Abstract: Policymakers are obsessed with how trade flows create dependencies and vulnerabilities. They’ve spent billions of dollars on subsidies and deployed industrial policy to decrease their foreign dependence and remove chokepoint vulnerabilities that their adversaries could exploit. Where are these vulnerabilities? Have they become more concentrated in chokepoints over time? Will industrial policy to reshore production via automation actually decrease dependence? We answer all three questions for the United States using an original measure of dependence constructed at the most fine-grained detail possible, building on the work of Mangini and Chaudoin (2026). The measure is microfounded in structural models of the welfare gains from trade, and it incorporates downstream input-output consequences as well as alternative products and suppliers. We first show how U.S. dependence originates from some products associated with so-called chokepoints and many others that are not. The degree to which U.S. dependence is concentrated in chokepoints has not increased over time. We then use data from the wave of industrial automation in the 1990s and 2000s to assess whether automation decreased dependence. We show that, if anything, automation increased dependence. Industries that automated more quickly were a source of greater dependence in later years. These findings have major implications for where we identify chokepoints and what can be done about them.
Moderator: Lauren Prather