Robert Shiller on Narrative Economics

Episode Summary

Executive Summary: Russ Roberts and Robert Shiller discuss Shiller’s idea of “narrative economics”: the study of how contagious stories shape beliefs and, in turn, economic behavior. They debate whether narratives truly cause macro outcomes or simply reflect them, using examples like the Great Depression, WWII, automation fears, Trump, Greta Thunberg, and policy responses after crises.

Main Topics: Definition and scope of narrative economics (Priority: 5/5): Shiller defines narrative economics as studying popular stories that influence economic behavior over time, arguing that people transmit stories rather than equations and that these narratives can go viral and matter economically. Causality vs. correlation in economic storytelling (Priority: 5/5): Roberts repeatedly challenges whether narratives are causal or just post-hoc explanations, pressing Shiller on how to distinguish stories that move behavior from those that merely describe it. Data revolution and tracking narratives (Priority: 4/5): Shiller argues digitized text, search tools, and future neuroeconomics will allow economists to systematically measure narratives in newspapers, books, sermons, and other sources, making the field more empirical. Great Depression, WWII, and competing narratives (Priority: 5/5): The conversation uses the Great Depression and World War II to illustrate how multiple narratives—bank runs, tariffs, monetary contraction, New Deal, war, pent-up demand—can coexist and influence interpretation of macroeconomic history. Automation, technological unemployment, and recurring fears (Priority: 4/5): Shiller traces the long history of fears that machines replace workers—from ancient references and Luddites to AI today—and argues these narratives can alter saving and spending decisions even if the fear is often false. Political leaders, self-presentation, and economic sentiment (Priority: 3/5): The guests discuss how figures like Trump, Greta Thunberg, and Keynes shape public behavior through charisma, symbolism, and storytelling, not just through formal policy levers. Policy implications and crisis management (Priority: 4/5): Shiller concludes that central banks and governments already use narrative management implicitly during crises, but economists should formalize this knowledge in research and teaching.

Key Arguments: Shiller’s central claim is that economic fluctuations are substantially driven by contagion of oversimplified, easily transmitted narratives. Roberts argues that many narratives may be entertaining or psychologically true without being clearly causal in macroeconomic outcomes. Shiller responds that narratives can be studied empirically using digitized historical text and computational tools, improving on older methods that lacked searchable archives. The Great Depression shows how stories about catastrophe, unemployment, and technological displacement can shape consumer and investor behavior. World War II and the postwar period illustrate the problem of attributing macro outcomes to a single story, since multiple narratives and measurement issues are always present. Automation fears have recurred for centuries and often prove false, yet they can still affect confidence, savings, and investment when widely believed. Political and cultural narratives can matter economically because people respond to status, identity, and social perception, not just prices and incentives. Policy makers already try to manage narratives during crises, such as by rescuing banks to prevent bank-run stories from spreading. Economics should become more interdisciplinary, combining text analysis, neuroscience, and experimental methods to understand behavior better. Both economists and the public are prone to self-reinforcing stories about policy, markets, and national identity that may be only partly grounded in evidence.

Data Points: Podcast date: December 6, 2019 - Introductory metadata for the EconTalk episode. Previous EconTalk appearance: September 2008 - Shiller previously appeared to talk about housing and bubbles. Great Depression / GDP official series start: 1929 - Roberts notes official GDP numbers begin in 1929. UK government rescue example: 2007 - Shiller cites the Northern Rock collapse and rescue during the financial crisis. Employment growth: 266,000 jobs - Roberts references the unexpectedly strong U.S. employment report released that Friday. Unemployment rate: 3.5% - Roberts cites the current U.S. unemployment rate from the same employment report. Great Depression car-sales drop: about 86% to close to 90% - Shiller cites Ford car sales collapsing during the Depression as evidence of changing sentiment. Historical span of automation narrative: over 2,000 years - Shiller traces fears of machines replacing labor back to Aristotle and Homer. Third edition of Ricardo’s work: 1830s - Roberts mentions Ricardo adding a chapter on machines in the third edition of his Principles. Key Keynes book publication: 1936 - Roberts notes Keynes’s The General Theory of Employment, Interest, and Money. Great Depression start of unemployment narrative: 1937–38 - Shiller says the recession narrative became prominent after the 1937–38 recession.

Pivotal Quotes: "Narrative economics is the study of popular narratives that relate to economic behavior." — Robert Shiller: Shiller defines the concept at the start of the interview. "economic fluctuations are substantially driven by contagion of oversimplified and easily transmitted variants of economic narratives." — Russ Roberts (quoting Shiller's book): Roberts challenges Shiller on the strength of the book’s main claim. "We are undergoing a data revolution now." — Robert Shiller: Shiller explains why narrative economics may become more rigorous through digitized text and computational search.

Implications: Listeners should expect economics to incorporate more text analysis, psychology, and history. The episode suggests that policy, markets, and public confidence are shaped not only by incentives but by contagious stories that can amplify or dampen economic change.