Introduction
In order to determine if analysts’ earnings forecast is useful, this essay will argue whether analysts’ earnings forecast is biased, and if their forecasts are consistent with recommendation.

Analysts’ earnings forecast vs. random walk earnings forecast
According to the paper (1), they want to determine if analysts’ annual EPS forecasts are superior to those from time-series models. The empirical tests are based on forecasts of annual earnings over horizons ranging from 1 through 36 months. They compare the performance of simple random walk (RW) annual earnings forecasts and analysts’ annual earnings forecasts. They find:
First, for longer forecast horizons, analysts’ forecasts of future earnings are not consistently more accurate than time-series models, even when analysts have timing and information advantages.
Second, for forecast horizons where analysts are more accurate than RW forecasts (that is, forecast horizons of several months), the differences in accuracy are economically small.
Third, RW forecasts are more accurate than analysts’ forecasts for estimating 2-year-ahead earnings in approximately half of the forecast horizons analyzed, and RW forecasts strongly dominate analysts’ forecasts of 3-year-ahead earnings.
Fourth, over longer forecast horizons, analysts’ forecast superiority is prevalent only in limited settings, such as when analysts forecast negative changes or small absolute changes in EPS.
The evidence that time-series forecasts perform as well or better than analysts’ forecasts suggests that the generalizability of research typically confined to firms with available analysts’ forecast data (that is, large, mature firms) would be enhanced by incorporating time-series forecasts, which permits an expansion of firms available to be examined.

In additional analyses, they compare the performance of RW and of analysts’ forecasts of long-horizon forecast earnings to short-horizon forecast. It finds that this extrapolation of analysts’ 1-year-ahead forecasts provides the most accurate forecast over almost every horizon and among all subsamples.

Consistency between analysts' earnings forecast and recommendation
The paper (2) examines whether valuation estimates based on analysts' earnings forecasts are consistent with their stock recommendations. They think earnings forecasts are linked to value and recommendations reflect analysts' opinions of value relative to current price, earnings forecasts and stock recommendations should be linked in a predictable manner.

They consider four possible valuation models of how earnings forecasts and stock recommendations are linked. These models include two specifications of the residual income model, a price-earnings-to-growth (PEG) model, and analysts' projections of long-term earnings growth.

The results provide little evidence that analysts' recommendations are explained by either residual income model specification. However, both the PEG model and analysts' projections of long-term earnings growth explain analysts' stock recommendations. The relation between the valuation models and future returns is also examined. Analysts' projections of long-term earnings growth have the greatest explanatory power for stock recommendations, but investment strategies based on these projections have the least association with future ex-cess returns.

Conclusion
This essay has compared analysts’ earnings forecast vs. random walk earnings forecast, and argue the consistency between analysts' earnings forecast and recommendation.

Analysts’ forecasts of annual earnings are superior to time-series forecasts is not fully descriptive.