Disseminating Innovative Resources and Technologies to Smallholders (DIRTS)

How to cite

Udry, Christopher; Kolavalli, Shashidara; Fosu, Mathias; Karlan, Dean. 2014. "Disseminating Innovative Resources and Technologies to Smallholders (DIRTS)." AEA RCT Registry. AEARCTR-0000190. https://doi.org/10.1257/rct.190-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Christopher Udry* , Shashidara Kolavalli , Mathias Fosu , Dean Karlan

* Primary investigator

Registry ID
AEARCTR-0000190
DOI
https://doi.org/10.1257/rct.190-1.0
First registered
Last updated
Status
On going
Trial dates
2014-01-01 – 2016-03-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/190

Abstract

In Ghana’s Northern Region, smallholder farmers cultivate rainfed crops, face significant risk of weather shocks, chronically underinvest in input technologies, achieve just a fraction of potential yields, maintain limited liquid savings and may be food insecure. Innovations for Poverty Action (IPA), the International Food Policy Research Institute (IFPRI), the Savannah Agricultural Research Institute (SARI), and the Ghana Ministry of Food and Agriculture (MoFA) will examine the barriers to smallholder farmer adoption of intensified cultivation practices and risk management tools. IPA will further measure the impact of three innovative, potentially scalable programs on farm production and profitability, consumption and food security, intra-household labor allocation, asset holdings and rural household resilience. The Disseminating Innovative Resources and Technologies to Smallholders (DIRTS) project will use the randomized controlled trial methodology to measure the impact of providing assured rural access to (1) improved information flows through Android-based extension applications with community extension agents, (2) increased access to improved-yield agro-input packages, and (3) commercial drought index insurance. The baseline survey for DIRTS will be completed in May 2014 covering 3,240 households in 162 communities. Communities, and households within communities, will be randomized into one of four treatment groups: insurance and extension; insurance and agricultural inputs; insurance, extension, and agricultural inputs, and insurance only. Two main evaluation tools will be used to study DIRTS households: comprehensive annual surveys to collect household and plot level data (from year 1), and weekly tracking of household labor surveys during the agricultural season (from year 2).

Intervention

The Disseminating Innovative Resources and Technologies to Smallholders (DIRTS) project will use a randomized controlled trial to measure the impact of improved flows of extension information, access to agricultural input packages, and rainfall index insurance on agricultural intensification, specifically the use of fertilizers and improved seeds.

To further examine the importance of weather-related risk to farm investment, all farmers in our study will be able to purchase a commercial rainfall index insurance product, developed by the Ghana Insurers Association (GIA). In 2014 we will offer free insurance to a randomly-selected subset of our sample, and actuarially-fair insurance to everyone else.

Based on past experience, we expect take-up to be at or near 100% for the free insurance, and low for actuarially-fair insurance. We anticipate a strong investment response to the free insurance. In year 2, given that communities now have experience with the insurance, we expect that there will be strong demand at actuarially fair and higher prices. We will also experiment with marketing by local notables, to see if this increases demand.

Second, to test the importance of unsure, untimely and costly access to agricultural inputs, DIRTS will make commercial fertilizer, improved seeds and other agricultural inputs available to selected communities at different points during the year instead of just after land preparation. These inputs will be sold at market price by existing agro-input dealers, who are based in the districts. The project will facilitate linkages between these suppliers and the communities by organizing marketing events and subsidizing transport. The primary linkage will be provided by community marketing agents (CMA) based in each community.

Third, to test the importance of imperfect farmer knowledge of farming best practices, randomly selected communities will be provided with more intensive extension through a Community Extension Agent (CEA). CEAs will be based in their own communities and will use Android phones to deliver standardized weekly extension messages, and will use mobile technology (pre-loaded database) to offer appropriate and time-sensitive advice to the farmers.

Experimental design

A two-step randomization was followed for this study: a) community level, and b) household level. The community level randomization was stratified on two variables: (1) access to market and (2) whether insurance had been marketed to that community in 2013. 162 eligible communities were randomly assigned to four treatment groups balanced on (1) number of compounds/community size, and (2) average household size. The four groups are: (i) Treatment 1- Insurance and Extension; (ii) Treatment 2- Insurance and Agricultural Inputs (Marketing); (iii) Treatment 3- Insurance, Agricultural Inputs (Marketing), and Extension; (iv) Treatment 4- Insurance only. The extension treatment is administered at the household level, whereas the marketing and insurance at market price interventions are at the community level.

Second, twenty households were randomly selected per community for the evaluation, balanced on 1) whether the household head was also the head of the compound, 2) the size of the household, and 3) total acreage. The sample is comprised of 3,240 households. Surveys were administered to two respondents per households: the household head (R1), and a second respondent (R2) selected using the following order: priority given to a female spouse, any other female adult if a female spouse was unavailable, and and if no female adult was present, then any other male adult. For the extension advice intervention and the free insurance, ten households each in the relevant communities were randomly selected (with replacement) from the 20 randomly selected households for the evaluation.

At the household level, this means that in:

Treatment 1- Insurance and Extension: all households have access to insurance at market prices. Twenty households are randomly selected to be surveyed as a part of the evaluation. Of these 20 households, 10 households each are randomly selected into two different treatment groups: 10 are selected for a grant of free insurance and 10 (with replacement) are selected for extension advice.

Treatment 2- Insurance and Agricultural Inputs (Marketing): all household have access to insurance at market price and to the marketing treatment. Twenty households are randomly selected to be surveyed as a part of the evaluation. Of these 20 households, 10 households are randomly selected to receive the grant of insurance.

Treatment 3- Insurance, Extension and Agricultural Inputs (Marketing): all households have access to insurance at market price and to the marketing treatment. Twenty households are randomly selected to be surveyed as a part of the evaluation. Of these 20 households, 10 households each are randomly selected into two different treatment groups: 10 to receive the grant of insurance and 10 to receive extension advice.

Treatment 4- Insurance only: all households are given the opportunity to buy insurance at market price. Twenty households are randomly selected to be surveyed as a part of the evaluation. Of these 20 households, 10 households are randomly selected to receive the grant of insurance.

Furthermore, in order to test the effect of Community Based Marketers (CBM) on the uptake of insurance, we assign each community to one of three sub-treatments (balanced on number of compounds, average household size and access to market):

Village headman in 54 communities
Women Organizer in 54 communities
Based on merit in 54 communities

Design details

Randomization method
Randomization done in office by a computer using Stata
Randomization unit
Community and household level randomization. The full explanation is provided in the experimental design above.
Clusters
162 communities
Observations
3240
Arms
Treatment 1 (insurance + extension): 51 communities, 1020 households Treatment 2 (insurance + inputs): 30 communities, 600 households Treatment 3 (insurance + extension + inputs): 30 communities, 600 households Control (insurance only): 51 communities, 1020 households
Primary outcomes
Agricultural productivity, Farm production, Profits (Farm and non-farm business), Food security, Income, Consumption, Access to Agricultural inputs, Asset holding, Intra-household labor allocation, Rural household resilience

Cited by

No citation data available for this registration yet.

Keywords