HS Section deep-dives

The Harmonized System (HS) — administered by the World Customs Organization and the basis for every country’s tariff schedule since 1988 — groups merchandise trade into 21 sections and 99 chapters. Each section is its own economy: chemicals behave nothing like apparel, machinery nothing like minerals. Click any section for thirty years of world trade, exporter concentration, within-section product detail, and an average complexity read.

The HS is one of three standard product classifications. ISIC Rev.4 (UN 2008) assigns activities to industries; NAICS 2022 (US Census & Statistics Canada / INEGI) does the same at North American level; and the UN’s Broad Economic Categories (BEC Rev.5, 2016) map HS lines onto System-of-National-Accounts end-uses (intermediate, capital, consumption). The section grid below sits in HS space; the BEC column below gives the modal end-use for each section, and the ISIC / NAICS correspondence is maintained in the concordance tables published alongside HS 2022.

1. Trajectories by macro grouping

Merchandise trade has trebled in nominal terms since 1995, but the trajectories differ. Machinery and electronics (HS84–85, section 16) drove the post-WTO rise; mineral fuels (HS27, section 5) track oil prices; textiles (HS50–63, section 11) flatten after the 2005 Multi-Fibre Arrangement expiry (Brambilla, Khandelwal & Schott 2010, Journal of International Economics); vehicles (HS86–89, section 17) absorb the 2008–09 Great Trade Collapse (Bems, Johnson & Yi 2013, AER).

2. Growth rates are not parallel

The CAGR ranking decomposes the three-decade climb. Sections with the fastest compound growth are those where global demand outran GDP (machinery, instruments) or where commodity prices lifted values sharply (minerals); the slowest are legacy labour-intensive blocks already large in 1995 (textiles, arms).

2b. Log-linear CAGR with one-sigma regression bands

The simple endpoint CAGR in Figure 2 uses only the first and last year, so a single anomalous year at either end moves the estimate. A log-linear trend CAGR = exp(beta) − 1 from a regression of log(value) on year uses the full time series and comes with a standard error. We plot the point CAGR and the plus/minus one-sigma band from OLS SE(beta), the usual asymptotic trend band used in cross-sector growth comparisons (Hausmann, Hwang & Rodrik 2007, Journal of Economic Growth; Imbs & Wacziarg 2003, AER).

3. Exporter concentration

Some sections are winner-take-most (one or two dominant exporters); others are broadly distributed. We compute the Herfindahl-Hirschman Index (HHI, DOJ/FTC merger guidelines scale, 0–10,000) across country shares of each section’s world exports and plot the top-3 exporter share. Hausmann & Hidalgo’s (2011, Complexity Economics) finding that more complex sections concentrate in fewer countries shows up cleanly here.

3b. Section-level RCA leaders — top-10 concentration

Balassa (1965, Manchester School) defined revealed comparative advantage as RCA_c,s = (X_c,s / X_c,world) / (X_world,s / X_world,world): a country’s share of a section normalised by the section’s share of world trade. RCA > 1 means the country specialises in that section relative to its own export basket. Hausmann & Hidalgo (2011) use this as the building block for the product-space graph. Summing the world-export shares of the top-10 section-RCA leaders gives a finer concentration read than HHI or top-3 alone — it tells you how far down the specialization distribution you have to walk before you cover most of world supply.

4. What the world actually trades

Weight by final value rather than CAGR: which sections dominate the latest snapshot? The answer tracks the HS-section composition that drives the Broad Economic Categories (BEC) balance of intermediates vs capital vs consumption. Section 16 (machinery/electronics) alone is about a quarter of everything crossing borders.

5. Who gained share, who lost it

Shifts in section share between 1995 and the latest year show the structural rebalancing of world merchandise trade. Gainers generally reflect the capital-goods intensification associated with global value chains (Antràs & Chor 2013, Econometrica; Baldwin & Lopez-Gonzalez 2015, The World Economy); losers reflect saturated or substituted markets.

6. Who gained trend growth in the second half

Figure 5 shows the structural change in shares over the full window; here we ask a sharper question: which sections’ trend growth accelerated relative to others after the 2008 GVC-trade-slowdown reset? We split the sample at 2012 (Hoekman 2015, The Global Trade Slowdown, CEPR), compute a log-linear CAGR within each window for each section, rank sections descending by CAGR in each half, and plot the rank change. Positive values mean the section moved up the growth league table; negative means it slipped. This is a Shift of the growth distribution, not of the level distribution — a section can rise in rank while still losing share (e.g. small base, fast acceleration).

7. Tradability density: how much of the HS6 catalogue actually trades

Each HS section carries a different number of HS6 lines in the WCO catalogue: chemicals (HS section 6) holds 760 codes, art (section 21) only 9. The interesting question is how many of those catalogued lines actually moved across a border in 2024: the extensive margin of trade aggregated to section-level (Hummels & Klenow 2005, American Economic Review 95(3): 704–723; Cadot, Carrère & Strauss-Kahn 2011, Review of Economics and Statistics93(2): 590–605). A high activation rate flags a section where the catalogue is fully populated by world trade; a low rate flags sections with many narrow, rarely-traded lines.

Section navigator

Bars are scaled to the largest section for visual comparison; click a section name for the per-section deep-dive page.

Classification crosswalk

The HS is the workbench’s primary axis because tariffs, rules of origin, and customs declarations are all filed on HS codes. For labour-market and productivity work the ISIC Rev.4 (UN Statistics Division 2008) or NAICS 2022 (US Census) activity classifications are more natural; the UN maintains correspondence tables that map HS6 to ISIC and to BEC end-use categories. Trade economists should know all three axes exist and be deliberate about which one the current question lives in: HS for policy (tariffs, RoO), ISIC/NAICS for matching to firm-level production data, and BEC for the intermediate/capital/consumption split used in GVC decomposition (Johnson & Noguera 2012, JIE; Koopman, Wang & Wei 2014,AER).

References

Antràs, P. & Chor, D. (2013). “Organizing the Global Value Chain.” Econometrica 81(6): 2127–2204. Baldwin, R. & Lopez-Gonzalez, J. (2015). “Supply-Chain Trade.” The World Economy 38(11): 1682–1721. Bems, R., Johnson, R. C. & Yi, K.-M. (2013). “The Great Trade Collapse.” American Economic Review 103(3): 298–302. Brambilla, I., Khandelwal, A. K. & Schott, P. K. (2010). “China’s Experience under the Multi-Fiber Arrangement and the Agreement on Textiles and Clothing.” Journal of International Economics 82(2). Hausmann, R. & Hidalgo, C. A. (2011). The Atlas of Economic Complexity. Johnson, R. C. & Noguera, G. (2012). “Accounting for Intermediates.” Journal of International Economics 86(2): 224–236. Koopman, R., Wang, Z. & Wei, S.-J. (2014). “Tracing Value-Added and Double Counting in Gross Exports.” American Economic Review 104(2): 459–494. UN Statistics Division (2008). International Standard Industrial Classification of All Economic Activities, Rev.4. UN Statistics Division (2016). Classification by Broad Economic Categories, Rev.5. US Census / Statistics Canada / INEGI (2022). North American Industry Classification System. WCO (1988/2022). Harmonized Commodity Description and Coding System.

Source: CEPII BACI 202501, HS6 mapped to HS92 section codes via products.parquet. BACI values stored in thousands of USD; multiplied by 1,000 for display. CAGR computed over annual world-export values, nominal, no deflator. Country codes are ISO3 uppercase throughout.