The Reference Point Decides

Expected utility theory evaluates an outcome as a state of the world: you end up with this much total wealth, which is worth this much to you.

Kahneman and Tversky's observation, published in 1979 in the most cited paper in the history of economics, was that this is not how anybody thinks. People evaluate changes (gains and losses relative to wherever they currently are) and not final states.

That sounds like a technicality. It is the whole ballgame, because it makes the reference point decisive, and the reference point is not fixed by the facts. It can be set by the status quo, by an expectation, by what you were told, or by whatever the person describing the situation chose.

Three components follow, and each is separately testable.

The first is loss aversion. Losses loom larger than equivalent gains, by a factor usually estimated at around two to one. This is why most people refuse a coin flip that wins £110 or loses £100, even though the expected value is positive and the amounts are small enough that diminishing marginal utility explains nothing. It is why negotiations stall over concessions that are arithmetically symmetrical, and why any change with concentrated losers and diffuse winners is politically harder than the sums suggest.

The exact coefficient is genuinely debated, and there are domains where it seems weaker. The direction of the asymmetry is not in dispute.

The second is the shape of the value function: concave in gains and convex in losses. Concave in gains means the second thousand pounds of profit thrills you less than the first, producing risk aversion when you are ahead. Convex in losses means the second thousand of loss hurts less than the first, producing risk seeking when you are behind, which is why people double down to get back to even, and why a trader sitting on a loss holds it and sells the winner.

The third is probability weighting. People overweight small probabilities and underweight moderate ones. That single function produces the lottery ticket and the insurance policy in the same person, without needing a wiggly utility curve, and it produces a specific and consequential failure: risks around one in a thousand get treated as either roughly one in ten or as effectively zero, with very little in between. That is precisely the range most safety and catastrophe policy lives in.

Then there is the effect that turned prospect theory from a laboratory finding into something you can watch happen.

Kahneman, Knetsch and Thaler gave half a class of students mugs and set up a market. Sellers, on average, would not part with a mug for less than about seven dollars; buyers would not pay more than about three. Standard theory says the mugs should end up with whoever values them most and roughly half should trade. Far fewer did.

The endowment effect: owning something raises its value to you, because giving it up is coded as a loss against the new reference point. It has been demonstrated with mugs, pens, lottery tickets, sports tickets and hunting licences, and it predicts something practical: that free trials convert well not because people discover they like the product but because they now own it.

The cleanest demonstration of how much the frame does is the Asian disease problem.

Participants are told six hundred people face a disease outbreak. In one version, programme A saves two hundred for certain and programme B has a one-third chance of saving everyone and a two-thirds chance of saving nobody. Most choose A.

In the other version, programme C means four hundred die for certain and programme D has a one-third chance that nobody dies and a two-thirds chance that six hundred die. Most choose D.

A and C are the same programme. B and D are the same gamble. The only difference is whether the outcomes are described in lives saved or lives lost, which sets the reference point at zero saved or at six hundred alive, putting the chooser in the concave region where certainty is attractive, or the convex region where a gamble to avoid a sure loss is attractive.

Whoever writes the description chooses the frame, and there is no neutral way to write it.

Which is the point at which this stops being academic. The American credit card industry lobbied successfully for merchants to be permitted a "discount for cash" rather than a "surcharge for card" Arithmetically identical, and one is a forgone gain while the other is an imposed loss. They understood exactly what they were buying.

Every price that is presented as a saving, every fee that is bundled rather than itemised, every subscription that renews by default is a decision about the reference point. Once you have the concept, you cannot stop seeing it, and that is probably the most useful thing prospect theory offers a reader who is not an economist.