Aggregation in economics presents problems. In fact, a case can be made that aggregation contradicts and even undermines our understanding of markets. See Christopher Coyne's "Economics as the study of coordination and exchange" in Boettke's Handbook on Contemporary Austrian Economics.
We study markets because decentralized exchange is the only way we know to solve the immensely difficult and complex problem of getting uncountable scarce resources to where they are most valued. Coyne argues we have to focus on exchange rather than pose mechanical "allocation" problems which can be solved via aggregation and modeling.
Think about the idea of "capital". Das Kapital denotes a nefarious thing that is accumulated solely for the sake of profiteers -- and at the expense of everyone else. But neo-classical economists also aggregate to get the "capital" that we see in their models.
But these approaches ignore and obscure the interesting problem. Society provisions itself for a better future if and only if some of today's consumption is deferred in the service of worthy projects that will pay off (be rewarding and rewarded) at some future date. These simple words do not do evoke the scope and the difficulty of the task. Tremendous risk-raking and coordination are involved.
Readers of this blog know that I view cities the same way. The managers of firms are part of various supply chains. In each case, they must decide how much of the supply chain is within vs beyond the firm. Many of them locate their firm in cities and make the concurrent decision of how much of the supply chain (within as well as beyond the firm) takes place nearby or not (in the vicinity or not, in the same city or not, etc.). Attending to and forming any supply chain is also a coordination problem. In fact, I often note that there are supply chains for ideas as well as things. Expanding useful knowledge is a key part of the problem. This complicates matters even further.
We have to think of coordination problems that involve fine-grained spatial as well as temporal dimensions. Markets (always "imperfect" as in straw man) are the only game in town.
But this is not the way most analysts approach the problem. Google scholar comes up with only 14,500 hits for "coordination problem" entered along with "markets". "Keynesian" gets 169,000. "Cities" along with "coordination problem" gets 6,160. "Marx" and "cities" gets 212,000.
These are crude measures, but cities' role in the coordination problem is, in my view, much more interesting than the cities' role in whatever Marxists write about these days.