Modern agricultural development largely focuses on raising the productivity of small scale
farmers. Organization mission statements and budgets reflect this priority: the Food and
Agriculture Organization of the United Nations’ (FAO) mandate is to “raise levels of nutrition,
improve agricultural productivity, better the lives of rural populations and contribute to the
growth of the world economy” (emphasis added), while the United States Agency for
International Development (USAID) identifies increased productivity as a key to “inclusive
agriculture-led growth” (USAID, 2013). And the Bill & Melinda Gates Foundation Agricultural
Development Program states its core goal as “to reduce hunger and poverty for millions of
farming families in Sub-Saharan Africa and South Asia by increasing agricultural productivity
in a sustainable way” (BMGF, 2014; emphasis added). Over the past several decades
governments, non-profits and others have invested billions to raise crop yields in the pursuit of
productivity related goals (Conway, 2012).
But despite good intentions, common yield-based crop output measures – used ubiquitously
to identify successful interventions and target needy households – are a poor proxy for
smallholder farm productivity. Various measures including land productivity, labor productivity,
and total factor productivity are used to measure agricultural progress (Fuglie, 2008; Alston et
al., 2010). But crop yield – defined as a simple ratio of quantity harvested divided by area
harvested (e.g., kg/ha) – is most often used as the primary productivity indicator (Fermont &
Benson, 2011). The merits of this indicator (henceforth “common crop yield”) include its relative
ease of calculation and intuitive interpretation, acceptance among agronomists and agricultural
policymakers, and the relatively widespread availability of time-series data on crop production
and harvested area, allowing monitoring and comparisons of yield estimates over time (FAO,
2014). However this common yield measure is also known to be imperfect – both in the
reliability of the data used to generate yield (measurement error), and the appropriateness of the
measure itself as an indicator of agricultural productivity (measure validity).