Nowcasting world GDP growth with high‐frequency data

Although the Covid-19 crisis has shown how high-frequency data can help track economy in real time, we investigate whether it improve nowcasting accuracy of world GDP growth. To this end, build a large dataset 718 monthly and 255 weekly series. Our approach builds on Factor-Augmented MIxed DAta Sampling (FA-MIDAS), which extend with preselection variables. We find that markedly enhances performances. This also outperforms LASSO-MIDAS—another technique for dimension reduction mixed-frequency setting. Though FA-MIDAS outperform other models relying or quarterly data, point to asymmetries. Models have indeed performances similar during “normal” times but strongly them “crisis” episodes, above all period. Finally, model annual growth incorporating give timely (one per week) accurate forecasts (close IMF OECD projections 1- 3-month lead). Policy-wise, provide an alternative benchmark episodes when sudden swings make usual (IMF's OECD's) quickly outdated.