Every 8-digit commodity code India traded above a million dollars last year, both directions, nine years, value and quantity — 7,786 lines, searchable in the page. It also shows you where the data lies: which codes were renumbered, which quantity units silently switched from tonnes to kilogrammes, and which series do not span the period. Those flags are the point.
India’s 8-Digit Trade Book, Searchable
What this is.
- 7,786 HSN-8 commodity lines — every code India traded above $1 million in 2025-26, in either direction.
- Nine years, FY2017-18 to FY2025-26, value and quantity, exports and imports side by side.
- Search by product name or code, sort by size or growth, and read the eight-year shape from the sparkline.
- Every row carries its data-quality flags. That is the part worth reading.
| HS code | Commodity | FY26 $mn | 8-yr change | Series | Unit value | Other side $mn |
|---|
Taking a code further. copy puts the 8-digit code on your clipboard. Paste it into TradeStat’s commodity-wise all-countries export report to get the country breakdown this page cannot show. TradeStat is a POST-only form with no deep links, so a direct URL per code is not possible — copy-and-paste is the shortest honest route.
RECODED the series does not span 2017-18 to 2025-26 — almost always a nomenclature change, not a trade collapse.
UNIT the implied unit value jumps by more than 300× between adjacent years, the tonnes-to-kilogrammes signature. Do not compute a price from this row.
RETIRED DGCI&S marks the commodity name with an asterisk.
Eight-year change is left blank where the first year is under $1 million — a percentage off a rounding-dust base is not a trend. Sparkline spans FY2017-18 to FY2025-26, scaled to each row’s own maximum. The export series has no FY2021-22; that gap is drawn as a break, not interpolated.
Why the flags matter more than the numbers
An 8-digit trade series looks like a time series and mostly is not one. Three things break it, and all three are flagged on every row above.
Renumbering. India revises its national 8-digit sub-divisions regularly. 31.8 per cent of 2025-26 export value sits on codes that did not exist in 2017-18, and on the import side it is worse still at about 39 per cent. Smartphones are the clearest case: the largest single export line in the country, $29.4 billion, on a code that begins in 2022-23. Sort by growth and the top of the list would be nonsense without the flag.
Silent unit switches. The source carries one unit string per code with no year axis, so a change of measure mid-series is unrepresentable. Petroleum crude is labelled KGS throughout while its implied unit value moves from 316.1 to 0.5 dollars per “kilogramme”. Half a dollar a kilo is about $68 a barrel, which is right; $316 a kilogramme is not. The quantities switched from tonnes to kilogrammes and the label did not. Rows where that signature appears are marked, and their unit-value column is left blank rather than filled with a fiction.
Retirement. DGCI&S marks dead codes with an asterisk on the commodity name. It is a weak marker — it catches only about a quarter of codes that actually stop trading, and 46 asterisked codes still carry value — so the flag is shown but the series test is what the tool actually relies on.
How to read a row
| Column | What it is | What it is not |
|---|---|---|
| FY26 $mn | Value in the latest year, US$ million | Not rupee crore; TradeStat’s country tables use crore, these do not |
| 8-yr change | First year to last, on the value series | Meaningless on a RECODED row — that is what the flag is for |
| Series | Sparkline, scaled to the row’s own maximum | Not comparable between rows; each has its own vertical scale |
| Unit value | Change in value per unit of quantity, first year to last | Blank where the unit is unreliable. Machinery counted in pieces is excluded outright — a locomotive line reads 37 then 4,198,750 then 20 dollars per unit |
| Other side | The counterpart flow under the same code | Two-way trade usually means quality differentiation, not substitutability |
Things worth searching for
- smartphone — the largest export line in the country, on a four-year-old code.
- 71023910 — cut diamonds, $23.5 billion to $12.1 billion on a code that never changed. The one unambiguous structural loss in the book.
- 03061790 then 03061720 — “other shrimps” falling 95 per cent while Vannamei shrimp appears from nothing. One product, two codes, zero collapse.
- 85423100 — monolithic digital ICs. $17.5 billion of imports against $160 million of exports, the single largest one-directional dependency in the data.
- 27090010 — crude petroleum, flagged both RECODED and UNIT. India’s biggest import line and the least usable series in the file.
- human hair — two codes, both rising fast on both price and volume. Search the phrase rather than the number.
Sources, and what is missing
The underlying data is the DGCI&S TradeStat EIDB commodity-wise export and import reports at 8-digit HS level, pulled for FY2017-18 to FY2025-26. Values are US$ million. The export series has no FY2021-22; the sparkline draws that as a break rather than joining across it, and any comparison spanning it is a two-year change, not an annual one.
What this dashboard cannot tell you is where a product goes. Destination is the single most important question in choosing something to export — a product selling to forty countries is a different proposition from one selling to one — and commodity-wise data does not carry it. That sits in a separate TradeStat report, country-wise by commodity, and it changes conclusions: dressed human hair looks outstanding here and turns out to send 75.8 per cent of its value to a single destination.
The dataset behind this page is a single JSON file, about two megabytes, served from herrrickshaw.github.io/masaladeutsch/data/trade8.json. It is plain and freely downloadable; if you want to do your own analysis, take it rather than scraping this page.
Sources and caveats
All figures are from DGCI&S commodity-wise export and import data at 8-digit HS level, retrieved from the TradeStat EIDB portal (tradestat.commerce.gov.in/eidb), FY2017-18 to FY2025-26, with FY2025-26 provisional. Values are US$ million as published; no deflation or currency conversion has been applied. The dashboard holds every code with more than $1 million of trade in FY2025-26 on either side — 7,786 of roughly 12,700 codes in the source. Codes below that threshold are omitted for file size and carry a negligible share of total trade. The three quality flags are computed here, not supplied by the source. RECODED means the value series for that direction does not have a non-zero first and last year. UNIT means the implied unit value jumps by more than 300 times between adjacent years on a weight-denominated code, which is the tonnes-to-kilogrammes signature; codes counted in pieces, sets or carats are excluded from that test because large swings there are genuine product-mix effects. RETIRED reflects the asterisk DGCI&S puts on the commodity description. The churn figures quoted — 31.8 per cent of FY2025-26 export value on codes absent in FY2017-18, about 39 per cent on the import side, and the fall to under one per cent at 4-digit level — are from an independent audit of this dataset carried out for this blog. The crude petroleum unit-value example (316.1 to 0.5 dollars per stated kilogramme) is from that audit and is reproducible from the file. Unit-value changes shown in the table are capped: anything outside a factor of eight over eight years is left blank rather than displayed, on the grounds that it is more likely a measurement change than a market. This is a deliberate choice that hides some real price moves in order to avoid publishing many false ones. The destination-concentration example for dressed human hair is from the TradeStat commodity-wise all-countries export report for 2025-26, a separate query not included in this dataset. Nothing here is investment or trade advice. Anyone acting on a single line should pull that code directly from TradeStat and check its own history for a revision before relying on it.
About this article: Researched, written and edited by Umashankar Triplicane Dwarakanathan, with AI research assistance; every figure is meant to trace to the primary source cited. See the Editorial Policy for how sourcing, AI use and corrections work.
.jpg){kind=link}