Asset Bubbles: How Prices Detach From Value and Crash

Fundamental value: the yardstick a bubble departs from

To say a price has "detached" you first need a value to detach from. In finance, the fundamental value of an asset is the present value of the cash it will ever deliver — dividends, coupons, rents — discounted for time and risk. For a share paying a dividend growing at rate g, discounted at required return r, the Gordon growth formula gives:

P* = D₁ ÷ (r − g)

Example: a stock expected to pay D₁ = $2 next year, with dividends growing g = 3% and investors demanding r = 8%, is worth P* = 2 ÷ (0.08 − 0.03) = $40. Everything about the company that matters is compressed into those three numbers. If the price is $40, no bubble. If it trades at $120 while D₁, g, and r haven't budged, then two-thirds of the price is not a claim on cash — it is a claim on the belief that someone will later pay even more. That residual is the bubble component B = P − P*.

The subtlety: P* is unobservable and forecast-dependent. Bubble skeptics (in the tradition of Eugene Fama) argue that a soaring price may just mean the market revised g up or r down — you can always tell a fundamentals story after the fact. That's why bubbles are easy to name in hindsight and genuinely hard to prove in real time. The honest working definition: a bubble is a large, self-reinforcing price rise that is later reversed by a comparably large fall not explained by any change in cash flows.

The mechanism: a self-reinforcing feedback loop

What makes a bubble a bubble is positive feedback. In a normal market, a rising price chokes off demand — higher cost, fewer buyers (see market equilibrium). In a bubble, rising prices attract demand, because recent gains become the reason to buy. The loop runs:

• Price rises on some real trigger (a new technology, low interest rates, a policy change) — Kindleberger's "displacement."
• Early buyers profit, and their returns are visible. Extrapolative expectations kick in: people forecast the future from the recent trend, not from cash flows.
• New money floods in — often borrowed money. Leverage amplifies both the buying pressure and the eventual unwind.
• Higher prices validate the earlier buyers, drawing the next cohort. Repeat.

The engine can't run forever, because it depends on a growing inflow of new buyers and credit. When the marginal buyer is reached — the last person willing to pay the going price — there is no one left to sell to at a profit. Prices stop rising, extrapolators see the trend break, and the same feedback runs in reverse: falling prices trigger selling, margin calls force leveraged holders to dump, and the descent feeds itself. Because leverage and fear are asymmetric, the crash is typically faster and overshoots below P* — the "panic" phase, and often a Minsky moment where a sudden collapse in confidence forces a scramble for cash.

The rational-bubble model: why smart money need not pop it

A tempting objection is: if everyone knows the price exceeds value, why doesn't a rational trader sell (or short) and collapse the bubble immediately? The rational bubble literature (Blanchard & Watson, 1982) shows a bubble can survive even among fully rational agents. The condition is that the bubble must grow fast enough to reward holders for the risk it bursts.

Suppose the bubble component Bₜ either survives with probability π, in which case it grows, or collapses to zero with probability 1 − π. For a risk-neutral investor requiring return r each period, the bubble must satisfy:

(1 + r)·Bₜ = π·Bₜ₊₁ + (1 − π)·0

so Bₜ₊₁ = Bₜ·(1 + r) ÷ π. If r = 5% and each period carries a 1 − π = 10% crash chance, the surviving bubble must grow by 1.05 ÷ 0.90 − 1 ≈ 16.7% per period. Holding the asset is perfectly rational: you're paid a premium for bearing crash risk. This is why "the smart money" often rides bubbles rather than fighting them — famously, hedge funds and Isaac Newton alike.

Two important riders. First, this requires an infinite horizon; on a finite-life asset (a bond that matures, a company that liquidates) backward induction rules the bubble out. Second, in reality even rational arbitrageurs face limits to arbitrage (Shleifer & Vishny, 1997): short-selling is costly, can be squeezed, and — as Keynes warned — "the market can stay irrational longer than you can stay solvent." So even non-believers may decline to bet against the bubble.

A worked case: Japan's asset bubble, 1985–1991

Numbers make the detachment concrete. After the 1985 Plaza Accord, the Bank of Japan cut its discount rate to 2.5% and credit expanded aggressively. The Nikkei 225 rose from about 13,000 in 1985 to 38,916 on 29 December 1989 — roughly a 3× gain in four years. Valuations left any fundamentals story behind: the market's price-to-earnings ratio hit around 60×, versus a long-run norm near 15×, implying earnings yields of under 2% while bonds paid more.

Real estate was even wilder. At the peak, the land under the Imperial Palace in Tokyo was said to be worth more than all the real estate in California; nationwide land value reached roughly 4× Japan's GDP. The feedback loop was textbook: rising land collateral let banks lend more, the loans bought more land and stock, which raised collateral again.

When the BoJ raised rates to cool it (to 6% by 1990), the marginal buyer vanished. The Nikkei halved within a year and, remarkably, had not recovered its 1989 high even three decades later — reclaiming 38,916 only in February 2024. Land prices fell for over a decade. The overshoot below fundamentals and the debt overhang produced Japan's "Lost Decade(s)": a cautionary tale that the crash's damage scales with how much leverage financed the boom.

Evidence bubbles are real: the laboratory and history

Skeptics can rationalize any historical episode, so economists built bubbles in a controlled lab. In the classic Smith, Suchanek & Williams (1988) experiment, subjects trade an asset with a known lifespan and a known random dividend, so the expected fundamental value is computable and falls in a straight line to zero at the end. Yet across hundreds of replications, prices reliably balloon to 2–3× fundamental value mid-experiment, then crash near expiry. With common knowledge of the value and no uncertainty about cash flows, standard theory predicts no bubble — but bubbles appear anyway, driven by uncertainty about others' rationality. Give traders more experience with the same group and the bubbles shrink; shuffle in newcomers and they return.

History supplies the field data, with a consistent shape:

• Tulip mania (1636–37): single rare bulbs traded for the price of a canal house before collapsing over days.
• South Sea Bubble (1720): shares rose ~8× then crashed; Newton, having sold early then re-bought, lost £20,000 and reportedly said he "could calculate the motions of the heavenly bodies, but not the madness of people."
• 1929: the Dow, up on margin-fueled speculation, fell 89% from peak to its 1932 trough.
• Dot-com (2000): the NASDAQ hit 5,048 in March 2000 on companies with no earnings, then fell ~78% by 2002.
• US housing / subprime (2006–08): the Case-Shiller index roughly doubled 2000–2006, then fell ~27%, triggering the global financial crisis.

Misconceptions, limits, and where it shows up

Misconception 1 — "A bubble is just an overvalued asset." Overvaluation is a level; a bubble is a dynamic — a self-reinforcing price process sustained by resale expectations and typically credit. A stock can be modestly overpriced with no bubble mechanics; a bubble is defined by the feedback and its eventual reversal.

Misconception 2 — "Bubbles are obvious, so you can just short them." Identification is genuinely hard in real time (the fundamentals-story problem), and even when you're right, timing and solvency can wipe you out first. Julian Robertson's Tiger Fund shorted dot-com stocks — correctly — but closed in early 2000 after losses, weeks before the peak.

Misconception 3 — "It contradicts the Efficient Market Hypothesis, so one must be wrong." Not quite. The efficient-market hypothesis says you can't reliably beat the market, which limits-to-arbitrage explains even amid a bubble. Bubbles are the hard case that behavioral finance and EMH argue over, not a knockout for either.

Limitation of the theory: because P* is unobservable, no bubble test is fully clean — this is a real critique of the whole field, not a footnote.

Where it shows up today: the 2021 meme-stock episode (GameStop), crypto cycles (Bitcoin's 2017 and 2021 peaks and ~65–80% drawdowns), zero-rate-era "everything" valuations, and recurring debate over whether AI-related equities are in a boom or a bubble. Regulators respond with macroprudential tools — loan-to-value caps, margin and capital requirements (Basel III), and stress on leverage — because it is credit, more than optimism, that turns a price detachment into a systemic crash.

Rational bubble vs. behavioral (greater-fool) bubble — two accounts of the same price detachment

Feature	Rational bubble	Behavioral / greater-fool bubble
Who is holding the asset	Rational agents who know price > value	Optimists or noise traders who misjudge value
Why they hold it	Bubble grows fast enough to compensate for crash risk	Belief price keeps rising; or plan to resell to a "greater fool"
Key condition	Expected return on bubble ≥ required return each period	Persistent inflow of new buyers / credit
Can it start on a worthless asset	No — needs infinite horizon; ruled out in finite-life assets	Yes — even provably worthless assets bubble (lab, meme coins)
What ends it	Any positive per-period probability of collapse compounds	Marginal buyer exhausted; sentiment or credit reverses
Signature model	Blanchard–Watson (1982) stochastic bubble	Kindleberger–Minsky; De Long et al. noise traders (1990)