310. Are We Running Out of Ideas?

Episode Summary

Executive Summary: The episode argues that productivity growth is increasingly constrained because good ideas are harder and more expensive to find. Using evidence from semiconductors, agriculture, medicine, and firm-level R&D, economists Bloom and Van Reenen claim innovation remains powerful but yields are falling. The second half tests that pessimism against emerging technologies and policy ideas that could revive growth.

Main Topics: Ideas as the engine of productivity (Priority: 5/5): Steve Levitt and the host frame successful firms and economic growth as depending on a small number of powerful simple ideas, making the search for ideas central to productivity. Are ideas getting harder to find? (Priority: 5/5): Bloom and Van Reenen's paper argues that growth is slowing because innovation now requires much more R&D input for the same output, suggesting diminishing returns to research. Sector case studies: semiconductors, agriculture, and medicine (Priority: 5/5): The episode examines Moore's Law, agricultural yields, and life expectancy improvements to show that progress continues but often requires vastly more resources than before. Firm-level R&D returns are declining (Priority: 4/5): CompuStat data show a positive link between R&D and sales, employment, and market value, but the payoff per dollar of R&D has weakened over time. Techno-optimist rebuttal via emerging technologies (Priority: 4/5): Kelly and Zach Wienersmith argue that technologies such as bioprinting, AR, robotics, and some AI uses could still trigger major productivity gains, though they are cautious about claims of easy breakthroughs. Growth, politics, and policy (Priority: 5/5): The discussion ends with concerns that pension and budget assumptions rely on unrealistic growth rates, and with policy proposals to increase public and private R&D and broaden access to invention. Who becomes an inventor? (Priority: 4/5): Research on 1.2 million inventors suggests children from the top 1% are far more likely to become inventors than those from the bottom half, implying large lost innovation potential among disadvantaged groups.

Key Arguments: A single simple idea can create outsized economic value, but firms and societies often underinvest in generating ideas. Bloom and Van Reenen argue that the core reason productivity growth has slowed is not a lack of effort, but that ideas are becoming harder to discover. In semiconductors, huge increases in R&D spending have been needed just to keep pace with historical improvement rates. Agricultural productivity experienced a large mid-20th-century burst, but today gains are much harder to extract from a mature sector. Medical innovation has produced major life-expectancy gains, yet the cost of adding a human life-year has generally risen over time. Across publicly listed firms, R&D still raises sales and market value, but the effect per dollar has fallen since the 1960s. The Wienersmiths argue that some emerging technologies could still meaningfully raise productivity, especially AR, robotics, biotech, and perhaps some forms of AI. Policy responses should include more R&D incentives, more public research funding, and reducing barriers that prevent talented people from becoming inventors. Slower long-run growth has implications for pensions, deficits, and social stability because many systems assume stronger future growth than may materialize.

Data Points: Moore's Law productivity improvement: About 35% yearly improvement in semiconductors - Used to illustrate technological progress in chips and the high R&D needed to maintain it. R&D input increase in semiconductors: 25-fold increase - Bloom says resources devoted to semiconductor innovation have risen roughly 25 times to sustain similar progress. Agricultural productivity burst: 1930s to 1960s - Period when hybrid seeds, fertilizer, and farm management produced major gains in crop yields. Human life-year cost trend: Increasing over most of the period studied - Medical R&D became more expensive per unit of life extension, though late-1980s/1990s saw a surge in productivity. Firm-level return to R&D in the 1960s: About 3 times as much impact on sales per $1 million of R&D compared with the 1980s - CompaStat analysis showing declining R&D productivity over time. Inventor likelihood by family income: Top 1% children are 10 times more likely to become inventors than children from the bottom 50% - From Van Reenen's 'The Life Cycle of Inventors' research. Inventor dataset size: More than 1.2 million U.S. inventors - Dataset analyzed for the inventor mobility and family background study. Space travel cost: Around $10,000 per pound - Current cost to escape Earth's gravity, used to assess economic promise of reusable rockets. Robotic construction speed: Three times faster - Sam, the semi-automated mason, works three times faster when paired with a person than a person alone. House construction estimate: About 24 hours and about $5,000 - Contour crafting estimate for printing a house, cited as a potentially productivity-boosting technology.

Pivotal Quotes: "it's just getting harder and harder to find new ideas" — John Van Reenen: Central thesis of the episode explaining slowing productivity growth. "we might destroy our species as we know it while chasing total factor productivity" — Kelly Wienersmith: Concern about the social and existential risks of productivity-enhancing technologies and competitive adoption. "the adventure that was started sometime in the 18th century is probably still continuing" — Zach Wienersmith: Closing note arguing against declaring technological progress finished.

Implications: If innovation is getting harder, growth forecasts, pensions, and budgets may be too optimistic. Policymakers may need more R&D support and broader access to invention, while businesses should expect slower gains unless new technologies deliver unexpectedly large breakthroughs.