Costco beat: what lasts beyond the tariff refund?
By @engtmk · Research as of 25 September 2026
I own Costco. Before Thursday evening’s results, my latest portfolio snapshot showed the position down 2.7%. I did not add to it on Tuesday, 22 September.
Neither my purchase price nor that unrealised loss tells me what the business is worth today. The useful question after these results is whether the evidence for sustainable earnings has improved enough to justify committing more capital.
My assessment: an encouraging quarter whose earnings beat survives the disclosed tariff-refund adjustment. But the refund also helped fund lower prices, so subtracting it from earnings does not recreate a quarter in which it never happened. Spending, membership and operating execution still need to support the investment case at the valuation paid.
This analysis uses Costco’s release, management’s earnings-call remarks and Reuters’ reported LSEG consensus. Calculated adjustments and my investment interpretation are identified below. It is a post-results assessment, not a price target or instruction to trade.
The beat after the refund adjustment
Costco reported its 16-week fiscal fourth quarter ended 30 August. The following bridge uses the company’s release and the estimate basis reported by Reuters on 24 September.
| EPS bridge | Dollars per diluted share |
|---|---|
| Reported EPS | $6.75 |
| Disclosed non-recurring refund benefit, net of partial reinvestment | −$0.15 |
| EPS excluding that benefit: 6.75 − 0.15 | $6.60 |
| LSEG consensus reported by Reuters | $6.53 |
| Ex-benefit surprise: 6.60 − 6.53 | +$0.07, approximately 1.1% |
Sources: Costco release and Reuters coverage. The percentage surprise is my calculation: 0.07 ÷ 6.53.
The release’s wording matters: the 15 cents is already after partial reinvestment in member value. Deducting a further estimate of that reinvestment would double-count an adjustment. Equally, the resulting EPS is an ex-disclosed-benefit measure, not proof that every remaining dollar will recur.
Consensus varies between providers and snapshots. I use the Reuters/LSEG comparison consistently rather than mixing an estimate from another service with this adjustment.
The broader principle is covered in Why headline profit is often the wrong number. Correctly reported earnings and earnings suitable for forecasting are different questions.
A temporary refund can have a longer commercial effect
Reuters reported that management received $184 million associated with tariff refunds during the quarter and used some proceeds for price reductions. That is a cash event with a commercial consequence, not merely an accounting mark. Reuters.
There are two possible channels to evaluate. The retained benefit can raise current profit; the portion passed to shoppers can change purchasing behaviour. Lower prices might encourage visits, increase quantities purchased or strengthen the perceived value of membership.
Those are plausible mechanisms, not measured contributions. A price reduction also lowers revenue per unit. Whether the net effect on sales is positive depends on the response in volumes and product mix. I cannot label a particular portion of comparable-sales growth “refund-driven,” or subtract an invented amount to produce supposedly clean sales.
Nor must every benefit disappear when the refunds end. If customers develop lasting shopping habits, part of the commercial effect could persist. If demand depends on continuously subsidised prices, sustaining it may require a greater ongoing sacrifice in margin. That distinction is the investment test.
Non-recurring does not mean confined to one quarter
CFO Gary Millerchip said a similar amount had already arrived in fiscal Q1 2027 and that Costco intended to reinvest most future proceeds in member value. He said these items would continue affecting fiscal 2027 results. Call transcript, prepared remarks around 10:20.
The implication is that investors should track each quarter’s separately disclosed net impact. A finite recovery of earlier costs can arrive in instalments without becoming an ordinary annual earnings stream. I would not mechanically assume the same EPS adjustment next quarter, or extrapolate any future refund benefit indefinitely.
Timing also matters for cash analysis. Cash received, the accounting benefit and spending passed through to members need not produce identical amounts in the same reporting period. The next disclosure should explain the relationship rather than invite an inference from one headline number.
Comparable sales: stronger than the estimate, still measured in dollars
| Fourth-quarter operating snapshot | Result |
|---|---|
| Total comparable-sales growth | 9.4% |
| Comparable growth excluding gasoline-price and currency effects | 6.7% |
| Reuters/LSEG expectation on that adjusted basis | 6.11% |
| Membership fee income, current / prior-year quarter | $1.850bn / $1.724bn |
| Operating income, current / prior-year quarter | $3.801bn / $3.341bn |
Sources: Costco release and income statement, Reuters for consensus.
Removing gasoline-price and currency effects makes the comparison more useful, but it does not convert sales into a pure volume measure. Other product prices, the mix of purchases and transaction size still matter. Nor does this adjustment remove the potential behavioural effect of price reductions funded by refunds.
Management reported worldwide traffic growth of 3.3% and adjusted average-ticket growth of 3.3%. Call transcript, around 11:22.
My interpretation is that demand has more support than an aggregate sales figure alone can show. Even so, higher traffic does not establish why people visited, and a larger ticket does not reveal whether they bought more units or a more expensive mix. Sustained growth across visits and spending would provide stronger evidence than a single quarter’s beat.
Membership is the second test
Membership income grew approximately 7.3%, calculated as 1.850 ÷ 1.724 − 1. It is a useful companion to merchandise sales because it tests whether customers continue to pay for access to the proposition.
Millerchip said Q4 was the final quarter with a year-over-year benefit from the 2024 fee increase. He reported 84.1 million paid members, up 3.8%, with renewal rates of 92.3% in the U.S./Canada and 89.8% worldwide. Call transcript, around 12:27.
My focus now is the composition of future fee growth. New paying members, retention and upgrades to a higher membership tier are different drivers from a fee change. A rising fee-income total is encouraging, but the number alone cannot establish how much comes from a larger customer base versus more revenue per member.
I will also resist treating membership income as a standalone profit figure. Supporting the value customers renew for requires merchandise, people, premises and services. The economic strength of the model should appear in the combined business, not in an assumption that one revenue line has no associated costs.
Profitability needs a careful denominator
Operating income increased approximately 13.8%, calculated from the table above. That supports a positive reading of the quarter, but reported operating growth should not automatically be labelled recurring growth when a temporary benefit affects the accounts.
The useful margin questions are more specific: are merchandise economics improving, are expenses being controlled, and how much of an apparent ratio improvement comes from the sales denominator? A jump in fuel prices can increase revenue without an equivalent change in the underlying operating effort. Margin percentages then become harder to interpret without the corresponding dollar figures and mix.
I would examine comparable definitions over time rather than switching between reported and adjusted margins to favour whichever looks better. I would also avoid dividing an after-tax EPS adjustment by sales to invent an operating-margin adjustment: the tax basis and accounting location must first be reconciled.
Cash conversion is the next check. Profit growth is more valuable when it is supported by cash after necessary investment, with working-capital movements explained. Refund cash should be distinguished from the ongoing cash generated by selling merchandise and serving members. This article does not supply a full cash-flow reconciliation, so I do not infer free-cash-flow growth from EPS.
The debate on the call
UBS analyst Michael Lasser questioned whether pricing reinvestment was changing the sales trajectory. Millerchip emphasised resilient spending and described the reinvestment as returning value to members. Q&A, around 25:14–28:05.
That exchange captures the main disagreement. The optimistic case is that value returned now builds loyalty and profitable spending later. The cautious case is that the investment maintains competitiveness without producing a lasting improvement in growth.
My next check is whether visits, member retention and operating economics remain healthy as the timing and scale of refunds change. Neither the analyst’s question nor management’s answer supplies a quantified causal estimate of the pricing programme’s return.
A beat does not settle valuation
Reuters described shares as flat in Thursday’s extended trading. That is a timestamped initial reaction, not a description of Friday’s session or proof that the market reached a settled conclusion. Reuters.
A quiet reaction can coexist with good results when expectations were already demanding. It can also reflect attention to future growth, margins or unrelated market conditions. Price action alone cannot identify which explanation dominates.
For a prospective addition, I would separate the earnings forecast from the multiple paid for it. As a purely illustrative sensitivity, 10% earnings growth with a 10% contraction in the P/E multiple produces a price change of approximately −1%: 1.10 × 0.90 − 1, before dividends. This is not a forecast or a claim about Costco’s current multiple. It demonstrates why improving earnings need not guarantee a positive shareholder return.
A defensible valuation would require a dated share price, a consistent forward earnings estimate and assumptions about how growth and the multiple evolve. I am not presenting that model here. The quarter improves the evidence about the business; it does not, by itself, establish a margin of safety.
My own position being below cost is equally insufficient. Averaging down merely because a holding is red would confuse portfolio history with prospective return. I did not add on Tuesday, and this report alone does not establish a reason to increase the position.
What would change my assessment?
| Test | Evidence that strengthens the case | Evidence that weakens it |
|---|---|---|
| Member spending | Visits and spending remain healthy across successive reports | Growth depends increasingly on promotions without durable demand |
| Membership | Paying-member growth and renewals support fee income | Slower acquisition or retention weakens the recurring base |
| Operating execution | Comparable margins and cash conversion improve after identified temporary effects | Sales growth requires a disproportionate rise in costs or investment |
| Refund treatment | Clear reconciliations distinguish receipts, reinvestment and net earnings impact | Reported growth becomes difficult to separate from temporary benefits |
| Valuation | Sustainable earnings offer adequate returns at the price considered | A demanding price leaves little room for normalisation or execution misses |
Management identified 7 October, after market close, as the next September sales release. Call transcript, closing prepared remarks. That will provide another demand observation; monthly sales will not settle quarterly profitability or cash conversion.
There is a useful connection to my Target research: attracting customers and earning an adequate return from them are related but separate achievements. Costco’s membership structure adds another test. Its results do not establish that every retailer is seeing the same consumer or earning the same return.
I see a business delivering encouraging results, with a beat that survives the disclosed adjustment. I still want evidence that spending and membership economics can support durable earnings at a price worth paying.
Sources and freshness
- Company facts: Costco’s 24 September release and accompanying unaudited statements for fiscal Q4 2026. Release figures take precedence over minor transcription differences in the call transcript.
- Management commentary: the 24 September earnings call, accessed through MarketBeat’s transcript. Plans and explanations are management statements, not independently demonstrated outcomes.
- Expectations and initial reaction: Reuters’ 24 September coverage, using LSEG estimates. The after-hours observation is historical, not a live quote.
- Author calculations: ex-benefit EPS and surprise; membership and operating-income growth; explicitly hypothetical P/E sensitivity. No full valuation, cash-flow model or multi-quarter chart series is supplied.
- Portfolio disclosure: I hold COST; the pre-results −2.7% snapshot and absence of a Tuesday addition are my reported portfolio facts. They are not current performance figures or an independent account audit.
My portfolio and views, not investment advice.