Vertical, horizontal and L-shaped retention on an illustrative $1,000M conduit, and why the horizontal piece is far thicker than 5 per cent of the deal.
US CMBS risk retention requires the sponsor, or a B-piece buyer standing in for it, to hold 5 per cent of the deal's credit risk, and for the horizontal option that 5 per cent is measured on fair value, not face. On an illustrative $1,000M conduit whose bonds are worth $1,003.26M, the retained slice must be worth $50.16M. Because junior bonds trade at deep discounts, buying that value from the bottom of the stack takes $77.96M of face, 7.80 per cent of the deal, against $50.0M for a vertical slice.
Worked in full in CMBS and CRE CLOs by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
The conduit is the illustrative $1,000M deal used in how a CMBS waterfall works: eight principal classes paid sequentially and an interest-only class, X-A, on a $787.5M notional. The prices below stand in for the fair values a sponsor would derive from a discounted cash flow at disclosed discount rates. They are illustrative, not market levels. The rule set is the US one, Regulation RR under section 15G of the Exchange Act, in force for CMBS since December 2016: the sponsor must retain at least 5 per cent of the credit risk, and an eligible horizontal residual interest is measured at fair value under US GAAP, with the method and key inputs disclosed to investors. The European and UK test is set beside it at the end.
| Class | Face | Price | Fair value |
|---|---|---|---|
| A | 700.0 | 102.0 | 714.00 |
| A-S | 87.5 | 100.5 | 87.94 |
| B | 45.0 | 99.0 | 44.55 |
| C | 40.0 | 96.0 | 38.40 |
| D | 52.5 | 90.0 | 47.25 |
| E | 25.0 | 78.0 | 19.50 |
| F | 20.0 | 68.0 | 13.60 |
| G | 30.0 | 48.0 | 14.40 |
| X-A (notional 787.5) | 3.0 | 23.62 | |
| All ABS interests | 1,000.0 | 1,003.26 |
The IO class has no face but it is an ABS interest, so its $23.62M of value counts in the denominator. That is the first reason the requirement is not simply $50.0M.
Step 1: the requirement in value. Five per cent of the fair value of all ABS interests.
Required fair value = 5% × 1,003.26 = $50.16M
Step 2: fill it from the bottom. An eligible horizontal residual interest must be the most subordinate claim, first to absorb losses. Take Class G, then F, then E, at their fair values, until $50.16M is reached.
G + F + E = 14.40 + 13.60 + 19.50 = $47.50M of value on $75.0M of face
Still needed: 50.16 − 47.50 = $2.66M of value, from Class D at a price of 90.0: 2.66 / 90.0% = $2.96M of face
Horizontal face = 75.0 + 2.96 = $77.96M, 7.80% of the deal
Excel, with fair values in D2:D9 from Class A down to G: a running total from the bottom in E9 =SUM(D9:D$9), filled up; the deepest class is the first row where it reaches the target, and its partial face is =(Target-E10)/(C9/100) for that row, with the price in column C
The face is 1.56 times the 5 per cent a reader of the headline would expect, because every dollar of first-loss face is bought at well under par. In practice the deal is structured so that the retained classes land exactly on the requirement, often with a separate retention class carved out of the next class up, rather than a fraction of Class D.
| Option | What is held | Face held | Fair value held |
|---|---|---|---|
| Vertical | 5% of every class, IO included | 50.0 | 50.16 |
| Horizontal | G, F, E and part of D | 77.96 | 50.16 |
| L-shaped (2.5 / 2.5 split) | 2.5% vertical + horizontal worth 2.5% of fair value | 70.71 | 50.16 |
All three tie up the same $50.16M of capital at these prices; what differs is the risk. The L-shaped option can use any split whose vertical percentage and horizontal fair value percentage add up to 5; the even split here holds 2.5 per cent of each class ($25.0M of face, $25.08M of value) and a horizontal slice worth $25.08M: all of G and $15.71M of F. A 3 per cent pool loss, $30.0M, falls entirely on the horizontal holder, while the vertical holder bears 5 per cent of it, $1.50M. That asymmetry is why a horizontal holder is paid a B-piece yield and a vertical holder an average of the stack. CMBS is the one US asset class where the horizontal piece can be held by up to two third-party purchasers, in practice B-piece buyers, rather than the sponsor: each must pay cash for it without sponsor financing, review the credit of every loan in the pool, and accept an operating advisor on the deal. The initial purchaser must hold for at least five years and can then sell only to another purchaser that meets the same conditions; hedging the retained risk is prohibited. The pricing of that first-loss slice is worked in how to price a CMBS B-piece.
The face needed is a function of the price of the bottom classes. Shift the prices of D to G and recompute both the denominator and the slice.
| Price shift, D to G | FV of all interests | Required FV | Face needed | % of deal | Deepest class |
|---|---|---|---|---|---|
| +10 points | 1,016.01 | 50.80 | 70.23 | 7.02% | E |
| Base | 1,003.26 | 50.16 | 77.96 | 7.80% | D |
| −10 points | 990.51 | 49.53 | 86.91 | 8.69% | D |
| −20 points | 977.76 | 48.89 | 98.41 | 9.84% | D |
When spreads on junior paper widen, the required value barely moves but the face needed to reach it grows by more than $20M. The retention holder ends up owning deeper into the investment-grade mezzanine, which changes its return profile and its control rights.
The EU Securitisation Regulation (Article 6) and the UK regime require the originator, sponsor or original lender to retain a material net economic interest of at least 5 per cent, held continuously and not hedged, and every option is measured on nominal value, not fair value. Its first-loss option is a tranche of at least 5 per cent of the nominal value of the securitised exposures: 5 per cent of $1,000.0M is $50.0M of face, here all of G and F. At the illustrative prices that slice is worth $28.00M, little more than half the US horizontal requirement. The two regimes are separate tests, not a single standard: a deal sold to US and to EU or UK investors has to meet each one, with a retainer that qualifies under each, and a B-piece buyer that satisfies the US rule is not automatically an eligible retainer in Europe.
The common mistake is sizing the horizontal piece at 5 per cent of face: $50.0M of the bottom classes. That slice is worth $28.00M and fails the US test by $22.16M of value. The second is leaving the IO class out of the denominator, which understates the requirement whenever premium or IO bonds are in the deal. The third is fixing the face at pricing and forgetting that the fair value calculation, its discount rates and its assumptions must be disclosed before sale and the final figure after closing, so the number will be checked.
The book treats risk retention at fair value alongside the first-loss bid, and the free workbook for this case prices the B-piece against a target yield and a loss scenario.
It is the most subordinate slice of the deal, retained so that it absorbs losses before any other class, with a fair value of at least 5 per cent of the fair value of all ABS interests issued. In this illustrative $1,000M conduit that means $50.16M of value: Classes G, F and E in full and $2.96M of Class D, $77.96M of face in all.
Yes. US Regulation RR lets a CMBS sponsor satisfy the horizontal requirement through up to two third-party purchasers, typically B-piece buyers, if each pays cash without sponsor financing, reviews every loan, and the deal has an operating advisor. The initial purchaser must hold for at least five years, then may sell only to another qualifying purchaser. In this example it pays the fair value of the horizontal slice, $50.16M, priced to its own target yield.
Not in the US. Vertical retention is 5 per cent of each class, which is $50.0M of face on a $1,000M deal. Horizontal retention is measured as 5 per cent of fair value, so it depends on bond prices: here it needs $77.96M of face, and $98.41M if junior bond prices fall by 20 points. The EU and UK rules, by contrast, measure every option on nominal value.
This article is one calculation from CMBS and CRE CLOs. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: real estate investing, finance and fund management → · All 681 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.