How each figure was built, what was calculated and what was deliberately not calculated. A documented gap is worth more than an estimate.
The figure that carries the report is not an average tariff but the actual value of every European import classified by the regime under which it entered. Eurostat publishes it in the PREFERENCES folder of Comext, at eight-digit granularity, with no registration or key required.
The file mixes detail rows with aggregate rows, and summing both doubles the value exactly. The script takes only the detail and checks that it reconstructs the published aggregate: if it does not match, it fails rather than publishing a bad figure. The total so computed matches, to the million euros, the sum of the 97 chapters of the main Comext dataset, which rules out using a biased subset.
The imported value is official Eurostat data. The percentages (duty-free share, share that would have entered free with no agreement, preference utilisation) are our own ratios over that data, and are labelled as own calculation on every chart. Derived figures are never presented as official.
The euro to dollar conversion uses the annual average of the European Central Bank reference rate, which is published data and is cited as such. Using the wrong year's average shifts the gap between records by billions.
The European and Mexican records do not agree, and not by a little. The cause is documented in both institutions' methodology: Mexico assigns the trading partner by the destination declared on the export document, and the European Union by the country of origin of the goods.
This report uses Eurostat as the governing source for the main series and presents the Mexican record as an explicit contrast. The two figures never share a chart line, because they do not measure the same thing.
The European rapid alert system changed platform and taxonomy around 2020. The historical file and the current system label the same kind of event differently, so they are published separately and no line is ever drawn across that methodological break.
The current system's listing does not expose the hazard category, which is only in each notification's detail. That is why the hazard breakdown is published only for the historical system, and it is stated.
The table of authorised countries was extracted from the regulation's consolidated text through the European Union Publications Office data interface, because the human-facing portal does not respond to automated requests.
The parser respects each cell's position. Flattening the table to text would make empty cells disappear and misalign the columns, with the result of reading that a country is authorised for a category when it is not. An empty cell means not authorised, and that is information, not missing information.
No estimate of exports by state to the European Union is published. That cross does not exist in official sources: the state-level data has no destination country, the data with destination country has no state, and the only source that crosses both measures sales by the exporter's fiscal address, a different concept that the authority itself warns should not be confused.
Nor is any figure published for Mexican establishments approved to export to the European Union, because that data exists only in a European system with no documented programmatic access and could not be obtained reproducibly.