|
Abstract:
|
Despite the declining manufacturing share of most Western economies, industrial production is one of the most carefully tracked macroeconomic indicators. In this work, we rely on unconventional data sources to nowcast the year-on-year growth of Finnish industrial production, for different industries. We use real-time truck traffic volumes measured automatically in different geographical locations around Finland, as well as electricity consumption and production data which are available almost in real-time. In addition to standard time-series models, we look into the use of machine learning techniques to compute the predictions. We find that the use of atypical data sources such as the volume of truck traffic is beneficial in terms of predictive power, giving us substantial gains in nowcasting performance compared to a simple autoregressive model. This holds true for all industries considered, with particularly strong gains for the forest industry and the paper industry. On the other hand, the adoption of machine learning techniques does not improve substantially the accuracy of our predictions in comparison to standard linear models.
|
