About the Russian Consolidated Statements Database#

The Russian Consolidated Statements Database (RCSD) is the first open, representative collection of firm-level consolidated IFRS (International Financial Reporting Standards) financial statements for groups of companies operating in Russia.

RCSD is a continuation and a companion project to the Russian Financial Statements Database (RFSD) — an open, harmonized collection of annual unconsolidated financial statements for the universe of Russian firms. While RFSD’s unconsolidated data is sufficient for standalone entities, it can be grossly misleading for corporate groups. For example, relying on unconsolidated statements for the parent entity of Gazprom severely underestimates its wage bill and overestimates labor productivity, as the bulk of its operations and employees are spread across hundreds of subsidiaries. RCSD bridges this gap by providing the much-needed consolidated view, enabling accurate measurement of firm productivity and markups while avoiding double-counting within corporate groups.

The Data#

The database contains over 5,800 annual financial reports covering 993 unique companies. From these PDF reports, we successfully extracted 20,644 individual financial tables using Vision-Language Models (Ministral 3 14B 2512) tailored for table structure recognition. The data spans from the year 2000 to 2025.

For each firm-year, we focused on extracting four key tables:

  1. Balance Sheet (5,248 tables extracted)
  2. Cash Flow Statement (5,191 tables extracted)
  3. Statement of Changes in Equity / Capital Statement (5,116 tables extracted)
  4. Income Statement (5,089 tables extracted)

While the vast majority of our reports feature all four statements, inconsistencies in reporting formats, rotated layouts, missing grid lines, and multi-level headers present significant extraction challenges. As a result, there are gaps in the coverage. Of the 5,892 reports, 386 reports currently remain as gaps where the VLM models could not reliably parse the table structure, or the firm did not cleanly report the standalone statement within the PDF.

Data Sources#

The consolidated reports and their tables originate from four primary sources:

  • Interfax e-Disclosure (e-disclosure.ru): The largest corporate disclosure portal in Russia, serving as the source for the vast majority of the filings (5,385 reports).
  • INION RAS (Institute of Scientific Information for Social Sciences, reportcollection.inion.ru): Provided the foundational collection of filings, offering extensive historical coverage for the largest groups of companies (447 reports).
  • AZIPI Information Agency (e-disclosure.azipi.ru) & SKRIN (disclosure.skrin.ru): Additional authorized corporate disclosure websites used to fill remaining data gaps and ensure representative coverage (61 reports combined).

Metadata and Enrichment#

Beyond the raw financial tables, RCSD enriches the reports with robust firm-level metadata. By matching entities using their Tax Identification Numbers (INN) and Primary State Registration Numbers (OGRN) against official corporate registries and open data sources, we have augmented the database with:

  • Industry classifications: Standardized Russian OKVED codes and their English translations.
  • Geospatial data: Geocoded exact coordinates (latitude and longitude) of firm headquarters (their legal addresses).
  • Open Knowledge Identifiers: Links to official websites and Wikidata URLs.
  • Entity status: Operational tracking (active status and year of inactivation, if applicable).

This metadata creates a framework that enables researchers and analysts to filter, aggregate, and examine Russian corporate financials by sector, region, and operational status.

Acknowledgements#

The RCSD is built by researchers at the European University at St. Petersburg. The project is an ongoing effort — from scraping thousands of PDFs to harmonizing table lines and conducting internal and external articulation validation. By making this dataset publicly available, we aim to support economic research, financial analysis, and further advancements in the table structure recognition domain.

Team (in alphabetic order): Nikita Bugrin, Yaroslav Kashkarev, Galina Loskutnikova, Polina Rysakova.

Supervision: Dmitriy Skougarevskiy (EUSPb), Nika Zykova (Yandex).

Russian Consolidated Statements Database - © CC BY 4.0