Basil Halperin on Macroeconomic Policy in an Age of Transformative AI

In this episode of Macro Musings, host David Beckworth engages in a deep dive with Basil Halperin, an assistant professor of economics at the University of Virginia, to explore the macroeconomic implications of transformative artificial intelligence. Halperin challenges the widely circulated dystopian narrative from a 2028 'macro memo' predicting mass unemployment and demand collapse due to AI-driven labor displacement. He argues that financial markets, particularly real interest rates, provide a crucial reality check: if transformative AI were imminent, we would already see much higher real interest rates—potentially 30% GDP growth would push long-term rates far above current levels of 2.75%. The absence of such a spike suggests markets are not pricing in a near-term AI revolution. Halperin also unpacks the economic logic behind this, emphasizing that both rapid growth and extinction risk raise real interest rates through consumption smoothing and intertemporal substitution. The conversation then shifts to policy, where Halperin advocates for nominal GDP targeting as an 'eclectically optimal' monetary policy, especially under sticky prices driven by menu costs rather than Calvo frictions. He explains why stabilizing nominal income—rather than inflation—better handles productivity shocks, allowing only affected firms to adjust prices. The discussion extends to fiscal policy, wealth accumulation, and global inequality, warning that AI-driven growth could exacerbate debt burdens and create new crises in emerging markets if interest rates rise without commensurate growth. Despite the complexity, Halperin remains cautiously optimistic about AI’s potential to deliver widespread prosperity, provided institutions adapt and redistribute gains effectively. Key takeaways include: (1) Real interest rates are a powerful early-warning signal for transformative AI—current levels suggest we’re not yet in that regime; (2) Nominal GDP targeting, not inflation targeting, is better suited for economies with menu-cost-driven price stickiness; (3) AI-driven growth may not solve fiscal problems if interest rates rise faster than GDP; (4) Policy must focus on redistribution and institutional adaptation to avoid dystopian outcomes; (5) Even with advanced technology, behavioral frictions like money illusion and sticky information will persist, requiring ongoing policy attention. The episode underscores the importance of grounding macroeconomic narratives in empirical signals and robust models.