The scale of Kenya’s varietal stagnation is striking. On-farm, the area-weighted average age of maize hybrids stands at 18 years; at the point of sale, it is 20 years. Some hybrids have remained in cultivation for 15, 20, even 30 years, even though roughly 16 new hybrids are licensed annually through Kenya’s National Performance Trials and about 25 hybrids released within the past decade are available through agro-dealers. Newer varieties, bred for shifting rainfall patterns, more frequent drought, and emerging pests and diseases, have shown higher yields both in trials and in farmers’ own fields, with recent evidence indicating newer hybrids out-yield older ones by up to 147 kilograms per hectare in the long rains season.

Why do farmers stick with the familiar? The researchers emphasize that the preference is not irrational. Older hybrids often carry traits farmers genuinely value, including preferred grain taste and texture, better storability for household food security, and stover quality for livestock. Risk aversion, long-standing beliefs, liquidity pressures, and seed availability constraints all reinforce the status quo. But the team identified one binding constraint that has received little empirical attention: the near-total absence of independent, credible, up-to-date information on how commercially available hybrids actually perform under real farming conditions. Unlike in many high-income countries, where publicly funded variety trials guide seed selection, no equivalent system exists in most of Africa, leaving seed companies to market their products largely unchallenged. Popular varieties persist as cash cows, dominating market share for decades because companies need to recoup investments, not because they outperform.

To test whether credible data could break this pattern, the researchers intercepted 1,194 farmers, 46 percent of them women, at 36 agro-dealer shops in five towns just before the 2024 long rains planting season. Only farmers intending to buy at least two 2-kilogram bags of certified seed were enrolled, a threshold designed to exclude opportunistic free-riders while retaining genuine smallholder buyers. All participants received a voucher for a free 2-kilogram bag of maize seed, redeemable only at that shop, along with a message encouraging experimentation on a small portion of their land. The voucher covered roughly 40 percent of the seed farmers planned to plant that season.

The randomization was deliberately simple: predefined serial numbers fed into handheld tablets assigned farmers alternately to treatment and comparison groups. Both groups then received a placebo-style engagement, but for the treatment group this was replaced with a laminated chart showing farmer-reported yields for ten commercially available hybrids grown in their county during the 2023 long rains. The chart listed four columns: product name, years since release, county average yield, and the average yield of the top 25 percent of farmers. Enumerators reviewed the charts with farmers but did not let them take copies home, minimizing spillovers. Crucially, the data was brand-agnostic and promotional in no way; farmers were simply shown how products already on the market had performed for their neighbors.

The effects on stated intentions were dramatic. Only 7 percent of comparison farmers said they intended to use their voucher on one of the top two performers, compared with 25 percent of treatment farmers, an increase of 19 percentage points, or 271 percent above the baseline. Purchase uncertainty also fell sharply: while 5 percent of comparison farmers did not know what to buy, only 2 percent of treatment farmers were undecided. Overall, 62 percent of comparison farmers intended to buy a listed product versus 76 percent of treatment farmers. Demand for unlisted products fell correspondingly, from 35 percent among comparison farmers to 20 percent among treatment farmers.

Actual purchases told a slightly attenuated but still powerful story. Voucher redemption was nearly universal, with only about 2 percent of farmers in either group failing to buy seed, mostly because their desired variety was out of stock. Among those who redeemed, 5 percent of comparison farmers bought a top-two product compared with 11 percent of treatment farmers, a 135 percent increase over baseline. Stockouts clearly constrained the full effect: 25 percent of treatment farmers had intended to buy a top performer, but only 11 percent managed to do so. Comparison farmers, by contrast, shifted upward from their stated intentions inside the shop, likely steered by agro-dealer recommendations, while treatment farmers held closer to their informed choices.

Two additional findings strengthen the case. First, the information genuinely pushed farmers toward experimentation rather than simply subsidizing familiar choices. While 24 percent of comparison farmers intended to use the voucher on a product outside their original purchasing plan, 40 percent of treatment farmers did so, and virtually no comparison farmers planned to use the voucher on a variety different from their general seed purchases, compared with 11 percent of treatment farmers. Second, and unexpectedly, the effect spilled over into unassisted purchasing: treatment farmers were significantly more likely to buy a top-two hybrid with their own money for their main seed needs, rising from 6 to about 9 percent, and were more likely to switch from their original plans to top performers. The researchers caution that this exploratory result warrants further study, but it suggests performance information may reshape preferences more broadly than subsidized experimentation alone.

Farmers themselves rated the information highly. Ninety-five percent found it easy to understand, 91 percent considered it trustworthy, and 99 percent called it useful for seed selection. Half of treatment farmers singled out the top-25-percent yield column as the most meaningful, while 40 percent valued the county average and only 10 percent prioritized product age. Challenges were modest: 13 percent found it hard to differentiate between products and 9 percent struggled to choose from the list, hinting that clearer visual aids or more tailored presentation could help a minority of users. Notably, treatment farmers cited agro-dealer influence less often than comparison farmers, suggesting the data reduced reliance on shop-floor persuasion.

The authors are careful about the limits of their findings. Because both groups received the voucher, the results capture the marginal effect of information conditional on partial liquidity relief, not the effect of information alone. The sample comprised farmers buying certified seed from formal agro-dealers, so conclusions may not extend to the informal and intermediate seed systems where many smallholders operate. Agro-dealer interactions with comparison farmers and stockouts likely compressed the measured gaps, meaning the estimates are conservative lower bounds on the true informational effect. Still, the policy implications are clear: an independent, sustained system for evaluating and disseminating varietal performance could shift seed markets from branding-driven to quality-driven competition, provided testing reflects real smallholder conditions, data covers farmer-relevant traits such as drought tolerance and maturity, and supply chains can actually stock the varieties farmers come asking for. The question, the study concludes, is no longer whether farmers can use performance information to choose better seed, but whether the institutional infrastructure to deliver that information at scale can be built and sustained.

Subject of Research: The effect of independent varietal performance information on maize seed choices among smallholder farmers in Kenya

Article Title: Does varietal performance information drive better seed choices? Experimental evidence from Kenya

Article References: Ndegwa, M. K., Bulinda, C. M., Kariuki, S., Donovan, J., Rutsaert, P., & Jaleta, M. (2026). Does varietal performance information drive better seed choices? Experimental evidence from Kenya. Journal of Agriculture and Food Research, 31, Article 103334. https://doi.org/10.1016/j.jafr.2026.103334

Image Credits: AI Generated

DOI: Not provided

Keywords: maize, seed choice, varietal turnover, Kenya, randomized controlled trial, yield performance data, smallholder farmers, agro-dealers, seed systems, information intervention, hybrid varieties, agricultural economics