A study of one brokerage house’s recommendations to its individual customers during the 1960s suggests that they were genuinely valuable, even after allowing for transactions costs and risk. On the other hand, the recommendations were useful in selection, rather than in market timing: The ratio of buys to sells varied little over the period studied.
Abnormal returns were associated primarily with buy, rather than sell, recommendations. Positive in the six months prior to the recommendation, they peaked in the month of recommendation and remained essentially zero thereafter. One possible explanation is that, in the months prior to being recommended, companies enjoyed abnormal prosperity accompanied by a series of favorable news items. Their prosperity caught the attention of the brokerage house, whose research staff then uncovered additional positive news, encapsulating in recommendations to customers what might otherwise have been several more months of slowly emerging information.
If large positive returns in the month of the recommendation were merely the result of trading pressure induced by the recommendation, those returns would have been followed by reversals. The absence of such reversals suggests that the brokerage house’s recommendations were associated with genuine changes in the value of the securities.