Loss Aversion

Loss aversion is a core concept in behavioral economics, formalized by Daniel Kahneman and Amos Tversky in their 1979 paper introducing prospect theory (Econometrica, "Prospect Theory: An Analysis of Decision under Risk"). It describes the empirical regularity that the psychological pain of losing a given amount is greater than the pleasure of gaining the same amount. In their experiments and subsequent work, Kahneman and Tversky estimated a "loss aversion coefficient" of roughly 2 to 2.25, meaning a $100 loss feels about twice as bad as a $100 gain feels good.

The concept departs from the expected-utility framework of classical economics, which assumes agents evaluate outcomes based on final wealth states rather than gains and losses relative to a reference point. In prospect theory, the value function is:

• Reference-dependent: outcomes are coded as gains or losses relative to a status quo or expectation.
• Concave for gains, convex for losses: producing risk aversion in the gain domain and risk-seeking in the loss domain.
• Steeper for losses than for gains: the defining feature of loss aversion.

Loss aversion underpins several documented phenomena, including the endowment effect (Thaler, 1980; Kahneman, Knetsch, and Thaler, 1990), the status quo bias (Samuelson and Zeckhauser, 1988), and the disposition effect in finance (Shefrin and Statman, 1985), where investors hold losing stocks too long and sell winners too early. Kahneman received the 2002 Nobel Memorial Prize in Economic Sciences in part for this work; Tversky had died in 1996 and was ineligible.

For political and IR researchers, loss aversion has been applied to foreign-policy decision-making by scholars such as Jack Levy and Rose McDermott, who argue leaders take greater risks to avoid losses (e.g., territorial concessions, prestige decline) than to achieve equivalent gains—helping explain escalation, sunk-cost behavior in wars, and resistance to negotiated settlements. Some replications, however, have found the 2:1 ratio is context-dependent and weaker for small or hypothetical stakes.