Skip to main content
Updated Mar 07, 2026

Macroeconomic Scenarios and Post-Model Adjustments

In Lessons 3 and 4, you built the three components of the ECL formula — PD, LGD, and EAD — and calculated facility-level ECL. But those calculations assumed a single set of economic conditions. IFRS 9 requires something more: banks must estimate ECL under multiple macroeconomic scenarios and calculate a probability-weighted average. And when the models still fall short — when they cannot capture an emerging risk, a structural change, or a once-in-a-generation event — banks must apply Post-Model Adjustments.

This lesson completes the IFRS 9 picture. After this lesson, you will have every component needed to calculate a full IFRS 9 ECL from raw data to final provision — and you will understand why the number your model produces is almost never the number that appears in the financial statements.

Why Multiple Scenarios?

IFRS 9 paragraph 5.5.17 requires that the ECL measurement reflect "an unbiased and probability-weighted amount that is determined by evaluating a range of possible outcomes." A single-scenario ECL — even a "most likely" scenario — is not compliant because it fails to capture the asymmetric distribution of credit losses.

The key insight is non-linearity: credit losses do not increase linearly with economic deterioration. A 10% decline in GDP does not produce twice the losses of a 5% decline — it may produce three or four times the losses, because default rates accelerate and recovery rates collapse simultaneously. This means:

The Non-Linearity Principle

The expected credit loss is NOT the credit loss under the expected scenario.

If the base case ECL is $45M and the adverse case ECL is $78M, the probability-weighted ECL will always be higher than $45M — even if the base case has the highest probability weight. This is because the adverse scenario's impact is disproportionately large.

Designing Macroeconomic Scenarios

IFRS 9 requires a minimum of two scenarios (base plus at least one adverse), but best practice among major banks is three to five scenarios:

ScenarioDescriptionTypical Weight
UpsideBetter-than-expected economic outcomes10-20%
BaseMost likely economic trajectory35-50%
AdverseModerate economic downturn25-35%
Severe AdverseExtreme but plausible stress event5-15%

What Drives Each Scenario?

Each scenario is defined by a set of macroeconomic variables that affect credit risk. The most common variables are:

VariableEffect on ECLMechanism
GDP growthHigher GDP = lower PDs, lower ECLEconomic growth reduces defaults
Unemployment rateHigher unemployment = higher PDsJob loss is the primary driver of retail defaults
Property pricesLower prices = higher LGDs (less collateral coverage)Mortgage and CRE portfolios are directly affected
Interest ratesHigher rates = higher PDs for variable-rate borrowersDebt service burden increases
Oil/commodity pricesSector-specific PD impactEnergy, mining, and commodity-linked sectors

Scenario Calibration Example

A bank designs four scenarios for its 2025 year-end ECL calculation:

ScenarioGDP GrowthUnemploymentProperty PricesProbability Weight
Upside+3.5%3.8%+5%15%
Base+1.8%4.5%+1%40%
Adverse-1.0%6.5%-10%30%
Severe-3.5%9.0%-25%15%

Probability weights must sum to 1.0 (100%). The weights reflect the bank's assessment of the likelihood of each scenario — not an objective probability, but an informed judgment that must be documented and approved by the credit committee.

Calculating Probability-Weighted ECL

Once the bank has calculated the ECL under each scenario (using the PD, LGD, and EAD methodology from Lessons 3-4, with each scenario's macroeconomic variables feeding into the PD model), the probability-weighted ECL is:

Probability-Weighted ECL = Sum of (Weight x ECL) across all scenarios

Worked Example

A commercial real estate portfolio produces the following ECL under each scenario:

ScenarioWeightPortfolio ECLWeighted Contribution
Upside15%$28M$4.20M
Base40%$45M$18.00M
Adverse30%$78M$23.40M
Severe15%$145M$21.75M
Total100%$67.35M

Probability-Weighted ECL = $67.35M

Notice: the base case ECL is $45M, but the probability-weighted ECL is $67.35M — 49% higher. This is the non-linearity effect in practice. The adverse and severe scenarios contribute disproportionately because their ECL amounts are much larger than the base case. Even though the base case has the highest probability weight (40%), the tail scenarios pull the weighted average significantly above the most likely outcome.

This is why IFRS 9 requires multiple scenarios. A bank reporting only the base case ECL of $45M would be understating its expected losses by $22.35M.

Sensitivity Analysis

Changing the probability weights materially changes the ECL:

Weight ScenarioUpsideBaseAdverseSevereWeighted ECLChange
Original15%40%30%15%$67.35MBaseline
More optimistic20%45%25%10%$57.45M-15%
More pessimistic10%35%35%20%$77.10M+14%

This sensitivity demonstrates why probability weights are among the most scrutinised assumptions in bank financial statements. A bank that assigns optimistic weights reports lower provisions and higher profits. Auditors, regulators, and analysts challenge weight assignments rigorously.

Post-Model Adjustments (PMAs)

Even with multiple scenarios, ECL models cannot capture everything. Models are built on historical data and statistical relationships. When something happens that is outside the model's historical experience — a pandemic, a geopolitical shock, a structural industry change — the model output must be adjusted.

Post-Model Adjustments (PMAs) are management overlays applied on top of the model ECL to capture risks that the model cannot quantify. They are the bridge between what the model says and what management believes the actual expected loss to be.

When Are PMAs Needed?

SituationExamplePMA Type
Model limitationModel not recalibrated for recent data; historical loss rates do not reflect current conditionsTechnical PMA
Emerging risk not in modelCovid-19 pandemic: no historical data for global lockdownsEvent PMA
Sector concentration30% of portfolio in commercial real estate during a property downturn; model treats CRE like other commercialSector PMA
Data quality issueRating system migration in progress; some borrowers have stale ratingsData PMA
Climate/transition riskExposure to carbon-intensive industries facing regulatory transition; model does not include climate variablesClimate PMA

PMA Governance: The Audit Focus

PMAs are the single most scrutinised element of IFRS 9 reporting. Because PMAs represent management override of quantitative models, they are inherently subjective — and auditors treat them with heightened scepticism. Every PMA must satisfy strict governance requirements:

Governance RequirementWhat It Means
Documented rationaleWritten explanation of why the model output is insufficient and what the PMA corrects
Committee approvalApproved by the credit committee or equivalent senior management body — not by the modelling team alone
Quantified impactSpecific dollar amount, with methodology for calculation (not "approximately $10-20M")
Time-limitedEvery PMA must have an expiry date or review trigger — PMAs are temporary adjustments, not permanent model changes
Quarterly reviewRe-assessed every reporting period; removed when the underlying issue is resolved or incorporated into the model
Back-testingWhere possible, compared against actual outcomes to validate the adjustment was appropriate
Why PMAs Matter for Auditors

PMAs are the primary audit focus because they represent the area of greatest management judgment and the greatest risk of earnings management. A bank could use a PMA to artificially increase or decrease its provisions. The governance framework exists to ensure that PMAs are genuine risk adjustments, not earnings manipulation. External auditors will challenge the rationale, quantum, and continued relevance of every material PMA.

PMA Example: Covid-19

During Q1 2020, banks globally applied Covid-19 PMAs because:

  1. No ECL model had been calibrated for a global pandemic with simultaneous lockdowns
  2. Macroeconomic scenarios designed in December 2019 did not include a pandemic scenario
  3. Government support schemes (furlough, payment holidays) obscured true credit quality — borrowers on payment holidays were technically current but potentially distressed

Major banks applied PMAs of $1-10 billion on top of their model-calculated ECL. As the pandemic progressed and models were recalibrated to include pandemic data, these PMAs were gradually released — but the process took 18-24 months.

Putting It All Together: The Complete IFRS 9 ECL

After five lessons, you now have all the components for a complete IFRS 9 ECL calculation:

StepComponentLessonPlugin Skill
1Stage classification (1/2/3)Lesson 3ifrs9-staging
2PD calibration (TTC to PIT, term structure)Lesson 4ifrs9-ecl
3LGD estimation (downturn, collateral, cure rates)Lesson 4ifrs9-ecl
4EAD calculation (drawn + CCF x undrawn)Lesson 4ifrs9-ecl
5ECL = PD x LGD x EAD (per facility, per scenario)Lesson 3ifrs9-ecl
6Macroeconomic scenario weightingThis lessonifrs9-scenarios
7Post-Model AdjustmentsThis lessonManagement judgment
8Final reported ECL = Weighted ECL + PMAThis lessonCombined output

The /bank-ecl command chains steps 1-6 automatically. Step 7 (PMAs) requires professional judgment that the agent cannot automate — it can draft the PMA documentation using the ifrs9-scenarios skill, but the quantum and rationale must come from management.

Exercise 2: GCC Corporate Portfolio

Apply macroeconomic scenario analysis and PMA design to this GCC-region corporate portfolio:

BorrowerSectorBalance ($M)StageBase ECL ($M)Adverse ECL ($M)Severe ECL ($M)
AlFahad GroupTrading5010.82.14.5
Gulf PetrochemicalsOil & Gas12011.55.815.2
Emirates ConstructionReal Estate8028.518.032.0
Doha HealthcareHealthcare3010.30.61.2
Riyadh MetalsMining4524.29.518.0
Kuwait Retail CorpConsumer2510.41.02.5

Scenario weights: Upside 15% (assume ECL is 70% of Base), Base 40%, Adverse 30%, Severe 15%.

Tasks:

  1. Calculate the upside ECL for each borrower (Base ECL x 0.70).
  2. Calculate the probability-weighted ECL for each borrower.
  3. Sum to get the portfolio-level probability-weighted ECL.
  4. Design a PMA for the Oil & Gas sector — Gulf Petrochemicals has $120M exposure and oil prices have dropped 25% in the last quarter. The base case macro scenarios were calibrated with oil at $80/barrel; current price is $60/barrel. Estimate the PMA amount and document the rationale.
  5. Calculate the final reported ECL (probability-weighted ECL + PMA).

Using the Banking Plugin

The ifrs9-scenarios skill calculates probability-weighted ECL across multiple macroeconomic scenarios and drafts PMA documentation. Here is a worked example.

Worked example. Priya Sharma, Chief Economist at Crescent Bank, needs to run a four-scenario overlay on her commercial real estate portfolio. She types:

Calculate probability-weighted ECL for this CRE portfolio:
Portfolio balance: GBP 800M
Scenario ECLs:
Upside: GBP 28M (weight 15%)
Base: GBP 45M (weight 40%)
Adverse: GBP 78M (weight 30%)
Severe: GBP 145M (weight 15%)
Also draft a PMA rationale — oil prices dropped 25%
and our base macro was calibrated at $80/bbl, now $60/bbl.
Jurisdiction: UK PRA

The agent routes through ifrs9-scenarios and responds:

GOVERNING STANDARD: IFRS 9 — UK PRA
DOMAIN: IFRS 9 ECL — Scenario Weighting + PMA
JURISDICTION: United Kingdom — PRA Rulebook / UK CRR

PROBABILITY-WEIGHTED ECL
Upside: 15% x GBP 28M = GBP 4.20M
Base: 40% x GBP 45M = GBP 18.00M
Adverse: 30% x GBP 78M = GBP 23.40M
Severe: 15% x GBP 145M = GBP 21.75M
Weighted ECL: GBP 67.35M

Non-linearity impact: Weighted ECL is 49% above base case
ECL (GBP 45M) due to disproportionate adverse tail.

PMA DRAFT
Type: Sector PMA — Oil price shock
Rationale: Base macro calibrated at $80/bbl; current
spot $60/bbl (-25%). CRE portfolio has 18% exposure
to energy-dependent tenants. Model does not incorporate
oil price as an explicit variable.
Suggested PMA: GBP 3.2M (methodology: stress the
energy-linked sub-portfolio PDs by 1.5x CCA uplift)
Governance: Requires credit committee approval,
quarterly review, 6-month expiry.

NOTE: The professional reviews the scenario weights
and the PMA quantum; the agent calculated the weighted
ECL and drafted the PMA documentation.

Priya reviews the scenario weights against the bank's latest economic outlook and validates the PMA quantum by cross-checking the energy-linked sub-portfolio exposure against her internal concentration report.

Try With AI

Use these prompts in Claude or your preferred AI assistant to explore this lesson's concepts.

Prompt 1: The Non-Linearity Demonstration

A bank's ECL under three scenarios is:
- Base case: $45M (probability weight 50%)
- Upside: $28M (probability weight 20%)
- Adverse: $78M (probability weight 30%)

Calculate the probability-weighted ECL.

Then explain to a CFO why the probability-weighted ECL
is higher than the base case ECL, even though the base
case has the highest probability weight.

Use a simple analogy to illustrate the non-linearity
principle. Why is the expected credit loss NOT the
credit loss under the expected scenario?

What you are learning: The non-linearity principle is the reason IFRS 9 requires multiple scenarios. By calculating and explaining the difference between the base case and weighted ECL, you build the ability to communicate IFRS 9 provisioning decisions to non-technical stakeholders — a critical professional skill for credit risk officers and finance directors.

Prompt 2: PMA Design and Governance

A bank has $500M in hospitality sector loans (hotels,
restaurants, tourism). A pandemic has been declared.
The bank's ECL models were calibrated on pre-pandemic
data and do not include a pandemic scenario.

Design a PMA for this exposure. Include:
1. The rationale: why is the model output insufficient?
2. The methodology: how would you estimate the PMA
amount? What data would you use?
3. The governance documentation: what must be documented
for audit?
4. The review schedule: when should this PMA be
reassessed?
5. The exit criteria: under what conditions should the
PMA be removed?

Explain why this PMA would be the primary focus of the
external audit engagement.

What you are learning: PMAs are where professional judgment meets quantitative modelling. By designing a PMA yourself, you develop the ability to bridge the gap between what models produce and what the financial statements should reflect. This is the highest-value skill in IFRS 9 practice — it requires both technical knowledge (what the model misses) and governance awareness (how to document management judgment defensibly).

Prompt 3: Scenario Weight Sensitivity

A bank reports these ECL amounts under four scenarios:
- Upside ($25M), Base ($40M), Adverse ($70M), Severe ($130M)

Calculate the probability-weighted ECL under three
different weight schemes:

Scheme A (Optimistic): 25/45/20/10
Scheme B (Neutral): 15/40/30/15
Scheme C (Pessimistic): 5/30/35/30

Show the weighted ECL for each scheme. What is the
range between the most optimistic and most pessimistic?

Then explain: why do regulators and auditors challenge
probability weights? What stops a bank from using
optimistic weights to report lower provisions?

What you are learning: Probability weight selection is one of the most contentious areas in IFRS 9 reporting. By calculating ECL under different weight schemes, you see how material the weight choice is — and why regulatory and audit scrutiny focuses on whether banks are assigning appropriate probabilities to adverse scenarios. This prepares you for the professional judgment required in real ECL reporting.

Flashcards Study Aid


Continue to Lesson 6: Basel III/IV Capital Adequacy — CET1, Tier 1, Total Capital →