Analysts' Weight of Forecasts in Stock Recommendations - from The Recursive Partitioning Approach

Abstract

I utilize the recursive partitioning method to extract analysts’ weight of forecasts assigned in their stock recommendation decisions. My findings suggest that in addition to analysts’ earnings forecasts, the non-earnings forecasts, such as sales forecasts and net income forecasts, also play an important role in explaining stock recommendation decisions. The overall weight from six different types of forecasts is positively related to the effectiveness recommendations and the profitability of trading on recommendations. Homogeneity in the styles of using forecasts is a result of analysts’ learning, which leads to variations in analysts’ capability of producing influential recommendations and their career outcomes.

Keywords: Analyst Stock Recommendation and Forecasts, Decision Tree Model, Analyst Learning, Market Reaction, Investment Profitability, Career Outcomes

JEL Classification: G20, G24, G30, M41