Provisions are balance-sheet liabilities created to recognise obligations that are probable but uncertain in amount or timing. Under Ind AS 37, a provision is recognised when: a present obligation exists (legal or constructive), an outflow of economic benefits is probable and a reliable estimate can be made.
Each criterion involves judgment. And the judgment, applied across the full population of potential obligations, can produce a provision balance that either reflects the company’s true exposure or manages it.
The Engineering Mechanisms
Timing manipulation
A company that knows in September that a litigation settlement is probable but waits until January to recognise the provision has shifted the cost from one fiscal year to the next. The obligation existed in both periods. The recognition occurred when management chose to recognise it.
The timing decision is not always manipulation. Ind AS 37 requires recognition when the criteria are met, not when they are convenient. But the criteria include probability assessment and reliable estimation, both of which are judgments that management can credibly argue are not met until a later date.
Under-provisioning (current-period inflation)
A company that provisions Rs 10 crore for a litigation exposure that a reasonable assessment would value at Rs 25 crore has understated its obligation by Rs 15 crore. The current period’s profit is inflated by the under-provision. The future period absorbs the excess when the obligation settles.
Over-provisioning (cookie-jar reserves)
A company that provisions Rs 30 crore for a restructuring when the actual cost will be Rs 18 crore has created a Rs 12 crore reserve. The reserve sits on the balance sheet, available for release in a future period when operating performance is weak. The release, presented as a “reversal of excess provision,” inflates the future period’s profit by Rs 12 crore.
The creation and release pattern, large provisions in good years, released in weak years, is the signature of cookie-jar accounting.
Best estimate vs worst case
Ind AS 37 requires the “best estimate” of the expenditure required to settle the obligation. The best estimate is the amount the entity would rationally pay to settle the obligation or transfer it to a third party.
Management can interpret “best estimate” as the most likely outcome (resulting in a lower provision) or as a probability-weighted average of all outcomes (which may result in a higher provision if the tail risks are significant). The interpretation determines the provision.
The Forensic Methodology
In Northrop Management Private Limited’s forensic and financial reporting work, provision analysis examines the full lifecycle of each material provision.
Step 1: Track provisions over time. For each category (litigation, warranty, restructuring, environmental, onerous contracts), plot the opening balance, additions, utilisations, reversals and closing balance for each of the last five years.
Step 2: Compare provisions to actual outcomes. For each provision that has settled (the uncertainty has resolved), compare the provision to the actual cost. Systematic under-provision (actual consistently exceeding provision) indicates aggressive accounting. Systematic over-provision (actual consistently below provision, with frequent reversals) indicates reserve building.
Step 3: Examine reversal timing. When are reversals booked? If reversals are concentrated in periods of weak operating performance, the pattern suggests that the provisions are being used to smooth earnings rather than to recognise uncertain obligations.
Step 4: Test the probability assessment. For each contingent liability disclosed but not recognised (classified as “possible” rather than “probable”), assess whether the probability classification is defensible. A litigation exposure that management classifies as “possible” but that the company’s own legal counsel has assessed as “likely to result in a payment” has been misclassified to avoid balance sheet recognition.
Step 5: Assess management incentives. Is management compensated on a metric affected by provisioning? Net profit, EPS, EBITDA after provisions: each creates an incentive to manage the timing and amount. The incentive does not prove manipulation. But it identifies the direction in which any judgment bias would operate.
Ashish Chaudhary, frames the analytical principle directly: “Provisions are not simply expenses. They are estimates of uncertain future outcomes, and uncertainty creates room for judgment. A provision recognised early is prudence. A provision deferred to protect current earnings is management. A provision created excessively and reversed to boost a future period is engineering. The pattern of creation, utilisation and reversal reveals which description applies.”
Questions for the Boardroom
- For each major provision category, how does the provision compare to actual cost over the last five years?
- Have any significant provisions been reversed in periods of weak operating performance?
- Do any contingent liabilities disclosed as “possible” have characteristics that suggest they should be classified as “probable”?
- What is the total provision balance as a percentage of net worth, and has that percentage been stable, increasing or declining?
- If an independent expert estimated each provision using the most current evidence, would they reach the same amount as management?
Closing Implication
Provisions are where accounting meets uncertainty, and uncertainty creates room for judgment. The judgment can reflect genuine estimation of future obligations, or it can reflect a management preference about when and how much of the obligation appears in the reported results.
The Provision Engineering Question examines the pattern: are provisions created, utilised and reversed in a manner consistent with the resolution of uncertainty, or in a manner consistent with the management of reported earnings? The answer is visible in the data. The question is whether anyone is asking it.