Demographers abandon methodological individualism by studying whole populations. But they clearly don't think there's some kind of Borg-brain that drives reproduction decisions. They just think that individuals make reproductive decisions, but they make those decisions in consistent enough patterns that it is amenable and useful to study it at an aggregate level.
This is the mistake I think Don Boudreaux is making in this post. You can't think of the community as an economic actor just because it's an important unit of analysis, as Don interprets Krugman's claim here. It's just like (moving away from methodological individualism in the micro direction) a neuroscientist doesn't think a small section of the brain "acts" just because it is an important unit of analysis for studying human action.
Methodological individualism is wrong because it confuses the nature of human action with the imperatives of social science. Methodological individualists should not project their mistakes onto non-methodological individualists. Just because we often think in terms in aggregates does not mean agency is being ascribed to aggregates. You get muddled when you think about it that way. Aggregates don't have agency. But who the hell said we only had to study things with agency?
btw - here's a little Kenneth Arrow on methodological individualism.