NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Disseminating market information via mobile phones to cashew producers: an impact evaluation in Guinea-Bissau
Initial registration date
September 20, 2019
September 24, 2019 11:40 AM EDT
This section is unavailable to the public. Use the button below
to request access to this information.
Bissau Economics Lab
Other Primary Investigator(s)
Bissau Economics Lab
University of Oxford
Ministry of Economy and Finance, Guinea-Bissau
Additional Trial Information
With raw cashew nuts exports accounting for 25% of the GDP and over 90% of Guinea-Bissau’s exports, cashew nuts production represents over 40% of annual income for over half of the country’s population. Recent studies have shown that the farm gate price, at which producers sell their raw cashew nuts, is a critical variable with a huge impact on the country’s GDP growth. Disperse ownership of small and medium cashew plantations across the population makes the impact of farm gate prices on poverty levels even larger. The increased variability of international raw cashew nuts prices in recent years has introduced an important source of inefficiency in this market: accurate information about short-term fluctuations in prices rarely reaches producers. This implies that producers frequently end up selling their cashew nuts at a much lower price than they could, were they to be well informed. We propose a clustered intervention with two different treatments: the promotion and training in the use of a service to sending weekly market and prices updates by mobile phones, both for free and at a small cost for the producers. An improved understanding of market dynamics, combined with better information, should improve farmers’ awareness of feasible prices, improving their capacity to negotiate better deals for their production. Our experimental design will allow us to study the relevance of spillover effects and that of subsidizing the price of the service.
Martínez Zavala, Tatiana et al. 2019. "Disseminating market information via mobile phones to cashew producers: an impact evaluation in Guinea-Bissau." AEA RCT Registry. September 24.
Our intervention seeks to provide timely and accurate market information to cashew producers during their marketing season (usually between March-July). On the day of the baseline survey, after the within-village randomization into treatments, an enumerator will explain the producer the characteristics and timing of the treatment he or she has been assigned to. During the cashew campaign, March to August, the selected producer will receive price information and market advice on their phones (for free, or at a small fee).
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Increased farm gate prices.
Primary Outcomes (explanation)
Primary outcome: Increased farmgate prices. Selling cashew nut production at a higher price should directly imply higher household revenue in this context, as we expect a null or very small effect of improved market information and higher prices on quantities sold (to be explored in the analysis). A key feature of our intervention is that will provide information on the more appropriate time to sell (which should allow for some temporal arbitrage). With high poverty incidence and revenues from cashew amounting to approximately 40% of total household income for cashew producers (WB 2017), we expect a significant impact of this increased revenue on important economic outcomes associated to improved household welfare.
Secondary Outcomes (end points)
Improved economic information and market understanding and higher household welfare.
Secondary Outcomes (explanation)
Secondary outcome: Improved economic information and market understanding. Together with measuring the use of mobile market information service (through, for example, the number of messages listened, subscriptions made), the information outcome will be measured with an index constructed from a set of questions related to market understanding, including the state of the cashew season and prices (both practiced and expected). Tertiary outcome: Higher household welfare. Higher food security of the household and improved access to education and health (including psychological health).
Sampling: we plan to use as sampling frame the latest available census data from the National Institute for Statistics, together with a comprehensive list of geo-referenced villages and the latest World Bank survey on cashew producers.
a. We will randomly select 282 villages from the sampling frame, trying to maximise distance between any two randomly selected villages.
Screening: Upon arrival to a selected village, the data-collection team will check with the village leader the randomly re-ordered list of households to identify the first up to 20 households that own a cashew plantation, our key screening condition.
The data-collection team will visit and seek consent to participate in the study from the first 8 (4 in control villages) households identified in the list of the -up to- 20 potential participants, and replace any household that actually does not meet the full list of eligibility criteria (see below) with one of the -up to-12 replacement households.
Eligibility mechanism: One by one, each household who has been randomly selected as a potential candidate for the study will be visited by an enumerator (until enough eligible households from the randomly generated list are found in the corresponding village). The enumerator will briefly explain the purpose of her visit and check that the following eligibility conditions are satisfied: 1. The household owns at least one cashew plantation (we know that about 50% do). 2. The household head or, alternatively, someone with good-enough knowledge of and involvement in the family’s cashew business (cultivation, collection and trading) who lives permanently in the house is present. 3. Someone permanently living in the house owns a well-working mobile phone, charged, with signal at home and knows how to use it. The enumerator will check the phone works correctly by calling the number and by sending a text message and reading it herself. If these conditions are met, the enumerator will explain the candidate the basic elements of the study, including that a within-village lottery determining who will participate in the program will take place. If the candidate agrees to participate, he will be asked to sign a consent form. Those who consent will then complete a baseline survey.
The last stage of the randomization will take place on the spot, within-villages. All those individuals who granted consent and are assigned to a treatment will be informed of their selection into the program by the enumerator, who will explain the corresponding treatment during a training session lasting approximately 1 hour and 30 minutes.
Experimental Design Details
With households as the unit of observation, the randomization will be done in three stages, following the next steps: 1. A random selection of 282 (+ replacements) villages/neighborhoods (geographical clusters) across all regions of Guinea-Bissau (using a combination of data sources, including the 2009 census the registry of geo-coded villages from the ministry of education and possibly the latest WB cashew survey) and randomization of treatments at the cluster level: 94 of these villages will be allocated to the control group and 94 to each of the two treatment arms. 2. Within-village random selection of households to be contacted to check for their eligibility (combining a pre-visit randomization with a list provided by the village leader). These households are visited in person by enumerators, who check for eligibility, register consent for participation in this study and complete the baseline survey for the required number of eligible households: 4 per “control village” and 8 (4 treated + 4 untreated) per “treated village” (details below). 3. Once households complete the baseline surveys, an on-the-spot within-village lottery (with each farmer drawing one of two kinds of goodies from a bag) among eligible households will randomly determine 4 treated and 4 non-treated households in “treated villages”. The treated farmers will then receive the training about the service. Stratification by literacy seems a potentially important element for the randomization, either at the household or village level. In control villages no lottery will take place.
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
- 94 villages with 376 households control.
- 94 treated villages with 376 treated households and 376 untreated households, treatment 1.
- 94 treated villages with 376 treated households and 376 untreated households, treatment 2.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
The main detectable effect size for the main outcome: the difference in the average selling price, with an estimated average of 500 FCFA/kg and standard deviation 100 FCFA/kg, is 30 FCFA. In percentage, compared to the average, this is a 6% of the expected average price, and 30% of its expected standard deviation.
INSTITUTIONAL REVIEW BOARDS (IRBs)
Partnership for Economic Policy (PEP)
IRB Approval Date