Research indicates that investors will repurchase familiar stocks that are easily recalled, such as those in the news (attention bias) or those that previously resulted in a profit (a simple learning approach based on a pleasant experience). The authors expand this literature and test whether investors are more likely to repurchase their most recently sold stock rather than some other previously owned stock. They find that a recent sale significantly lowers the likelihood that the investor will repurchase another stock, even if that other stock was traded for a profit (evidence of a sequential learning process). The authors determine that these repurchasing decisions can result in inferior portfolio strategies.

This research highlights that investors' repurchasing activity could limit their investment opportunities. The authors find that a recent roundtrip trade in a stock decreases the propensity to repurchase any other previously sold stock by 23%. A prior profit on an older trade increases the propensity to repurchase that stock by an insignificant 1%.

In addition, the authors’ research suggests that repurchasing is a suboptimal trading activity. The final sales of repurchased stocks do not display any positive abnormal returns, and repurchasing involves additional commissions, which significantly decrease performance. They note that sophisticated investors and active traders repurchase less often and have more diversified portfolios than naive general traders. Investors could improve their performance by broadening their choices for stocks and by forgoing their trading activity.

The dataset is composed of 1.9 million trades made by approximately 80,000 households at a large discount brokerage house from January 1991 to November 1996. The authors find that 40% of the households engaged in at least one repurchase and repurchases accounted for 80% of all purchase trades. Households are categorized according to their self-reported knowledge of investing and their level of trading activity. Investors who rate themselves as knowledgeable repurchase less often than the less knowledgeable, and active traders repurchase less often than less active naive traders.

Investors display a preference for repurchasing stocks with large positive returns on the previous day and a few large-cap stocks in a few industries. Of the repurchases, 36% are in the technology sector and about 14% are in the pharmaceutical and retail sectors. These stocks are easy for investors to recall because of wide analyst coverage, heavy advertising, or news headlines.

The authors construct probit regressions with dummy variables for various stock and investor characteristics. The dummy variables have a value of 1 if the characteristic is present and 0 if it is not. In this way, the authors test repurchased stocks for characteristics that might proxy for a high degree of attention and test for types of investors who engage in repurchases.

Although there is a marginally significant investor preference for repurchasing stocks that the investor previously sold for a profit and for repurchasing stocks that were in the previous day’s highest-return decile, the most significant result is the recency effect. The authors find no evidence that the repurchases are informed trades. Using several different formulas to isolate alpha, the authors find no positive abnormal returns on the final roundtrip of a repurchased stock. In addition, they explain how the observed returns can be overstated.

The authors measure the diversification of the portfolios of repurchasing households. The data indicate that investors who repurchase more often tend to have less-diversified portfolios—fewer stocks with higher average correlations in fewer sectors. Sophisticated investors and active traders have more diversified portfolios and hold more stocks on average than naive general traders (8 as opposed to 3).

The research is very interesting and well written and includes thorough performance and diversification measures. The authors cite a study indicating that volatile stocks may be conducive to informed trading. They form a variable for high idiosyncratic volatility and find that it marginally reduces the recency effect. Isolating the performance of the repurchased volatile stocks could be a very interesting extension of this work.