Commenter J-D asks an interesting question of John Quiggin: when economists say that prices don't reflect the full opportunity costs, whose opportunity costs do they not reflect?
Society is not a good answer because societies do not choose. So who is being left out?
The opportunity cost of a choice is the best alternative not taken. It is inherently tied to the choice of the individual: the opportunity cost to me of someone else choosing not to give me a dollar is not one dollar. The opportunity cost to me is not even zero: I made no choice and thus bore no opportunity cost.
In welfare economics, prices reflecting all the opportunity costs means that all the net utility has been squeezed out of some constrained choice. Prices ideally encourage people to do as much of something as makes everyone better off and to encourage them to pursue alternatives once doing that thing can no longer make everyone better off. What this requires is that the price of every good and service be equal to the marginal cost of consuming it.
Frankly I think P = MC is a really confusing idea. P is in dollars, and MC is in utils. How are they be compared? Prices "reflect" opportunity cost. What does that mean? I know what it means at extrema, but if we are not at an extremum, as John posits, then the idea just confuses me.
We could speak about prices reflecting opportunity costs if we could translate every MC into a P via a market. That would require complete markets, which is certainly not the case if we are talking about market failure. So while I can make sense of optimality in these terms, I can't make sense of suboptimality. Efficiency makes sense, but inefficiency does not.
I suggested in the comments that normally in economics opportunity cost refers to the best foregone alternative in some choice, but in welfare economics opportunity cost is not tied to choice at all. Rather, when we speak of an opportunity cost to society, and so presumably when we speak of prices failing to reflect opportunity costs, we simply refer to some alternative imaginable arrangement of resources. There is an opportunity cost to society, not because society made a choice, thus forgoing its next-best alternative, but because things could be another way.
This runs contrary to the normal meaning of opportunity cost. Normally opportunity cost has no existence in being but only in doing. (Later I will argue that Pigou in fact intended no contradiction. His use of opportunity cost is the normal use of opportunity cost, or at least it is an attempt to be.)
Not just any imaginable superior alternative is supposed to be proof that prices don't reflect opportunity cost. Only a mere rearrangement of resources is. That is, imagining how better off we would be with cheaper steel is not proof of market failure. But being able to imagine that steel could be reallocated to the benefit of all is proof of market failure.
As I suggest above, market failure (prices not reflecting full opportunity costs) only makes sense in the context of complete markets. This isn't as bad as it sounds. We've learned a thing or two about the creation of markets since Pigou. In particular, we've learned that markets don't exist because making them exist is costly. In other words, there is a true and meaningful sense in which all markets exist...but most of them are too expensive to buy from.
So why don't prices reflect all opportunity costs? Because doing so is costly. Therefore, is it wrong to say that prices do reflect all opportunity costs, including the cost of reflecting opportunity costs?
What does this say about treating market failure as a misallocation of resources? When the supply of steel is less than infinite, there is no misallocation of resources even though everyone could be made better off if the supply of steel were infinite. When the supply of markets is less than infinite, that suggests the choices necessary to achieve the imagined superior alternative allocation of resources costs more than the imagined alternative would yield in benefits if realized. In fact, that is just what Coase demonstrated in his famous 1960 paper, "The Problem of Social Cost."
Moving from scarcity of steel to abundance of steel is a Pareto improvement, but infeasible, and thus efficient. Moving from scarcity of markets to abundance of markets is a Pareto improvement, but infeasible. Why is it inefficient?
More on this to come.