Econometric analysis of productivity with measurement error: Empirical application to the US Railroad industry

This paper analyzes the productivity in the US rail industry for the period 1980 - 2006. I propose a value-added production framework to circumvent the problem of measurement error in one input. I find evidence showing that aggregate productivity gains can be attributed to returns to scale and the reshuffling of resources to more efficient firms. However, productivity slows down for the period 1995 - 2000 after important concentrations. I also look at the correlations between firm-level productivity and the operating environment. My results show that failing to control for the omitted price variable bias leads to an overestimation of productivity gains.