Forecasting earnings with combination of analyst forecasts

We propose a regression-based method for combining analyst forecasts to improve forecasting efficiency. This method significantly reduces the bias in earnings forecasts, and generates forecasts that consistently outperform consensus forecasts over time and across firms of different characteristics. Incorporating firm-level and macroeconomic information in the model further improves earnings forecasting performance. Forecasting gains increase with the dispersion and bias of analyst forecasts, and the degree of under/overreactions to earnings news. Moreover, the combination forecast produces larger earnings response coefficients, weakens the anomaly of post-earnings-announcement drift, and provides a better expected profitability measure that has higher power to predict stock returns.

Automated summary

We propose a regression-based method that combines analyst forecasts to enhance forecasting efficiency. This method effectively reduces earnings forecast bias and consistently outperforms consensus forecasts across different firms and over time. Additionally, incorporating firm-level and macroeconomic information improves the accuracy of earnings forecasting. The gains in forecasting performance are greater with higher dispersion and bias in analyst forecasts, as well as greater under/overreactions to earnings news.