Second Thoughts About Hierarchies

By Alistair Brett

Eric D. Beinhocker in The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics pp 154 - 159, Harvard Business School Press, 2006 notes that managers are urged to flatten organizations and to de-layer hierarchies for improved efficiency. “But counterintuitively hierarchy can serve to increase adaptability by reducing interdependencies and enabling an organization to reach a larger size before gridlock sets in.”

We are used to the idea of hierarchy as a successive set of subsystems. We have a boss; our boss has a boss, we are the boss of someone else, and so forth. The term hierarchical as used to describe innovation systems does not necessarily imply that one level is more important than another as it tends to in common usage.

The recommendation to de-layer hierarchies argument has strengths and weakness. It is based on reducing the numbers of connections between ‘nodes’ in a network (see figure below). Not everyone needs to be copied by default on every e-mail, not everyone needs to take part in every face to face or online meeting. Amazon’s CEO Jeff Bezos was reported as telling his management team to spend less time communicating; small teams should get on with achieving things rather than constantly checking with one another. Such behavior requires a high level of trust among managers. In turn, high levels of trust have the benefit of also reducing the costs of transactions between network nodes.



On the left, each of n nodes is linked to every other one; optimum connectivity. On the right, connecting nodes in a hierarchy produces fewer links. For very large numbers of nodes in extensive networks the number links in a fully connected network becomes n2 and for hierarchies it remains n.

However, it’s not quite so simple. We know from our own experience that not all connections are equally relevant or efficient. If I participate in a business meeting about developing new products but my expertise is in managing the company’s IT systems then the connection between me and the meeting group is not relevant. Likewise if I participate in the same meeting and I know only a little about new product development (plus, I probably don’t know what I don’t know) then my presence is not efficient. The converse obviously holds in both examples.

The trick is to achieve the right balance between on the one hand over-connecting people, resulting in inefficient and slow decision making or even gridlock, but on the other hand not allowing too much authority to reside in a hierarchical few. And, making sure connections are both relevant and efficient. As you might suspect, a certain amount of trial and error experimenting will be needed to achieve the desired balance.

Hierarchies need to be identified. An illustrative case comes from a small country which, as part of an economic development strategy, set about improving commercialization of research from the nation’s universities. To get started many activities were set up with a focus on training for university Technology Transfer Office (TTO) staff in the basics of technology commercialization. Their focus was on setting up training (a common situation, but not necessarily effective) and mentoring (often neglected, but frequently more effective) for TTO personnel. This was a necessary initiative but not a sufficient one. Missing at a higher level of hierarchy was a necessary policy and practical commitment to support TTOs. Only short term contracts were given to TTO staff, there were no incentives or reward structure in place, and TTO staff felt that the university had no confidence in their abilities or dedication. Missing at the lower level were mentors with hands on industry experience. The effort faltered.

This story illustrates the importance of inputs from lower levels being a part of a system’s hierarchical design. Tacit knowledge (knowledge which is mostly in our heads) resided at lower levels where employees often have knowledge and information that higher level management does not. Furthermore, when decisions are made at lower levels those making the decisions are likely to want to implement them.

So, before we condemn hierarchies as something to pull down we may want to re-think where they are beneficial – or as the section title in Beinhocker’s book puts it “Two Cheers for Hierarchy!”