The disposition effect describes investors' tendency to sell winning stocks too early while holding losing stocks too long. In Are Investors Reluctant to Realize Their Losses?, Odean (1998) analyzed 10,000 discount brokerage accounts from 1987 to 1993. He found investors realized gains at a 14.8% rate versus 9.8% for losses—a ratio of more than 1.5 to 1.
What the Study Found
The proportion of gains realized (PGR) was 14.8% for the full 1987–1993 period. The proportion of losses realized (PLR) was 9.8%, a difference significant at t > 35. From January through November, the gap was even wider: PGR 15.2% versus PLR 9.4% (t = 38). In December, the pattern reversed—PLR rose to 12.8% and PGR fell to 10.8% (t = 4.3). Sold winners subsequently outperformed retained paper losses by 3.4 percentage points over the following year (p = 0.001).
Methodology
The dataset consists of trading records for 10,000 randomly selected accounts at a nationwide U.S. discount brokerage. The sample spans January 1987 through December 1993 and includes 97,483 transactions from 6,380 accounts with CRSP-matched stocks. PGR and PLR count realized and unrealized gains and losses on every sale day in a multi-stock portfolio. Controls include partial-sale exclusions to filter rebalancing, no-repurchase exclusions to filter diversification trades, and price and return partitions to isolate transaction-cost effects.
Key Statistics
| Metric | Finding | Context |
|---|---|---|
| PGR (Proportion of Gains Realized) | 14.8% | Full year, 1987–1993, all accounts |
| PLR (Proportion of Losses Realized) | 9.8% | Full year, 1987–1993, all accounts; t > 35 |
| PGR/PLR ratio | ~1.5× | Full year; a winner is 50% more likely to be sold than a loser |
| PGR/PLR ratio in January | 2.1× | Highest monthly ratio; declines to 0.85× by December |
| PGR/PLR ratio in December | 0.85× | Below 1.0; tax-loss selling reverses the bias |
| PLR – PGR in December | +2.0 pp | t = 4.3; only month PLR exceeds PGR |
| 1-year excess return: sold winners vs. paper losses | +3.4 pp | p = 0.001; winners sold outperform losers held |
| 2-year excess return: sold winners vs. paper losses | +3.58 pp | p = 0.014 |
| PLPA (Proportion of Losses Purchased Again) | 13.5% | vs. PGPA 9.4% for winners; t = 19 |
| Average account PGR (account-level test) | 0.57 | 1,893 accounts equally weighted; mean PGR – PLR = 0.21, t = 19 |
| Average account PLR (account-level test) | 0.36 | 1,893 accounts equally weighted |
| Formula: PGR | Realized Gains ÷ (Realized Gains + Paper Gains) | Equation (1); measures gain realization rate |
| Formula: PLR | Realized Losses ÷ (Realized Losses + Paper Losses) | Equation (2); measures loss realization rate |
Why This Matters
The disposition effect imposes a direct cost on taxable investors: selling winners rather than losers generates capital gains tax liability that could be deferred. At the aggregate level, the effect may stabilize prices near historical purchase levels by reducing the supply of sellers below investors' reference points. Selling the loser instead of the winner could raise an investor's annual return by approximately 4.4%. The bias holds across frequent and infrequent traders, the full sample period, and multiple price and return partitions.