Multinational enterprises (MNE)
Introduction
Multinational Enterprises (MNEs) – in other words, companies with business operations in multiple countries – play a key role in the EU economy.
Across all EU Member States, they contribute substantially to the production of goods and services, employment, and investment. For this reason, statistics on MNEs provide important insights about the EU economy and are of growing interest to the EU citizens. Starting from 2023, Eurostat has been publishing experimental statistics on multinational enterprise groups and their structure in the EU, using data extracted directly from the EuroGroups Register.
Eurostat seeks to ensure high accuracy of the data on corporations in the EuroGroups Register. To this end, the EuroGroups Register entries are compared with other publicly available information on MNEs, including among others content extracted from web sources by the Web Intelligence Hub (WIH).
Methodology
Sources
This use case aims to retrieve content from public websites containing information on MNEs that could be used to either cross-check the accuracy of EuroGroups Register’s entries, or to complete its missing data.
The WIH is currently extracting content from public web sources on around 1,500 MNEs and already has a set of developed algorithms for the content collection and its processing.
The current source of the WIH MNE use case is Wikipedia. It offers a broad range of publicly available information on MNEs, including company history, key figures, and URLs to official websites, serving as a source for further exploration. Wikipedia is a rich source of data on MNEs, given that most such enterprises have a well-populated Wikipedia article.
A dedicated WIH pipeline has already been developed for data from Wikipedia, including a workflow, validation rules, data and metadata extraction and processing.
Eurostat plans to extend this use case to additional sources, in particular Wikidata, DBPedia, the Global Legal Entity Identifier Foundation (GLEIF), and the official websites of MNEs.
Feasibility Study
To kick-start its work on MNEs, Eurostat carried out the feasibility study "Smart data for multinational enterprises (MNEs)". The study set out to retrieve public information from the Web for about 199 selected MNEs operating in the EU and EFTA countries, as a proof-of-concept. These included some enterprise groups headquartered outside the EU.
The feasibility study covered five phases:
1. Scoping
It began with a detailed review and assessment of various web sources, focusing on their potential to provide information on MNEs. This phase compared these sources with existing information to identify gaps and evaluated the availability and relevance of specific variables for producing reliable official statistics on MNEs.
2. Web Content Retrieval
It then looked into suitable methods and tools to scrape, retrieve, and extract information from the content retrieved from these web sources, so that it can be transformed into data that can be used by statisticians.
3. Information Extraction
This phase focused on identifying key fields in the raw content, parsing those fields into meaningful variables, resolving inconsistencies, and formatting it into standardized datasets. The variables extracted included company identifiers, employment, and financial metrics.
4. Transformation
This phase focused on combining the data extracted from the content from the several sources, ensuring consistency and coherence across datasets. Data entries were standardized into formats suitable for analytical applications and visual representations, enabling statisticians to generate insights into MNE structures and activities efficiently.
5. Documentation
This phase compiled the previous steps into a report summarizing insights, best practices, and lessons learned. It also provided tools and instructions to support future data collection, processing, and analysis.
Use of MNE data
MNE data is currently available to the national data suppliers of the EuroGoups Register, via the corresponding unit at Eurostat. In the next step, a dataset with an enlarged number of MNEs should be available to be used for other statistical purposes.