A1436
Title: Optimal pooling and finite mixture distribution combinations of probabilistic forecasts
Authors: Giulia Mantoan - The Alan Turing Institute (United Kingdom) [presenting]
Abstract: The combination of two or more density forecasts has a long tradition in the statistics and forecasting literature. However, comparatively little attention in econometrics has been given to the finite mixture distribution as a statistical model for combining density forecasts. Combination procedures based on a mixture density distribution are able to account for parameter uncertainty in addition to combination's weight uncertainty, which are features normally not considered in traditional ``two-step'' approaches to density forecasts combination. The aim is to compare the ``one-step'' mixture approach to the more traditional ``two-step'' approach thereby endowing the decision maker with a tool to elicit the best aggregation method. The superiority of the "one-step" approach is accessed through analytical analyses, several Monte Carlo simulations to account for structural breaks and a macroeconomic application.
