Measuring and Decomposing Agricultural Productivity and Profitability Change
Speaker
Abstract
| Total factor productivity (TFP) is usually defined as the ratio of an aggregate output to an aggregate input. This definition implies that any TFP index can be written as the ratio of an output quantity index to an input quantity index. This paper demonstrates that, in theory, any such TFP index can be decomposed into unambiguous measures of technical change, technical efficiency change and scale efficiency change. Importantly, there is no allocative efficiency change component and no residual or unexplained component. The paper also shows that profitability change, which can be trivially expressed as a TFP index multiplied by an index measuring the change in the terms of trade, can be decomposed into unambiguous measures of technical change, technical efficiency change, scale efficiency change and allocative efficiency change. Again, there is no residual component. Practical implementation of these decompositions involves estimating the boundary of the production possibilities set, which leads naturally to the computation of Malmquist- and Konus-type indexes. The Malmquist TFP index that is popular in the empirical efficiency and productivity literature is incompatible with the definition of TFP as a ratio of quantity aggregates. Consequently, decomposition of that index leads to poor measures of the components of TFP change. In contrast, a version of the Malmquist index proposed by Bjurek (1994, 1996) is perfectly consistent with the conventional definition of TFP and can be neatly decomposed. To illustrate these and related ideas, the paper uses FAO data to make international comparisons of agricultural productivity change and its components. This empirical illustration is also used to show how transitive Malmquist and Konus indexes can be computed using a methodology developed in a consumer context by Geary (1958) and refined by Neary (2004). It is also used to illustrate how measures of reliability for different indexes and their components can be computed using Bayesian methodology |
| Contact information: |
| Prof.dr. B.M. Balk |
