CCI's recommendation for charter cities is that it would be best to develop the initial infrastructure buildout of a charter city with private capital rather than with public resources, the idea being that a private developer will be more responsive to market forces and take a long-term view of their investment. So this is one type of cost, let's call them infrastructure costs.
The second set of costs are the resources dedicated to creating and implementing the governance innovation. This is drafting and negotiating legislation or a concession agreement, determining institutional arrangements, identifying governing officials, and so on. Let's call these governance costs.
I think you can discount the infrastructure costs because this is private capital that would go into a variety of other projects, presumably a large chunk of it in real estate, in the absence of the charter city project. And since we're talking about lower income countries, those alternative investments are probably not creating much social value, or are going somewhere else entirely. If the charter city is successful, the costs are recouped by the developer through the increase in land value brought on by economic activity. If the charter city does not take off, the capital is lost, which is bad, but seems like this consideration should be relatively unimportant from an EA point of view. And even if the charter city doesn't take off, a power plant built for the charter city, for example, could obviously continue to function for the surrounding region. Note that if there was government financing involved in this first area, then I think the calculation changes because the state is then being forced to make tradeoffs with more obvious implications for social cost.
So if you're willing to accept that we can discount those infrastructure costs for the reasons outlined above, you're left with the governance costs. The value proposition then becomes: for the governance costs outlined above, a successful charter city creates a sustained high-growth environment that reduces poverty and leads to broader reform. I think that for a few million dollars (with some variance in either direction from case to case) you can get most of the elements mentioned above in place. From the point of view of an EA looking to direct resources, these are the only set of costs they're going to interact with.
Charter cities are certainly high-risk and high-uncertainty, but the return from even just one successful charter city is quite large. And after there's a direct proof of concept rather than various linkages to similar examples from the recent past (Shenzhen, etc), the risk and uncertainty decline. I appreciate that this calculated bet won't be attractive to a lot of EAs, but I think it's worth exploring. Recreating just a fraction of the successes of China, the Asian Tigers, Botswana, Vietnam, etc in countries still seeing low or highly variable growth is a better counterfactual than the path these countries are on now and EAs should be trying to do something about it, charter cities or otherwise. I think this fits into a larger discussion about calcification within EA, but that's a topic for another thread.
It's interesting to me that your takeaway is that pushing for governance innovation is cheaper in existing polities rather than newly created polities. If your definition of "cheaper" factors in the cost of building new cities, then I get your point. I think there are good reasons to discount that cost that are discussed in the report, but regardless, do you think there is something unique about German institutional arrangements that allows for cities to better overcome collective action problems?
Cities in the US and most (all?) developing countries seem to really struggle with overcoming the collective action problems that stand in the way of achieving anything that isn't extremely marginal. For example, it seems as though if there were significant public support for infrastructure project X or major policy reform Y in <pick an African megacity> that it's pretty unlikely that the project can be attempted, let alone successfully completed.
Maybe it's a function of state capacity? High capacity states (I think of Germany as having greater state capacity than the US, for instance) are better equipped to overcome collective action problems than low capacity states because they can credibly commit to a course of action and then efficiently execute on that vision.
Anyway, glad to hear you're interested in charter cities!
Here's a corrective: https://twitter.com/JeffJMason/status/1511663114701484035?t=MoQZV653AZ_K1f2-WVJl7g&s=19
Unfortunately I can't do anything about where it shows up. Elon needs to get working on that edit button.