The disposition effect describes investors' tendency to sell winning positions too early while holding losing positions too long. An investor who holds a stock down from $50 to $40 often refuses to sell, waiting to break even instead. Shefrin and Statman (1985) found redemption ratios of 0.93 in gain months versus 0.74 in loss months across 252 months of Broker/Dealer data (1961–1981). Their paper, "The Disposition to Sell Winners Too Early and Ride Losers Too Long," identifies four drivers: prospect theory, mental accounting, regret aversion, and self-control.
What the Study Found
In the Schlarbaum et al. individual investor data (1964–1970), approximately 40% of all realizations were losses across every round-trip duration category. The gain-realization ratio was 0.58 for 0–1 month holds, 0.57 for 2–6 months, and 0.59 for 7–12 months. In Broker/Dealer mutual funds (month t−1 specification), the mean redemption ratio was 0.93 in capital-gains months versus 0.74 in capital-losses months. The t-statistic was 1.69, significant at the 0.05 level — the only significant result across all 9 fund-type and timing combinations tested. No-Load and Direct-Seller funds showed directionally similar but insignificant differences (t-statistics ranging from −0.09 to 1.62).
Methodology
The individual investor stock data came from Schlarbaum, Lewellen, and Lease, covering brokerage panel trades from 1964 to 1970. Mutual fund data came from the Investment Company Institute: monthly purchases and redemptions for three fund types, January 1961–December 1981 (252 months). No-Load fund data ended in December 1973; post-1973 money market funds used accrual accounting, making loss realization structurally impossible. The study partitioned individual trades by round-trip duration (0–1, 2–6, 7–12 months) and compared redemption ratios across 30 highest-gain and 30 highest-loss months.
Key Statistics
| Metric | Finding | Context |
|---|---|---|
| Gain-realization ratio, 0–1 month round-trips | 0.58 | Schlarbaum et al. individual investor data, 1964–1970 |
| Gain-realization ratio, 2–6 month round-trips | 0.57 | Schlarbaum et al. individual investor data, 1964–1970 |
| Gain-realization ratio, 7–12 month round-trips | 0.59 | Schlarbaum et al. individual investor data, 1964–1970 |
| Mean redemption ratio, Broker/Dealer funds, capital-gains months (month t−1) | 0.93 | 30 highest-gain months, January 1961–December 1981 |
| Mean redemption ratio, Broker/Dealer funds, capital-losses months (month t−1) | 0.74 | 30 highest-loss months; t = 1.69, p < 0.05 |
| Mean redemption ratio, Direct-Seller funds, capital-gains months (month t−1) | 0.92 | 30 highest-gain months, January 1961–December 1981 |
| Mean redemption ratio, Direct-Seller funds, capital-losses months (month t−1) | 0.66 | 30 highest-loss months; t = 1.62 (not significant) |
| t-test for redemption ratio difference | t = (X̄_G − X̄_L) / sqrt((N_G·S_G² + N_L·S_L²) / (N_G + N_L − 2)) · sqrt(N_G·N_L / (N_G + N_L)) | N_G = N_L = 30; tests H₀: X̄_G = X̄_L against H₁: X̄_G > X̄_L (Table II) |
Why This Matters
The disposition effect challenges purely tax-based models of investor behavior. Even when realizing a loss would be optimal under Constantinides' normative strategy, investors systematically avoid it. The framework integrating prospect theory, mental accounting, regret aversion, and self-control offers a richer explanation of observed trading patterns than rational models alone. December loss concentrations, the paper argues, reflect self-control deadlines rather than rational optimization, with implications for tax-loss harvesting and year-end return seasonality.