Don Boudreaux looks at cases where prices should be above marginal cost.
Jones builds the bridge and charges tolls to pay for it. When the bridge is not congested, the marginal cost of allowing each driver access to the bridge is zero. Is the optimal toll zero? According to textbook theory: yes. According to the much-wiser Coase: no. If Jones were forced, by whatever means, to charge a price equal to his marginal cost of zero, clearly he would not recover his cost of building the bridge. Equally importantly, other investors would have no way of knowing if, and how much, additional investment is appropriate in building bridges to span the Mississippi.
If the future is going to see more of our GDP provided as bits rather than atoms, then we are going to have to develop a lot more tolerance for pricing above marginal cost. We also are going to have to develop a lot more tolerance for price discrimination--a point which was impressed on me by reading the classic Information Rules by Carl Shapiro and Hal Varian.
I do not think that there are easy answers in these cases. For example, with the bridge, you might have a toll that varies by time of day, to reflect congestion costs. You charge $1 at off-peak times, and $5 at peak times. You get 10,000 off-peak riders per day and 1000 peak-time riders per day, for $15,000 a day in revenue. Suppose that breakeven revenue is $6,000 a day.
Now, suppose that a competitor opens a bridge. Then my guess is that the toll will be competed to zero when there is no congestion, so that both bridge-owners become dependent on the congestion charge to recover fixed costs. At peak time, price competition is less fierce, because riders are willing to pay a little extra to be on a less congested bridge. However, there are only 1000 people willing to pay $5 a day for the privilege of a peak-time ride, so now neither bridge can recover its costs.
Another issue cited in Boudreaux's post is predatory pricing, and Armen Alchian's apparent doubt in the existence of such. I think that from a competitor's point of view, predatory pricing may exist, although from a consumer's point of view it is much rarer. That is, if you were Netscape, then you got driven out of the browser market. But even after the competition disappeared, the Microsoft browser was not priced above marginal cost.
For Discussion. In the bridge example that I laid out, what is the socially optimum number of bridges?