Nowcasting GDP in Real Time: A Density Combination Approach

Knut Are Aastveit,Karsten R Gerdrup,Anne Sofie Jore,Leif Anders Thorsrud

doi:10.1080/07350015.2013.844155

Abstract

In this article, we use U.S. real-time data to produce combined density nowcasts of quarterly Gross Domestic Product (GDP) growth, using a system of three commonly used model classes. We update the density nowcast for every new data release throughout the quarter, and highlight the importance of new information for nowcasting. Our results show that the logarithmic score of the predictive densities for U.S. GDP growth increase almost monotonically, as new information arrives during the quarter. While the ranking of the model classes changes during the quarter, the combined density nowcasts always perform well relative to the model classes in terms of both logarithmic scores and calibration tests. The density combination approach is superior to a simple model selection strategy and also performs better in terms of point forecast evaluation than standard point forecast combinations.

Highlights

Policy decisions in real-time are based on assessments of the recent past and current economic condition under a high degree of uncertainty
The density nowcasts are combined in a two-step procedure
Our results extends the findings in the earlier nowcasting and model combination literature along several dimensions: First, we show that the log score of the predictive densities for the model combination and all three model classes increases almost monotonically as new information arrives during the quarter, while the densities seem well-calibrated at each point in time

Summary

Introduction

Policy decisions in real-time are based on assessments of the recent past and current economic condition under a high degree of uncertainty. The academic literature on nowcasting has been focusing on developing single models that increase forecast accuracy in terms of point nowcast, see among others Evans (2005) and Giannone et al (2008). This differs in two important ways from policy making in practice. Policy makers are often provided with several different models which may provide rather different forecasts. To ensure appropriate monetary policy decisions, central banks must provide suitable characterizations of forecast uncertainty. Density forecasts provide an estimate of the probability distribution of the forecasts.

Objectives

Methods

Results

Conclusion