Increasing Malawian Smallholder Farmers Access to Improved Storage Technology and Credit

Last registered on August 19, 2021

Pre-Trial

Trial Information

General Information

Title
Increasing Malawian Smallholder Farmers Access to Improved Storage Technology and Credit
RCT ID
AEARCTR-0008077
Initial registration date
August 16, 2021

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
August 19, 2021, 10:40 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Region

Primary Investigator

Affiliation
Purdue University

Other Primary Investigator(s)

PI Affiliation
Purdue University

Additional Trial Information

Status
Completed
Start date
2018-04-24
End date
2019-06-30
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Seasonal commodity price fluctuations can potentially offer farmers inter-temporal arbitrage opportunities to increase their income. However, smallholder farmers in most of sub-Saharan Africa often do not exploit these opportunities to the fullest extent possible. We administered a randomized control trial (RCT) among 1,739 smallholder farmers in Malawi to evaluate the impacts of storage and commitment constraints on farmers’ legume storage decisions (soybeans and groundnuts). The treated groups received (i) an improved storage technology in the form of two hermetic (airtight) bags (T1: technology only); (ii) the same improved storage technology under the condition that farmers store collectively with their farmer club members in their village, (T2: village storage program) and (iii) the improved storage technology under the condition that farmers store collectively at a centralized association warehouse (T3: warehouse storage program). We analyzed the impacts of these treatments on the following outcomes: quantity stored, number of weeks stored before the largest sale, net sales quantity and net sales revenue. Our results showed that all three storage interventions helped farmers store more legumes at harvest, store longer, and increase revenue from legume sales compared to control households. Our results also showed that the village storage program was relatively more effective at incentivizing farmers to store legumes compared to the warehouse storage program. The finding suggests that providing improved storage technologies and encouraging farmers to store with others in their village may be more effective strategies for them to capture intra-seasonal arbitrage opportunities than promoting larger-scale warehouse receipt-type systems.
External Link(s)

Registration Citation

Citation
Nindi, Tabitha and Jacob Ricker-Gilbert. 2021. "Increasing Malawian Smallholder Farmers Access to Improved Storage Technology and Credit." AEA RCT Registry. August 19. https://doi.org/10.1257/rct.8077-1.0
Experimental Details

Interventions

Intervention(s)
Our intervention included three treatments, described below. Treatment assignment was random, made at the village (club) level, and stratified by NASFAM association. A control group and three treatment groups including: (I) the storage technology-only treatment, (ii) the storage technology + village storage treatment, and (III) the storage technology + warehouse storage treatment.

The physical storage technology (Treatment 1)
In treatment 1 (T1: technology intervention), households were trained about the hermetic storage technology and given two 100-kilogram bags for free. The hermetic bags, called the Purdue Improved Crop Storage (PICS) bag is a 3-layer airtight storage bag that effectively protects grain from pests and molds without the use of chemicals, simply by hermetically sealing its contents. The PICS bags have proved to be effective at storing maize as well as legumes including cowpeas, soybeans and groundnuts (Baributsa et al., 2017; Sudini et al., 2015; Williams et al., 2014). The treatment was designed to help smallholder farmers overcome the storage technology constraint they face from insects and molds.

We chose to provide only two 100-kilogram bags to avoid creating an incentive for sharing bags across households, which could result in treatment spillover or contamination. However, the two 100-kilogram bags allowed farmers to effectively store a substantial share of the average harvest for legumes, which was 520 kg at baseline. The training included in this treatment informed smallholder farmers about the benefits of using PICS bags, as well as the prospects it presents for exploiting seasonal price arbitrage opportunities. This treatment was, therefore, expected to help reduce the expected quality and quantity losses for farmers and thus induce them to store more at harvest, so that they could sell good quality grain at a higher price later in the year.

The village storage program (Treatment 2)
In Treatment 2 (T2: The technology + Village group storage arrangements), households received the same training and two 100-kilogram PICS bags provided in T1 and agreed to store their legumes with fellow club members within their villages. To receive the bags, participants had to agree to also participate in the village storage program. Each club selected a stock-keeper responsible for the club’s stocks based on trust and storage ability (i.e. enough and secure space to store all members’ grain). This treatment was designed to help farmers overcome the storage technology constraint as well as the behavioral challenge associated with individual storage of grain in homes where farmers often face social pressure to share, along with impatience and limited self-control problems (Aggarwal et al., 2018; Ashraf et al., 2006; Baland et al., 2011; Brune et al., 2011).

The group storage arrangement allowed farmers to separate and deposit part of their grain stocks in a club-managed stock that was stored away from home for liquidation when prices rise. Each club independently agreed on storage length, a reservation price, and procedures for early grain withdrawal, which included getting the club’s consent and/or a penalty. Farmers may have been influenced to store longer through village group storage arrangements than they would have on their own. In addition, the amount of grain deposited into the group stocks by an individual farmer is likely to be influenced by his or her peers in the group depending on the groups’ anticipated gains of storage. Given self-control and other problems that may influence farmers to liquidate stocks early, we designed this storage intervention to understand if group storage arrangements implemented locally within the village with a relatively small number of other farmers would induce people to store more grain at harvest.

The warehouse storage program (Treatment 3)
In Treatment 3 (T3: The technology + Warehouse group storage arrangements), farmers received the same training and two 100-kilogram PICS bag given to households in T1, as well as an invitation to participate in a group storage arrangement. The group storage arrangements different from those in T2 in three ways. First, farmers in T3 received some information on financial management. We provided farmers information about the benefits of storing grain (a form of savings) and strategically marketing their products to exploit better prices. Second, storage was at centralized NASFAM warehouses within Group Action Centres (i.e., at the community level) rather than within the farmers’ village. Unlike the village storage program, this implied that more than one club stored in each centralized warehouse (i.e. between 5 to 10 farmers stored together per club for T2, while between 10 to 15 clubs stored together per warehouse in T3, with 5 to 10 farmers per club). Third, clubs using the same warehouse were required to synchronize their grain deposit and withdrawal conditions, which were more stringent than the village storage program’s. This requirement stemmed from the intended use of the stored crop as a guarantee for the loan and the standard loan repayment conditions.

The warehouse storage locations used in this treatment arm had the disadvantage of being much further away from the villages than storage locations in T2 (eg: 10 to 35 km away in T3, versus 1 to 5 km away in T2), and required smallholder to store with a larger group of people from a wider geographic region, with whom they may have had fewer social connections. However, the benefit of storing at a larger warehouse with more farmers in T3 was that this treatment helped farmers assemble their legume for easy off-taking by big traders and processors. This potentially facilitated more trading opportunities at potentially higher prices for participants. This larger collectivization by farmers in T3, potentially increased their bargaining power for higher prices compared to the more localized village collectivization that occurred in T2.

Control group
The control group included farmers that did not receive any treatment but resided in the same area as treated farmers and were also members of NASFAM clubs. Farmers in the control group were included in all follow-up data collection efforts throughout the intervention period. The farmers in this group were asked whether they purchased PICS bags on their own before the baseline, and whether they stored their legumes in groups. Only 12 households in the control group reported having bought PICS bags, with the number of bags bought per household ranging from 1 to 10 bags. All 12 of these households reported storing maize in their PICS bags and not legumes, which does not bias our impact estimates.
Intervention Start Date
2018-04-25
Intervention End Date
2018-05-31

Primary Outcomes

Primary Outcomes (end points)
The main outcomes of interest in this study are:
i) households’ quantity of legumes stored at harvest,
ii) number of weeks stored before the largest legume sale,
iii) total sales revenue from legumes,
iv) quarterly legume inventories,
v) net sales quantities, and
vi) net value of legume sales.
Primary Outcomes (explanation)
The main outcomes of interest in this study are: i) households’ quantity of legumes stored at harvest, ii) number of weeks stored before the largest legume sale, iii) total sales revenue from legumes, iv) quarterly legume inventories, v) net sales quantities, and vi) net value of legume sales. Legume inventories represent the total household’s inventories of legumes in a given quarter including legumes stored at home plus with the group. The net sales quantity is the difference between quantity sold and quantity purchased in a given quarter. Net value of sales is the value of legume sold minus the value of legumes purchased in every quarter.
We estimate treatment effects on outcomes measured both yearly and quarterly. Yearly treatment effects are measured for outcomes that we measured twice, one year apart, in the baseline and in the third follow-up survey: quantity stored at harvest (kg), number of weeks stored before largest sale and total sales revenue (MK). We use an ANCOVA specification to estimate intention to treat (ITT) effects on outcomes measured yearly (McKenzie, 2012)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Like most sub-Saharan African countries, Malawi has an active network of smallholder farmer organizations. As mentioned in the introduction, we worked with members of the National Smallholder Farmers’ Association of Malawi (NASFAM), a farmer-based organization with membership throughout the country. NASFAM has 43 Farmer Associations across Malawi. An average NASFAM Association covers an entire EPA, which typically comprises multiple communities. In each Farmer Association, NASFAM is organized in Group Action Centers (GACs), which generally match the community or Section level. The distance between these communities or sections range between 10 and 35 kilometers. On average, NASFAM Associations count about 21 GACs each, and each GAC counts about 15 farmer clubs each. A club is made of about 10 farmers who reside within the same village; villages are between 1 to 8 kilometers distance from each other. Although villages that fall within the same community are very similar, they are sufficiently far apart to limit possible treatment contamination.

Three NASFAM Farmer Associations were randomly selected for the study: Chioshya, Mikundi and Mpenu. Since the clubs in each Farmer Association are grouped into GACs, we randomly selected 12 GACs in each of the targeted Farmer Associations. Then, within each of the selected GACs, we randomly selected 12 clubs. Since our main study focus was on legumes, we excluded farmers that did not plant legumes in the 2018 cropping season before sampling.

We implement a clustered RCT with three treatments. Our intervention included three treatments, described below. Treatment assignment was random, made at the village (club) level, and stratified by NASFAM association. A control group and three treatment groups including: (I) the storage technology-only treatment, (ii) the storage technology + village storage treatment, and (III) the storage technology + warehouse storage treatment.

The physical storage technology (Treatment 1)
In treatment 1 (T1: technology intervention), households were trained about the hermetic storage technology and given two 100-kilogram bags for free. The hermetic bags, called the Purdue Improved Crop Storage (PICS) bag is a 3-layer airtight storage bag that effectively protects grain from pests and molds without the use of chemicals, simply by hermetically sealing its contents. The PICS bags have proved to be effective at storing maize as well as legumes including cowpeas, soybeans and groundnuts (Baributsa et al., 2017; Sudini et al., 2015; Williams et al., 2014). The treatment was designed to help smallholder farmers overcome the storage technology constraint they face from insects and molds.

We chose to provide only two 100-kilogram bags to avoid creating an incentive for sharing bags across households, which could result in treatment spillover or contamination. However, the two 100-kilogram bags allowed farmers to effectively store a substantial share of the average harvest for legumes, which was 520 kg at baseline. The training included in this treatment informed smallholder farmers about the benefits of using PICS bags, as well as the prospects it presents for exploiting seasonal price arbitrage opportunities. This treatment was, therefore, expected to help reduce the expected quality and quantity losses for farmers and thus induce them to store more at harvest, so that they could sell good quality grain at a higher price later in the year.

The village storage program (Treatment 2)
In Treatment 2 (T2: The technology + Village group storage arrangements), households received the same training and two 100-kilogram PICS bags provided in T1 and agreed to store their legumes with fellow club members within their villages. To receive the bags, participants had to agree to also participate in the village storage program. Each club selected a stock-keeper responsible for the club’s stocks based on trust and storage ability (i.e. enough and secure space to store all members’ grain). This treatment was designed to help farmers overcome the storage technology constraint as well as the behavioral challenge associated with individual storage of grain in homes where farmers often face social pressure to share, along with impatience and limited self-control problems (Aggarwal et al., 2018; Ashraf et al., 2006; Baland et al., 2011; Brune et al., 2011).

The group storage arrangement allowed farmers to separate and deposit part of their grain stocks in a club-managed stock that was stored away from home for liquidation when prices rise. Each club independently agreed on storage length, a reservation price, and procedures for early grain withdrawal, which included getting the club’s consent and/or a penalty. Farmers may have been influenced to store longer through village group storage arrangements than they would have on their own. In addition, the amount of grain deposited into the group stocks by an individual farmer is likely to be influenced by his or her peers in the group depending on the groups’ anticipated gains of storage. Given self-control and other problems that may influence farmers to liquidate stocks early, we designed this storage intervention to understand if group storage arrangements implemented locally within the village with a relatively small number of other farmers would induce people to store more grain at harvest.

The warehouse storage program (Treatment 3)
In Treatment 3 (T3: The technology + Warehouse group storage arrangements), farmers received the same training and two 100-kilogram PICS bag given to households in T1, as well as an invitation to participate in a group storage arrangement. The group storage arrangements different from those in T2 in three ways. First, farmers in T3 received some information on financial management. We provided farmers information about the benefits of storing grain (a form of savings) and strategically marketing their products to exploit better prices. Second, storage was at centralized NASFAM warehouses within Group Action Centres (i.e., at the community level) rather than within the farmers’ village. Unlike the village storage program, this implied that more than one club stored in each centralized warehouse (i.e. between 5 to 10 farmers stored together per club for T2, while between 10 to 15 clubs stored together per warehouse in T3, with 5 to 10 farmers per club). Third, clubs using the same warehouse were required to synchronize their grain deposit and withdrawal conditions, which were more stringent than the village storage program’s. This requirement stemmed from the intended use of the stored crop as a guarantee for the loan and the standard loan repayment conditions.

The warehouse storage locations used in this treatment arm had the disadvantage of being much further away from the villages than storage locations in T2 (eg: 10 to 35 km away in T3, versus 1 to 5 km away in T2), and required smallholder to store with a larger group of people from a wider geographic region, with whom they may have had fewer social connections. However, the benefit of storing at a larger warehouse with more farmers in T3 was that this treatment helped farmers assemble their legume for easy off-taking by big traders and processors. This potentially facilitated more trading opportunities at potentially higher prices for participants. This larger collectivization by farmers in T3, potentially increased their bargaining power for higher prices compared to the more localized village collectivization that occurred in T2.

Control group
The control group included farmers that did not receive any treatment but resided in the same area as treated farmers and were also members of NASFAM clubs. Farmers in the control group were included in all follow-up data collection efforts throughout the intervention period. The farmers in this group were asked whether they purchased PICS bags on their own before the baseline, and whether they stored their legumes in groups. Only 12 households in the control group reported having bought PICS bags, with the number of bags bought per household ranging from 1 to 10 bags. All 12 of these households reported storing maize in their PICS bags and not legumes, which does not bias our impact estimates.
Experimental Design Details
Randomization Method
Random assignment of clubs to treatment group done in office by computer while random selection of farmers within club done using coin flip in presence of club in the field.
Randomization Unit
Farmer clubs or groups of the National Smallholder Farmers Association of Malawi (NASFAM) which are at village level
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
A total of a total of 300 farmer clubs with 75 clubs per experimental arm and 5 households/farmers per club
Sample size: planned number of observations
1500 households or farmers in total
Sample size (or number of clusters) by treatment arms
75 clubs per experimental arm and 5 households/farmers per club with a total of 300 farmer clubs
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Power calculations indicated that 75 clubs per experimental arm and 5 households per club would provide a minimum detectable effect of 0.33 standard deviations in outcomes comparison between two arms of the experiment. This is between what is generally considered small and medium effect size (Duflo et al., 2008). Calculations were based on an intra-cluster correlation coefficient (ICC) of 0.1, 80 percent power, and a 95 percent confidence level. Estimates of the intra-cluster correlation coefficient, means and standard deviations for our outcomes were based on calculations using the World Bank’s 2015/16 Living Standard Measurement Survey data (agricultural survey) for Malawi. In order to account for possible attrition, we aimed to include 85 clubs or clusters per experimental arm.
IRB

Institutional Review Boards (IRBs)

IRB Name
Purdue University HRPP IRB
IRB Approval Date
2018-04-24
IRB Approval Number
Protocal No. 1802020251

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
Yes
Intervention Completion Date
June 30, 2019, 12:00 +00:00
Data Collection Complete
Yes
Data Collection Completion Date
June 30, 2019, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
377 farmer clubs
Was attrition correlated with treatment status?
Yes
Final Sample Size: Total Number of Observations
1739 farmers households
Final Sample Size (or Number of Clusters) by Treatment Arms
377 farmer clubs. Of the 377 clubs (1,739 farmers) sampled to be part of the study, 103 clubs (540 farmers) were assigned to the control group, 85 clubs (387 farmers) were randomly assigned to the technology-only treatment, 89 clubs (389 farmers) were randomly assigned to the technology + village storage treatment, and 100 clubs (423 farmers) were randomly assigned to the technology + warehouse storage treatment. We show in the results section that the random assignment was balanced along a large number of group and farmer characteristics at baseline (F-test that all coefficients in a multinomial probit model are jointly equal to zero=86, p=0.126).
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials