Daron Acemoglu on Innovation and Shared Prosperity

Episode Summary

Executive Summary: Russ Roberts and Daron Acemoglu debate whether technology automatically raises broad living standards. Acemoglu argues it does not: competition, labor bargaining power, and a mix of automation plus new tasks must be present for gains to spread. They explore medieval monopolies, the Industrial Revolution, U.S. manufacturing, unions, monopsony, tax policy, and how AI could tilt the economy toward excessive automation.

Main Topics: Technology is not automatically pro-worker (Priority: 5/5): Acemoglu argues technological progress can raise output and profits without broadly improving wages or living conditions unless institutions and market structure channel gains to workers. Historical cases: medieval technology and the Industrial Revolution (Priority: 5/5): They discuss windmills, enclosure, child labor, urban disease, and the slow spread of benefits during industrialization as evidence that technology’s gains were often delayed or captured by elites. Labor power, unions, and quasi-rents (Priority: 4/5): Acemoglu distinguishes monopsony from broader bargaining problems, arguing workers need voice and coordination to share the surplus created in employment relationships. Automation vs. creation of new tasks (Priority: 5/5): The core analytical claim is that automation can reduce labor’s role unless innovation also creates complementary tasks for workers; prosperity depends on the balance between the two. U.S. industrialization and labor shortages (Priority: 4/5): The United States is presented as an example where scarce skilled labor pushed firms toward machinery designed for unskilled workers, helping wages rise for those workers. Policy levers: taxes, digital ads, antitrust, and complementary AI (Priority: 4/5): Acemoglu proposes changing tax distortions, encouraging human-complementary innovation, taxing harmful digital-ad business models, and revisiting antitrust. AI and market concentration (Priority: 3/5): They end by noting that AI could either intensify concentration or create competition, but the current path may strengthen big incumbents like Microsoft and Google.

Key Arguments: Technology has historically improved welfare only when competition exists, labor is not coerced, and innovation complements rather than replaces workers. Industrialization did not automatically lift workers; many saw worse health, longer hours, and lower autonomy for decades before institutions changed. Trade unions and democratic worker voice can help distribute quasi-rents created in employment relationships, especially in large firms. Monopsony exists in some labor markets, but Acemoglu thinks broader bargaining and power asymmetries matter more than monopsony alone. Automation can lower prices and raise profits while still reducing labor’s share unless new tasks are created for workers. U.S. firms historically used technology to make unskilled labor productive because skilled labor was scarce, helping attract immigrants and support higher wages. Government should not centrally pick technologies, but it can shape incentives through tax policy, R&D competitions, ad taxes, and antitrust enforcement. Digital ad-driven platforms may impose negative social and political externalities by optimizing for outrage and engagement rather than productive communication.

Data Points: Years after industrialization began: 90–120 years - Acemoglu cites 1840 reports of child labor and harsh conditions long after the Industrial Revolution began around 1750 or even earlier textile mills around 1720. Child labor hours: 18 hours - British parliamentary report described children as young as five or six working this long in deep coal mines. Child labor age: 5 or 6 years old - Referenced in the 1840 parliamentary report on conditions in England. Chartist signatures: 3 million - Acemoglu uses this to show the scale of political demand for reform in Britain. Life expectancy at birth: Below 30 years - He says urbanization and industrial conditions may have pushed life expectancy below 30 in Britain. Marginal tax rate on labor: About 25% - In a paper with Andrea Manera and Pascual Restrepo, cited as a distortion pushing firms toward automation. Marginal tax rate on capital: About 5% - Same discussion of tax policy biasing firms toward capital over labor. Wages in the U.S. vs. the U.K.: Higher in the United States - Used to explain why immigrants and low-skilled workers found the U.S. attractive. Trade union legalization timing: After the Master and Servant Acts - Acemoglu argues worker conditions improved once unions were legalized and able to organize.

Pivotal Quotes: "there is nothing automatic about technological progress bringing widespread prosperity" — Daron Acemoglu: Core thesis of the episode: technology alone does not ensure broad gains. "the factory of the future will have two employees, a man and a dog" — Daron Acemoglu: Illustrative example of automation reducing labor’s role in production. "the social and the privately optimal amount of automation is different" — Daron Acemoglu: Used in the policy discussion to justify intervention when labor markets are imperfect.

Implications: Listeners should expect AI and automation to redistribute gains unevenly unless policy, competition, and worker voice are deliberately strengthened. Firms may optimize for capital and data unless incentives shift toward human-complementary innovation.