Austan Goolsbee on Central Banking as a Data Dog

Episode Summary

Executive Summary: Austin Goolsbee argues that central banking should be guided by data, causality, and robustness to supply shocks rather than rigid rules or accounting identities. The conversation spans inflation, money and velocity, shadow banking and stablecoins, housing affordability, the Fed’s structure, and AI’s likely but uncertain productivity effects.

Main Topics: Inflation, supply shocks, and the limits of rules (Priority: 5/5): Goolsbee says post-pandemic inflation cannot be understood as purely demand-driven; supply-side disruptions, cross-country patterns, and the supply-chain healing in 2023 mattered. He warns against central banking frameworks that overfit past demand-led recessions. Money, velocity, and why M2 became less informative (Priority: 5/5): He argues that M2’s link to inflation weakened because financial innovation and electronic payments changed money velocity. He is skeptical of turning the MV identity into a policy rule or of relying on simplistic monetary growth targeting. Stablecoins, shadow banking, and financial stability (Priority: 5/5): Goolsbee is cautiously supportive of innovation but deeply worried about money-like private deposits without insurance or controls. He frames stablecoins as potentially bank-run-prone shadow banking and emphasizes the need for prudential safeguards. Housing affordability and construction productivity (Priority: 4/5): He treats housing as a long-run relative-price puzzle rather than just a recent bubble, pointing to decades of house-price growth outpacing goods, weak construction productivity, and possible regulatory frictions even outside dense urban areas. How the Fed should be organized and managed (Priority: 4/5): He defends the Federal Reserve System’s regional structure, arguing that multiple reserve banks reduce groupthink and preserve national representation. Internally, he emphasizes delegation, marginal cost, and information-sharing as management tools. AI: real promise, but hallucination and timing risks (Priority: 5/5): Goolsbee thinks AI could become a general-purpose technology with important productivity effects, but warns against over-anticipation, hallucinations, and the belief that larger models or more data automatically solve judgment problems. Fed, CBDCs, and the boundary of the safety net (Priority: 4/5): He is uneasy about retail CBDCs, skeptical that the Fed should become a consumer-facing payments provider, and worries that extending the safety net too far could create new instability and political backlash.

Key Arguments: Central banks should distinguish supply shocks from demand shocks; post-pandemic inflation featured major supply-side components, so rule-based responses can misfire. M2 became less useful because electronic payments, cards, and financial innovation changed velocity; the accounting identity M×V=PY is not itself a policy rule. Cross-country inflation patterns, including high inflation in countries with different stimulus profiles, suggest the inflation episode was not purely U.S. demand driven. Stablecoins backed by Treasuries may still resemble run-prone money-like deposits unless they are insured or tightly restricted. Keeping banks legally special can push activity into shadow banking, but higher capital at large banks still improved resilience in 2023. Housing affordability is a long-run structural issue: house prices have risen much faster than goods prices for decades, and construction productivity appears weak or negative. AI may boost productivity over 10–20 years if it diffuses like earlier general-purpose technologies, but current hallucinations and hype make short-run forecasts unreliable. A multi-bank Fed structure helps prevent monoculture and groupthink; regional presence adds information and legitimacy to monetary policy and supervision.

Data Points: Inflation decline without recession (2023): almost as much as it ever fell in one year without a recession - Goolsbee cites 2023 as evidence against simple demand-driven recession rules. M2 growth: 40% over a few-year period - Used as evidence that money aggregates surged during the pandemic period. U.S. imports share of personal consumption: 10–12% - He uses this to argue the U.S. is primarily domestically driven in inflation discussions. House price inflation before COVID: 3.5%–4% per year for 12 years - He contrasts long-run housing inflation with deflation in goods prices. Goods price inflation before COVID: around -1% per year - Used in the housing relative-price puzzle. Relative housing price increase: 4%–5% per year for a decade and a half - Supports the view that housing affordability problems are long running. Number of Federal Reserve Reserve Banks: 12 - He defends the regional Fed structure and representation. FOMC size: 19 people - Seven governors plus 12 reserve bank participants. Federal Reserve governors: 7 - Political appointees named by the president and confirmed by the Senate. cash in Chicago Fed vault: many tens of billions of dollars - He references the vault to illustrate the scale of physical cash operations. cash in/out operations: hundreds of millions a day - Used to show the scale of cash logistics handled by the Fed. AI productivity surge period: last 2 years - He notes productivity has risen above the pre-COVID trend recently. Assistant professor quiz comparison: 50-question quiz - Hypothetical used to discuss whether O1 Pro could outperform new professors.

Pivotal Quotes: "I’m not one of the birds. I’m in the data dogs." — Austin Goolsbee: He describes his approach to macroeconomics and Fed decision-making as data-driven rather than ideological. "I think we got to be careful over committing." — Austin Goolsbee: His warning against rigid monetary-policy rules that assume all shocks look like past demand shocks. "The Fed will always be the fuddy-duddy of the financial system. And that’s how it should be." — Austin Goolsbee: He explains why central banking should be cautious about stablecoins, CBDCs, and risky innovation.

Implications: Listeners should expect the Fed to remain data-first and cautious about AI, stablecoins, and CBDCs. The transcript suggests future policy will hinge on flexibility, financial stability, and not mistaking identities or hype for economic law.