World Food Day 2026: Where Food Loss Data Breaks Down in the Supply Chain

The global food loss figure was 13.0 per cent in 2015. In 2023 it was 13.3 per cent. FAO's own explanation for the rise is that national production data was updated and country estimates were revised, and that the movement sits within the expected fluctuation of the model.

Eleven years of monitoring, and the headline number moved for reasons that have nothing to do with food. FAO states the finding plainly: the world has made no apparent progress in reducing food losses since 2015.

That is not a criticism of FAO, which publishes its limitations in more detail than most agencies would. It is the starting point for anyone trying to build a loss figure for their own supply chain, because the boundaries that make the global number fragile are the same boundaries that will break your own.

The Day And The Theme

Today is World Food Day, marked on 16 October, and FAO's 2026 theme is "Innovate today. Nourish tomorrow," with activity across around 150 countries and campaign materials in up to 50 languages.

The loss picture underneath it varies enormously by region. On FAO's index, Northern America and Europe moved from 9.4 to 10.0 per cent between 2015 and 2023, while sub-Saharan Africa moved from 22.1 to 23.0 per cent. Least developed countries sit at 19.9 per cent and small island developing states at 19.0. Fruits and vegetables are the worst commodity group, rising from 23.2 to 25.4 per cent.

The Split At The Retail Door

Target 12.3 reads as one sentence and is measured as two things by two agencies.

Sub-indicator 12.3.1a is the Food Loss Index, with FAO as custodian. It covers quantities that completely exit the post-harvest, slaughter or catch supply chain without re-entering another use, starting on the production site with post-harvest operations and running up to, and excluding, the retail level. Storage, transport and processing are in, including for imported quantities.

Sub-indicator 12.3.1b is the Food Waste Index, with UNEP as custodian, and it picks up at retail and runs through consumption.

The target language is asymmetric in a way worth noticing. It commits to halving per capita food waste at retail and consumer level by 2030, and to reducing food losses along production and supply chains, with no number attached. The half of the problem that sits upstream of the retail door has no quantified target, two of the weakest data environments, and the larger share of the tonnage in most low and middle income countries.

What The Global Index Does Not Count

The UN metadata for 12.3.1a is unusually candid, and five exclusions matter for anyone borrowing the methodology.

Pre-harvest losses are out of scope. Harvest losses are excluded from the international index, and countries may include them at national level only, which means a national figure and the global figure are not the same measure.

Quality and economic losses are not measured at all. A consignment that arrives downgraded, sells at half price and is eaten in full registers as zero loss.

The index is a fixed-base index, current-year loss percentage divided by base-year percentage. It tracks change against 2015, not the level of loss in any absolute sense.

Country-level publication is not supported. The metadata states that the underlying data does not allow it, so results appear at regional level by commodity group, and each group covers only two commodities.

And the data is thin at the root. In 2016, 23 of 185 countries reported loss data for at least one commodity, and most loss factors in the supply and utilization accounts are estimated rather than officially reported.

Boundary By Boundary

Four boundaries do most of the damage in a company's own accounting.

The farm boundary. A crop left unharvested because the price fell below the cost of picking is not in the global index, and it is often the largest single quantity in a fresh produce chain. If your contract specifies grade and you reject at the field, the loss happens on land you do not own, to a crop you did not buy, and it appears in nobody's accounts.

The storage boundary. Weight loss and food loss are different things. Grain drying from 14 to 12 per cent moisture loses weight without losing food. Shrink, in the retail sense, bundles theft, damage, administrative error and spoilage into one number, and only part of it is food. A storage figure that has not separated moisture from spoilage from shrinkage is not a loss figure.

The transport boundary. Rejection at the receiving gate is the hardest case, because the event is recorded by the buyer and the consequence falls on the supplier. Whether that consignment becomes animal feed, a discounted secondary sale or a roadside dump determines whether it was a loss at all, and the buyer who rejected it usually never finds out. A rejection log with no destination field produces a tonnage that cannot be classified.

The retail boundary. Here the destination question decides everything. Marked down and sold is not loss. Donated and eaten is not loss. Sent to anaerobic digestion is loss with energy recovery. Three outcomes, one tonnage, and the choice of category changes the headline by more than most interventions do.

The FLW Standard And Its Scope Dimensions

The Food Loss and Waste Accounting and Reporting Standard, published in 2016 with the World Resources Institute as secretariat and a steering committee including FAO, UNEP, WRAP, WBCSD and the Consumer Goods Forum, resolves several of these problems by making you declare your choices rather than inherit them.

It requires an entity to define four scope dimensions. Timeframe, with start and end dates. Material type, which is food only, associated inedible parts only, or both. Destination. And boundary, itself split into food category, lifecycle stage, geography and organisation.

Two features deserve attention from anyone who thinks the global indicator is the ceiling.

The destination list has eleven categories, and one of them is "not harvested or plowed-in." The corporate standard has a home for exactly the quantity the SDG index excludes. If you want to count the crop you left in the field, the accounting vocabulary already exists.

The material type dimension is the one that quietly destroys comparability. Peel, bones, shells and trimmings are associated inedible parts. Two companies reporting tonnes of food loss, one counting them and one not, are reporting different quantities under the same label, and neither is wrong.

The standard also requires you to report the physical amount as weight, describe the quantification methods used, and provide a qualitative or quantitative assessment of uncertainty. That uncertainty requirement is the one most often skipped and the one that makes a number usable.

A Worksheet For Quantities And Causes

The column that makes this a worksheet rather than a table is the fourth one. Quantities without causes tell you the size of a problem and nothing about what to do.

Recording Causes Properly

Two levels of cause are worth separating, because confusing them produces action plans that fail.

The immediate cause is the physical mechanism. Crush damage. Mould. Temperature excursion. It is what an inspector sees.

The underlying cause is the decision or constraint that allowed it. Packaging specified for a shorter journey. A cold chain with no redundancy at the transfer point. A harvest window set by labour availability rather than maturity. An order pattern that pushes the supplier to pick early.

Logging only the immediate cause produces a register full of mould and a plan to buy better crates. Logging the underlying cause is harder, because it often implicates the buyer's own ordering, specification or pricing behaviour, which is precisely why it is worth the effort.

India: Two Studies, Two Boundaries

India illustrates the boundary problem cleanly, because its own national study measures something different from the global index.

The NABCONS study commissioned in 2022, reported through a government release in August 2025, covered 54 crops and commodities with a 2020 to 2022 reference period, and it splits losses into farm level and market level. Cereals showed 3.22 to 4.75 per cent at farm level and 0.56 to 1.17 per cent at market level. Fruits ran 3.77 to 11.59 per cent at farm and 1.64 to 3.46 per cent at market. Guava was the worst performer at both stages, at 11.59 and 3.46 per cent. Tomato reached 8.37 per cent at farm level.

Note what that structure does. It reports a farm-level figure, which the international index excludes, and it uses a farm and market split rather than the post-harvest to pre-retail boundary that 12.3.1a defines. Both are legitimate. Neither can be subtracted from or added to the other, and a company blending the two into one supply chain figure has produced a number with no definition behind it.

What To Do With This

Declare the boundary before you collect anything. Which stages, which commodities, which geography, which organisational units, and whether inedible parts are in or out. Write it at the top of the sheet, and write the date range next to it.

Record destination on every line, because destination is what separates loss from diversion and because it is the field most commonly missing from existing records.

Keep the uncertainty statement. A figure with a stated method and an honest error range is more useful to a buyer, an auditor and a regulator than a precise-looking number with neither.

And expect the total to rise when you start doing this properly. A loss figure that falls in its first year of serious measurement usually means the boundary moved, not the food.

General guidance only. SDG indicator definitions, custodianship and metadata are revised periodically and should be read in the current UN metadata rather than in summary. FAO's Food Loss Index is a modelled index tracking change against a 2015 base and is not published at country level; its figures are not comparable with national post-harvest loss studies that use different boundaries, reference periods or commodity baskets. The NABCONS ranges cited are drawn from a government release summarising that study and should be checked against the study itself before use; one quantity figure in that release appears to carry a unit error and has been excluded here. The FLW Standard requirements are summarised and should be applied from the published standard and its quantification guidance. Confirm reporting obligations and acceptable methods with your own advisers, your assurance provider and the relevant authority.

Sources

Food and Agriculture Organization of the United Nations, United Nations Statistics Division, FAO, United Nations Environment Programme, Food Loss and Waste Protocol, World Resources Institute, WRAP, Consumer Goods Forum, NABARD Consultancy Services, Ministry of Food Processing Industries, Press Information Bureau

This article is intended for general professional information and does not constitute legal, financial, or investment advice.