SA-CCR: replacement cost and the multiplier · RiskAtoms Academy

By the end you will be able to

1Compute SA-CCR's replacement cost for unmargined and margined netting sets, with the threshold, the MTA and NICA.
2Compute the multiplier and explain why it never falls below 5%.
3Apply the cap on margined netting sets and say when a CSA lowers EAD and when it does not.
4Work through RiskAtoms Bank plc's four netting sets to their EAD and RWA, and show the effect of the UK's alpha of 1 from 2027.

Why it matters

Corriemoor Bank has the bank's biggest derivatives with one name: two swaps and a currency forward with notionals of £400m and $130m. Yet its exposure at default is only £5.69m, the smallest of the four. Take away its margin agreement and the same trades give £16.63m, 2.9 times as much. Ebbsworth also has a margin agreement, but it does nothing for capital. Both answers sit in two pieces of SA-CCR: the replacement cost and the multiplier.

Core explanation

Today's loss, and a credit for collateral

Illustrative dataRiskAtoms Bank plc's capital book

The same made-up UK bank as Level 3: £67bn of assets and £4bn of equity at 31 December 2025, in £ millions. Module 4.1 adds its regulatory layer: own funds, deductions, risk-weighted assets by risk type, buffers, the leverage exposure and MREL, computed from the Level 3 credit, trading, derivatives and operational risk books. Every figure is illustrative, not any bank's data.

Pro forma. The book applies Basel 3.1, the 2017 finalisation of Basel III, to the 31 December 2025 balance sheet. On that date the bank still reported under the old UK CRR. The UK applies Basel 3.1 from 1 January 2027, and the internal model approach for market risk from 1 January 2028 (PRA PS1/26, paras 1.23 and 1.24). So the book shows the bank as it would look under the new rules.

The counterparty data. The four OTC derivative netting sets of the Level 3 derivatives book (M3.4: Counterparty credit risk): Halvergate Moor Manufacturing (HAL, a manufacturer), Corriemoor Bank (COR), Saltney Capital (SAL, a hedge fund) and Ebbsworth Pension Trustees (EBB), all invented. Module 4.1 gave them SA-CCR exposure of £36.9m and counterparty credit risk RWA of £31.2m. That figure is SA-CCR on these four netting sets only, at Basel's α of 1.4.

Cleared trades and repo funding are left out; later lessons in this module add them (L4.4.03: The internal model method and SFTs, L4.4.04: Exposures to central counterparties).

One formula, two parts

L3.4.12: Managing counterparty risk introduced the standardised approach for counterparty credit risk, SA-CCR, and computed it for Halvergate Moor. Every netting set gets an exposure at default:

\text{EAD} = \alpha \times \big(\text{RC} + \text{PFE}\big), \qquad \text{PFE} = \text{multiplier} \times \text{AddOn}^{\text{aggregate}}

The replacement cost is what the bank loses if the counterparty defaults today. The potential future exposure is what it could lose if the default came later, after the trades had moved. This lesson takes the replacement cost and the multiplier; the add-on, built trade by trade, is L4.4.02: SA-CCR: the add-ons. Alpha is 1.4 (CRE52.1), carried over from the internal model method when SA-CCR was adopted in 2014 to replace the older current exposure and standardised methods. There it covers what an expected exposure model misses, such as wrong-way risk and a small number of counterparties (CRE53.15 names these as reasons for a higher α).

Replacement cost without a margin agreement

Without a CSA, RC is the value of the trades less the collateral the bank holds after haircuts, floored at zero: $\text{RC} = \max(V - C, 0)$ (CRE52.10). $C$ includes the net independent collateral amount, NICA: initial margin and other collateral the counterparty has posted, less any unsegregated collateral the bank has posted, which it could lose in the counterparty's insolvency (CRE52.17). Halvergate Moor has no CSA and both its trades are at market today, so its RC is zero. Floors and caps like this one are the subject of M-A11: Min, max, floors and caps.

Replacement cost with a margin agreement

Margin changes the question. Even when the trades are at market today, the counterparty may let the exposure grow up to its threshold, and then a little more, up to the minimum transfer amount, before any margin is called. So the margined RC is the largest exposure that would not trigger a call (CRE52.19):

\text{RC} = \max\big(V - C,\ \text{TH} + \text{MTA} - \text{NICA},\ 0\big)

Independent collateral comes off the middle term: initial margin covers the first part of any exposure. The three margined sets show each term in turn.

Which term sets RC? Guess, then open

The multiplier

Over-collateralisation should reduce capital (CRE52.21). The add-on, though, is built from the trades alone and ignores collateral. The multiplier brings it in: when V − C is negative, because the set is out of the money or the counterparty has posted extra collateral, the multiplier falls below 1. It decays exponentially (M-A05: The number e and continuous growth) in V − C measured against the add-on, and stops at a floor of 5% (CRE52.22): however much collateral the bank holds, 5% of the add-on stays. At V − C equal to minus one add-on the multiplier is 0.611; at minus three add-ons 0.246.

Saltney Capital shows it. The bank holds a put on the equity index worth £18.98m; Saltney has posted variation margin equal to that value and £3m of initial margin, so V − C = −£3m. Against an add-on of £6.878m the multiplier is 0.8051: the initial margin takes about a fifth off the PFE.

Margined or not: the cap

A margined set differs in two ways. Its RC can include TH + MTA, and its add-on uses a maturity factor based on the margin period of risk instead of the trade's maturity: $\tfrac{3}{2}\sqrt{10/250} = 0.3$ at the ten-day floor, against 1 for an unmargined trade of a year or more (L4.4.02: SA-CCR: the add-ons). So Corriemoor's add-on falls from £11.88m to £3.56m, and its EAD from £16.63m to £5.69m, even after the £0.5m of RC.

A large threshold reverses this. Ebbsworth's £10m threshold puts £11m into the margined RC, and its margined EAD, £20.36m, is above the £16.53m it would have with no CSA at all. The rules do not let a margin agreement raise the exposure: a margined set's EAD is capped at its unmargined EAD (CRE52.2). The cap binds, so Ebbsworth reports RC of zero and the unmargined PFE. Module 4.1's table (and L3.4.12: Managing counterparty risk) show the margined RC of £11m; the EAD of £16.53m is the capped one.

Ebbsworth's threshold	Margined RC	Margined EAD	Unmargined EAD	EAD
£10m	11	20.358	16.526	16.526
£5m	6	13.358	16.526	13.358
£2m	3	9.158	16.526	9.158
£0m	1	6.358	16.526	6.358

£ millions, MTA £1m. The CSA starts to lower Ebbsworth's EAD once the threshold is below about £7.3m.

Sort itDoes it lower SA-CCR EAD?

Each change is made to a margined netting set that is at market today. Sort it.

The counterparty posts £5m more initial margin, segregated with a custodian
The bank posts £5m of initial margin that the counterparty holds unsegregated
The CSA's threshold rises from zero to £10m
The bank posts £5m of initial margin held by a bankruptcy-remote custodian
Margin calls in the past two quarters led to three disputes lasting longer than the MPoR

The formal version

SA-CCR's EAD (CRE52.1, 52.2, 52.10, 52.18, 52.22-52.23; the formulas are images on bis.org and were read as text in the Basel Committee's March 2014 SA-CCR standard, paras 128, 136, 144 and 149):

\begin{aligned} \text{EAD} &= \alpha \times (\text{RC} + \text{PFE}), \qquad \alpha = 1.4 \\ \text{RC}_{\text{unmargined}} &= \max(V - C,\ 0) \\ \text{RC}_{\text{margined}} &= \max(V - C,\ \text{TH} + \text{MTA} - \text{NICA},\ 0) \\ \text{PFE} &= \text{multiplier} \times \text{AddOn}^{\text{aggregate}} \\ \text{multiplier} &= \min\Big\{1,\ \text{Floor} + (1 - \text{Floor}) \exp\Big(\frac{V - C}{2\,(1 - \text{Floor})\,\text{AddOn}^{\text{aggregate}}}\Big)\Big\}, \quad \text{Floor} = 5\% \\ \text{EAD}_{\text{margined}} &\le \text{EAD}_{\text{unmargined}} \end{aligned}

$V$ : the value of the derivatives in the netting set; $C$ : the haircut value of net collateral held, variation margin plus NICA (CRE52.10).
$\text{TH}$ : the counterparty's threshold; $\text{MTA}$ : its minimum transfer amount; $\text{NICA}$ : collateral posted by the counterparty other than variation margin, segregated or not, less unsegregated collateral posted by the bank (CRE52.17-52.18).
$\text{AddOn}^{\text{aggregate}}$ : the sum of the asset class add-ons (L4.4.02: SA-CCR: the add-ons). For a margined set, the add-on uses the MPoR-based maturity factor.
The last line is the cap of CRE52.2: compute a margined set both ways and take the lower EAD. The multiplier always lies between 0.05 and 1: the floor and the min are the two limits of M-A11: Min, max, floors and caps.

Try it

Start on Saltney. Remove its initial margin (NICA to zero) and watch the multiplier go to 1 and RC become the MTA. Then pick Ebbsworth and lower the threshold until the margined EAD drops below the unmargined one. Switch Corriemoor's CSA off, and try the UK alpha on Halvergate.

Try itReplacement cost, the multiplier and the cap

Netting set (resets to the book)

Saltney Capital (SAL)

Value of the trades less variation margin held, V − VM (£m)£0.00m

Net independent collateral held, NICA (£m)£3.00m

Margin agreement (CSA)

Threshold TH (£m)£0.0m

Minimum transfer amount MTA (£m)£0.25m

Alpha

The point is this set: (V − C) ÷ AddOn = −0.44.

Margined EAD

£7.75m

Unmargined EAD

£30.07m

V − C−£3.00m

RC = max(V − C, TH + MTA − NICA, 0)£0.00m

Multiplier (AddOn 6.878)0.8051

PFE = multiplier × AddOn£5.538m

EAD at α = 1.4£7.75m

RWA at 100%£7.75m

The margined EAD applies: £7.75m, below the unmargined £30.07m.

Worked example

RiskAtoms Bank plc's four netting sets

£ millions, 31 December 2025, Basel α = 1.4. Add-ons from L4.4.02: SA-CCR: the add-ons.

From collateral to RWA

Step 1 of 5

1
Halvergate Moor: unmargined
$\begin{aligned} \text{RC} &= \max(0 - 0,\ 0) = 0, \qquad \text{multiplier} = 1 \\ \text{EAD} &= 1.4 \times (0 + 1 \times 4.9531) = 6.934 \end{aligned}$

Regulatory lens

Basel: CRE52, from the Committee's March 2014 SA-CCR standard, which set an effective date of 1 January 2017 and replaced the current exposure and standardised methods. Jurisdictions as at 3 November 2026 as at 3 Nov 2026. The wider 2027 changes are in L4.1.08: Basel 3.1 from 2027: UK, EU and US.

In the UK

From 1 January 2027 the PRA's Counterparty Credit Risk (CRR) Part, Article 274(2), sets α = 1 for non-financial counterparties, pension scheme arrangements and entities that compensate pension scheme members. The PRA had found SA-CCR “significantly above” the internal model method for these counterparties, and paired the cut with removing their CVA exemptions (CP16/22, paras 7.65-7.77). On the bank's book, Halvergate and Ebbsworth qualify: their EAD falls to £4.95m and £11.80m, total EAD to £30.20m and RWA to £24.98m (−£6.21m).

Trades entered into before 2027 with counterparties that stay under the CVA transitional exemption carry an add-on of 60%, 40% and 20% of the difference between the EAD at 1.4 and at 1 in 2027, 2028 and 2029 (Art 274(2A)); it does not apply to the leverage ratio (Art 274(2B)). Treating all of Halvergate's and Ebbsworth's trades as pre-2027 and still under the CVA exemption, RWA would be £28.70m, £27.46m and £26.22m. Module 4.1 keeps Basel's 1.4, and so does the rest of this module.

In the EU

CRR Article 274(2), in the consolidated text as at November 2026, keeps α = 1.4 for every counterparty. On this book the EU figure is the Basel one, £31.19m.

In the US

The agencies' SA-CCR rule took effect on 1 April 2020, with mandatory compliance from 1 January 2022 for large, internationally active banking organisations; others may opt in. It removes the 1.4 alpha for derivative contracts with commercial end-users that use them to hedge or mitigate commercial risk (FDIC fact sheet; Federal Reserve press release, 19 November 2019).

Spark Atoms

Spark AtomA margin agreement that costs capital

Without the cap, Ebbsworth's £10m threshold would raise its EAD from £16.53m to £20.36m. The 3/2 in the margined maturity factor is the reason a margined set can come out above an unmargined one for short trades too, and CRE52.2's cap exists to stop that. A CSA with a large threshold is, for capital, no CSA at all.

Spark AtomWhy 5% and not zero

With no floor, enough collateral would drive the PFE to nothing. But collateral can lose value, and trades can jump before it is called. The Committee floored the multiplier at 5% of the add-on “for prudential reasons” (CRE52.22). Saltney's 80.5% is far from it; even at V − C of minus six add-ons the multiplier is still 9.0%.

Key takeaways

EAD = α × (RC + multiplier × AddOn) for each netting set, with α = 1.4 (CRE52.1).
Unmargined RC = max(V − C, 0). Margined RC = max(V − C, TH + MTA − NICA, 0): the largest exposure that would not trigger a margin call. NICA is independent collateral held, less unsegregated collateral posted.
The multiplier falls exponentially as V − C turns negative, between 1 and a 5% floor. Saltney's £3m of initial margin gives 0.805.
A margined set's EAD is capped at its unmargined one (CRE52.2). Corriemoor's CSA cuts its EAD from £16.63m to £5.69m; Ebbsworth's £10m threshold makes its CSA worthless, so the cap binds at £16.53m.
The book: EAD £36.90m, RWA £31.19m. In the UK from 2027, α = 1 for Halvergate and Ebbsworth gives RWA £24.98m, with a transitional add-on to 2029.

Reaction

Check what stuck

Question 1 · objective 1

An unmargined netting set is worth £8m to the bank, and the counterparty has posted £3m of collateral. What is RC, in £m?

£m

Question 2 · objective 1

A margined netting set has V − C = −£1m, a threshold of £5m, an MTA of £0.5m and NICA of £2m. What is RC, in £m?

£m

Question 3 · objective 2

V − C is −£10m and the aggregate add-on is £5m. What is the multiplier? (Four decimals.)

Question 4 · objective 3

Ebbsworth's margined EAD is £20.36m and its unmargined EAD £16.53m. What does the bank report?

A£16.53m: the margined EAD is capped at the unmargined oneB£20.36m: a margined set always uses the margined formulaCThe average of the twoDZero, because the trade is at market

Question 5 · objective 4

From 1 January 2027, which of the bank's counterparties take α = 1 under the PRA's rules?

AHalvergate Moor (a manufacturer) and Ebbsworth (a pension scheme)BAll fourCCorriemoor and Saltney, which post collateralDNone: the UK keeps 1.4

Glossary

Exposure at default (SA-CCR): Exposure at default (EAD) under SA-CCR: α × (RC + PFE) for each netting set (CRE52.1). The risk weight of the counterparty multiplies it.
Replacement cost (RC): Replacement cost (RC): what the bank would lose if the counterparty defaulted today, after collateral, never below zero (CRE52.10, 52.18).
Potential future exposure (PFE): Potential future exposure (PFE): multiplier × aggregate add-on, a supervisory estimate of how far the exposure could grow (CRE52.20).
NICA: Net independent collateral amount (NICA): collateral other than variation margin that the counterparty has posted, segregated or not, less any unsegregated collateral the bank has posted (CRE52.17).
Threshold (TH): Threshold (TH): the exposure the counterparty may run up before it must post variation margin (CRE52.18).
Minimum transfer amount (MTA): Minimum transfer amount (MTA): the smallest margin call either side will make (CRE52.18).
Multiplier: Multiplier: a factor between 5% and 1 that cuts the add-on when the netting set is out of the money or over-collateralised (CRE52.21-52.23).
Margined cap: Margined cap: the EAD of a margined netting set may not exceed its EAD calculated as if it were unmargined (CRE52.2).
Alpha (α): Alpha (α): 1.4, carried over from the internal model method to SA-CCR. In the UK from 1 January 2027, 1 for non-financial counterparties and pension schemes (PRA Art 274(2)).
Margin period of risk (MPoR): Margin period of risk (MPoR): the time from the last good margin exchange to closing out a defaulted counterparty; at least 10 business days for bilateral daily-margined trades (CRE52.50).

Deep Dives & Origins

Deep Dives

Basel Committee on Banking Supervision, The standardised approach for measuring counterparty credit risk exposures, March 2014 (rev. April 2014): the rationale for each piece and worked examples in its Annex 4. Intermediate.
Prudential Regulation Authority, CP16/22, chapter 7: the case for the UK alpha of 1. Intermediate.
M6.5: Counterparty risk and XVA: exposure simulation, the internal model method and XVA. Advanced.

Origins

Basel Committee on Banking Supervision, CRE52: Standardised approach to counterparty credit risk, paras 52.1-52.23 and 52.50-52.53, version effective 1 January 2023, bis.org (also the Swiss SIF copy).
Basel Committee on Banking Supervision, The standardised approach for measuring counterparty credit risk exposures, March 2014, paras 128, 136, 144 and 149.
Basel Committee on Banking Supervision, CRE53: Internal models method for counterparty credit risk, para 53.15 (Swiss SIF copy, version effective 1 January 2023).
Prudential Regulation Authority, PRA Rulebook: Counterparty Credit Risk (CRR) Part, Articles 274-275, version from 1 January 2027; CP16/22, November 2022, chapter 7; PS1/26, 20 January 2026, para 1.23.
Regulation (EU) No 575/2013 (CRR), Article 274, as amended by Regulation (EU) 2024/1623.
FDIC, Federal Reserve and OCC, Standardized Approach for Calculating the Exposure Amount of Derivative Contracts, fact sheet, fdic.gov; Board of Governors of the Federal Reserve System, press release, 19 November 2019.

Watch this lesson

Video version

Prefer to watch? The same lesson as a 12-minute video will sit here, made from this page's script.