Abstract
This paper evaluates the methodological foundations of some recent attempts to estimate econometrically the degree of market power and the degree of returns to scale in manufacturing. The method discussed is based on estimating the aggregate production function in growth rate form. It is argued, following an argument made in another context by Phelps Brown, Shaikh, and Simon, that as the data used in empirical analyses are in value terms (i.e., monetary values at constant prices), the parameter derived as a mark-up can be reinterpreted simply as a coefficient from the income accounting identity, which takes a value of unity subject to omitted variable bias. Thus, it cannot be unambiguously interpreted as a mark-up. It is also shown that the large estimates of the degree of increasing returns to scale are similarly flawed. The argument also has implications for understanding cyclical fluctuations of the Solow residual, which turns out to be largely the result of the procyclical fluctuations of the profit rate. We conclude by questioning whether the aggregate production function can ever be statistically tested, or, in other words, whether it is capable of being refuted, as opposed to its parameters being merely estimated.