Incomplete Information, Costly Adolescent-Elder Interactions, and Miscoordination in Agricultural Households: Evidence from School-based Agricultural Extension in Liberia

I design a 2 x 2 randomized experiment among 1000 households with students in 50 schools that join the school-based agricultural program (4-H Liberia) in 2021. Our household-level interventions encourage students and their elders to imagine, at the beginning of the program (and of the first rainy season), the consequences of students' active participation in the program.

The first randomization aims to encourage elders (in students' households) to learn about and support their students' potential growth in farming skills, attitudes, and commitments. Half of the households receive an invitation to treatment video sessions; the other half receive an invitation to placebo video sessions. For every household, the representative elder who attends the session must be a farmer growing crops in the promoted categories (root crops or vegetables). Each video session has 5 elders and lasts around 1.5 hours: baseline questions; video display on a laptop under palava huts; and follow-up questions. Treatment video sessions display a 14-minute video that (i) provides an overview of the program and (ii) summarizes program impact on students’ farming skills, attitudes, and livelihoods from past participants (in communities that are not included in the sample). Placebo video sessions display a 4-minute video that ONLY provides an overview of the program. Both videos are taken from footage shot during late 2020 - early 2021 by a third-party media production company in Liberia. The treatment video is a long version that encompasses everything in the placebo video. We also independently randomize who in the households attend the video sessions (both treatment and placebo): for half of the households, a male is invited to the sessions; for the other half of the households, a female is invited.

The second randomization aims to overcome potential coordination difficulties between students and their elders by encouraging students to reach out to their elders (to introduce the program, what they have learnt, suggest new farming ideas, and propose to manage farms, etc).
During the video sessions, we collect (before showing videos) elders' beliefs about students' current farming skills, attitudes, and commitments; and (after showing video) elders' forward-looking expectations about the same attributes of students in 1 year. We ask if elders would give their consent for IPA to reveal their positive expectations to students. We then manipulate students' information about their elders' expectations in a follow-up survey. In half of the households, we deliver (i) revelation messages that reveal expectations from elders to students and (ii) encouragement messages that ask students to take on the role as an agent of diffusion of new practices. In the other half of the households, we deliver ONLY encouragement messages.

2021-06-21

2021-07-31

Please refer to our pre-analysis plans.

I test the significance of my hypothesized mechanism - information asymmetry in changes of students' farming skills, attitudes, and commitments - in a 2 x 2 experiment among 1000 households with students in 50 schools that join the school-based agricultural program (4-H Liberia) in 2021.

We identify our effects as follows. I test the first hypothesis - that there is room to encourage elders in students' households to learn about the positive impacts of the program (on students' farming attributes) - using the first randomization. Households that are invited to treatment video sessions will be compared to households that are invited to placebo video sessions.

I test the second hypothesis - that even when elders in students' households have positive expectations about students' growth in farming attributes, there are difficulties for elders to reveal these expectations to students (e.g. because farming ideas and knowledge about the program belong to students) - using the second randomization. Households that receive the "revelation + encouragement" treatment will be compared to households that receive the "encouragement" treatment.

The selection of our household sample is done in schools. Students are randomly drawn from Grades 4, 5, 7, and 8 (stratified by school and gender). Informed by pilot survey data, we implement a series of screening criteria in selecting 20 students per school who are likely to participate in the program. Selected students must be aged 12-20; must have at least one elder in the household who is planting root crops or vegetables; and must be from separate households. We prioritize students who had farming experience or joined cooperative labor groups (kuu) before 2021. We also prioritize students who have both male and female elders in the household who are planting root crops or vegetables.

All baseline surveys and household-level interventions (same or next day after baseline surveys) will be conducted in June - August 2021. A follow-up survey after the major rainy season (typically April - October) will take place in November 2021 - January 2022. Two further follow-up surveys might take place in April - June 2022 and September - November 2022 to track long-run outcomes. Students' participation in program activities will be monitored by our partners 4-H Liberia and AgriCorps; these data will be linked to our baseline surveys.

Randomization done in IPA office by a computer, using student ID (which is sorted according to schools, grades, and gender of students). Randomization was done before students are recruited into the sample, and was stratified based on schools and gender of students.

For video treatment: household
For revelation treatment: household

No

1000 households in total.

1000 students (500 male, 500 female) and 1000 elder farmers (500 male, 500 female).

250 households receiving treatment video and "revelation + encouragement" treatment;
250 households receiving placebo video and "revelation + encouragement" treatment;
250 households receiving treatment video and "encouragement" treatment;
250 households receiving placebo video and "encouragement" treatment.