Size and family structure are different definitions
The 2021 farm research uses less than two hectares for small farms. Family farms are defined through organisation and family labour and can be much larger. The estimates combine census information from different years.
A production share in value terms is different from a share of tonnes, calories or agricultural land. This distinction should remain visible when describing why small and family farms matter.
Keep report editions together
SOFI 2026 reports around 645 million people experiencing hunger in 2025. Hunger means undernourishment; moderate or severe food insecurity is a broader experience-based measure. Their populations overlap and cannot be added.
Earlier estimates can change between editions. The first chart uses SOFI 2026. The other datasets are from SOFI 2025 and retain their separate edition labels; their revised estimates should not be combined.
INDEPENDENT RESEARCH · 2022–2024 · SOFI 2025
Global hunger eased, with recovery still incomplete
Global population; prevalence of undernourishment
| Measure | % of population |
|---|---|
| 2022 | 8.7 |
| 2023 | 8.5 |
| 2024 | 8.2 |
World / indicated regions. Same report edition. Show the 2022–2024 range without extrapolation or blending earlier editions' revised values.
SOFI 2025 joint-agency findingsFood loss and food waste are separate
FAO’s food-loss measure covers the post-harvest chain before retail and uses a weighted methodology. UNEP’s food-waste measure covers retail, food service and households, including inedible parts. The years and denominators differ.
Do not add the percentages or display them as slices of one total. Commodity-specific project measurement needs its own defined quantity, stage, quality and baseline.
INDEPENDENT RESEARCH · 2015 and 2023 · FAO
Losses before retail: regional infrastructure matters
Food supply chains after harvest and before retail
| Measure | % food loss, FAO weighted measure |
|---|---|
| World · 2015 | 13 |
| World · 2023 | 13.3 |
| Sub-Saharan Africa · 2015 | 22.1 |
| Sub-Saharan Africa · 2023 | 23 |
| Northern America and Europe · 2015 | 9.4 |
| Northern America and Europe · 2023 | 10 |
| Eastern and Southeastern Asia · 2015 | 13.8 |
| Eastern and Southeastern Asia · 2023 | 13.9 |
World / indicated regions. Global row overlaps regional rows; compare bars or paired observations, not pie slices. Small changes do not demonstrate a statistically meaningful deterioration.
12.3.1 Global food losses | FAO SDG Indicators Data PortalThe agrifood system is broader than farming
FAO’s 2023 emissions boundary includes farm-gate activity, land-use change, and pre- and post-production processes. The rounded components total 16.5 Gt CO₂e.
Sector statistics describe a global system. A borrower’s footprint needs its own activity data, method and reporting boundary. A financing programme cannot claim the sector’s global emissions as its own reducible impact.
INDEPENDENT RESEARCH · 2023 · FAO release 2025
Where agrifood emissions arise
Global agrifood system emissions: farm gate, land-use change and pre/post-production.
| Measure | Gt CO₂e |
|---|---|
| Farm gate | 8.1 |
| Pre/post-production | 5.2 |
| Land-use change | 3.2 |
World. Rounded components total 16.5 Gt CO₂e. Agrifood systems are broader than farming alone. This chart is sector context, not a WAB footprint.
FAO · Agrifood emissions 2001–2023From external evidence to an operating baseline
External research helps identify priorities. Local evidence then establishes the customer, commodity, infrastructure and service need. Operating data can eventually show activity and performance; outcome measurement requires a credible baseline and comparison.
The site’s charts are fixed, dated research snapshots reviewed in October 2026. They are linked to the original publications and are not live market feeds.
READ THE SOURCE