IWhen clients call
Services / 10 — Financial instruments & IFRS 9
A model nobody else can replicate is a model that will be challenged.
Classification and measurement, expected credit loss modelling, fair valuation under IFRS 13, hedge accounting and IFRS 7 disclosure — for Pakistani, Gulf and international reporting.
Expected credit loss and fair value are the most judgement-heavy numbers in a set of financial statements, which makes them the most argued over. The difference between a model that is accepted and one that is contested is almost never the mathematics. It is whether the reasoning is documented well enough for somebody else to follow.
Auditor challenging your ECL model?
Send us the model and the audit query. We will tell you whether the challenge is about methodology, documentation or inputs — they need different answers, and the wrong one costs a reporting cycle.
Our auditor says our provision is not IFRS 9 compliant.
A percentage applied to ageing brackets is not an expected credit loss model, however long it has been accepted.
We built a model in Excel and cannot explain it any more.
Models that only their author understands fail the first time the author is unavailable.
We have interest-free loans to related parties on the balance sheet.
Below-market and interest-free lending requires fair value measurement at initial recognition, and the difference has to go somewhere.
We issued convertible instruments to an investor.
Convertibles usually contain separable components, and getting the split wrong distorts both equity and profit.
We hedge foreign currency but do not apply hedge accounting.
Economic hedging without hedge accounting produces volatility in profit that the hedge was meant to remove.
Our Gulf subsidiary needs an ECL model for local reporting.
Receivables portfolios in Saudi Arabia and the UAE, modelled to IFRS as endorsed in the relevant jurisdiction.
We granted share options and do not know how to value them.
Option valuation requires a defensible model and documented assumptions, not a nominal figure.
A buyer's advisers are questioning our provisioning.
Under-provisioning found in diligence is treated as an earnings adjustment, and it is priced.
IIWhat we do
Scope of the financial instruments practice
Technical accounting work delivered with the auditor's questions already in view.
01
Classification and measurement
Establishing the right measurement category before anything else follows from it.
- Business model assessment
- SPPI testing — solely payments of principal and interest
- Amortised cost, FVOCI and FVTPL determination
- Reclassification assessment
- Embedded derivative identification
- Accounting policy documentation
02
Expected credit loss modelling
The general approach, for portfolios where a provision matrix is not sufficient.
- PD, LGD and EAD estimation
- Three-stage staging framework design
- Significant increase in credit risk criteria
- Forward-looking macroeconomic overlays
- Multiple scenario probability weighting
- Model documentation and governance
- Back-testing and periodic recalibration
03
Simplified approach and provision matrices
For trade receivables, contract assets and lease receivables — where most corporates actually sit.
- Provision matrix construction from loss history
- Segmentation by customer, product or geography
- Forward-looking adjustment of historical rates
- Credit-impaired receivable identification
- Write-off policy design
- Sensitivity analysis
04
Fair valuation under IFRS 13
Valuation with the hierarchy level established and the inputs sourced and referenced.
- Level 1, 2 and 3 hierarchy determination
- Valuation technique selection and rationale
- Observable and unobservable input sourcing
- Unquoted equity investment valuation
- Investment property and non-financial asset support
- Valuation memoranda for audit
05
Derivatives and structured instruments
Where the instrument does not fit a standard template.
- Forward, swap and option valuation
- Convertible instrument component separation
- Compound instrument liability and equity split
- Embedded derivative separation
- Preference share classification
- Puttable instrument assessment
06
Amortised cost and modifications
The mechanics that quietly misstate profit when they are done by convention rather than by standard.
- Effective interest rate computation
- Transaction cost treatment
- Concessional and interest-free loan fair valuation
- Related-party and shareholder loan measurement
- Substantial versus non-substantial modification assessment
- Derecognition and modification gain or loss
07
Hedge accounting
Documentation prepared at inception, because it cannot be applied retrospectively.
- Hedge designation and documentation
- Economic relationship demonstration
- Hedge ratio determination
- Effectiveness assessment and rebalancing
- Fair value, cash flow and net investment hedges
- Discontinuation accounting
08
Disclosure and independent review
Including review of models prepared by someone else.
- IFRS 7 disclosure preparation
- IFRS 13 fair value disclosure
- Credit risk and sensitivity disclosure
- Independent review of management's model
- Remediation of audit findings on provisioning
- Share option and employee benefit valuation
IIIWhere models get challenged
Eight things auditors query, in order of frequency
Almost none of these are arithmetic. Nearly all are judgements that were made but never written down.
| The query | What sits behind it |
|---|---|
| The provision matrix is just last year's percentages | Historical loss rates are the starting point, not the answer. IFRS 9 requires them to be adjusted for current conditions and forward-looking information. A matrix carried forward unchanged for three years is the single most common finding we see. |
| Segmentation is not explained | Receivables have been grouped — by customer type, product or geography — with no documented basis for why those groupings share credit risk characteristics. |
| Forward-looking information is asserted, not evidenced | The model says macroeconomic factors were considered. Nothing shows which factors, what source, what weighting, or what the effect was. |
| Staging criteria are undefined | Under the general approach, the trigger for a significant increase in credit risk must be defined in advance. Deciding case by case at reporting date is not a staging framework. |
| Related-party balances are excluded without reason | Intercompany and shareholder receivables are frequently assumed to carry no credit risk. That assumption requires support, particularly where the counterparty is loss-making. |
| The fair value hierarchy level is not stated | Level 3 measurement carries specific disclosure obligations. Valuations presented without an explicit level determination invite the auditor to assume the most onerous. |
| Unobservable inputs have no source | Discount rates, growth assumptions and credit spreads appear in the model with no reference to where they came from or why they are appropriate. |
| Nobody can rebuild the model | The spreadsheet works, and the person who built it has left, and the logic is undocumented. At that point the number cannot be supported regardless of whether it is right. |
IVHow we work on models
Six things that make a model survive audit
01
The memorandum is written alongside the model, not after it
Methodology, judgements, inputs and their sources documented as the model is built. Documentation assembled afterwards reads as justification rather than reasoning, and auditors can tell the difference.
02
Every input is sourced and referenced
Loss history, macroeconomic data, discount rates and credit spreads traced to a source. Unobservable inputs identified as unobservable rather than presented as facts.
03
Built so somebody else can rebuild it
Logic separated from data, assumptions in one place, no hard-coded values buried in formulas. The test is whether a reviewer with no prior involvement can follow it — because that is exactly who will.
04
Independently reviewed inside the firm before release
A second qualified professional reviews the model logic and the judgements before it leaves. Modelling errors are systematic rather than random, which is what makes single-reviewer work dangerous here.
05
Sensitivity is run and disclosed
Which assumptions actually move the answer, and by how much. This is a disclosure requirement, and it is also the fastest way to find out whether the model is stable.
06
Proportionate to the entity
A trading company with two hundred customers does not need a bank's PD and LGD framework. Over-engineering a model is not conservatism — it is a maintenance burden that produces its own errors.
These are the firm's controls applied to modelling work specifically. The full set, applied to every engagement of any type, is published.
VWho needs this
Which approach applies to you
IFRS 9 gives two routes to expected credit loss. Most corporates need the second one, and many are applying neither properly.
Route oneGeneral approachA three-stage model with twelve-month expected credit loss on performing exposures and lifetime loss once credit risk has increased significantly. Required for loans, debt investments and most financing receivables.
Route twoSimplified approachLifetime expected credit loss from initial recognition, with no staging assessment. Available for trade receivables, contract assets and lease receivables — which is where most corporates sit.
NeitherWhat many still doA fixed percentage applied to ageing brackets, unchanged for years, with no forward-looking adjustment. It resembles the old incurred-loss model and is not compliant.
01
Corporates with receivables
Trading, manufacturing, services and construction businesses applying the simplified approach — the largest group by number, and the one most often carrying a non-compliant matrix.
02
Groups with intercompany lending
Shareholder loans, intercompany balances and concessional lending, where initial fair value measurement and credit risk assessment are frequently ignored entirely.
03
Entities with financing portfolios
Lending and financing operations requiring the general approach, staging criteria and a documented PD, LGD and EAD framework.
04
Businesses with structured instruments
Convertibles, preference shares, options and derivatives, where classification between liability and equity determines both the balance sheet and reported profit.
05
Gulf entities
Saudi and UAE reporting entities, including IFRS as endorsed in the Kingdom of Saudi Arabia, delivered from Islamabad alongside their Pakistan reporting.
06
Groups reporting internationally
Where a parent requires IFRS 9 numbers, or the equivalent under FRS 102 or ASC 326, prepared to the group's timetable and format.
The independence boundary, stated plainly
Valuation and expected credit loss modelling involve significant judgement. Where we act as your statutory auditor, we do not perform this work in a way that would create a self-review threat under the ICAP Code of Ethics — for public interest entities, material subjective valuations are provided only to non-audit clients.
We will tell you which side of that line your engagement falls on before we quote, rather than after you have engaged us. Where we cannot act, we will say so and explain why.
VICommon questions
Questions we are asked
Is a percentage applied to ageing brackets an acceptable ECL model?
Generally no, not on its own. Historical loss experience is the starting point, but IFRS 9 requires it to be adjusted for current conditions and reasonable, supportable forward-looking information. A matrix carried forward unchanged for several years resembles the old incurred-loss approach and is the most common non-compliance we are asked to remediate.
What is the difference between the general and simplified approaches?
The general approach uses a three-stage model — twelve-month expected credit loss while the exposure is performing, moving to lifetime loss once credit risk has increased significantly. The simplified approach recognises lifetime expected credit loss from the outset with no staging assessment, and is available for trade receivables, contract assets and lease receivables. Most corporates qualify for and should be using the simplified approach.
Do we need to provide against intercompany receivables?
Intercompany and shareholder balances are within the scope of IFRS 9 and are not automatically risk-free. The assumption that they are needs support — particularly where the counterparty is loss-making, where repayment is not demanded, or where the balance has been outstanding for years. This is an area auditors increasingly challenge.
How should an interest-free loan to a related party be measured?
Financial instruments are measured at fair value on initial recognition, which for a below-market or interest-free loan is not the amount advanced. The difference has to be accounted for according to its substance — frequently as a capital contribution or distribution — and the loan is subsequently measured at amortised cost using the effective interest rate. Getting this wrong misstates both equity and finance income.
Can we apply hedge accounting retrospectively?
No. Hedge accounting requires formal designation and documentation at inception of the hedging relationship. If the documentation was not prepared then, hedge accounting cannot be applied to that period — which is why the documentation matters more than the hedge itself from an accounting standpoint.
Our model was built by someone who has left. What do we do?
This is common and it is recoverable. We rebuild the model with documented logic and sourced inputs, reconciling to the prior position so the movement is explainable. Where the previous model was wrong, we quantify the effect and advise on whether it is a correction of error or a change in estimate — which have different accounting and disclosure consequences.
Can you review a model we prepared ourselves?
Yes, and independent review is often the more efficient engagement. We assess methodology, documentation and inputs against the standard, identify what an auditor is likely to challenge, and set out what needs strengthening — without rebuilding work that is already sound.
How is this work priced?
Model construction and valuation engagements are fixed-fee against a defined scope, established after an initial review of the portfolio and the current position. Independent review of an existing model is normally a smaller fixed fee. Recurring recalibration and annual update work can be scoped as an annual engagement.
Related
Where this work usually connects
Let's work together
Send us the model, or the audit query.
A partner will tell you whether the issue is methodology, documentation or inputs — and what it takes to close it before the next reporting date.