Demand and Supply Constraints to Improved Sorghum Technology Adoption and their Gender-Differentiated Effects in Burkina Faso

How to cite

Porter, Maria; Dillon, Andrew. 2017. "Demand and Supply Constraints to Improved Sorghum Technology Adoption and their Gender-Differentiated Effects in Burkina Faso." AEA RCT Registry. AEARCTR-0002119. https://doi.org/10.1257/rct.2119-1.0

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

Maria Porter* , Andrew Dillon

* Primary investigator

Registry ID
AEARCTR-0002119
DOI
https://doi.org/10.1257/rct.2119-1.0
First registered
Last updated
Status
On going
Trial dates
2014-02-15 – 2018-06-15
Countries
Private
JEL codes
Q12, O13, J22
Registry record
http://www.socialscienceregistry.org/trials/2119

Abstract

Supply and demand constraints reduce adoption of improved sorghum technology in the West African Sahel. We will work with sorghum breeders and agro-input suppliers in Burkina Faso to compare alternative mechanisms to encourage adoption of improved seed and fertilizer micro-packs. A demand side treatment will be targeted by social network characteristics to understand the information effects of farmer take-up and spillover based on social network characteristics from a randomized distribution of micro-packs. A social network census will reveal the extent to which villagers insure one another against idiosyncratic risk specifically through exchange of seed, use of complementary inputs, intahousehold labor substitution and assets. The supply side of the randomized control trial will test whether consistent market supply, credit constraints and farmer commitment explain low adoption and potential supply side marketing mechanisms to increase adoption.

Experimental design

The demand side interventions will target a seed and fertilizer micro-packet to farmers based on their social network characteristics. In treatment groups A, B, and C villagers will also receive a free training on improved sorghum technology. Free micro-packets will be randomly distributed to 20 farmers in each village in treatment group A, while in treatment groups B and C, free distribution of the micro-packet will be distributed to those 20 farmers with the highest degree of centrality or betweeness, respectively. Degree of centrality and betweeness are two measures of social network characteristics that measure the connectivity of a farmer and the influence of a farmer respectively. Degree of centrality measures the total number of social network connections a farmer has within his or her village. Betweeness measures the share of the shortest paths from all pairs of households in the network that are connected to that household. On the supply side of the experiment, we experiment with different ways of organizing agro-input fairs, where suppliers sell fertilizer and seeds to farmers. We compare farmers' purchases across three input fairs in 36 rural villages: just after harvest (January), at planting (June), and at planting with a price subsidy.

Design details

Randomization method
Randomization was done in the office by a computer using STATA. The sampling process started with a complete listing of all villages in the 3 provinces using data collected by IPA visits to villages, commune level data, and the 2006 census data. The latter includes the following variables: region, province, commune, village, number of households per village, population size per village and village locality (urban or rural). Note IPA only visited villages in designated rural areas. The total number of villages in designated rural areas in the three provinces is 925. We then applied the following selection criteria: 1- Keep villages where the population size is between 190 and 800, provided the number of households does not exceed 120 if known. This leaves us with 226 villages. This criterion is used in order to avoid villages that would be too large for collecting social network census data. 2- Obtain the final sample by randomly selecting 71% of villages from each of the three provinces, resulting in an initial sample of 160 villages. 3- From this sample of 160 villages, we randomly assign villages to 6 treatment arms and 1 control group. We then conducted village enumerations in all sampled villages. In all villages in Group A and the control, approximately 30 households were randomly selected for a detailed baseline household survey. Of these, approximately 15 households were randomly selected to receive treatment in villages in Group A.
Randomization unit
Treatments were randomized by village. Within the 80 villages receiving random distribution of free kits, randomization was by household. In households with multiple sorghum farmers, randomization was by individual sorghum farmer.
Clusters
168 villages
Observations
16,800 households across all villages in the study. For the detailed household survey conducted in treatment A and the control, there would be 100 villages and roughly 3000 households.
Arms
80 villages in Treatment A - random distribution of free microdosing kits and training 16 villages in Treatment B - distribution of free microdosing kits based on degree of centrality 16 villages in Treatment C - distribution of free microdosing kits based on betweeness 12 villages in Treatment D - early commitment offer at fixed "market" price 12 villages in Treatment E - late commitment offer at fixed "market" price 12 villages in Treatment F - late commitment offer at discounted price 20 village in control group
Primary outcomes
Labor supply, sorghum yield, fertilizer use, use of microdosing, knowledge of microdosing

Cited by

No citation data available for this registration yet.

Keywords