Testing for non-stationary technology and constant returns to scale in the context of the inter-country agricultural production function

Abstract

This paper uses a panel data production function .framework with stochastic technology trends to re-examine the inter-country agricultural production function. The focus is on the dynamies of technology and the type of (aggregate) returns to scale. The study is implemented both at the world level and by groups of countries using a panel data set on quality-adjusted agricultural inputs and outputs for 110 countries over a period of 31 years. Instead of usual deterministic time trends, a dynamic error-components model is used to model technology. This specification results in a well-defined common factor dynamic process for agricultural labor productivity. The hypothesis of non-stationary technological levels and constant returns to scale is tested by means of a LR test carried out on the Within residuals. The results support stationary technology processes and decreasing (aggregate) returns to scale. The accuracy of the error-components specification for technology is, finally, evaluated by testing its implied common factor restrictions. Some evidence in favor of this specification is found.