Events

Professor Zhaoyang Gu
Professor
School of Accountancy
The Chinese University of Hong Kong

We examine how firms’ global sales network (GSN), measured using machine-learning based textual analyses of 10-K narratives describing worldwide sales, affects analysts’ cost of information acquisition and processing. We find that when the firms within an analyst’s portfolio have overlapping GSN, the analyst tends to form larger portfolios and issue more timely and more accurate forecasts, consistent with overlapping GSN allowing analyst more efficiently acquiring geography-specific information. The effects of GSN on analyst forecasts are greater for more experienced analysts, and when portfolio firms’ earnings are easier to forecast. Forecast-level analyses suggest that analysts learn from portfolio companies’ earnings surprise and actively revise their forecasts of other portfolio firms with similar GSN. Further suggestive evidence show that analysts’ cultural and ethnic origin is a likely cause for analysts’ heterogeneous degree of GSN. Our paper provides evidence on a previously unexplored dimension of learning by analysts.