Vintage data, or retrospective data, play a crucial role in assessing the accuracy of macroeconomic forecasting models and the decisions based on these models. These data represent a snapshot of information available at a particular point in time in the past and reflect how the economic situation was perceived at that time. It is important to realise that it is on the basis of vintage data, not revised and refined indicators, that forecasts and economic policy decisions were made. As an example, we show that the GDP growth paths in the projection period may diverge by as much as 1 pp.

This paper presents a set of vintage data on Russian GDP and its components by the expenditure and production approaches. The dataset consists of revisions of nominal and real quarterly data for the period from December 2005 to the present. In addition to such data, the paper describes some properties of real and nominal GDP indicator revisions (number, frequency, magnitude) and its expenditure components, as well as the methodology for collecting historical indicators using the Wayback Machine, which enables data collection even in the absence of a saved history of their releases.