Eight years after IFRS 9 came into effect, many banks still appear to treat expected credit loss modelling mainly as a compliance requirement. IFRS 9, or International Financial Reporting Standard 9, is the accounting standard that governs how financial instruments such as loans and investments are classified, measured and reported. Its expected credit loss requirements ask banks to recognise potential credit losses before a borrower actually defaults. Yet the framework has much wider potential.

For operational researchers working in financial services, IFRS 9 sits at the intersection of forecasting, risk modelling, scenario analysis and portfolio decision-making. The same models used to calculate provisions can also help banks identify deteriorating loans earlier, price credit more accurately and make more informed decisions about where capital should be deployed.

IFRS 9 replaced the previous incurred-loss approach with a forward-looking model based on expected losses. Rather than waiting for evidence that a loss has already occurred, lenders must assess how credit risk may develop over time.

Loans are placed into stages according to changes in their credit quality. Performing exposures begin in Stage 1, while loans showing a significant increase in credit risk move into Stage 2. Credit-impaired assets are placed in Stage 3. Each transition affects how expected losses are calculated and how much provision the lender must hold.

This structure creates a modelling problem, but it also creates a valuable source of management information.

Turning loan monitoring into early intervention

One of the clearest benefits of IFRS 9 is the discipline it brings to loan monitoring.

Banks must continually assess whether borrowers remain financially stable or show signs of deterioration. This may involve tracking payment behaviour, changes in credit scores, sector conditions, collateral values and wider economic indicators.

Operational research techniques can help institutions identify patterns within that data and estimate the likelihood that an exposure will move between stages. Transition models, classification methods and early-warning systems can all support decisions about when intervention is needed.

When these signals are incorporated into management dashboards, lenders can act sooner. A borrower showing early signs of difficulty may be offered a restructuring or revised repayment arrangement before the account moves into default. Collections activity can also be prioritised according to expected recovery, exposure size and the probability of successful intervention.

Recovery data provides another important source of insight. It can reveal whether collateral is properly documented, how long assets take to realise and which recovery routes produce the strongest outcomes. That information can then be fed back into future loss estimates and lending policies.

Using scenarios to understand uncertainty

Expected credit loss calculations depend not only on historical performance but also on assumptions about the future.

Banks may consider economic variables such as growth, inflation, unemployment, interest rates and exchange rates. Rather than relying on a single forecast, they are expected to assess several plausible scenarios and weight them according to their likelihood.

This makes scenario design particularly important.

Too few scenarios may fail to capture meaningful risks, while too many can make models difficult to interpret and maintain. The assumptions used must also remain internally consistent. A scenario combining falling interest rates, rising inflation and strong growth, for example, needs a credible economic explanation.

Operational researchers can contribute by designing scenarios, testing their sensitivity and examining how changes in probability weights affect expected losses. Simulation can also be used to explore a wider range of outcomes than traditional base, upside and downside cases.

The results can support more than financial reporting. They can feed into stress testing, capital planning and portfolio reviews, helping banks understand where risk is concentrated and how their balance sheets may respond under adverse conditions.

Expected credit loss models therefore become part of a broader decision-support system. They help institutions consider where lending should expand, where exposure should be reduced and where additional capital may need to be held.

Incorporating climate-related risk

Climate risk adds another layer of uncertainty.

Physical risks, such as flooding, extreme heat or storm damage, may reduce the value of collateral or disrupt a borrower’s operations. Transition risks may arise as industries adjust to new regulation, changing technologies and shifts in consumer demand.

The available data remains incomplete, and many climate scenarios operate over longer time horizons than conventional credit models. Even so, banks can begin incorporating climate-related assumptions into portfolio analysis and expected loss calculations.

Scenario models can help lenders assess which sectors, regions or borrowers are most exposed. They may also support decisions about lending limits, collateral requirements and the financing of lower-carbon activities.

This does not mean that climate risk can be reduced to one precise number. The more useful aim is to understand possible pathways, identify vulnerabilities and test how robust current lending decisions remain under different conditions.

That is a familiar operational research challenge: supporting decisions where the evidence is incomplete but the consequences of ignoring uncertainty may be significant.

Treating expected loss as a pricing input

Pricing is another area where IFRS 9 models can provide practical value.

Banks already consider funding costs, operating costs and required returns when setting loan prices. Expected credit loss should form part of that calculation because it represents a real cost associated with taking risk.

IFRS 9 encourages lenders to divide portfolios into groups with similar characteristics. These might include borrower type, industry, geography, loan size or collateral structure. Historical default rates, recovery patterns and forward-looking assumptions can then be estimated for each segment.

This allows pricing to reflect differences in risk more accurately.

A lower-risk borrower may justify a more competitive rate, while a higher-risk exposure may require a larger margin or additional security. The challenge is to balance competitiveness with profitability and financial resilience.

Operational research models can help evaluate these trade-offs. Pricing decisions can be examined alongside capital constraints, expected returns, default risk and strategic objectives, rather than being made through separate processes.

This creates a stronger link between front-office lending decisions and the bank’s wider risk appetite.

Supporting portfolio and capital decisions

At portfolio level, IFRS 9 data can reveal how risk is distributed across products, sectors and customer groups.

Banks can use this information to test alternative lending strategies. They might compare the effect of expanding into one market, tightening criteria in another or changing the mix between secured and unsecured lending.

These decisions involve competing objectives. A bank may want to increase revenue while limiting expected losses, maintaining capital ratios and avoiding excessive concentration in a particular sector.

Portfolio optimisation can help decision-makers examine those trade-offs systematically. Rather than considering expected loss as a fixed accounting output, institutions can use it as an input to choices about growth, diversification and capital allocation.

This is where the strategic value of IFRS 9 becomes most visible. Provisioning models can help answer not only how much risk the bank currently holds, but also whether that risk is aligned with its objectives.

Governance still matters

None of this value can be realised without strong governance.

Expected credit loss models depend on judgement as well as data. Decisions must be made about segmentation, scenario weights, default definitions, model adjustments and the point at which an increase in credit risk becomes significant.

Boards and senior managers need to understand how these choices affect reported provisions and business decisions. Clear ownership is essential, alongside independent validation, regular performance monitoring and thorough documentation.

Data quality is equally important. Automated systems linked directly to source data can improve consistency and reduce manual errors, but technology alone will not solve weak processes. Institutions also need staff who understand the models, can challenge their assumptions and explain their limitations.

Progress remains uneven. Some lenders have moved from spreadsheets and fragmented processes to integrated modelling systems, while others continue to depend heavily on external advisers. A proportionate approach may be necessary, particularly for smaller institutions, but the underlying objective should remain the same: models that are reliable enough to support both reporting and management decisions.

From accounting model to strategic tool

IFRS 9 was introduced to make credit loss recognition more forward-looking. Its wider potential lies in how that forward-looking information is used.

For operational researchers, the framework offers a practical application of forecasting, optimisation, simulation and decision analysis. It involves modelling borrower behaviour, testing economic scenarios and balancing profitability against risk and resilience.

Banks that treat expected credit loss modelling solely as an exercise for auditors may meet the technical requirements of the standard, but they miss much of its value.

Those that embed the models into loan monitoring, pricing, capital planning and portfolio strategy can turn a regulatory obligation into a stronger decision-making capability.

References

https://ifrs9.com/about-ifrs-9

https://www.elibrary.imf.org/view/journals/005/2026/004/article-A001-en.xml