How to evaluate the forecasting performance of a macroeconomic model

Abstract

This paper provides an answer to the question of how to improve the forecasting performance of a macro model to better account for economic developments and how to evaluate the forecasting uncertainty.The main tool in this assessment is stochastic simulation.Stochastic simulations in this paper involve both endogenous and exogenous variables.These simulations also allow us to assess the linearity of the model.Alternative dynamic simulations may, in turn, give some idea of the stability of the model.Finally, the forecasts may be improved by comparing the outcomes from the macro model and from a leading indicators' model. This kind of exercise is particularly useful in assessing the developments in the short run, in which case the macro models typically perform rather poorly. Keywords: forecasting, macro models, simulation