The three ways a bank can meet the 5 per cent risk retention rule in a synthetic securitisation, priced in capital released on one fictional corporate portfolio.
In a synthetic SRT where the bank keeps the senior, random selection costs less capital than a vertical slice at ordinary CET1 targets, and retaining the first loss does not work at all. On the fictional Northwall 2026-1, keeping 5 per cent of the junior tranche leaves 191.46 million of capital after the trade; keeping 200 million of randomly selected loans outside the pool leaves 189.35 million, a saving of 2.10 million at a 13.50 per cent CET1 target. The saving shrinks as the target rises and turns negative above 15.89 per cent.
Worked in full in Significant Risk Transfer by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
The originator of a securitisation must keep a material net economic interest of at least 5 per cent. Article 6(3) of the EU Securitisation Regulation lists five ways to hold it. For a static synthetic deal on a corporate book three are realistic:
The amounts are the book's, in euro millions, and illustrative.
| Input | Value |
|---|---|
| Loans available for the pool | 4,000 |
| IRB RWA of those loans (density) | 3,077.08 (76.93%) |
| Junior tranche, detaching at | 7.00% |
| Senior risk weight, SEC-IRBA | 35.34% |
| CET1 target | 13.50% |
| Margin paid on the protection | 10.50% |
| Capital before the trade | 415.4 |
Vertical slice. The junior is 7.00 per cent of 4,000, or 280.0. The bank keeps 14.0 and sells 266.0. The senior, 3,720 at 35.34 per cent, needs 177.46 of capital at 13.50 per cent, and the slice is deducted at par, which is cheaper than weighting it at 1250 per cent (see deduct or risk-weight the retained sliver).
Capital after = 177.46 + 14.0 = 191.46; released = 415.4 − 191.5 = 223.9
Random selection. The bank keeps 200.0 of loans out and securitises 3,800. The junior, still 7.00 per cent, is 266.0 and the fund buys all of it, so the notional sold and the protection cost do not change. The 200.0 kept carry the pool's average density.
Capital on the loans kept = 200.0 × 76.93% × 13.50% = 20.77
Senior = 3,534 × 35.34% × 13.50% = 168.58
Capital after = 20.77 + 168.58 = 189.35; released = 226.1, 2.10 more than the slice
Excel: =Kept*Density*Target+(Pool-Kept)*(1-Detach)*SeniorRW*Target
The book keeps 5 per cent of the whole 4,000, which is 5.26 per cent of the 3,800 securitised. The rule asks only for 5 per cent of the securitised exposures, so the minimum is 4,000 × 5% ÷ 1.05 = 190.48. Kept at that size, capital after is 188.79 and the saving 2.67, with 266.7 sold.
First-loss tranche. Keeping 5 per cent of 4,000 at the bottom means keeping 200.0 of the 280.0 junior and selling 80.0. That is 71.4 per cent of the positions weighted at 1250 per cent, far above the 20 per cent the first-loss test allows (worked in how to pass the first-loss test). The transfer is not significant.
Capital released = 0; margin still paid on 80.0 = 8.40 a year
Even if it were recognised: 177.5 + 200.0 deducted = 377.5 after, 37.9 released
| Option | Retained | Sold | Capital after | Released | Margin a year | After-tax cost per euro released |
|---|---|---|---|---|---|---|
| Vertical slice | 14.0 | 266.0 | 191.5 | 223.9 | 27.93 | 9.35% |
| Random selection, 5% of 4,000 | 200.0 | 266.0 | 189.4 | 226.1 | 27.93 | 9.27% |
| Random selection, minimum | 190.5 | 266.7 | 188.8 | 226.6 | 28.00 | 9.27% |
| First-loss tranche | 200.0 | 80.0 | 415.4 | 0.0 | 8.40 | - |
At an 11.5 per cent cost of equity and a 25 per cent tax rate, the slice creates 4.81 a year of value and random selection 5.05. The difference looks small against the deal, but it costs nothing: the same notional is sold at the same margin.
The headline saving is noted in passing in the deduction article linked above; what matters for structuring is where it comes from and where it reverses. The two options differ in what the bank holds. The slice is 14.0 deducted, one euro of capital per euro kept. Random selection is 200.0 of ordinary loans at 10.39 cents of capital per euro, less the capital on the 186.0 of senior that is no longer in the pool. In RWA terms it costs 153.85 of loans kept against 65.72 of senior avoided, a net 88.13 of RWA. The slice costs 14.0 whatever the target; the loans cost the target times 88.13. The two meet at 14.0 ÷ 88.13 = 15.89 per cent.
| CET1 target | Vertical slice | Random selection | Random selection saves |
|---|---|---|---|
| 10.50% | 171.1 | 175.8 | +4.75 |
| 12.00% | 197.5 | 200.9 | +3.42 |
| 13.50% | 223.9 | 226.1 | +2.10 |
| 15.00% | 250.4 | 251.2 | +0.78 |
| 15.89% | 266.0 | 266.0 | +0.00 |
| 17.00% | 285.6 | 284.7 | −0.98 |
Other things equal, a pool with a higher RWA density moves the break-even down, because every loan kept carries more RWA. The operational cost runs the other way: random selection needs a documented draw, a population of at least 100 eligible exposures, and loans that stay outside the pool for the life of the deal, where they must not be hedged or sold.
The common mistake is to treat the retention as a pure compliance item and default to the first-loss tranche because it looks like the strongest alignment of interest. In a synthetic SRT it collides with the first-loss test and the capital relief disappears, while the coupon is still owed. The second mistake is to compare the options on the amount retained: 200.0 of loans looks fourteen times worse than a 14.0 slice, but the slice is deducted at par and the loans are not.
Both the vertical slice and the random selection are computed side by side in the free workbooks for Northwall 2026-1, with the CET1 target in a blue cell.
Article 6(3) of the Securitisation Regulation allows five: a vertical slice of at least 5 per cent of each tranche sold, a seller's share for revolving pools, randomly selected exposures worth at least 5 per cent of the securitised ones (from at least 100 potential exposures), the first-loss tranche, or a first-loss share of every exposure. In an illustrative synthetic SRT the two that work are the slice and random selection.
Because the first-loss test caps the bank at 20 per cent of the positions weighted at 1250 per cent. On an illustrative 4,000 million pool with a 280 million junior tranche, a 5 per cent first-loss retention is 200 million, 71.4 per cent of the junior. The transfer is not significant, no capital is released, and the bank still pays 8.40 million a year of margin on the 80 million it sold.
When the bank's CET1 target is high. The slice costs its face in deducted capital; random selection costs the target ratio on the retained loans' RWA, net of the smaller senior. On an illustrative pool that net RWA is 88.13 million, so the two cost the same at a 15.89 per cent target. Below it random selection wins: 2.10 million at 13.50 per cent, 4.75 million at 10.50.
This article is one calculation from Significant Risk Transfer. 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: credit, private credit and distressed debt → · All 1340 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.