Episode Summary
Executive Summary: The episode centers on three linked macro themes: why inflation and weak real income growth helped shape the 2024 election, how transformative AI could radically raise growth and real interest rates, and why U.S. productivity has stagnated since the 1970s. Zachary Maslisch argues that median households really did experience poorer real income outcomes under Biden once timing, composition, and total income are measured correctly; that truly transformative AI would likely push real rates sharply higher; and that most of the Great Stagnation reflects ideas getting harder to find rather than Baumol’s cost disease.
Main Topics: Inflation, real incomes, and the 2024 election (Priority: 5/5): Maslisch argues voters were not confused: once the comparison starts in 2021 and uses appropriate income measures, most households experienced real income declines or weak growth under Biden, especially after taxes and transfers. Median wage change vs. change in median wage (Priority: 5/5): He distinguishes the change in the median wage from the median change in wages, arguing the latter better captures whether typical workers are better off over time and is more relevant for election analysis. Annual and after-tax income matter more than hourly wages (Priority: 4/5): Hourly wages can mask reductions in hours and compositional shifts; weekly and especially annual post-tax/post-transfer income show a much weaker picture for households during Biden’s term. Transformative AI and real interest rates (Priority: 5/5): Maslisch, Halperin, and Chow frame transformative AI as either explosive productivity growth or existential catastrophe, and show that if growth expectations surge, equilibrium real rates should rise substantially. Asset prices, policy, and market signals under AI (Priority: 4/5): The discussion explores how higher growth expectations could affect stocks, treasuries, housing, and monetary policy, while emphasizing that market prices may already reflect some probability of an AI boom. Decomposing the Great Stagnation (Priority: 4/5): A short research note attributes about one-quarter of the post-1973 TFP slowdown to Baumol’s cost disease and about three-quarters to within-industry slowdown, i.e. ideas becoming harder to find.
Key Arguments: The post-COVID real wage story changes materially depending on the start date; starting in 2020 can overstate gains relative to starting in 2021, which is the relevant election baseline. No income quintile experienced faster real wage growth under Biden than in the 2012-2020 period; lower quintiles did not do better than their historical benchmark, and higher quintiles often did worse. Hourly wage data are misleading because they ignore changes in hours and workforce composition; weekly and annual earnings provide a more complete picture and look worse for Biden. Households care about total income, not just hourly pay; after-tax and after-transfer income better matches voter experience and showed a sharp deterioration after 2020. The median change in wages/income is more informative for welfare and elections than the change in the median, because the latter can rise even when most people are worse off. The Atlanta Fed’s compositionally adjusted earnings data show Biden-era median weekly real earnings growth was far below the Trump and Obama comparison periods. Transformative AI, if it delivers sustained growth well above normal rates, should raise real interest rates through consumption smoothing and higher borrowing demand for investment. Empirically, the paper finds a strong positive relationship between expected growth and real rates, with coefficients near one, suggesting a roughly one-for-one pass-through. Even in a high-growth AI world, stock prices could move ambiguously because higher expected dividends would be discounted at higher real rates. The Great Stagnation is mostly not a Baumol story; the dominant explanation is within-industry slowdown, consistent with the idea that ideas are getting harder to find.
Data Points: Real hourly wage growth by income quintile: No quintile grew faster under Biden than during 2012-2020 - Comparing Biden’s presidency to Obama’s second term and Trump’s term using CPS-based wage measures Lowest-income quintile performance: Did not outperform prior presidential periods in a clear way - Used to test claims that running the economy hot disproportionately helped lower-income workers Compositionally adjusted median weekly real earnings growth: 0.6% annualized during Biden; 1.9% under Trump; 2.0% under Obama’s second term - Atlanta Fed series used as a within-person, adjusted earnings benchmark Post-1973 TFP slowdown: About 1 percentage point annual slowdown vs. post-WWII to 1973 - Great Stagnation decomposition Baumol’s cost disease contribution: About 0.25 percentage points of the 1 percentage point slowdown - Roughly one-quarter of the productivity slowdown Within-industry ideas/getting-harder-to-find contribution: About 0.75 percentage points of the 1 percentage point slowdown - Roughly three-quarters of the productivity slowdown Transformative AI growth threshold: Around 30% annual real GDP growth - Benchmark used to define transformative AI in the paper Simple Ramsey/Euler model implication: Real interest rates could rise to roughly 30% per year if expected growth were 30% - Illustrative extreme-case calculation, not a central forecast Estimated real rate-growth relationship: Coefficient around 1 - Empirical regressions of real rates on expected growth over the past 30 years Potential real-rate implication of 10% growth: About a 10 percentage point increase in real rates - Rule-of-thumb interpretation of the estimated relationship Median growth in after-tax annual income in 2022: Possibly the worst year in the data - Attributed in part to the timing of pandemic stimulus and inflation Inflation-adjusted annual income distribution comparison: Biden distribution lies to the left of Trump distribution - Describes the distribution of after-tax annual income changes
Pivotal Quotes: "In no part of the income distribution did wages grow faster while Biden was president than they did in 2012 through 2020." — David Beckworth summarizing Zachary Maslisch's paper: Opening summary of the article’s main empirical claim "The change in median wages is not what matters. It's the median change in wages that does." — David Beckworth: Introduces the core conceptual distinction in the wage analysis "Transformative AI has been defined as a situation where either real GDP growth over 10 Xs and rises to something like 30% per year, or where this AI kills us all and destroys humanity as we know it." — Zachary Maslisch: Definition of transformative AI used in the AI/interest-rate discussion
Implications: For voters, the transcript argues that inflation-driven declines in real purchasing power were real and politically salient. For markets and policymakers, transformative AI could materially lift growth and rates. For productivity research, the main challenge remains generating new ideas, not just reallocating activity.
About Macro Musings
Hosted by David Beckworth of the Mercatus Center, Macro Musings pulls back the curtain on the important macroeconomic issues of the past, present, and future.