In accounting, as in life, what undoes us is rarely the outright lie. It is the softly spoken optimism, the revenue not yet earned but already recognised, the cost that might reduce next quarter, the contract that should turn profitable if only we wait. These are not deceptions in the criminal sense, but rather seductive half-truths. They inhabit the footnotes of annual reports and the judgements column of audit working papers. And nowhere did such financial hopefulness do more damage, more quietly, than in the collapse of Carillion.

At its peak, Carillion was one of the UK’s largest construction and facilities management firms. It built hospitals, managed schools, cleaned rail stations. It was entrusted with public infrastructure and private contracts alike. And yet, behind its solid exterior and polite press releases, the company was slowly disintegrating, dragged down not by a single catastrophic event, but by a long series of unchallenged estimates.

When Carillion went into compulsory liquidation in January 2018, it left behind £7 billion in liabilities, thousands of lost jobs, and a stunned government scrambling to reassign public services. It was, in every sense, a slow-motion implosion.

 Construction Contracts and the Art of Estimation

The heart of the accounting issue was the way Carillion recognised revenue and profit on long-term contracts. Under the now-superseded IAS 11: Construction Contracts, companies were permitted, indeed, encouraged, to use the percentage-of-completion method. This meant that revenue could be recognised before the project was completed, based on estimates of how much work had been done, what the final costs would be, and what margins could be expected. In theory, this makes sense. In practice, it opens the door to a world of illusion.

Carillion had a portfolio of complex, long-term construction projects. Many were over-budget or under performing. But the company consistently overestimated the profitability of these contracts. It routinely assumed that overruns would be recovered, that variations would be approved, that margins would remain intact. Loss-making projects were not recognised (IAS 37) as such, because management believed (or needed to believe) that recovery was just around the corner.

The profits were not earned in the ordinary sense of the word. They were predicted, then booked. The result was a series of annual reports that portrayed stability and growth, when, in fact, the foundation was crumbling.

Image :Blower

KPMG’s Failure to Interrogate

For all the shortcomings of Carillion’s management, the failure was not theirs alone. The auditors, KPMG, signed off on Carillion’s financial statements for nearly two decades. And year after year, they accepted the accounting estimates presented to them without sufficient scepticism. A parliamentary inquiry, launched in the aftermath of the collapse, accused KPMG of being “complicit in the misrepresentation of Carillion’s financial reality.” More precisely, they failed to challenge management’s assumptions regarding: Cost-to-complete forecasts on major projects, Contract recoveries that were unlikely to materialise and impairment testing on goodwill and other intangible assets.

When contract costs ballooned and deadlines slipped, Carillion simply updated its estimates to maintain the illusion of profitability. KPMG, rather than testing these estimates with third-party evidence or independent benchmarks, largely relied on management’s own internal calculations.

At one point, Carillion’s management even reversed previously recognised losses, claiming they had “renegotiated terms” or “improved margins.” These reversals were based on forecasts that never came to fruition. And still, they passed audit scrutiny. The result was not so much an oversight as a systemic failure of professional doubt.

Why Auditing Estimates is So Difficult, And So Dangerous

Unlike cash balances or inventory counts, accounting estimates are not observed; they are constructed. They involve assumptions about the future, interest rates, project timelines, customer behaviour, litigation outcomes. They depend on models, inputs, and professional judgement. And crucially, they are susceptible to confirmation bias and managerial pressure. An auditor, therefore, must tread a delicate path. To audit an estimate is not simply to verify a number, it is to challenge a belief. One must step into the company’s assumptions and ask whether the optimism is warranted, whether the future being imagined is even plausible.

This is technically difficult, but it is also psychologically difficult. Because estimates are often defended with confidence, they are dressed in spreadsheets and buttressed with boardroom rationalisations. And to question them, properly and firmly, is to disrupt the social harmony of the audit-client relationship. It requires intellectual courage. The Carillion story is not merely one of accounting failure, but of unchecked optimism mistaken for truth.

The very mechanisms designed to reveal reality, estimates, audits, standards, became tools to conceal it. In place of professional scepticism, there was deference; in place of challenge, complicity. And so the deeper lesson is not technical but human: our greatest errors often arise not from deceit, but from the quiet comfort of believing what we want to be true. In Carillion’s case, that shield held for far too long. Until, as with all things built on soft foundations, it gave way. And perhaps, in that sense, the lesson is not just technical. It is human. We all live among estimates, about our careers, our relationships, our futures. The danger lies not in making them, but in refusing to test them.