Decision-focused agricultural research
Agriculture provides most of our food and many other products. It also affects ecosystem services, such as waterregulation, soil protection, and biodiversity conservation. Decision-makers on agricultural systems, from farmers toagricultural ministers, should consider all these functions and their trade-offs, but this rarely happens. Many of agriculture’sproducts and services are regularly ignored in decision-making, mainly because they are difficult to appraise.This easily leads to decisions with adverse side effects, such as land degradation, pollution, or loss of cultural heritage.‘Holistic’ decision-making needs decision support approaches that consider factors that are difficult to quantify.Decision Analysis can potentially solve this problem. It recognizes that rational decisions do not normally requireprecise information on all factors of interest. Decision Analysis harnesses the knowledge of system experts to produce ahigh-level model of a decision, which reflects the best available information on plausible decision impacts. The modelshould include all factors experts consider relevant and all important decision impacts, regardless of data availability.Since most variables cannot be precisely quantified, experts estimate their state of uncertainty as confidence intervalsor probability distributions. These allow an initial model run, in which these inputs are translated into decision impactforecasts. These are imprecise, but they allow estimating a plausible range of decision outcomes. Often, this is sufficientfor selecting one of the decision alternatives. When no clear recommendation emerges, Value of Information analysiscan identify key uncertainties that decision-supporting research should address.Decision Analysis solves the problem of data gaps, which has often prevented research from comprehensivelyand holistically forecasting decision impacts. It also allows explicit consideration of risks and variability. We presentseveral applications of Decision Analysis in agricultural development, demonstrating its ability to convey a holisticunderstanding of likely decision impacts, in the face of risk and imperfect information.