False Contagion and False Covergence Clubs in Stochastic Growth Theory

Abstract

Three hypotheses exist for the bimodal, cross-country, income distributions emerging in the data since 1960 and dubbed convergence clubs. One is that economies in each mode are interacting with one another - true clubs. A second is that the bimodality is caused by lock-in or 'true contagion' - success at innovations breeds further success, failure breeds failure. These are false clubs because no interaction is assumed. A third (ours) generates false clubs but with no lock-in (false contagion), in a model with two stochastic processes, one from the consumer side, the other the usual Poisson process for innovations. Monte Carlo simulations show that bimodals can be caused by effects from either side. With non-conformist consumers for example, as R&D productivity is varied from increasing to decreasing returns, the bimodal distribution can move to an almost normal curve, with both few rich and few poor economies. All the derived Markov transition matrices mimic the dominant diagonals of the post-1960 data so that these are no proof of either clubs or lock-in.

Keywords: stochastic endogenous growth, false convergence

JEL Classification: O41, C15