We’ve recently added network visualisations or “node-link” charts to Flourish, so it’s now easy to make these from your spreadsheets. To demonstrate what they can do, I’ve visualised the corporate networks of some large UK companies. The networks are derived from the new Persons of Significant Control data, which requires companies registered in England & Wales to declare any other companies that control them, as an owner or shareholder.
This fascinating data gives an insight into the increasingly sprawling structures of modern corporations. However, the raw data at Companies House is complicated to unpick. For example, BP Marine Limited lists its PSC as BP International Limited, which in turn lists its PSC as BP PLC. Long, complex chains are common! And there’s no way to see all the companies controlled by another company. This is where network graphs come in handy.
One company with a complex structure is Persimmon PLC, the housebuilder – in the news recently because its CEO received a £74 million bonus. It’s the “ultimate” controller of no fewer than 587 other companies at Companies House.
Clicking through the slides, you can see how Persimmon has grown over the years by acquiring numerous other housebuilders – Ideal Homes, Beazer Homes, Westbury, and Prowting. And the chart shows how more than half the network are special purpose vehicles controlled by Persimmon Homes Limited, mostly related to a single site.
Creating this visualisation in Flourish required no coding skills – the only hard part was getting the data. Learn how to make a network graph without coding in Flourish.
I featured Persimmon because it controls lots of other companies – the third most of any FTSE 100 company, after Barclays and HSBC. (FTSE 100 companies themselves are publicly listed, so don’t generally declare any PSCs, since they’re controlled by lots of smaller shareholders.)
The Flourish bar chart below lists the companies with the largest networks – click or tap on the bars to see the numbers.
The median number in the FTSE 100 is about 50 companies. But note that these numbers are underestimates, because:
- A few companies fail to declare their PSCs, so will be missing from this analysis (about 3% of companies, as of October 2017).
- Companies outside England and Wales are excluded from the Companies House data, so for example RBS doesn’t appear because it’s registered in Scotland. This also include companies headquartered in Jersey, Isle of Man etc.
- If a company owns less than 25% of another company, it won’t be declared.
Here’s a fairly “typical” company structure. It’s Rio Tinto, the mining group, which is the ultimate controller of around 50 companies registered at Companies House.
But the top end of the chart looks very different. Five out of the top 10 companies are housebuilders, or involved in land – Persimmon, Berkeley Group, Barratt Homes, Taylor Wimpey and Land Securities are all in the top 10. Most of these look like the Persimmon chart above, with many small groups of special purpose companies set up for individual sites.
BP is another interesting case. The visualisations below show how BP itself has a fairly “typical” structure, but when it acquired a $200 million stake in solar developer Lightsource, it also acquired corporate control over more than 200 new companies:
It would be interesting to know whether company structures are generally getting more complex.
Once I’d got the data (which I did by crunching the raw PSC data with Python), making the network graphs was as easy as uploading two CSV files, then choosing some settings. You can try it with your data here.
Of course, corporate networks aren’t the only thing you can visualise with network graphs – they’re also good for data from social networks like Twitter and LinkedIn, and social structures generally. If you’d like to read more about network charts, Lynn Cherney’s overview is a great place to start.
You’re welcome to embed any of the charts above in your own website. Just click on the “Made with Flourish” links above, then click the “Embed” button on the header.