The wonderful Arnold Kling discusses the issue in a recent blog post, "Methodological Individualism, Libertarianism, and Conservatism":
I believe that there is a weakness in libertarian thought, but I am not sure that methodological individualism is the culprit. Consider an analogy. Chemists want to talk about atoms as being fundamental. But in the spirit of the quote above, one could argue that we do not have a substance until we have many atoms bound together. The relationships binding the atoms are what is really fundamental in chemistry. It is impossible to have a substance with just one atom.
I would say that it is useful for chemists to think in terms of atoms, and by the same token it is useful for social theorists to think in terms of individuals. But it is important for chemists to understand the various bonding mechanisms among atoms, and it is important for social theorists to understand the various bonding mechanisms among humans.
I think that what makes conservative social theory of the Burke/Tocqueville/Yuval-Levin sort distinctive is its emphasis on multiple modes of human interaction and bonding mechanisms, including families, organized religion, civic associations, and business enterprises. Libertarians tend to focus almost entirely on free trade as a mode of interaction, and progressives tend to focus almost entirely on the central government as a bonding mechanism.
That the individual is the sole unit of society is a big point made by the Austrian crowd. And they have a point. If one is to suppose that all analysis of social phenomena should be parsed in terms of human action, it does follow, pretty much by definition, that the fundamental unit of society should be the only element of society capable of action, that is the individual. In my estimation, the Homo-agens argument is sound.
However, I do think that we should be suspicious of talk about fundamentals and concerns about parsing social phenomena in terms of human action alone. After all, being a complex system, society emerges out of the interaction of individuals. It is more than the sum of individual action. We can see this in basic institutional analysis. Yes, when analyzing institutions, we should always care about thinking about the costs and benefits that a particular agent weighs in deliberating his future actions, but we will generally have to treat of institutions as transcendent to that deliberation. It will always be the individual plus institutions, with those institutions existing at an emergent level above, so to speak, the level of a single individual's actions.
There are a lot of reasons to be skeptical of those who deny methodological individualism beyond concerns about foundations. Usually when people deny methodological individualism, they're trying to gloss over the differences between people. It's a lot easier to deal with disagreements in politics when one's conversation entirely neglects the level at which disagreement happens. In this fashion, methodological individualism serves as a instrumental prophylactic against error rather than a statement about the foundations of economics