As we’ve explained previously, a modern corporation (even a small one) is actually a complex network of related legal entities connected together in a web of relationships. At OpenCorporates we call these corporate networks.
But what’s the nature of these relationships, how do we identify them, and what do they really mean? That’s what this series of posts is designed to explain. We hope it will act as a useful primer on corporate structures. It aims to be both technically accurate and understandable to novices, and to explain why open data is key to creating accurate corporate network data.
A corporate relationship between two companies is often described as one company being a subsidiary of another. The notion of a subsidiary is fairly well established in law across the world. In the US, for example, it is “an affiliate controlled … directly, or indirectly through one or more intermediaries.”
But can we unpick this notion of control? What does it mean? And where does power to exercise the control come from?
The best-known method of control is via share ownership. That’s the way most owners of small and medium companies control their companies — specifically, through shares that have voting rights. It’s the only formal definition of control most people (including some government people) are aware of.
It’s not just people that can own shares, but companies too, and this is a common mechanism for control in corporate networks. There are different types of voting (and non-voting) shares, but equity control is usually exercised through what are known as ‘ordinary shares’ or ‘common stock’. In most companies, many decisions can be carried with a simple majority of over 50% of the votes.
Using this threshold definition of control, let’s consider the corporate control network of Ingham Mora Limited, a random New Zealand company. New Zealand is a great country to investigate, because all its shareholding information is available as open data in OpenCorporates (in fact we consider it to be probably the best major company register in the world for access to data).
Every statement about a shareholding has its own page on OpenCorporates. For example, here’s a statement showing 600 shares issued by Refund Me Limited to Ingham Mora. Notice how we record date information, the source of the information, and the confidence we have in that information.
Overall, the graph of control for Ingham Mora, with a 50% threshold for control, looks like this:
There are 32 relationships of control in the network shown. Each arrow shows the direction of control.
But it turns out, it’s not as simple as that. A minority share can sometimes give you control.
What if a company has 30% of the voting shares, and 70 other people have 1% each? It’s quite likely that such a company can effectively control decisions, by using its size to encourage or pressurise the minority shareholders, or simply because of the difficulties of small shareholders acting together. In fact, 30% is the threshold deemed to give a company “effective control” in the City of London’s Code on Takeovers and Mergers. Using this definition of control would increase the Ingham Mora network to 39 relationships.
But it turns out 30% is quite a conservative threshold to define control. In US banking regulations, control is defined as having 25% of any voting shares; the same level implies control in UK money laundering regulations. Using a threshold of 25% would extend the Ingham Mora network to 40 relationships.
The threshold model is not the only way of attributing control within regulatory frameworks. The US banking code also suggests a relative control model: where the presumption is that any person with 10% of voting shares exercises control, as long as no-one else has more than them. At a 10% relative control cutoff point, Ingham Mora has 112 relationships in its network.
Finally, you might quite reasonably consider that control accrues in proportion to the share of equity an actor has – the one-share-one-vote model. In other words, someone with 3% of equity wields 3% of the power. In this model, there are 163 relationships in the network:
So how would you define control? Academic studies of control tend to use a range of 10% – 20%. If you’re a journalist looking for story ideas, you might be interested in the one-share-one-vote model, where merest suggestions of influence or control are potentially relevant. Some credit risk models might work on simple majority control using the threshold method with a share of 51%.
What makes the OpenCorporates network data unique is that you can pick your preferred model of control – and you can verify our sources to make sure they’re right. Play with the Ingham Mora data yourself by setting desired percentages and confidence levels in our network visualisation tool, and look out over the next few months for more data and more ways to query it.