Standards in practice: GRAP 104
Why the GRAP 104 simplified approach gives a truer picture of consumer debtors than the old provision methods
A flat percentage on old debt told council very little. A provision matrix built on the municipality's own collection history says what the debtors book is actually worth, and why.
For years the impairment of consumer debtors in most municipalities came down to a policy line: provide a fixed percentage on everything older than 90 or 120 days, or provide in full on anything handed over. It was quick, it was easy to explain, and it produced a number every June. What it did not do was tell anyone what the debtors book was actually worth. The revised GRAP 104, which applies to the 2025/26 financial statements onward, replaces that habit with an expected credit loss approach, and for consumer debtors the practical route is the simplified approach using a provision matrix. This article explains why that is a better answer, not just a different one.
What the old methods were really doing
The previous version of GRAP 104 was built on an incurred loss model. An impairment was recognised only once there was objective evidence that a loss had already happened, such as a debtor in arrears beyond a set point or a hand-over for collection. Until that trigger, the receivable sat at full value even when everyone in the finance office knew that a large part of it would never be collected.
In practice, municipalities layered a shortcut on top of that model. The most common one was a flat percentage by age bucket, set in the accounting policy and rarely revisited. Another was to provide 100 percent on balances older than a fixed number of days and nothing on the rest. Both are easy to apply. Both also share the same weaknesses.
They ignore who owes the money. A government department 150 days in arrears and a household 150 days in arrears got the same provision, although one is almost certain to pay and the other is not. They ignore the municipality's own collection record, so a town collecting 95 percent of its billing and a town collecting 60 percent could carry the same provision rate if they copied the same policy. They lag reality, because a loss was only recognised once the account had aged into the trigger bucket, which meant the year in which collection actually deteriorated showed a healthier book than it should have. And they invited year-end adjustment, because a single percentage is easy to move when the surplus needs help.
What the simplified approach changes
Under the revised standard, the question is no longer "has a loss happened yet" but "how much of this balance do we expect not to collect over its life". For receivables like consumer debtors, the standard has the loss allowance measured at lifetime expected credit losses from the start, without tracking changes in credit risk account by account. That is the simplified approach.
The provision matrix is the practical tool for it. Debtors are grouped by characteristics that actually drive collection: the type of debtor (households, businesses, organs of state, indigent accounts), the service billed, and sometimes the area. For each group, the municipality's own history shows what share of the balance in each age bucket was eventually written off or never collected. Those historical loss rates are then adjusted for what is reasonably expected to change, such as a new credit control drive, a change in the indigent register or a local employer closing. The result is a rate per cell of the matrix, applied to the balance in that cell at year end.
| Illustrative segment: household water and sanitation | Current | 31 to 90 days | 91 to 180 days | 181 to 365 days | Over 365 days |
|---|---|---|---|---|---|
| Historical loss rate | 4% | 18% | 41% | 67% | 92% |
| Forward-looking adjustment | +1% | +2% | +3% | +3% | 0% |
| Rate applied | 5% | 20% | 44% | 70% | 92% |
Swipe sideways to see the rest of the table.
The figures above are an example of the shape of a matrix, not a benchmark. Every municipality's rates come from its own collection history.
Why this is fairer presentation, not just more work
Fair presentation under GRAP means the statements faithfully represent the effects of transactions and events. Applied to consumer debtors, that means the receivable shown on the statement of financial position should be the amount the municipality can reasonably expect to recover, and the impairment charged to the statement of financial performance should reflect the losses that belong to the year. The simplified approach does this in four specific ways that the old methods could not.
The number is anchored to evidence
A provision matrix is built from the municipality's own billing, receipting and write-off data. The rate in every cell can be traced to a calculation, and the calculation can be traced to the ledger. A flat 60 percent on debt over 90 days came from a policy document, and the policy document came from somewhere nobody could remember. Auditors test evidence, and an allowance that rests on evidence is an allowance that survives the audit.
Losses are recognised when the risk arises, not when the account ages
Because every age bucket carries a rate, including current balances, a deterioration in collection shows up in the year it happens. Council and the audit committee see the trend in the year they can still do something about it. Under the incurred loss model the same deterioration only surfaced a year or more later, when the accounts had aged into the trigger bucket and the options had narrowed.
Different debtors are treated differently
Segmenting the book means the allowance on organs of state reflects how organs of state pay, the allowance on indigent accounts reflects the municipality's indigent policy and equitable share, and the allowance on business accounts reflects business collection. The total is no longer an average that is wrong for everyone. This matters for revenue management as much as for reporting, because the matrix shows exactly where the losses are concentrated.
The judgement is visible and disclosed
The revised standard requires the inputs, assumptions and techniques behind the allowance to be disclosed, together with a reconciliation of the movement in the loss allowance. A reader of the statements can see how the number was built and how it changed. That is the opposite of a single policy percentage, where the only disclosure was the percentage itself.
The objections, and what they are really saying
The first objection is that the matrix will increase the impairment. Sometimes it does, and when it does, the old method was understating the loss. Sometimes it reduces the allowance, particularly on organs of state and on well-collected services, because the flat percentage was overstating it. Either way the movement is a correction towards the true position, and it is far better to make that correction through a documented change than to have the auditors make it for you.
The second objection is that the data is not good enough. That is usually true in the first year and it is a reason to start, not to wait. Building the matrix forces the age analysis, the receipting history and the write-off records into one reconciled view, and that view is worth having for credit control regardless of the accounting standard. Where history is thin, the standard accepts reasonable and supportable estimates, provided they are documented.
The third objection is that the old policy percentage was "accepted by the auditors before". It was accepted under a standard that no longer applies. From 2025/26 the auditors are testing an expected credit loss model, and a flat percentage without a data basis is the first thing they will challenge.
Two things the matrix does not replace
Property rates, traffic fines and other amounts arising from legislation are statutory receivables under GRAP 108 and need their own impairment assessment. The data and discipline of the matrix carry across, but the assessment has to be done and documented separately. And a good matrix does not replace credit control. It measures the loss; it does not reduce it. The municipalities that get the most from the approach are the ones that read the matrix as a management report and act on the segments where the rates are climbing.
Moving from a flat percentage to a provision matrix
- Reconcile the debtors age analysis to the general ledger at year end, by service and debtor type.
- Decide the segments on the basis of how the groups actually pay, and write down why.
- Calculate historical loss rates per segment and age bucket from the municipality's own receipting and write-off history.
- Document the forward-looking adjustments and who approved them.
- Prepare the movement reconciliation and the disclosure notes from the same model, so they agree to the rand.
- Have the model reviewed by someone who did not build it.
Where we help. Ambition Partners Advisory builds and reviews GRAP 104 and GRAP 108 impairment models for municipalities and municipal entities, and 360DebtorsPro produces the provision matrix, the movement reconciliation and the disclosure notes from the municipality's own age analysis, with the audit trail the AGSA asks for. Read about the service or message us on WhatsApp.
Discuss your municipality’s requirements
Ambition Partners Advisory is a Chartered Accountant (SA) led practice working with municipalities, municipal entities, government departments and public entities across South Africa. Tell us where you are in the reporting cycle and we will send you a scope, timeline and fee.