Loading workbench page
Fetching primary parquet sources and computing exhibits.
Loading chart...
Fetching primary parquet sources and computing exhibits.
Across 2007-2024, the world market for Electronic integrated circuits: processors and controllers, whether or not combined with memories, converters, logic circuits, amplifiers, clock and timing circuits, or other circuits sits at an HHI of 1,171 on the 0-10,000 scale. The single largest supplier, Chinese Taipei, accounts for about 23% of global exports: if that origin goes dark, the world loses roughly that share of supply overnight. The ranked alternative suppliers, their unit values, tariff exposure into USA, and distance to destination are tabulated below.
The Herfindahl-Hirschman index (Hirschman 1945; Herfindahl 1950) sums squared exporter shares of world trade for this HS6. Higher HHI means fewer sellers dominate. The US Department of Justice and Federal Trade Commission Merger Guidelines (DOJ/FTC, 2023, section 2.1) treat markets with HHI above 1,800 as 'highly concentrated', and HHI between 1,000 and 1,800 as 'moderately concentrated', on the 0-10,000 scale used here. These firm-market benchmarks are descriptive references here: country export shares are not firm shares in a defined antitrust market, and this chart makes no merger-law determination. DOJ/FTC Guideline 1.
Cross-section snapshot: top-10 exporter shares in 2024, annotated with the same country's share 15 years earlier in 2009. The gap between the two reveals rising and falling suppliers, a diagnostic emphasised in Baldwin & Freeman (2022, Annual Review of Economics) on global-value-chain restructuring after the 2018 tariff shock and the pandemic supply break.
For each exporter with at least 1% of the world market in 2024, we compute global share, export unit value (USD per BACI canonical quantity unit; reads as USD per metric ton for product codes where tonnes is the canonical unit, otherwise interpret as USD per the HS-specified unit), the simple-average applied tariff facing that origin in USA (from the WITS-TRAINS preferential-tariff compilation), and the recorded bilateral distance to USA in kilometres. Distance comes from Legacy gravity_bilateral table (exact release provenance unverified); its construction and population weighting are not verified. Missing distances are not estimated.
A summary 'diversification score' ranks these origins for a buyer trying to split away from the dominant supplier. The score combines current share, distance penalty, and a quality tilt anchored on the median exporter's unit value:
score = share × [ 1 / (1 + d_km / 10,000) ] × q q = clip( UV_i / UV_median , 0.5 , 1.5 )
Grossman & Rossi-Hansberg (2008, AER) and Baldwin (2012) on the 'second unbundling' framed sourcing concentration as an unpriced risk in the gains-from-trade calculus. The OECD geodepdatabase (Arjona, Connell & Pisu, 2023, ECO/WKP 1775) operationalises that risk per importer × HS6 as the share of imports coming from the single largest origin. This display uses 40% as a descriptive reference, not a validated risk cutoff. The geodep extract groups the EU as EUN; its aggregate share is not assigned to individual EU members. It does not record HS revision, so cross-revision code equivalence is unverified.
For each of the 2024 top-10 exporters, the compound annual growth rate of HS 854231exports between 2019 and 2024. Rising stars and has-beens separate cleanly in markets with shifting comparative advantage, as in the semiconductor-capacity reshuffling surveyed by Bown (2020, 'How the United States Marched the Semiconductor Industry into Its Trade War with China,' East Asian Economic Review 24(4)).
High HHI says few sellers share the market. It is silent on whether those sellers have held their positions or are constantly reshuffling. Davis, Haltiwanger & Schuh (1996,Job Creation and Destruction, MIT Press, Ch. 2) define a reallocation index as the sum of absolute share changes, a one-sided analogue of gross job flows applied here to HS6 exporter shares: Rt = (1/2) · Σi | si,t − si,t-1 |. The coefficient sits on [0, 1] where 0 is a frozen ranking and 1 is a complete takeover by new exporters within one year. Low values say the same origins hold position; high values say market share churns between countries from one year to the next.
Rauch (1999, Journal of International Economics48(1): 7-35) distinguishes three product types by price-discovery mechanism: goods traded on organised exchanges (oil, wheat, copper), reference-priced goods (base chemicals, steel), and differentiated goods (machinery, apparel). Anderson & van Wincoop (2004, JEL 42(3): 691-751) connect that typology to trade-cost elasticities: homogeneous goods price in world markets, differentiated goods do not. A natural by-product is that exporter concentration should rise with homogeneity, fewer countries can profitably produce the unit good at scale. The figure below tests that: trade-weighted mean HHI within each Rauch class,1996-2024, across all HS6 products. This places the single HS 854231trajectory from Figure 1 against its class-wide benchmark.
HHI squares every exporter's share, so a moving HHI can reflect either a single dominant seller or many mid-sized ones reshuffling. The concentration ratio of the top-5 exporters (CR5) strips that ambiguity: it is simply the combined world-export share held by the five largest suppliers in a given year. Bain (1956, Barriers to New Competition, Harvard UP) introduced CRk as the workhorse concentration statistic for industrial organisation, and Curry & George (1983, Journal of Industrial Economics 31(3): 203-255) survey its use in applied trade and IO. A rising CR5 across 29 years means a tightening oligopoly of origins; a falling CR5 signals genuine entry by new suppliers below the top tier.
HHI and CR5 both weight large players. A complementary read is the Theil entropy of exporter shares Ht = −Σi si,t ln si,t and its exponentiated form exp(Ht), the 'effective number of exporters' (Hill 1973, Ecology 54(2): 427-432; Jost 2006, Oikos113(2): 363-375). A market with five equal-share exporters has effective N = 5 regardless of how many tail-share exporters file customs returns. The gap between the raw exporter count and the effective count is the dominance gap; if a market gains tail entrants but the leader takes more share, raw N rises while effective N falls. Cadot, Carrère & Strauss-Kahn (2011, Review of Economics and Statistics 93(2): 590-605) use this transform on the demand side for export diversification; it reads symmetrically on the supply side here.
Market concentration for HS 854231 in 2024 sits at HHI = 1,171 on the 0-10,000 scale used by the DOJ/FTC Merger Guidelines (2023). That is in the 1,000-1,800 moderately-concentrated band; the single top supplier accounts for 23% of global flows. When a dominant exporter holds this much of a line, a political or policy shock at that origin propagates directly into every downstream importer. The OECD geodep view in Figure 4 shows how that risk concentrates: 4 of the 12 importers with observed geodep shares, among the 15 largest importers, exceed the 40% descriptive reference. Missing observations do not establish low exposure.
WITH t AS ( SELECT year, country_code, SUM(export_value) AS v FROM 'country_year_product_ext/**/*.parquet' WHERE product_code = '854231' AND revision = 'HS07' AND export_value > 0 GROUP BY year, country_code ), s AS (SELECT year, v / SUM(v) OVER (PARTITION BY year) AS share FROM t) SELECT year, SUM(share * share) * 10000 AS hhi FROM s GROUP BY year ORDER BY year;
SELECT c.iso3, SUM(cyp.export_value) / tot.t AS share
FROM 'country_year_product_ext/year=2024/*.parquet' cyp,
(SELECT SUM(export_value) AS t FROM 'country_year_product_ext/year=2024/*.parquet'
WHERE product_code='854231' AND revision='HS07') tot
JOIN 'countries.parquet' c ON c.code = cyp.country_code
WHERE cyp.product_code = '854231' AND cyp.revision='HS07'
GROUP BY c.iso3, tot.t ORDER BY share DESC LIMIT 10;| $4,827,886 |
| n/a |
| 15.3K km |
| 0.032 |
| 5 | USA United States | 5.79% | $319,079 | n/a | 1.2K km | 0.026 |
| 6 | ISR Israel | 2.84% | $6,524,102 | n/a | 9.1K km | 0.022 |
| 7 | CRI Costa Rica | 1.15% | $9,262,343 | n/a | 3.6K km | 0.013 |
| 8 | IRL Ireland | 2.83% | $2,288,689 | n/a | 5.1K km | 0.011 |
| 9 | JPN Japan | 4.26% | $1,967,700 | n/a | 10.9K km | 0.010 |
| 10 | HKG Hong Kong SAR | 1.59% | $3,912,467 | n/a | 13.0K km | 0.007 |
| 11 | PHL Philippines | 1.80% | $1,011,422 | n/a | 13.7K km | 0.004 |
| Unranked | MYS Malaysia | 12.10% | $9,368,999 | n/a | n/a | Unavailable |
| Unranked | VNM Vietnam | 6.88% | $6,611,093 | n/a | n/a | Unavailable |
| Unranked | DEU Germany | 1.56% | $96,801 | n/a | n/a | Unavailable |
Top-supplier share unavailable: Germany (DEU), Ireland (IRL), Czechia (CZE).
-- Per-year exporter shares (same as Figure 1 source), then within-exporter year-to-year diff:
WITH t AS (
SELECT CAST(year AS INTEGER) AS year, country_code, SUM(export_value) AS v
FROM 'country_year_product_ext/**/*.parquet'
WHERE product_code = '854231' AND revision = 'HS07' AND export_value > 0
GROUP BY year, country_code
), s AS (
SELECT year, country_code, v / SUM(v) OVER (PARTITION BY year) AS share FROM t
)
-- The reallocation index is computed in the page body:
-- R_t = 0.5 * sum over countries of |share_{t} - share_{t-1}|.WITH hhi AS (
SELECT year, product_code, country_code,
SUM(export_value) AS v
FROM 'data/parquet/country_year_product/**/*.parquet'
WHERE export_value > 0
GROUP BY year, product_code, country_code
), s AS (
SELECT year, product_code,
v / SUM(v) OVER (PARTITION BY year, product_code) AS share,
SUM(v) OVER (PARTITION BY year, product_code) AS world_v
FROM hhi
), p AS (
SELECT year, product_code,
SUM(share * share) * 10000 AS hhi,
MAX(world_v) AS world_v
FROM s GROUP BY year, product_code
)
SELECT year,
CASE WHEN SUBSTR(product_code,1,2) IN ('09','10',...) THEN 'w'
WHEN SUBSTR(product_code,1,2) IN ('11','13',...) THEN 'r'
ELSE 'n' END AS rauch,
SUM(hhi * world_v) / NULLIF(SUM(world_v), 0) AS mean_hhi
FROM p GROUP BY year, rauch ORDER BY year, rauch;WITH t AS (
SELECT year, country_code, SUM(export_value) AS v
FROM 'country_year_product_ext/**/*.parquet'
WHERE product_code = '854231' AND revision = 'HS07' AND export_value > 0
GROUP BY year, country_code
), ranked AS (
SELECT year, country_code, v,
ROW_NUMBER() OVER (PARTITION BY year ORDER BY v DESC) AS rnk,
SUM(v) OVER (PARTITION BY year) AS tot
FROM t
)
SELECT year, SUM(v) / tot AS cr5
FROM ranked WHERE rnk <= 5 GROUP BY year, tot ORDER BY year;WITH t AS ( SELECT year, country_code, SUM(export_value) AS v FROM 'country_year_product_ext/**/*.parquet' WHERE product_code = '854231' AND revision = 'HS07' AND export_value > 0 GROUP BY year, country_code ), s AS ( SELECT year, v / SUM(v) OVER (PARTITION BY year) AS sh FROM t ) SELECT year, -SUM(sh*LN(sh)) AS theil, EXP(-SUM(sh*LN(sh))) AS eff_n, COUNT(*) AS n_exp FROM s WHERE sh > 0 GROUP BY year ORDER BY year;