The Psychology of Bubbles: Why Smart People Buy at the Top

In the autumn of 1720, Sir Isaac Newton sold his shares in the South Sea Company for a profit of seven thousand pounds, a sum that would have set him comfortably for several years. Then he watched the price continue rising. His friends were getting richer by the week. The most brilliant scientific mind of his age, the man who had decoded planetary motion and invented calculus, bought back in at near the peak and lost twenty thousand pounds when the bubble collapsed. He reportedly said afterward that he could calculate the motions of heavenly bodies but not the madness of people.

Newton’s loss is typically cited as evidence that even geniuses are not immune to speculative excess. But this framing misses the actual lesson. Newton did not lose money because he was foolish or because his intellect failed him. He lost money because he made a perfectly rational inference from the social information available to him. His friends were making extraordinary returns. The price signal was strong and persistent. The decision to reinvest was not a lapse in reasoning; it was the result of reasoning applied to a corrupted information environment. Newton was not irrational. He was solving the wrong problem.

The Information Problem at the Heart of Every Bubble

Financial bubbles are not primarily psychological failures. They are information failures that produce psychological effects. The distinction matters enormously because it changes what we think the remedy is.

During a genuine bubble, the price of an asset rising rapidly transmits a signal to observers: people who bought this earlier made money, which means they knew something, which means the asset may still be worth buying. This is not a cognitive error. In normal markets, rising prices do contain information about underlying value. A stock that has risen fifty percent over two years may have done so because analysts have correctly identified a business model that will be enormously profitable. Following price momentum in normal conditions is sometimes a sensible heuristic.

The problem is that in a bubble, price momentum becomes entirely self-referential. The asset rises because people expect it to rise, and people expect it to rise because it has been rising, and the very act of buying to capture that rise pushes the price further, confirming everyone’s expectations for another cycle. The price no longer conveys information about fundamental value. It conveys information only about the distribution of belief among current market participants. But because price ordinarily does convey fundamental information, observers rationally update toward the bullish interpretation.

Newton saw the South Sea Company’s price and assumed other investors had analyzed the company’s prospects and found them excellent. They had not. Most of them were also watching the price and making the same inference. The market had become a hall of mirrors in which everyone’s belief was derived from everyone else’s behavior, and no one’s belief was anchored in anything external to the market itself.

This is the precise mechanism that distinguishes a bubble from a legitimate boom. In a legitimate boom, rising prices reflect genuine improvements in underlying economics, and observers who follow the price signal get approximately the right answer. In a bubble, rising prices reflect only the distribution of speculative expectation, and observers who follow the price signal are reasoning from a corrupted input.

Why Skeptics Capitulate: The Social Cost of Being Right Too Early

One of the most consistent features of historical bubbles is that skeptics who identify the mania early are almost universally punished before they are vindicated. This is not a minor detail. It is the central mechanism that sustains the mania long enough to cause serious damage.

Consider the Dutch tulip mania of 1637. A professional trader who correctly identified that tulip bulbs could not sustain their nominal prices would have faced a specific social and economic problem: being right too early is commercially identical to being wrong. If you refuse to participate in a market that continues rising for another year, you have underperformed your peers by the full magnitude of that year’s gains. Your clients, partners, and colleagues observe your underperformance, not your correct analysis. In the mercantile culture of seventeenth-century Amsterdam, underperformance carried reputational costs that could end careers.

The rational response to this incentive structure is to participate in the market while privately doubting it, to buy the tulips while knowing they are overpriced, because the cost of being visibly wrong in your skepticism exceeds the cost of riding a bubble and selling before it bursts. The flaw in this strategy, of course, is that no one can reliably identify when a bubble will burst, and the attempt to time the exit concentrates enormous risk into the final stages when the reversal is fastest and most violent.

John Maynard Keynes formalized this dynamic with his famous observation that markets can remain irrational longer than investors can remain solvent. But the insight cuts deeper than solvency. Even a fully solvent investor faces social and institutional costs from prolonged skepticism. Fund managers who underperform their benchmarks lose assets under management. Corporate executives who refuse to pursue acquisitions in a boom era lose their positions to successors who will. Central bankers who raise rates to cool a mania face political pressure from constituencies who are currently enjoying rising asset prices.

The social structure of professional finance systematically penalizes correct early skepticism and rewards participation in trends regardless of their fundamental justification. This is not a correctable flaw in individual psychology. It is an institutional design problem that has never been solved across four centuries of organized capital markets.

The Role of Narrative: Why Bubbles Need Stories

Pure price momentum is not sufficient to sustain a bubble for long. Prices can rise rapidly and then fall without ever reaching the scale of a true mania. What transforms a speculative episode into a civilization-level event is the development of a narrative that reframes the speculation as fundamental insight.

Every major bubble in history has been accompanied by a powerful explanatory story. The South Sea Company promised to refinance British national debt through the profits of a monopoly trade with Spanish America. The narrative was plausible enough, and complex enough, that most investors could not definitively refute it. The 1920s American stock market bubble was sustained by the story of a new technological era in which electricity, automobiles, and radio would permanently raise corporate earnings. The story was not entirely wrong; those technologies did transform the economy. The error was the valuation, not the technological assessment.

Narratives serve the bubble in two ways. First, they provide a fundamental justification for prices that cannot be justified on traditional metrics. When a stock trades at fifty times earnings, someone who points to the price-to-earnings ratio as evidence of overvaluation can be answered: yes, but this company is entering a new phase of growth that historical metrics cannot capture. The narrative makes the conventional valuation tools seem obsolete.

Second, narratives create a moral dimension to skepticism. During the 1840s British railway mania, skeptics of railway investment were characterized as opponents of progress, as people who could not see that the steam age was transforming Britain. This moral framing raised the social cost of skepticism beyond mere financial underperformance. To doubt the railways was to reveal yourself as a backward-looking reactionary. Similar dynamics appeared in every subsequent technology bubble. The skeptic is not just wrong about prices; they are wrong about the future. They are standing athwart history.

This narrative mechanism explains why bubbles are so often associated with genuine technological change. A new technology provides the raw material for a compelling story, and the story allows speculation to detach from fundamentals while maintaining the appearance of disciplined investment thesis. The most dangerous bubbles are not the ones built on pure fraud or obvious irrationality. They are the ones built on real technologies and genuine business model innovation, because the story is partially true, and the partial truth makes the error invisible until it is not.

Crashes as Information Cascades, Not Panics

The conventional account of bubble crashes emphasizes panic: investors suddenly lose confidence, selling begets selling, rationality is overwhelmed by fear. This account is not wrong, but it is incomplete in ways that matter for understanding why crashes are so difficult to prevent or arrest.

The more accurate model is an information cascade running in reverse. During the bubble, each new buyer’s action conveyed apparent information about the asset’s value, encouraging the next buyer. At some point, a threshold is crossed where the price becomes too high for even the most optimistic buyer to justify at the margin. When buying stops, the absence of buyers conveys information. Holders who assumed others were buying because they knew something valuable now update toward the possibility that no one knows anything useful. The information cascade reverses: each seller’s action implies that others are discovering the same problem, which encourages the next seller.

What makes this cascade so fast and violent is the asymmetry of leverage. During bubbles, participants frequently borrow money to amplify their returns. When prices fall, leveraged holders face margin calls that force selling regardless of their fundamental assessment. This forced selling is not panic in the psychological sense. It is mechanically compelled by contractual debt obligations. The mechanical selling drives prices lower, which triggers more margin calls, which drives prices lower still.

The 1929 crash accelerated so rapidly in part because call loans, the standard mechanism for financing stock purchases, could be recalled by brokers on twenty-four hours’ notice. When brokers called their loans, investors had no choice but to sell. The speed of the crash reflected the speed of forced deleveraging, not the speed of psychological contagion. The psychology was secondary to the balance sheet mechanics.

What Cannot Be Fixed About Human Markets

Newton’s loss points toward a conclusion that is uncomfortable for both market advocates and market skeptics. The problem is not that markets are populated by irrational people who can be educated into better behavior. The problem is that markets are social institutions that generate information from the aggregation of private beliefs, and that process of aggregation is irreducibly vulnerable to self-referential loops.

When every participant’s belief is partly derived from other participants’ observed behavior, and when that behavior is expressed through the same price signal that the belief is supposedly informing, the system can sustain arbitrarily large divergences from fundamental value for arbitrarily long periods. There is no architectural fix that preserves the information-discovery function of markets while eliminating this vulnerability. The price signal that normally works so well is the same mechanism that makes bubbles possible.

What this means practically is that bubble prevention is not a solvable engineering problem. It is an ongoing political and institutional management problem, requiring constant judgment calls about when speculative excess has become dangerous enough to justify regulatory intervention, knowing that intervention will be politically unpopular, will be timed imperfectly, and will be resisted by constituencies with strong financial interests in the continuation of the boom.

The people who manage this problem best are those who understand, at a structural level, that they are not trying to correct individual irrationality. They are trying to manage institutional incentive structures that predictably produce collective irrationality even when populated entirely by intelligent, well-intentioned people. Newton was intelligent and well-intentioned. He lost twenty thousand pounds anyway, not because he failed to apply his considerable intellect but because he applied it to a problem that was structurally set up to defeat intelligence. Markets remain, four centuries after the tulip mania, structurally capable of producing the same outcome. The next Newton is already buying near the top of something, and the analysis is almost certainly rigorous, sophisticated, and wrong.