#INVESTOR BEHAVIOR
"A loss hurts twice as much as an equivalent gain." The reference meta-analysis puts the figure at 1.955. A re-analysis of the very same dataset drops it to 1.07 as soon as gains and losses are the same size.
Published on September 9, 2026
In June 2024 the Journal of Economic Literature published a meta-analysis by Brown, Imai, Vieider and Camerer: 607 estimates of the loss aversion coefficient, drawn from 150 articles published between 1992 and 2017, giving a mean of 1.955 and a 95% interval running from 1.820 to 2.102. In 2025 Yechiam and Zeif took that same dataset and sorted it by experimental protocol. In studies where gains and losses are the same size and are not presented in ascending order, the coefficient is roughly 1.07 and no longer distinguishable from 1.
How much more does a loss weigh than an equivalent gain?
About twice as much across the pooled literature, and about the same in protocols where the stakes are symmetric.
In other words: the most quoted coefficient in behavioural economics also measures how the question was asked.
Five caveats that belong with the number
- The widely quoted 2.25 does not come from Kahneman and Tversky 1979 but from their 1992 paper, based on 25 students.
- Both figures come from the same dataset: the second is a sort of the first by protocol, not a fresh collection of studies.
- That sort covers only 84 of the 150 articles, or 163 estimates and 149,218 participants; the rest did not document their protocol well enough.
- Loss aversion still shows up when the losses offered are smaller than the gains, or when amounts are presented in ascending order. The dispute is about how general it is.
- These coefficients come from lottery experiments, not from real saving decisions.
Why it still matters: a bias can be solid on average and vanish the moment the stakes are made equal.
For you
- When you are told you feel losses twice as keenly, where does that figure come from?
- Does a coefficient that moves with the presentation of amounts measure a preference, or a format?
Source: Brown, Imai, Vieider and Camerer, Journal of Economic Literature, June 2024; Yechiam and Zeif, Journal of Economic Psychology, 2025.
View the source →