Modeling Technology as a Dynamic Error Components Process: The Case of the Inter-country Agricultural Production Function

In this paper, we propose a dynamic error-components model to represent the unobserved level of technology. This specification implies a well-defined common factor dynamic model for per capita output that can be tested explicitly. The model is applied to data on aggregates of agricultural inputs and outputs for groups of countries from the OECD, Africa (AF), Latin America (LA) as well as centrally planned countries, over a period of 31 years. We find that the proposed model fits the data better than alternative static specifications and satisfies the implied common factor restrictions in two of the samples. The results suggest that although technological change seems to have been a faster process for less developed countries relative to the OECD countries, it has not been fast enough to reduce appreciably the enormous differences in average technological levels that still persist between them.