Our study sheds light on five important areas of macroeconomic forecasting. First, we do believe that the macroeconomic variable release affects the financial market (stock and bond markets, for instance) volatility; however, the effect is significant when the actual release is different than the market expectation. Second, the importance of an individual forecaster, who is better than consensus, is increased given that the market moves more when actual release is different than market consensus. That being said, an individual forecaster, who is better than consensus, will provide more opportunities to his firm to make money (or reduce loss). Third, in short-term forecasting (one month ahead) actual release timing of the target variable (dependent variable), as well as predictors, is very important and needs to be considered in model specification. Fourth, traditional forecast evaluation methods (R-squared, Adjusted R-squared, Root Mean Square Error, etc) are necessary but we recommend more. That is, directional accuracy along with traditional forecast evaluation methods is a better approach. Finally, one model specification won’t remain accurate for ever. For example, one model specification will produce good forecast for a while (one year or so) and later completely fails, unable to generate good forecast.
Therefore, even a good forecasting model may need to be revised at some point in the future and hence, short-term forecasting is an evolving process.
Key Words: Macroeconomic Surprises; Consensus; Individual Forecaster.
JEL Classification; C32; E37.