The concern is not the profit figure itself, but how some of the underlying judgements were made.
Picture: Devina Haripersad/The Citizen News
The most reliable warning sign in a company’s accounts is seldom a wrong number. It is a judgement that cannot be tested from the outside – where the statements disclose what management concluded, but not the contemporaneous evidence that would show whether that conclusion was reasonable when reached.
That is a failure of disclosure, not arithmetic, and in a quarter-century of forensic work, it is the point at which fraud, if it exists, stops being independently checkable.
Eskom’s decade is a case of exactly that.
Eskom has reported a R30.3 billion profit, and called it operational recovery, energy security and financial sustainability.
The figure is reported in audited financial statements, although the 2026 audit opinion was qualified over irregular-expenditure disclosure.
I do not dispute the R30.3 billion figure. But a single year is the wrong lens.
Across the decade, the material discretionary choices examined here repeatedly fall in the direction that flatters the result.
That does not make the reported position false. It means the boundary between what the accounts assert and what an outsider can independently establish has widened.
Accounting standards leave room for management judgement in recurring places: what to recognise as an asset, what to impair, what to value by estimate and what future assumptions to use.
Judgement becomes significant when it repeatedly favours the reported result and later evidence – a derecognition, restatement, collectability write-down, counterparty figure or audit finding – tests or contradicts it.
None of what follows alleges fraud.
Each accounting treatment may have a defensible explanation. The forensic question is whether the contemporaneous evidence supporting those judgements can be reconstructed.
The tax asset that disappeared two years before the profit
A deferred tax asset represents future tax benefits that can be used against taxable profits. It is recognised only when sufficient future taxable profit is expected to be available.
In 2024, Eskom derecognised R36.6 billion in deferred tax assets after concluding that it would not generate sufficient taxable income to use R135.7 billion in accumulated tax losses.
The adjustment contributed R29.5 billion to the tax charge, turning a R25.5 billion loss before tax into a R55 billion loss after tax.
Eskom attributed the adjustment to the separation of the transmission business into the National Transmission Company of South Africa (NTCSA). After the transfer, it no longer expected sufficient taxable profit against which the losses could be used.
The issue is not that the write-off was impermissible, but whether the forecasts supporting recognition before the write-off were reasonable and adequately supported at the time.
The timing matters: a R36.6 billion tax asset disappeared two years before Eskom declared a R30.3 billion profit and financial sustainability.
The bad-debt charge that shrank as the profit appeared
The same question arises with expected credit losses.
In 2025, Eskom recorded a R7.3 billion impairment charge on financial assets. In 2026, that fell to R281 million, even as gross municipal arrears rose by R17 billion to R111.6 billion.
That does not mean the accounting is wrong.
Expected credit loss rests on probability of default, loss given default, recovery and forward-looking assumptions – precisely the judgements that require testing.
The forensic question is therefore not “arrears rose, so the provision should have risen”.
It is: What changed in Eskom’s recovery, default and loss assumptions sufficiently to move the impairment charge from R7.3 billion to R281 million while municipal arrears reached R111.6 billion?
Valuations, finance costs and figures revisited
Roughly two-thirds of the R17.5 billion improvement in profit before tax came from below-operating movements.
A R9.3 billion favourable swing occurred in the net fair-value, foreign-exchange and deferred-income line, including a much smaller fair-value loss than in the prior year.
Net finance costs fell about R2.2 billion, helped by government debt-relief support and lower debt.
Neither of those is equivalent to selling more electricity.
The accounts also show earlier positions being revisited: a 2022 correction to the hedges’ “valuation curve methodology”, prior-period errors identified through audit, and a restated 2021 loss.
The FY2026 audit records that the statements initially submitted for audit contained material misstatements and omissions in areas including fair values, impairment, going concern and capital management, which were subsequently corrected.
These are not, individually, evidence of wrongdoing. They are evidence of where the underlying accounting record needs to be reconstructed.
Where the physical evidence enters
The diesel matter is different in kind because the accounting record can be compared with physical evidence.
Eskom has characterised R41 billion in diesel payments as out-of-contract and irregular – R3 billion relating to current contracts and R38 billion to previous contracts – and says the matter was reported to its auditors and National Treasury.
In one case, Eskom’s SAP records showed 40 million litres received, while the terminal operator reported only about five million litres physically held in its tanks. The balance was described as available through the supply arrangement.
Eskom has disputed the inference that this meant the fuel did not exist, explaining that “available” could include fuel stored elsewhere and available on request.
That distinction matters. The evidence does not establish a loss, still less that the R30.3 billion profit is misstated.
It establishes something narrower and important: the accounting record and the physical evidence require reconciliation. Only the underlying payment, delivery and inventory trail can establish what actually happened.
Why the direction is worth noticing
A reported recovery brings reputational, institutional and political benefits and can affect executive performance assessment.
Eskom’s turnaround is a national programme, while government has provided R64 billion in debt-relief support in FY2025 and R80 billion in FY2026, subject to statutory conditions and the broader recovery programme.
None of that establishes motive or intent.
It establishes incentive – and incentive is why repeated favourable judgements deserve examination rather than assumption.
What the public record can – and cannot – show
This is a pattern described, not fraud alleged or found.
The issue is not that auditors did nothing. They did.
The 2026 auditors tested management’s going-concern assessment, forecasts, assumptions and sensitivities, and concluded that management’s assessment was reasonable. They nevertheless reported a material uncertainty related to going concern and qualified their opinion over irregular expenditure.
The narrower point is that the published record does not disclose enough of the underlying contemporaneous evidence for an outsider to reproduce that testing independently.
Nor has this article scored every material estimate Eskom made over 13 years.
The claim is narrower: the material judgements examined here repeatedly fall in the direction that flatters the reported result, while later evidence creates reasons to revisit them.
That is a forensic risk marker – a signal of where to look, not a finding that anything is wrong.
The records required to test it are identifiable: the profit forecasts supporting each tax-asset recognition and what happened against those forecasts; impairment and valuation working papers; the diesel approval, payment, delivery and SAP trails; and reconciliation of the payments Eskom calls irregular to the amounts recognised in its accounts.
This calls for a targeted independent examination and oversight probe, not another annual audit opinion.
The Auditor-General and the relevant parliamentary portfolio committee are the appropriate public oversight mechanisms to require that examination, obtain the underlying records and test management’s explanations against the evidence.
What turns on it
If such an examination found material judgements that were unsupported when made, the consequences could include restatement, implications for reported profit and fair-value or impairment balances, and consequences for covenants, credit ratings or debt-relief arrangements.
Findings of unsupported or deliberately misleading financial reporting could also raise questions of accountability for directors, executives and others involved, including the recovery of incentives and referral to the appropriate investigative or prosecutorial authorities.
And if the recovery ultimately proved less substantive than reported, the public case for tens of billions in support would also have to be reconsidered against the underlying evidence.
None of that is asserted here. The proposition is simpler: the public record establishes Eskom’s reported result, but does not yet give an outsider enough contemporaneous evidence to reproduce the judgements that produced it.
That is not an allegation of fraud. It is a reason to look.
In plain terms
Eskom has announced a large profit and told the country it has turned the corner. As a reported result, that profit is real.
But accounts involve judgement – what to count as an asset, when to write it off, what it is worth and what future assumptions to use.
A worsening business can honestly produce losses and writedowns, so direction alone proves nothing.
What deserves attention is that, across more than a decade of Eskom’s audited figures, the material judgements examined here repeatedly fall the way that flatters the result while later evidence pulls the other way.
A set of scales that repeatedly tips in one direction – across changing management, standards and grid conditions – deserves a second look.
A recovery being built is one thing. A recovery merely booked is another.
The records Eskom holds can tell them apart.
* The financial figures are drawn from Eskom’s audited Integrated Reports and Annual Financial Statements across 2013/14 to 2025/26 – including the 2020, 2022 and 2026 Annual Financial Statements – together with its public results releases for 2023 and 2024 and its diesel-investigation statement of 27 August 2026; diesel-volume detail is attributed to amaBhungane. Nothing in this article asserts that any accounting treatment described was impermissible under the applicable standards; the question is whether the underlying judgements and estimates were reasonable and adequately supported when made. Nothing here is an allegation or finding of fraud or wrongdoing against Eskom or any person.
Bart Henderson is a veteran fraud risk specialist and forensic investigator with nearly three decades of experience at the highest levels of financial crime detection, investigation, and litigation support across South Africa and beyond.
This article was republished from Moneyweb. Read the original here.
