An Expected Credit Loss framework may have strong models, good data, robust controls and mature governance forums, yet still remain unstable if the policy foundation beneath it is weak. This is because policy is not just a formal document that sits behind the process. It is the institutional grammar of ECL. It defines what the organisation means when it says default, significant increase in credit risk, credit-impaired, overlay, stage override, reasonable and supportable information or material judgment.
This is why an ECL Policy Drafting Framework deserves a pillar article of its own.
Many institutions underestimate policy design. Some assume the policy can simply restate the accounting standard in internal language. Others treat it as a static compliance document produced once and updated only when auditors ask. Both approaches are weak. A strong ECL policy is not a paraphrase of the standard, nor a mere manual of current process steps. It is a controlled institutional framework that links accounting requirements, risk interpretation, model architecture, governance expectations, controls and reporting obligations into one coherent set of rules and principles.
The policy is where the institution decides what is in scope, how deterioration is interpreted, what evidence counts, which methods are allowed, who has authority, how judgment is controlled and how the reporting estimate is anchored to institutional rules. Without a strong policy, all these questions get answered elsewhere: in model files, committee memory, analyst practice or audit debate.
The accounting standard provides the external framework. The policy must go further. It must explain how the institution itself will apply that framework in its own context.
One of the first and most important sections of the policy should define scope — not only in accounting terms, but in terms of actual institutional portfolios and exposure classes.
A strong ECL policy should set out the high-level methodological architecture of the framework. It should identify which portfolios use stage-based approaches, which use simplified approaches, which are assessed collectively, which may be assessed individually and which measurement approaches are approved for which classes of assets.
Critical terms that should typically be defined clearly include default, cure, credit-impaired, significant increase in credit risk, origination date or initial recognition, lifetime expected credit loss, 12-month expected credit loss, write-off, recovery, restructuring or modification, overlay, post-model adjustment, materiality and override.
The policy should define what constitutes default, which quantitative and qualitative indicators matter, how portfolio-specific applications differ if relevant, how cure is identified and how default and cure are reflected across systems and processes.
A stronger policy should explain the framework for SICR determination, including the role of relative deterioration from origination, the role of backstop indicators such as delinquency, the place of internal ratings, behavioural signals or watchlist triggers and when qualitative factors may override quantitative indicators.
The policy should explain not only that the institution uses Stage 1, Stage 2 and Stage 3, but how it operationalises those stages, including the existence of stage overrides and the approval and review process for overrides.
Segmentation is so central to ECL that the policy should address it directly by defining the principle of segmentation, the kinds of characteristics considered relevant, the responsibility for segment design and review and the approval process for material segment changes.
A mature ECL policy must explain how forward-looking information enters the estimate, including the use of macroeconomic information, the principle of reasonable and supportable assumptions, the role of multiple scenarios where relevant, and the authority for scenario approval.
A stronger policy should explain when overlays may be used, what kind of issues justify them, what documentation is required, who can propose them, who reviews and approves them, how double counting is considered and when they should be revisited or released.
Where individual assessment is relevant, the policy should define the criteria for selecting exposures for individual review, the circumstances that trigger extraction from collective pools and the broad methodology used in individual assessment.
A strong ECL policy should explicitly address credit-driven modifications, their implications for staging, treatment of restructured currentness and requirements for post-modification monitoring.
One of the most valuable parts of a policy is the roles and responsibilities section, identifying who owns the policy, who owns model development or maintenance, who owns stage assignment processes, who reviews and approves overlays, who validates the framework, who approves final booking and which committees oversee what.
A strong drafting framework distinguishes between policy, which defines principles, rules, governance and approved institutional approaches, and procedure, which defines detailed operational steps, timelines, file sequences and practical workflows.
The policy should be consistent enough to establish one institutional ECL framework, but also allow portfolio-specific variation where risk characteristics genuinely differ.
A mature ECL policy should include review frequency, triggers for interim review, approval requirements for material policy changes and linkage with model and methodology change.
Many institutions benefit from a core policy document supported by annexes such as portfolio scope mapping, role matrix, glossary of defined terms, approved methodology by asset class and committee structure.
Recurring drafting mistakes include making the policy too generic, making it too operational, omitting difficult areas such as overlays, restructurings or overrides, failing to align definitions across policy, model and process documents, and not updating the policy as the framework evolves.
A policy that simply states that the institution applies expected credit loss in accordance with the relevant accounting standard, uses 12-month and lifetime expected loss where appropriate, considers macroeconomic information and applies judgment in determining credit deterioration may be technically correct, but operationally weak if it does not define default, SICR governance, overlay approval, committee structure or treatment of modified assets.
A strong ECL policy often benefits from a structure such as purpose and objective, scope, definitions, governing principles, portfolio methodology architecture, default and cure framework, SICR and stage framework, collective versus individual assessment, forward-looking information and scenarios, overlays and post-model adjustments, restructuring treatment, roles and responsibilities, governance and approval structure, review and change management, and appendices or annexes.
An ECL Policy Drafting Framework is ultimately about turning the impairment model into an institutional standard. A strong institution treats the ECL policy not as a document to be filed, but as a living backbone of the framework.
