FIFO, weighted average, and why you cannot switch on a whim

Three permitted methods

IFRS (IAS 2) and ASPE (Section 3031) permit the same three ways of assigning cost to units sold: specific identification for goods that are not interchangeable (a dealer's boats, each with a serial number), and for everything else a choice between FIFO and weighted average. LIFO is not permitted under either framework. Several US-authored textbooks teach it, and US GAAP allows it; in Canada it is not an option, not even for a private company.

Worked example — one month of maple syrup

Kettle Creek Foods Inc. starts April with no syrup, buys 100 cases at $8 on 3 April and 200 cases at $9 on 17 April, and sells 220 cases during the month. Cost of goods available: $800+$1,800 = $2,600 for 300 cases.

FIFO sends the oldest cost out first: the 220 sold are the 100 at $8 and 120 of the $9 cases.

$${\text{COGS}}_{FIFO}=100\times 8+120\times 9=800+1,080=1,880$$

Ending inventory is the 80 remaining cases at $9=\ast \ast$720**. Check: $1,880+$720 = $2,600$.

Weighted average pools the costs first:

$$\text{Unit cost}=\frac{2,600}{300}=8.6667{\text{COGS}}_{WA}=220\times 8.6667=1,906.67$$

Ending inventory is $80\times 8.6667=693.33$. Check: $1,906.67+693.33=2,600$.

In a month of rising prices, FIFO's COGS is lower ($1,880versus$1,906.67), so FIFO reports the higher gross profit and the higher ending inventory. In falling prices the ranking flips. The two methods never change the total cost of $2,600; they only change how much of it is expensed now and how much waits on the balance sheet.

Consistency — and the two things that are not a policy change

Once Kettle Creek picks weighted average for syrup, it keeps using it. Consistency is a component of comparability (L1.5): a reader comparing April with March must not be comparing FIFO with weighted average. A company may change an accounting policy only when a standard requires it or when the new policy gives more relevant and reliable information — not because this year's number looks better under the other method. And when it changes, the change is applied retrospectively: prior-period figures are restated as if the new policy had always applied, and opening retained earnings is adjusted for the cumulative effect. If Kettle Creek moves from FIFO to weighted average on 1 January, its comparative statements for last year are re-presented on weighted average too.

Two neighbours are easy to confuse with a policy change:

• A change in estimate — a new residual value for a truck, a revised warranty rate, a different expected selling price for syrup — is applied prospectively, from the date of the change forward. Nothing is restated, because the old estimate was the best information at the time.
• An error — a miscount, an unrecorded invoice, a formula mistake — is corrected by restating the affected prior periods (Module 2's last two lessons). An error is not a policy change; a policy change is not an error.

The midterm tests the three as a set: policy → retrospective; estimate → prospective; error → restate.