PRISM is the operating framework Ezelman applies to every stress mandate. It governs the delivery sequence from the moment the EBA methodological note lands on the CRO's desk to the moment the P2G dialogue closes with the JST. This is the external technical note — a 4-6 page methodology appendix written for CROs, CFOs, Heads of Model Risk and ex-supervisors.
PRISM is the framework we invoke on every stress-test engagement. It exists because stress testing is routinely approached as a data-production exercise — templates filled, numbers published — when it is really a decision-grade diagnostic about capital adequacy, P2G formation and the MB buffer strategy.
The order of the five steps is non-negotiable. Pressure must precede Reverse; Reverse must precede Iterate; Scenario and Map close the loop. Skipping a step does not compress delivery — it produces a submission that fails methodology review, fails challenger reconciliation or fails the P2G defence.
PRISM is cycle-agnostic. The same five steps run on an EBA EU-wide exercise, on an SSM thematic test (e.g. climate, cyber, geopolitical), on an ICAAP internal run, on a reverse stress test for the Board, and on an ORSA-adjacent exercise for an insurance affiliate. The step labels do not change; the inputs and deliverables do.
Each step is a discrete phase with a named partner-in-charge and a hard gate at close. The gate produces an artefact; no subsequent step opens until that artefact is signed.
Catalogue real pressure points, not the ICAAP ones
Cataloguing the bank's genuine pressure points — the exposures, cohorts and second-order effects most likely to break the CET1 path — not the comfortable list written into last year's ICAAP. Runs inside ANCHOR™ for the T-0 build. Outputs a ranked pressure register: top-10 exposures, top-5 behavioural assumptions, top-3 off-balance-sheet risks.
The scenario that breaks, not the one that validates
Genuine reverse stress: start from the failure point (CET1 below hurdle, MDA breach, AT1 coupon cancellation) and engineer backwards to the macro path that produces it. Most banks treat reverse stress as a comfort exercise. PRISM treats it as the single highest-leverage artefact in the SREP dialogue.
Calibration bench · second-order effects
Iteration is where second-order effects surface: credit losses compress NII via lower-yielding refinanced books; RWA density rises as defaults migrate; the Output Floor bites harder at adverse than baseline. Without this bench, the bank's projection is first-order and a single standard deviation below peer.
P&L, BS, RWA a CFO recognises
Translate scenarios into outputs the CFO actually uses: the P&L in three scenario years, the balance-sheet evolution, the RWA walk, the capital plan. If the output does not drop into the bank's existing capital-planning template, it will not inform the MB decision.
ICAAP · MB buffer · P2G dialogue
Integrate the EBA output into the ICAAP, into the MB buffer logic, and into the P2G dialogue. A stress test that does not change a capital decision is not a stress test — it is a regulatory return. Map is where PRISM converts numbers into capital-plan consequences.
Each step consumes specific inputs and produces specific deliverables. If an input is missing, the step cannot close. If a deliverable is missing at close, the subsequent step does not open.
Five principles apply across every PRISM step. They are not values statements — they are delivery constraints that shape who does what, when, and with which counter-signature.
Every PRISM step has a named partner in charge. Partners do not delegate to juniors in the defence. The CRO briefing is by a partner; the EBA call is with a partner on the line.
Every calibration is grounded in the EBA methodological note, ESRB macro variables, or ECB public guidance. If a choice has no regulatory basis, we can't defend it — so we don't make it.
Credit, market, operational, NII, liquidity run on shared macro drivers and a single scenario-translation matrix. No islands, no double counts, no silo calibrations.
Every artefact is built assuming the JST will read it tomorrow. Written clarifications within 24h. Challenge briefing books ready on submission day, not after the first call.
We publish benchmarks. Where a client's projection sits in the 25-50-75 percentile distribution of peer submissions — and how far from the EBA top-down — is known before we defend it.
ANCHOR and PRISM run together. The two are not separable on a stress-test mandate; attempting to do one without the other produces either a well-governed dataset that never makes a capital decision (ANCHOR without PRISM) or a projection built on sand (PRISM without ANCHOR).
In one line: PRISM is the delivery sequence; ANCHOR is the starting-point protocol inside step P. You cannot run a PRISM engagement without ANCHOR, and ANCHOR by itself does not produce a capital decision.
Four mandatory gates. A PRISM engagement cannot submit without every gate closed and signed.
Translation matrix from macro drivers to PD, LGD, market shocks and NII paths, CRO-signed before any production run.
Production projection vs independent challenger — variance < 8% on adverse CET1 end-point. Material variance explained line by line.
EBA output mapped into Pillar 2, MB buffer, capital plan — CFO-countersigned before submission.
Outlier cards, anticipated-question matrix, response scripts — ready on submission day, not after the first call.
What a client receives at each PRISM step. Every deliverable is partner-signed and archived in the post-submission pack.
Ranked pressure register, materiality thresholds, briefing to CRO. Typically 15-25 pages. Signed by partner and Head of Stress Testing.
Reverse stress scenario, macro path, implausibility score, Board briefing deck. Typically 20-30 pages. Signed by CRO.
Second-round effects note, Output Floor bite analysis, calibration changes tracked. Typically 10-15 pages.
Full P&L, BS, RWA, CET1 walk. EBA submission templates. CFO-countersigned. 40-60 pages + data.
ICAAP integration, MB buffer recalibration, P2G dialogue pack, MDA-distance note. 25-35 pages. Partner + CRO + CFO countersigned.
Tamper-evident archive: every input, every sign-off, every challenge response. Delivered to the Risk Committee within 10 working days of the EBA close.
PRISM™ is an Ezelman proprietary framework. This technical note is released for reader reference; the framework may be cited with attribution but not re-used as the house methodology of another adviser.
If you spot an error, a misreference to an EBA paragraph or a calibration we should update, write to contact@ezelman.com. Substantive corrections are acknowledged in a dated revision log. Figures are sourced to the EBA 2018 / 2021 / 2023 / 2025 transparency disclosures and the Ezelman EU G-SIB stress-test dataset.