What Is a CLO? A Practitioner's Guide to Collateralized Loan Obligations

A collateralized loan obligation is a securitization vehicle that buys a diversified portfolio of corporate loans and funds those purchases by issuing several classes of debt plus a residual equity class, each ranking differently in the order in which cash reaches them. The loans are almost always broadly syndicated, first-lien, senior secured facilities lent to companies rated below investment grade. The classes issued against them run from AAA at the top down to unrated subordinated notes at the bottom.

What separates a CLO from a simple pool of loans is that it is actively managed and structurally leveraged at the same time. A collateral manager buys and sells loans inside the vehicle for years after it closes, within limits written into the indenture. Meanwhile the tranche structure means a modest slice of first-loss capital supports a much larger amount of highly rated debt. Both features drive everything else that follows, including why the instrument is difficult to price and why the data around it is harder to source than the data around a corporate bond.

What sits inside a CLO

A typical new issue broadly syndicated loan CLO is roughly 500 million dollars in the United States or 400 million euros in Europe, and holds somewhere in the region of 150 to 250 individual loans, sometimes more. Industry primers generally describe at least 90 percent of the pool as below investment grade first-lien senior secured loans, with the average collateral rating around single-B and limited room for second-lien or unsecured positions. The borrowers are large companies, conventionally with EBITDA above 250 million dollars, which is the line that separates broadly syndicated CLOs from middle market CLOs backed by smaller directly originated loans.

Three properties of that collateral matter for everything downstream.

The loans are floating rate. Coupons reset periodically against a reference rate, currently SOFR in the United States, plus a contractual spread. Because the coupon adjusts as rates move, the price of a performing floating rate instrument tends to sit close to par and its sensitivity to interest rates is minimal. Credit is what moves the price.

The loans are senior and secured, which historically has meant materially better recovery in default than unsecured claims. Moody's long-run data puts first-lien bank loan recoveries around 65 percent on a trading price basis against roughly 38 percent for senior unsecured bonds, with ultimate recovery on term loans higher still. That differential is the reason a pool of single-B loans can support a AAA-rated liability at all.

That said, the historical average is not a safe forward assumption, and this is worth stating plainly because a good deal of CLO marketing material still recites it as though nothing has changed. S&P Global Ratings has said it expects recoveries on rated first-lien debt to come in below historical averages, and that empirical first-lien recoveries in the United States, Canada and Europe have already degraded noticeably. Four forces explain it. Borrowers increasingly finance themselves almost entirely with first-lien debt, so there is little junior capital beneath the loans to absorb loss. More than 90 percent of large-cap leveraged loans now lack maintenance covenants, so lenders intervene later in a deterioration, when less value remains. Liability management exercises, meaning out-of-court restructurings such as uptiering exchanges and drop-down financings, can subordinate lenders who do not participate. And a growing share of leveraged borrowers are asset-light software and services companies whose enterprise value can erode faster than hard assets in distress. Anyone modelling a CLO on long-run recovery averages is using a number the rating agencies themselves have moved away from.

The loans are diversified across unrelated borrowers and industries, subject to concentration limits in the indenture. This point looks mundane and is the single most important structural fact about the asset class, for reasons that become clear when we come to the comparison with CDOs. Diversification is not unlimited in practice: S&P has reported that the 250 most widely held obligors account for roughly half of assets under management across the US broadly syndicated CLOs it rates, with the largest ten around 7 percent.

The manager does not simply buy a portfolio and hold it. Eligibility criteria constrain what can be purchased and portfolio tests govern the aggregate: limits on single-obligor exposure, on industry concentration, on the weighted average rating of the pool, on the weighted average spread it must earn, on CCC-rated holdings, and on the weighted average life it may carry. The manager trades within that box, and the tests are measured continuously rather than at purchase.

The capital structure

A CLO issues a stack of securities against that loan portfolio, and the defining feature of the stack is that losses hit the bottom first while cash reaches the top first.

At the top sits the AAA class, which in a conventional structure is around 60 to 65 percent of the deal and pays the lowest spread. Below it come mezzanine classes rated AA, A, BBB and BB, sometimes with a B class below that, each considerably smaller at roughly 4 to 12 percent apiece. At the bottom sit the subordinated notes, generally called the equity, typically 9 to 10 percent of the structure, carrying no rating and no promised coupon.

Two mechanisms convert that ordering into credit protection.

The first is subordination. Every class benefits from the classes beneath it, because those classes absorb losses first. Contemporary CLOs are generally issued with par subordination of around 35 percent below the AAA class, meaning the collateral pool would have to sustain principal losses of that magnitude before a AAA holder was touched. That cushion has grown by roughly ten percentage points since the financial crisis. For scale, the leveraged loan market's peak default rate during 2007 and 2008 was in the region of 11 percent, and defaults are not the same as principal losses once recoveries are applied.

The second is overcollateralization. The portfolio is deliberately larger than the debt issued against it, so the vehicle holds more collateral par than it owes. That cushion is monitored through overcollateralization tests, which compare the adjusted par value of the collateral to the outstanding balance of each class down to a given level. Interest coverage tests work similarly on the income side, comparing interest collections to interest owed.

These tests are the enforcement mechanism, and the way they enforce is where the interests of different classes diverge sharply. Cash entering a CLO is not distributed pro rata. It runs down a waterfall, paying senior fees and expenses first, then interest on the AAA class, then interest on each class in turn, with a test at each level. Pass the test and cash continues down. Fail it and cash is diverted from its intended recipient and used instead to pay down the most senior outstanding notes until the test is cured.

For a senior noteholder, a failing test is a protective feature that accelerates repayment. For the equity, it means distributions stop entirely while the structure deleverages above them. The same cashflow event is good news at the top of the stack and bad news at the bottom, which is why a single price for a CLO deal is a meaningless concept. Each tranche is a distinct instrument with a distinct payoff.

Behind all of this sits an event of default test as the final backstop for the senior class. If the adjusted par balance of the collateral falls below a threshold set close to the AAA balance itself, the deal can be liquidated and the AAA principal repaid.

CLO equity

The subordinated notes deserve separate treatment because they behave nothing like the debt above them.

Equity holders receive what remains after the vehicle pays its fees and the contractual interest on every debt class. That residual is a spread arbitrage. The portfolio earns a weighted average spread over the reference rate, the liabilities cost a weighted average spread over the same reference rate, and the difference on a leveraged base accrues to the equity. When the arbitrage is wide, distributions can be substantial. When collateral spreads compress, or when defaults reduce the performing balance, or when a coverage test diverts cash upward, distributions can fall to nothing.

Equity is also the first-loss piece, so principal is at risk before any rated class is touched. And equity has no maturity in a meaningful sense, since it receives whatever is left at the end of the deal's life after every note has been repaid.

That combination makes CLO equity closer in character to a levered fund interest than to a bond. It is valued on projected cashflows under assumed default, prepayment and recovery paths rather than on a spread to a curve, and reasonable people using reasonable assumptions arrive at materially different numbers. Anyone valuing it should expect wider dispersion between marks than they would tolerate anywhere else in the structure.

The life of a CLO

CLOs move through phases, and knowing which phase a deal occupies tells you most of what you need to know about how it will behave.

It begins in a warehouse. Before the deal exists as a securitization, the manager accumulates loans using a short-term facility provided by a bank, funded with a small slice of first-loss capital. The warehouse is where the portfolio is assembled and where the economics of the eventual deal are largely determined, since loans bought cheaply into the warehouse improve the arbitrage for the equity later.

The deal then prices and closes. Notes are issued, the warehouse is repaid, and the vehicle begins life as a standalone entity. It is usually not yet fully invested, so a ramp-up period of roughly three to six months follows while the manager buys the remaining collateral to reach target par.

Next comes the reinvestment period, and this is the phase that makes a CLO an actively managed vehicle rather than a static pool. Most deals issued since 2013 have carried four or five years, with five years emerging around 2017 as the prevailing standard for broadly syndicated loan CLOs. Principal received from loans that repay or prepay does not flow to noteholders during this window. The manager reinvests it in new collateral, subject to the eligibility criteria and portfolio tests. A manager can trade to improve credit quality, harvest gains, or sell a deteriorating position before it defaults, and skill in doing so is the reason manager selection matters to the price of a tranche.

When the reinvestment period ends, the deal enters amortization. Principal proceeds now pass to noteholders sequentially from the top, retiring the AAA class first, then the class below it, and so on. The AAA is commonly repaid in full by around year six or seven. The structure deleverages steadily, which improves credit support for the remaining senior notes while shortening the life of the deal.

Two refinancing mechanics interrupt this sequence in practice. After a non-call period that conventionally runs two years, a deal can be refinanced, meaning individual tranches are repriced at tighter spreads without otherwise restructuring the vehicle. It can also be reset, meaning the reinvestment period is extended and the capital structure substantially rewritten. Both are ordinary in active markets, and most deals are refinanced or reset at the end of the non-call period before any amortization occurs. As the senior tranches retire and excess spread compresses, remaining debt eventually becomes uneconomic to keep outstanding, and equity holders can collapse the deal well short of its stated legal maturity.

The practical consequence for anyone maintaining records is that the tranche you analyzed last year may carry a different spread, a different maturity and a different reinvestment horizon today. Reference data that fails to capture a refinancing or reset will misprice the instrument regardless of how good the pricing model is.

CLOs and CDOs are not the same instrument

The two acronyms differ by one letter and the instruments differ almost entirely. Since collateralized debt obligations are remembered chiefly for their role in the 2008 crisis, the distinction matters to anyone approaching CLOs for the first time.

A CLO holds corporate loans. Each borrower is a company with its own business, its own leverage and its own industry, and while corporate defaults cluster in recessions, the borrowers are not exposed to one identical risk.

The CDOs that failed catastrophically were mostly asset-backed CDOs, and what they held was tranches of other securitizations, predominantly subprime residential mortgage-backed securities. That construction embedded two problems. The collateral was already tranched, so the CDO applied leverage on top of leverage. And every piece of that collateral was ultimately exposed to the same variable, the performance of US housing. A structure whose diversification is nominal rather than real offers protection that evaporates precisely when it is needed, because a single macro shock impairs the entire pool at once. Where CDOs held tranches of other CDOs, the effect compounded further.

The performance record reflects that difference. According to S&P Global Ratings data, no AAA-rated CLO tranche has defaulted in the recorded history of the market, and cumulative defaults across investment grade CLO tranches rated AAA through BBB between 1996 and 2023 amounted to fewer than 0.1 percent, some nineteen instances across nearly 23,000 tranches. The pre-crisis CLO 1.0 generation has now fully paid down, and of more than four thousand tranches rated across that era, forty defaulted, fifteen of which had been investment grade at issuance. No S&P-rated investment grade CLO issued since 2010 has defaulted.

Post-crisis structures carry more credit enhancement than their predecessors, particularly at the top of the stack, and they largely exclude the structured product buckets that caused the damage. Contemporary CLO collateral pools are almost entirely senior secured corporate loans, with none of the synthetic exposure or resecuritization that defined the instruments that failed.

The lesson worth carrying forward is that the tranching technology was never the problem. What mattered was whether the collateral underneath the tranches was genuinely diversified.

How CLOs actually get priced

Almost nothing about a CLO can be priced from an exchange, because CLO tranches do not trade on one. Nor do they benefit from the trade reporting infrastructure that publishes post-trade prices in other markets. Trade reporting utilities generally do not cover bank loans or complex structured finance, which means the transparency a corporate bond investor takes for granted does not exist here. Valuation is therefore evaluated rather than observed, which places CLO tranches in the Level 2 category of the fair value hierarchy.

SQX's approach begins with market observations rather than with a model. The primary inputs are dealer runs and quotes, circulated from the sell side to the buy side, which SQX parses using proprietary software. Two-sided quotes carrying size are weighted more heavily than indications without it, on the straightforward basis that a dealer showing both a bid and an offer in size is making a more meaningful statement about value than one showing neither.

From there the method addresses the problem that makes CLOs distinctive. Two tranches with identical ratings and identical positions in the capital structure can be worth different amounts because they are managed by different managers. So SQX assigns a rating to each CLO manager, then combines manager rating with seniority in the capital structure to form a grid, and derives a baseline discount margin or yield from the intersection. A tranche is then adjusted off that baseline for its own specifics: the structural and waterfall features written into its indenture, the quality of the collateral behind it, and the identity of the manager itself.

Default, prepayment and recovery assumptions are implied from parsed market data rather than imposed, cashflows are generated under those assumptions, and the result is discounted.

One further step addresses the connection between a tranche and the portfolio underneath it. SQX prices the underlying collateral to establish the net asset value of the loan portfolio, and uses that valuation to determine how much overcollateralization is genuinely available to each tranche. A tranche's price can then be adjusted for the coverage actually supporting it. This matters because the protection a mezzanine class enjoys is a function of what the collateral is worth today, not what it was worth at issuance, and a tranche that looks comfortably covered on original par may look considerably less comfortable on current value.

Quality assurance runs across three axes: parsed prices for the same tranche, other tranches in the same deal, and other deals from the same manager. A price that disagrees with all three of those reference points warrants review.

Staleness is handled by instrument rather than uniformly. A CLO tranche is permitted a longer time to staleness than a G5 government bond, because a week without a quote means something different in a market where a given tranche might trade a handful of times a year than it does in one of the most liquid markets on earth. Applying a single staleness rule across a fixed income universe either discards good CLO data or accepts bad government bond data.

Why CLO data is harder than it looks

Several difficulties compound in this asset class, and they are worth naming because they explain why CLO holdings are disproportionately responsible for valuation disputes, audit questions and reporting exceptions.

Identification is unreliable. Structured finance instruments are issued by special purpose vehicles, frequently in multiple currencies and jurisdictions, often with several identifiers attached to what is economically one tranche, and sometimes with identifiers that change through a refinancing. Matching a position in a portfolio system to a security in a reference file is materially harder here than for a corporate bond with a stable issuer and a single identifier.

Terms change during the deal's life. Refinancings reprice tranches and resets rewrite structures. A reference record that captures issuance terms and never updates will drift away from the instrument it purports to describe.

Trade evidence is thin and unevenly distributed. Senior tranches of large deals from well-known managers trade with some regularity. Mezzanine tranches of smaller deals may go long stretches without a quote. Any pricing process must therefore work across a wide range of observability, and a process calibrated only to the liquid end will fail exactly where a valuation is hardest to defend.

Nothing about a CLO can be inferred reliably from its rating alone. Two BB tranches with different managers, different structural protections and different collateral quality are different instruments in every respect that determines value.

Where this leaves investors or operations teams

For anyone holding CLO tranches, the practical implications reduce to a few points. Position matters more than the deal, since the tranche determines the payoff. Manager identity is a pricing input rather than a footnote. The phase of the deal's life shapes how principal behaves. The equity is a different kind of instrument from the debt and should be valued and reported as such. Recovery assumptions inherited from long-run averages deserve a fresh look. And because prices are evaluated rather than observed, the defensibility of a valuation rests on the methodology behind it and on the market evidence feeding that methodology.

SQX prices fixed income across the structured finance spectrum, including CLOs, using the evaluated approach described above. If you would like to discuss how it applies to a specific portfolio, just contact us!

Want more SQX updates? Follow SQX on LinkedIn and subscribe to our
LinkedIn newsletter to stay ahead of the curve.