In October 2024, Cambodia launched the Cambodia Agricultural Survey (CAS) for the sixth consecutive year. That might not sound like headline news, but it is. Cambodia became the first country in the 50x2030 Initiative to reach this milestone, and it signals something important about how seriously the country is taking evidence-based agricultural policy.
I joined this programme as the FAO statistician working alongside colleagues at the National Institute of Statistics (NIS) and the Ministry of Agriculture, Forestry and Fisheries (MAFF), and have been part of its growth since. The 50x2030 Initiative is a global partnership aiming to help 50 countries produce regular, reliable agricultural data by 2030. Cambodia's consistency stands out. Running a nationally representative agricultural survey every year, with integrated data collection across crops, livestock, fisheries, and household economics, requires serious institutional commitment. It also requires the kind of mundane, unglamorous work that rarely gets attention: harmonising questionnaires, training enumerators in remote provinces, cleaning messy datasets, and building systems that don't fall apart when key staff rotate.
The link between good agricultural data and food security is direct but often underestimated. You can't target nutrition programmes, allocate irrigation investments, or plan climate adaptation without knowing what farmers are actually growing, earning, and struggling with. Cambodia's agriculture sector still employs a significant share of the labour force, and the decisions made about it affect millions of households. Those decisions should be grounded in evidence, not assumptions.
CAS 2025 was fully funded by the Cambodian government's national budget — a major milestone. When a country commits its own resources to a statistical programme that was initially donor-supported, that is real national ownership. The data now feeds directly into Cambodia's Pentagonal Strategy national indicators and broader national planning. But the most important thing remains simple: data collected must be data used. Otherwise all the effort — the enumerator training, the fieldwork logistics, the quality assurance — is in vain. Now that there is a steady workflow for data production, the focus shifts to advanced statistical analysis utilising AI and machine learning. That is where the efforts lie now, and it is where I believe the next real gains will come from.