The trick is that in large modern states jurisdictional boundaries cover many communities with highly variable levels of cooperativeness. So here’s a question. Will the quality of the government of widest scope tend to average out unequal levels of cooperativeness, such that high-cooperativeness communities will tend to get worse governance than they could provide alone (maybe even worse than they could do without a state) and low cooperativeness communities will tend to get better governance than they could provide alone? It seems that the answer has to be “yes,” but what does this imply?
I think the answer has to be “no.” Cooperative efficacy surely cannot be summed and averaged in the way Will suggests. Cooperation is about the relationships between people, not some characteristic of people themselves: the lines rather than nodes of a graph. It would be possible for a supergroup (no, not that kind of supergroup) to be either more or less cooperative than any of its constituent subgroups. To take an extreme example, Israelis and Palestinians may both have high internal cooperative efficacy, but a polity encompassing both communities (in a very loose sense of the word) is bound to be less cooperative than either on its own. I think it’s probably a fairly general trend that larger aggregations will have less cooperative efficacy than smaller communities. There are good reasons to believe that ethnic cleavages are associated with conflict and low-quality governance.