The Russian Financial Statements Database (RFSD) is an open, harmonized collection of annual unconsolidated financial statements of the universe of Russian firms:
-
🔓 First open data set with information on every active firm in Russia.
-
🗂️ First open financial statements data set that includes non-filing firms.
-
🏛️ Sourced from two official data providers: the Rosstat and the Federal Tax Service.
-
📅 Covers 2011-2024, will be continuously updated.
-
🏗️ Restores as much data as possible through non-invasive data imputation, statement articulation, and harmonization.
The RFSD is hosted on 🤗 Hugging Face and Zenodo and is stored in a structured, column-oriented, compressed binary format Apache Parquet with yearly partitioning scheme, enabling end users to query only variables of interest at scale.
The accompanying paper provides internal and external validation of the data: https://doi.org/10.1038/s41597-025-05150-1.
Here we present the code used to create the data set, as well as the instructions for importing the data in an R or Python environment and use cases.
You have two options to ingest the data: download the .parquet files manually from Hugging Face or Zenodo or rely on 🤗 Hugging Face Datasets library.
It is as easy as:
from datasets import load_dataset
import polars as pl
# This line will download 6.6 GB+ of all RFSD data and store it in a 🤗 cache folder
RFSD = load_dataset('irlspbru/RFSD')
# Alternatively, this will download ~540 MB with all financial statements for 2023
# to a Polars DataFrame (requires about 8 GB of RAM)
RFSD_2023 = pl.read_parquet('hf://datasets/irlspbru/RFSD/RFSD/year=2023/*.parquet')We provide a file in aux/descriptive_names_dict.csv which can be used to change the original names of financial variables to user-friendly ones, e.g., B_revenue and CFo_materials in lieu of line_2110 and line_4121, respectively. Prefixes are for disambiguation purposes: B_ stands for balance sheet variables, PL_ — profit and loss statement, CFi_ and CFo — cash inflows and cash outflows, etc. (One can find all the variable definitions in the supplementary materials table in the accompanying paper and consult the original statement forms used by firms: full is KND 0710099, simplified — KND 0710096.)
# Give suggested descriptive names to variables
renaming_df = pl.read_csv('https://raw.githubusercontent.com/irlcode/RFSD/main/aux/descriptive_names_dict.csv')
RFSD = RFSD.rename({item[0]: item[1] for item in zip(renaming_df['original'], renaming_df['descriptive'])})Please note that the data is not shuffled within year, meaning that streaming the first n rows will not yield a random sample.
Importing in Python requires pyarrow package installed.
import pyarrow.dataset as ds
import polars as pl
# Read RFSD metadata from local file
RFSD = ds.dataset("local/path/to/RFSD", partitioning="hive")
# Use RFSD_dataset.schema to glimpse the data structure and columns' classes
print(RFSD.schema)
# Load full dataset into memory
RFSD_full = pl.from_arrow(RFSD.to_table())
# Load only 2019 data into memory
RFSD_2019 = pl.from_arrow(RFSD.to_table(filter=ds.field('year') == 2019))
# Load only revenue for firms in 2019, identified by taxpayer id
RFSD_2019_revenue = pl.from_arrow(
RFSD.to_table(
filter=ds.field('year') == 2019,
columns=['inn', 'line_2110']
)
)
# Give suggested descriptive names to variables
renaming_df = pl.read_csv('local/path/to/descriptive_names_dict.csv')
RFSD_full = RFSD_full.rename({item[0]: item[1] for item in zip(renaming_df['original'], renaming_df['descriptive'])})Importing in R requires arrow package installed.
library(arrow)
library(data.table)
# Read RFSD metadata from local file
RFSD <- open_dataset("local/path/to/RFSD")
# Use schema() to glimpse into the data structure and column classes
schema(RFSD)
# Load full dataset into memory
scanner <- Scanner$create(RFSD)
RFSD_full <- as.data.table(scanner$ToTable())
# Load only 2019 data into memory
scan_builder <- RFSD$NewScan()
scan_builder$Filter(Expression$field_ref("year") == 2019)
scanner <- scan_builder$Finish()
RFSD_2019 <- as.data.table(scanner$ToTable())
# Load only revenue for firms in 2019, identified by taxpayer id
scan_builder <- RFSD$NewScan()
scan_builder$Filter(Expression$field_ref("year") == 2019)
scan_builder$Project(cols = c("inn", "line_2110"))
scanner <- scan_builder$Finish()
RFSD_2019_revenue <- as.data.table(scanner$ToTable())
# Give suggested descriptive names to variables
renaming_dt <- fread("local/path/to/descriptive_names_dict.csv")
setnames(RFSD_full, old = renaming_dt$original, new = renaming_dt$descriptive)- 🌍 For macroeconomists: Replication of a Bank of Russia study of the cost channel of monetary policy in Russia by Mogiliat et al. (2024) —
use_cases/interest_payments.md - 🏭 For IO: Replication of the total factor productivity estimation by Kaukin and Zhemkova (2023) —
use_cases/tfp.md - 🗺️ For economic geographers: A novel model-less house-level GDP spatialization that capitalizes on geocoding of firm addresses —
use_cases/spatialization.md
To the best of our knowledge, the RFSD is the only open data set with up-to-date financial statements of Russian companies published under a permissive licence. Apart from being free-to-use, the RFSD benefits from data harmonization and error detection procedures unavailable in commercial sources. Finally, the data can be easily ingested in any statistical package with minimal effort.
We provide financials for Russian firms in 2011-2024. We will add the data for 2025 by July, 2026 (see Version and Update Policy below).
Although the RFSD strives to be an all-encompassing database of financial statements, end users will encounter data gaps:
-
We do not include financials for firms that we considered ineligible to submit financial statements to the Rosstat/Federal Tax Service by law: financial, religious, or state organizations (state-owned commercial firms are still in the data).
-
Eligible firms may enjoy the right not to disclose under certain conditions. For instance, Gazprom did not file in 2022 and we had to impute its 2022 data from 2023 filings. Sibur filed only in 2023, Novatek — in 2020 and 2021. Commercial data providers such as Interfax's SPARK enjoy dedicated access to the Federal Tax Service data and therefore are able source this information elsewhere.
-
Firm may have submitted its annual statement but, according to the Uniform State Register of Legal Entities (EGRUL), it was not active in this year. We remove those filings.
We use Nominatim to geocode structured addresses of incorporation of legal entities from the EGRUL. There may be errors in the original addresses that prevent us from geocoding firms to a particular house. Gazprom, for instance, is geocoded up to a house level in 2014 and 2021-2023, but only at street level for 2015-2020 due to improper handling of the house number by Nominatim. In that case we have fallen back to street-level geocoding. Additionally, streets in different districts of one city may share identical names. We have ignored those problems in our geocoding and invite your submissions. Finally, address of incorporation may not correspond with plant locations. For instance, Rosneft has 62 field offices in addition to the central office in Moscow. We ignore the location of such offices in our geocoding, but subsidiaries set up as separate legal entities are still geocoded.
Why is the data for firm X different from https://bo.nalog.ru/?
Many firms submit correcting statements after the initial filing. While we have downloaded the data way past the April 1, 2025 deadline for 2024 filings, firms may have kept submitting the correcting statements. We will capture them in the future releases.
We provide the source data as is, with minimal changes. Consider a relatively unknown LLC Banknota. It reported 3.7 trillion rubles in revenue in 2023, or 2% of Russia's GDP. This is obviously an outlier firm with unrealistic financials. We have manually reviewed the data and flagged such firms for user consideration (variable outlier), keeping the source data intact.
We should stress that we provide unconsolidated financial statements filed according to the Russian accounting standards, meaning that it would be wrong to infer financials for corporate groups with this data. Gazprom, for instance, had over 800 affiliated entities and to study this corporate group in its entirety it is not enough to consider financials of the parent company.
The data is provided in Apache Parquet format. This is a structured, column-oriented, compressed binary format allowing for conditional subsetting of columns and rows. In other words, you can easily query financials of companies of interest, keeping only variables of interest in memory, greatly reducing data footprint.
The below figure explains how we constructed the data set. An annotated Makefile documents the process (with renv taking care of creating virtual R environment with packages required for this project). Please be aware that in order to replicate it one has to have an access to the fee-based API of the Federal Tax Service of the Russian Federation, and also be in possession of the panels of all active organizations and their classification code, and geocoded addresses. We built them outside of this project from the official sources — the Uniform State Register of Legal Entities (EGRUL) and Rosstat's Statistical Register of Economic Entities — and do not provide here.
├── aux
│ └── descriptive_names_dict.csv
├── code
│ ├── 1_financials
│ │ ├── 1a_collect_rosstat_data.R
│ │ ├── 1b_build_rosstat_panel.R
│ │ ├── 2a_collect_fns_xmls_ids.R
│ │ ├── 2b_collect_fns_xmls.R
│ │ ├── 2c_parse_fns_xmls.R
│ │ ├── 2d_build_fns_panel.R
│ │ ├── 3_build_filing_panel.R
│ │ ├── 4_combine_rosstat_fns_panels.R
│ │ ├── 5_build_articulation_panel.R
│ │ ├── 6_adjust_values.R
│ │ └── helpers
│ │ ├── check_articulation_functions.R
│ │ ├── lines_tags_dict.R
│ │ └── parsing_functions.R
│ └── 2_geocoding
│ ├── 1_set_up_nominatim_server.sh
│ ├── 2_export_addresses.r
│ ├── 3_query_nominatim.r
│ ├── 4_create_final_mapping.r
│ └── 5_join_results_to_financials.R
├── use_cases
│ ├── external_data
│ │ └── VVP_god_s_1995-2024.xlsx
│ ├── interest_payments.md
│ ├── interest_payments.Rmd
│ ├── spatialization.md
│ ├── spatialization.Rmd
│ ├── tfp.md
│ └── tfp.Rmd
├── figures
│ ├── dataset_construction.png
│ ├── filing_by_month.png
│ ├── interest_figure2-1.png
│ ├── interest_figure3-1.png
│ ├── interest_figure3alt-1.png
│ ├── kommersant_logo_upscaled.png
│ ├── kz_table_1.png
│ ├── mogilyat_figure2.png
│ ├── mogilyat_figure3.png
│ ├── rbc_logo_upscaled.png
│ ├── spatialization_gardenring-1.png
│ ├── spatialization_moscow-1.png
│ ├── spatialization_moscowkummu-1.png
│ ├── spatialization_spb-1.png
│ ├── spatialization_spbcentre-1.png
│ ├── whats_new_2.0.0_filing.png
│ └── whats_new_2.0.0_geocoding.png
├── renv
│ └── activate.R
├── renv.lock
├── AUTHORS
├── CITATION.cff
├── LICENCE
├── Makefile
└── README.md
Version (SemVer): 2.0.3.
We intend to update the RFSD annualy as the data becomes available, in other words when most of the firms have their statements filed with the Federal Tax Service. The official deadline for filing of previous year statements is April, 1. However, every year a portion of firms either fails to meet the deadline or submits corrections afterwards. As the figure below shows, filing continues up to the very end of the year but after the end of April this stream quickly thins out. Nevertheless, there is obviously a trade-off between minimization of data completeness and version availability. We find it a reasonable compromise to query new data in early June, since on average by the end of May 96.7% statements are already filed, including 86.4% of all the correcting filings. We plan to update RFSD annualy in late July — early August.
All notable changes to this project will be documented below. The format is based on Keep a Changelog.
- Added previously missing
okopf,okfc, andokpovalues for new firms entering in 2024.
- Improved
okopfcompleteness: previously the data on this classification code came only from the Rosstat's Statistical Register of Economic Entities whereas now it is sourced from the Federal Tax Service's EGRUL and GIR BO filings (with EGRUL taking precedence). As a result, we were able to fill missingokopffor <1% of firms, futher slightly improvingeligibleclassification. - Impovements in eligibility classification prompted by reduction
okopfmissingness allowed us to removed about 7 thousand non-filing organisations from the data. Now that we had theirokopfwe could confidently classify them as non-eligible non-filers (they had had missingokopfbefore and were treated as eligible non-filers to be on the safe side). Those organisations are primarily government or municipal agencies or religious entities that are not required to file their financial statements.
- Added
regionvalues for years 2011–2013. - Interpolated
region_taxcodevalues for years 2011–2013. This information is only available from the EGRUL from 2014 onwards. To obtain pre-2014 information we use values from the earliest available EGRUL entry of the firm, under the assumption that region of registration has not changed in the 2011-2013 period.
- Resolved a bug in the
regionvariable value labels.
- Removed
zerovariable marking all-zero financial statements.
- Financial statements for 2024 have been added, totaling approximately 2.25 million observations.
- More than 315,000 new statements (as well as 55,000 new non-statements by eligible firms) for 2011–2023 have also been included:
- More than 24,000 statements for 2023 were imputed using the 2024 filings.
- Approximately 3,400 statements were either added to GIR BO retrospectively or missing in v1.0.1 due to the absence of the corresponding organizations in the EGRUL. When we updated the EGRUL, we were able to reconstruct those filings.
- The remaining 286,000+ statements are the ones that we had previously imputed from firms' future statements but erroneously dropped due to a bug (see Fixed).
- Resolved a bug in the final filtering of the panel. We removed all non-eligible non-filers, as intended, but inadvertently excluded observations of non-eligible non-filers whose statements we were able to reconstruct from the following-year filings. Fixing this filter returned about 286,000 statements to the panel.
- The
region_taxcodenow reflects a firm's region of incorporation. Earlier it was simply derived from the first two digits of the firm tax identifier (inn) and did not account for reincorporations. The format has been extended from 2 to 4 digits, allowing for differentiation between Arkhangelskaya oblast ("2900") and Nenetskiy AO ("2983"), which the former includes. - We have revised the statements of all firms previously marked as
outlierand set the flag to 0 where anomalous revenue has been corrected in the GIR BO database retrospectively. - The outlier detection procedure has been updated. Before, we manually reviewed top-20 firms in terms of revenue or total assets within each 2-digit industry (excluding financial firms). Now, we conduct the outlierness review of 2024 filings at OKVED section level, manually examining the top-30 firms by revenue within each OKVED section.
- The
filedflag for observed but all-zero statements (all fields, even Equity, are 0) statements is now set to 0 as it is clearly erroneous. The change this brings is reported below:
- The
exemption_criteriafor eligible organizations is now set to"none"(previouslyNA). - Enhanced geocoding quality (see #1): improvements in address pre-processing procedure have enabled us to upgrade geocoding quality from the city level to the street level for firms accounting for ~8% of revenue from 2022 onwards:
- Fixed a bug in summation of negative lines when calculating line 2400 (net profit). The bug was identified in #7 and the fix is explained in #8.
- Fixed a bug in adjustment of line 1300 (total capital and reserves) and 2500 (result of the period). See #9 for an explanation.
The updated lines 2400 are quite different from the original values. The value of line 2400 changed in 6–11% of observations in 2011-2018 and in about 25% observations in 2019–2023, the summed difference in the original and new values ranges from 5% to 110% depending on year. The fix for sign inconsistency implies revising scripts for all calculations where negative-only, those ()-ed in statement forms, variables were used.
Below is our To-Do list, we will be grateful for any contributions you can make. If you spot a bug, just raise it as a GitHub issue.
- Improve geocoding quality (#1)
- Better outlier detection procedure (#2)
- Better detection of financial firms (#3)
- Improve next-year imputation procedure (#4)
- Explain differences with Mogilyat et al. (2024) (#5)
- Explain differences with Kaukin and Zhemkova (2023) (#6)
Creative Commons License Attribution 4.0 International (CC BY 4.0).
Copyright © the respective contributors, as shown by the AUTHORS file.
Please cite as:
@article{bondarkov2025rfsd,
title={{R}ussian {F}inancial {S}tatements {D}atabase},
author={Bondarkov, Sergey and Ledenev, Victor and Skougarevskiy, Dmitriy},
journal={Scientific Data},
doi={10.1038/s41597-025-05150-1},
year={2025},
volume="12",
issue="1"
}
Kommersant, a leading Russian business daily (January, 2025): https://www.kommersant.ru/doc/7443485
RBC, a leading Russian business information website (January, 2025): https://www.rbc.ru/spb_sz/25/01/2025/678df2229a79470e3b19affb- Business FM Saint Petersburg (February, 2025): https://bfmspb.ru/proekty/intervyu/v-evropejskom-universitete-predstavili-rossijskuyu-bazu-buxgalterskoj-otchyotnosti/
Data collection and processing: Sergey Bondarkov, [email protected], Viktor Ledenev, [email protected]
Project conception, data validation, and use cases: Dmitriy Skougarevskiy, Ph.D., [email protected]