Abstract

The compilation of Supply Utilization Accounts (SUAs) and Food Balance Sheets (FBSs) is one of the key activities of the Food and Agriculture Organization (FAO) to provide timely, coherent and internationally comparable statistics on food supply and utilization. Covering over 400 food items in 190 countries, these datasets form the basis for important indicators, such as dietary energy supply, import dependency ratios, food self-sufficiency, and resilience measures. SUAs follow a balancing identity whereby supply (production, imports and stock variation) is matched against utilization (exports, food, feed, seed, processing, industrial use, losses and residuals) for each calendar year. The balancing process is complex since it ensures both vertical consistency (between primary products and their derived items through extraction rates) and horizontal consistency (between total supply and all uses).
Traditionally, FAO relies on a balancing algorithm that compiles for each country and each year the entire SUA dataset taking into account the latest agri-food information from new collected data. While effective for maintaining coherence, this approach is resource-intensive and time-consuming, particularly when there is the need for revisions of past time-series for a limited set of items. A small revision in the figures for agricultural wheat production, for example, could require the recalculation of the entire dataset, since changes may cascade through commodity links (i.e., commodity trees) such as flour, pastry, or even products like alcohol produced from wheat or different multiple crops.
To improve SUA/FBS compilation efficiency, a new methodological tool named Single-tree plug-in has been developed. It consists of an algorithm which allows analysts to isolate a specific commodity tree and recompile its SUA across many years simultaneously, without recalculating the entire system. It applies the same balancing logic as the standard approach but restricts recalculations to the target item and its immediate derivatives at the first level of processing. This enables time-consistent revisions of a single commodity tree, significantly accelerating the task of revising historical data.
The single-tree plug-in is especially valuable when the scope of revisions is limited, such as when updated country data are available for a specific crop or processed product. In such cases, the country analysts can assess directly the impact of revisions on derived products and aggregated indicators. It also provides more analytical control, since the effects of a given change can be traced and evaluated more transparently than within a system-wide recalculation.
Nonetheless, the method introduces distinct challenges. The country analyst's definition of parameters and the tree level is essential, particularly in complex chains where an item has several processing connections. Analysts must evaluate whether a change propagates upwards, altering available raw materials, or downwards, modifying derived products. Moreover, the single-tree approach is not always preferable: if revisions affect multiple interconnected trees or propagate widely, the standard global balancing algorithm remains the more efficient and reliable option.
Rather than replacing the standard methodology, the single-tree plug-in complements it. Together, the two approaches provide a flexible toolkit: comprehensive recalculation for wide-ranging revisions, and a focused, efficient update for narrow revisions. The introduction of the single-tree algorithm strengthens FAO’s ability to manage time-series revisions, optimize resources, and increase responsiveness to new data submissions or corrections.
By enabling faster updates without sacrificing methodological soundness, the single-tree plug-in contributes to a more agile and transparent food balances compilation system. This advance supports the broader goal of delivering high-quality food statistics to guide evidence-based policy, monitor progress on food security, and respond to shocks in the global food system.