What’s the State of the Evidence for Digital Agriculture?

This post is written by Jonathan Mockshell and Brian King.

Digital innovations are cropping up across agricultural sectors, from extension services to trade, marketing to consumption. Precision agriculture technologies have helped industrial farms optimize resource use for decades, and today even on small farms there’s an app that enables smallholders to book tractors to reduce the financial burden of equipment, and interactive voice response systems that send agro-climatic information to help farmers increase crop yields and income. The growing number and diffusion of these digital innovations stand to help us build more resilient, adaptive, sustainable global food systems, improve welfare, and promote economic development.

There is, however, a perennial evidence gap in digital agriculture. This is to be expected with any class of emerging technology, but it is important to monitor how--and how much--these innovations are transforming our world if we wish to use them to their full potential. 

This is not to imply there is *no* evidence. Indeed, a literature review by our team found several cases that meet the “gold standard” of evidence. We found several randomized control trials (RCTs), quasi-experimental designs, machine learning techniques, and narratives of change evaluation that have demonstrated solid evidence in support of digital interventions:

• A study on a picture-based insurance project indicates that streams of pictures of individual smallholder fields, taken using inexpensive smartphones, can support crop modeling, extension and insurance schemes (Hufkens et al, 2019). In turn, these outcomes can increase resilience to production risk and enhance food security in smallholder agricultural systems.

• A second study developed a picture-based smartphone app called Time-Tracker that allows data recording in real time to avoid recall biases. The study compared data recorded with the Time-Tracker app to data collected with 24-hours recall questions. The evidence highlights recall biases, suggesting that using the app leads to valid results. 

• Another study on the impact of mobile phone technology on agricultural extension services in India indicates that digital technology has improved the quality and speed of service delivery. Farmers in the study demonstrated greater knowledge and awareness of new agricultural practices, farmers’ aspiration to try new technology in the future, and access to credit.

These are great studies, but we also need to be sure we are not overly selective about evidence amid the inherent complexity of food systems. For example, another study using RCT highlights the need to use a nuanced lens to examine each intervention. The study shows that while households in Nigerian villages with increased access to mobile phone technology planted a more diverse basket of crops (especially marginal cash crops grown by women), this did not increase the likelihood of selling these crops or the farm-gate price received (Aker & Ksoll, 2015). Results like these suggest that cohesive interventions that address challenges at various points along the food system continuum are needed to improve farmers’ welfare, and that digital technologies can only be one piece of the puzzle.

At the CGIAR Platform for Big Data in Agriculture, we are of the view that systematic evidence collection and review needs to happen across the array of food system entry points. The food system framework is important for being able to examine interventions in light of consumption and nutrition, but also provides a way to begin to better define digital intervention types and build more commonality among study designs. With more standard data, definitions and approaches, we will be able to begin to measure and understand the full complexity that must be managed by farmers, industry, funding partners, and policymakers seeking to drive transformative change.  

In September 2019 we launched the Digital Food Systems Evidence Clearing House to begin to fill this gap. Featured interventions, which are reviewed by the BIG DATA Platform’s community of experts, must provide a description of their services, the estimated number of active users, and credible evidence of impact broken down into economic, environmental, social, technical components, as well as the overall impact on food systems efficiency.  

As we call for more evidence and populate the database, we will allow users to see global trends in the digital innovation and, once sufficient impact evidence is collected, the Clearing House will be used to periodically generate synthesis reports and meta-analyses of the ‘state of the evidence for digital food systems’.

As researchers, development experts, start-ups, policymakers and funders it’s in our hands to improve the state of evidence of digital intervention in the food system to guide program designs and scale out mature innovations. 

Information on how to submit your intervention and evidence to the Digital Food Systems Evidence Clearing House can be found here.