Forecasting the real price of oil using online search data

New models to forecast the real price of oil on the basis of macroeconomic indicators and Google search data are proposed. A large–scale out–of–sample forecasting analysis comparing the different models is performed. It is found that models including both Google data and macroeconomic aggregates statistically outperform the competing models in the short term, while multivariate models including only Google data perform best also for medium and long term forecasts up to 24 months ahead. This finding is confirmed by different robustness checks.