Accurately estimating future firm earnings is crucial for evaluating corporate profitability and stock valuation. Conventional market consensus measures based on professional analysts' forecasts are prone to errors due to analysts’ systematic behavioral biases, necessitating effective debiasing methods. In this paper, we propose a new approach to debiasing by focusing on analysts' silent behaviors and uncovering the underlying forecasts of silent analysts through elaborate modeling of their self-selected silent behavior during the process of estimating and releasing earnings forecasts. We formulate analysts' selective forecasts as a data-missing-not-at-random problem and develop a joint likelihood optimization method to infer silent analysts' unrevealed forecasts and consolidate them with the observed ones to achieve more accurate estimations for annual earnings per share. Using a large sample of firms over twenty years, we evaluate our method and find that it outperforms market consensus forecasts in terms of forecast bias and forecast accuracy, both statistically and economically.
Jin, Enze; Wang, Cong; Guo, Kai; Guo, Xunhua; and Ke, Bin, "Modeling silent behavior for synthesizing analysts’ earnings forecasts: a joint likelihood approach" (2023). PACIS 2023 Proceedings. 138.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.