Tariff Impact on Prices: What Importers Can Prove From Their Records
Chair Kevin Warsh listed tariffs among recent shocks as he described an FOMC discussion of whether price increases could become broader inflation. Importers can test direct pass-through by linking the tariff amount on an entry to its inventory cohort, later SKU price, and a non-tariff comparison.
Primary lensTrade policy
Sub-topicPolicy monitoring
Evidence base17 records used
Use casePolicy monitoring
Warsh's opening statement put tariffs inside a harder question about inflation. The question is no longer confined to whether a duty raises the cost of an imported product. It is whether a price shock stays close to its source or spreads into prices further away.
The distinction is visible in the official record. The Federal Open Market Committee statement held the federal funds target range at 3.50 to 3.75 percent on July 29. It said inflation remained elevated in part because of supply shocks, including energy. It did not name tariffs or explain why three members preferred a quarter-point increase.
Chair Kevin Warsh went further in his preliminary opening statement. He listed substantial tariff increases among several recent shocks considered in the discussion, alongside supply-chain disruptions, military conflicts, energy disruptions, and investment related to artificial intelligence. He framed the effects on output, employment, and broader inflation as questions. That was a description of the debate, not a committee finding that tariffs caused the current inflation rate.
For an importer, the same distinction separates a defensible price claim from a convenient story. The published and realized effective rates can diverge, while the estimated duty deposit may differ from the liability fixed at liquidation. An entry record alone still does not identify the inventory later sold. A higher SKU price supports a tariff explanation only when these records can be linked.
The propagation question is separate from the tariff rate
Most tariff debates begin with a rate and end with a price index. The records between those points determine whether the comparison means anything.
Federal Reserve staff research on real-time tariff effects illustrates the problem. One line of work estimates the first-round effect of tariffs on core goods or household prices. Another studies the retail price movement of highly exposed products. Reserve Bank surveys ask businesses what they plan to pass through. These estimates can differ without contradicting one another because they use different products, denominators, periods, and points in the supply chain.
The question described in Warsh's opening statement requires a second distinction. A direct price effect remains attached to the product, input, or route exposed to the duty. Propagation begins when the adjustment reaches prices with a more distant connection to that entry. A domestic substitute may reprice. A supplier may reset a contract. A producer may recover a tariffed input through a wider product line. A household may trade down and alter demand in another category.
Those movements cannot be read from one company invoice. Yet company records can show whether the first links exist and where the explanation stops. That makes the firm's evidence useful without asking it to prove an economy-wide result.
The border ledger connects entry to liquidation
The published tariff rate is a legal input. The entry record begins with a calculated duty and an estimated deposit tied to a specific entry line, product classification, origin, customs value, and entry date. The final liability can change at liquidation.
Federal Reserve Board staff have documented a material gap between announced tariff rates and the realized effective tariff rate, measured as Census calculated duties owed divided by customs value. Front-loading, sourcing changes, product mix, origin mix, USMCA claims, foreign-trade zones, bonded storage, and shipment timing can all move that measure. The same headline tariff can therefore produce different entry-level obligations across firms and across months.
The border ledger should preserve the entry number, line number, HTS classification, origin, entered value, duty type, calculated duty or estimated deposit, and any exclusion or preference claimed. It should also preserve later liquidation or correction data. CBP requires an entry summary and an estimated duty deposit, while the amount fixed at liquidation can differ. Dividing the entry-level calculated duty by customs value produces an entry-specific effective rate. It does not produce a retail pass-through rate.
The measurement boundary is easy to miss. Bureau of Labor Statistics methodology excludes tariffs from the transaction price used in import and export price indexes, even though tariff policy can affect price trends indirectly. A customs duty record and an import price index therefore answer different questions.
Inventory starts a second clock
A shipment can clear customs today while the affected units reach customers months later. Inventory decides when the border cost becomes commercially visible.
The useful link is not the month of importation alone. It is the receipt, lot, batch, or cost layer that carries the entry into inventory and then into a sale. The exact identifier depends on the firm's systems and costing method. The objective stays the same. It must be possible to say which sold units bore the duty and which units still reflected pre-tariff stock.
This is why a price increase after a tariff announcement may be too early, too late, or attached to the wrong goods. An importer may have front-loaded inventory before the effective date. A distributor may sell older stock while a new shipment sits in a warehouse. A contract may delay repricing until a quarterly reset. A retailer may hold a price until a promotion ends.
A research note published by the Minneapolis Fed describes inventory depletion and contract resets as reasons tariff effects can remain in the pipeline. The clock can keep running even if the tariff rate does not change. It can also restart when a later tariff action reaches a new entry cohort.
A monthly inflation chart will combine these cohorts. A company should not. Each material tariff change needs its own effective date, entry population, inventory path, and sell-through window. Otherwise the effect of an older measure can be assigned to a new one, or the new cost can disappear inside a blended average.
Contracts decide whether the duty is absorbed
The duty attached to inventory still may not reach the customer price. The importer can absorb it in margin, obtain a supplier concession, change the product, shift origin, reduce services, add a surcharge, or raise the base price.
The contract record shows which path was available. Purchase terms determine whether a supplier shares the cost. Customer agreements determine whether a tariff surcharge is permitted, when a notice can take effect, and whether the charge is tied to a named government action. Price-approval records show whether management recovered one affected SKU or used the event to reset a wider category.
Survey evidence supports treating these choices as variables rather than a fixed pass-through rule. Authors in research published by the Boston Fed found that expected tariff duration was associated with planned pass-through. Authors analyzing the Cleveland Fed's SORCE survey linked greater import exposure with greater expected pass-through. These are business expectations and reported responses, not a committee forecast.
The same commercial record may serve different legal and economic questions. Our analysis of the four-link proof chain for IEEPA refund claims asks whether a customer charge can be tied back to a shipment and CBP payment. An inflation audit continues in the other direction. It follows the entry into the inventory sold, the approved price action, and the actual transaction.
The SKU record tests direct pass-through
The cleanest unit of analysis is usually the lowest product level at which entry exposure, inventory, cost, and price can be reconciled. For many businesses that will be the SKU, customer item, or contract line.
The calculation must use the same unit and cohort on both sides. A unit-level test compares the change in net transaction price per unit with tariff cost per unit for the same sold cohort. The price change should be adjusted for the comparison product and for recorded non-tariff costs, discounts, rebates, and promotions. An aggregate test can instead use the corresponding net-revenue change and total tariff cost for the same sold units. Any liquidation change already known belongs in the tariff amount. Mixing a unit price with a cohort's total tariff cost produces a ratio with no meaning.
A comparison product is the next test. It should have similar demand, seasonality, channel, and non-tariff input exposure but a different tariff exposure. If both products move together, energy, freight, exchange rates, wages, supplier repricing, or a category-wide demand shift may offer a better explanation. If the tariff-exposed product moves when its affected cohort sells through and the comparison does not, the direct pass-through claim becomes stronger.
No comparison will control every difference. The record should state what remains unmatched. It should also separate a temporary surcharge from a durable base-price change. That distinction helps explain whether a later reversal reflects tariff removal, competitive pressure, a new sourcing route, or a pricing decision unrelated to customs.
Federal Reserve Board staff research on 2025 retail prices found gradual adjustment rather than an immediate full reset. Other Federal Reserve staff work estimated first-round effects while expressly excluding retaliation, wages, expectations, and broader channels. These limits are useful. They show why a direct SKU estimate should not be presented as the entire inflation effect.
Intermediate inputs can outlive the first repricing
Final goods and intermediate inputs do not have to produce the same price path. Federal Reserve staff modeling found a larger but shorter inflation response for disruptions to final goods and a smaller, more persistent response for intermediates. The study links some persistence to production-efficiency losses, subject to the limits of its model and cross-country evidence.
For a firm, an intermediate input may enter many finished products over several production runs. Its tariff cost can be diluted in any one item, then remain in the cost base after the first visible price action. A supplier using the same input may reprice later. A domestic producer may respond to the new competitive range even without paying the duty.
The audit trail therefore needs one more relationship for inputs. It should connect the tariffed component to the bill of materials or other usage record for each finished item. It should preserve the supplier price before and after the tariff, any concession, and the timing of each finished-goods price decision.
This is also where sourcing analysis and inflation attribution part company. Our Traverse Analysis publication 20260729, China Plus One Tariff Costs and the Continuity Premium separates the current landed-cost spread from the cost of maintaining restart-ready capacity. That is a sourcing decision. The inflation question begins after the chosen source and duty enter production, then asks where the cost appears and how long it remains.
Reconcile each tariff cohort across the records
A usable result begins with the entry-level tariff amount and later liquidation. It must be matched to the units sold and the price the customer actually paid. None of these records is sufficient on its own.
The join key does not need to be one field in one system. It can be a controlled crosswalk that connects entry line, purchase order, receipt or lot, SKU or bill of materials, sales transaction, and price approval. The crosswalk should retain the dates at each handoff. That preserves the separate clocks for legal effectiveness, customs entry, inventory consumption, and repricing.
The review should begin with a cohort, not a calendar average. Select the entries affected by one tariff action. Trace their duty into inventory. Identify the transactions that depleted the affected layer. Compare their net prices with the prior cohort and with a credible comparison product. Then reconcile the observed change with supplier concessions, freight, currency, promotions, and other input costs recorded for the same window.
The reconciliation may show absorption in margins, partial pass-through on exposed units, or a wider category increase that came before the affected inventory sold. Sometimes the apparent tariff effect disappears once a non-tariff comparison is added.
Management can use the finding in its next price review, forecast, supplier negotiation, and customer explanation. Keeping the record at cohort level also allows a later revision when an entry liquidates, an exclusion applies, or an old inventory estimate proves wrong.
The answer can change without another tariff action
Evidence does not arrive at once. A tariff can remain unchanged while pre-tariff inventory runs out, contracts reset, suppliers reprice, or consumers switch products. A new measure can start another cohort before the earlier one has finished moving through prices.
The policy record will also change. The July FOMC statement did not establish a tariff explanation for the three dissents, and the July 28 to 29 meeting minutes are scheduled for release on August 19, 2026. New consumer-price, producer-price, spending, inventory, and expectations data can alter the broader assessment. New exclusions, court decisions, or corrections to entry-level duty and liquidation data can alter the firm-level result.
The finding should carry an as-of date and revision log. It should identify the entries and sales included, the comparison used, and any unresolved costs. That record lets a firm answer the part of Warsh's propagation question that its own entries and transactions can support.
From reading to review
Run the numbers on your lane.
The duty calculator runs the current stack for any HTS code and origin. A free account opens full tool output, AD/CVD detail, Chapter 98 processing, and available exports.