The research team, led by Stephen Prah and including John N. Ng’ombe, Bright Owusu Asante, and Jonathan Masasi, framed their analysis around a deceptively tricky statistical problem. Farmers who adopt climate-smart technologies are not a random sample of the farming population. They tend to be better educated, better connected to extension agents, and more likely to have access to credit. Any naive comparison between adopters and non-adopters therefore risks conflating the effects of the technologies themselves with the pre-existing advantages of the people who choose them. To untangle this, the authors deployed a dose-response model with Heckman selection correction, a two-stage econometric approach that explicitly models both the decision to adopt and the intensity of adoption, while correcting for the selection bias that would otherwise distort the estimates.

The dose-response framework is what makes this study unusual. Most adoption studies in agricultural economics treat technology use as a binary variable: a farmer either adopts or does not. But in the field, adoption is a continuum. One household might plant a small portion of its land with drought-tolerant maize while continuing business as usual elsewhere; a neighbor might restructure the entire farm around a full suite of climate-smart practices. By measuring adoption intensity, the share of recommended technologies and practices a household actually implements, the researchers could trace out how welfare outcomes change across the entire spectrum of engagement, from minimal uptake to comprehensive adoption.

The headline finding is a U-shaped relationship between adoption intensity and maize yield. At low levels of adoption, yields actually decline relative to non-adopter baselines, a pattern the authors attribute to the transition costs and learning curves that come with new practices. Partial adoption can mean partial disruption: new seeds planted without the complementary fertilizer schedules, or mulching introduced without adjusted tillage, may initially perform worse than the familiar system they replace. But as adoption intensity climbs past a threshold, the curve turns upward, and heavily adopting households pull decisively ahead. On average across the sample, higher adoption intensity was associated with a 21 percent increase in maize yield compared with what farmers would have achieved without the technologies.

Productivity, however, was only half of the story. The researchers also examined household asset accumulation, a slower-moving but arguably more meaningful indicator of long-term welfare. Assets, livestock, equipment, housing quality, and similar holdings, capture whether productivity gains translate into durable improvements in living standards rather than being consumed away season by season. Here again the relationship was nonlinear and positive: greater adoption intensity corresponded to an average 19.5 percent increase in household assets. The parallel results suggest that climate-smart farming is not merely boosting one season’s harvest but is reshaping the economic trajectory of adopting households over time.

Perhaps the most policy-relevant insight concerns heterogeneity. The effects of adoption were not uniform across the sample. Households differed in how much they benefited depending on their resource endowments, education levels, and exposure to climate stress. This matters because development programs often roll out technology packages with the implicit assumption that benefits will diffuse evenly. The Ghanaian evidence suggests otherwise: the same technology bundle that transforms one farm’s fortunes may deliver only marginal gains to another, and the distributional consequences of promotion campaigns deserve as much attention as their average effects.

The study also identified the factors that push farmers toward adoption in the first place. Education emerged as a consistent driver of both the likelihood and the intensity of adoption, reinforcing a well-established pattern in which human capital determines the capacity to absorb and implement complex agronomic information. Access to credit mattered for obvious reasons: many climate-smart inputs, from certified drought-tolerant seed to soil amendments, require upfront cash that cash-constrained households cannot easily mobilize. Contact with agricultural extension services was another strong predictor, since extension agents remain the primary conduit through which technical knowledge reaches rural communities. Finally, farmers who perceived drought as a significant and growing threat were more likely to invest in adaptive technologies, suggesting that climate risk awareness is itself a lever for change.

These determinants point toward a practical agenda. If adoption intensity is what drives the welfare gains, and if the drivers of intensity are education, credit, extension, and risk perception, then policy should aim squarely at deepening adoption rather than simply counting adopters. Programs that celebrate the number of farmers reached with a new seed variety may be missing the point: a farmer who plants 10 percent of her land with drought-tolerant maize may see no benefit at all, while one who reorganizes her whole operation around climate-smart principles reaps gains on the order of a fifth of her yield and assets. The difference between shallow and deep adoption is, on this evidence, the difference between wasted subsidy and transformative investment.

For Ghana, where maize is a staple crop and rainfed agriculture dominates, the stakes are high. Climate projections for West Africa point to rising temperatures, more erratic rainfall, and increased frequency of drought stress, all of which threaten the productivity of smallholder systems that already operate close to the margin. Climate-smart agriculture has been promoted by international initiatives as a triple-win strategy, raising productivity, enhancing resilience, and reducing emissions where possible, but the evidence base for its welfare effects has remained thin, particularly regarding how benefits scale with the depth of adoption. This study helps fill that gap with rigorous causal identification and a design that respects the reality of partial, incremental, and bundled technology use on real farms.

The broader lesson may resonate well beyond Ghana’s maize belt. As governments and donors channel billions into climate adaptation for agriculture, the question of dosage, how much adoption is enough, is becoming central to program design everywhere from the Sahel to South Asia. The Ghanaian evidence offers a clear if demanding answer: partial measures may disappoint, and the transition period may even hurt, but sustained, intensive adoption of climate-smart technologies pays off in both harvests and household wealth. Getting farmers to that threshold will require the unglamorous work of building credit systems, strengthening extension networks, and convincing rural households that a changing climate makes the investment worthwhile. The science of the technologies, in other words, is only half the equation; the economics of getting farmers fully on board is the other half, and it is where the battle for climate-resilient agriculture will be won or lost.

Subject of Research: Effects of climate-smart agricultural technology adoption intensity on maize yield and household welfare in Ghana

Article Title: How much climate-smart agriculture is enough? A dose-response analysis of technology adoption effects on farmer welfare in Ghana

Article References: Prah, S., Ng’ombe, J. N., Asante, B. O., & Masasi, J. (2026). How much climate-smart agriculture is enough? A dose-response analysis of technology adoption effects on farmer welfare in Ghana. PLOS Climate, 5(10), e0000886. https://doi.org/10.1371/journal.pclm.0000886

Image Credits: AI Generated

DOI: 10.1371/journal.pclm.0000886

Keywords: climate-smart agriculture, Ghana, maize, adoption intensity, dose-response analysis, farmer welfare, crop yield, household assets, extension services, credit access, drought resilience, agricultural economics