What is the loss aversion coefficient, and how does the 2:1 ratio of losses to gains impact portfolio construction?

The loss aversion coefficient (lambda) quantifies the asymmetry in how individuals experience gains versus losses. Kahneman and Tversky's (1979) foundational research estimated lambda at approximately 2.25, meaning a $1 loss generates approximately 2.25 times the psychological disutility of a $1 gain's utility.\n\nProspect Theory Value Function:\n\nThe value function is defined as:\n- v(x) = x^alpha for gains (x >= 0)\n- v(x) = -lambda x (-x)^beta for losses (x < 0)\n\nwhere alpha = beta approximately equals 0.88 (diminishing sensitivity), and lambda approximately equals 2.25 (loss aversion coefficient).\n\nThe function is:\n- Concave for gains (risk averse)\n- Convex for losses (risk seeking)\n- Steeper for losses than gains (loss aversion)\n\nWhere Lambda = 2.25 Comes From:\n\nKahneman and Tversky calibrated lambda through experiments asking subjects to accept or reject gambles. The typical finding:\n\nA 50/50 gamble of winning $X or losing $Y is rejected unless X >= approximately 2.25 x Y.\n\nFor example, most people reject a coin flip between winning $100 and losing $50 (even though the expected value is +$25). They require winning approximately $112-$125 to accept a potential $50 loss.\n\nImpact on Asset Allocation:\n\n`mermaid\ngraph TD\n A[\"Loss Aversion λ ≈ 2.25\"] --> B[\"Investors demand higher
compensation for downside risk\"]\n B --> C[\"Equity Risk Premium
higher than rational models predict\"]\n B --> D[\"Under-allocation to equities
relative to mean-variance optimal\"]\n B --> E[\"Preference for guaranteed returns
even at lower expected value\"]\n C --> F[\"Mehra-Prescott Equity Premium Puzzle
partly explained by loss aversion\"]\n D --> G[\"Home bias, cash hoarding,
excessive bond allocation\"]\n E --> H[\"Popularity of structured products
with principal protection\"]\n`\n\nWorked Example — Equity Premium Under Loss Aversion:\n\nUsing the Benartzi-Thaler (1995) myopic loss aversion model:\n\nAssume an investor evaluates their portfolio annually. Historical equity returns have:\n- Mean annual return: 7%\n- Standard deviation: 16%\n- Probability of negative annual return: approximately 33%\n\nFor a loss-averse investor (lambda = 2.25), the prospective value of holding equities:\n\nPV = 0.67 x v(+7%) + 0.33 x v(-9%)\n= 0.67 x (7%)^0.88 + 0.33 x [-2.25 x (9%)^0.88]\n= 0.67 x 0.058 + 0.33 x [-2.25 x 0.074]\n= 0.039 - 0.055\n= -0.016 (negative prospective value)\n\nFor the loss-averse annual evaluator, equities have negative prospective value despite positive expected returns. They would only invest if the equity premium increases to approximately 6-8% — close to the historically observed premium.\n\nBy contrast, a rational expected utility maximizer with standard risk aversion would only require a premium of approximately 1-2%. This gap is the equity premium puzzle.\n\nPortfolio Construction Implications:\n\n1. Asset allocation tilt: Loss-averse clients allocate 15-25% less to equities than mean-variance optimization suggests\n\n2. Evaluation frequency matters: If the same investor checks their portfolio monthly instead of annually, the probability of observing a loss increases, and loss aversion intensifies the demand for safe assets. This explains why daily portfolio monitoring damages long-term returns.\n\n3. Product design: Financial products with capital protection features (structured notes, buffered ETFs, variable annuities) exploit loss aversion by eliminating the feared loss scenario, even when the expected return is significantly reduced.\n\n4. Disposition effect connection: Lambda > 1 directly creates the disposition effect: the pain of realizing a loss (v(loss) weighted by lambda) exceeds the pleasure of realizing an equivalent gain, causing asymmetric sell behavior.\n\nVariations in Lambda:\n- Professional traders: lambda approximately equals 1.5 (reduced but not eliminated)\n- Retail investors: lambda approximately equals 2.5-3.0 (higher than the experimental average)\n- During crisis periods: lambda can spike to 4.0+ as fear amplifies loss sensitivity\n\nMaster behavioral portfolio theory in our CFA course.