Forecasting national activity using lots of international predictors

Autori

• Sandra Eickmeier

Parametri

• Economia e affari

Maggiori informazioni sul libro

This study explores how large international datasets enhance forecasts of national activity, focusing on New Zealand as a small open economy. It employs various data-rich factor and shrinkage methods to efficiently manage numerous predictor data series from multiple countries. The techniques discussed include principal components, targeted predictors, weighted principal components, partial least squares, elastic net, and ridge regression. The analysis reveals that leveraging a wide array of international predictors significantly improves forecasts of New Zealand's GDP growth compared to traditional models reliant on smaller datasets. Notably, even though New Zealand's survey data captures a considerable amount of predictive information, the inclusion of international data yields the greatest forecasting accuracy gains, particularly at longer forecast horizons. The performance of data-rich methods varies, with shrinkage methods and partial least squares yielding the best results. Additionally, the study evaluates which types of international data provide the most valuable predictive insights for New Zealand's economic growth throughout the sample period.

Lingua: Inglese

Pubblicato: 2009