Using public financial data, developed a searchable database on the subsidiaries and industry-presence of US-based multinational companies.
This project served to highlight the size and scope of multinational companies located in the US.
By law, public companies are required to submit financial information, including annual 10-K financial reports, to the US Security and Exchange Commission. Within these annual reports, companies will provide an Exhibit-21 document listing the subsidiaries they own.
In starting this project, we used the Edgar package (https://pypi.org/project/edgar/) to develop a program that would automatically pull and classify Companies and their subsidiaries. Following initial development, we opted to instead use the dataset provided by Corpwatch API (http://api.corpwatch.org/) a non-profit organization that had pulled and classifed companies in the SEC. Seeing as how Corpwatch utilized the same strategy of pulling 10-K and EX-21 documents to highlight parent-child relationships, we felt confident in our instincts for tackling this problem.
Using Corpwatch datasets, we created a master dataset allowing us to query information on names, locations, and industries. Using a command line interface, users can query specific companies or industies.
- When querying by company, the program will display any companies related to it (either parent or children). For multinational companies, the program would also generate a report that would show all subsidiaries, locations, and industries associated with that company.
- When querying by industry, the program will identify multinational companies that owned subsidiaries in the selected industry. Using Pandas, we would then generate a list of the top 25 multinationals in that field based on number of subsidiaries owned, usually identifying unexpected relationships or stakes.