High-Mixed-Frequency Dynamic Latent Factor Forecasting Models for GDP in the Philippines

Abstract

The paper considers constructing high-frequency forecasting models for GDP growth in the Philippines, in the form of dynamic time-series models that combine latent factors with a parsimonious set of indicators that are observable at different frequencies. The forecast performances of the estimated models also are compared with other alternative current modeling approaches - e.g., Mixed Data Sampling Regression (MIDAS), Factor Analytic Models, and Current Quarterly Modeling (CQM) with Bridge Equations.