How to Read Caviar Production Statistics

China produced an estimated 104,280 tonnes of farmed sturgeon in 2020, according to EUMOFA's February 2023 report. That is fish, not caviar. Confuse the two and a striking statistic becomes a misleading headline.

The FAO and EUMOFA sources cited yield no single, complete official total for world caviar production. They count different things: harvested fish, processed roe, estimated farm output and goods crossing borders. To read the numbers well, first establish the product, activity, unit and reporting year.

Start with the statistical question

What, exactly, are you counting? Roe harvested on farms, salted caviar leaving processors, caviar sold and recorded exports each tell a different story. They are not interchangeable measures of one activity.

Write the intended measure above your spreadsheet before downloading anything. Specify the geography, whether the world, the EU or one country, and establish whether the period is a calendar year. For the industry's scale rather than its measurement, turn to the global overview. The caviar industry hub connects this method to related trade and market topics.

Which sources measure what?

Choose the source that answers your question. Do not add every row into one total. The dates below record the documentation check, not the year of the underlying observations.

FAO FishStat: fish tonnage is not caviar tonnage

A tonne of harvested sturgeon is not a tonne of salted eggs. FAO's aquaculture collection records annual production by country or area, species item, farming environment and fishing area. For aquatic animals, it generally expresses production in tonnes of live weight. A sturgeon series therefore measures harvested fish; its species item may also group species rather than identify one individually.

There is no universal percentage for converting that fish total into caviar. The relationship depends on which fish were harvested and why, including whether they were harvested for meat. Any conversion is a model: state its assumptions rather than presenting the result as an observation.

FAO separately collects production of preserved and processed fishery commodities, alongside commodity trade. To find out whether a processed commodity series identifies sturgeon caviar separately, check the dataset, commodity description and reporting coverage.

EUMOFA: follow the source beneath the report

The source note beneath an EUMOFA table is essential reading. EUMOFA, the European Market Observatory for Fisheries and Aquaculture Products, analyses production, trade and consumption within and outside the EU. Its caviar reports assemble sources, rather than replace them with a single global census.

Its 2021 caviar study uses producer-federation information for EU production and distinguishes customs data from CITES trade data. Its February 2023 sturgeon study also draws on FAO and Eurostat production series.

Read the adjustments, too. The 2021 caviar report filtered customs observations to reduce suspected inclusion of other roe products. That explains why a cleaned analytical series may differ from an unfiltered customs download, even when both carry the same commodity code.

CITES: compare reporters, not just totals

One movement of caviar can generate two accounts: the exporter's and the importer's. Keep them separate. Adding exporter-reported and importer-reported quantities counts accounts of trade, not independent supplies.

The CITES Trade Database records international trade reported by Parties to the wildlife convention. For caviar, select trade term CAV. Retain the taxon, unit, exporter, importer and country of origin, along with source codes, which classify documented origin, and purpose codes, which describe the transaction's purpose.

Reporter differences can reflect reporting dates, permits issued rather than quantities actually traded, missing reports or different product descriptions. A blank unit field means a number of specimens, not kilograms. It does not establish a count of tins: convert it to weight only with supporting records. Re-exports can record the same caviar at another border, so separate direct exports from subsequent movements.

Check completeness before interpreting recent years. Annual reports arrive after the trade year, and late submissions can change the picture. Save the extraction date and inspect reporting coverage before calling a fall a market contraction. See CITES and caviar and the source code guide for the regulatory context.

Customs statistics: select the right product and flow

The right product code is the first filter. UN Comtrade disseminates national goods-trade returns; Eurostat's Comext provides detailed EU trade statistics. In the 2026 EU Combined Nomenclature, CN 1604 31 00 is caviar and CN 1604 32 00 is caviar substitutes. Their corresponding six-digit Harmonised System subheadings are 160431 and 160432.

Select heading 1604 and you bring in other prepared or preserved fish alongside those products. Select the substitutes subheading and you measure other fish roe, not sturgeon caviar. Misclassified declarations can also put other roe into a series labelled caviar. These customs classifications do not provide a complete statement of consumer food naming law.

Use net mass, the weight without packaging, where that is the relevant field. Keep supplementary quantities separate. For the EU, distinguish trade between Member States from trade across the EU's external border. For world exports, remember that UN Comtrade's total exports include re-exports.

Before joining historical downloads, record the classification edition and inspect code changes. Country of dispatch, country of origin and destination are distinct variables. A matching product code alone does not make two country totals comparable.

National statistics and industry estimates

National figures are only as useful as their definitions. In the UK, HMRC's overseas trade statistics provide commodity-level weights and quantities, not a census of farm output. Its methodology distinguishes Great Britain and Northern Ireland collection arrangements, so check for breaks in the series.

In the US, USDA's 2023 Census of Aquaculture includes a caviar category in Table 20, Miscellaneous Aquaculture Production and Sales by Type. Read the units and sales headings literally. Preserve confidentiality symbols: a suppressed entry is not zero.

One movement of caviar can generate two accounts: the exporter's and the importer's.

For China, separate official statistics from sector commentary. FAO GLOBEFISH's 2019 Chinese industry article reports caviar production separately from fish production. It is an industry article, however, rather than a downloadable national statistical table. Identify the underlying authority wherever available.

Producer associations, the World Sturgeon Conservation Society or consultancies may estimate gaps. CIC, for example, estimated China's caviar sales volume in 2025 at 436.8 tonnes in a commissioned estimate published in a 2026 company prospectus. The measure is estimated sales, not an official production census. For any estimate, record the organisation, method, reference year and coverage, and check whether announced capacity has been excluded.

Worked example: fish production is not roe production

China's share of sturgeon fish production makes the distinction clear. EUMOFA's February 2023 report gives China's estimated sturgeon aquaculture production in 2020 as 104,280 tonnes, against approximately 123,476 tonnes worldwide. Divide the former by the latter and the result is about 84.5%, consistent with the report's rounded statement of 84%.

That calculation describes China's share of recorded sturgeon fish production in that year. Calling it China's share of world caviar output silently changes the product being measured.

Even the punctuation matters. The report uses points as thousands separators: its 104.280 tonnes means 104,280 tonnes, not a little over 104 tonnes.

Common mistakes to catch before publication

• Mixing units: Never sum a kilogram series and a count series. Convert only compatible weights, keeping the original unit alongside the conversion.
• Double counting movements: Re-exports record goods crossing another border. Decide whether you are measuring border traffic or output from the originating country.
• Calling capacity production: Announced annual capability is not measured output. Label forecasts, installed capacity and realised production separately.
• Mixing meat and caviar: Both come from sturgeon, so loose headings can mislead. Preserve the dataset's product definition rather than shortening everything to caviar.
• Equating trade year with harvest year: A shipment record dates movement, not necessarily production. Assign a harvest year only when another record establishes it.

How to cite a figure properly

A good citation lets another reader reconstruct the number. Name the organisation, dataset or report, exact table or saved query, reference year, geography, product definition, unit and retrieval date. For trade, add the flow, reporter basis, commodity or CITES term, and treatment of re-exports. State any conversions or exclusions.

For the worked example, cite EUMOFA, Sturgeon meat and other by-products of caviar: production, trade, and consumption in and outside the EU, February 2023, section 2.2 and Table 6, FAO-based sturgeon aquaculture production for 2020, tonnes of fish, checked November 2026. Identify 84.5% as a calculation, not a separately reported caviar statistic.

Keep the raw file and query settings with your notes so the result can be reproduced. When two totals disagree, reconcile their definitions before choosing one. Turn to market trends only once the underlying series is comparable.