I think people in social sciences tend to ignore the most obvious factor: military power. A successful campaign needs rapid coordinated efforts, which are best delivered by hierarchical models. And if you suck at war, chances are that your civilization won't stick around for very long.
Here's the paper:<p><a href="http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0024683" rel="nofollow">http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjourna...</a><p>The crucial idea:<p>> Our simulation results support the hypothesis that socioeconomic stratification may have spread due to its effects on the demography of small groups—i.e. by demic diffusion—rather than cultural adoption. [...] In variable environments, stratified groups migrate more and are less likely to go extinct than egalitarian groups.<p>The main contribution of the paper seems to be the identification of a survival advantage for unequal societies in unstable environments.<p>To make this more concrete, imagine two neighboring villages A and B, hit by a catastrophic crop failure that has reduced harvest by 50%. In village A, there is a strong ethos of equality; the dwindled resources are divided equally, and basically midway through winter everyone dies of starvation.<p>In village B, the inhabitants strongly believe, on the contrary, that tradition (gods, spirits, ancestor cults etc.) orders the elevation of some above the others. The favored individuals get their normal full share of the reduced harvest, and the un-favored get none. This results into an immediate 50% mortality; but the favored survive through the full winter.<p>After a series of adequate harvests, the B population increases enough to split off a migrating group that takes over the (now abandoned) site of village A. The ethos of inequality is taken with them, and thus spreads at the expense of egalitarianism.