What bad records cost at year-end filing

Monthly filings can absorb a surprising amount of disorder. Figures get estimated, a credit gets deferred, something gets classified approximately, and the return goes in. Nothing visibly breaks.

Year end is different because it is where everything has to agree with everything else at once. The annual accounts have to tie to the monthly filings. The closing balances have to tie to the bank, to the stock count and to the receivable and payable ledgers. This year's opening balances have to match last year's closing ones. Each of those is a constraint, and disorder that survived twelve monthly filings individually cannot survive being checked against itself simultaneously.

This is why the year-end experience of a business with poor records is qualitatively different rather than just busier. The work is not recording, it is reconstruction: deciding what happened, from incomplete evidence, months later, under a deadline, usually by someone who was not there.

The costs that follow divide into three groups. Tax you did not have to pay. Money paid to professionals to rebuild what should have been recorded. And exposure, meaning the positions you are left defending if anyone asks. The first two are near-certain. The third is contingent, and it is the one people worry about, which gets the emphasis backwards.

The largest recurring loss is the simplest: amounts you were entitled to claim and did not, because nothing supported them.

An expense with no document is an expense a careful preparer excludes. Your taxable profit rises by the full amount and your tax by that amount at your applicable rate. Note who bears the risk in each direction, because it explains the outcome: the professional signing the return carries the risk of an unsupported claim, and you carry the cost of an omitted one. Given that asymmetry, omission is the rational choice for them and the expensive one for you.

Input tax credit fails in a harder way, because it depends on a cut-off. Entitlement to claim credit for a financial year is subject to a time limit tied to a date in the following year and the filing of the annual return, and the precise formulation has been amended, so it should be checked in the current provision rather than assumed. The consequence is structural: a mismatch discovered at year end may be discovered after the point at which it could be fixed. Credit lost this way is not deferred, it is gone, and the tax has then been paid twice, once to the supplier and once to the government.

Both losses share the property that makes them persist. Neither generates a notice. A return omitting a deduction is a valid return, so the feedback that would prompt better records never arrives, and the business concludes that its records are adequate because nothing complained.

When records cannot support a figure, someone estimates it, and estimates behave predictably: they are made cautiously by whoever bears the risk, which means against you. Closing stock estimated high raises profit. Expenses estimated low raise profit. Estimated figures also tend to be round, and a set of accounts containing round numbers where measured ones belong is a set of accounts that invites a question.

When a question comes, the named provisions matter more than the general fear. Under the Income-tax Act, section 270A deals with under-reported income and provides for a penalty of fifty per cent of the tax on the under-reported amount, rising to two hundred per cent where the case falls within the misreporting instances the section specifies. Under GST, section 122 of the CGST Act covers offences including incorrect invoicing and wrongly availed credit, with a penalty structure generally set at the higher of ten thousand rupees or the tax involved, per offence. Interest on any shortfall runs from the original due date rather than from the date the error was found, which is why a late-discovered problem costs more than the same problem found early even when the tax is identical.

The distinction between an error and misreporting is where records do their most valuable work. Books that show a genuine mistake, with contemporaneous documents and a traceable cause, support a very different characterisation than an absence of records. The difference between those two positions is a multiple of the tax, and it is decided by evidence created long before anyone asked.

All of these figures and their conditions have been amended and should be confirmed in the current text.

Reconstructing a year is a distinct service from preparing accounts, and it is priced differently because it is a different activity.

What it involves: obtaining statements for every account including ones nobody remembered, working through every line to determine what it was, requesting duplicate invoices from suppliers who may not provide them, deciding how to treat what remains unidentified, and producing accounts that tie together despite resting partly on judgement. It is investigative work under a deadline shared with every other client of the practice, since year ends cluster.

The fee difference between this and a clean engagement is real and it is significant. It is also not quotable in general terms, and this is worth being direct about, because a specific multiple attached to a business that does not exist is a fabricated statistic even when it sounds plausible. What the difference is for you depends on your transaction volume, the state of your records, how many accounts are involved, how much is cash, and the practice you use. The way to find out is to ask your own accountant for both quotes: one to prepare from complete records, one for what they would charge given the records you actually have. That conversation takes ten minutes and produces a number that applies to you.

There is a second cost in the same category. Reconstruction consumes your time too, because you are the only source for what a payment was for, and that time comes out of running the business during a period you did not choose.

The cost people miss entirely is that a badly closed year does not end. It becomes the starting position of the next one.

Closing balances become opening balances. If closing stock was estimated, next year's opening stock is that estimate, and it feeds directly into next year's cost of goods sold and therefore next year's profit. If a receivable ledger was never agreed to the total, the discrepancy carries forward and grows. If a bank reconciliation was closed with a plug entry, that entry is now in the opening position and will never resolve, because there was never a real transaction behind it.

An asset register that drifted has a longer tail. Depreciation is calculated from it each year, so an asset scrapped and never removed produces a wrong charge indefinitely, and the error is invisible from inside the calculation because the arithmetic is correct.

The compounding matters more than any single item. Each year built on an approximate opening position is harder to close than the last, because the reconciliations that would have caught new errors are already failing for old reasons nobody can separate from new ones. Businesses reach a state where the books cannot be tied to evidence at all, and the entry point is usually one year closed with plugs and estimates under time pressure. That is the year worth avoiding, and the decision to avoid it is made in month two, not month twelve.

The case for good records is strong enough that it does not need overstating, and overstating it invites the wrong conclusions.

Clean records do not prevent scrutiny. Selection for examination is not purely a function of tidiness, and a well-documented business can receive a notice. What changes is what happens next: the question is answerable with evidence, quickly, and far more likely to close without an adjustment. That is worth a great deal and it is not immunity.

Clean records do not make a position correct. A complete file supporting a deduction that is not allowable establishes only that the payment was made. Deductibility, credit admissibility and rate classification are questions of law, and no filing standard resolves them.

And clean records do not mean complete records. This is the failure that hides inside good practice. Every check inside a set of books compares records against other records, or against a bank statement that only saw what passed through the bank. A cash sale never entered disturbs nothing: the trial balance balances, the bank reconciles, the sequence has no gap, and the revenue is missing.

So the honest description of what any bookkeeping system offers, software included, is this. It can capture, categorise consistently, reconcile against a statement, flag a missing document or a gap in a sequence, compare a purchase register against portal data, and refuse to let a period close with an unexplained difference hidden. That is most of the mechanical labour. It cannot know what it was never told, count your stock, or decide a question of law.