Acurácia nas previsões dos analistas: um estudo sobre os fatores influenciadores

Abstract

Analysts' forecasts are considered to be the main sources of information available to various investors, customers and other financial market participants. Therefore, it is desirable that the forecasts are consistent, reliable and impartial. However, it is intriguing that analysts may be subject to behavioral characteristics of illusion of control, confirmation bias and overconfidence that can influence the issuance of their forecasts and cause errors to exist, which can generate losses for individuals and for companies organizations. In addition, there may be demographic characteristics such as gender, education level, area of education, working time as an analyst, main sector of activity, coverage and country that can be factors that influence the accuracy of the forecasts. Thus, the thesis aims to identify the factors that may influence the accuracy estimate of analysts' financial forecasts. To this end, the initial sample consisted of 6.387 analysts who submitted their forecast information to the Thompson Reuters® database and a structured questionnaire was applied to identify respondents' behavioral and demographic characteristics to verify possible influencing factors. The accuracy estimation variables and the number of companies the analyst performs the coverage were collected in the database. 434 analysts participated. However, 35 were excluded because they were anonymous, 58 because they did not have complete information to estimate the accuracy of the system and 73 due to inconsistency in completing the questionnaire. The final sample consisted of 268 analysts. Regarding the methodology, this study has the characteristic of being a survey, with qualitative and quantitative analysis, and as a method, factor analysis, multiple linear regression and correlation were used. The results show that most are male, are 36 to 50 years old, have a master's degree and a background in economics. In addition, most of them have more than 10 years of experience, mostly working in the construction and energy sector, for over 10 years and participants are from 53 different countries. Although evidence and studies show that analysts have behavioral biases, and that their predictions are biased and ineffective, the three behavioral biases studied have shown that they are not able to influence the average accuracy estimate. The influencing variable was the main sector, which indicates that analysts of non-cyclical firms are more accurate than analysts of cyclical firms. An interesting finding is that the greater the illusion of control, the greater the confirmation bias and the higher the confirmation bias, the greater the overconfidence. Thus, research allowed a better understanding about the relationships between agents and the dynamics of the financial markets, as well as presented an overview of the behavioral and demographic characteristics of the participants. Thus, this study is expected to be the beginning of a reflection so that new research can be developed.