Kahneman and Tversky on losses
📚SOURCE
💎PERMANENT NOTES
🧵SYNTHESIS
Paraphrase, not quotation: 🔬 Prospect Theory - An Analysis of Decision under Risk, with the algebra skimmed.
The economics I was taught says a rational person weighs outcomes by their probability and cares about final wealth. This paper shows, with a series of simple choice problems, that people do neither, and offers a description of what they do instead.
Two ideas carry it. First, outcomes are felt as gains and losses from a reference point, not as states of wealth. The reference point is usually where you stand now, and it can be moved by how a choice is worded. Second, the curve that turns money into felt value is steeper for losses than for gains, and flattens in both directions. So a loss hurts more than an equal gain pleases, people play safe when choosing between gains, and they gamble when choosing between losses.
They add that we overweight small probabilities and prize certainty. That explains insurance and lottery tickets with one curve.
✨KEY IDEAS
People take a sure gain over a larger probable one, and take a probable loss over a smaller sure one.
What is valued is the change from a reference point, and the reference point can be shifted by description.
The value curve is steeper below the reference point than above it.
A reduction of risk from small to zero is worth more to people than an equal reduction elsewhere on the scale.
People strip out what two options share and compare what is left, so the same choice split differently gets a different answer.
📝READING NOTES
Why I read it after the workshop and not before
I taught 🧩 Loss aversion at Brightwater on Friday from memory and from the chapters of Kahneman's book I have reached. On Saturday it nagged at me that I had never read the paper I was citing. So this is a check on what I said. I got the idea right and one attribution wrong.
What the paper does
- Shows the standard theory failing. A set of paired gambles, mostly put to students and university staff as hypothetical questions. Majorities choose in ways that no consistent weighting of wealth and probability can produce.
- Proposes two phases. Editing: we simplify the options, set a reference point, cancel shared parts. Evaluation: we weigh what remains.
- Gives two curves. A value function over gains and losses, kinked at the reference point. A weighting function that turns stated probabilities into decision weights, too heavy at the small end.
What I used on Friday
🤝 Brightwater Studio - pricing workshop. Their question was pricing; the detail is in the client note. The three points I leaned on:
- A price rise is a loss from the current price. A new, larger package at a new price has no reference point to lose from.
- Removing something from a plan hurts more than its price would suggest. Do not take things away; stop adding them to the cheaper tier.
- Cancel any time is a certainty effect: it removes a small risk completely.
I said losses hurt about twice as much, Kahneman and Tversky, 1979. The 1979 paper argues for the asymmetry and sketches the curve. The number is from their 1992 follow-up. Nobody noticed. I have corrected the slide.
My doubts
- Hypothetical choices. Nobody lost anything. The authors defend this, and later work with real stakes broadly supports them, but the 1979 evidence on its own is thinner than its reputation.
- The reference point does all the work and the paper does not say how it gets set. In pricing that is the entire question.
- It describes; it does not say what I ought to do. Knowing I am loss-averse has not made me less so.
Closer to home
The sabbatical fund. I can move 900 euros into it without a flicker and the thought of spending 900 from it produces something like vertigo. Same money. The reference point is the balance. Two sentences on that became 💎 I fear a loss more than I want the same gain, which is still a seed.
Added 21 December: did the sum on the fund today. The seed has not grown, but I understand the vertigo better. Leaving the status at processing until I write the second half.
Questions it left me with
- Is my reluctance to spend the fund loss aversion, or is it ordinary prudence that I am dressing up in a famous name?
- Where does a reference point come from? If it is habit, then the balance of the fund becomes harder to spend every month it grows.
- In the workshop I used this to help someone raise prices. Is there a line between framing an offer and exploiting a bias, and did I stay on the right side of it?
Still To Do
- Done: Correct the attribution on the workshop slide
- Still open: Read the 1992 follow-up for the actual number
- Still open: Write the second half of the permanent note
Reading record
- Read for: work
🗣️IN SHORT
- In : Prospect Theory: An Analysis of Decision under Risk (1979)
- Kahneman and Tversky argue that : people evaluate risky choices as gains and losses from a reference point and not as final wealth, and weigh losses more heavily than gains
- by stating that : observed choices violate expected utility in consistent ways, which a value function that is kinked at the reference point and a weighting function that distorts probabilities can reproduce