Nairobi residents buy fruits and vegetables at neighborhood market stalls called mama mboga in Swahili. Credit: Global Alliance for Improved Nutrition (GAIN).
A healthy diet should be affordable, but cost is not the only factor that matters. Case in point: Kenya’s per-capita income tripled between 2000 and 2015, and education campaigns emphasized the health benefits of eating fresh produce. Actual fruit and vegetable consumption levels, however, remained virtually stagnant—and well below the World Health Organization’s recommendations—during this time.
To change that trend, the Global Alliance for Improved Nutrition (GAIN) asked associate professor Charles Nicholson to identify strategies for boosting vegetable consumption that go beyond consumer education. To help prioritize future funding decisions, the NGO also wanted to learn how well these strategies would likely work in practice.
Nicholson turned to a tried-and-true process called group model building. Developed 30 years ago by business strategists, it assembles stakeholders from different backgrounds who jointly develop strategies for reaching a shared goal. Nicholson had previously used the group model building process to help establish and better operate rural dairy cooperatives in Mexico and Ethiopia.
In this case, Nicholson and Eva Monterrosa, his GAIN collaborator, invited 16 stakeholders to a September 2019 workshop in Nairobi, Kenya: representatives of the Ministry of Agriculture, consulting companies and NGOs in the agriculture and nutrition sectors; researchers with different areas of expertise; and a farmer and private vegetable distribution company.
With Nicholson’s guidance, the participants developed a food systems model of the main supply chain agents: farmers who grow the vegetables; wholesale intermediaries who transport, distribute and sell them to small-scale retail vendors; and consumers who buy them from the vendors. Using leafy greens as an example, Nicholson placed these interconnected agents into a complex network diagram.
“Developing the model was a learning process because the diagram was a new way of looking at a familiar problem,” says Nicholson. “The participants were surprised about the many connections and feedback loops they had not previously considered, which helped them appreciate that the supply chain was not as simple as they thought it was.”
At a follow-up workshop six months later, the diagram helped the stakeholders agree on three potential strategies for raising vegetable consumption levels: increasing consumer awareness of health benefits, reducing perishability and increasing farm yields.
Nicholson’s next task was to quantify the likely impact of these three interventions. He used software to simulate how each strategy would affect urban Kenya’s vegetable supply chain over the course of five years, based on the diagram and empirical data about the flow of products within the chain.
To illustrate, consider farmers, intermediaries and vendors as three agents who each manage two types of product stocks: orders placed from their respective supplier (or plants growing in the field for farmers) and existing inventory. An agent’s profit expectation determines the size of placed orders. The time between placing and receiving orders depends on the growing period for leafy green vegetables and the delivery times from farmer to intermediary and intermediary to vendor. Previous studies in Kenya provided relevant parameter estimates: the typical size of orders and inventory and the respective delivery times.
The vegetables are either sold to consumers by vendors or lost to spoilage while traveling through the supply chain. Consumer prices depend on the relationship between sales and losses: When the inventory falls faster than expected, prices go up; higher-than-usual inventories result in lower prices.
Consumer purchasing decisions depend on prices and the perceived health benefits of vegetables. Each agent’s profit is the difference between the cost of managing their inventory and the income from selling it. Relevant parameter estimates included typical daily loss rates, management costs and sales prices.
In reality, key parameter values are not known precisely, and the simulation software helps account for this uncertainty. Each simulation round randomly sampled combinations of parameters from their range of possible values, as specified by the stakeholders. Averaging the outcome of interest—daily mean per-capita vegetable consumption from 2019 to 2024—across many simulation rounds allowed Nicholson to compare the likely impact of the three interventions.
He found limited benefits of the first strategy—increasing consumer awareness—because knowledge levels were already high. A cultural factor that helps explain stagnant vegetable consumption levels, says Nicholson, is that Kenyans are more likely to spend extra income on foods with higher prestige values than vegetables, such as animal meat.
The second intervention—reducing perishability—reduced rather than increased consumption levels. This was due to the costs of storage improvements exceeding the savings from avoided losses. Supply chain agents can only recover those costs by increasing sales prices, which would decrease consumption.
“Another explanation for this seemingly counterintuitive finding is the erroneous assumption that farmers would still produce the same amount of food when spoilage is reduced,” says Nicholson. “That’s generally not the case because the reduced perishability is accounted for in placing orders upstream in the supply chain.”
Vendors, he explains, would order less product from intermediaries, say an extra 5% instead of 10% above expected sales to account for spoilage. Intermediaries would make similar adjustments, eventually causing the farmers at the top of the supply chain to produce fewer vegetables.
The largest increase in consumption—from an average 131 g/person/day in 2019 to 171 g/person/day in 2024—came from increasing farm yields, which would decrease consumer prices. However, this would also reduce farm profits unless government subsidies cover the cost of fertilizers or yield-boosting agricultural technology.
Additional funding allowed Nicholson and colleagues to explore alternative strategies. Consumer choice experiments helped them estimate people’s willingness to pay for higher-quality vegetables and faster or more convenient buying options.
“These are examples of what economists call a demand curve shift, which would not reduce the profits of supply chain agents by lowering prices,” says Nicholson. “We think this approach may be a more feasible alternative for increasing vegetable consumption levels.”