For the past several years I’ve been writing about criminal justice externalities at the local level, pulling on the loose thread of the correctional free lunch—the observation that local officials make the decisions that send people to prison but the state pays for prison out of general revenues. The phrase was coined by Gordon Hawkins and Frank Zimring in the Scale of Imprisonment; Hadar Aviram provides a great introduction to the theory here (the very introduction, in fact, that piqued my interest). From 2000-2009, counties in California used prison at very different rates, but only 3 percent of the variance can be explained by differences in rates of violent crime. An ongoing question is how to internalize these externalities—I’ve proposed breaking up the state and unifying the pieces of criminal justice at the county level (Leon Neyfakh’s summary of my argument in Slate is here).   My reasons for picking the county level are, inter alia, that DA’s (and judges) are elected at the county level, jails are county institutions, and the Sixth Amendment says juries should come from the county and district of the crime (though I’ve also made the point that crime does not confine itself to county boundaries).
Two recent articles from Colorado underscore the idea that there are also important municipal/intra-county dimensions to criminal justice externalities. In Colorado, several large cities/towns have banned recreational marijuana sales (most notably Colorado Springs). This then means that little towns just outside the city limits can make money from sales (including, notably, from state sales tax rebates). In one instance, a marijuana-selling town also contracts law enforcement out to the county sheriff–meaning that all costs are externalized (assuming crime goes hand in hand with marijuana sales, where the evidence is mixed) but all the benefits are internalized. This strikes me as ripe for a race to the bottom where every jurisdiction ends up authorizing sales because it will end up paying for the costs anyway without reaping any of the benefits.
At the same time, there is also evidence from Colorado about action within the municipal level.  Denver is, apparently, siting marijuana businesses in rich white neighbor–oh, I mean poor neighborhoods of color (sorry, just wanted to break up the predictability a bit). So there, the costs are apportioned intra-city but the benefits (in the form of sales tax rebates) accrue at the city level. It’s the opposite internality/externality problem.
I don’t know that I have a one-size-fits-all prescription about the optimum state/county relationship in criminal justice. (I’m still waiting for the criminal justice equivalent of a theory of optimum currency areas.)  In case you want to read more from me, you can check out my answer about about why we have state prisons—and how they used to be a revenue center, not a cost center–or about how the question about the proper relationship between county and state criminal justice in California (so important after realignment) has actually been with us since statehood.  If  you want to read someone else, a great place to start is  Lisa Miller’s excellent work the Perils of Federalism. In general, though, I think discussions of criminal law in the legal academy tend to focus too much on the federal system and not enough on the state and local—even though there are so many fascinating issues to be explored there.
[Cross-posted at The Reality-Based Community]
