Here is another interesting post from Econtalk, where Russell Roberts conductes a very interviewing Greg Mankiw.
Mankiw is a very interesting and bright guy. However, I did have one serious point to make with Russell. When Greg made the claim about the Coase theorem only applying when transaction costs are zero with the note that when transaction costs are positive it matters who we give the property rights to. That is not quite right. The Coase theorem implies that if the transaction costs are less than the gains from trade it doesn't matter who we assign property rights to. Similarly, I wish that Russell had raised the point that the market will solve externality problems, not just when transaction costs are zero, but when they are less than the gains from trade.