Shockwave

Universal-closure scenario, first-order spillover accounting — not general equilibrium. Trade-diversion gains are not modelled under universal closure: every supplier faces the same wall, so no untaxed alternative exists (target one partner to see diversion). Goods trade only; services and remittance exposure are flagged, not counted. Rankings sample the imputed inputs — the “holds top-3 in N% of input draws” figure is that test, and only that test: it resamples inputs with the model’s structure held fixed, so the omissions listed above contribute nothing to it. The least-affected list carries no such figure; it is an argmax over a restricted set where hits cluster near zero, and is the more fragile of the two rankings.