In their paper “Do Analyst Forecasts Vary Too Much?” Russell J. Lundholm and Rafael Rogo (2016) present very interesting evidence of variation in analyst forecasts that is inconsistent with the belief that analysts use rather obvious information from the time-series of earnings when making their forecasts. Using as a benchmark the variance in one-year-ahead quarterly earnings, they report that the time-series variation in analysts' individual forecasts violates this bound 12 to 17 percent of the time and that cross-sectional variation in forecasts (or forecast dispersion) violates this bound 5 to 8 percent of the time. Lundholm and Rogo conjecture that there are behavior reasons and strategic reasons for this excess volatility. I will first comment on how surprising the evidence is. I will then comment on the sample and study design. Finally, I will comment on the implications of the findings for future research.
In terms of what we have...
