This doesn't always happen but when it does, assuming economic efficiency and perfect information, there are some good reasons:
* Managers' actions take effect across the entire team, and therefore having a 1x manager vs a 1.1x manager on a 10 person team is like if any individual (assuming evenness in the team) goes from 1x to 2x.
* It is harder to find someone who can effectively manage human beings than someone who can effectively solve technical problems. Mildly supportive evidence is the percentage of people who complain about their managers vs. the percentage of people who complain about how hard the problems they have to solve are. Your maximum attainable compensation is the minimum of the value you provide and the cost to replace you, and good managers are rare.
* In practice, few people are pure managers - they also solve technical problems. Often, they will participate in architecture questions, but not implementation. If they do so, their technical contributions are also on a lever.
However, there are secondary factors too:
* Managers tend to have more experience. More experience, until a point, leads to higher compensation because successful experienced people are rarer
* The depth to which humans perceive contribution to success is limited. A CEO will see whether his engineering division is effectively delivering value and reward or punish its organizational leader. Likewise all the way down the chain. This is leverage in terms of responsibility and risk.
But the short form of my theory is that where it's true, it's often because they deliver comparatively higher value to the organization.
If it's true, a consequence would be that the organizations that have a culture of self-organization and alignment will have managers that command smaller multipliers solely on their management skills while those whose members require substantial management (for mediation, communication, or prioritization) will place a premium on managers.