ECL on contract receivables and retentions

Concentrated exposure to a handful of large government or mega-project clients, combined with slow certification cycles and long retention releases, means ECL must reflect the real economics of chronic delay and negotiated haircuts, not just a formal-default probability that is rarely triggered by these counterparties; contract assets and retentions carry the same lifetime ECL requirement as invoiced receivables. Under-providing here overstates net assets and hides a genuine, often material, concentration and liquidity risk that materializes as cash, not accounting, stress.

What percentage of receivables, contract assets and retentions is concentrated in a small number of government or mega-project clients? What is this client's historical certification-to-cash cycle length, and does the ECL model reflect the time-value cost of that delay even absent formal default? Are amounts under active commercial dispute assessed with a specific loss rate or haircut rather than treated as clean current balances? Is the same lifetime ECL rigor applied to contract assets and retentions as to invoiced trade receivables?

1) Apply the IFRS 9 simplified approach (mandatory for trade receivables and contract assets without a significant financing component, and available as an accounting policy choice otherwise): recognize a loss allowance for lifetime expected credit losses from initial recognition, with no need to track stage transitions. 2) Segment the receivables/contract-asset/retention book by client (or client tier) and by ageing, since concentration in a small number of government or mega-project clients means a portfolio-wide default-rate assumption will misstate risk, assess large individual exposures separately where warranted. 3) Build a provision matrix or client-specific model using the entity's own historical loss experience (actual write-offs, negotiated haircuts, and the economic cost of multi-year payment delays), adjusted for forward-looking macroeconomic factors relevant to public infrastructure spending and the specific client's fiscal position. 4) Treat prolonged non-certification, repeated payment deferrals, or amounts under active commercial dispute as credit-risk indicators requiring a higher loss rate or specific assessment, not as 'clean' current balances merely because formal default has not occurred. 5) Apply the same lifetime ECL requirement to contract assets (unbilled revenue) and to retentions, not just to invoiced trade receivables, all three exposures arise from the same underlying credit risk to the same client. 6) Reassess the model at least annually (more frequently if a client's payment behavior or dispute status changes materially) and recalibrate loss rates against actual subsequent collections.

ECL is measured as the probability-weighted present value of expected cash shortfalls (the difference between contractual/expected cash flows and the cash flows the entity actually expects to receive), discounted at the effective interest rate or an approximation thereof. For long-delayed collections (a defining feature of government/mega-project receivables), the expected shortfall should reflect both the time-value-of-money cost of delay (even absent a formal default) and a probability-weighted haircut for amounts under negotiation or dispute, modeling delay as a form of loss even when full nominal recovery is still expected. Retentions carry the same lifetime ECL requirement as other contract exposures, but the loss-given-delay assumptions should be calibrated specifically to retention release history (often materially slower than ordinary invoice payment terms), not borrowed wholesale from the general trade receivables provision matrix. Subsequent measurement: remeasure the loss allowance each reporting period through profit or loss (impairment gains/losses), reflecting updated ageing, changes in the client's fiscal/payment position, resolution or worsening of disputes, and actual amounts collected relative to the model's assumptions; use actual collection outcomes to recalibrate the model prospectively.

Disclose credit risk concentration by client/counterparty, quantifying the largest exposures as a percentage of total receivables, contract assets and retentions; disclose the ECL methodology (provision matrix, client-specific assessment, or a combination), the key inputs (historical loss rates, forward-looking adjustments) and any significant judgments; disclose an ageing analysis covering receivables, contract assets and retentions together, and separately quantify amounts embedded in active disputes; disclose the loss allowance reconciliation (opening balance, amounts recognized/reversed, write-offs, closing balance) and a sensitivity analysis where concentration risk is high; disclose the entity's assessment of sovereign/government counterparty risk and why a zero or near-zero loss rate is, or is not, appropriate given actual payment-cycle experience.

• Assuming zero or negligible ECL on government exposure purely because the counterparty is sovereign, ignoring extensive regional evidence of multi-year payment delays that carry a real economic cost
• modeling ECL only on formal default probability while ignoring the time-value-of-money cost of routine multi-year payment cycles that are not formal defaults but are genuine economic losses
• classifying disputed balances as current, performing receivables when they should carry an elevated, specifically-assessed loss rate or a dispute-related haircut
• applying a single portfolio-wide loss rate to a book that is in fact heavily concentrated in one or two mega-project clients, masking concentration risk
• failing to extend the same ECL rigor to contract assets and retentions that is applied to invoiced trade receivables, leaving a large part of the credit exposure effectively unprovided.