Who measures conversion: testability, blindness and the missing denominator in Brazil's critical-minerals doctrine

Brazil's critical-minerals doctrine names domestic processing as its goal and supplies no measure by which a government could later establish whether it occurred. This paper proposes a falsifiable test of conversion and then asks a prior question: on how many of Brazil's public trade lines can conversion be measured at all? Applying four conditions — sufficient mass, a declared grade, a reference price independent of the exporter, and a classifiable counterparty — to Comex Stat monthly data for 2016–2025, conversion proves testable on one of seven critical-mineral lines. It is the line on which conversion has already failed. Of the six remaining failures, three are caused by nomenclature, two by the absence of a domestic industry, and one by market structure. Only the first family is corrected by a customs code. A second route requires no licensed price series. Destination concentration, relationship duration, rotation and unit-value dispersion separate an industrial offtake from a commodity sale using public data alone, and do so more sharply than the price route. The rare-earth case is examined in detail. In 2025 Brazil exported 677.9 tonnes under NCM 2846.90.90 to China for US$ 11.96 million: 99.83 percent of the national line, declared from a single municipality. Unit value on that destination fell 51 percent as volume rose elevenfold, and the public statistic cannot establish whether price, grade or product changed. Following the 3 September 2026 closing of Serra Verde into USA Rare Earth, that line becomes an intra-group transfer price with no residual market portion. The trade data extract, the derived indicators and the extraction script are included with this deposit.