Essays on plant -level productivity, market structure, and enterprise growth

This dissertation explores methodological issues in productivity measurement and the dynamics of market structure, competition and plant-level productivity growth in Colombia's manufacturing sector. Recent econometric developments have proposed significant improvements over traditionally used methods, offering simultaneity bias and selection bias corrections in estimating total factor productivity. However, the magnitude of this bias on production function estimates and its resulting effect on productivity calculations is unknown. We compare productivity estimates from six widely used techniques using both a gross output production function and the classical value added functional form. The dataset comprises a panel of Colombian manufacturing plants observed from 1993 to 2002. We find evidence of large disparities across methods in both productivity levels and growth estimates: correlations are low and there are large variations in the estimates of productivity dispersion across methods. The methods offer a contrasting view on the sources of productivity heterogeneity. Parametric estimates suggest that aggregate industry productivity is mainly the result of reallocation of resources to more efficient producers, while the semi-parametric methods attribute productivity differentials to within plant changes. Functional form specification, type of industry, and time horizon matters. The productivity estimates are used to examine the relationship between competition and productivity growth. While existing studies estimate measures of central tendency, we model plants' conditional productivity distribution using regression quantiles. The effect of competition on productivity growth varies widely based on plants' location in the conditional productivity growth distribution. We show that, after controlling for unobserved plant and industry heterogeneity, at upper ends of the productivity growth distribution a 'Schumpeterian' effect of higher competition lowering growth dominates. For firms at the lower tails an 'escape competition' effect of productivity improvements prevails. We also show that, within an industry the relationship between competition and productivity growth is correlated with existing levels of competition. The diversity in the competition-productivity relationship at the industry level suggests that industry orientation is fundamental in defining competition policy targets. While removing entry barriers and tightening antitrust/collusion policy could increase aggregate growth in some industries, higher competition need not necessarily favor growth of all firms in other industries.