Auto Parts and Computers: What the General Column Misses in Chapters 84, 85 and 87

Last updated October 12, 2026·Data: USITC HTS + Chapter 99 exports, 2026-10-09

Machinery and electronics are where duty math misleads most often. Buyers see “Free” or “2.5%” in a lookup tool and treat it as the whole story. Then the entry summary arrives and the numbers differ — sometimes by orders of magnitude. The gap between the two figures is not an error; it is the second layer of the schedule, and it has its own logic. This page walks through both layers with every count recomputed from the raw official export, plus the import flows that show where the exposure concentrates.

Layer 1: what the General column says

We parsed the full USITC schedule export — 23,193 administrative lines across 97 chapters — and isolated the three headings buyers of machinery, electronics and vehicle parts touch daily:

Three facts stand out. First, 8471 (computers) and 8542 (integrated circuits) are entirely Free in the base schedule: all 34 and 34 lines respectively print “Free” in the General column. Second, 8708 (motor-vehicle parts) has no high rates at all: among its 1 ad-valorem lines, every single rate found is 2.5%, with the remaining 57 lines (40.1%) Free. Third, this cheapness is not typical of the wider schedule, where only 8,504 of 23,193 lines (36.7%) are duty-free — machinery and electronics are structurally privileged inputs, which is precisely why policy interventions target them through the other layer.

A counting note before anyone re-uses these figures: the schedule publishes ten-digit statistical suffixes under every eight-digit heading, and our 23,193 rows count those administrative lines, not distinct products. Chapter 87's 142 lines resolve to 85 distinct eight-digit headings, chapter 84's computer heading group to 20 and chapter 85's IC heading to 5 — suffixes inherit the parent rate, so the rate distribution is unaffected either way.

Layer 2, part A: where the dollars concentrate

Base-rate arithmetic only matters where goods actually flow. UN Comtrade's public preview gives the US import picture for exactly these sectors:

The concentration is the risk. Motor-vehicle parts arrive above all from Mexico ($31.2B) and China ($11.6B); computers from China ($55.8B) and Mexico ($36.8B); integrated circuits from Malaysia ($14.2B). Together the three sectors represent roughly $254B of annual US imports in the reference year — flowing over a base schedule that prints mostly Free. When policy wants leverage over technology trade, it therefore does not rewrite chapters 84/85/87; it writes new rows somewhere else.

Layer 2, part B: the Chapter 99 rewrite

That somewhere else is Chapter 99. In the same USITC export, 243 Chapter 99 rows carry an additive percentage duty (“base rate + X%”). We screened every row whose scope text cites a subheading inside chapters 84, 85 or 87: 18 measures match, spanning several distinct program families — and their rates are not rounding errors. The heaviest anchor: rows 9903.41.15 and 9903.41.20, each providing a 100% additional rate for automatic data processing machines provided for in subheadings 8471.30 , 8471.41 and kin — the laptop and server classes that sit on otherwise-Free lines. Trailer parts get the same treatment via a note-31 provision that adds 100% on top of the applicable subheading rate for certain 8716.39.00/8716.90.30/8716.90.50 articles.

The mechanism matters as much as the level: Chapter 99 rows apply by their own scope tests. Origin country, timing windows and program exceptions decide whether a given shipment is caught. That is why two importers bringing identical connectors under the identical 10-digit line can lawfully pay different totals — and why a screenshot of the base rate is not research.

Which program families reach these chapters

Aggregating the screened rows by their provision type shows the intervention surface at a glance:

Read the table as a map of political economy rather than arithmetic: information technology sits under a blanket full-value provision, while vehicle-related goods appear under note-specific provisions whose scope tests reference particular truck and trailer classes. None of these rows touch the base-schedule Special column you see in most lookup tools — which is exactly how totals get underestimated.

Worked examples (illustrative arithmetic)

All three scenarios use only numbers present in the datasets above; they illustrate the arithmetic, not legal advice for any specific entry.

1. $100,000 of servers-class ADP machines (8471.30/.41) from a covered origin under 9903.41.15. Base schedule: Free. Add-on: 100%. Duty exposure: $100,000 × 100% = $100,000 — the add-on is the entire bill.
2. $40,000 of steel bumpers (8708.10.30.20) at 2.5%. Base duty: $1,000. If a Chapter 99 provision matching the origin applies at +25% of customs value, that is another $10,000 — eleven times the base layer. The 2.5% headline never told you anything about the total.
3. $20,000 of ICs (8542) from Malaysia. Base: Free, and no screened Chapter 99 row cites 8542 subheadings in the current snapshot — exposure comes out at the base layer only. The point of screening is knowing which case you are in.

A five-minute workflow before you sign a supply contract

1. Resolve the exact 10-digit line from the chapter 84, chapter 85 or chapter 87 index.
2. Note the General rate — but treat it as the floor, not the quote.
3. Screen Chapter 99 provisions citing your subheading (origin- and date-specific).
4. Recompute landed cost with the duty and landed-cost calculator, including add-on layers.
5. Set a quarterly re-check — or let the monitoring tier watch the line for you.

• Open the landed-cost duty calculator.
• Open the HS chapter 84 tariff lookup.
• Open the HS chapter 85 tariff lookup.
• Open the HS chapter 87 tariff lookup.
• Open the how Section 301 China tariffs work line by line.
• Open the monitoring plans and pricing.

Methodology and data freshness

Base-schedule counts derive from the USITC HTS REST export captured 2026-10-09 (23,193 lines). Chapter 99 measures derive from the same REST endpoint's 9900–9999 range, captured 2026-10-10; measure-to-subheading linkage is text-evidence derived as documented in our data build notes. Trade flows are UN Comtrade public-preview values for reference year 2022 (reporter USA, imports, partner-partner detail excluding World aggregate). Counts were recomputed at page-generation time directly from these datasets; nothing is transcribed by hand. Trade-flow previews are indicative rather than definitive final statistics. Nothing here is legal advice — for binding decisions consult your broker or counsel under the governing regulations.

Frequently asked questions