Many of the largest numbers on a company’s balance sheet are not observed facts. They are estimates, produced through methodologies that require assumptions, populated with inputs that management controls, and presented with a precision that implies certainty where none exists.
Impairment values. Useful lives of depreciable assets. Expected credit loss provisions. Warranty obligations. Litigation provisions. Fair values of financial instruments. Deferred tax assets. Pension and gratuity obligations. Revenue from contracts with variable consideration.
Each of these involves a chain of assumptions: future cash flows, discount rates, probabilities, utilisation rates, growth rates, customer behaviour, legal outcomes and market conditions. Change the assumptions, and the number changes. In many cases, it changes materially.
The question for any board, auditor, investor or lender is straightforward: how much of this company’s reported net worth depends on assumptions that management controls?
The Methodology
For each significant accounting estimate, the forensic examination follows a structured chain:
Estimate → assumptions → management incentives → historical accuracy → external evidence → sensitivity → subsequent realisation
Assumptions
What inputs did management use, and how do they compare to independently observable data? A depreciation estimate based on a 15-year useful life for machinery in an industry where technological change makes equipment obsolete in eight years is an assumption that flatters the balance sheet. A provision for expected credit losses based on a 1% default rate when the historical default rate is 3% is an assumption that flatters the income statement.
Management incentives
Does management benefit from the estimate being higher or lower? An impairment estimate that avoids a write-down preserves reported profit and net worth. A provision that is lower than warranted inflates earnings. A useful life that is longer than economic reality reduces depreciation expense. The direction of the bias is predictable, because the incentives are consistent: management benefits from estimates that increase reported profit and net worth.
Historical accuracy
How accurate have management’s prior estimates been? A company whose provisions have been consistently inadequate (subsequent costs exceeding provisions by 30% to 50% year after year) is systematically underestimating. A company whose provisions have been consistently excessive and subsequently reversed may be building reserves that can be released to smooth future earnings.
Historical accuracy is the most objective test of management’s estimating quality. It requires comparing the estimate at the time it was made to the actual outcome when the uncertainty resolved. The track record reveals whether management’s judgment is conservative, neutral or optimistic.
External evidence
Is there independent evidence against which the estimate can be tested? Industry default rates for credit loss provisions. Market transactions for fair value estimates. Actuarial tables for pension obligations. Legal precedent for litigation provisions. External evidence provides a benchmark that is independent of management’s judgment.
Sensitivity
How much does the reported number change if the key assumption changes? An impairment test where a 5% reduction in the growth rate assumption triggers a Rs 200 crore write-down is a number that depends more on an opinion than on a fact. A provision where a 20% change in the probability assumption changes the liability by Rs 50 crore is a number whose precision is illusory.
Subsequent realisation
After the reporting period, what actually happened? A warranty provision of Rs 10 crore followed by actual warranty claims of Rs 25 crore reveals that the estimate was inadequate by a factor of 2.5. This does not necessarily indicate manipulation. But it indicates a bias in management’s estimating process that should inform how the board evaluates future estimates.
The Governance Implication
The board’s role is not to second-guess every accounting estimate. It is to understand which estimates are material, which assumptions drive them and whether those assumptions are reasonable in light of available evidence.
In Northrop Management Private Limited’s forensic and financial reporting work, the accounting estimate audit is a core component of every governance review, because the largest numbers on the balance sheet are frequently the ones most dependent on management judgment and least scrutinised by the board.
Ashish Chaudhary, frames the principle directly: “An accounting estimate with a precise number is still an opinion. The question is not whether the number is exact. It is whether the assumptions behind it are honest.”
Questions for the Boardroom
- Which five accounting estimates on our balance sheet have the largest impact on reported net worth, and what assumptions drive each of them?
- For each of those estimates, how would the reported number change if the key assumption moved by 10% in the unfavourable direction?
- What is our track record of estimating accuracy: have our prior-year estimates been consistently adequate, consistently inadequate or variable?
- Does management have a financial incentive (through compensation, covenants or market expectations) for any of these estimates to be higher or lower than an independent party would set them?
- If an independent expert prepared each of these estimates using externally observable data, would they reach the same conclusions as our management?
Closing Implication
The precision of a financial statement number is not the same as its reliability. A number can be calculated to the rupee and still be wrong by crores, because the calculation is only as reliable as the assumptions behind it.
Accounting estimates are the point where financial reporting meets management judgment. The board that understands which estimates matter, what assumptions drive them and whether those assumptions are reasonable is governing with full information. The board that accepts the numbers at face value is governing with the assumption that management’s opinions about the future are facts about the present.
They are not. And treating them as such is the most common governance failure that nobody calls a governance failure.
