Endogenous productivity: a new Bayesian perspective

This study develops a methodology to address the endogeneity of productivity in the cost minimization framework where input demands and productivity itself depend on input prices and desirable and undesirable outputs. Specifically, we model toxic chemical releases (emissions) as an undesirable output in the production process. We apply our theoretical cost system approach to a panel data set of 2462 US manufacturing facilities over the period 1958–2007, which we estimate via Bayesian Markov Chain Monte Carlo semi-parametric methods subject to theoretical regularity conditions. The empirical findings reveal a non-linear inverted-U-shaped productivity curve concerning toxic emissions. This has important policy implications as the reduction in toxic emissions can be achieved without a decrease in productivity growth. The empirical findings are also consistent with productivity “divergence” across the U.S. manufacturing sectors and the formation of individual productivity clusters.