Value turns on evidence no one publishes before commitment

In October 2026 a headline reported that North Carolina had put up $120 million to beat Ohio for a small rare-earth magnet startup, Vulcan Elements. FalconBridge used the award to test a narrower and more useful question: can a transparent scorecard judge, before public money is committed, whether a strategic-manufacturing incentive is likely to deliver risk-adjusted public value? The study tests a decision method; it is not an audit of the award or a review of the company. The headline merges three layers of government. The state’s own commitment is a Job Development Investment Grant of up to $17,584,500 to the company over 12 years, paid only after annual verification approved on 18 November 2025, eleven months before the article. About $94.7 million — some 79 per cent of the headline — is county and town property-tax grants over 15 years, and about $8 million is training and road support. On the reported terms (the local agreement text is not yet public), about 93–94 per cent of the package is paid only on performance or as a share of new taxes. Applied to this award, the FalconBridge Strategic Incentive Scorecard — nine scored dimensions, an explicit “unknown” grade, confidence levels and three gates — establishes with confidence that the public’s downside is bounded and that the price is in line with comparable awards. It establishes the two terms that dominate value, whether the money changed the location decision and whether the recipient can deliver, only at low-to-medium confidence. The evidence that would settle them sits with the competing state, the awarding committee in closed session, the local governments, the federal lender and private investors, none of whom publish it. At commitment the profile reads “safe but likely poor value” on the public evidence: the exposure gate passed clearly, the capability gate failed on the public record, and the attribution gate failed on the state layer. Eleven months on, a securities filing recording about $430.7 million of equity sold and a finalised federal CHIPS award lift capability to a marginal pass; the remaining concern is confined to the low-cost state layer, at low confidence. The state’s commitment, about $17,600 per announced job, is the lowest in a 25-award comparator set and the headline sits at the median, so the “new baseline” claim is not supported for the state. Five external-data gates remain open: the local agreement text, Ohio’s offer, the federal loan’s financial close, the committee’s but-for analysis and the grant schedule. The test does not show whether the award is good or bad policy, or whether the company will succeed; scores are judgement against published rubrics, not findings about any party’s merit. States could consider decomposing strategic awards by layer and instrument type, recording the counterfactual reasoning behind but-for findings, and adding reporting covenants for private early-stage recipients. Companies could negotiate timing terms as carefully as amounts, and investors could read public support as a signal to test rather than a guarantee. Before the next commitment, the named missing evidence is where diligence could most usefully be directed.