Are analysts biased? an analysis of analysts’ stock recommendations for stocks that perform contrary to expectations

The finance literature suggests that analysts’ stock recommendations have negligibleimpact on market prices. Some studies suggest this lack of market impact may be partlydriven by the affiliations between investment banks and the firms their brokerage armscover (conflicts of interest). However, most of these studies fail to take into accountother factors including institutional and trading issues and psychological biases whichmay well be just as important in influencing analysts when they gather, process andinterpret information about stocks.The aim of the current study is to establish the factors which are associated withanalysts issuing stock recommendations that lack market impact. I find thatnonconforming analysts’ stock recommendations are associated with overconfidencebias (as measured by optimism in the language they use) and representativeness bias (asmeasured by previous stock price performance, market capitalisation, book-to-marketand change in target price). Thus, stocks that receive a buy rating and subsequentlyunderperform the respective benchmark are associated with a high level of optimism inthe tone of the language used by analysts in their investment reports that they prepare tojustify their recommendations, have positive previous price momentum, have largemarket capitalisation, have low book-to-market ratio and have their target priceschanged in the same direction as the stock recommendation. Not surprisingly, there isalso a relationship between the investment bank issuing the recommendation and thefirm. In addition, stocks that are awarded sell rating and subsequently outperform thebenchmark have characteristics opposite to those of nonconforming buys.Finding that potential conflicts of interest significantly predict analysts’ nonconformingstock recommendations supports recent policy-makers’ and investors’ allegations thatanalysts’ recommendations are driven by the incentives they derive from investmentbanking deals. These allegations have led to implementation of rules governinganalysts’ and brokerage houses’ behaviours. However, finding that cognitive biases playa major role in the type of recommendation issued suggests that these rules may workonly in as far as regulating conflicts of interest, but will have a limited role in regulatingpsychological bias, as my results suggest that analyst bias is inherent in their work.Surveys of what fund managers expect of analysts indicate low rankings of analysts’investment advice as manifested in their recommendations (e.g., All-America ResearchTeam Survey 2002). My results further indicate that fund manager concern is likely tocontinue because not all behavioural factors in analyst stock recommendations can be controlled.