Risk aversion

• 1. Introduction

  ›

  Risk aversion is the tendency people have to prefer a sure thing over a gamble, even when the gamble pays the same on average. Imagine being handed a choice: take $50 right now, guaranteed, or flip a coin for $100 or nothing. Both options carry the same expected value. Yet most people reach for the certain $50. That preference reveals something deep about how human beings weigh uncertainty against reward. Why do people sacrifice potential gains just to avoid the unknown? How does that preference ripple through markets, government policy, children's playgrounds, and the bargaining table? And what happens to risk aversion when the stakes get very large, or when the losses replace the gains?

• 2. The Coin Flip That Explains Everything

  ›

  A guaranteed $50 against a coin flip for $100 or nothing is the simplest possible classroom for risk aversion. Both scenarios share an expected value of $50, meaning that a person completely indifferent to risk would treat them as identical. But people are not identical in how they respond.

  A risk-averse person would accept a guaranteed payment of less than $50, perhaps $40, rather than face the coin flip. A risk-neutral person would be indifferent between the certain $50 and the gamble. A risk-loving person would prefer the gamble even if the guaranteed offer rose above $50, say to $60.

  The figure that sits at the heart of this framework is the certainty equivalent: the smallest guaranteed dollar amount that leaves a person indifferent compared to a risky bet of the same average value. For a risk-averse individual, the certainty equivalent is smaller than the expected value of the gamble. The gap between those two figures is the risk premium. For risk-averse individuals, that premium is positive; for risk-neutral persons it is zero; for risk-loving individuals it is negative.

  This is not merely an abstract exercise. It tells us that a risk-averse person would willingly sacrifice money, in expected terms, to escape uncertainty. In the coin flip example, if a person's certainty equivalent is $40, their risk premium is $10. That $10 is what they are willing to pay to know in advance exactly what they will receive.

• 3. Utility and the Curved Function

  ›

  Expected utility theory gives risk aversion a mathematical home. Each possible dollar amount a person might receive maps onto a utility value through a function written as u(c), where c is the money or goods received. An agent is risk-averse if and only if that function is concave.

  Concavity means the function curves downward: each additional dollar adds less utility than the dollar before it. If u(0) equals 0 and u(100) equals 10, then u(50) might be 6 rather than 5. The midpoint of the dollar range does not land at the midpoint of the utility range, because the first dollars matter more than the later ones. Using those sample values, a 50-50 bet between $100 and $0 yields an expected utility of 5, which matches the utility of a certain $40. That makes $40 the certainty equivalent, and the risk premium comes out to $10, or 25% of the expected value of $50.

  Wealthier individuals tend to show a flatter curve over small amounts. For someone with substantial savings, the utility function across a modest range like $0 to $100 is nearly linear, which is why the same $100 gamble carries less psychological weight for them. The two defining properties of the utility curve for gains are an upward slope and concavity: more money is always better, and each additional unit of money matters less and less.

• 4. Measuring How Risk-Averse Someone Is

  ›

  Kenneth Arrow and John W. Pratt gave economists a tool for quantifying risk aversion beyond a simple yes-or-no. Their measure of absolute risk aversion, known as the Arrow-Pratt measure or the coefficient of absolute risk aversion, captures the curvature of the utility function in a way that holds up even when the function is rescaled or shifted.

  The measure is expressed in units of inverse dollars. A separate, dimensionless measure is the coefficient of relative risk aversion, which makes comparisons across different wealth levels and different currencies possible without conversion.

  Several families of utility functions have been developed around these measures. Constant absolute risk aversion, or CARA, describes functions where risk aversion does not change as wealth changes. Constant relative risk aversion, or CRRA, describes functions where risk aversion is stable as a proportion of wealth. Hyperbolic absolute risk aversion, or HARA, is the most general class typically used in practice and encompasses both CARA and CRRA as special cases, along with quadratic utility.

  Empirical evidence mostly supports decreasing absolute risk aversion: as wealth rises, people become less risk-averse in absolute dollar terms. One finding that complicates the picture, however, is that wealth is not a reliable proxy for risk aversion when studying risk-sharing contracts between principals and agents. Tests built on that assumption tend to produce unreliable results.

• 5. Portfolios, Bargaining, and the Brain

  ›

  Modern portfolio theory uses risk aversion to explain why investors hold combinations of assets rather than putting everything into whichever single asset promises the highest return. Risk is measured in this framework as the standard deviation of an investment's return, and a risk-averse investor demands a higher expected reward before accepting more of it. The risk-return spectrum follows directly from this dynamic.

  At the bargaining table, risk aversion tends to work against the person who has it. Studies drawing on both the von Neumann-Morgenstern framework and Nash Game Theory find that risk-averse individuals are willing to accept a smaller share of any deal than their less risk-averse counterparts. Because a risk-averse person's utility function is concave, each additional unit of gain matters less to them, so they settle sooner. Opponents prefer to face the most risk-averse person in a negotiation.

  Inside the brain, a 2009 study by Christopoulos and colleagues found that activity in the right inferior frontal gyrus correlates with risk aversion. Participants with higher risk premia, meaning they demanded more compensation to accept uncertainty, also showed stronger neural responses to the safer options. Experiments in which that brain region was artificially modulated caused participants to shift their choices in the direction of the modulation, becoming more or less risk-averse depending on whether activity in the area was increased or decreased.

• 6. Where Expected Utility Breaks Down

  ›

  Matthew Rabin's calibration theorem exposed a flaw in how expected utility theory handles small-stakes decisions. The theorem shows that a risk-averse person who consistently turns down a 50-50 gamble of losing $100 or gaining $110, starting from any level of wealth, would also refuse a 50-50 bet of losing $1,000 against gaining any finite sum of money. That implication, as Rabin pointed out, is implausible. No real person behaves that way, yet the math of diminishing marginal utility generates exactly that prediction.

  Prospect theory, developed by Daniel Kahneman and Amos Tversky, offers one resolution. Rather than measuring outcomes against total wealth, prospect theory evaluates gains and losses relative to a reference point, usually the status quo. That reframing allows the model to match observed behavior more closely.

  Kahneman and Tversky also identified what they called the reflection effect: a systematic reversal of risk preferences when the sign of the outcomes changes. Given a choice between a certain gain of 3,000 and an 80% chance of gaining 4,000, most people choose the certain gain. Posed the same problem in the domain of losses, most people prefer an 80% chance of losing 4,000 to a certain loss of 3,000. Risk aversion flips to risk-seeking when losses are on the table. The psychological mechanism behind this, according to Kahneman and Tversky, is the overweighting of certainty: options framed as certain are weighted more heavily than the probabilities alone would justify. Later research found that the reflection effect is most pronounced when either very small or very large amounts and extreme probabilities are involved.

• 7. Risk Aversion in Everyday Life

  ›

  John von Neumann and Oskar Morgenstern first laid out the mathematical framework for expected utility in their book Theory of Games and Economic Behaviour. Their model, the von Neumann-Morgenstern utility theorem, builds risk aversion into the structure of preferences itself rather than treating it as an add-on. If an individual's preferences satisfy four key axioms, a utility function reflecting those preferences can be derived directly.

  The model's power shows up in decisions that seem irrational under simpler frameworks. Consider a risk-averse person with $20,000 in savings who is offered a 30% chance of winning $100,000. The traditional expected value calculation suggests the gamble is worth taking. Yet many people in that position would refuse it, because the prospect of losing $20,000 in savings outweighs the mathematical appeal of the payoff. The von Neumann-Morgenstern model accommodates that refusal by assigning a specific utility to each outcome and incorporating the weight the individual places on keeping what they already have.

  Beyond formal finance, risk aversion shapes how governments write safety rules. Agencies like the Health and Safety Executive operate under mandates that are fundamentally risk-averse, requiring risks to be minimized even when that means forgoing the utility of a risky activity. That framing can distort outcomes: a law written purely around risk reduction may ignore the opportunity cost of not acting.

  Children's playgrounds are a concrete example. Many have been fitted with impact-absorbing matting intended to prevent death from direct falls. Critics note that the matting is expensive, which leaves fewer resources for other safety investments, and that children who know the surface is soft may attempt riskier behavior than they otherwise would. Sheila Sage, an early years school advisor, has said that children who are only ever kept in very safe places are not the ones who are able to solve problems for themselves. One experimental study using the television game show Deal or No Deal found that people behave more risk-aversely in front of a live audience, cameras, and a game show host than they do in a standard computerized laboratory setting, pointing toward how social visibility shapes financial decisions.