This article reviews the different methods of constructing multilateral output and productivity indices for agriculture in cross-country panel studies. We show that various multilateral output indices used by different researchers can have considerable disparities, thus rendering the comparison of the final results problematic. The production indices produced by the Food and Agricultural Organization of the United Nations (FAO) are increasingly used by researchers as a unique source of data for cross-country panel studies. The paper examines the properties of the FAO index, and finds them deficient in paying little attention to the problem of loss of characteristicity in a highly heterogeneous panel. It is shown that the FAO production indices lead to unacceptably large deviations from domestically based production indices in the case of low-income countries. It is argued further that the use of the FAO production index can lead to spurious results in econometric studies of the links between productivity growth, per capita income, and price levels.
Research Fellows
Massoud Karshenas
Emeritus Professor of Economics, Department of Economics,...