This is a good summary of the Pigou vs. Coase debate on externalities. I have one comment: That the author should have brought Hayek into the picture. After all, the piece was published in 1996. What follows is a kind of executive summary.
Demsetz sees the debate on externality as one between two ideals: An ideal state (with perfect information) and an ideal market (also with perfect information and zero transaction cost). Taken to the limit, both models do not generally produce identical solutions. It is well known that the initial distribution of wealth and income affects market outcomes. Change that distribution and the economy rests somewhere else. With Pigovian state intervention, one also needs to factor in the initial distribution of wealth and income as a determinant of political process. Still, it is reasonable to imagine that both models arrive at the same end-point if they started with the same initial conditions.
Demsetz then concludes, based only on theoretical considerations, that the choice between the two models is one determined by preferences for freedom and the final (and/or initial) distribution of incomes and wealth.
Once we depart from the ideal to actual governments and markets, the choice between the two solutions would then have to take into account how much information there is (available) in the competing models, and how well they would reduce transaction cost. Here, Hayek would pronounce in favor of Coase, if only because Hayek believes that the market is more capable of ‘discovering’ such phenomena as efficient technologies and consumer preferences. Transaction cost can be seen as another form of externality, so we start to run the risk of arriving at a proverbial slippery slope.
Nonetheless, Demsetz is essentially right. Transaction cost is not at the kernel of Coase; and neither did Pigou ignore transaction cost. What was being debated was who should have the property rights to the externality, a question that economists usually avoid but one that Coase faced head on.