Endogeneity bias and other problems in the estimation of production functions

The Institute of Economics will hold a webinar meeting as part of its Seminar Series on Tuesday, April 27, 2021: Jesus Felipe from the Asian Development Bank will present the paper "Endogeneity bias and other problems in the estimation of production functions".

Abstract:

This paper rescues a very much neglected argument about production functions that was first outlined in the literature in the 1950s. It has implications for understanding production function estimations and the potential problem of regressors’ endogeneity. This argument is that the series of output, labor and capital used to estimate production functions are linked through a distribution accounting identity. The reason is the data used in empirical exercises are not physical quantities (for output and capital) but value data. This identity can be rewritten as a form that resembles exactly as a production function (Cobb-Douglas, CES, translog). The authors estimate a typical Cobb-Douglas function with five estimators and show that the identity is responsible for the good statistical fits obtained and for the fact that the elasticities are often close to the factor shares. The authors also show that the possibility of regressors’ endogeneity bias that has been discussed for decades is, in reality, caused by the omission of a term in the identity. It is concluded that the use of value rather than physical data poses an insoluble problem for the estimation of production functions. This is, consequently, far more serious than any supposed endogeneity problems.

All interested participants are welcome to join online at the following link. External participants need to contact the organisers via email to grant access to the seminar.