Empirical Research On Securities Analysts' Earnings Forecasts In China

Whether securities analysts'earnings forecasts are more accurate and a better surrogate for market expectations of earnings than those generated by univariate time-series models are two basic questions among the research on securities analysts'earnings forecasts. And there is a large body of research on these two questions in America. But the conclusions from American research are not applicable in China because of the differences between the market conditions in America and in China, especially the differences between the institution of securities analysts in America and in China. So, we should explore whether securities analysts'earnings forecasts in China's A-share markets are more accurate and a better surrogate for market expectations of earnings than those generated by univariate time-series models.In 2005, using forecasted 2001 annual earnings for listed A-share firms in China, which is provided by Guotai Junan Research Institute on its web-site at the end of 2001 or at the beginning of 2002, Wu and Xue(2005a) firstly study these two questions in China and find that securities analysts'earnings forecasts in China's A-share markets are more accurate and a better surrogate for market expectations of earnings than those generated by the random walk model.Then, using securities analysts'earnings forecasts for listed A-share firms in China provided by Wind, this paper furthers the research. This paper differentiates four forecast horizons, and uses the mean, the median, and the most current forecast to aggregate analysts'earning forecasts into a single expectation, and adopts quarterly earnings models besides the random walk model. The major conclusions of this paper indicate that annual earnings forecasts generated by quarterly earnings models are more accurate and a better surrogate for market expectations of earnings than those generated by the random walk model, and that securities analysts'earnings forecasts are more accurate in the forecast horizons near the announcement of annual report but not a better surrogate for market expectations of earnings than those generated by univariate time-series models. Generally, these conclusions could help researchers and investors to obtain accurate earnings forecasts and a good surrogate for market expectations of earnings.