Convergence, endogenous growth, and productivity disturbances

Kelly (1992) has recently shown that evidence on convergence cannot be taken as evidence against endogenous growth in general. This study uses a well-known class of stochastic growth models to show other difficulties with traditional empirical studies of convergence. Key parameters typically cannot be estimated consistently in cross-section regressions. When the parameters are assumed known, implications for convergence arc unavailable except under restrictive and economically unmotivated assumptions. Those same assumptions that relate key parameters to cross-country convergence render cross-section regressions impossible to estimate consistently.