Crop yield or the quantity of crop output harvested (say in tonnes) per unit of area cultivated (say in hectares) is the most important measure of agricultural performance worldwide. Crop yields are central to discussions of growth of output, technological change and the dissemination of scientific and technical information, the impact of climate change, the relative cultivation performance of small and large farmers, food security, and many other contemporary debates. The analysis and measurement of crop yields are central to the work of a wide range of people, including agronomists, economists, and policy makers. An accurate measure of crop yield is critical to all those concerned with food and agriculture.

The definition of crop yields, however, is hardly debated in the literature. I refer here to conceptual problems with the definition of yield, not of methods of estimation (such as crop-cutting experiments or farmer recall). I shall raise two concerns about the definition of yields.

First, measuring output or crop harvested per unit of area is very difficult in the case of intercropping and mixed cropping. Intercropping refers to a situation where more than one crop is grown on a particular field (such as alternate rows of cotton and pigeon pea in central India). Mixed cropping refers to a field where two or more crops are intermingled (such as when seeds of two types of pulses are mixed and then sown). In recent years, intercropping has been recommended on ecological grounds (soil health) and to promote nutritional diversity of the crop-mix, among other things.

Official statistics in India do not report yields for intercropping. In the surveys conducted as part of the Project on Agrarian Relations in India (PARI), we do not report yields separately for crops on an intercropped field (as the area is combined), but separate costs of production based on the ratio of the gross value of output of the two or more crops grown on the same plot.

A second problem, raised in a recent paper by Reymonds, Anderson, Slakie and Gugerty (2015), is that of a possible difference between area cultivated or sown and area harvested. Such a difference can arise for many reasons: the irregular shape of plots, non-planted areas with trees or bushes or other obstructions, partial crop loss or poor germination, and so on. Such differences will, of course, matter more for small cultivators than for large cultivators. For example, on a one-acre plot in West Bengal, as a colleague pointed out, field channels for irrigation can take up as much as 0.01 acre or one-tenth of the area sown. If the yield on such a plot is measured as production per acre, the figure for the yield will obviously be lower than if yield were measured as being equal to production per 0.9 acres!

In their study of smallholder rice growers in Tanzania, Reymonds et al. show that the difference between yields defined on area planted and area harvested is statistically significant. Harvested area was less than planted area in 23 per cent of sample plots. The divergence was on account of drought, rain, insects, crop theft, and disease. The difference was larger for female-headed households than male-headed households.

Reynolds, T. W., Anderson,C. L., Slakie, E., and Gugerty, M. K. (2015), “How Common Crop Yield Measures Misrepresent Productivity among Smallholder Farmers,” paper presented at the International Conference of Agricultural Economists, Milan, July 29, available here.