If the overwhelming majority of Californians want a socialist nanny-state, then that is okay with me. If Mississippians want zero government, low education levels and poor public health, that's fine with me as well. Sometimes people will say, "If you don't like X, (policy, law, politician) leave the country." That's not really feasible. When it comes to the state level, it's vastly more flexible. It's not perfectly ideal, even better would be decentralization to the city level which would allow for simpler Tiebout sorting. However, some programs rely on economies of scale, and are simply not feasible for the lowest level of government.
Here are the Tiebout assumptions
1. Mobile Consumers: Consumers are free to choose where they live. There are no costs associated with moving. Obviously there are costs associated with moving, but national mobility doesn't exist really. State does, and municipal certainly.
2. Complete information Obviously complete information is difficult, but it is easy to research social programs, tax levels, and public services. Ie- Property taxes, public school levels, etc.
3. Many Communities to chose from There's one nation. 50 states. Thousands of municipalities.
4. Commuting is not an issue The Tiebout model reflects more municipal living, not Federalism, so I ignore this one.
5. Public Goods do not spill over of benefits/costs from one community to the next This is an issue, particularly for municipalities. Suburbs are often free riders of city programs.
6. An optimal city size exists: Economies of scale Likewise to states.
7. Communities try to achieve "optimal size" Likewise to states.
8. Commuties are rational and try to keep the public 'bad' consumers away. This assumption reflects more the racial/monetary sorting within municipalities, and isn't really valid for state-to-state migration.