Asset Pricing When Traders Sell Extreme Winners and Losers (Digest Summary)

The author builds on existing research in behavioral finance that shows that the transaction rate for shareholders of a company increases when the price changes by larger amounts. This effect is stronger for increases in price than for decreases, and in both cases, the selling rate changes by more than the buying rate. She graphically represents these facts as an asymmetrical V-shaped response to profit and loss. She shows the predictability of stock returns based on the magnitude of unrealized gains and losses and attributes this predictability to higher returns arising from a depressed stock price caused by this V-shaped net selling pressure.

Cross-sectional momentum in its simplest form has become a part of financial orthodoxy. The author highlights that some nonlinearity exists in the response to price action that is not fully consistent with this traditional view of momentum. This information can help shape practitioners’ understanding of how momentum originates and thus whether they should be looking at it in their process. It might also have implications in terms of the definition of momentum that is appropriate to a given application.

Portfolio managers might be inspired to reflect on the timing of their continued investment decision; a price action–agnostic process might result in more rational reevaluation of investment theses. The disposition effect is a subtle but well-evidenced feature of financial behavior that may be hard to identify in one’s own process.

The author begins by using brokerage data to establish the relationship between buying and selling and to establish a set of control variables, including the positive and negative returns to stocks in the dataset. Her results are consistent with prior literature and confirm the significance of the change in sales and purchases with respect to change in price, the asymmetry of that response, and the greater impact on sales. These results motivate her price impact hypothesis as well as the capital gains overhang and V-shaped net selling propensity that are later used as control variables.

The author establishes the relationship between these control variables and the subsequent stock performance by using double portfolio sorts and regression analysis. Both show that stocks with larger gains and losses have higher returns. The capital gains overhang effect is strongly seasonal (presumably because of year-end tax effects), and the effects are stronger for subsets of the dataset that are more speculative in nature—that is, smaller companies and those with lower institutional ownership. The author reports that the key results are robust to changes in treatment of stock splits and dividends, in universe selection, in frequency of analysis, and in control variable construction.

The author presents a thorough and detailed description of a simple and intuitive phenomenon along with strong support from wider literature. Perhaps the most interesting implication of this work is how it informs the assumed causes of the momentum effect in the cross section of stock returns. Behaviorally motivated selling as identified in this work contributes nonmonotonically, which leaves much to the momentum effect to be explained by purchases by new shareholders or by causes not corresponding to transactional price pressure. An understanding of the origins of equity factors is increasingly important as a tool to filter the growing number of identified pricing anomalies.