Global Academy of Agriculture and Food Systems

Caution urged over updates to key food dataset 

Changes to methods used to calculate national consumption of key commodities gives rise to discrepancies. 

Updates to a widely used international database of food consumption can lead to inconsistencies when applied to research, scientists have found. 

Researchers should exercise caution when interpreting calculations based on food balance sheets published by the Food and Agriculture Organization of the United Nations (FAO), which monitor long-term trends in national food supply, the research suggests. 

An update to the methodology used to determine the year-by-year data has led to discrepancies between values determined using the previous and current methods. 

Researchers and policymakers should be aware of the potential for irregular outcomes in applications of the dataset, scientists suggest. 

Dataset updates 

Scientists examined the FAO’S large-scale, standardised food and agriculture dataset for over 245 countries and territories, known as food balance sheets (FBS), which contain annual data on food production and consumption.  

The datasets, dating from the 1960s, are vital to monitoring long-term trends in national food supply and tracking trends in hundreds of edible commodities - such as meats, fruits and vegetables - and are regularly used to inform policy and for research. 

Recent changes to the methodology used to generate the data has created a discrepancy between two periods of time, the team found.  

One methodology is applied to years 1961-2013 and a second method for years 2010–2020, without a common mechanism to conduct trend analyses across the two. 

Comparing values for the overlapping four-year period of 2010–2013 highlighted substantial differences between equivalent values, demonstrating the mismatch between the methods. 

For example, milk production for Belgium in 2010 is recorded as 232kg per person annually in the old dataset, compared with 59kg using the newer dataset. 

Researchers suggest that organisations using the data should be aware of the potential mismatch and take steps to standardise their calculations where possible. Users of the data may have to develop bespoke methods to account for the discrepancies in the figures. 

The study was published in Lancet Planetary Health. 

This international dataset is widely used to inform policy advice and decisions. Our research shows that analysis combining the multiple Food Balance Sheet datasets may produce incorrect results. We hope user awareness of potential discrepancies can help limit their likelihood.

Alexander VonderschmidtPostgraduate Researcher

Related links 

Scientific publication 

Image credit: Milada Vigerova / Unsplash