[empyrrhicist]

> Obviously this system is very inefficient but nobody can deny that some CEO are better than others and that the best ones can make decisions which bring 10-100x or more money to the company than whatever they are paid.

There's nothing obvious about this to me - how would you distinguish that from survivorship bias? Why then do shareholders and boards often come into conflict on this issue?

Claiming that someone is worth paying 100m per year is an outrageous claim that requires commensurate evidence, not a wishy-washy statement about boards thinking its worthwhile. Boards are made of people in a very small oligarchic circle - their behavior is more easily explained by remembering that they're social animals in a hierarchical context than pretending this is all perfect economic rationality.

[ztrww]

> There's nothing obvious about this to me - how would you distinguish that from survivorship bias?

So success of companies is entirely random? That seems statistically unlikely…

I mean are you really saying that there are no decisions that CEOs regularly take due to which a company might lose/gains up to millions to billions of dollars? Why wouldn’t you pay a CEO whose actions can bring the company billions a 100m or so? Seems like a good deal..

e.g. if you put a random highly competent, educated and very hardworking person in charge of Apple back in 1997 is it more or less likely that he would have done better than Jobs?

I mean, yeah I agree with you in part. In most cases it’s hard to distinguish real impact (even after a few years) from survivorship bias which is why this whole process is so inefficient. I’m sure that quite a few companies are just as likely (if not more likely) to hire a 100+ mil CEO who’ll be a net negative on as one who’s action will bring 10x+ in additional revenue compared to what most other candidates would have.

[empyrrhicist]

> So success of companies is entirely random? That seems statistically unlikely…

You're claiming that it's worth paying CEOs 100M+, and my point is that whatever level of competence/skills/judgement/character you desire for performance should be achievable at an order of magnitude lower salary offering. My point is that this astronomical compensations represents a really outrageously strong market claim about someone's performance, but the only evidence you're offering is that some companies do better than others. That is exactly what we'd expect in a market with many companies trying different things - aka, survivorship bias.

This is super common in tech, where people assume that successful entrepreneurs (for example) are visionaries, when in fact they had to be BOTH visionaries and ALSO incredibly lucky, and any structural advantages they had to acquire capital also help a lot.

> I mean are you really saying that there are no decisions that CEOs regularly take due to which a company might lose/gains up to millions to billions of dollars? Why wouldn’t you pay a CEO whose actions can bring the company billions a 100m or so? Seems like a good deal..

This is disconnected - that does indeed happen, and it's been interesting to watch e.g. Elon Musk set fire to several billion dollars along with his reputation over the past few years. My argument though isn't about whether or not CEOs are important, but whether or not their compensation is actually tied to their real value, or whether it reflects their membership in an oligarch class and their skill/luck in navigating that socioeconomic environment. This is a question about opportunity cost. Note - I would also be willing to believe that the perception of having a genius CEO who is "appropriately" compensated on a cosmological pay scale is valuable as a social signal in and of itself, and I would also object that this is irrational. In an economic system that capped CEO pay, that social signal could still exist unmodified at much lower absolute levels of compensation.

> e.g. if you put a random highly competent, educated and very hardworking person in charge of Apple back in 1997 is it more or less likely that he would have done better than Jobs?

Weird example, since Jobs took a salary of $1 during that time.

> I mean, yeah I agree with you in part. In most cases it’s hard to distinguish real impact (even after a few years) from survivorship bias which is why this whole process is so inefficient.

Couple that problem with ANY reasonable prior on the distribution of abilities in human populations and the distribution of salaries, and you end up with exactly what I'm proposing - CEO salaries are easier to explain by looking at human societies than any possible collection of individual attributes that such a CEO could possess.