Global Sales Network and Analyst Forecasts

Abstract

In this paper, we use an analyst’s coverage portfolio as the unit of analysis and argue that the linkage among the portfolio firms can provide useful information for analyst forecasts at the portfolio level. In particular, the geographic overlap in the portfolio firms’ global sales network (GSN) can generate important information spillover within the portfolio and create synergies in an analyst’s information production. Based on several measures of GSN similarity from textual analysis and machine learning algorithms, we find that greater GSN similarity of an analyst’s portfolio is associated with more frequent and more accurate sales forecasts. Further, greater GSN similarity is associated with more accurate earnings forecasts, especially when portfolio firms have high operating and financial leverages and when the earnings forecasts are accompanied by sales forecasts. Additional analyses corroborate analysts’ learning from portfolio firms’ earnings surprises when they are closely related through GSN.

Keywords: Analyst forecasts, sales network, machine learning