Parts Pay 62.5 Per Cent. The Finished Flashlight Pays 25.

The United States charges more to import the parts of a flashlight than to import the flashlight. A finished hand-held torch enters under HTS 8513.10.20 at 12.5 per cent, plus the 12.5-point Section 301 forced-labour action that took effect on 24 July 2026: 25 per cent. Its parts enter under 8513.90.20 at 12.5 per cent, plus 25 points of Section 301 List 3, plus 25 points of metals-derivative duty: 62.5 per cent. A lithium-ion cell on the same invoice enters under 8507.60.00 at 53.4 per cent. Parts therefore cost 2.50 times what a finished light costs in duty. The cell costs 2.14 times.

Read against 2025’s Chinese import bill, that classification gap is worth USD 231.75 million a year — 26.9 per cent of the entire US first-sale market, which the US Flashlight Market report sizes at USD 860.0 million in 2025 and USD 1,150.1 million by 2033 at 3.7 per cent. It is decided by which ten-digit code a customs broker selects on one line of an entry summary. On the evidence assembled in that report, it is the single largest variable in the profitability of selling a flashlight in America — larger than the lumen war, larger than the channel shift, larger than the brand.

The schedule, line by line

The finished line is a clean 25 per cent. The parts line carries 37.5 points more than the thing the parts are for. That is the inversion, and it survives every simplification you throw at it. The metals-derivative component is commonly assessed on the aluminium content value rather than the whole customs value, so 62.5 is a ceiling and not a flat rate — but the finished line carries no metals-derivative duty at all, so even at the low end the parts route is at least 25 points worse than the finished route. The gap does not close under any reading of the schedule.

Why the code splits this way

HTS 8513 covers portable electric lamps: 8513.10 for the complete article, 8513.90 for the parts. The structure encodes an assumption that made sense when the schedule was written — that a flashlight is assembled where it is sold, that assembly is the labour, and that the way to protect the assembly is to make the components cheap to bring in and the finished product expensive.

That assumption stopped holding when the light became a battery with an LED on it. The value in a modern flashlight sits in the cell, the driver and the emitter, and those are manufactured in the same places the finished light used to be. The schedule still taxes the components as inputs and the assembly as value-add, which means it now taxes the domestic-assembly route more heavily than the import route. A company that brings in a finished light pays 25 per cent and is done. A company that brings in a body, a bezel, a switch and a cell pays 62.5 per cent on the parts and 53.4 per cent on the cell, and then pays American wages to screw them together.

Sixty economies are covered by the forced-labour action that sets the finished rate. The parts line predates it and was not adjusted to match. That is how a schedule arrives at a 37.5-point premium on the act of assembling the product in America without anyone deciding that it should.

The repeal that did not repeal it

The rate on the finished line has been through three legal bases in eighteen months, and the sequence matters more than any single number in it.

• February 2025. IEEPA duties take the stacked rate on a finished light to 42.5 per cent.
• 20 February 2026. Learning Resources v. Trump strikes the IEEPA layer down.
• Four days later. Section 122 bridges the gap at 10 per cent.
• 24 July 2026. A Section 301 forced-labour action covering 60 economies — 10 or 12.5 per cent depending on the economy — lands at 25 per cent on the finished line.

The net effect is that the effective rate ended up 7.5 points below where it started the year, on entirely different legal footing, with a refund channel still open on the first layer. The schedule was not repealed so much as rebuilt, and the rebuilt version is the one that produces the inversion. Anyone forecasting off the pre-February 2025 numbers is forecasting a tariff regime that no longer exists; anyone forecasting off the post-repeal vacuum is forecasting one that never existed for long.

The evidence that the penalty lands on the assembler

The clearest confirmation is not in the tariff schedule. It is in how the surviving American manufacturers describe themselves. SureFire states on its own site that all of its illumination products are “designed, perfected, and manufactured in the USA with the finest materials and components sourced worldwide.”

Read that sentence twice. It is a statement of a manufacturing model — American assembly, global components — and it is precisely the model the schedule charges a 37.5-point premium to operate. The companies best positioned to make the tariff an advantage are the ones importing finished goods and putting their name on them. The companies worst positioned are the ones doing what the industrial policy says it wants.

That produces a predictable shape in the trade data. Through 2025, imports from Vietnam rose by roughly USD 43 million and imports from Thailand by roughly USD 23.5 million, which is what a schedule looks like when it works on the flow rather than on the factory: the sourcing shifts, the assembly does not come home.

Rechargeables remove the blade, not the bulb

The second force is a business-model change that gets discussed as a technology change. An alkaline flashlight is a razor-and-blade business — a low-margin body and a recurring battery sale that carries the profit. A sealed light with an integrated lithium-ion cell has no blade at all. It is a durable good with a ten-year replacement cycle, sold once, and the integrated-versus-removable cell decision is the most consequential engineering choice in the product.

Which is why the forecast is split the way it is: 1.0 per cent volume against 2.7 per cent price and mix. Units are close to flat — US households already own the lights they need, and a sealed light does not get replaced because a battery died. Value grows on specification, professionalisation and price, not on more objects. That is not a growth story dressed as maturity; it is a category where the volume line and the value line have genuinely separated, and a supplier modelling both at the same rate will be wrong in both directions at once.

Amazon turns the shelf into a price list

The third force is distribution, and it is the one that turns the other two into a visible number. E-commerce accounted for 17.1 per cent of US retail in the second quarter of 2026 and grew 12.2 per cent against 6.7 per cent for retail overall, according to the Census series the Global E-Commerce Market coverage tracks in detail.

A flashlight’s entire specification is four numbers: lumens, runtime, beam distance and IP rating. On a shelf that is packaging design. In a search result it is a comparison table, and a comparison table is the whole product. The consequence is that a listing carrying a 25 per cent landed-cost disadvantage cannot hide it behind brand or shelf position, because the buyer is sorting by number and the number is identical. The channel does not create the tariff problem. It removes the retail margin that used to absorb it.

That is also why the tool platforms matter more than the category’s own brand list. A cordless-tool battery platform sold the work-light function into a system the flashlight brands do not compete in at all — the light becomes an accessory to a battery the buyer already owns. The same logic runs through the Household Cleaning Tools and Supplies market, where the platform owner took the consumable.

Where the number comes from

USD 860.0 million is a first-sale figure, not a retail one: USD 770 million of imports, plus an estimated USD 230 million of domestic factory shipments, minus USD 140 million of exports. The import and export figures are official. The domestic shipment figure is an estimate and the report labels it as the weakest number in the document.

At retail, applying a 1.8 to 2.0 times channel multiplier, the same perimeter is roughly USD 1.55 to 1.72 billion — stated as a memo line rather than a headline, because the multiplier is an assumption and the first-sale figure is not. Compounding 860.0 at 3.7 per cent across eight intervals gives USD 1,150.1 million in 2033; the low case at 3.0 per cent gives USD 1,089.4 million, and that spread is the honest error bar rather than a confidence interval dressed up as one.

The one thing the report will not do is print a unit series. No citable source publishes US flashlight units, and the forecast is value-only with the method note saying so. A volume figure invented to fill a table would have been the easiest number in the document to write and the only one nobody could check.

What would move this beyond its band

Not a tariff rate. A classification event. The 37.5-point gap exists because two similar objects sit on opposite sides of a line drawn in 8513, and the size of the prize — USD 231.75 million a year on the 2025 bill — is large enough that someone will eventually litigate a reclassification, either of integrated-cell lights as something other than portable lamps, or of populated boards and sub-assemblies as parts rather than components.

Three things to watch, in order of consequence: whether CBP issues a ruling that moves an integrated-cell light off 8513.10; whether the metals-derivative assessment base is clarified as content value in a way that makes 62.5 per cent a routine rate rather than a ceiling; and whether the refund channel on the struck-down IEEPA layer produces recovered duty that behaves like a one-off margin event for importers who filed.

The number to watch

Thirty-seven point five. It is the distance between the policy the schedule states and the policy the schedule runs. A tariff code that charges 2.50 times as much to import the parts of a flashlight as to import the flashlight is a policy to import finished flashlights, and it will show up in the trade series long before it shows up in anyone’s forecast — because the forecast is built on the market, and this is a decision about which side of a border the market’s product gets assembled on.

The full sizing, the six chapters covering the competitive landscape, value chain, investment activity, technology, risk and regulatory position, and the year-by-year series are in the US Flashlight Market report. The upstream constraint on the rechargeable shift — the cell — is covered in the graphite and nickel mining reports, which are the anode and the cathode of the same story.