The herd instinct is a well-known aspect of group behavior in many social and economic contexts, and it is also reflected in various financial decisions. As it is based on an individual's conscious decision to imitate the behavior of others, it is very important to distinguish it from a behavior cluster similarity of individuals behavior that results from unexpected but parallel external factors. Such a cluster can arise for instance when investors or financial advisors (analysts) have access to general databases and use similar methods to analyze the data. In such cases individuals may be prompted to act in the same way and at the same time without being aware of each other's behavior. Choosing the same time to act in any way (such as analysts' recommending a particular share) resulting from interaction between the individuals does not necessarily lead to imitation of the content itself. The analysis presented in this study, based on a unique database of about 2,780 recommendations by analysts in the years 1999-2004 regarding shares traded on the Tel Aviv Stock Exchange, attempts to identify the herd instinct in the actual recommendations as reflected in the target prices forecast by the analysts. The results of the study show that it is essential to divide the analysts' recommendations by timing, because the timing alters the conclusions regarding the existence of the herd instinct in forecasting share prices. This study supports the theoretical forecast that indicates the possibility of two equilibriums, from the aspect of the timing of the publication of analysts' recommendations in an endogenous environment (in other words, when the order of the recommendations is not predetermined), by defining those equilibriums via empirical tests. Thus, recommendations of analysts made at times that are independent of the recommendations of other "neutral" recommendations) are over-optimistic, and the target price quoted in their forecasts deviates from the consensus by nine times the deviation derived from the model of rational expectations. Despite the inferiority of these recommendations, measured by the accuracy of the forecast and the extent of new information they contain, the market does not differentiate between them and others, and reacts to their publication in the same way. In contrast, recommendations of analysts who are part of a time cluster ("followers") are more accurate than those of the neutrals,and they have a higher informative content (i.e., the new information they convey). That said, an examination of the herd effect in these recommendations shows that together with the creation of a cluster at the time they are published, the forecasts of the target price converge towards the consensus. The fact that the last recommendation in the cluster, published up to 14 days after the first one, also has a significant effect on the market price of the shares intensifies the market distortion that results from the herd behavior. The study shows that the herd effect in analysts' recommendations becomes stronger the smaller the incentive to introduce a bias and the greater the number of analysts in the cluster. The current study offers explanations for this, and suggests ways of improving the efficiency of the information market. The conclusions of the research are that the new arrangement in the field of financial analysis in Israel that focuses on revealing how the analysts operate, is necessary, but not sufficient.

For a PDF of the full article in Hebrew