I gave the third lecture of my Cours Bachelier at IHP yesterday, which followed the lectures that I described here and here.
I first reviewed the work of Frank Allen and Douglas Gale on the role that financial intermediation has in creating bubbles and crises. The idea is that when investors buy assets with borrowed money, the possibility of defaulting on the original loans makes the equilibrium price higher than it would be if they had to buy with their own money. Essentially, the loan plays the role of a "call option" whose convex payoff drives the asset price up. The key insight is that this type of bubble can be created by uncertainty in the amount of credit itself, rather than in the real economy. They show that in some cases a crises can occur even when credit is expanding, essentially because it didn't expand as much as necessary to keep fueling the overinflated asset price. In my view this is a neat toy model for the Minsky story about financial instability.
For the second part of the lecture, I focused bubbles caused by heterogeneous beliefs. The essential idea is simple: when short sales are possible, the views of optimists are balances by those of pessimists and asset prices reflect fundamental value, whereas if there are short sale constraints, pessimists sit on the side lines, prices reflect the views of optimists and a bubble ensues. This mechanism was described for discrete-time models by Harrison and Kreps in a paper on speculative behavior.
One possible source for heterogeneous beliefs is overconfidence, as proposed in a continuous-time model by Scheinkman and Xiong, which I reviewed towards the end of the lecture.
All references for the course can be found here.