9.1.2007
New research––The Herd Instinct in Analysts’ Recommendations: Evidence from Israel’s Capital Market
The views expressed in this paper do not necessarily reflect those of the Bank of Israel
 

  Analysts’ recommendations tend to over-optimism: about 72 percent of their recommendations (on average) regarding shares are “buy.”
  Recommendations published independently as far as their timing is concerned, in other words, at a time when recommendations of other analysts are not published (about 45 percent of all recommendations), are over-optimistic, and contain forecasts that deviate from the existing market consensus. In contrast, share-price forecasts in “cluster” recommendations, i.e., published at approximately the same time, converge to the consensus and indicate the existence of a herd instinct with regard to the target price.
  Despite the “inferiority” of such independent recommendations, measured by the relative accuracy of the forecast and the extent of their new economic informative content, the market does not differentiate between them and others, and reacts to their publication in the direction recommended. In the case of cluster forecasts too the market does not discern the herd instinct in the target price and the reduced relevance of their informative content, and reacts strongly to these recommendations too. Investors’ reaction was found to be significant even in the case of the last recommendation in the cluster, published fourteen days later than the first one.
  Revealing analysts’ professional relationships and possible conflicts of interest with the aim of counteracting the incentive to bias investors is indeed necessary, but it is not sufficient. The study suggests adding a weighted rating of analysts based on––in addition to their accuracy and the extent of their effect on the markets––the timing of their recommendations. Such an arrangement would probably solve the issue of the herd instinct and enhance the informative efficiency of the markets.
  The Bank of Israel announces that research carried out by Konstantin Kosenko of the Research Department analyzes the herd instinct reflected in analysts’ recommendations regarding shares on Israel’s capital market. The study is based on 2,780 recommendations by analysts in the years from 1999 to 2004 regarding shares traded on the Tel Aviv Stock Exchange. Their recommendations were found, on average, to converge to an optimism expressed in “buy” recommendations (about 72 percent of all recommendations on average), even in times of recession. The timing of the publication of their recommendations (within a given quarter) was found to be of critical importance with regard to their effect on the conclusions drawn. Thus, analysts’ recommendations at an independent time, i.e., a time without recommendations by other analysts (about 45 percent of the sample) with their forecasts of share prices, were characterized by a bias and deviation from the market consensus target price; the accuracy of these recommendations and their new informative content were low. That said, it was found that investors considered them of value, and their publication was accompanied by share price changes in the direction predicted. In contrast, forecasts of target prices in recommendations published in a time cluster, i.e., published in close proximity to each other, have a higher informative content and are more accurate than independent-timing recommendations. The study postulates that these results derive from the publication of the cluster recommendations at a time close to the arrival of new extensive external information and that their accuracy rises as a result of the learning process and the delay of the recommendation. It was found, however, that the forecast target prices in cluster recommendations tend to converge to the consensus price, and their informative contribution declines as the cluster grows larger. Nevertheless, the herd instinct factor in the recommendations of analysts who are part of the cluster is not identified by the market, and investors respond significantly also to the last recommendation in the cluster, fourteen days later than the first one.
  The research found that reducing the incentive to introduce a bias in addition to expanding the cluster of recommendations have the effect of intensifying the herd instinct effect in the forecast target price, and regulation in the field of financial analysis aimed at revealing analysts’ potential conflicts of interest does not resolve the problem of the herd effect in their recommendations in the long run.
  The study proposes the use of a weighted rating of analysts based on their accuracy, the extent of their effect on the markets, and the timing of their recommendations; this would differentiate between the quality of their forecasts, and thereby enhance the informative efficiency of the markets and the situation of the investors.
  The Bank of Israel emphasizes that the results of the research are based on a large sample of analysts (about 2,800 recommendations by some 50 analysts in Israel’s capital market) and their recommendations in the years from 1999 to 2004. Thus, no definite conclusions should be drawn regarding any specific analysts or institutions that engage in analysis.
 
Information Ratio of Recommendations (“Buy” recommendations)

 
Market Reaction to Recommendation of Last “Follower” Analyst in Cluster (Cumulative abnormal return (CAR), “buy” recommendations)

 
​​