Of course, that’s not how we do things. We may live in a market sea, but that sea is dotted with many islands that we call firms, some of them quite large, within which decisions are made not via markets but via hierarchy — even, you might say, via central planning. Clearly, there are some things you don’t want to leave up to the market — the market itself is telling us that, by creating those islands of planning and hierarchy.
and here:
The thing is, however, that for a free-market true believer the recognition that some things are best not left up to markets should be a disturbing notion. If the limitations of markets in providing certain kinds of shared services are important enough to justify the creation of command-and-control entities with hundreds of thousands or even millions of workers, might there not even be some goods and services (*cough* health care *cough*) best provided by non-market means even at the level of the economy as a whole?In a lot of ways, the corporation is a huge challenge for theories of free markets. They are large organizations that have a great deal of political power and are famous for being able to survive sub-optimal decisions (see Dilbert, especially the Wally character). This use of political power can result in firms being successful due to favorable regulation (consider agricultural subsidies).
This is quite different than the nation of shopkeepers envisioned by Adam Smith. Without all business being set up in small units, the net result is that new entrants to the market can be strangled by existing firms (if nothing else, it is expensive to fend off spurious lawsuits). The Sears experiment that Krugman references is another good piece of evidence that there can be returns to cooperation as well as competition. Heck, a student of history who has read Caesar's account of his war against the Gauls should be highly suspicious that cooperation can be extremely important in the success or failure of groups.