Cut-off testing for purchases: procedures and worked examples

Purchase cut-off testing establishes that purchases and the payables they create are recorded in the period in which control of the goods or services passed to the client. It is tested by comparing three dates on transactions either side of the year end — date of receipt, invoice date and the date recorded in the books — and resolving disagreements by reference to the terms of delivery rather than to whichever date is most convenient. Both directions have to be tested: purchases recorded in the year that belong to the next, and purchases in the next year that belong to this one.

Cut-off is a control question, not a date question

The temptation is to treat cut-off as an arithmetic exercise on dates. It is not. The question is when control of the goods transferred, and the dates are only evidence about that. Terms of delivery decide the answer, and the same three dates can point to opposite conclusions under different terms.

The three dates, and which field they live in

Before designing the test, confirm which date the client's system actually records and which one the purchase register was filtered on. In Tally and similar systems the voucher date and the supplier's invoice date are separate fields, and a register run on one will not contain the same transactions at the year end as a register run on the other.

• Date of receipt: from the goods receipt note, gate inward register, security register or transporter's delivery record. This is the field that usually does not exist in the accounting system and has to be obtained separately.
• Invoice date: the supplier's date on the face of the tax invoice. Relevant to the input credit period and to the terms, but it does not determine when control passed.
• Date recorded: the voucher date in the books. This is the only one under the client's control, and it is the one being tested.

Sizing the window

A fixed number of days either side of the year end is a habit, not a method. The window should be long enough to cover the client's normal lag between receipt and recording, which you can measure from the register rather than guess.

1. 01Measure the normal lagFor a few hundred transactions during the year, compute the days between the invoice date and the voucher date. Look at the distribution rather than the average. If 90% are recorded within eleven days and the tail runs to forty, a seven-day window tests almost nothing.
2. 02Set the window to cover the bulk of the tailCover the period within which the great majority of that client's transactions are recorded, in both directions from the year end. Record the measurement and the resulting window in the working paper, because that is the sample basis.
3. 03Add the goods in transit population separatelyPurchases on ex-works terms dispatched before and received after the year end will not appear in either direction of the date test, because they are not recorded in either period. They have to be found from the supplier's dispatch documents and the transporter's records.
4. 04Add a search for unrecorded liabilitiesReview payments made in the first weeks after the year end and identify those relating to goods or services received before it. Review the file of unprocessed invoices, and the vendor statements received after the year end.

The procedure

1. 01Select the last transactions recorded before the year endFrom the purchase register, by voucher date, running backwards from 31 March across the window. For each, obtain the receipt evidence and the invoice, and establish the date control passed.
2. 02Select the first transactions recorded after the year endFrom the next year's register, running forwards from 1 April. Same test, opposite direction. This is the leg most often skipped, and it is the one that finds understated liabilities.
3. 03Agree both selections to the goods inward recordsWork from the client's inward record, not from the invoice. The inward record is the independent evidence of when the goods arrived; the invoice date is the supplier's choice.
4. 04Check the last few document numbers in each sequenceThe last goods receipt note number before the year end and the first after it, against the client's own count-sheet cut-off if a physical count was carried out. These should reconcile. A receipt note numbered before the cut-off but recorded after it is an exception whether or not the dates look right.
5. 05Test consistency with stockAn item treated as a purchase this year must either be in closing stock or have been consumed or sold. An item excluded must not be in closing stock. Cut-off errors that are consistent on both sides do not affect profit; ones that are inconsistent do, and those are the material ones.
6. 06Look at the direction of the errors foundA scatter of errors in both directions is a process weakness. A run of errors all in the same direction at the year end is a different finding and should be escalated as one.

Four worked examples

Input credit does not follow the accounting cut-off

An accrual raised for goods received in March against an invoice dated April is correct for the financial statements, and it does not entitle the client to the input credit in March. Credit under the CGST Act depends on conditions including possession of the invoice, and the invoice is dated April. The two treatments legitimately differ, and the reconciliation between the purchase register and GSTR-2B will show the difference as a timing item.

Concluding

The conclusion on cut-off has to state what the testing showed and what the aggregate effect of the errors found was, in both directions and gross rather than netted. Netting a March purchase wrongly excluded against an April purchase wrongly included produces a small number that hides two separate control failures.

• Window measured from the client's own recording lag, and recorded.
• Both directions tested, with the post year-end leg evidenced.
• Goods in transit population identified separately, from dispatch and transport records.
• Receipt sequence agreed to the physical count cut-off where a count was performed.
• Consistency with closing stock checked on each item.
• Errors summarised gross and by direction, and carried to the summary of uncorrected misstatements.
• Input credit timing differences cross-referenced to the GSTR-2B reconciliation rather than raised twice.