In this post the other day, pointing out that the real value of the minimum wage was higher in parts of California than it was in parts of Alabama, my primary point was an econometric one. If you insist on doing cross-sectional or time-series first-approximations of the minimum wage effect (I don't think you should - we celebrate Card and Krueger for a very good reason) it ought to be anchored in some way to the local price level.
But it has policy implications too. Most Californians probably see their state minimum wage (rightfully) as an assertion of state-level prerogative to be a laboratory of democracy, but they also probably (wrongly) think of it as an experiment in a higher minimum wage than the federal government. In many places, of course, it may be (and some cities have even higher living wages - fine - the principles of federalism don't end with the state legislature), but for a lot of California it's actually an experiment in a minimum wage that is lower than the federal level.
There's something to be said for federal policies that lift up depressed regions of the country with transfer payments and investments. If we think there may be low wage equilibria and weak employment effects of the minimum wage some federal wage floor might be justified. But overall we're probably better off with a lower federal minimum wage and more state minimum wages.
Perhaps something like an inflation-indexed federal minimum wage set at some not-too-high scalar times the median wage of the twentieth poorest state, or something like that. Then we can be done with these discontinuous hikes and there would be a lot of room for state experimentation. It would leave less for econometricians to do, but that's OK. We'll survive.