In a recent paper Ganguli and Yang  demonstrate, that there can exist multiple equilibria in a financial market model á la Grossman and Stiglitz  if traders possess private information regarding the supply of the risky asset. The additional equilibria differ in some important respects fromthe usual equilibrium of the GrossmanStiglitz type which still exists in this model. This note shows that these additional equilibria are always unstable under learning. This is true for both eductive learning following Guesnerie  and adaptive learning via leastsquares estimation (cf. Marcet and Sargent  or Evans and Honkapohja ). Regarding the original GrossmanStiglitz type equilibrium, the stability results are less clear cut, since this equilibrium might be unstable under eductive learning while it is always stable under adaptive learning.
Recursive Least Squares Learning Eductive Stability Rational Expectations Private Information