Tariff refunds: who gets to keep the money?

US companies are recovering tariffs struck down by the Supreme Court. Customers who paid higher prices have a straightforward question: will any of that money come back to them?

On August 20, 2026, Walmart announced that it had received nearly $2.9 billion in refunds of duties imposed under the International Emergency Economic Powers Act, or IEEPA. It says customers will benefit. FedEx’s accounts take a different approach: much of the money received is matched by an obligation to repay customers. The distinction matters more than the size of the headline refund.

A tariff can travel through the economy in one direction and return by another. The importer pays customs, then may recover part of the cost through higher prices. When the duty is refunded, the government follows the customs record. It does not reconstruct every subsequent retail transaction.

The question is who remains out of pocket after the government has paid the money back. Customs documents, company filings and consumer lawsuits help establish part of the answer. They do not yet support a reliable estimate of how much households will ultimately recover.

From customs records to store prices

On February 20, 2026, the Supreme Court held in Learning Resources v. Trump that IEEPA did not authorize the president to impose the tariffs at issue. The ruling concerns that statutory authority. It did not abolish every US tariff or adjudicate individual shoppers’ claims against retailers. Supreme Court decision, holding and opinion.

US customs regulations make the duty a liability of the importer. The first phase of the Consolidated Administration and Processing of Entries (CAPE) refund system opened on April 20, 2026, accepting declarations from importers and authorized brokers. Customs rules also allow a designated party to receive a refund. These are mechanisms for processing an identifiable customs transaction. 19 CFR 141.1, importer liability. CAPE launch. Designated refund recipient.

They cannot, by themselves, identify everyone who absorbed its economic cost. A foreign supplier might cut its price. An importer might accept a lower margin. A retailer might charge its customers more. That distribution is called tax incidence: several parties may bear the cost even though the government’s payment record names one.

Research published by New York Fed economists on February 12, 2026, using import data through November 2025, estimated that nearly 90% of the burden of the 2025 tariffs studied fell on US businesses and consumers together. That is not a finding that households alone paid 90% of every tariff. Nor does it establish the retail pass-through of any particular company. Original study and methodology.

Those qualifications are essential when trying to work out who should benefit from a refund. The legal payer and the ultimate bearer of the cost need not be the same person.

The same $20 refund can produce three different outcomes

Consider a deliberately simplified example. A business imports and sells one item. A $20 tariff is imposed. Sales volume, all other costs and the original pricing conditions remain fixed. Ignore income taxes, interest and wider commercial effects.

The business could absorb the full $20. Its margin would fall by that amount. A later $20 refund would restore its original position. In this scenario, the shopper paid no tariff-related price increase to be reimbursed.

Alternatively, the business could raise the selling price by $20. Its margin would be preserved while the tariff was in force. A subsequent government refund, with nothing returned to the buyer, would leave the business $20 better off than in the no-tariff case. The customer would still have spent the extra $20. Passing through half the duty produces an intermediate result.

This hypothetical example isolates the question that a company-level analysis must answer: how much of the cost had already been recovered from someone else? Customs receipts cannot answer that without evidence about prices, margins and any losses the importer absorbed.

Real businesses may also have lost sales, changed suppliers or financed more expensive inventories, incurring inventory carrying costs. Refunding the principal does not automatically reverse those effects. Interest may compensate for the use of money over time; it is not a complete valuation of disruption.

Dollar Tree’s refund runs through earnings

Dollar Tree’s quarterly filing, for the period ended August 1, 2026, records approximately $369 million in principal and $14 million in interest received during the quarter. The principal reduces cost of sales; interest is recorded in other income, net. These are received funds, not simply the value of an outstanding application. Form 10-Q, “Tariff Refunds.”

The August 27 earnings release provides a reconciliation of the items the company groups into the net impact of tariff refunds. That grouping includes reinvestment expenses and certain other duties. A useful reading needs all of those items, rather than a comparison between the gross inflow and one selected promotional expense. Earnings release, footnote 1.

The $22 million covers markdowns recorded in cost of sales; reinvestment also includes $15 million in selling, general and administrative expenses. The $13 million in other duties is a separate customs charge. The calculated $333 million before tax measures this group of items’ quarterly earnings effect. It does not measure unspent cash or improper enrichment.

Dollar Tree also expects further reinvestment in the third and fourth quarters, including markdowns and freight expenses. Later spending could therefore offset some of the immediate earnings benefit. That declared timetable does not establish what the original buyers will recover. Form 10-Q, discussion of cost of sales and tariff refunds.

There is a related trap for investors. Reported diluted earnings per share of $2.70, both under US generally accepted accounting principles (GAAP) and on an adjusted basis, include a company-defined $1.31 net benefit per share from tariff refunds. In this release, “adjusted” does not mean “excluding the refund.” Quarterly results and non-GAAP reconciliation.

Readers therefore need to check what the adjustment leaves in before extrapolating this quarter’s earnings into a lasting profit run rate.

Walmart bets on lower prices

Walmart frames its approach around lower prices and customer experience. Its August 20 release covers the quarter ended July 31, 2026, and says price investments only partly offset the refund benefit in that period. Earnings release, page 1.

Its disclosures also separate some of the exceptional boost from the underlying business. Adjusted operating income at constant currency grew 17.4%, including approximately 750 basis points, or 7.5 percentage points, of net contribution from refunds after the quarter’s price investments. That is a contribution to the growth rate of operating income, not a 7.5-point increase in the operating margin. Official results overview.

Management says related spending continues into the following quarter and expects the financial effects of receipt and reinvestment to be largely contained within the current fiscal year. This describes timing and intentions. It does not verify a retrospective repayment to identified buyers. August 20 earnings call, pages 6 and 8.

The strongest defense of Walmart’s approach is straightforward: lower prices can help households. Dismissing a real discount because it is not a check would make little economic sense. But a discount answers a different question from compensation for an earlier charge.

The first difference is the recipient. A shopper who did not pay the earlier premium may benefit from today’s offer. Someone who paid it may never return to the store. The second difference is measurement. Counting discounted products does not reveal the dollars transferred to consumers. That requires the actual price reduction, quantities purchased and a credible estimate of the price that would otherwise have been charged.

Tomorrow’s discount does not necessarily repay yesterday’s buyer. It may be a valuable commercial policy, a response to competition and a useful deployment of refunded money. None of those possibilities makes it a retrospective refund ledger.

FedEx records what it owes customers

FedEx illustrates why receiving money and earning it are not interchangeable.

As of May 31, 2026, its annual filing reported approximately $800 million in cash refunds received and a $749 million current liability for estimated customer refund obligations associated with the amounts received. A liability is an obligation to pay, not profit. Form 10-K, Note 19, page 112.

The filing also explains that some duties advanced on customers’ behalf had not been collected and had been written off as credit losses. Recoveries of those amounts can reduce bad-debt expense. Subtracting $749 million from an approximate $800 million and labeling the difference “profit retained” would therefore be unjustified. MD&A, “Global Trade Policies,” page 42.

The practical difference is traceability. When a carrier pays a shipment’s duty and then charges an identifiable customer, there is a transaction to match against the government’s refund. A duty paid on a container of goods, later incorporated into months of store prices, does not offer the same direct match.

FedEx says disbursements have begun. Associated Press also reports the start of carrier repayments, drawing in part on the carriers’ statements. That coverage is not an independent audit of completed transfers. FedEx FAQ, accessed September 12. AP reporting.

We did not find a current, aggregate figure for FedEx’s completed customer disbursements as of September 12 in the materials reviewed. A balance-sheet amount dated May 31 is not a live reading of the refund portal. Recording an obligation and executing the payment remain distinct events. FedEx itself also discloses customer litigation over these refunds: a recorded liability does not settle whether repayments are adequate or timely. Form 10-K, Note 19.

Consumer lawsuits test the missing return leg

The initial complaint filed in Oregon on May 8, 2026, in Caldwell et al. v. Nike advances the double-recovery argument directly. Plaintiffs allege that price increases shifted tariff costs to shoppers while Nike sought to recover the same duties from the government. They seek, among other remedies, restitution of alleged overcharges. Those are allegations, not judicial findings. Complaint, Document 1, paragraphs 1–12.

The Justia docket, last retrieved on June 29, 2026, records an order on June 17 consolidating this action with Dunn v. Nike. That record does not establish the complete subsequent procedural history. Docket.

The document also requires a distinction between anticipated refunds and verified cash receipts. Its allegations are not a basis for placing Nike alongside Walmart or Dollar Tree in a table of refunds already collected.

A separate action against Dollar Tree was filed on September 3, 2026, in the Eastern District of Virginia, case 2:26-cv-00932. The public docket, last retrieved on September 4, confirms the filing. It does not establish the merits of the claim. Docket reproduced by Justia.

The legal question cannot be settled by adding together two receipts. The case must address what the customer was told, which charge is being challenged and the basis for requiring its return. An agreed retail price is not necessarily a conditional reimbursement of every cost incurred by the seller. Conversely, a specific representation about a surcharge may be important evidence. The customs ruling does not resolve that separate dispute.

Economic analysis and legal liability therefore need to stay apart. A set of accounts may be compatible with a transfer of income without proving fraud. Corporate communications can be inadequate without guaranteeing that a proposed class action will succeed.

The missing disclosure concerns the final allocation

The documents establish part of the picture. At the two retailers, refunds support current-quarter earnings alongside actual and planned spending or price reductions. At the carrier, a large customer repayment obligation is expressly recorded. These disclosures do not support a national percentage of refunds “returned to Americans.” Their periods, beneficiaries and accounting categories differ. (Walmart; Dollar Tree; FedEx.)

A convincing account of the ultimate distribution would reconcile refunds actually collected, price increases attributable to the duties, the share absorbed in margins, and money or price reductions subsequently passed to buyers. It would follow comparable products, distinguish previous purchasers from new customers and avoid treating every commercial investment as a consumer payment.

Such evidence could defeat an accusation as readily as support it. A business that absorbed most of the shock may simply recover a lost margin. Another may deliver measurable savings to customers. Elsewhere, a household may remain permanently out of pocket even after the business has recovered the duty.

At the end of the chain, repairing the government’s account with the importer does not automatically repair the distribution of costs across the economy. The repayments may be real, lower-price commitments sincere, and the original shoppers still lack identifiable compensation. A balance sheet or a commercial promise cannot, on its own, document that final step.

This article complements our analysis of the funding behind Trump’s proposed tariff dividend: federal revenue and importer refunds involve different accounts and recipients.

Sources

Sources accessed on September 12, 2026. Links throughout the text lead to the underlying documents; the locators below identify the relevant passages.

Customs and incidence. Supreme Court, Learning Resources v. Trump, February 20, 2026, holding and opinion; 19 CFR 141.1, importer liability; CBP, CSMS 68397554, April 20, 2026, CAPE launch; CBP, CSMS 67648307, February 6, 2026, designated recipients; New York Fed research, February 12, 2026, analysis of 2025 import data.

Companies. Walmart’s August 20 results, release, page 1 and official call transcript, pages 6 and 8; Dollar Tree’s Form 10-Q through August 1, “Tariff Refunds” and cost-of-sales discussion; August 27 release, footnote 1 and thirteen-week results; FedEx’s Form 10-K through May 31, Note 19, page 112, and MD&A, page 42; FedEx refund FAQ. Commercial commitments are attributed to the companies, not independently certified as aggregate consumer benefits.

Proceedings and corroboration. Caldwell v. Nike, complaint filed May 8, Document 1, court-stamped copy hosted by a third party; Fennessy v. Dollar Tree, September 3 filing, public procedural record; Associated Press, carrier refunds, reporting on the start of repayments, drawing partly on carrier statements; not treated as an independent audit or exhaustive tally.

Limitations

This is a document-based investigation, without interviews or access to individual transactions. Dollar Tree’s quarterly financial statements are unaudited. Its adjusted reconciliations, and Walmart’s, reflect company definitions. We found no public verification sufficient to quantify, company by company, the share of refunds corresponding to costs previously passed to households.

The initial Nike complaint and the docket retrieved through June 29 were reviewed. For the new Dollar Tree action, the filing record rather than the full complaint was verified. These materials are not an exhaustive review of the litigation dockets. No final decision establishing the alleged liabilities was identified in the records consulted.

The illustrations distinguish a hypothetical exercise, a calculation using rounded disclosures, and a dated balance-sheet position. No total combines the companies or their different periods. Future spending commitments are not treated as completed expenditures. Text and infographics: CC BY 4.0.