Title: Advances in nowcasting economic activity
Authors: Ivan Petrella - University of Warwick (United Kingdom)
Thomas Drechsel - LSE (United Kingdom) [presenting]
Juan Antolin Diaz - Fulcrum Asset Management (United Kingdom)
Abstract: The nowcasting performance of a Dynamic Factor Model which extends the existing framework along various dimensions is evaluated. First, building on our prior work, we incorporate low-frequency movements to the growth rate and the volatility of the variables. Second, we consider the impact of relaxing the assumption of homogeneous dynamics by allowing for lags of the factor to enter the measurement equation. Third, we endogenously model seasonal fluctuations, which can be particularly useful whenever there is suspicion that residual seasonality is present in the data, and outliers, by allowing for fat tailed distributions in the idiosyncratic disturbances. Using a fully real-time database covering the period 2000-2017 for the G7 economies, we conduct a comprehensive forecasting evaluation exercise, with particular attention to probabilistic (density) forecasts, and assess the importance of these features for the real time tracking of economic activity.