EcoSta 2017: Start Registration
View Submission - EcoSta2017
A0600
Title: Dynamic mixed frequency synthesis for GDP nowcasting Authors:  Kenichiro McAlinn - Temple University (United States) [presenting]
Abstract: A new method is developed for mixed frequency modeling. We call it mixed frequency synthesis (MFS), and it utilizes the newly developed framework of Bayesian predictive synthesis (BPS). MFS synthesizes the information from multiple frequencies in a theoretically coherent Bayesian framework. We demonstrate the efficacy of MFS by a topical macroeconomic exercise of nowcasting GDP using higher frequency data. This study highlights insights into dynamic relationships among multiple frequencies, as well as the potential for improved forecast accuracy at multiple horizons.