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...