For two centuries, engineers have understood a simple truth about stability: any system left unchecked will eventually destroy itself. James Watt's centrifugal governor, invented in 1788, solved this problem for steam engines by creating a closed loop of negative feedback. When the engine raced too fast, weighted arms rose, throttling the steam supply. When it slowed, the arms fell, opening the flow. The machine regulated itself.
Human beings, unfortunately, come with no such governor.
This is not merely a poetic observation. It is the central finding of behavioral economics, a field that has spent four decades systematically dismantling the rational-agent model upon which classical economics was built. The evidence is now overwhelming that people make large, predictable, and expensive mistakes. The question is no longer whether we err, but whether we can engineer our own salvation—designing policies, products, and personal strategies that function as external governors for the flawed cognitive machinery inside our skulls.
The 3.4 Per cent Problem#
Consider the humble individual investor. Researchers examining large brokerage accounts have tracked what happens when a person sells one stock and buys another on the same day—a clear signal that the investor believes the new purchase will outperform the old holding. The results are a quiet catastrophe. One year later, the stock that was sold has, on average, outperformed the stock that was bought by 3.4 per cent.
This is not a rounding error. Compounded over a lifetime, the "cost of having an idea" becomes a staggering drag on wealth. Yet the behaviour persists, driven by a cluster of cognitive biases that no amount of market education seems to cure.
The most potent of these is overconfidence. Psychological research is unequivocal: confidence is not a reliable indicator of accuracy. It is primarily a feeling about the coherence of the story one tells oneself. A plausible narrative generates confidence regardless of the quality of the underlying information. And because the stock market is an irregular environment that offers slow, noisy feedback, genuine expertise is nearly impossible to develop. The investor who feels certain is almost certainly deluded.
This delusion expresses itself most destructively in the disposition effect—the systematic tendency to sell winning stocks while holding onto losers. The logic is purely emotional: selling a winner feels like a success; selling a loser requires acknowledging failure. People choose pleasure over pain, even when that choice is financially irrational. The original purchase price, which a rational investor would treat as irrelevant, becomes an anchor that distorts every subsequent decision.
The Shadow of Loss#
Behind these specific errors lies a deeper asymmetry. Behavioural economists have established that the pain of a loss is roughly twice as intense as the pleasure of an equivalent gain. This is loss aversion, and it is the single most important feature distinguishing behavioural from standard economics.
Its consequences are bizarre. Offer someone a coin toss where losing costs 1,000 euros, and they will typically demand a potential gain of more than 2,000 euros before accepting the gamble. Rejecting a 50-50 bet with an expected value of 1,500 euros is mathematically irrational, yet most people do exactly that. They view each gamble in isolation—a cognitive error known as narrow framing—rather than as part of a long sequence where occasional losses are inevitable.
A fully rational agent would adopt a policy: accept all positive-expected-value gambles and let the law of large numbers work its magic. But human beings are not rational agents. They are loss-averse, myopic creatures who feel each loss as a wound.
Time Inconsistency and the Ipecac Solution#
If loss aversion distorts decisions across space, time inconsistency distorts them across time. Standard economics assumes exponential discounting—a consistent, steady rate at which future value depreciates. Humans, however, use hyperbolic discounting: we value today vastly more than tomorrow, but treat next month and next year with near-equivalence.
This generates the classic self-control failure. Ninety per cent of smokers say they want to quit. Ninety per cent fail. They wake up, feel terrible, resolve to quit tomorrow, and have one more cigarette today. The same pattern holds for dieting, exercising, and saving for retirement.
The most profound evidence of this irrationality is not found in laboratories but in the desperate lengths people go to constrain their future selves. Smokers make punitive financial bets with friends. Dieters throw away junk food before a moment of weakness. And in the most striking example, some alcoholics drink syrup of ipecac—a medicine that induces violent vomiting if alcohol is consumed later. In a rational world, such self-punishment would be unnecessary. Its existence proves that people know their own willpower is unreliable and are willing to suffer to enforce their long-term goals.
Engineering the Governor#
If individuals cannot govern themselves, can institutions do it for them? The answer, increasingly, is yes—but only if policy abandons the fiction of the rational agent.
Consider cigarette taxation. Traditional models, which consider only "externalities" (the costs smokers impose on others), produce an optimal tax of roughly 50 cents per pack. But smokers inflict enormous harm on themselves—the "internality." Each cigarette shortens life by about seven minutes, imposing health costs of roughly $40 per pack. Because time inconsistency proves that smokers are making a cognitive mistake rather than a rational choice, society can justify taxing the behaviour to correct the error. When internalities are included, the optimal tax rises to $10 per pack.
Even more elegant is the nudge—a change in choice architecture that steers behaviour without restricting options. In a famous study of corporate 401(k) plans, changing the default from opt-in to opt-out raised participation among young workers from 20 per cent to 80 per cent. No choices were banned. No incentives were changed. A single tweak to the paperwork secured the retirement futures of thousands of people.
The Therapeutic Role of Advisors#
For individuals who cannot rely on paternalistic policy, the solution lies in building personal governors. The primary role of a financial advisor, contrary to popular belief, is not to pick winning stocks. It is therapeutic and educational: to press clients toward broad framing, encourage them to trade less, and—crucially—to check their results less frequently. Frequent tracking triggers emotional reactions and promotes unnecessary policy changes, which reliably produce worse outcomes.
Anticipating regret is another powerful tool. Discussing volatility and potential losses in advance acts like a vaccine, reducing the severity of the reaction when losses inevitably occur. The investors who fare best are not those who avoid losses but those who accept them as the cost of a sound policy.
The Architecture of Freedom#
The integration of psychology into economics forces a reckoning with what freedom actually means. For decades, the assumption of perfect rationality provided a convenient excuse for inaction. If every choice was mathematically optimal, then poverty, addiction, and chronic disease were simply the market expressing individual preferences.
Behavioural economics reveals this as a comforting illusion. Our choices are unstable, heavily influenced by framing, and constantly undermined by our hyperbolic desire for immediate gratification. We are not calculating machines. We are flawed, loss-averse creatures who need external structures to achieve our own internal goals.
When policymakers design defaults that automatically enrol workers in retirement plans, or when they tax cigarettes to offset myopia, they are not overriding free will. They are acting as architects of choice—designing the cognitive environment so that our better angels have a fighting chance against our weaker impulses. Just as Watt's governor maintains engine speed by adjusting to fluctuations while preserving the long-term goal, well-designed institutions can maintain human welfare by compensating for our predictable errors.
The alternative is to continue pretending that human beings are rational. That pretence has a cost. It is roughly 3.4 per cent per trade, a lifetime of forgone retirement savings, and millions of cigarettes smoked by people who desperately wish they could stop. The engine of human behaviour will never run smoothly on its own. The question is whether we have the wisdom to build the governor it so plainly needs.
References#
Deng, S., Ji, J., Wen, G., & Xu, H. (2020). Delay-induced novel dynamics in a hexagonal centrifugal governor system. International Journal of Non-Linear Mechanics, 121,.
Di Gennaro, S. (2023). Fundamentals of linear systems: analysis and control. Instrumentation and Control Systems for Nuclear Power Plants,.
Goldstein, D. (1989). Chapter 22 Control Theory Applied to Stress Management. Advances in Psychology, 62.
Huebner, L., & Zacher, H. (2023). The role of mean item ratings, topic distance, direct leadership, and voice climate in action planning after employee surveys. Acta Psychologica, 238.
Ormandy, R. (2024). From synapses to ephapsis: Embodied cognition and wearable personal assistants. Artificial Intelligence in the Age of Neural Networks and Brain Computing.
Sotomayor, J., Mello, L., & de Carvalho Braga, D. (2008). Stability and Hopf bifurcation in an hexagonal governor system. Nonlinear Analysis: Real World Applications, 9,.
Wang, P. (2000). Soft modeling for a certain class of intelligent and complex systems. Information Sciences, 123.
Wolf, M. (2017). Electronic Computers. The Physics of Computing.