CMBS Structure: Tranches and Subordination

Introduction

A single pool of commercial mortgages can be sold as bonds rated anywhere from AAA to unrated, and the loans themselves never change. What changes is the order in which each bond is paid and the order in which each bond absorbs losses. That ordering, called subordination, is the entire engine of commercial mortgage-backed securities. Stack the claims so the top bond is paid first and loses last, and a pool of ordinary fixed-rate property loans yielding perhaps 6% produces a AAA-rated security sitting behind roughly 20% to 30% credit enhancement, the share of the pool that must be wiped out before that bond loses a dollar. At the bottom sits a first-loss piece that takes the initial hit and earns equity-like returns for the privilege. CMBS is less a lending product than a machine for cutting one credit risk into many, and understanding the machine is the foundation for everything else in securitized real estate debt.

A CMBS deal begins with an originator (an investment bank, a conduit lender, or a balance-sheet lender clearing inventory) assembling a pool of commercial mortgages, anywhere from a single large loan to fifty or more smaller ones. Whether the pool holds one big loan or many small ones is the central distinction between the two CMBS sub-markets, conduit and single-asset, single-borrower deals, and it changes both how the deal is structured and how its risk diversifies. That pool is sold into a trust, which issues bonds backed by the loan cash flows. The trust is structured as a REMIC (Real Estate Mortgage Investment Conduit), a tax election that lets the cash flows pass through to bondholders without an entity-level tax, the same pass-through logic that makes the whole securitization economic.

A single very large loan is often too big to sit in one conduit pool without overwhelming its diversification, so it is split into multiple pari passu notes of equal priority and placed across several deals. The Cunard Building's $250 million mortgage, for instance, was divided into a $130 million note and a $120 million note that went into separate conduit trusts. Each trust holds an equal-ranking claim on the same loan, which lets one borrower's debt diversify across pools while keeping every note at identical seniority, the same instinct that drives pooling, applied in reverse to a loan too large to pool cleanly.

The bonds the trust issues are not identical claims on the pool. They are layered into tranches, each with its own rating, coupon, and priority. The defining feature is that the tranches are paid and impaired in a fixed order, which is what allows a single pool to support securities of wildly different risk. The top of the stack is a super-senior AAA bond; the bottom is an unrated first-loss certificate, often called the B-piece. Between them sits a ladder of investment-grade and below-investment-grade classes.

Tranche

A distinct class of bonds issued from a securitization, each with its own priority in the cash flow and loss waterfall, its own rating, and its own coupon. Tranches are paid in order of seniority and absorb losses in reverse order, which is how one pool of identical-quality loans can back both AAA and unrated securities.

The reason to tranche at all is investor demand. A pension fund or insurer mandated to hold AAA paper will buy the senior bonds; a high-yield credit fund wants the subordinate classes for their double-digit returns. By slicing the pool, the originator sells each slice to the buyer who values it most, raising more total proceeds than selling the whole-loan pool to any single buyer would. The capital-stack logic mirrors the bond markets generally, which is why the broader machinery is covered in the debt capital markets guide; CMBS is the real estate application of the same idea.

Subordination is the percentage of the pool balance held by tranches junior to a given bond. If the AAA bond sits above tranches representing 25% of the pool, it has 25% subordination, meaning a quarter of the entire pool must default with zero recovery before that AAA bond loses any principal. That cushion is the bond's credit enhancement, and it is what earns the rating. Nothing about the underlying loans makes them AAA; the rating comes entirely from the structural protection beneath the bond.

Subordination

The share of a securitized pool held by tranches junior to a given bond, expressed as a percentage of the total pool balance. It measures how much of the pool must be lost before that bond is impaired, and it is the primary form of credit enhancement in CMBS.

The rating agencies set these levels. Their role in a CMBS deal is to opine on how much subordination each bond needs to achieve a target rating, given the credit quality, diversity, and leverage of the underlying loans. Several agencies compete for the work (Moody's, S&P, Fitch, KBRA, and DBRS Morningstar), and an issuer typically secures ratings from two or more, a structure that drew criticism before the crisis for encouraging issuers to shop for the most favorable opinion. A pool of low-leverage loans on diversified, well-located assets needs less enhancement to reach AAA than a pool of high-leverage loans on a single property type, which is why the LTV, DSCR, and debt yield of every loan in the pool feed directly into the subordination the agencies demand. The main levers an agency weighs include:

The agencies, in effect, design the structure: the originator brings the loans, and the agencies tell it how thick each tranche must be.

The arithmetic of an attachment point is simple. A bond's subordination, or attachment point, is the cumulative balance of everything below it divided by the pool:

Structured-credit desks describe the same idea with a loss band: every tranche has an attachment point and a detachment point. The attachment point is the cumulative pool loss at which the tranche begins to take losses, and for a given bond it equals its subordination, since losses must first chew through everything junior before reaching it. The detachment point is the cumulative loss at which the tranche is fully wiped out, equal to its attachment point plus its own thickness. A junior AAA bond with 20% subordination and a 10% slice, for example, attaches at 20% of pool loss and detaches at 30%, so it absorbs the loss band between those two points and nothing outside it.

A representative conduit deal stacks up roughly as follows, though exact levels move with collateral quality and market conditions:

The single most important fact in this table is the difference between the GFC era and now. Legacy pre-crisis AAA bonds carried subordination of around 12%; post-crisis CMBS 2.0 AAA bonds carry roughly 20%, a 67% increase in structural protection, and super-senior classes can sit behind 30% or more. The market rebuilt the cushion after learning in 2008 that thin enhancement on aggressively underwritten loans was no protection at all.

The mechanism that makes subordination real is a pair of opposing waterfalls. Cash flows are paid from the top of the stack down, and losses are allocated from the bottom up. These two directions, working against each other, are what concentrate risk in the junior bonds and shield the senior ones.

Cash flows: paid from the top down

On the cash side, interest is generally paid to every outstanding class in order of seniority each period, and scheduled and prepaid principal is directed first to the most senior bonds. This is the sequential-pay structure: the AAA classes are retired first, which steadily shortens their average life and, more importantly, increases the percentage subordination beneath the remaining senior bonds as the pool amortizes. A senior bond actually becomes safer over time as the deal pays down, because the dollar cushion beneath it shrinks more slowly than the bond itself. This is the opposite of how a corporate bondholder experiences time, and it is one reason senior CMBS appeals to liability-matched buyers who value a self-deleveraging position.

Because sequential pay determines when each class is retired, the buyers who price these bonds focus on weighted average life rather than the stated maturity, since that is when the bond actually returns capital. WAL weights each principal repayment by the time it is expected to arrive and divides by the total principal:

Here each principal payment of size $P_i$ lands at time $t_i$ (in years), so the senior classes, which receive principal first, carry the shortest WAL and the most predictable duration, while junior classes that wait at the back of the line show a longer and far more uncertain one.

Losses: absorbed from the bottom up

The loss waterfall runs the other way. When a loan defaults and the collateral cannot repay the balance, the shortfall is a realized loss, and that loss is written off against the bonds starting from the most junior outstanding class.

The timing of all this matters as much as the order. A realized loss is recognized only when a defaulted loan is actually liquidated, which can come years after the borrower first stops paying. In the interim, the master servicer advances the missed principal and interest to the bonds out of its own capital, so senior holders keep receiving cash even on a non-performing loan, and those advances are repaid off the top when the asset finally resolves, ahead of every bondholder. The obligation lasts only as long as the servicer judges the advances recoverable from eventual proceeds. Once an updated appraisal shows the collateral has fallen too far, an appraisal reduction curtails the advances flowing to the most subordinate classes, cutting off their interest before the loss itself is even crystallized. This is why distress surfaces in junior CMBS cash flows well before the headline realized-loss figures move.

This is also why the senior and junior tranches behave like completely different instruments despite sharing one collateral pool. The AAA bond trades like a high-grade corporate bond, sensitive mostly to interest rates and spreads. The first-loss piece trades like equity, sensitive to every default and recovery in the pool. The same loans, sliced differently, become products for opposite ends of the investor universe.

Two further features round out the structure and explain why CMBS bonds behave so differently from residential mortgage securities. The first is prepayment protection. A homeowner can refinance freely, which makes residential MBS cash flows unpredictable; commercial mortgages, by contrast, are heavily locked down so the bonds deliver the stable, defined cash flows that institutional buyers price. Most CMBS loans carry a prepayment lockout for several years, followed by either yield maintenance (the borrower pays a make-whole penalty that compensates investors for lost interest) or defeasance (the borrower substitutes a portfolio of government securities that replicates the remaining loan payments, releasing the property from the lien). Either way, the bondholders keep their expected yield, which is part of why senior CMBS trades like a high-grade fixed-income instrument rather than a callable one.

The second feature is the interest-only strip, and it rests on the deal's excess spread. Excess spread is the pool interest collected in a period minus the coupons owed to the bonds minus the servicing and trust fees, and it is the first cushion the deal builds before subordination is ever touched, since a missing dollar of loan interest eats into that spread before it dents any bondholder's coupon. The loans in a pool carry a higher weighted-average coupon than the blended coupon paid out across the rated bonds, and that excess interest is itself securitized into an interest-only class (often labeled the X class) that receives the spread rather than any principal. The IO strip lets the originator monetize the difference between what the loans pay and what the bonds cost, and it trades as its own instrument with its own buyers, sensitive primarily to how fast the pool prepays or defaults.

The investor base sorts itself by risk appetite, and the split is worth knowing because each buyer's motivation shapes how the deal trades and how it is worked out in trouble.

The senior buyers

The AAA and high investment-grade classes go to the largest, most rating-constrained pools of capital: insurance companies, money managers, banks, and pension funds that need high-grade, liquid paper and cannot or will not hold credit risk. These buyers are price-takers on credit and care mostly about yield relative to other AAA assets, prepayment behavior, and average life. They are the deep, stable demand that makes the senior bonds cheap to issue, and their presence is the whole reason the structure raises more in aggregate than a whole-loan sale would: this capital will pay up for the safest slice precisely because the subordination beneath it removes the credit work they are unwilling to do. Within the AAA stack, the split between the super-senior class and the junior AAA matters to these buyers, because the super-senior sits behind even the original AAA bond and was created after the crisis specifically to give the most conservative buyers an extra layer of protection. The depth and reliability of this senior demand is why CMBS issuance can scale into the hundreds of billions in a strong year.

The mezzanine and subordinate buyers

The AA-through-BBB- classes attract insurance companies reaching for yield and dedicated CMBS credit funds that underwrite the pool's loans more closely. These buyers do real credit work, because their bonds can be impaired in a moderately bad scenario, and they price the subordination cushion explicitly.

The first-loss buyer and the controlling class

The bottom of the stack is the most consequential. The unrated first-loss certificate, and often the lowest-rated classes above it, are bought by specialized B-piece buyers who underwrite every loan in the pool because they will absorb the first dollar of any loss. In exchange for that risk, the structure usually hands the most subordinate non-written-down class extraordinary control rights.

Controlling class

The most subordinate outstanding bond class in a CMBS deal, which holds the right to appoint and replace the special servicer and to approve major workout decisions on defaulted loans. The role typically belongs to the B-piece buyer, giving the first-loss investor outsized influence over how troubled loans are resolved.

Because the controlling-class holder bears the first losses, the structure lets it direct the workout: it names the special servicer (frequently an affiliate of the B-piece buyer itself) and signs off on loan modifications, extensions, and foreclosures. That alignment is intentional, since the party with the most to lose has the strongest incentive to maximize recoveries. It also gives the B-piece buyer real power over the deal, a dynamic explored in depth in the B-piece buyer, risk retention, and credit selection, and it sets up the special servicing and workout process that governs defaulted CMBS loans.

The structure described above is the post-crisis version, and the differences from the pre-2008 market are deliberate. The legacy CMBS market failed in 2008 because thin subordination, deteriorating underwriting, and misaligned incentives let aggressively structured deals carry AAA ratings they could not support. The rebuilt market, generally called CMBS 2.0, raised subordination levels sharply and tightened underwriting in concrete ways. Issuers moved away from sizing loans to optimistic pro-forma income and back toward in-place, trailing cash flow; interest-only concentrations were scrutinized rather than waved through; and the rating agencies, chastened by the crisis, demanded the thicker cushions that took AAA subordination from roughly 12% to roughly 20%. The market also added an appraisal-reduction mechanism, under which a defaulted loan whose collateral value has fallen triggers a reduction in the interest advanced to the most subordinate classes, accelerating the moment the first-loss holder feels the pain and protecting the trust from advancing money it will never recover. Each of these changes pushed risk back toward the parties best able to bear and price it.

The rebuilt market also added a governance counterweight to the concentrated power of the controlling class: the operating advisor. Every conduit CMBS 2.0 deal appoints an independent operating advisor who monitors the special servicer, reviews major workout decisions against the servicing standard, and gains expanded rights once the controlling class has been written down to zero. The logic addresses a specific pre-crisis flaw, that a wiped-out first-loss holder or its affiliated special servicer could keep steering workouts with nothing left to lose. When losses climb high enough to exhaust the B-piece, the operating advisor can recommend replacing the special servicer by a vote of the remaining bondholders, restoring some accountability to the parties now actually bearing the losses.

The most important structural change came from Dodd-Frank risk retention, effective at the end of 2016. The rule requires the sponsor of a securitization to retain 5% of the credit risk of the deal, so that the party creating the bonds keeps real skin in the game rather than originating purely to distribute.

These reforms changed the economics of issuing CMBS and elevated the importance of the first-loss buyer, but they did not change the underlying machine. Subordination still does the work; the waterfalls still run in opposite directions; the senior bond is still safe because the junior bonds stand in front of it. The crisis rewrote how thick each tranche must be and who must keep skin in the game, not the basic logic of slicing one pool of loans into a ladder of differently ranked claims.

A final point on geography: this securitized model is overwhelmingly a US phenomenon. The European CMBS market never recovered the depth it briefly had before 2008, and most European commercial mortgage risk stays on bank balance sheets or funds through covered bonds rather than the public tranche structure described here. A banker working cross-border deals should expect the CMBS playbook to apply in the US and to thin out quickly elsewhere, where the securitized channel gives way to the balance-sheet lenders mapped in the CRE debt universe.