Research On Security Analysts’ Overconfidence Bias

Analysts have long time been considered as an information intermediary in the financial markets. They act as a communication bridge between investors and listed firms. Based on their analyses of the overall macroeconomic situation, industry development trends, and firm’s growth potential, analysts make earnings predictions. Analyst report is generally an important reference for investors’decision-making. However, some academics find that analysts are not neutral when making forecasts. And their predictions may be influenced by themselves, their brokers, management of the firm they analyzed and other factors. As a result, analyst predicitons tend to be systematically biased.Overconfidence is one of the causes that may lead to distortions in analysts’behavior. By establishing a dynamic framework of overconfidence and using the quarterly analyst forecast data covered from 2001 to 2010, this paper tries to provide empirical evidences for the existence of analyst overconfidence. The two main research questions in this paper are as follows. First, does overconfidence actually exist when analysts make earnings forecasts? Second, if yes, what factors may influece the degree of analyst overconfidence?Using the model same as the one of Hilary and Menzly (2006), this paper fisrt tests whether overconfidence exists in analyst forecast behavior. The result shows that analysts who have predicted earnings more accurately than the median analyst in the previous quarters tend to be less accurate in their subsequent earnings predictions, but not tend to deviate further from the consensus. The result partly supports the overconfidence hypothesis.Secondly, from the perspective of financial crisis, this paper tries to further explore the impacts on the degree of analyst overconfidence by certain environmental factors. The result shows that the association between past success and current forecast errors is strengthened against the background of financial crisis. Overconfidence still exists, and the crisis increases the degree of analysts’ exposure to past success.Finally, given the short-term dynamic overconfidence may be affected by the consecutiveness of analyst forecasts behavior, and simultaneously to check whether the empirical results are robust, this paper performs a robustness test regarding the main findings. The results of robust test are consistent with previous findings.