How does Japan’s trade profile look? Level of exports and imports over thirty years, structural composition, partner concentration, and the position on the economic-complexity frontier.
Japan: merchandise exports and imports, 1996-2024 (incl. 2025 preliminary)
Exports grew from $430.1B in 1996 to $762.9B in 2024, a CAGR of 2.1%. The merchandise balance sits at $40.9B, which is 5% of exports , a simple scale normalisation over the available BACI series. Note that exports are f.o.b. and imports c.i.f., so a meaningful portion of the gap reflects freight and insurance baked into the import valuation rather than a real shortfall in receipts (CIF/FOB spreads typically run 5-10% per UNCTAD/IMF BOP methodology, but country-specific values vary). The more conventional open-economy metric is the current- account balance as % of GDP, reported on the /macro/JPN page. The short 2024-2025 segment is preliminary, from Japan’s own UN Comtrade filing rather than BACI: exports of $738.2B (+4.4% vs 2024) and imports of $755.4B (+1.7%), with both years of the segment on that same as-reported basis. Levels differ from the BACI lines because BACI harmonises mirror flows and re-exports; the 2025 BACI release will supersede these points.
Source: CEPII BACI 202601 (retrieved 2026-06-01), f.o.b. exports and c.i.f. imports, summed across all reporters. 2025 points: UN Comtrade annual, as reported (retrieved 2026-08-18), preliminary.Query
SELECT year, total_exports*1000 AS exports_usd, total_imports*1000 AS imports_usd, trade_balance*1000 AS balance_usd
FROM 'data/parquet/country_year_totals.parquet'
WHERE country_code = 392 ORDER BY year;
The structural change in what a country exports, from agriculture to manufactures to services, or between HS sections, is one of the most studied transitions in development economics (Imbs & Wacziarg 2003; Hausmann, Hwang & Rodrik 2007). The chart below shows the five largest HS sections of Japan’s exports in 2024, plotted back thirty years.
Figure 2
Top-5 HS sections of exports, 1996-2024
Figure 2b
Export basket by HS chapter, 2024 · top-30 $733.0B of $762.9B total
Figure 2c
Import basket by HS chapter, 2024 · top-30 $643.7B of $722.0B total
Who it sells to
Figure 3a
World map of export destinations, 2024
Figure 3b
Top 12 export destinations, 2024
What moved exports between years
The line charts above show levels over time; they do not answer the analyst’s next question: when exports changed between 2023 and 2024, which partners and which products accounted for how much of it? The two bridge charts below answer that directly. Contributors are ordered by absolute dollar change, largest first, and every line outside the top 8 is collapsed into one residual bar so the columns still sum exactly to the total move.
Partners
Figure 3c. Export change by partner, 2023-2024
Exports moved from $787.7B to $762.9B (−$24.8B).
Year totalsIncrease on continuing linesEntry from zeroExit to zero / declineAll other residual
Tables used: bilateral_year + countries.parquet. Contributors are ordered by absolute dollar change, largest first; the residual bar is appended last. After the top 8 named contributors, every remaining line is collapsed into one exact arithmetic residual bar so the bridge reconciles to the total change. Entry bars are partners that rose from zero; exit bars fell to zero, so no growth rate is printed for them.Source: CEPII BACI 202601 (retrieved 2026-06-01), bilateral_year export flows for 2023 and 2024, values scaled from thousands USD to USD.
Products
Figure 3d. Export change by product, 2023-2024
Exports moved from $787.7B to $762.9B (−$24.8B).
Year totalsIncrease on continuing linesEntry from zeroExit to zero / declineAll other residual
Tables used: country_year_product + products.parquet. Contributors are ordered by absolute dollar change, largest first; the residual bar is appended last. After the top 8 named contributors, every remaining line is collapsed into one exact arithmetic residual bar so the bridge reconciles to the total change. Entry bars are HS6 lines that were zero in 2023; exit bars fell to zero in 2024.Source: CEPII BACI 202601 (retrieved 2026-06-01), country_year_product export flows for 2023 and 2024, values scaled from thousands USD to USD.
Economic complexity
The Economic Complexity Index (ECI), building on Hidalgo & Hausmann (2009, PNAS) and formalized as the eigenvalue (second-eigenvector) index in the Atlas of Economic Complexity, ranks countries by the productive knowledge embedded in their export basket. Countries that export many products, and whose products are also exported by few others, score high. ECI is predictive of subsequent income growth and structural transformation; see Hausmann et al. (2014, The Atlas of Economic Complexity, MIT Press) for the full methodology and the comparative country atlas. For a non-linear alternative that ranks fast-diversifying economies differently (China, for instance, sits far higher on fitness than on ECI), compare the Economic Fitness metric of Tacchella et al. (2012).
Figure 4a
Japan: ECI trajectory, 1996-2024
Current ECI: 2.27, ranked #1 of 226 economies in 2024. ECI is zero-centered; positive values mean above-median complexity.
Method: ECI = second eigenvector of the reflections matrix on the BACI RCA≥1 matrix, z-standardized (eigenvalue method of the Atlas of Economic Complexity, Hausmann et al. 2014). Framework: Hidalgo & Hausmann (2009) “The Building Blocks of Economic Complexity,” PNAS 106(26): 10570-10575, whose method of reflections is equivalent to the eigenvector form (Mealy, Farmer & Teytelboym 2019, Science Advances).Figure 4b
Export concentration (HHI) and product count, 1996-2024
2024 HHI = 0.016, equivalent to about 64.5 equally-sized product lines.Japan exports 4,609 distinct HS6 products. Rising HHI or falling effective-N indicates loss of diversification.
Revealed comparative advantage (Balassa 1965) says a country is specialised in a product when its share of that product’s world exports exceeds the country’s share of all world exports. A stronger version asks: which HS6 lines does the country lead the world in? The table below lists the ten largest export lines (by value) where Japan ranks in the world top-5 in 2024, restricted to products with at least US$10M of global trade so tiny niches don’t crowd out economically meaningful positions. This is the “niche leadership” view: products the country is not just diversified in, but competitive at the frontier.
Figure 5
Japan: top 10 HS6 lines with world top-5 rank, 2024
Peer countries by structural profile
Which economies share the closest structural profile to Japan? Each country is placed in a three-dimensional space of economic complexity (ECI), log GDP per capita, and log total exports, each standardised to zero mean and unit variance in 2024. The five nearest neighbours by Euclidean distance in that space are Japan’s closest structural peers, similar on productive-capability, income level, and scale of external trade. This is a trade-specific adaptation of the synthetic-control “donor pool” logic (Abadie, Diamond & Hainmueller 2010).
Figure 6
Japan: five closest structural peers, 2024
Closest structural peer: Rep. of Korea (ECI 2.18, GDP-pc $36,239, exports $764.9B). Bar length encodes peer proximity in the standardised (ECI, log GDP-pc, log exports) space, shorter bars are closer structural matches. This is the peer set to benchmark trade-policy outcomes against, rather than regional or income-group aggregates. The metric matches on capability frontier and scale, not on sector mix, so countries with similar export composition but different ECI (e.g. higher-complexity apparel exporters) will not appear here.
Margins of export growth
Hummels & Klenow (2005, “The variety and quality of a nation’s exports,” American Economic Review95(3): 704-723) introduced the split of a country’s exports into an extensive margin (the set of goods it ships) and an intensive margin (how much of each). In their own measure the extensive margin is nota raw line count: following Feenstra (1994), each category is weighted by its importance in world trade. The figure below applies a simplified, count-based version of that distinction to one country’s export growth over time, so the levels are a proxy, not the Feenstra-weighted index. The two margins still respond to different levers: trade-cost reductions and discovery push the extensive margin (Melitz 2003), while productivity and demand push the intensive.
Figure 7
Margins of export growth, 2014-2024
Over 10 years, total exports grew at 0.0%/yr. Decomposition: HS6 line count moved from 4,721 to 4,609 (extensive margin -0.2%/yr); average value per line moved from $161.0M to $165.5M (intensive margin 0.3%/yr). The intensive margin dominates, which under the extensive/intensive logic indicates growth via deepening sales of existing lines (productivity, scale, or quality).
Method: simplified count-based margin decomposition. The extensive/intensive distinction is from Hummels & Klenow (2005) AER 95(3): 704-723, whose own extensive margin is Feenstra (1994) world-trade-weighted, not a line count. Here, extensive margin = CAGR of distinct HS6 export lines; intensive margin = CAGR of (total exports / line count); total CAGR ≈ extensive + intensive (log-linear approximation).
Export basket on the complexity frontier
Figure 8
Japan: export value vs. product complexity (PCI), 2024
Related
Value-chain position, domestic vs foreign value added in Japan’s exports, VAX ratio, and upstream/downstream index (OECD TiVA)
Economic Fitness, the non-linear Tacchella-Pietronero complexity ranking, and how it diverges from ECI
Research index, analytical pieces grounded in BACI flows and gravity covariates
Sector monitor, quarterly deep dives on 12 HS-defined sectors
In 2024, the top three HS sections account for 71% of merchandise exports that map to a HS96 section, a rough measure of sectoral concentration at the coarsest classification level. Shares here are over BACI exports that carry a valid HS96 section mapping, so the five-section series does not sum to 100% of total exports: some HS6 codes (newer HS revisions, non-standard codes) fall outside the 21-section mapping. Note that the top-5 sections are selected by 2024 share, so a section that was dominant in 1996 but has since fallen out of the top 5 will not appear on the chart.
Source: CEPII BACI 202601 (retrieved 2026-06-01) crossed with HS96 section codes. Shares are over the HS6-mapped universe; HS6 codes without an HS92 section fall outside the 21-section totals.
Cell area is proportional to 2024 export value. The treemap shows the top 30 HS chapters (HS2 level) of Japan’s exports; cells are colour-coded by HS section so structurally related chapters share a hue (e.g. minerals, machinery, textiles). The largest chapter, HS 87 , Vehicles; other than railway or tramway rolling stock, and parts and accessories thereof, represents 22.5% of the top-30chapter total shown. Compared to Figure 2’s five-section trajectory, this view exposes which specific chapters within the dominant sections drive the overall composition.
Source: CEPII BACI 202601 (retrieved 2026-06-01), f.o.b. exports of HS6 codes aggregated to HS2 chapters, 2024. Restricted to top 30 chapters by value.Query
SELECT SUBSTRING(product_code, 1, 2) AS chapter_code,
ANY_VALUE(p.chapter) AS chapter_name,
p.section,
SUM(cyp.export_value) * 1000 AS value_usd
FROM 'data/parquet/country_year_product/year=2024/*.parquet' cyp
JOIN 'data/parquet/products.parquet' p ON p.code = cyp.product_code
WHERE cyp.country_code = 392 AND cyp.export_value > 0 AND p.section IS NOT NULL
GROUP BY chapter_code, p.section
ORDER BY value_usd DESC LIMIT 30;
Cell area is proportional to 2024 import value. The treemap shows the top 30 HS chapters (HS2 level) of Japan’s imports; cells are colour-coded by HS section so structurally related chapters share a hue. The largest import chapter, HS 27 , Mineral fuels, mineral oils and products of their distillation; bituminous substances; mineral waxes, represents 25.7% of the top-30 chapter total shown. Read alongside Figure 2b to compare what Japan sells abroad against what it buys in.
Source: CEPII BACI 202601 (retrieved 2026-06-01), c.i.f. imports of HS6 codes aggregated to HS2 chapters, 2024. Restricted to top 30 chapters by value.Query
SELECT SUBSTRING(product_code, 1, 2) AS chapter_code,
ANY_VALUE(p.chapter) AS chapter_name,
p.section,
SUM(cyp.import_value) * 1000 AS value_usd
FROM 'data/parquet/country_year_product/year=2024/*.parquet' cyp
JOIN 'data/parquet/products.parquet' p ON p.code = cyp.product_code
WHERE cyp.country_code = 392 AND cyp.import_value > 0 AND p.section IS NOT NULL
GROUP BY chapter_code, p.section
ORDER BY value_usd DESC LIMIT 30;
Each country is shaded by the bilateral value of Japan’s exports to it in 2024, binned into quintiles. Japan itself is tinted in amber as the origin; white indicates destinations with no recorded bilateral flow. The map makes regional dependency structures visible, neighbours, former colonial ties, bloc partners, in ways a top-list ordering hides.
Partner concentration measures market exposure. The single largest destination absorbs 19% of Japan’s exports; the top 3 together take 43%. Concentrated partner bases make bilateral shocks (trade wars, recessions) first-order.
Source: CEPII BACI 202601 (retrieved 2026-06-01), bilateral exports 2024. Shares are over total exports.
Japan holds a top-5 world-export position in 10of its largest export lines (HS6 products with ≥ US$10M world trade, 2024). The leading niche is HS 870390 , Vehicles: for transport of persons (other than those of head , where it ranks #2 globally with 13.9% of world exports. Dominant niches are the revealed-comparative-advantage anchors a trade-shock or industrial- policy analysis should start from.
Source: CEPII BACI 202601 (retrieved 2026-06-01), HS6 exporter rankings, 2024. Restricted to products with ≥ US$10M world trade. Method: Balassa (1965) RCA taken to its rank-based extreme, top-5 world rank is a stringent specialisation test.Query
WITH prod_year AS (
SELECT product_code, country_code, SUM(export_value) AS v
FROM 'data/parquet/country_year_product/year=2024/*.parquet'
WHERE export_value > 0 GROUP BY product_code, country_code
), ranked AS (
SELECT product_code, country_code, v,
RANK() OVER (PARTITION BY product_code ORDER BY v DESC) AS world_rank
FROM prod_year
WHERE product_code IN (SELECT product_code FROM prod_year GROUP BY product_code HAVING SUM(v)*1000 >= 1e7)
)
SELECT product_code, v*1000 AS value_usd, world_rank
FROM ranked WHERE country_code = 392 AND world_rank <= 5
ORDER BY v DESC LIMIT 10;
Method: z-score each dimension (ECI, log GDP-pc, log total exports) across the universe of countries with all three observations in the latest year; rank by Euclidean distance. Abadie, Diamond & Hainmueller (2010) “Synthetic Control Methods for Comparative Case Studies,” JASA 105(490): 493-505.
Query
SELECT h.year, h.num_products, t.total_exports*1000 AS exports_usd
FROM 'data/parquet/country_hhi.parquet' h
JOIN 'data/parquet/country_year_totals.parquet' t USING(country_code, year)
WHERE h.country_code = 392 AND h.year IN (2014, 2024);
Each dot is one HS6 product in Japan’s basket with more than US$1M in exports in 2024 (long-tail marginal exports below that floor are filtered out so the scatter is legible; the filter drops many HS6 codes for small economies and few for large ones). Horizontal = export value (log), vertical = Product Complexity Index (PCI). Dots upper-right are high-value, high-complexity products (machinery, precision instruments). Lower-left products carry less productive knowledge per dollar. A basket shifted toward the upper-right correlates with higher ECI and higher income.
Method: PCI = product loadings from the second-eigenvector complexity decomposition of the country-product RCA matrix (eigenvalue method of the Atlas of Economic Complexity, Hausmann et al. 2014; framework from Hidalgo & Hausmann 2009, PNAS; eigenvector form equivalent to the method of reflections, Mealy et al. 2019). Basket restricted to HS6 exports > US$1M in the latest year.
where Japan is exposed
Telephones for cellular networks or for other wireless networks (851712): imports 86% from China. Ranked by import dependency (GeoDep 2022). Single-supplier concentration is the exposure. Model a supply shock →
Iron ores and concentrates: non-agglomerated (260111): imports 64% from Australia. Ranked by import dependency (GeoDep 2022). Single-supplier concentration is the exposure. Model a supply shock →
Data processing machines: portable, digital and automatic, weighing not more than 10kg, consisting of at least a central processing unit, a keyboard and a display (847130): imports 99% from China. Ranked by import dependency (GeoDep 2022). Single-supplier concentration is the exposure. Model a supply shock →
Vaccines: for human medicine (300220): imports 99% from European Union (EU27). Ranked by import dependency (GeoDep 2022). Single-supplier concentration is the exposure.
Tricycles, scooters, pedal cars and similar wheeled toys: dolls' carriages: dolls: other toys: reduced-size (scale) models and similar recreational models, working or not: puzzles of all kinds (950300): imports 82% from China. Ranked by import dependency (GeoDep 2022). Single-supplier concentration is the exposure. Model a supply shock →
where Japan could grow
Rubber: synthetic, acrylonitrile-butadiene rubber (NBR), latex, in primary forms or in plates, sheets or strip (400251): proximity 0.51 to current capabilities, complexity PCI +1.87, world market $1.0bn. Ranked by density × complexity (product space 2024). See product →
Photographic film: for colour photography (polychrome), in rolls, sensitised, unexposed, without sprocket holes, of a width exceeding 610mm and of a length exceeding 200m (370241): proximity 0.54 to current capabilities, complexity PCI +1.71, world market $21.3m. Ranked by density × complexity (product space 2024). See product →
Acids: aromatic polycarboxylic acids: dinonyl or didecyl orthophthalates (291733): proximity 0.45 to current capabilities, complexity PCI +2.04, world market $1.1bn. Ranked by density × complexity (product space 2024). See product →
Complex cyanides (283720): proximity 0.42 to current capabilities, complexity PCI +2.09, world market $70.6m. Ranked by density × complexity (product space 2024). See product →
Acids: saturated acyclic monocarboxylic acids: vinyl acetate (291532): proximity 0.51 to current capabilities, complexity PCI +1.68, world market $1.6bn. Ranked by density × complexity (product space 2024). See product →
Japan trade-riskGlobal risk rank #98 of 200 (board)
supplier concentration83.7%
top chokepointStrait of Malacca (3.8%)
price stress 12m9.6%
price coverage14.2%
Dependency diagnosis
Telephones for cellular networks or for other wireless networks(851712): you import 85.6% from China. UNKNOWN World concentration unavailable for this product code.
Iron ores and concentrates: non-agglomerated(260111): you import 64.2% from Australia. STRUCTURAL Structural scarcity, world export HHI 0.41. Switching suppliers will not fix this; the world itself is concentrated (top alternatives: BRA 20.3%, ZAF 4.5%, CAN 3.4%, IND 1.6%). Policy: stockpile, substitute the input, or invest in allied/domestic production.
Data processing machines: portable, digital and automatic, weighing not more than 10kg, consisting of at least a central processing unit, a keyboard and a display(847130): you import 98.6% from China. STRUCTURAL Structural scarcity, world export HHI 0.51. Switching suppliers will not fix this; the world itself is concentrated (top alternatives: VNM 12.8%, NLD 3.1%, DEU 1.9%, S19 1.8%). Policy: stockpile, substitute the input, or invest in allied/domestic production.
Vaccines: for human medicine(300220): you import 99.4% from Unknown. MODERATE Moderate, world HHI 0.16; some concentration. Alternatives: BEL 28.6%, IRL 19.7%, USA 13.2%, FRA 8.4%.
Tricycles, scooters, pedal cars and similar wheeled toys: dolls' carriages: dolls: other toys: reduced-size (scale) models and similar recreational models, working or not: puzzles of all kinds(950300): you import 82.5% from China. UNKNOWN World concentration unavailable for this product code.
Textiles: made up articles (including dress patterns), n.e.s. in chapter 63, n.e.s. in heading no. 6307(630790): you import 85.4% from China. STRUCTURAL Structural scarcity, world export HHI 0.30. Switching suppliers will not fix this; the world itself is concentrated (top alternatives: MEX 4.6%, DEU 4.5%, VNM 4.3%, USA 3.4%). Policy: stockpile, substitute the input, or invest in allied/domestic production.
Air conditioning machines: comprising a motor-driven fan and elements for changing the temperature and humidity, window or wall types, self-contained(841510): you import 90.1% from China. STRUCTURAL Structural scarcity, world export HHI 0.47. Switching suppliers will not fix this; the world itself is concentrated (top alternatives: THA 17.6%, MYS 3.1%, KOR 1.7%, ITA 1.3%). Policy: stockpile, substitute the input, or invest in allied/domestic production.
Footwear: n.e.s. in heading no. 6402, (other than just covering the ankle), with outer soles and uppers of rubber or plastics(640299): you import 66.3% from China. STRUCTURAL Structural scarcity, world export HHI 0.32. Switching suppliers will not fix this; the world itself is concentrated (top alternatives: VNM 17.8%, IDN 5.9%, DEU 4.1%, ITA 2.2%). Policy: stockpile, substitute the input, or invest in allied/domestic production.
Compound exposures
These flows carry multiple simultaneous stressors on the same product; the averaged risk score understates them.
Wrist-watches: whether or not incorporating a stop-watch facility, with automatic winding from Switzerland: single-source (96.5%) + transits Strait of Malacca.
Vegetable oils: palm oil and its fractions, other than crude, whether or not refined, but not chemically modified from Malaysia: single-source (79.9%) + price +15% over 12m.
Wrist-watches: (not electrically operated), automatic winding, whether or not incorporating a stop-watch facility, case of precious metal or of metal clad with precious metal from Switzerland: single-source (95.2%) + transits Strait of Malacca.
Nuts, edible: almonds, fresh or dried, shelled from USA: single-source (97.4%) + transits Strait of Malacca.
GeoDep 2022, world shares latest BACI year, Pink Sheet 2026M04. No real-time signal is implied.