The Economics of Superstition

In 1928, a London insurance broker named Edward Lloyd began requiring ship captains to report the results of voyages made on Fridays. He had noticed, or believed he had noticed, that Friday departures produced worse outcomes than departures on other days, and he was considering adjusting premiums accordingly. Lloyd’s colleagues thought this was absurd. The data, when properly analyzed, showed no such effect. But the belief in the unluckiness of Friday sailings was so widespread among British sailors — who would sometimes refuse to depart on a Friday regardless of contracts or instructions — that it functioned as a real economic variable. Whether or not Fridays were objectively unlucky, the belief that they were unlucky changed sailor behavior enough to potentially affect outcomes, which meant an insurer who ignored the belief was ignoring a genuine risk factor. Lloyd was wrong about the cause and right about the effect. The sailors’ superstition was irrational as meteorology and entirely rational as coordination device.

This is the puzzle at the heart of superstition economics: beliefs that are empirically false can be socially functional, and beliefs that are socially functional tend to persist even when evidence against them accumulates. Standard economic models have difficulty with this because they assume that information is processed accurately and that beliefs update in response to evidence. But humans are not belief-updating machines. We are social animals who use beliefs instrumentally, who inherit belief systems as packages rather than evaluating them item by item, and who pay costs for deviating from group beliefs that often exceed any benefit from accuracy. Understanding why irrational beliefs persist requires understanding not the cognitive failures that produce them but the social and economic functions they serve.

The Insurance Function of Ritual

The most fundamental function of superstitious belief is insurance against decision paralysis in conditions of genuine uncertainty. This sounds paradoxical — how does adding a false belief help anyone make better decisions? — but the logic is straightforward once you stop thinking about belief accuracy and start thinking about decision costs.

Imagine you are a medieval farmer trying to decide when to plant. You know from experience that planting too early risks frost and planting too late means you miss the optimal growing window. You don’t have a thermometer, a long-range weather forecast, or any way to measure soil temperature with precision. What you have is a set of traditional indicators: the flowering of specific plants, the behavior of particular animals, the position of stars, the date of certain saints’ days. Some of these indicators have genuine predictive value. Others are folk correlations that happen to track with real variables — a saint’s feast day that falls in mid-April is a reasonable approximation of average last-frost date in many European climates. And some are simply false but functionally useful as coordination devices: if everyone in your village plants on the same day because of a shared ritual calendar, you can pool labor, synchronize planting and harvest, and avoid the conflicts that arise when people are doing incompatible agricultural activities in shared fields.

The farmer who refuses all superstitious planting indicators on rational grounds and tries to calculate optimal planting date from first principles is in a worse position than the farmer who follows the village calendar. Not because the village calendar is accurate in any deep sense, but because the coordination benefits of shared belief exceed the costs of any inaccuracy in the belief’s specific content. This is the insurance function: ritual and superstition provide a cheap decision rule in conditions where calculating the optimal decision is prohibitively expensive.

Why Smart People Believe False Things

The cognitive science literature on superstition has spent decades documenting the specific mental shortcuts that produce superstitious beliefs — pattern recognition in random sequences, the clustering illusion, confirmation bias, the availability heuristic. These are genuine cognitive phenomena, and they do contribute to the persistence of false beliefs. But the research program has a systematic blind spot: it treats superstition as a cognitive failure to be explained away rather than as a behavioral strategy to be understood.

Consider the famous studies of basketball players and the “hot hand” — the belief that a player who has made several consecutive shots is more likely to make the next one. The original 1985 analysis by Gilovich, Vallone, and Tversky concluded that the hot hand was a cognitive illusion: sequential shots are statistically independent, and the apparent clustering of makes was within chance variation. For twenty years this was cited as the definitive debunking of superstitious belief in sport. Then in 2015, economists Joshua Miller and Adam Sanjurpe identified a statistical error in the original analysis — the Gilovich team had used a biased estimator that would find no hot hand even if one existed. Subsequent reanalysis found robust evidence for hot-hand effects in several sports contexts.

The point isn’t that the hot hand is definitely real. The point is that the confident debunking of a widespread belief turned out to rest on methodological error, while practitioners — coaches, players, sports bettors — who believed in the hot hand were tracking something real even if their causal model of what was happening was wrong. The folk belief was a better model of reality than the academic debunking. This is not an unusual situation. People who work intimately with complex systems — farmers, fishermen, traders, clinicians — frequently develop folk theories that are empirically superior to the formal models of outside experts who have studied the system less. The folk theory is superstition from the outside. From the inside, it’s compressed expertise.

The Social Enforcement Problem

Superstitious beliefs do not persist solely because they are useful. They persist because they are enforced. Every functioning social group has mechanisms for punishing deviance from group beliefs, and these enforcement mechanisms apply to factual beliefs as well as to behavioral norms. The person who publicly announces that they don’t believe in Friday bad luck and sails on Fridays to prove a point is not just making a private choice about scheduling. They are signaling a willingness to deviate from group norms, which triggers social sanctions that have nothing to do with ships or weather.

This enforcement dynamic explains why superstitious beliefs are so resistant to evidence. You might think that demonstrating repeatedly that Friday sailings go fine would undermine the belief. In practice, it doesn’t. The person who successfully sails on Fridays is subject to motivated reinterpretation: they got lucky, or the rule applies to other circumstances, or they’re the kind of person who courts disaster and will eventually pay for it. The confirmation bias at work here is not fundamentally cognitive — it is social. People have much more to lose from updating their beliefs in a way that marks them as deviant than from ignoring evidence that would require such updating.

This analysis has uncomfortable implications for how we think about the spread of scientific knowledge. The naive model of scientific communication assumes that evidence against false beliefs, when clearly presented, will gradually update those beliefs. The social enforcement model predicts that this will only work when the social costs of updating are low — when abandoning a false belief doesn’t mark you as a traitor to your community. When beliefs are identity-constitutive — when believing X is part of what it means to belong to group Y — evidence against X will be processed as an attack on group Y, not as information about X. The cognitive resistance to vaccination skepticism, climate denial, or young-earth creationism is not primarily about cognitive capacity. It’s about the social cost of updating.

When Irrational Beliefs Become Self-Fulfilling

The most economically interesting category of superstitious belief is the self-fulfilling kind — beliefs that create the outcomes they predict by changing behavior. Financial markets are saturated with these. Technical analysis — the practice of predicting future price movements from chart patterns — has no reliable predictive value when tested against market outcomes over long periods. Chart patterns are, in general, noise. But if enough traders believe that a particular chart pattern predicts a price breakout, and they all buy when they see that pattern, the resulting buying pressure will create the breakout. The prediction comes true not because the pattern had genuine predictive value but because belief in the pattern was widespread enough to make it a coordination device.

This self-fulfilling dynamic operates far outside financial markets. Fashion is largely driven by self-fulfilling predictions: a designer or editor predicts that a certain look will be popular next season, this prediction circulates within the fashion industry, and enough people act on it that the prediction comes true. The prediction had no epistemic basis — nobody knows what will be fashionable — but the social authority of the predictor made it self-fulfilling. Academic prestige works similarly: departments that are believed to be excellent attract better graduate students and faculty, producing better research, confirming the belief in their excellence. The initial reputation was often based on historical accidents, but once established, it tends to self-perpetuate.

The Schelling-point analysis of focal equilibria explains why specific superstitious beliefs become established rather than others. If a coordination device is needed — a day to rest, a practice for beginning important activities, a ritual for sealing agreements — the specific content of the device matters less than its universal recognition within the relevant community. Any day could be the Sabbath; the important thing is that everyone knows which day is the Sabbath. Any handshake could be the deal-sealing gesture; the important thing is that everyone knows what it means. Once a belief is established as the focal point for coordination, its specific content becomes irrelevant to its function, and the belief can persist long after the original reason for choosing it has been forgotten.

The Rationality of Irrationality

The conclusion that emerges from taking the economics of superstition seriously is not that humans are irrational and can’t help it. It is that what looks like irrationality at the individual level is often rationality at the social level, and that the two can diverge substantially. A hunter-gatherer who performs pre-hunt rituals is not being stupid. They are maintaining the social cohesion and psychological preparation of the hunting group, coordinating shared attention, and signaling commitment to the group enterprise — all at extremely low cost. The ritual’s cosmological claims might be empirically empty, but its social functions are real.

This reframing has practical consequences. Campaigns to eliminate superstitious beliefs by replacing them with accurate information will fail unless they simultaneously address the social and coordination functions those beliefs serve. Vaccine skepticism cannot be reduced by better information campaigns alone because the beliefs aren’t primarily about vaccine safety — they are about group identity, distrust of institutions, and the social costs of updating. Agricultural development programs that dismiss traditional farming practices as superstition routinely destroy functional coordination mechanisms that the farmers couldn’t have articulated but were relying on. Public health campaigns that ridicule food taboos without understanding what nutritional or social functions those taboos serve predictably fail to change behavior.

The harder and more productive question is never “why do people believe false things?” The harder question is “what functions does this belief serve, and what would need to be true before the belief became more expensive to hold than to abandon?” Edward Lloyd’s sailors wouldn’t stop believing in Friday bad luck until sailing became sufficiently routinized, controlled, and de-skilled that the coordination functions of the superstition were no longer valuable. That took industrial technology and professional naval organization, not debunking. The superstition didn’t yield to evidence. It yielded to a world in which evidence could finally be cheaply processed by instruments and institutions rather than by the shared folk knowledge of a crew. That is, characteristically, how irrational beliefs end: not with a refutation, but with a structural change that makes the belief’s function obsolete.