Core Close Activities: Part 3

2026-08-31

The processing gap: an accrual that isn't really an estimate

Payroll runs on a monthly cycle: a full month's attendance gets finalised, salary is calculated, and disbursement follows a day or two later. A company closing July's books on 31 July, with ₹40,00,000 in gross salary for 50 employees actually hitting bank accounts on 1 August, has to decide where that ₹40,00,000 belongs.

It belongs to July, the work was done in July, and when the cash moves doesn't change that. But the more precise question is what to call the liability sitting there overnight. It looks like an accrual (expense recognised, cash to follow), but it behaves like the GR/IR liability from cut-off procedures: the amount is already certain, HR has finalised attendance, the salary contract is exact, there's no estimating involved, not an accrual in the sense of a number that still needs a guess. The more precise name is Accrued Salary / Salaries Payable, not a generic accrual, and unlike an estimated accrual, there's no true-up needed the next day: the ₹40,00,000 booked on 31 July and the ₹40,00,000 paid on 1 August match exactly.

Dr Salary Expense ₹40,00,000 / Cr Salaries Payable ₹40,00,000, reversed the next day against Bank.

Statutory dues: whose money is it, and what happens if it's late

What actually reaches an employee's account is never the full gross figure, deductions and employer contributions sit in between, and they behave differently enough to need separate accounts.

Take an employee with a ₹25,000 Basic Pay component (part of a larger gross salary). Provident Fund rules require both employee and employer to contribute 12% of Basic: ₹3,000 each. The employee's ₹3,000 is a deduction from their own gross salary; it was always part of what they were owed, simply routed to PF instead of to their bank account. The employer's ₹3,000 is a separate, additional cost the company bears on top of the gross salary, the same way a shopkeeper topping up a worker's retirement fund out of their own pocket, over and above the wage already promised, is a genuine extra expense, not a slice of the wage itself. For a workforce with combined Basic Pay of ₹2,50,000, the employer's own PF contribution is ₹30,000, and the company's total cash outflow for a ₹5,00,000 gross payroll is ₹5,30,000, not ₹5,00,000.

The employer's 12% itself splits further, into two schemes with different character:

• EPS (Employee Pension Scheme): 8.33% of Basic Pay, capped at a Basic of ₹15,000 regardless of actual Basic Pay. This funds a pooled, formula-based pension, not an individual account, and the cap exists because EPS was designed in 1995 as modest, guaranteed retirement income for lower- and middle-income workers, not a wealth-accumulation vehicle for high earners.
• EPF (Employees' Provident Fund): whatever remains of the employer's 12% after EPS. This is an individual account: contribution plus interest, withdrawn as exactly what accumulated. The employee's own 12% has no such split, it goes to EPF in full.

For a Basic Pay of ₹25,000 (above the ₹15,000 EPS ceiling): EPS = 8.33% of ₹15,000 (the cap, not actual Basic) ≈ ₹1,250; EPF = ₹3,000 − ₹1,250 = ₹1,750. For a Basic Pay of ₹10,000 (below the ceiling, so the cap never binds): EPS = 8.33% of ₹10,000 ≈ ₹833; EPF = ₹1,200 − ₹833 = ₹367.

The cap is worth understanding rather than just applying, because it produces a genuinely counter-intuitive result: removing it wouldn't help higher earners. EPS pays out on a fixed formula, not strictly in proportion to contribution, and it's a shared, pooled fund rather than an individual balance. EPF, by contrast, returns exactly what was put in plus interest, a direct, proportional, individually-owned outcome. Sending more of a high earner's contribution into the pooled, formula-based EPS instead of the individual, guaranteed EPF would leave them worse off, not better; the ceiling actually protects their own money by keeping more of it in the account with the more favourable, individually-owned return.

Full entry:

Two separate liability accounts for PF, Employee Payable and Employer Payable, rather than one combined figure, for the same transparency reason a reclass entry kept accrued income and revenue visible as distinct lines rather than netted: the money's source is different, and a regulator reading the disclosure needs to see the split even though both amounts get deposited in a single challan.

That deposit has a due date, the 15th of the following month for PF, and missing it isn't like a late vendor payment. A vendor payable involves only two parties, and a late payment is, at worst, an awkward conversation. A statutory due involves a third party, the regulator, with its own automatic, pre-set consequences: interest accrues from the due date regardless of intent, a penalty can stack on top of that, and repeated or severe delays can expose company officers to personal liability, not just the company's balance sheet. This is why statutory dues don't get the materiality relief that a small, immaterial vendor variance would; a ₹500 shortfall on a due date is still a due date missed, and the size of the amount doesn't change what the deadline triggers.

Bonus and variable pay: earned over the year, paid on one date

Some payroll cost accrues steadily but crystallises at a single point in time, and the two kinds of bonus a company might promise need different treatment.

Fixed, guaranteed bonus. A policy of "one month's Gross Salary as bonus, paid every December, no conditions attached" is a known total, ₹5,00,000 for a company with ₹5,00,000 monthly gross payroll, spread evenly across the 12 months it's earned in: ₹41,667/month, accumulating to ₹2,91,667 by the end of July. This is mechanically identical to any fixed, contractual accrual: the total is certain, so the monthly rate is certain too, and it doesn't move unless the policy itself changes.

Performance-linked bonus. A policy of "10% of annual Basic Pay if the full-year sales target is hit, scaled proportionally below that" is a genuinely different animal, because the total isn't fixed, it depends on a metric that's still unfolding. Suppose the annual target is ₹60,00,000 and, going into the year, the company assumes it will be hit, accruing ₹25,000/month (10% of ₹30,00,000 annual Basic Pay, spread over 12 months) starting in January. By July, seven months of actual sales data, ₹35,00,000, annualises to exactly ₹60,00,000, meaning 100% achievement, matching the original assumption. In this specific case the flat monthly accrual happens to still be right by the time July closes, but that's a coincidence of the forecast not having moved, not evidence that flat, unchanging accrual is the correct method in general.

The more useful case is what happens when the forecast does move. Say by the end of April, actual sales are running behind, ₹15,00,000 in four months, annualising to ₹45,00,000, or 75% of target. At 75% achievement, the full-year bonus pool is 7.5% of ₹30,00,000 = ₹2,25,000, and the cumulative liability that should exist after four of twelve months is ₹2,25,000 × 4/12 = ₹75,000. But ₹1,00,000 has already been accrued (₹25,000 × 4 months, at the old, too-optimistic rate), ₹25,000 more than the revised estimate now supports.

This is where a performance-linked accrual behaves differently from a fixed asset's useful-life estimate. Depreciation's prospective-only rule works because each past month already consumed a real, distinct slice of a fixed total cost, nothing about a later re-estimate undoes what those months genuinely used up. A bonus accrual isn't allocating a fixed, already-determined cost across separate periods; it's building toward one number that crystallises once, in December, and every month's accrual is only ever an interim guess at that eventual figure. When the guess changes, the correct move is a cumulative catch-up: recompute what the year-to-date liability should be under the new estimate, and let the current month's expense be the difference between that and what's already on the books, in April's case, a ₹25,000 reversal (Dr Bonus Payable ₹25,000 / Cr Bonus Expense ₹25,000), bringing cumulative recognised down to the correct ₹75,000. If performance held steady at 75% into May, the same logic gives May's expense as (₹2,25,000 × 5/12) − ₹75,000 = ₹18,750, a number that only makes sense once the catch-up framing is in place.

Leave encashment: a liability that can run in either direction

Unused paid leave that a company allows employees to carry forward or encash is a liability, valued at the employee's current pay rate, not the rate in effect when each day of leave was originally earned. An employee with 20 days of accumulated, encashable leave and a current Basic + DA of ₹36,000/month (a ₹1,200 daily rate, on the common 30-day convention) carries a liability of ₹24,000, regardless of whether some of those 20 days were earned two years ago at a lower salary. The entitlement is valued at what it would cost to settle today, not at a blended historical rate, the same logic that put the bank's balance, not the cash book's, on the financial statements: what matters is the position as it stands right now, not how it accumulated.

This liability doesn't move in only one direction the way most accruals do. In a month where an employee earns 1.5 new days but uses 2, the balance, and the liability, falls slightly. Company policy typically also caps how many days can be carried forward or encashed (commonly 30 to 45 days); anything earned beyond that cap simply lapses rather than becoming a liability, the same "read the policy's literal boundary" discipline that decided whether a ₹5,000 prepaid fell inside or outside a capitalisation threshold.

Full and Final Settlement: several payroll threads converging at once

When an employee exits, several separate calculations land in a single net settlement, and, like closing out a disposed asset, everything has to be brought current before the final number means anything.

An employee resigning with a 60-day notice requirement but serving only 30 days leaves a 30-day shortfall the company is entitled to recover; leave encashment for unused days becomes payable in full; and the exit month's salary is pro-rated for days actually worked. Different components can use different salary bases, notice-pay recovery and pro-rata final salary typically use the full Gross rate, while leave encashment uses Basic + DA, which matters for the arithmetic, not just the labels.

For an employee with Gross Monthly Salary ₹48,000 (₹1,600/day) and Basic + DA ₹36,000/month (₹1,200/day), who worked 10 days in the exit month, served 30 of a required 60 days' notice, and has 20 days of unused leave:

• Final month's pro-rata salary: 10 × ₹1,600 = ₹16,000 (owed to the employee)
• Leave encashment: 20 × ₹1,200 = ₹24,000 (owed to the employee)
• Notice-period shortfall: 30 × ₹1,600 = ₹48,000 (recoverable from the employee)

Net settlement: ₹16,000 + ₹24,000 − ₹48,000 = −₹8,000, the departing employee owes the company money, not the other way round. Nothing about the individual components was unusual; the net direction only becomes clear once all of them are actually computed and combined, which is exactly why F&F can't be estimated in a hurry from just one or two of these pieces.

Payroll Register reconciliation

Payroll typically runs through its own dedicated software, separate from the core accounting system, precisely because of the statutory complexity above, and that separation creates the same register-versus-GL gap seen with accruals, prepaids, and fixed assets. What the payroll system calculated and what actually got posted into the GL can diverge through a manual entry error, a partial upload, or a rounding difference, and the payroll register, gross salary, each statutory deduction, net pay, by employee, is the source document that reconciliation ties the GL back to, the same role the accrual register and fixed asset register play elsewhere.

A simple but effective companion check is headcount: 50 employees paid last month, 3 new joiners and 1 exit this month, should reconcile to 52 paid this month. A mismatch here is the payroll equivalent of a ghost or unrecorded asset, someone paid who shouldn't have been, or someone owed who wasn't included, caught by comparing against an independent count rather than trusting the system's own total.

Current tax provision: the same cumulative catch-up, applied to tax

A company's income tax provision is a running estimate toward one number, the actual tax liability once the full year's taxable profit is known, built the same way the performance-linked bonus was: re-estimated as better information arrives, with the current period's provision being the gap between the newly-implied cumulative liability and what's already been provided.

At a 25% tax rate, a Q1 estimate of ₹80,00,000 full-year taxable profit implies ₹20,00,000 annual tax, and a Q1 provision of one quarter of that: ₹5,00,000. If H1 actuals push the full-year estimate up to ₹1,00,00,000 (₹25,00,000 annual tax), the cumulative liability that should exist after two of four quarters is ₹25,00,000 × 2/4 = ₹12,50,000. With ₹5,00,000 already provided, Q2's provision is ₹12,50,000 − ₹5,00,000 = ₹7,50,000, visibly larger than a flat quarterly ₹5,00,000 would suggest, because it's catching the year-to-date estimate up to the improved outlook, exactly as the bonus accrual did in April.

Deferred tax: when book and tax depreciation genuinely disagree

Deferred tax arises from timing differences between accounting profit and taxable profit, and depreciation method choice, from fixed assets, is one of the cleanest sources of exactly this kind of difference. Indian tax law generally mandates Written Down Value depreciation for tax purposes, regardless of what method a company uses for its books.

A ₹10,00,000 machine depreciated straight-line over 10 years for books (₹1,00,000/year) but at 15% WDV for tax purposes charges ₹1,50,000 of tax depreciation in year one, ₹50,000 more than the book figure. That extra ₹50,000 of tax-deductible expense lowers this year's taxable profit (and tax paid) below accounting profit, but it's a timing effect, not a permanent saving: because WDV shrinks its own base while straight-line stays flat, book depreciation will eventually overtake tax depreciation in later years, reversing the gap and pushing future tax payments higher than the accounting profit in those years would otherwise suggest. Recognising that future reversal now is what a Deferred Tax Liability does: 25% (tax rate) × ₹50,000 (the timing difference) = ₹12,500, a liability representing tax that hasn't been avoided, only deferred.

GST: output liability, input credit, and its own cut-off trigger

GST works as a running net position each period: Output GST collected on sales, less Input Tax Credit (ITC) available on purchases, with the difference payable to the government. ₹10,00,000 in sales at 18% generates ₹1,80,000 of output GST; ₹1,20,000 of ITC from purchases nets to a ₹60,000 GST payable, due by a fixed statutory date each month.

GST liability itself crystallises based on "time of supply" rules, generally the earliest of invoice date, delivery date, or payment date, the exact same cut-off logic already established for revenue, just with a tax-specific trigger event rather than an accounting one.

TDS on vendor payments: a different kind of consequence

Tax Deducted at Source doesn't stop at salary, it applies to professional fees, contractor payments, and rent, each with its own statutory deadline (typically the 7th of the following month). The consequence of missing it is sharper than PF or ESI's interest-and-penalty structure: under the Income Tax Act, failing to deduct or deposit TDS correctly can get the underlying expense itself disallowed for tax purposes, turning a compliance slip into a direct hit on the company's own taxable profit, not just an interest charge on the missed deposit.

Reconciliation: letting the regulator's own records do the checking

Two reconciliations here go further than an internal register ever could, because the independent source is the regulator's own system rather than an internal log. GSTR-2B, auto-populated by the GST portal from vendors' own filings, should tie to the ITC claimed in the company's books; if a vendor hasn't filed their return, that ITC won't appear in GSTR-2B, and the company may need to reverse a credit it had already claimed. Form 26AS, the income tax department's own record of TDS deducted on the company's income and TDS the company has deducted from others, plays the same role for TDS. Both are stronger completeness checks than an internal contract log or Open PO report, precisely because they're compiled entirely outside the company's own control.

Bundled arrangements: one price, several performance obligations

A single sale can bundle distinct things that get delivered, and recognised, at different times. A machine sold with a 2-year extended warranty for a combined ₹12,00,000, where the machine alone would sell for ₹10,00,000 and the warranty alone for ₹4,00,000 (₹14,00,000 combined standalone value against a ₹2,00,000 bundle discount), requires the discount to be allocated proportionally across both pieces rather than dumped entirely on one: the machine gets (10,00,000/14,00,000) × 12,00,000 ≈ ₹8,57,143, recognised immediately at delivery; the warranty gets (4,00,000/14,00,000) × 12,00,000 ≈ ₹3,42,857, recognised over the 2-year warranty period (roughly ₹14,286/month), the same straight-line-over-time allocation used for prepaids and depreciation, now applied to revenue rather than expense.

Variable consideration: recognising revenue net of what's expected to come back

Revenue should be recognised net of expected returns, rebates, or discounts, using the same estimation toolkit built for accruals. A company with a historical 3% return rate on ₹50,00,000 of sales this month books revenue net of the expected return, keeping the estimate in its own visible liability account: Dr Accounts Receivable ₹50,00,000 / Cr Revenue ₹48,50,000 / Cr Refund Liability ₹1,50,000, a historical-average estimate, structurally identical to the electricity accrual's trailing average.

Percentage of completion: another cumulative catch-up

Long-term contracts, construction, custom software builds, recognise revenue as work progresses rather than waiting for final delivery, using the same cumulative-catch-up mechanic already seen twice this part. A ₹1,00,00,000 contract with an estimated ₹80,00,000 total cost to complete, against ₹20,00,000 of costs actually incurred to date, is 25% complete by cost, and recognises 25% of ₹1,00,00,000 = ₹25,00,000 of revenue to date. If the total cost estimate later revises upward, the project turning out more complex than expected, the percentage complete, and therefore the cumulative revenue that should have been recognised, both need recalculating, with the current period absorbing the difference exactly as the bonus and tax provisions did.

Contra-revenue accounts: keeping the gross figure visible

Returns and discounts get their own accounts, Gross Sales, Less: Returns, Less: Discounts, rather than being silently netted into a single revenue line, for the same transparency reason employee and employer PF stayed in separate accounts: collapsing them loses information a reader might need, even though the final net number would come out the same either way.

Costing methods: the same physical stock, different reported values

Inventory bought at different prices over time needs a costing convention to decide which cost attaches to what's sold versus what remains, and the choice changes the reported numbers even though the physical stock is identical. Opening stock of 100 units at ₹100, followed by purchases of 100 units at ₹110 and 100 units at ₹120 (300 units, ₹33,000 total cost), with 150 units sold during the period:

FIFO treats the oldest stock as sold first: 100 units at ₹100 plus 50 at ₹110 = ₹15,500 cost of goods sold, leaving 150 units (50 at ₹110, 100 at ₹120) worth ₹17,500 in closing inventory.

Weighted Average blends everything into one rate, ₹33,000 ÷ 300 = ₹110/unit, giving ₹16,500 cost of goods sold and ₹16,500 closing inventory.

Same 150 physical units, a ₹1,000 difference in reported closing value, purely from the costing convention, the inventory equivalent of choosing straight-line versus WDV depreciation. In a rising-price environment like this one, FIFO reports a higher closing inventory and lower cost of goods sold than weighted average, since it leaves the more recently, more expensively bought units in stock.

Lower of Cost and NRV: inventory's version of impairment

Inventory is carried at the lower of its cost and its Net Realisable Value, expected selling price less remaining costs to sell, the direct inventory equivalent of the impairment logic already built for prepaids and fixed assets: an asset's carrying value can't outrun what it can actually still deliver.

The 150 units above, carried at ₹110/unit under weighted average, facing a market shift that drops expected selling price to ₹100/unit and estimated selling costs of ₹5/unit (NRV = ₹95/unit), must be written down: (₹110 − ₹95) × 150 = ₹2,250, bringing the carrying value to ₹14,250. The mechanism is identical to a prepaid or fixed asset's recoverable-amount test; only the asset type and the specific inputs differ.

Slow-moving and obsolete provisions

Stock that hasn't moved in a defined period, 180 days, 365 days, typically gets provisioned at policy-set percentages, an ageing-based estimate structurally similar to a bad-debt provision: not a certainty that the stock is worthless, but a systematic, policy-driven recognition that older, unsold stock is increasingly unlikely to realise its full carrying value.

Consignment and goods-in-transit: ownership versus possession

Inventory counting isn't just about what's physically on the shelf, it's about what the company actually owns, and the two don't always match, which is exactly the ownership-versus-possession distinction already built through FOB Shipping Point.

Goods shipped FOB Shipping Point and still in transit at period-end already belong to the buyer, even though the buyer hasn't physically received them yet: the seller should have already recognised the sale and removed the goods from their own inventory, while the buyer needs to include those in-transit goods in their own inventory figure even though a warehouse count would miss them entirely, a completeness gap a physical count alone can't catch. Consignment inventory runs the opposite mismatch: goods sitting on a retailer's shelves, fully within their physical count, that legally still belong to the manufacturer or consignor until an end customer actually buys them, meaning the retailer should exclude consignment stock from their own inventory balance even though it would otherwise show up in any count taken on-site. Both cases are the same lesson cut-off procedures established: the location of the goods is not, on its own, the answer to who owns them.

Current versus non-current: reclassifying the next 12 months

A long-term loan's classification isn't fixed for its whole life, whatever portion falls due within the next 12 months has to move from non-current to current liabilities at each period-end. A five-year loan of ₹50,00,000, with ₹20,00,000 in remaining principal by the close following its third year (two more annual instalments of ₹10,00,000, in years four and five), reclassifies the ₹10,00,000 due in year four to current liabilities, while the final ₹10,00,000 due in year five stays non-current. This directly affects the current-ratio calculations flagged earlier in this study, one more reason the classification isn't a formality.

Interest accrual

Loan interest accrues for the days elapsed regardless of the actual payment schedule, the same mechanic behind every other accrual in this study, just applied to a lender relationship: interest paid quarterly still needs accruing monthly, for whatever portion of the period has actually elapsed by close.

Debt covenants: a compliance check with real teeth

Loan agreements frequently require maintaining specific financial ratios, a Debt-to-Equity ceiling of 2:1, say, and someone has to actually calculate that ratio each close and confirm compliance, the same named-ownership discipline built for reconciliations generally. A breach carries a consequence beyond the ratio itself: with Total Debt of ₹80,00,000 against Total Equity that has fallen to ₹35,00,000 (a 2.29:1 ratio, over the 2:1 limit), the loan can become technically callable, and without a formal waiver from the lender obtained before the reporting date, the entire long-term loan, not just next year's instalment, may need reclassifying to current liabilities, overriding the normal 12-month rule entirely. The lender's waiver letter is the authoritative document that prevents this, playing the same role a signed confirmation played in cut-off procedures: without it, in writing, before the reporting date, the technical breach stands.

Dividends: proposal versus declaration

A board proposing a dividend on 25 July, ahead of a 31 July period-end, but with shareholder approval only coming at an Annual General Meeting on 15 August, has not created a liability as at 31 July. Under Ind AS 10, a dividend is only recognised once formally declared by the body actually authorised to do so, the boardroom discussion is not the authoritative trigger event, the same way a verbal agreement was never sufficient for cut-off purposes without a timestamped, authoritative document behind it. Until the AGM approval, the proposed dividend sits in disclosure notes only, not as a recognised liability.

Retained earnings: where everything else in this section lands

Every equity movement discussed in this part, net income for the period, dividends once actually declared, and any prior-period error corrections, flows through the same Statement of Changes in Equity already built out in full in the fixed-asset retrospective-restatement discussion. Nothing new happens mechanically here; it's the same statement absorbing a wider set of inputs.

Ten of the fourteen items are now covered. Four remain for the next part: intercompany, FX, journal entries, and post-close adjustments. Cut-off testing's open thread from Part 1, what a repeated pattern across several invoices should mean, and manipulation risk more broadly, is still unresolved.