Kahneman and Tversky’s prospect theory They postulate that individuals have loss aversion. What is loss aversion?
This means that people experience losses more intensely than gains of the same magnitude; For example, the psychological impact of losing a certain amount of money is greater than the pleasure derived from winning that same amount. A key question is how much Do individuals experience gains more intensely than losses?
To formalize things, prospect theory assumes the following utility function:
The most cited estimates for these parameters come from Tversky and Kahneman (1992)In that paper, they found that loss aversion λ=2.25 and α=β=0.88. We can graph the utility function with this parameterization in the following graph as shown below.
A key issue, however, is that Tversky and Kahneman (1992) Estimates of loss aversion were obtained from a single study of 25 graduate students at an elite American university. To what extent can these results be generalized? Is there a better estimate of loss aversion?
A paper Brown et al (2024) aims to answer this question by conducting a meta-analysis of loss aversion estimates from all studies published between 1992 and 2017. They found 607 empirical estimates of loss aversion across 150 articles. The studies came from a variety of disciplines (e.g., economics, psychology, neuroscience) and a variety of data types. Most studies (53%) were based on a laboratory experiment design, but 26.5% of the identified articles came from a field experiment with other field data; 42% of the studies came from Europe and 30% from North America.
The unadjusted results (shown below) estimated a median loss aversion of 1.69 and a mean loss aversion of 1.97. After applying a random effects meta-analytic distribution, the mean loss aversion coefficient was found to be 1.955 with a 95% probability that the actual value is between 1,820 and 2,102.
These results are somewhat lower, but not very different from those Tversky and Kahneman (1992) Estimate 2.25. We can also compare the results with two previous meta-analysis studies on loss aversion. Neumann and Böckenholt 2014– which examined consumer aversion using 33 studies on consumer brand choice – reported a base model estimate of λ = 1.49 and an “enhanced model” estimate of λ = 1.73; Walasek, Mullett and Stewart (2018)– which examined 17 studies of financial lotteries with wins and losses – estimated that λ = 1.31. In summary, the results of Brown et al. They are higher than previous estimates, but lower than those of Tversky and Kahneman.
You can read the full article here.
Key references
- Brown, Alexander L., Taisuke Imai, Ferdinand M. Vieider, and Colin F. Camerer. “A meta-analysis of empirical estimates of loss aversion.” Magazine of economic literature 62, No. 2 (2024): 485-516.
- Neumann, Nico, and Ulf Böckenholt. 2014. “A Meta-Analysis of Loss Aversion in Product Choice.” Journal of Retailing 90 (2): 182–97.
- Tversky, Amos and Daniel Kahneman. 1992. “Advances in Prospect Theory: Cumulative Representation of Uncertainty.” Journal of Risk and Uncertainty 5(4): 297–323.
- Walasek, Lukasz, Timothy L. Mullett, and Neil Stewart. 2018. “A meta-analysis of loss aversion in risk contexts.” http://dx.doi.org/10.2139/ssrn.3189088.
