Financial Analyst vs. Statistical Model:The Relative Accuracy of Analyst Forecasts and Its Determinants

Yue Heng,Lin Xiaochi

DOI: https://doi.org/10.3969/j.issn.1003-2886.2008.08.006

2008-01-01

Abstract:Whether analysts' forecasts are more accurate than forecasts of statistical models is the most fundamental issue of the research on financial analysts.If earnings forecasts of analysts are less accurate,then analysts are worthless;otherwise,if analysts can more accurately forecast earnings,then analysts can provide useful information to investors and contribute to the development of capital market.This paper uses analysts' forecasts from 35 stock brokers to investigate the relative accuracy of analysts' forecasts and forecasts of statistical models in China and the determinants of the relative accuracy.We find that analysts' forecasts are more accurate than statistical models based on historical annual data,but are less accurate than statistical models based on historical quarterly data.Further investigation indicates that earnings volatility,firm age,and the number of analysts following affect the relative accuracy of analysts' forecasts.Our research suggests that analysts in China need to improve their ability of forecasts and investors should be cautious when using analysts' forecasts.