Skip to content

Expanding the Planning Horizon: Behavioral Training to Scale Drought-Tolerant Maize and Index Insurance in Tanzania

Policy Context

Smallholder farmers across sub-Saharan Africa remain exposed to drought and invest below the agronomic frontier because they cannot manage tail risk. A bundle of drought-tolerant (DT) maize and index insurance offers a solution: the seed absorbs moderate mid-season dry spells, while insurance covers severe events at an estimated benefit–cost ratio near seven to one. Despite this proven value, the bundle has not scaled. The obstacle is a learning problem. The bundle reveals its worth only through a rare compound event—a farmer must buy it, a drought must strike, and a payout must arrive—so the experience that would teach that the technology works almost never occurs within the horizon over which a smallholder plans. Conventional explanations do not bind here: awareness is near-universal, and contract knowledge shifts with even light training. The researchers argue that the binding friction is temporal. Under short planning horizons, a farmer codes the premium as a certain near-term loss and underweights the multi-season variance reduction it actually delivers.

Study Design

Working in the Morogoro and Mvomero districts of Tanzania with implementing partner Agrinature, the research team recruited roughly 7,500 farmers across 50 villages into a five-arm, cluster-randomized trial, with approximately 1,800 followed through a midline survey. The arms span a continuum of temporal integration—how aggressively each draws a farmer into evaluating the bundle across many seasons rather than one. The headline AgriShield Game is a ninety-minute, five-period board game that compresses years of seed–payout decisions into a single session: farmers draw drought cards, choose seed and insurance for each simulated season, watch outcomes resolve, and carry their cash position forward. Drought Recall reconstructs a farmer’s own six-year drought history alongside the payouts insurance would have provided. Budget Planning, a creative-control arm, has farmers project the coming season’s income and expenses with no shock content, isolating the horizon mechanism from facilitator attention. Only sales touchpoints and pure control complete the design. Adoption is tracked through farmers’ actual seed and insurance purchases over two seasons.

Results and Policy Lessons

Early evidence reveals a clear dose-response gradient: adoption is ordered almost exactly by how aggressively each arm pulls farmers into multi-period evaluation, with the experiential AgriShield Game producing the largest gains, the lighter-touch arms in between, and the untrained control the lowest. The gradient is sharpest for index insurance—the less familiar, harder-to-value half of the bundle. The effect also appears durable: twelve months after a single session, the gap between trained and untrained farmers widens rather than decays, with little insurance drop-off among returning game-arm buyers. Belief data suggest that the “resilience-dividend” mindset can form in advance of any real drought.

The emerging policy lesson is that for proven technologies whose value is revealed only through rare, compound events, a brief and low-cost simulated experience may substitute for years of market exposure.

Full results forthcoming.

Countries
Tanzania