Thread by @MacRoweNick

Lucas '72, which is (sorta) where microfoundations all began, is a heterogeneous agent model.
It has: old agents (consumers); young agents (producers) on island 1; young agents (producers) on island 2; etc.
It doesn't work with only 1 type of agent (or 1 island). 1/n

Larry Ball's "yeoman farmer" interpretation of the NK model is also, in a sense, a heterogenous agent model. Each agent has a comparative advantage and specialises in producing a different type of fruit, which they trade.
It doesn't work if they all produce the same fruit. 2/n

Or, you can reinterpret the NK model as having literally identical households but heterogeneous firms, each specialising in producing a different type of fruit.
Again, it doesn't work if they all produce the same fruit. 3/n

None of these models is like an archipelago with 100 identical Robinson Crusoes who do not trade with each other, where you literally can say that the whole is exactly the same as the sum of the parts. 4/n

All of these models are models of coordination failures, where what is true and optimal for each agent is not always true and optimal for all agents. That's the whole point of these models. These coordination failure are not a Robinson Crusoe problem. 5/n

These models are "representative agent" models only in the same sense that symmetric Prisoners' Dilemma is a "representative agent" model. It's symmetric, but it doesn't work with only one Robinson Crusoe prisoner. 6/n

(When I said that Lucas '72 is (*sorta*) where microfoundations all began, that "sorta" is important. Because the search for microfounded explanations goes back way before Lucas. But it was usually piecemeal, or informal.
We want to explain *why* people do what they do.) 7/7