Impact of Biases on Analyst´s Accuracy: a Cluster-Based Analysis

Objective: This study investigates the impact of biases on the accuracy of financial analysts by distinguishing between analysts with higher and lower levels of accuracy through cluster analysis. Method: The analysis encompassed publicly traded companies in Brazil and the USA during the quarters of 2019, comprising 840 observations from 76 Brazilian firms and 16,402 observations from 880 U.S. firms. The cognitive biases examined included: anchoring, optimism, overconfidence, communal bias, representativeness, and realism, measured using Diction®. Data analysis was conducted using STATA® and SPSS®, employing analysis of variance (ANOVA), post-hoc ANOVA test, cluster analysis and multiple regression models. Results: Among the biases analyzed, anchoring—defined as the tendency to rely on past earnings as a reference point for future forecasts, was the only bias to exhibit consistent patterns across both accuracy groups and countries. This finding suggests that the stochastic effect of past profitability contributes positively to earnings forecasting accuracy. In the U.S. sample, cognitive biases were more strongly associated with analyst accuracy than in the Brazilian sample. Overconfidence and realism were particularly salient among the higher accuracy groups in the U.S. In Brazil, the only statistically significant result indicated a negative effect of optimism on forecasts issued by analysts in the higher accuracy group. The communal bias was found to negatively affect analyst accuracy in the U.S., regardless of accuracy level. Representativeness bias also had detrimental effect on analyst with lower accuracy in the U.S. Contributions: This study emphasizes the relevance of considering biases in the evaluation financial analysts and suggests that awareness of such biases may enhance investment decision-making. Specifically, analysts with overconfidence and realism in the U.S. tend to produce more accurate forecasts, whereas those influenced by optimism in Brazil, and communal or representativeness biases in the U.S., should be approached with caution.