On Herbert Simon's criticisms of the Cobb-Douglas and the CES production functions

Scott Carter (2011) reproduces and discusses correspondence between Simon and Solow in 1971, where Simon first outlined his critique that estimations of production functions merely capture an underlying accounting identity. This idea culminated in a paper published by Simon in 1979 in the Scandinavian Journal of Economics. We extend Simon's argument that production functions should ideally be estimated using physical data, and discuss the serious problems that arise when they are estimated using constant-price monetary data. Simon also suggested that the good statistical fits to production functions could be derived from a markup pricing model, but he did not follow this up. We show that this can indeed account for the very good statistical fits of the Cobb-Douglas and other production functions. We conclude by showing how estimates of cost functions suffer from the same problem.