Why do bubbles always feel rational at the time?

The short answer

The retrospective “how could anyone have believed this was sustainable?” is the wrong question. The right question is: given what was visible at the time, was buying actually irrational? The answer, almost always, is no. Bubbles persist because their internal logic is coherent within the time horizon of the marginal participant.

The complication is that local rationality and systemic rationality diverge. Every individual decision can be defensible while the aggregate outcome is unsustainable. A retail investor in 1999, a homebuyer in 2006, or an AI infrastructure ETF buyer in 2025 each receives positive reinforcement for their decisions for extended periods.

The collapse happens not when investors collectively recognize the bubble, but when the structural conditions that supported the marginal buy decision change — typically, when refinancing or fresh capital becomes scarcer.

→ New to behavioral finance? Financial education framework

What the data shows

Empirical performance of bubble-era investment theses (S&P, Refinitiv, FRED, period 1995-2026) shows just how much positive feedback investors received before the eventual collapse.

The numerical context (S&P, FRED, 1995-2026) :

• Magnificent 7 cumulative return 2016-2025: +875%, vs S&P 500 +234%
• Cisco Systems 1995-2000: +75,000% (peak), then -86% by 2002
• US housing prices: +89% from 2000 to 2006 peak (Case-Shiller national index)
• Bitcoin: +20,000,000% from 2010 to 2021 peak — making “HODL” empirically defensible for a decade
• S&P 500 in 2022: -19.4%, Magnificent 7 in 2022: -41.3% — showing the dispersion when the thesis pauses

The exception worth noting: rationality is not uniform across all bubble participants. Some buyers in 1999 were sophisticated — they understood the valuation distortion and traded accordingly. Others were retail momentum chasers without that framework. The local rationality argument applies most strongly to the second group, who optimized correctly given their information set.

→ Dataset: S&P 500 historical returns

Why it happens — the macro mechanism

Three psychological and structural channels make bubble participation feel rational.

The empirical reinforcement channel. Markets pay buyers in real time. A 2018 buyer of Nvidia at $50 saw it rise to $135 by 2021, then $145 by 2024 — a track record that any rational analyst would treat as confirming their thesis. Skeptics who shorted at $50 lost capital and credibility. The market literally rewards conviction over caution during the bubble phase. Equity valuation framework.

The narrative coherence channel. Contrary to the popular framing of bubbles as mass delusion, late-stage rallies are typically supported by genuinely transformative narratives — internet productivity (1999), housing as a national asset (2006), AI as the next general-purpose technology (2024). The narratives are not wrong; they are simply over-priced. The rationality lies in mistaking “this is real” for “this is correctly valued at any price.” This is the angle most often missed: bubbles do not require false beliefs, only mis-applied true beliefs.

The challenge is that distinguishing the two requires explicit valuation discipline that most participants do not exercise.

The institutional pressure channel. Professional asset managers face career risk for underperforming benchmarks, even temporarily. A manager who avoided tech in 1999 lost clients before being vindicated in 2002; a manager who avoids the Magnificent 7 in 2024-2026 faces the same dynamic. This forces institutional flows into the dominant theme regardless of individual conviction.

Synthesis by regime: in regimes where the dominant theme produces verifiable economic transformation (Cisco’s actual revenue growth in 1996-2000, Apple’s actual profit growth in 2010-2020), the rationality argument holds longest; in regimes where the theme is purely narrative (NFTs in 2021, certain SPACs in 2020-2021), the local rationality breaks faster as the next-buyer story collapses sooner. The transition parameter is the rate of new participant inflow: bubbles persist while inflow accelerates and end when it inflects, regardless of valuation level.

Bubbles do not require irrational investors — only rational investors who correctly anticipate that other rational investors will buy at higher prices.

→ Framework: Behavioral investing and cognitive biases

What it means for different economic actors

Savers often interpret skepticism toward late-stage bull markets as a personal failure to participate. The empirical track record of momentum strategies during 2016-2025 made conservative positioning feel like systematic underperformance.

Investors who built mental models of “I will exit when X happens” rarely execute, because X is itself a moving target — the bubble narrative continuously updates to incorporate new information that justifies the current price. Behavioral research suggests that even sophisticated investors retain bubble-era positions ~60% longer than their stated risk plan would predict.

Pension funds and endowments structurally cannot exit; their challenge is calibrating the level of bubble-era exposure they are willing to bear without explicit conviction in the underlying valuations.

A common error is to mistake post-collapse hindsight for pre-collapse insight. The investor who exited Cisco in 1999 looked foolish for nine months and prescient for the next 24. Most investors do not have the temperament to bear the first nine months.

Practical observation

Go deeper

Related questions

• What defines a financial bubble vs bull market?
• What is the greater fool theory?
• What is the Minsky financial instability hypothesis?
• How do you identify a bubble in real time?

Frequently asked questions

How does the local rationality framework differ from “irrational exuberance”?

Robert Shiller’s “irrational exuberance” framing emphasizes the psychological anchoring and herd behavior that produce mispricing. The local rationality framework, drawn more from Soros and Brunnermeier, emphasizes that the same prices can emerge from rational individual optimization given the information available. Both frames have empirical support; they describe different participant populations within the same market. Sophisticated investors typically operate under local rationality; less informed retail under irrational exuberance.

If bubbles are locally rational, why do they ever end?

The local rationality holds only as long as the marginal buyer can be expected to remain available. Bubble collapses correspond to inflection points where the supply of new buyers slows or reverses — typically driven by external shocks (Fed tightening, refinancing wall, regulatory change) rather than collective recognition. This is why bubbles persist long past the point at which fundamental valuation models flag them, and why they collapse violently rather than gradually.

How should the local rationality framework be applied to AI exposure in 2024-2026?

The framework suggests that current AI-related positions can be locally rational without being systemically defensible. Each individual buyer of an AI infrastructure ETF in 2024 received positive reinforcement through 2025; that reinforcement says nothing about the cumulative aggregate position. The honest internal question is not “is AI real?” — it likely is — but “at what valuation does my position implicitly bet on continued momentum versus underlying cash flows?”

Last updated — 19 May 2026