LesenForschungRadarAnlageframework
Anmelden / Registrieren
← Zurück zur Lektüre
NewsThema · 安德烈亚森泡沫模型Wesentlich3 Min. LesezeitDisconfirmAI Redaktion

Identifying Stock Bubbles: A New Method Separates Fundamentals from Speculation, Explaining 61% of Valuation Ratio Variation

Andreasen and Bro's study in the Review of Finance decomposes stock prices into expected dividends, expected returns, and a rational bubble, using investor survey data to distinguish fundamentals from speculation. Based on U.S. monthly data from 1971 to 2024, the model estimates that the bubble component explains 61% of variation in the S&P 500 price-dividend ratio, identifying multiple historical speculative episodes including a potential AI-related phase. The paper also finds that positive bubble shocks are followed by lower unemployment and higher monetary policy rates.

2026-10-20 11:17 GMT+8

Stock market bubbles are easy to identify in hindsight—the dot-com boom, the run-up to the global financial crisis, and the surge in technology stocks during the pandemic are all familiar examples. But identifying a bubble while it is forming remains one of the longstanding challenges in financial economics. Rising stock prices do not necessarily indicate speculation: investors may rationally expect stronger earnings growth, lower future returns, or improving economic conditions. Distinguishing these fundamental explanations from a speculative bubble is precisely the problem that Andreasen and Bro address in their 2026 paper published in the Review of Finance.

They develop a present-value model that decomposes stock prices into three components: expected future dividends, expected future returns, and a rational bubble. The key innovation is incorporating survey-based expectations about dividend growth and stock prices alongside traditional market data. This matters because a high price-dividend ratio alone does not establish that a bubble exists: when investors anticipate stronger dividend growth or lower required returns, high valuations can be rational. By incorporating information about what market participants actually expect, the model attempts to identify the portion of valuations that cannot be explained by these fundamental components. In this framework, a rational bubble is a non-fundamental component of prices consistent with the model's rational-expectations assumptions.

Anmelden / Registrieren
LesenForschungRadarAnlageframework

Using U.S. monthly data from January 1971 through December 2024, the authors find that the estimated bubble accounts for 61% of the variation in the S&P 500 price-dividend ratio. Expected returns account for another 42%, while expected dividend growth contributes little to explaining variation in the ratio over the full sample. These variance contributions are not additive shares of a fixed total and should not be interpreted as percentages summing to 100%. The findings suggest that speculative forces may play a much larger role in explaining stock market valuation fluctuations than traditional models without bubbles imply. The authors also show that removing the estimated bubble substantially reduces the excess volatility of stock valuations and eliminates the evidence of explosive dynamics detected in the unadjusted price-dividend ratio.

The model identifies several recognizable historical bubble episodes. During the early-1970s Nifty Fifty boom, the estimated bubble represents approximately 10% of the price-dividend ratio before declining to around 2% during the 1980s. It then rises sharply during the dot-com era, reaching 27% in January 2001. Another increase occurs around the global financial crisis, followed by a COVID-era bubble that peaks at approximately 19% in November 2021. Most notably, the estimated bubble reaches nearly 25% of the price-dividend ratio in December 2024, potentially reflecting investor enthusiasm surrounding artificial intelligence. This is a model-based historical estimate, not evidence that AI-related stocks were necessarily overvalued or that a subsequent market correction was inevitable.

Quellen

  • alphaarchitect.com↗
Thema

安德烈亚森泡沫模型

丹麦奥胡斯大学马丁·安德烈亚森(Martin M. Andreasen)与ATP机构耶佩·布罗(Jeppe Bro)在《金融评论》发表研究,提出一种利用投资者调查数据分离股市基本面与投机成分的现值模型。该模型基于1971至2024年美国月度…

2 Schritte1 BeiträgeAktualisiert 2026-10-21 10:53

Wie es dazu kam

  1. 2026-09-18《金融评论》刊发Andreasen和Bro的研究,指出模型识别出包括潜在AI相关阶段在内的多个历史投机时期,且AI泡沫占2024年底市价股息率约25%。revfin.org ↗
  2. 2026-10-20媒体解读该研究,强调其利用调查数据将价格分解为预期股息、回报率和理性泡沫,并发现泡沫冲击导致货币政策利率上升。alphaarchitect.com ↗

Was noch ungeklärt ist

  • 不同学科对“泡沫”的定义存在差异(如偏离基本面 vs. 价格路径特征),这是否影响模型结果的横向可比性?
  • 模型识别出的“AI泡沫”是否具有可持续性,还是仅反映短期情绪波动?
  • 泡沫冲击导致利率上升的具体传导机制在各国央行实践中是否一致?
Vollständiges Thema lesen →
Recherche fortsetzen →