When are Google data useful to nowcast GDP? An approach via pre-selection and shrinkage

We analyse whether, and when, a large set of Google search data can be useful to increase GDP nowcasting accuracy once we control for information contained in official variables. We put forward a new approach that combines variable pre-selection and Ridge regularization and we provide theoretical results on the asymptotic behaviour of the estimator. Empirical results on the euro area show that Google data convey useful information for pseudo-real-time nowcasting of GDP growth during the four first weeks of the quarter, when macroeconomic information is lacking. However, as soon as official data become available, their relative nowcasting power vanishes. In addition, a true real-time analysis confirms that Google data constitute a reliable alternative when official data are lacking.