Conjectures, learning, and equilibria in monopolistic competition (Q1117819)

The paper investigates the disequilibrium dynamics in a monopolistic competition market, in which firms cannot calculate ex-ante the Cournot- Nash equilibrium: this is due to lack of complete information or consistent beliefs at the initial condition. Starting with individual conjectures about the state of the market, producers make their maximising choices and if the market outcome falsifies the current conjectures, these are revised taking past observations into account. A situation in which all firms' beliefs are not falsified is a conjectural equilibrium, and it is shown that there is a continuum of such equilibria. An open subset of this continuum is formed by locally stable points with respect to the dynamic learning process envisaged, while the remaining equilibria are unstable (bifurcation analysis is not undertaken); stability depends on the number of firms, on their degree of memory, and on the parameters of the conjectures in an equilibrium point. It is shown that optimality criteria cannot be employed to reduce the range of feasible equilibria. The concept of stable (robust) conjectural equilibrium appears to bypass some common critiques - namely that conjectural equilibria stem from ad hoc assumptions on firms' beliefs.