The Organizational IQ Gradient
Leaders who hire people smarter than they cause what I call the organizational IQ gradient.
A direct and interesting consequence of the advice to hire direct reports smarter than you is that, if it happens at each level in a company, the smartest people are individual contributors and the dumbest are at the top, on average.
Most businesses have between three and seven layers in the hierarchy. Let’s assume that the intelligence at each level is a factor \(0< p < 1\) of the one that is immediately below it. The average intelligence at level \(L\) is therefore \(\mathrm{IQ}(L) = \mathrm{IQ}_{0}p^{L}\) with \(\mathrm{IQ}_{0}\) the IQ at the level of individual contributors. The more levels in the hierarchy, the greater the difference in IQ between ICs and leadership.
Note that I am talking about averages: individuals at level \(L\) in the corporate hierarchy can have higher IQs than anyone at \(L-1\). On average, the IQ at level \(L\) is a factor \(p\) lower than at \(L-1\) though. Bear with me on IQ as a measure of general intelligence. I have more to say on that in a moment.
If we have, say, \(L=4\) layers of management, the ICs are university educated with an average IQ of 114, and the IQ drops only 1% at each level (\(p=0.99\)), then the C-suite has an average IQ of 110, which happens to be in line with the average IQs of business administrators and accountants.
If your ICs are STEM graduates, their IQs probably hover around 125 and top management at such companies have 120 IQs with a 1% drop at each management layer. To reach an average executive IQ of 110 as before, \(p\approx0.97\).
I am aware that IQ as a measure of intelligence is extremely flawed, especially since it does not measure social intelligence, adaptability, or creativity. Emotional intelligence (EQ) is at least as important in leadership too. And a high IQ does not mean someone is necessarily smart, competent, or ethical. Still, the simple calculation matches the STEM-to-MBA intelligence gradient found in many tech companies.
Generalists vs specialists
Can the organizational IQ gradient be explained by the difference between specialist ICs and generalist managers?
There is no conclusive evidence that the IQs of generalists or specialists differ much. What is more, the line that separates generalists from specialists is fuzzy. Generalists do have an edge in established environments, whereas specialists are best in dynamic fields, where expertise is needed to identify opportunities.
Promotions are not merely related to expertise, but also impact, communication skills, the ability to navigate corporate politics, and dare I say luck? Beyond that, executives need the right relationships with investors, clients, and peers. These qualities are obviously not captured by IQ.
Emotional intelligence
The connection between IQ and EQ is questionable. Some studies report a small but positive correlation between cognitive abilities and emotional intelligence, while others report a small but negative correlation. What is clear is that the relationship is at best very weak. It therefore cannot explain the organizational IQ gradient: we would expect IQ to be negatively correlated with the skills and types of intelligence needed to be promoted.
Age-related decline
The C-suite is typically in their 50s. Age, not intelligence, that is. So, perhaps the organizational IQ gradient is a consequence of the age at which people reach higher rungs of the corporate ladder? Research on the decline of intelligence as a function of age is, however, not clear-cut. Certain types of intelligence do drop after the age of 20, but crystallized intelligence peaks much later in life. Ageing is therefore not the singular cause of the organizational IQ gradient.
Peter principle
The Peter principle states that people in a hierarchy rise to their level of incompetence. Success at a given level \(L=\ell\) does not imply success at a level \(L>\ell\), which is how people remain stuck at the level of their incompetence. While IQ is a predictor of success, it is by no means the only or even the main one. The problem of the Peter principle is that if IQ is only modestly correlated with success in an organization, you would expect the cream of the crop to be on top, at least on average. That is the exact opposite from what we observe.
If promotions are perfectly random, each layer in the hierarchy ends up with the same average IQ as the one immediately below it.
When each layer is sampled uniformly from the layer directly below it and the bottom one is a normally distributed, every layer's IQ will be normally distributed with the same mean and a scaled standard deviation.
If promotions are entirely dependent on intelligence, then each layer in the hierarchy must have a higher IQ on average than the one that preceded it.
This is because sampling from the top individuals of the layer directly below the current one leads to a truncated normal distribution at each layer.
That truncated normal distribution has a higher mean because of the way we sample (i.e. promote) individuals.
In both cases I assume we replace (i.e. hire) individuals with roughly the same IQ scores.
So, if promotions are somewhat correlated with IQ, but not fully, we are in a situation somewhere in the middle of both extremes, which means you would still expect the higher echelons to contain people with higher average IQs than what we actually observe.
So, what?!
While the causes for the organization IQ gradient still elude us, its existence is important for two reasons. First, leaders must be intelligent enough to recognize expertise in others to be able to hire people smarter than they. Anyone can be oblivious to their lack of knowledge, especially since individuals who lack basic knowledge tend to overestimate their overall knowledge. Second, employees managed by people with technical expertise are more engaged than those under administrators who lack specific technical domain knowledge. Business degrees do not confer such qualities onto managers.