s tranches is highly dependent on the correlations among PDs and LGDs within the pool.
These correlations are not static – they evolve in response to market shocks and macroeconomic scenarios. Therefore, by projecting both the credit quality and potential recovery values (in the event of default) under various conditions, investors can more accurately model both expected cash flows and mark-to-market variations across a wide range of feasible scenarios.
While the initial credit rating of a tranche reflects a default probability consistent with similarly rated instruments (bonds, loans, etc.), the probability of downgrades can diverge dramatically due to shifts in correlation under stress. Tables 1 and 2 (below) demonstrate this behavior exactly.
Table 1: Implied Default Probabilities for Each Tranche Over a 12-Quarter Horizon
Table 1 shows implied cumulative probabilities of default of various sample CLO tranches. These probabilities were derived by projecting the entire collateral loan pool across thousands of scenarios, each incorporating macroeconomic and market variables, as well as a variety of shocks that disrupt otherwise stable market behavior.
In scenarios involving credit crunches – either within specific industry sectors or across broader sectors – credit spreads widen and asset prices decline. This leads to a simultaneous increase in both the PDs and LGDs for individual loans, while also raising the copula correlations between defaults and downgrades.
A tranche is considered to have defaulted in any scenario where post-recovery losses exceed all subordinated tranches. Taking all of this into account, the implied cumulative probability of default within the first year aligns closely with the historical annual default rates published by major rating agencies.
However, the implied downgrade probabilities for CLO tranches tell a different story. For example, Table 2 shows the projected downgrade probabilities of a representative AA-rated tranche over the next 12 quarters, highlighting the sensitivity of tranche ratings to shifts in credit performance.
Table 2: Transition Probabilities of a Sample AA-rated Tranche
Source: Straterix Inc.
Transition probabilities are estimated by analyzing the proportion of forward-looking simulation paths in which the remaining subordination level is consistent with the initial subordination of a respective tranche’s rating.
It’s possible to reduce these downgrade probabilities with active management of the underlying collateral pool, based on early warning signals. As our analysis shows, it is critical to incorporate potential downgrades adequately, because they might cause a substantial increase in capital charges, across both banking and insurance investment portfolios.
CLOs – and especially their highly-rated senior tranches – are considered to be liquid instruments. Under normal market conditions, this assumption largely holds: their trading volumes are sufficient, bid-ask spreads are manageable, and MTM volatility is lower than other fixed income instruments with fixed-rate returns. However, this perceived liquidity can deteriorate rapidly in times of market stress.
It is therefore critical for investors and risk managers to look beyond the initial tranche ratings and conduct thorough analysis of the underlying loan pools across all CLO holdings in the portfolio. By doing so, investors can identify hidden pockets of risk that may not be visible through traditional risk metrics, proactively managing their exposure before market conditions turn adverse.
Alla Gil is co-founder and CEO of Straterix, which provides unique scenario tools for strategic planning and risk management. Prior to forming Straterix, Gil was the global head of Strategic Advisory at Goldman Sachs, Citigroup, and Nomura, where she advised financial institutions and corporations on stress testing, economic capital, ALM, long-term risk projections and optimal capital allocation.