Increasing Adoption and Varietal Turnover of Seed: The Role of Consumer and Producer Side Interventions

Last registered on February 20, 2024

View Trial History

Pre-Trial

Trial Information

General Information

Title

Increasing Adoption and Varietal Turnover of Seed: The Role of Consumer and Producer Side Interventions

RCT ID

AEARCTR-0010666

Initial registration date

December 20, 2022

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

January 03, 2023, 4:49 PM EST

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

Last updated

February 20, 2024, 3:36 PM EST

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

Locations

Country

Uganda

Region

Busoga

Primary Investigator

Name

Bjorn Van Campenhout

Affiliation

Ifpri

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Leocardia Nabwire

PI Affiliation

Ifpri

Contact Investigator

PI Name

Carly Trachtman

PI Affiliation

IFPRI

Contact Investigator

PI Name

Berber Kramer

PI Affiliation

IFPRI

Contact Investigator

PI Name

Gahaw Tadesse Abate

PI Affiliation

IFPRI

Contact Investigator

PI Name

Liesbeth Colen

PI Affiliation

University of Göttingen

Contact Investigator

Additional Trial Information

Status

Completed

Start date

2023-02-15

End date

2024-02-01

Keywords

Agriculture, Behavior

Additional Keywords

technology adoption, subsidies, sunk cost effect, demonstration

JEL code(s)

Q16, H24, O33, D91

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

To increase adoption of new agricultural technologies, both push (supply side) and pull (demand side) factors are important. As a push factor to increase adoption of a particular technology such as an improved seed variety, some level of initial subsidy is often offered. For instance, companies may offer free trial packs of new improved seed varieties; governments may offer subsidies to increase varietal turnover. However, it is also often argued that if something was subsidized (or even free), it may not be used for the intended purpose. In this paper, we first test the effectiveness of free trial packs by testing if farmers that receive a sample of a new improved seed variety are more likely to adopt it in the future than a control group of farmers who did not get a free sample. Furthermore we test whether farmers learn differently from seed that was obtained for free than if they had to pay a (small) price for it. This questions is investigated using BDM auction—essentially a two stage pricing design—such that we can disentangle the selection effect, whereby farmers that are prepared to pay a price are likely to be more motivated to learn from it for subsequent adoption decisions, and the sunk cost effect, where a product that has a price attached to it is valued more. In addition to the supply side intervention, we also test the relative effectiveness of a demand sided intervention for adoption of new or improved varieties—an area often overlooked in existing research. In particular, we cross-randomize an intervention where households are demonstrated how to prepare the new seed variety and get the ability to taste it.

External Link(s)

github repository for the project

Registration Citation

Citation

Colen, Liesbeth et al. 2024. "Increasing Adoption and Varietal Turnover of Seed: The Role of Consumer and Producer Side Interventions." AEA RCT Registry. February 20. https://doi.org/10.1257/rct.10666-5.0

Sponsors & Partners

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

Request Information

Experimental Details

Interventions

Intervention(s)

There will be two main treatments in this experiment, each corresponding to one of the two factors in the 2x2 factorial design (see experimental design below). For the first factor, the treatment level will consist of a seed trail pack that the household receives. This trial pack will be an improved seed variety (hybrid seed) that is available in the market but at the same time not yet widely adopted by farmers. The control level for this factor will simply be the absence of a seed trial pack, that is, these household will not receive a seed trial pack.
For the second factor, the treatment level consists of an event where farmers are shown how easy it is to prepare food grown with the improved variety and farmers are allowed to taste. Also here, control level for this factor will again be the absence of these events, that is, households in control villages will not be exposed to such events.
To investigate screening/sunk cost effects, we add a third interventions that involves a BDM auction for the seed trial packs (in a sense this can be seen as a subsidized seed trail pack treatment).

Intervention Start Date

2023-02-15

Intervention End Date

2023-08-15

Primary Outcomes

Primary Outcomes (end points)

seed used at endline

Primary Outcomes (explanation)

The main objective is to study varietal turnover and so we want to know is promoting seed via the production and/or consumption side increased adoption in the future. Note that we are not directly interested in the use of the seed trial pack for the main experiment (though we are for differentiating between sunk cost/screening effects); we are interested in knowing if farmers will adopt seed the season following the season where they were able to experiment with the seed trial pack.

Secondary Outcomes

Secondary Outcomes (end points)

production and consumption of improved seed varieties

Secondary Outcomes (explanation)

Experimental Design

The study will use a clustered 2x2 factorial design. The clusters will be villages, in which a fixed number of households will be selected to participate in the study. To study screening and sunk cost effects, we confine ourselves to the treatment factor of the seed trial intervention (that is, half of the sample that is used to power the 2x2 factorial design). In this subset, we match to each treated household a household that will be offered seed at a non-zero transaction cost. While all households in the treatment factor of the seed pack intervention will be subjected to a BDM, we will make sure that half of these individuals will receive a 100 percent discount (that is, make sure that the price the computer draws is zero).

Experimental Design Details

Randomization Method

randomization done in office by a computer

Randomization Unit

villages

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

27 clusters in each of the four treatment cells + 2 times 27 clusters for the sunk cost/selection effect experiment with the BDM.

Sample size: planned number of observations

3240 households

Sample size (or number of clusters) by treatment arms

540 pure control households in 27 pure control villages, 540 free seed pack households in 27 free seed trial pack villages, 540 subsidized seed pack households in 27 subsidized seed trial pack villages, 540 consumption intervention households in 27 consumption intervention villages, 540 consumption intervention + free seed trail households in 27 consumption intervention + free seed trail villages, 540 consumption intervention + subsidized seed trail households in 27 consumption intervention + subsidized seed trail villages

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

The primary outcome we used for power calculations is a binary indicator for use of improved seed at the farmer level. We used a baseline seed use rate of 32 percent. Inter cluster (within village) correlation for this outcome has been estimated to be 0.15. We assumed similar treatment effects for both the seed trail treatment and the consumption (a 13.5 percentage point increase). For the interaction effect, we assumed a 23.5 percentage point increase. We used HC3 standard errors clustered at the village level for the power calculations. With this setting, we are not powered to detect the three effects simultaneously. In only 66 percent of cases we are able to estimate a positive effect at the five percent significance level for both treatments and their interaction. However, if we consider the treatments separately, we hit conventional power levels for both treatments, and get up to 0.97 for the interaction effect. We are certain to identify at least one of the three parameters of interest (seed packs, consumer intervention, or the interaction).

Supporting Documents and Materials

IRB