Title: A real time investigation of short-term forecasting
Authors: Alain Galli - Swiss National Bank (Switzerland) [presenting]
Rolf Scheufele - Swiss National Bank (Switzerland)
Christian Hepenstrick - Swiss National Bank (Switzerland)
Abstract: Nowcasting methods to assess the current state of the economy are often evaluated in terms of their out-of-sample forecast performance for quarterly GDP. However, such an assessment heavily depends on the point in time, i.e. the information set, for which such an evaluation is conducted. Furthermore, it ignores that the target variable - the first release of quarterly GDP - is highly prone to revisions. Therefore, we assess such methods in an alternative way and focus on the convergence of each models forecast week by week to its current quarter nowcast at the time when the first estimate of GDP is published. In other words, we answer the question of what can be known when. Furthermore, we show that, for the case of Switzerland, within a period of two years after the respective quarter has ended, the real-time assessment of economic developments based on nowcasting approaches outperform first and subsequent GDP releases in terms of prediction ability for final GDP.