How complex is BP? – 1,180 companies across 84 jurisdictions going 12 layers deep

This post is by OpenOil’s CEO Johnny West, following a groundbreaking collaboration with OpenCorporates to open up corporate networks in the oil industry.

At OpenOil we deal with an industry with a reputation for secrecy. So when OpenCorporates put the idea on the table that we should experiment with whether it was possible to create a systemic view of a global oil major entirely from public filings, it was both challenging and irresistible. We settled on taking a look at the British oil giant BP Plc as a test case.

It took a bit of time to get to grips with the quirks of company filings – what you can find there, what you can’t, and how to get data into usable open formats – but by the end we had 1,180 affiliate companies for BP, registered in 84 jurisdictions around the world and in layers of ownership that run twelve layers deep.

We learned a number of things on the way.

The first is what we call the Socrates Principle – we know that we know almost nothing! We have a reasonable “Equity Map” – the shareholdings of BP Plc, the mother company, around the world. But what about the flow of money between all these companies, and into and out of the group network (BP turned over about $350 billion in 2013)? And the relationship between the money and BP’s extensive activities at all stages in the value chain, exploring for, producing, refining, trading and retailing oil products all over the world?

And yet even these exiguous structures throw up all manner of interesting questions. Why do all BP’s Cayman Island incorporations seem to be concentrated on pipelines and the Caucasus? Does it use a Belgian affiliate to handle earnings from the Rumaila oil field in southern Iraq, and if so is that for reasons of tax efficiency? Did a tiny company owned by Price Waterhouse Cooper take over legal compliance functions for huge parts of its network in mid-2010 as part of a “Macondo shield”, to insulate BP executives from possible liabilities over the Gulf of Mexico spill?

The core data by itself cannot answer these questions. But the mere fact of being able to zoom in to individual structures, then zoom out again to the group as a whole, allows questions to form which might never otherwise occur. It is our hope that the data and the visualisation are the spur to others to pursue these and other questions and gain greater insight about the way the oil industry works.

Which leads us to our second lesson – this map is not the end but the beginning of the quest. The map of company structures provides a skeleton from which to hang more information and deeper knowledge. For example, we have financial reporting for 150 of the BP subsidiaries going back a number of years. You need to be corporate accounting experts – and we are not – to detect trends in this data, and responses to real world events BP has faced, such as the incredible financial fall out from the Deepwater Horizon disaster, or current developments in and around Russia. But the open data principle means that if we collect and curate and post this data, maybe someone else can come in and start to make sense of it.

Third, if the open data principle is key, the OpenCorporates URI structure technique is in turn critical to creating the framework to hang information and insight on. Of the 1,180 companies we have found, about 370 of them currently have OpenCorporates entries, mainly because there are still too many company registries in the world that are not open. But all the BP companies we put into the network can be uniquely identified because they are anchored, in documents sourced from BP itself, to other affiliate structures whose incorporation details are fully known and posted in OpenCorporates.

Fourth is a principle we call “Click to Check”. Open data has one big advantage on closed systems – they can be clear about the provenance of the data – and we want to make it easy for users will need to be able to toggle easily between a visualisation layer and the source data that backs each atom of information in the network, each company “node”, each ownership “edge”. We have made a modest start to implementing that in the BP data.

BP network - blogpost image white

What next? We would like feedback, requests for more data coming from people in a position to use information when we find it, and suggestions for what to map next. This is just the beginning.

Finally we should say that there was no purpose in this project to single out BP as a company, nor did we expect to find, nor have we found, any smoking gun.

It is our belief that this degree of transparency, which provides system-level views, should be part of business as usual, that in fact it was part of business as usual in the pre-modern age, when the boundaries of corporate activity aligned much more with nation states, and the potential for public oversight. All we are seeking to do is to return to the balance in business between entrepreneurial freedom and public oversight that existed before global communications and instant capital flight tipped the scales.

Open data sets and visualisations of this kind are not anti-business. They are, in fact, good business, as is being increasingly recognised by policymakers all over the world.

The BP corporate network visualisation done by OpenCorporates can be seen here.
The BP map by OpenOil can be seen here.

There will be a Twitter chat on (#HowBigisBP) on Thursday 11th September, 3pm UK/ 4pm Berlin / 10 am EST Topic: “How can Open Data transform Big Oil – lessons from mapping BP”.