When a large firm has many different divisions, the CEO must often decide how much resource to invest in each one, in order to maximise the value created for the firm as a whole. This decision is rather like the decisions that investors in the stock market make about which shares to buy. For that reason, this type of setup within a large firm is often called an ‘internal capital market’. In theory, the CEO makes their decision on a purely rational basis, using the information available. In reality, however, they cannot know everything that is going on in every division. So they actually decide how much to invest by negotiating with their various divisional managers. This means that internal investment decisions are influenced by the relationship between each manager and the CEO.
We wanted to look into the effect on investment decisions when the CEO and managers had certain things in common. For example, what if they had the same career or educational background, if they were a similar age, or if they a similar career history within the firm?
Our theory was that the more that CEOs and managers had in common, the more likely they were to share similar beliefs, interpret information in the same way, share information and belong to the same ‘old boys’ network’. As a result of all these factors, they would trust each other more, putting the manager in a better position to lobby the CEO for more resources.
To test our ideas about these informal connections, we formed two hypotheses: a ‘trust hypothesis’ and a ‘bargaining hypothesis’.
The ‘trust hypothesis’ suggests that more resources will be allocated to divisions where the manager and CEO have a strong connection. In this situation, where there is trust between a manager and the CEO, the manager is confident that the CEO will distribute resources fairly. The managers of the most productive or ‘deserving’ divisions do not feel they have to resort to manoeuvring or underhand tactics in order to wrest resources away from less productive, ‘undeserving’ divisions.
So when the managers of ‘deserving’ divisions have strong connections to the CEO, we expect to see optimal resource allocation, and value being created for the firm as a result. When ‘undeserving’ divisional managers have strong connections, resource allocation will be suboptimal – and value creation will be impaired as a result.
The ‘bargaining hypothesis’ is based on the idea that the CEO allocates capital in such a way as to discourage divisional managers from rent seeking. The phrase ‘rent seeking’ means attempting to manipulate the environment in order to secure an economic advantage – for example, by restricting membership of a trade association. However, rent seeking can only help to secure more of a limited resource – it doesn’t create new value or new wealth.
In our context, rent seeking takes the form of managers putting too much of their efforts into trying to wheedle investments out of the CEO rather than doing ‘proper work’ that creates value for the firm. ‘Undeserving’ divisions get too much resource, because it is more profitable for their managers to focus on rent seeking, instead trying to improve performance. In this context, informal connections increase managers’ bargaining power with the CEO, making rent seeking easier. So while connections bring more investment to a division, they destroy value for the firm because they enable or encourage managers’ rent seeking behaviour.
We tested our hypotheses with a large sample of multi-division corporations in the US from 1996 to 2004. Within each firm, we collected detailed information on executives’ age, education and career history. We then used this to create a ‘connection index’ reflecting how closely a particular manager and their CEO were connected. For example, if manager and CEO joined the firm within the same two-year period, were born within four years of each other, attained their current position within the same two-year window and shared the same career or educational background, then they were defined as ‘connected’ in our model.
We had to rule out the possibility that CEOs chose to appoint connected managers to ‘deserving’ divisions – in other words, that connection is driven by performance, rather than the other way around. To do this, we looked at the prevailing weather conditions in the places where divisions were located. Our assumption was that when a CEO looks to fill a position, they will favour people with whom they have a connection by sending them to locations with a higher quality of living – all else being equal. We found a strong correlation between quality of life and connection, confirming that performance was not the key factor when CEOs made senior appointments.
Our results showed that connected divisional managers do indeed get allocated more resources – even if their division suffers from a cash flow shortfall in a particular year. Also, when the managers of ‘deserving’ divisions are well connected to the CEO, ‘undeserving’ divisions tend to receive less investment. This is consistent with our ‘trust’ hypothesis.
We also created a measure capturing the difference between ‘deserving’ and ‘undeserving’ divisions in terms of their managers’ connection with the CEO. We found that the more difference there was, the lower the valuation of the firm – particularly if the firm as a whole was very diverse in terms of its activities. This finding is in line with what other researchers have discovered, but we propose a different reason for it: that common ground between managers and CEOs has an important influence on the way capital is allocated within firms.
Our research provides further proof for the idea that commonality promotes trust between CEOs and divisional managers. It also helps to advance our understanding of internal capital markets and the ‘theory of the firm’ (economists’ ideas about why firms exist and how they work), and complements new ideas about the role played by influence and trust in economic decisions.
"The Role of Commonality between CEO and Divisional Managers in Internal Capital Markets", published in Journal of Financial and Quantitative Analysis