At the most basic level leadership is at a premium because there are fewer truly good leaders than there are developers.
Now, is every well paid CEO a truly good leader? No. But I’d argue it’s a lot more difficult to assess a good leader than a good developer. There’s a lot more people management, marketing and other soft skills involved. Not to mention the time scales for success are much larger.
I'd say it's actually easy to evaluate a leader, at least one with experience under their belt. You look at the performance of their organization. If they're a CEO, it's especially trivial: you look at the performance of the entire company. If they're a VP or Director of a smaller section of the org, you can still measure that section's performance, though it does get a little muddier.
At a decent-sized org it is nearly impossible to measure an individual developer's impact with any granularity.
I think this should also illustrate that evaluating a leader and evaluating a developer are two entirely different things that can't be directly compared anyway.
The thing is, it benefits the people in professional management roles to promote this point of view and institutionalize practices like outside hires in order to make it a self-fulfilling prophecy. They don't need to conspire about this (in the Adam Smith sense), just echo the notion of corporate management as such a distinct skillset that it needs to be sought out rather than grown and that will eventually come to have its own moat effect.
Eventually you get CEOs making statements that amount to 'I deliver 10x or 20x more value to this company than you, all my CEO peers say so.'
Now, is every well paid CEO a truly good leader? No. But I’d argue it’s a lot more difficult to assess a good leader than a good developer. There’s a lot more people management, marketing and other soft skills involved. Not to mention the time scales for success are much larger.