Grain Today, Gain Tomorrow: Evidence from a Storage Experiment with Savings Clubs in Kenya

How to cite

Robinson, Jonathan; Francis, Eilin; Aggarwal, Shilpa. 2017. "Grain Today, Gain Tomorrow: Evidence from a Storage Experiment with Savings Clubs in Kenya." AEA RCT Registry. AEARCTR-0002188. https://doi.org/10.1257/rct.2188-1.0

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

Jonathan Robinson* , Eilin Francis , Shilpa Aggarwal

* Primary investigator

Registry ID
AEARCTR-0002188
DOI
https://doi.org/10.1257/rct.2188-1.0
First registered
Last updated
Status
Completed
Trial dates
2015-08-01 – 2016-11-30
Countries
Kenya (Africa)
JEL codes
O12, O13, O16, D14, E21
Registry record
http://www.socialscienceregistry.org/trials/2188

Abstract

In many rural areas in Africa, staple food prices display predictable, sizeable price changes over the year, from post-harvest troughs to lean season peaks. We experimentally evaluate a group-based storage scheme with 139 treatment and control savings clubs in Kenya, designed to enable farmers to take advantage of these intertemporal price gains by storing maize after harvest. Treatment clubs were offered a savings product in which farmers could deposit a fraction of their harvest into a communal storage account, to be sold later in the season. We find that 58% of farmers took up the scheme, and that treatment farmers were 7-23 percentage points more likely to store maize for the hungry season and stored about 10% more maize than control farmers. We find that the vast majority of this maize was sold, leading to large effects on sales: treatment farmers were about twice as likely to sell maize. Finally, farmers realized gains from arbitrage – conditional on selling, the average sale by a treatment farmer was a month later than that of a control farmer, and was made at a 5 percent higher price.

Intervention

At an initial ROSCA meeting, each ROSCA was read a script about the benefits of setting maize aside after the harvest, of using inputs generally, and of saving. This basic script was augmented for group savings ROSCAs to also explain the group savings intervention. ROSCA members were encouraged to collectively set aside some portion of their harvest, and hold it to sell when prices had risen. ROSCAs were given hermetically sealed storage bags (called Purdue Improved Crop Storage, or PICS bags).

In addition to the bags, farmers were provided a heavily subsidized wooden stand to keep the maize elevated from the ground (and less susceptible to pests and water damage). Finally, ROSCAs were provided logbooks in which the treasurer could keep track of all deposits and withdrawals of maize by individual members. After describing the program, ROSCAs were given a month to think it over. Our field staff emphasized that not all members of a participating ROSCA were required to contribute maize for their ROSCA to qualify for the program.

Experimental design

In July 2015, we conducted a door to door census of 552 individuals in 17 villages spread across three counties in Western Kenya and identified 497 ROSCAs. The census collected basic identifying information about the ROSCA, as well as contact information for ROSCA officials. After identifying this list, we randomly sampled 274 ROSCAs for project inclusion. Enumerators called the treasurers of selected ROSCAs to schedule an initial meeting (at one of the normally scheduled ROSCA meetings).

We randomized ROSCAs into 3 treatment groups: (1) the Group Savings and Reinvestment Account (GSRA), which is the focus of this study; (2) control, and (3) an individual savings account group.

Of the 274 sampled ROSCAs, 163 were successfully reached. Since non-participation occurred before treatment was announced, it should not be possible that treatment affected project participation. However, due to random chance, more GSRA ROSCAs were reachable by phone than the other groups (of the 163 ROSCAs that were traced, 60 were GSRA, 52 were control, and 51 were ISRA). An additional 24 attrited before the intervention, leaving 139 ROSCAs. There are thus more GSRA ROSCAs (53) than control (41) and individual savings (45).

Design details

Randomization method
Randomization was done by Stata, in an office
Randomization unit
the unit of randomization was the ROSCA (Rotating Savings and Credit Association)
Clusters
139 ROSCAs
Observations
786 ROSCA members at baseline
Arms
53 group savings ROSCAs, 41 control, and 45 individual savings ROSCAs
Primary outcomes
storage, maize sales (quantities, revenues, dates, prices), take-up of experimental accounts

Cited by

No citation data available for this registration yet.

Keywords