Dynamics of Macroeconomic Forecasting Variation and Correlation Structure in the Business Cycle

Using a structural model, I analyze how changes in the distribution of signals about unknown economic conditions affect real aggregate macrovariables in the business cycle. I focus on two quantifiable properties of the distribution of signals, the signal accuracy and the correlation structure across signals, and analyze how time variation in these two properties affect an agent's decisions through his posterior beliefs. Since the exact signals agents use are difficult to observe empirically, I define two key concepts, uncertainty and disagreement, that capture dynamics in the distribution of signals and can be linked to data. Uncertainty is defined as the dispersion within each agent's forecasts about economic conditions. Disagreement is defined as the dispersion across agents in their mean forecasts about economic conditions. I show that uncertainty and disagreement affect an agent's controls through his first and higher order beliefs about economic conditions. Calibrating to US macrodata and the Survey of Professional Forecasters, I show empirically that the distribution of signals matters for aggregate dynamics and that my model mechanism can parsimoniously match the magnitude and sign of these effects. However, I find movements in the distribution of signals represent only a small fraction of the total variation in aggregate variables