Booms, Crashes, and Bubbles: A Fresh Look at What the Data and Theory Say

Booms don’t burst: Revisiting the predictability of stock market crashes

The main goal of the paper by William N. Goetzmann, Otto Manninen, and James Tyler is to establish, through an empirical examination of over two centuries of US stock market data, what actually happens after large price booms and whether those booms reliably predict crashes. The authors’ analysis operates at two levels:

1. Aggregate market. Using a monthly price index of 100 large-cap US stocks spanning 1792–2024, the authors count how often booms of varying magnitudes (50%, 75%, 100% over 12- or 36-month windows) were followed by further gains or full reversals, at 1-, 3-, and 5-year horizons.

2. Industry portfolios. Extending Greenwood, Shleifer, and You (2019), the authors apply the same framework to 49 Fama-French industry portfolios (1926–2024) and 68 Cowles Commission industry portfolios (1871–1938). A boom is defined as a doubling within 12 months; a crash as a halving.

Findings

• Bubbles are extremely rare. At the 50% three-year boom threshold, only 50 out of 2,700 overlapping observations ended in a full reversal, implying a frequency below 2%. At the 100% threshold, just 10 reversals occurred in 232 years.
• Booms do not reliably predict crashes. At nearly every threshold and horizon, booms are more likely to be followed by further booms than by full reversals. The popular belief that a sharp price run-up signals an impending crash finds no support in over two centuries of US data.
• ​Crashes are more likely to be followed by recoveries. At the five-year horizon, recoveries dominate further declines at every crash threshold. This finding is consistent with the long-run upward drift of equity markets. ​
• Booms predict higher volatility, not direction. Comparing conditional to unconditional return distributions shows that post-boom periods exhibit a wider spread of outcomes: both large gains and large losses become more probable. This reconciles the apparent paradox that crashes are more frequent after booms even though booms do not on average predict crashes.

The authors caution that the US market is an unusually successful single observation, and that inferences drawn from a handful of extreme, overlapping events should be treated as historical anecdote rather than precise probabilistic evidence.

​


Bubbles, Booms and Crashes in the US Stock Market 1792-2024
Authors: William N. Goetzmann, Otto Manninen, James Tyler
From: Yale University

Disagreement, not irrationality, drives asset bubbles

Gadi Barlevy’s goal in this article is to reframe the theoretical debate on asset bubbles. Rather than asking whether bubbles require irrational agents, as the conventional debate assumes, he argues the real dividing line is between models where agents hold identical information (which rule out bubbles) and models where agents hold different beliefs (which allow them), regardless of whether those agents are rational or not. The author looks  the major strands of the bubble literature, namely:

• Early rational models (Tirole 1982, 1985; Blanchard-Watson 1982): Barlevy explains how these frameworks demonstrate that bubbles are impossible or fragile when all agents share the same information, relying on backward induction and transversality conditions to rule out sustained price deviations from fundamentals.

• Models with irrational beliefs (Barberis et al. 2018; Bordalo et al. 2021; Scheinkman-Xiong 2003): Barlevy surveys how overconfidence, extrapolation, and diagnostic beliefs, each a departure from full rationality, can generate robust boom-bust dynamics in which prices rise well above and then collapse back toward fundamental value.

• Rational heterogeneous-beliefs models: Barlevy then shows that rational agents exposed to different information can produce the very same bubble dynamics attributed to irrational behavior, leading him to conclude that the rationality assumption is secondary to the belief-heterogeneity assumption as the true driver of bubbles.

• Empirical benchmark: Throughout, Barlevy draws on Greenwood, Shleifer, and You (2019), and implicitly on Goetzmann et al. above,  as an empirical benchmark, noting that nearly half of asset booms do not end in busts. He uses this finding to challenge definitions of bubbles that presuppose an inevitable collapse.

Conclusion

• The rationality vs. irrationality framing is a red herring. The key driver of bubbles is heterogeneous beliefs, not irrationality per se.
• Policy implications are largely unchanged by whether the underlying model assumes rational or irrational agents: the key variable for policymakers is the presence of heterogeneous beliefs and the speculative premium they generate.

Why Bubbles Occur: Revisiting the Rationality Debate
Author: Gadi Barlevy
From: Federal Reserve Bank of Chicago