On October 15, 2020, Rosstat published industrial production indices for September and also revised its data for all of last year and the first eight months of 2020. A rather significant revision has taken place — according to the new data, industrial production in Russia only fell by 2.6% from January to August this year compared to the same period in 2019, while the previous assessment assumed a decline of 4.5%. This sort of revision amounts to about 0.4 pps in terms of annual GDP growth, which is a lot objectively speaking, but is generally in line with Russia’s history of both GDP revisions as a whole and its individual components.
The first assessment of annual GDP growth published by Rosstat is lower than the final one in 74% of cases1. Furthermore, in 44% of cases the positive revision accumulated over several years is more than 0.5 pps (Fig. 1, right graph). Consequently, preliminary analysis of Russia’s economic growth based solely on early GDP data usually paints a pessimistic picture.
1ACRA analyzed data from 2004 to 2018; the cycle of planned revisions for 2019 has not yet been completed.
Sources: Rosstat, ACRA
In Fig. 1 and thereafter, calculations use real GDP growth in % vs. the corresponding quarter of the previous year.
1) Conservative assessments have been made almost every quarter for the past 15 years (Fig. 1, left graph).
2) Sometimes it is necessary to revise GDP estimates due to revised data from different sectors. From 2004 to 2011, positive revision of added value dynamics took place mainly in trade, the financial sector and other services segment. After 2014, the manufacturing industry and construction played an important role (Fig. 2).
3) Despite the infrequency of changes to methodological approaches, these revisions are mainly aimed at increasing initial estimates. This is due to the logic of the development of statistics: methodologies attempt to “catch up” with real life in terms of coverage of types of activities and agents, however, part of economic activity is usually not measured in full. On average, more than half of GDP growth revisions take place after the first two scheduled revisions that are related to completing the collection of data and its agreement. However, the initial assessment is usually increased even after the first revisions (Fig. 2).
A typical example of the existing underestimation is conditionally free online services. The added value which they create is not determined.
4) The above happens to the initial economic growth estimates calculated by a significant part of global statistical agencies (Fig. 3).
* Depending on the revision of the real rates of annual growth in added value produced in certain sectors.
Sources: Rosstat, ACRA
The possible consequences of the current conservative approach to initial economic growth assessments are as follows:
1) Superfluous incentives for carrying out state economic stimulus measures. As a rule, potential economic growth is estimated using long time series, while current GDP growth rates are systematically underestimated.
2) Shifting of economic forecasts to a conservative plane. In practice, the accuracy of forecasts is assessed shortly after the end of the calendar year, and to do this, often only the first reported data for the previous year is used. As a result, analysts have more incentives to predict and use for the purposes of analysis precisely the first, systematically underestimated estimate of economic growth rates.
* The country sample and analyzed horizon were determined taking into account the availability of data vintages.
Sources: Rosstat, OECD, ACRA
You will receive an email with a link to change your password
Полное использование материалов сайта разрешается только с письменного согласия правообладателя, АКРА (АО). Частичное использование материалов сайта (не более 30% текста статьи) разрешается только при условии указания гиперссылки на непосредственный адрес материала на сайте www.acra-ratings.ru . Гиперссылка должна быть размещена в подзаголовке или в первом абзаце материала. Размер шрифта гиперссылки не должен быть меньше шрифта текста используемого материала.