Aggregation is the price paid for modeling convenience. There is always the question about the nature of the bargain. Many economists worry that macro-economics assumes away most of economics -- including choice and discovery. But Paul Krugman cites evidence that Keynesians have been right (mostly); Tyler Cowen says not so fast.
There is little debate that no one predicted the depth of the recent recession. Krugman and others are scrambling to claim that they did get the slow recovery right. There is the problem of not just extreme aggregation but also the fact that the models pay no attention to public choice considerations. I doubt they can have much usefulness.
The Journal of Economic Perspectives (Fall 2014) includes a symposium on Social Networks. How can anyone resist Vasco Carvalho's contribution with this title: "From Micro to Macro via Production Networks?" The analysis involves taking a national input-output model and looking for supply chains among the 417 BEA sectors.
But (1) Supply chains are everywhere and formed at much greater levels of sectoral disaggregation than 417; and (2) Supply chains tend to be situated locally and much more interesting than what can be found from studying a national IO model. Industrial organization discussions ask how much of any supply chain is within the firm and how much is outsourced. The associated spatial organization question is how much of the supply chain that is outside the firm is local. Cities are agglomerations of supply chains (parts or whole) -- within and beyond local firms. Getting this part right would be a better micro-to-macro.