Centrally Planning Decentralization

Yesterday, Arnold Kling offered a suggestion for how to return America to federalism. He doesn't explain why he likes federalism so much, so I'm assuming he likes federalism for the same reasons most people do: increased government accountability, greater local autonomy, smaller jurisdictions acting as policy test tubes, etc. Underlying all of these reasons, and usually the desire for federalism itself, is a belief in the efficiency of decentralization. Locals are better informed about local issues and have more at stake than some far-away central government, and so are more likely to make the right decisions for the local population. I'm sympathetic to this line of thinking myself, but I have a few issues with Kling's plan.

1. Turn any area of 500,000 or more people within an area of 100 square miles or less into a city-state. City-states would be autonomous, other than their participation in the United States as a whole. […] Each city-state should be governed by a single elected individual, whose powers are limited by a Constitution, but who rules until death, resignation, or loss of a recall election. […]
2. Next, we want to create county-states. These would cover larger areas than city-states, and they would be governed the way states are today, with elected legislatures and governors. […] If there are many small counties adjacent to one another but not near a large county, then merge the small counties together. […]

First, this strikes me as an excessively centrally-planned approach to making things less centrally planned. Kling knows exactly how many people a city-state should have over how wide of an area, and if that means forcefully breaking up cities like New York or LA, so be it. Moreover, he knows exactly how they should be governed, with a one-size-fits-all constitution for all of his new city-states. Each is to be governed by a single individual, with no legislature. He has a very specific vision for this new government that allows no diversity of opinion and very few checks and balances.

Second, much of the plan seems simply arbitrary. (To be fair, he introduces the post as a pre-Superbowl "daydream sort of post.") Some current states and cities are "ridiculously large" while some are too small to warrant equal footing. County-states have more flexibility, and are allowed legislatures and further decentralized governments within the county-state. But that only stresses the arbitrariness of the dividing line between city-states and county-states. The ideal size for a city-state, according to Kling, is a population of 500,000 over 100 square miles, or a density of 5,000 people per square mile. (Interestingly, Kling apparently grew up in St. Louis, Missouri, a city which then had a little more than 500,000 and now has a little less than 500,000, with a density fairly close to 5,000 people per square mile.) Cities like Columbus are too spread out to be city-states, he says, while cities like New York or so huge they should be split up into as many pieces as necessary to make them fit his ideal city size. But it's just a number pulled out of thin air. The same is true for the governance structure. Why should the governor of a city-state be necessarily elected to a life term with no legislature in his/her way, while the governor of a county-state has to deal with a legislature? Not only is his approach excessively centrally-planned, it seems as if it's randomly planned.

For all I know, maybe New York would be better off split into multiple independent cities. Maybe Montana and Idaho would be better off merged into a single state. Maybe some cities would be better off without city councils. But shouldn't those be decisions made by the people who live there? Shouldn't they be allowed to decide for themselves how their governments are structured? What works for New York may not work for LA or Seattle or anywhere else for that matter. That's the advantage to federalism in the first place-- the efficiency of decentralization. Locals use local knowledge to make the best decisions for the local population.

As it stands, it's extremely difficult for specific jurisdictions to change their boundaries. Maybe the solution is to relax those constraints. If two jurisdictions want to alter their boundaries, let them. Make it easier for jurisdictions to change on their own, when they want, in the ways they want, and over time we'll see organic movements towards optimality. No one has to arbitrarily decide for the entire country what the local optima are. The local jurisdictions will do that on their own.