Install
$ agentstack add skill-panaversity-agentfactory-business-plugins-ifrs9-ecl ✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README, it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
CORE PRINCIPLE
ECL is FORWARD-LOOKING and PROBABILITY-WEIGHTED. It is NOT the incurred loss. It is the probability-weighted expectation of credit losses given ALL reasonable and supportable information, including future economic conditions. Never wait for objective evidence of impairment to recognise a loss.
STAGE SUMMARY
| Stage | Trigger | ECL Horizon | Interest Income | | ------- | --------------------------------------------- | ------------ | ------------------------------------ | | Stage 1 | No SICR since origination | 12-month ECL | On GROSS carrying amount | | Stage 2 | SICR since origination — see ifrs9-staging.md | Lifetime ECL | On GROSS carrying amount | | Stage 3 | Credit impairment occurred | Lifetime ECL | On NET carrying amount (gross - ECL) |
CRITICAL: Stage 3 interest is on the NET amount. Recognising Stage 3 interest on the gross amount is a material accounting error.
ECL FORMULAS
12-Month ECL (Stage 1): ECL12 = PD12 x LGD x EAD
Lifetime ECL (Stage 2 and 3): ECLlife = Sumt [ PDmarginalt x LGDt x EADt x DFt ] where t = each future period until maturity DFt = discount factor at the asset's effective interest rate
Scenario-Weighted ECL (REQUIRED): ECL = Sums ( Weights x ECLscenarios ) Weights must sum to 1.0 and reflect management's genuine scenario probability assessment
DISCOUNT FACTOR TREATMENT
The discount factor (DF_t) is calculated using the asset's effective interest rate (EIR). For floating-rate instruments: use current EIR at the reporting date. For fixed-rate instruments: use the EIR at initial recognition.
DF_t = 1 / (1 + EIR)^t
Discounting matters because:
- Lifetime ECL for long-dated assets (e.g., 25-year mortgages) extends decades
- Without discounting, future losses are materially overstated in present-value terms
- The discount effect is largest for Stage 2 facilities with long remaining maturity
Example: A 1% marginal PD at year 20 with LGD 30% and EAD 100k: Undiscounted: 1% x 30% x 100k = 300 Discounted at 4% EIR: 300 / (1.04)^20 = 137 The discount effect reduces the contribution by more than half.
PD ESTIMATION
TTC PD: Long-run average over a full economic cycle — starting point only. PIT PD: REQUIRED for IFRS 9. PIT PD = TTC PD x Credit Cycle Adjustment (CCA). CCA > 1.0 in recession (PDs higher than long-run average) CCA 80% Unsecured consumer: LGD ~ 65-80% Corporate unsecured senior: LGD ~ 40-60%
PORTFOLIO SEGMENTATION
ECL models must be segmented by portfolios with homogeneous risk characteristics:
| Segment | Typical PD Model | LGD Approach | Key Drivers | | ---------------------- | --------------------------------- | --------------------------------- | ------------------------- | | Retail mortgages | Behavioural scorecard | Property collateral + forced sale | LTV, income, employment | | Consumer unsecured | Behavioural scorecard | Statistical cure rate model | Utilisation, bureau score | | SME | Application/behavioural scorecard | Collateral-dependent | Revenue, leverage, age | | Corporate | Rating model (PD master scale) | Workout LGD | Financial ratios, sector | | Commercial real estate | Rating model | Property collateral | LTV, DSCR, vacancy |
EAD AND CREDIT CONVERSION FACTORS (CCF)
Term loans: EAD = scheduled outstanding balance at default Revolving facilities: EAD = Drawn balance + (CCF x Undrawn committed amount) CCF: unconditionally cancellable ~ 0-10%; committed revolving corporate ~ 50-75%
MACROECONOMIC SCENARIOS
Minimum: base + 1 upside + 1 adverse. Best practice: 4 scenarios. Weights must reflect genuine management view — equal weights rarely defensible. Common structure: Upside 15%, Base 40%, Adverse 30%, Severe 15%. See ifrs9-scenarios.md for full satellite model framework.
POST-MODEL ADJUSTMENTS (PMAs)
Required when known model limitation would cause material under/overstatement. Common types: pandemic PMA, sector concentration PMA, new product PMA, climate PMA. Each PMA must be: documented, committee-approved, time-limited, reviewed quarterly. Aggregate PMA amount must be disclosed in IFRS 7 notes. NEVER use PMAs to substitute management conservatism for model output.
PROVISION MOVEMENT TABLE
Build every quarter. Every line must trace to a documented source: Opening provision -> New business -> Stage 1->2 migration -> Stage 2->3 migration -> Cures -> Repayments -> Write-offs -> Model parameter changes -> Macro scenario changes -> PMA movements -> FX -> Closing provision
IFRS 7 MANDATORY DISCLOSURES (minimum)
- SICR criteria (quantitative and qualitative)
- Definition of default used
- Write-off policy
- Macroeconomic scenarios: names, weights, key variables
- Sensitivity analysis: single-scenario stress
- Stage distribution table: count and amount by stage, by product
- Stage migration table: movements with ECL impact
- Credit quality distribution: by rating grade or score band
- Post-model adjustments: aggregate amount and rationale
- Modified financial assets: amounts and conditions
OUTPUT FORMAT — ECL CALCULATION SUMMARY
ECL CALCULATION SUMMARY
Entity: [Bank / Group name]
Reporting Date: [YYYY-MM-DD]
Portfolio: [Segment name]
INPUTS:
Gross Carrying Amount: [Amount]
Stage Distribution: Stage 1: [X%] | Stage 2: [Y%] | Stage 3: [Z%]
PD (12-month, base): [X.XX%]
LGD: [X%]
EAD: [Amount]
Discount Rate (EIR): [X.XX%]
SCENARIO ECL:
Upside ([W1]%): [Amount]
Base ([W2]%): [Amount]
Adverse ([W3]%): [Amount]
Severe ([W4]%): [Amount]
WEIGHTED ECL: [Amount]
of which PMA: [Amount] ([description])
PROVISION MOVEMENT:
Opening: [Amount]
Change this period: [+/- Amount]
Closing: [Amount]
NEVER DO THESE
- NEVER calculate ECL using the incurred loss methodology
- NEVER apply a single macroeconomic scenario (must be probability-weighted)
- NEVER use TTC PD without PIT conversion
- NEVER use current market value of collateral (use downturn LGD)
- NEVER recognise Stage 3 interest on the gross amount
- NEVER omit the discount factor for lifetime ECL — undiscounted lifetime ECL materially overstates the provision for long-dated assets
- NEVER use PMAs as a substitute for fixing a known model deficiency — PMAs are temporary overlays, not permanent model corrections
ALL OUTPUTS REQUIRE REVIEW BY A QUALIFIED PROFESSIONAL BEFORE USE IN REGULATORY FILINGS OR BUSINESS DECISIONS.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: panaversity
- Source: panaversity/agentfactory-business-plugins
- License: Apache-2.0
- Homepage: https://agentfactory.panaversity.org/docs/Business-Domain-Agent-Workflows
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet, be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.