Comparing import concentration without confusing products and suppliers

An economy can have a diversified import basket while relying on one supplier for an essential component. It can also import mostly one commodity from several independent sources. Product concentration and supplier concentration describe different distributions. Combining them into an unexplained index can conceal the dependency that a business or policymaker needs to assess.

Comparisons among the United States, China, and the euro area introduce additional choices. Product codes have different levels of detail. Country of origin differs from country of shipment. A regional aggregate requires a decision about internal trade. Values depend on prices, quantities, and reporting conventions. A concentration estimate is meaningful only after those boundaries are fixed.

The analysis below uses official compilation documents to identify these choices, followed by original numerical examples. It does not present a current ranking of the three economies. Such a ranking would require harmonized observations for a specified period, a fixed membership perimeter, and consistent treatment of all residual and confidential categories. A precise number generated from inconsistent inputs would remain an inconsistent comparison.

The U.S. Census Bureau's trade-program guide generally assigns imports to the country where the merchandise was grown, mined, or manufactured under customs rules. When origin cannot be determined, the guide specifies attribution to shipment country. Thus, even a country-of-origin series requires attention to exceptions and unknown information. A supplier's shipping address is not automatically the import's reported origin. U.S. Census Bureau, international trade statistical program, import country of origin.

China's customs-statistics regulation, in the version published by the National Bureau of Statistics in August 2023, separately identifies origin, country of departure, and domestic destination for imports. The NBS's January 2025 explanation describes public trade queries by partner, trade mode, region, and commodity detail at two, four, six, and eight digits. These are distinct possible tabulations of the same trading system, not interchangeable definitions of dependence. NBS, customs-statistics regulation, articles 6 and 9; NBS, compilation and publication of import and export totals.

Eurostat's metadata distinguishes origin for extra-EU imports from consignment country for intra-EU imports. It also distinguishes the reporting territory and the physical movement being recorded. An analyst constructing a euro-area aggregate must therefore examine both the chosen region and the underlying partner variables, rather than assume that every member's reported import partner denotes ultimate production origin. Eurostat, international trade in goods metadata.

The euro area and the European Union are different perimeters. A member importing from an EU country outside the euro area creates an external euro-area transaction but an intra-EU observation. The partner concept available in that record can differ from the origin concept used for a non-EU source. Simply selecting all trade labeled extra-EU would not reconstruct all extra-euro-area imports.

This is a substantive measurement issue. A shipment routed through a regional distribution center may be recorded under a different partner concept from an otherwise similar direct shipment. The resulting change can reflect routing rather than relocation of production. A vulnerability assessment needs to specify whether it seeks production origin, dispatch location, contractual counterparty, or the location of final controlling ownership.

For a given product, define the supplier Herfindahl index as the sum of squared supplier shares. Shares are expressed as fractions, so the index ranges toward one as supply becomes concentrated in one source. The reciprocal can be read as the equivalent number of equal-sized suppliers under this measure, but it is not a count of technically interchangeable producers.

Suppose an importing economy buys two products of equal value. Product Alpha comes entirely from supplier A; product Beta comes entirely from supplier B. Across all imports, A and B each account for half. The aggregate supplier index is therefore 0.5 squared plus 0.5 squared, or 0.50. Within each product the supplier index is 1.00. The import-value-weighted average of product-specific indices is also 1.00.

Now compare another hypothetical economy that imports the same products and total values but buys half of each product from A and half from B. Its pooled supplier index is also 0.50. Its within-product indices are each 0.50. The aggregate statistics are identical, while the product-level sourcing structures differ materially.

This difference does not prove that the second economy is resilient. The suppliers may use the same upstream facility, depend on one transport route, or lack spare capacity. It demonstrates that the pooled index cannot establish product-level diversification. Where complementary inputs are required, the interruption of a small single-sourced component may constrain output far more than its import expenditure weight suggests.

A product-concentration index would ask another question: how import spending is distributed across Alpha and Beta, irrespective of source. Both economies in the example have equal product values and the same product-concentration index. That information describes the basket, but does not resolve whether either product has an alternative supplier.

Consider a hypothetical component worth one hundred. Its production origin is country A. Initially, all of it is shipped directly from A. Later, sixty is shipped through a hub B and forty remains direct. Assume that the hub does not undertake a transformation that changes origin, that the product is unchanged, and that there are no valuation differences.

The origin-based distribution remains one hundred percent A, with an index of 1.00. The dispatch-based distribution becomes sixty percent B and forty percent A, with an index of 0.52. A calculation using dispatch would show a large apparent diversification, although production remains entirely in A. This is a statistical-routing example, rather than an allegation about any actual trading relationship or an assessment of origin compliance.

The reverse can also occur. Goods from several production origins can be consolidated in one dispatch location. A high shipment-country concentration may then reveal dependence on a logistics route while overstating concentration of production. Both forms of concentration may matter, but they require separate scenarios and mitigation measures.

For the euro area, the mixed availability of origin and consignment information across trade systems makes this distinction especially important. For the United States and China, an analyst should still verify the selected published series and treatment of unknown origin. The existence of multiple fields in a declaration does not mean every public table uses the preferred field or provides all possible cross-classifications.

Detailed commodity codes distinguish products more finely, but changing the number of codes changes the distribution being measured. Suppose a broad product category accounts for sixty percent of imports and another accounts for forty percent. The product index is 0.52. If the sixty-percent category is split into two equal thirty-percent codes while the underlying transactions remain unchanged, the index becomes 0.34. The lower value results entirely from classification detail.

That arithmetic does not invalidate detailed analysis. It means that comparisons need a common product partition. A time series should account for code revisions, mergers, and splits. Where a one-to-many concordance is required, allocation assumptions should be stated rather than treating a mechanical code match as direct evidence of a changing economic structure.

Supplier concentration within a broad category can also conceal specialization. One country may supply a high-performance component and another a standard variant. They share a code but are not substitutes for a buyer's application. Greater detail helps, yet customs classifications remain designed for trade administration and statistics, rather than a complete description of engineering compatibility.

A procurement assessment should therefore link statistical categories to specifications, qualification requirements, and actual purchase contracts. A national index can identify where further examination is warranted. It cannot determine the lead time or cost of replacing a particular supplier.

Current-value shares can move when commodity prices change even if quantities and suppliers remain constant. A rise in one imported input's price increases its weight in the total basket and can alter aggregate concentration measures. That movement is economically relevant for expenditure exposure, but it does not necessarily describe a larger physical dependence.

Quantity measures create their own limits. Weight is often unsuitable for comparing technically different products. A kilogram of one input cannot be treated as equivalent to a kilogram of another. Within a narrowly defined product, a quantity series may be useful if units and quality are stable. Across products, the analyst must explain the aggregation rather than replacing money with an unexamined physical total.

Import expenditure is also an imperfect measure of production criticality. A low-cost control system can stop a high-value plant. An expensive commodity may be widely traded and substitutable. A useful vulnerability framework therefore combines concentration with inventories, substitution options, replacement lead times, and the role of the input in downstream production.

Domestic supply belongs in that assessment. A product can appear highly concentrated among imports but have substantial domestic production. Conversely, a low import share in national expenditure can conceal dependence at a particular firm with no domestic supplier. The relevant denominator depends on whether the question concerns imported sourcing, total consumption, or the inputs of a defined production network.

Correlation among suppliers changes the interpretation of the same index. Two equally sized suppliers give a supplier index of one half, whether their production is located on separate continents or in adjacent facilities dependent on one electricity substation. The numerical distribution is identical, but the consequence of a local outage differs. Conversely, geographically distant suppliers can share a specialized upstream input or software system. A risk assessment needs a scenario-specific map of those common dependencies. Concentration can organize the initial search, but a probability or loss estimate requires additional assumptions about disruption, substitution, inventories, and recovery. It should not be obtained by multiplying a concentration index by total imports and presenting the result as expected economic damage.

A euro-area aggregate should remove internal transactions from the chosen external-import perimeter. Summing all member imports without that adjustment counts trade within the area as external exposure. It can also count goods more than once when they move between members after entering the region. The trade perimeter must be fixed before calculating country shares or comparing with a single national economy.

Corporate ownership adds a further layer. Two exporting establishments in different countries may belong to one group and depend on one decision center. Conversely, several independent producers in a single country may face a common geographic shock but different commercial risks. Country concentration and corporate concentration are separate measures, neither of which can replace the other.

A transparent comparison should consequently publish its product detail, partner concept, regional membership, period, valuation basis, and formula. It should show pooled and within-product results where they answer different questions, and test whether the conclusion survives reasonable changes in classification. A result that depends entirely on one arbitrary aggregation should be reported as sensitive to that choice.

The most useful outcome is an explanation of the dependency being measured. A lower index can signal diversified production, alternative shipping routes, a different product mix, or merely revised codes. Identifying which of those occurred turns concentration statistics into evidence that can inform industrial and commercial decisions.

For the distinction between border flows and domestic production, see trading hubs and external exposure. For production-network accounting, see changes in trade geography.