Capitalize or Expense? An Auditor’s Decision Framework

The real question: what did the spending achieve?

Not what the invoice says — what the spending achieved. Does it keep something running as before, or does it create new future benefit? Replacing worn belts on an existing machine so it runs as before is an expense: it maintains the existing level of performance. Installing a new packaging line is capital: it creates a new asset that will serve the company for years. Upgrading a machine to double its capacity is also capital — future benefits have increased.

A three-question framework

1. 1.Does it create or extend the asset’s future benefits? If yes, consider capitalising. If no, expense it.
2. 2.Can the cost be measured reliably? If yes, capitalise.
3. 3.Is it maintenance that restores the original performance? If yes, expense — it keeps the machine as it was, it doesn’t make it better.

This won’t settle every case, but it turns an argument into a discussion of specific facts: did capacity increase? Did the life extend? Did output improve?

The error runs in both directions

Expensing the ₹38 lakh packaging line as repairs understates profit. Separately, treating a ₹22.5 lakh advertising campaign as work-in-progress overstates both assets and profit — advertising never creates a recognisable asset, however large the campaign, and should be expensed when incurred. The same test — what did the spending achieve — catches errors in both directions, and the larger misstatement here is the understatement: an auditor instinctively looks for overstatement first, but understatement is a misstatement too, and it can hide profit for other purposes.

When management pushes back — “the campaign will drive sales for years, so it’s an asset” — the answer is a question, not a refusal: expected future sales aren’t enough to recognise an asset. Can you point to something the company owns or controls? Advertising leaves no such thing, however much you hope it works.

Depreciation: vouch the date, not just the amount

A worked recomputation: a machine costs ₹1.2 crore, installed 1 October, 10-year life, straight-line. The annual charge is ₹12 lakh. In use for 6 months of the year, the charge is ₹12 lakh × 6⁄12 = ₹6 lakh. If the client charged a full year, it’s overstated by ₹6 lakh. If the client charged nothing yet, it’s understated by ₹6 lakh. Either error is found by the same recomputation — and the answer depends entirely on the installation date, which is exactly why the date gets vouched, not just the invoice amount.

Testing existence, not just additions

The fixed asset register agreeing to the ledger proves nothing about whether the assets actually exist — two client records agreeing with each other is the same trap as tying a trial balance to itself. Select items from the register and go find them on the floor; select items on the floor and find them on the register, to test completeness in both directions.

For disposals: a machine sold above book value records a gain. A machine scrapped but still on the register at full cost needs writing off, or the asset base stays overstated. A vehicle sold with proceeds banked but still on the register needs removing, with the bank receipt as the clue the sale happened. An idle but owned machine stays on the register, but its value should be reconsidered — idle time is a sign it may be impaired.

Watch estimate changes that move profit the same direction

Useful life, depreciation method, and residual value are all estimates management controls. Lengthening a useful life reduces the depreciation charge and raises profit — easy to do, and worth asking when the life was last reviewed and what supports the change. A change in estimate isn’t wrong in itself, but when several estimate changes move profit in the same direction in the same year, that pattern is exactly what bias testing looks for.

Common mistakes

• —Trusting the description on an invoice instead of reading the contract and asking what was achieved.
• —Depreciating from the invoice date instead of the date the asset was actually ready for use.
• —Testing only additions, and never checking disposals or existence.
• —Accepting a changed useful life without asking what evidence supports it.