This is the fourth in this year’s series of posts by PhD students on the job market
Low adoption of modern technologies holds back productivity and economic growth across developing countries. A common barrier is cost: many farmers cannot afford improved seeds and complementary inputs. To encourage adoption, governments aim to make technology cheaper for poor farmers. But policies that reduce prices may discourage technology providers from innovating. The trade-off: more adoption today, less innovation tomorrow.
This tension motivates my job market paper. I study price controls on genetically modified cotton seeds in India and show that they are an effective policy to raise adoption. At the same time, by compressing margins for technology providers, price controls disrupt innovation. On net, adoption gains dominate innovation losses: farmers’ welfare increases, especially among the poor. However, ignoring the innovation response vastly overstates the policy’s benefits for farmers. I then turn to policy alternatives: seed subsidies to farmers spur adoption without deterring innovation but are fiscally burdensome; grants to seed technology providers, tied to the productivity of new varieties, deliver the highest welfare for farmers.
Technology adoption and innovation in Indian cotton agriculture
I consider the case of Indian cotton, where the tension between affordability and innovation is especially salient. On the demand side, smallholding farmers face multiple constraints that keep take-up of existing technologies low. Policies that push technology prices down are a natural lever to increase adoption and welfare.
On the supply side, productivity growth in agriculture is driven by gradual improvements, tailored to local conditions. This is because agricultural technology requires continuous adaptation to local weather, soil, and pests. Capping prices may weaken incentives to develop new technologies and erode the welfare gains that farmers get from cheaper access.
The biggest breakthrough in Indian cotton was the introduction of the Bt technology. Developed by the multinational Monsanto in the 1990s and approved in India in 2002, Bt cotton is genetically modified to protect against the bollworm, a widespread and destructive pest, and so to limit crop loss. The genetic technology alone is not directly useful to farmers: domestic seed firms must embed it in locally adapted varieties for India’s ecological conditions. To do so, firms pay a royalty to Monsanto and then set retail prices.
Identification and measurement of causal responses to price controls
When Bt seeds were first commercialized in India, the per-packet royalty fee was set at ₹1,100, at a time when non-Bt seeds sold for roughly ₹500. Complaints over the high cost of the new Bt seeds led some states to cap prices: in 2006, the three cotton-growing states of Andhra Pradesh, Maharashtra, and Gujarat imposed a maximum retail price of ₹750 per packet. The unanticipated timing of the policy across states and the localized nature of innovation provide the variation needed to identify how farmer adoption and firm innovation respond to lower technology prices.
I overcome measurement challenges by combining existing and novel data. To measure adoption, I use panel surveys of about 500 cotton farmers to track exactly which seed varieties they plant over time (~1,700 farmer-years). To measure innovation, I digitized independent quality measures from experimental field trials, covering more than 600 varieties tested at 35 stations across India (~6,800 variety-trial observations). In this way, I can observe not only how adoption responds to lower prices, but also whether the quality of new seeds changes in the aftermath of the policy.
Leveraging the state-specific price caps, I estimate causal effects on both sides of the market: cotton farmers and seed firms. In a difference-in-differences design, I compare changes in observed outcomes in states with price caps versus states without price caps. I exploit the panel structure of my data to control for time-invariant characteristics of both farmers and firms as well as for common time shocks. Focusing on farmers cultivating cotton before the policy also rules out selection on entry into cotton.
Evidence of the adoption–innovation trade-off
The policy reached farmers: seeds became cheaper, and the new technology spread. As the government-mandated caps lowered cotton seed prices by 40%, adoption of genetically modified seeds rose by 23 percentage points. The technology delivered on its promised benefits: because the newly adopted seeds are pest-resistant, farmers cut pesticides and labor, pushing total cultivation costs down. Price-controlled states then saw a massive expansion in cotton acreage, mostly through the entry of new cotton farms.
The flip side is innovation: the quality of new seed varieties declined. I document that the number of varieties released in India plummeted a few years after price caps were imposed. To provide causal estimates of innovation responses, I leverage the fact that the same varieties are evaluated across different locations and compare experimental field trials in states with versus without price caps before versus after the policy. Yields of newly released varieties fell by 30% in price-controlled states, relative to other states. The estimates are robust to sampling-based and design-based approaches to clustered inference and survive alternative procedures to deal with few clusters. The observed yield drop is consistent with location-specific seed development shifting away from price-controlled states, where returns are lower. Benchmarked against agronomic estimates of productivity gains from the original genetic technology in India (Qaim and Zilberman, 2003), the innovation response to price caps wipes out roughly half of those gains.
Quantifying the net impact on farmer welfare
Lower seed prices lifted adoption but curbed innovation. What’s the bottom line for farmers’ welfare? I quantify it with a structural model of demand and supply for seeds. The net welfare impact of the policy depends on (i) how farmers trade off seed price versus expected yield and (ii) how costly it is for firms to provide yields. To recover unobserved preferences and costs, I use nationally representative individual choice data (~47,000 farmer-years) and adapt tools from empirical industrial organization to Indian cotton agriculture. Specifically, I combine (i) a discrete-choice demand model with heterogeneous farmers and (ii) a supply system with parametric cost functions that depend on yields.
I find that farmers are better off under the policy. Lower seed prices disproportionately benefit the poor, who are more price-sensitive and less likely to adopt the new technology absent the policy. However, farmers’ welfare is 31% lower relative to a naïve benchmark with no drop in yields. Failing to account for firms’ innovation response could lead researchers and policymakers to overstate a policy’s benefits for technology users.
Implications for the design of agricultural and innovation policies
I use the estimated model to design alternative policies that better balance affordability with incentives to innovate. To support technological progress, governments often weigh two policy instruments: seed subsidies to farmers and innovation subsidies to firms.
According to my structural estimates, seed subsidies are effective at lowering prices, while preventing the yield loss triggered by price controls. However, they are prohibitively expensive for budget-constrained state governments in India. Innovation subsidies tied to the productivity of new seed varieties are more cost-effective and achieve the highest welfare for farmers. This approach is particularly well-suited to agriculture: targeting on firm productivity – for example, using data on field-trial performance – alleviates the information and implementation challenges of targeting individual farmers.
Development economists and practitioners have often focused on farmers’ choices: adopting existing technologies, changing agricultural practices. I complement this focus by bringing the supply side into the picture: my findings highlight the decisive role of the private sector in developing new, locally adapted technologies for low-income settings. Maintaining incentives to both develop and spread affordable innovations is key to designing effective policies for inclusive growth and climate adaptation.
Matteo Ruzzante is a PhD student in Economics at Northwestern University. This paper is coauthored with Felipe Berrutti (Analysis Group).
Join the Conversation