Heterogeneity in inflation : analysing the performance of the mean, median and individual forecasters

Abstract

In this thesis heterogeneity in inflation forecasts is analysed. First, the Survey of Professional Forecasters and the Michigan survey are analysed with focus on the mean, median and individual forecasters. The forecasters are analysed with respect to forecast bias and forecast accuracy. The performance of the individual forecasters are analysed and compared to the performance of the mean and median forecasters. The mean and median forecasters are found to be less biased and more accurate than the individual forecasters. Second, a simple dynamic stochastic general equilibrium model describing an economy with heterogeneous consumers is analysed. The consumers in the model use two predictors and switch between the predictors based on their past performance. The predictors in the model are also analysed with focus on forecast bias and forecast accuracy. The mean predictor is found to be both less biased and more accurate than the two predictors used by the consumers in the model. However, the results are more mixed for the median predictor which is found to be less biased, but less accurate than the predictors used by the consumers.