Matching new agricultural technology with the demand for innovation of low-income farmers

Last registered on March 28, 2023


Trial Information

General Information

Matching new agricultural technology with the demand for innovation of low-income farmers
Initial registration date
November 01, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
November 05, 2021, 4:34 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
March 28, 2023, 11:31 AM EDT

Last updated is the most recent time when changes to the trial's registration were published.


Primary Investigator

Cornell University

Other Primary Investigator(s)

PI Affiliation
Cornell University

Additional Trial Information

On going
Start date
End date
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Why are farmers so slow to adopt seemingly profitable new technologies? A vast literature has examined this question, focusing mainly on demand-side factors affecting farmers. An under-appreciated aspect of this problem is that new technologies are not typically tailored to the needs of many of the world’s small-scale farmers. Although public research centers are the main supplier of agricultural technology in most low-income countries, they often lack resources and mechanisms to match new technologies with farmers’ demand for innovation. This constraint can limit the ability of agricultural R&D to prioritize and target new technology. Yet, we know little about the impact on technological change when the choices of resource-constrained agricultural researchers do not reflect the preferences of the farmers who must ultimately take up the new technologies. We propose a field experiment to estimate the effect of imperfect targeting on farmers’ take-up of improved beans varieties in Costa Rica. Commonly, plant breeding programs release a single new variety per cycle to supply a large and heterogeneous group of farmers. The challenge is that we only observe the varieties that breeders release, not the counterfactual varieties they may have chosen to develop had they been able to better target farmers’ preferences for new varieties. We recreate these counterfactual scenarios in an experiment that relaxes information constraints on breeders to estimate their effect on the adoption decisions of 840 small and medium-scale farmers.
External Link(s)

Registration Citation

Puerto, Sergio and Miguel Gómez. 2023. "Matching new agricultural technology with the demand for innovation of low-income farmers." AEA RCT Registry. March 28.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


Randomized control trial offering a new seed variety for purchase to low- and medium-scale farmers in Costa Rica.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
1) Adoption take-up rates, 2) Planted area under the adopted variety
Primary Outcomes (explanation)
1) Adoption takeup rates: we identify adoption with a dummy variable for each farmer will capture who adopts the new variety. Adoption rates across treatments are constructed using the average adoption takeup for all farmers in each group.
2) Planted area under the adopted variety: We use the total area planted using the new variety, and also the fraction of planted area with new varieties out of the total farm size.

Secondary Outcomes

Secondary Outcomes (end points)
3) Farm yield, 4) Input allocations (labor, land, and fertilizer), 5) Yield losses (biotic and abiotic losses in production), and 6) Price per unit sold
Secondary Outcomes (explanation)
3) Farm yield: net production (total production minus losses) over the planted area
4) Input allocation: distribution of inputs in the farm production calculated as total input use, and input use over total production
5) Yield losses: amount lost in production due to pest, extreme weather events, and non-idiosyncratic shocks
6) Price per unit sold: price received for quintal (35kg)

Experimental Design

Experimental Design
Our experimental design consists of two stages designed to answer four research questions. In the first stage, we will conduct trials with half of the farmers to test the on-farm performance of five new bean varieties. The trials also allow us to capture farmers’ revealed preferences for these seeds. Using a blind procedure, each farmer in the trial receives a random set of three varieties to evaluate their performance in several traits. Each farmer in the sample will then be advised about which variety is best for their plot, using the trial results and expert information from crop scientists that mimics their process for making real-world recommendations when a new variety is released. Farmers will also be asked to decide what variety they would like to buy for the next season.

In the second stage, we will test whether imperfect targeting limits the adoption of the new varieties. Our intervention consists of a randomized offer of one of the new varieties to each farmer. First, the benchmark group will be offered for purchase a fixed amount of a new variety of their preference from the set tested in the trials. This group represents an ideal but unrealistic situation in which the supply of new crop varieties matches exactly farmers’ revealed preferences for the new varieties. In other words, each farmer in the benchmark group faces their own synthetic market for seeds. On the other hand, the treatment group will be offered the variety recommended by the breeders, which is the case with many seeds systems in low-income countries, especially for minor crops. The comparison between the treatment and benchmark groups captures the mismatch that arises due to differences between plant breeders’ prediction of the variety that is better for the overall conditions of many farmers versus farmers’ choices. Thus, this design allows us to study realistic farmer-researcher interactions involved in the release and diffusion of new crop varieties.
Experimental Design Details
Randomization Method
Clustered sampling: Clusters were selected based on logistical and research criteria. A total of 7 individuals per cluster were randomly selected. Random samples were constructed using computer software and based on administrative data of the target population.
Randomization Unit
-First stage: stratified randomization (village level)
-Second stage: individual level randomization
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
120 villages
Sample size: planned number of observations
800 farmers
Sample size (or number of clusters) by treatment arms
200 farmers per treatment arm
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We estimate a conservative detectable effect size in this sample of at least a 10% difference in the take-up rates of the new varieties, assuming a standard deviation of 20%, 0.05 significance, and 0.9 power levels.

Institutional Review Boards (IRBs)

IRB Name
Cornell University Institutional Review Board
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials