The tax cross-check

There is one number in a set of accounts that a company finds unusually hard to fake, and it is the tax. A management team with an incentive to look better than it is can lean on a dozen judgements to lift reported profit — how complete a project is, how long an asset will last, how large a provision to hold — but at the tax authority it faces a second reader of the same numbers, one with no interest in flattering anyone and every interest in collecting. That is why tax behaves like a truth serum: a company can present a large book profit to its shareholders, but it usually cannot, at the same time, fake the cash it hands the government. The gap between the profit it shows you and the tax it pays is one of the most revealing reconciliations in the whole report. illustrative

The cross-check has three ties, and the forensic reader runs all three. The first is the effective tax rate — tax expense divided by profit before tax — set against the statutory rate the company should broadly pay. A rate that sits persistently and far below statutory, on a company with no holiday to explain it, raises the sharpest question in this module: is the book profit real, or is it not being taxed because it is not real cash? The second tie is the tax expense accrued in the P&L against the tax actually paid in cash, which the cash-flow statement discloses; a wide and widening gap flags a deferred-tax build-up or a stack of aggressive positions. The third is this year's rate against the run of prior years: a suddenly volatile rate is usually a lever being pulled, not a stable business paying a stable share.

And then the move that makes the number mean something: you interpret a low rate by sector, never in isolation. The same low figure is the legitimate, designed result for a business sitting on a genuine tax shield and the single sharpest flag for one with no shield at all. So the cross-check is the constant; the shield that would legitimise a low rate is what you check for in each business, and where there is none, the low rate is the finding.

Run the three cross-checks in the same order every time.

Effective rate against statutory. Divide tax expense by profit before tax and set the result beside the statutory rate. If they are close, the profit is being taxed as profit — quiet corroboration that it was earned. If the effective rate sits far below statutory year after year, with no tax holiday, no large exempt income and no unusual structure to explain it, you have the module's central question: a company earning genuine profit is normally taxed on it, so a profit that escapes tax may be a profit that is not real cash. The gap is not the verdict — an innocent exempt dividend or a one-year credit can pull the rate down — but a persistent, unexplained shortfall is exactly what to make the company account for.

Expense against cash paid. The P&L shows a tax expense on the accrual basis; the cash-flow statement shows the tax actually paid in cash. The two differ legitimately from year to year, but over time a company earning real profit pays cash tax roughly in line with the charge it books. When the accrued expense stays high while the cash paid runs at a fraction of it, and the shortfall accumulates as a deferred-tax liability that only ever grows, read that build-up carefully: it is where timing differences, aggressive positions and profit recognised ahead of the taxman's acceptance all collect. A deferred balance that reverses and unwinds is ordinary; one that swells one-directionally for years is reported earnings the company is not funding with cash, and may never.

This year's rate against its own history. The third check needs no benchmark beyond the company itself. Line up the effective rate for five or six years. A stable business pays a stable share, so a rate steady and near statutory is unremarkable. A rate that lurches — normal for years, then collapsing in the very year profit spiked, or swinging with no change in the business — is usually a lever being worked: a credit taken, a position adopted, an exempt item engineered, a book profit inflated without a matching taxable profit. You are not asking whether the rate is high or low but whether it moved in a way the underlying business does not explain.

Those three ties locate the gap. The next section is what decides whether the gap is a shield or a flag.

A low effective tax rate is not a flag in itself, because for whole categories of business it is expected and entirely legitimate — a power or infrastructure asset on a statutory holiday, an IT firm running SEZ units inside their window, a manufacturer with heavy accelerated depreciation on a freshly commissioned plant. In each case a named, disclosed tax shield accounts for the low rate, the tax note reconciles the gap line by line, and the low cash tax matches a real deferred-tax timing that later reverses. So the same low rate inverts in meaning across sectors, and the deciding factor is whether such a shield exists.

Tell a reader a company pays only a 10% effective rate and it means nothing until you know the business. For a power project on a holiday, throwing off accelerated depreciation from a new plant, that low rate is exactly what the structure is designed to produce, and the tax note reconciles every point of the gap. For a mature FMCG company with no holiday, no SEZ and an old, fully-depreciated asset base, the same 10% has no explanation — and the absence of an explanation is the finding. A reader who learns one sector's tax pattern as the universal rule will misread everywhere else, condemning a legitimate infra holiday as fraud or excusing an unexplained FMCG rate as clever planning.

Take a composite mid-cap that reported a spectacular year — profit before tax up from ₹360 crore to ₹610 crore, margins at a record, the commentary all momentum. On the headline it looks superb, and a reader who stopped at the post-tax figure would have been delighted. The forensic reader runs the tax cross-check first. illustrative

Start with the effective rate. Tax expense was ₹49 crore on ₹610 crore of pre-tax profit — an effective rate of 8%, against a statutory rate nearer 25% and the company's own 24% the year before. There is no holiday in the notes, no SEZ unit, no large exempt income, and the asset base is old and mostly depreciated, so no accelerated-depreciation shield either. That is the central question with a location: a record profit taxed at 8%, with nothing disclosed to explain why the taxman recognises so little of it. Second, tie expense to cash. Cash tax paid was ₹40 crore — barely above the prior year's ₹36 crore — so while book profit leapt 70%, the cash handed to the government hardly moved, and the difference piled up as a deferred-tax liability now growing for the third year running. Third, the history: the effective rate ran 23–25% for four years, then collapsed to 8% in the exact year profit spiked. Three ties, breaking together, in the same direction.

None of the three, on its own, is proof. A genuine one-off credit, a newly claimed incentive, a legitimate timing item — each could innocently pull one lever for one year. But the cross-check does not read the ties one at a time; it reads them together, and a rate that collapses precisely when profit spikes, cash tax that refuses to follow the profit, and a deferred liability swelling one-directionally are a single coherent pattern: book profit recognised faster than any taxable profit the authority will accept. The flattered post-tax profit is the noise; the cash tax that will not follow it is the signal. That is now the thing the company must explain, and its explanation — or the absence of one — is what turns the flag into a verdict. The habit to build is to run the cross-check before you admire the post-tax profit, not after.

When a company posts strong profit on a suspiciously low tax rate, the forensic analyst does not accuse; he asks the reconciliation into the open: "Your effective tax rate fell to about 8% this year from 24%, with no holiday I can see in the notes, and cash tax paid barely moved while profit rose 70% — can you walk us through what drove the rate down, and how much of the gap is a deferred-tax build-up versus cash actually saved?" The question names the two broken ties — rate versus statutory, expense versus paid — and asks the company to close them.

A good answer decomposes the low rate into named, checkable pieces and separates the one-off from the recurring:

"Fair question, and the drop is real this year. About twelve points of it is a one-time credit from a favourable tribunal order on a prior-year dispute, disclosed in the tax reconciliation note — that won't repeat, so we'd guide the normalised rate back to roughly 23% next year. The rest is accelerated depreciation on the new line commissioned in Q2, a genuine timing item: cash tax is low now, the deferred-tax liability builds, then reverses over the asset's life. The underlying rate is unchanged; this year carries a one-off credit and a depreciation timing effect, both in the note."

It closes the tie with specifics — a disclosed credit, a real depreciation timing — and guides a normalised rate you can hold it to. Compare the evasion:

"We're very comfortable with our tax position and it's fully compliant and audited. Tax moves around year to year and we manage it prudently. Our effective rate reflects our planning and our structure, and we don't guide on tax specifically. The balance sheet has never been stronger."

It reassures without reconciling. "Comfortable" and "tax moves around" are assertions, not the decomposition the question asked for; it names no credit, no depreciation item, no disputed demand, and reaches for "compliant and audited" and "momentum" in place of the numbers that should agree. That is the answer a company gives when the low rate will not survive being broken down. The follow-up that settles it — "of the fall in the rate, how much is a recurring shield versus a one-off, and how much of your tax expense over three years was actually paid in cash rather than deferred?" — forces the split between a repeatable reason and a one-time lever, and drags the deferred build-up into the open, where earnings the taxman has not accepted cannot hide behind "we manage it prudently." A company whose low rate rests on a real, disclosed shield usually volunteers the reconciliation without being pushed; the silence is the tell.

The cross-check locates the gap; it does not, by itself, name the cause. A low rate with cash tax that lags the charge is consistent with a legitimate undisclosed timing difference, an aggressive-but-defensible position, a credit the reader simply missed in the notes, or book profit inflated ahead of any taxable profit. Reconciling the rate tells you the tax line and the profit disagree; which explanation is true comes from the tax reconciliation note, the deferred-tax schedule, the contingent-liability disclosure of disputed demands, and the concall — the work of reading around the number once the check has told you to. Treating a low rate as proof of fraud is the paranoid error, and it stops the investigation exactly where it should begin.

Nor can a clean tax line prove honesty. Some manipulations keep the tie closed on purpose — a company can pay real cash tax on a genuinely taxable but fabricated related-party sale. A clean cross-check lowers suspicion; it does not eliminate it, because a determined deception can be built to satisfy the taxman too. It is simply expensive and hard to sustain, which is why keeping the tie closed is a real cost to a fraud and why the check catches the majority that will not pay it. And the check cannot forecast policy: holidays expire, SEZ benefits sunset, rates are cut or surcharged. A rate legitimately low today because of a holiday will rise when the holiday lapses. The cross-check tells you whether today's tax corroborates today's profit — not what tomorrow's regime will be.

The two ways readers get fooled follow from this. The first is admiring the post-tax profit and never asking the tax why it is so low — a record profit taxed at a fraction of statutory, with no shield to explain it, is a reported number contradicted by the one figure a company can least easily fake, yet it is convincing because it is specific and printed in bold. The second is mistaking the tax expense in the P&L for tax paid: a large charge that is mostly deferred and one that is mostly paid look identical on the face of the P&L and mean completely different things about how real the profit is. Both defences are mechanical — never accept a profit without asking whether the government is taxing it as profit, and never read the accrued charge without tying it to the cash tax paid.