Testing what works to accelerate varietal turnover: An experiment in Kenya

The turnover of improved crop varieties among smallholder farmers has been slow, and little is known about the drivers of varietal replacement and product substitution. Such low varietal turnover limits the extent to which investments in crop breeding programs can contribute to achieving the sustainable development goals. This cluster randomized trial, implemented in seven counties in Kenya, will therefore generate behavioral intelligence on what drives farmers', consumers' and private-sector decisions to adopt new varieties. Specifically, we will test whether varietal turnover can be accelerated through four interventions:
1) Providing trial packs of new varieties, to provide farmers with the opportunity to test new varieties at a lower cost (randomized at the village level, with additional randomization within villages in terms of who receives trial packs);
2) Providing personalized advisory services, to alleviate information constraints on how to cultivate these new varieties (randomized at the village level, with additional randomization within villages in terms of who is targeted with advisories);
3) Providing agricultural insurance subsidies, to reduce the financial risk associated with investing in more expensive seeds (randomized at the farmer level, within villages); and
4) Targeting the same set of farmers with a consumer-oriented intervention that addresses factors related to the taste, texture, nutritious value or other consumer-related characteristics of the final product associated with a new variety (randomized at the village level).

The trial is implemented in 160 villages spread across seven counties in Kenya, with 40 farmers per village recruited into the study. Results are expected to be available by December 2024.

External Link(s)

Registration Citation

Citation

Cecchi, Francesco et al. 2023. "Testing what works to accelerate varietal turnover: An experiment in Kenya." AEA RCT Registry. January 03. https://doi.org/10.1257/rct.10651-1.0

Sponsors & Partners

Interventions

Intervention(s)

We will test whether the adoption of new varieties can be accelerated through four interventions:
1) Providing trial packs of new varieties, to provide farmers with the opportunity to test new varieties at a lower cost;
2) Providing personalized advisory services, to alleviate information constraints on how to cultivate these new varieties;
3) Providing agricultural insurance subsidies, to reduce the financial risk associated with investing in more expensive seeds; and
4) Targeting the same set of farmers with a consumer-oriented intervention that addresses factors related to the taste, texture, nutritious value or other consumer-related characteristics of the final product associated with a new variety.

Intervention Start Date

2022-09-15

Intervention End Date

2023-08-31

Primary Outcomes

Primary Outcomes (end points)

Adoption of promoted varieties. Both aggregate and disaggregated by gender.

Primary Outcomes (explanation)

Whether the farmer chooses to plant any seeds of the varieties that were promoted through trial packs in the previous season. Because farmers who receive trial packs of non-hybrid varieties can replant these seeds, we also consider more stringent definitions for adoption, including a) whether the farmer obtains any seeds of the promoted varieties from sources other than their own recycled seeds, and b) whether the farmer plants more seed used of the promoted varieties, compared to what was planted in the season during which the farmer received trial packs.

Secondary Outcomes

Secondary Outcomes (end points)

Investments in seed, adoption of modern production technologies, investments in agriculture, agricultural productivity for target crops (yield per acre), total value of production, agricultural income. Both aggregate and disaggregated by gender.

Secondary Outcomes (explanation)

- Investment in seed: The total amount spent on seed across all varieties that a farmer may use.
- Adoption of modern production technologies: An indicator for whether the farmer uses fertilizer, pesticides, machinery, or any other modern production technology (including and excluding improved seeds).
- Investments in agriculture: The total amount spent on agricultural production, including any amount of money spent on seeds, and any hired labor. We will construct two investment measures: one excluding and one including the total value of family labor invested in agricultural production.
- Agricultural productivity for target crops: Average productivity across all target crops that a farmer has grown. To aggregate across different crops with different levels of average productivity, we normalize productivity for a given crop using baseline productivity levels. As weights, we use the area cultivated with a given crop.
- Total value of production: we multiply agricultural production for a farmer with the median output price within a county, and aggregate across all crops.
- Agricultural income: Total value of production minus total investments in agriculture (both excluding and including family labor).

Experimental Design

We randomized, at the village level, and with stratification by county, whether farmers were offered trial packs of new varieties and whether they would receive personalized advisory services. Within each treatment arm, we will further randomize whether or not the consumer-oriented intervention is implemented.

Within each village, we recruited 40 farmers. All farmers are offered insurance, but at the farmer level, the insurance premium discount level is randomized into high (25% of the sample), low (25% of the sample) or no discount (50% of the sample within a village). We cross-randomize whether or not a farmer is provided with a trial pack; 20 farmers per trial pack village receive a trial pack.

Experimental Design Details

Not available

Randomization Method

Randomization done in office by a computer

Randomization Unit

Village level randomization for trial pack, personalized advisory services, and consumer-oriented intervention
Individual randomization for insurance premium subsidies

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

160 villages

Sample size: planned number of observations

6400 farmers (40 farmers per village)

Sample size (or number of clusters) by treatment arms

1A. No trial packs, no personalized advisory services, no consumer intervention: 20 villages
1B. No trial packs, no personalized advisory services, consumer intervention: 20 villages
2A. No trial packs, personalized advisory services, no consumer intervention: 20 villages
2B. No trial packs, personalized advisory services, consumer intervention: 20 villages
3A. Trial packs, no personalized advisory services, no consumer intervention: 20 villages
3B. Trial packs, no personalized advisory services, consumer intervention: 20 villages
4A. Trial packs, personalized advisory services, no consumer intervention: 20 villages
4B. Trial packs, personalized advisory services, consumer intervention: 20 villages

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

Supporting Documents and Materials

IRB

Institutional Review Boards (IRBs)

IRB Name

Maseno University Scientific and Ethics Review Committee

IRB Approval Date

2022-07-25

IRB Approval Number

MUERC/00922/20

IRB Name

International Food Policy Research Institute

IRB Approval Date

2020-03-19

IRB Approval Number

MTID-20-0306

IRB Name

International Food Policy Research Institute

IRB Approval Date

2019-12-17

IRB Approval Number

MTID-19-1270

Analysis Plan