Nowcasting Inflation Using High Frequency Data

Abstract

Non-technical summary ECB Working Paper Series No 1324

Nowcasting GDP - A Scalable Approach Using DFM, Machine Learning and Novel Data, Applied to European Economies

“…Nowcasting methods have been successfully applied in a number of fields, including outside of economics. In economics, nowcasting methods have been applied not only in near-term economic activity prediction but also to nowcast inflation or employment(Knotek and Saeed, 2014;Modugno, 2011;Hutter, 2020). The focus of this work is nowcasting of quarterly GDP growth.©International Monetary Fund.…”

Nowcasting from cross‐sectionally dependent panels

“…As a predictor variable, we use consumer prices of petroleum products, net of duties and taxes, which are taken from the European Commission's Weekly Oil Bulletin (WOB). We use data for automotive gas oil, heating gas oil and Euro Super 95 gasoline, which we average together as in Modugno (2011). We use the same transformations as in the case of inflation, that is, the annual (52 week) and weekly log differences (for y‐o‐y and m‐o‐m, respectively).…”

“…Our second contrasting empirical application assesses how the PMIDAS model performs in nowcasting monthly inflation across a large set of European countries. We use weekly energy prices to provide a timely signal for tracking movements in inflation as in Modugno (2011Modugno ( , 2013. We therefore offer a new approach by nowcasting a panel of countries' inflation instead of single countries.…”

“…The earlier version of the paper(Modugno, 2011) analysed both US and Euro area inflation, whereasModugno (2013) focusses only on US inflation.…”

Nowcasting inflation with Lasso‐regularized vector autoregressions and mixed frequency data

“…Marcellino (2008) compare the predictive abilities of time‐varying models, nonlinear time series models and artificial neural network models against standard ARMA models in predicting US GDP growth. Modugno (2011) uses mixed frequency data and a factor modeling framework to separate the effect on forecast revisions due to the inclusions of new data releases from that attached to the high‐frequency dimension. Breitung and Roling (2015) propose a non‐parametric approach to high‐frequency forecasting using mixed frequency data and equations.…”