A Randomized Control Trial for Evaluating the Profitability of UDP Technology: A Discussion of the Research Methodology for the RCT
DemandSideConstraintstothe Adoptionof‘UreaDeepPlacement’forRice Production:A ProposedRandomizedControlTrialDr. Andrew Dillon, Michigan State UniversityDr. Lenis Saweda Liverpool-Tasie, Michigan State UniversityDr. Bolarin Omonona, University of IbadanDr. Yemisi Kuku-Shittu, International Food Policy Institute
Introduction• Technology adoption is influenced by manyfactors including theprofitability of the technology itself, but also how farmers learnabout the technology and understand how to use it.• Why don’t farmers use fertilizer?• Liquidity constraints• Farmers report not enough money to buy fertilizer• Uncertainty about returns• Experience and learning about the crop• Negative returns or heterogeneous returns• Complementary inputs• Transaction costs• In this study, we focus on farmer learning and what factors mayinfluence adoption including social networks and farmer riskpreferences building on recent work on UDP by IFDC and IFPRI.
Urea Deep PlacementTechnologyFarmers Fertilizer companies Farm Sector EnvironmentLower productioncosts, higher yields,higher plot levelprofitsProductdiversification,increased profitabilityIncreased riceproductivity, lowernational rice importsReduced nitrogen runoff and volatization• Many potential benefits described in previouspresentations from case studies.• These case studies illustrate important policy implicationsfor smallholders, firms, national agricultural policy and theenvironment.• Benefits only accrue if farmers adopt…..
Key Research Questions1. Does targeting based on social network and/or risk aversionfarmer characteristics increase take-up of UDP?2. What is the average productivity effect of UDP use relative tonon-UDP use among rice producers?• Does this vary by farmer characteristics such as experience,gender, social networks or risk aversion?3. What substitution effects on other inputs or labor areattributable to UDP take-up?
Key Research Questions4. What is the average effect on profits of UDP take-up relativeto non-UDP use among rice producers?• Does this vary by farmer characteristics such as experience,gender, social networks or risk aversion?5. Do differences in initial targeting based on farmer socialnetwork and risk aversion characteristics increase diffusionwithin villages?
TheImportanceofPrivateSectorPartners• Notore, a private fertilizer company, has refined productdelivery of UDP to rural zones.• EX: Repackaging fertilizer with smaller quantities and productsales at different periods, work with smallholders to link tomarkets. Working on innovative training strategies for farmers(time of day, seasonality, etc.)• Key sales force of knowledgeable community sales agent todescribe the new technology to farmers using advancedmarketing tools such as farmer testimonials, demonstrationplots, and extension advice.• Incentives of agents align with need to provide farmers withcomplete information about the product.
TheImportanceof Methodology• Due to many unobservable dimensions of the farmer’s decision totake-up a new technology, comparisons between those who take-upand those who don’t will yield biased estimates.• Existing empirical evidence suffers from problems with:• Omitted or unobserved variables• E.g. endowment effects: farmer’s initialassets/education/experience maybe correlated with higherrisk taking and take-up• Reverse causality• Do adopters gain higher income or do higher income farmersadopt? Or both?• To address these challenges, an encouragement design will be usedwhereby some randomly selected farmers will receive discounts,though all farmers can take-up if they want.
ResearchQuestionsand the Linkto theExperimentalDesignComparison1. Given standard marketing techniques, what is the averageproductivity effect of UDP use relative to non-UDP use among riceproducers? Does this vary by farmer characteristics such asexperience, social networks or risk aversion? A, Control2. Given standard marketing techniques, what is the average effect onprofits of UDP take-up relative to non-UDP use among rice producers?Does this vary by farmer characteristics such as experience, socialnetworks or risk aversion? A, Control3. Does targeting based on social network and/or risk aversion farmercharacteristics increase take-up of UDP and/or productivity? B, C, D, Control4. What substitution effects on other inputs or labor are attributable toUDP take-up?Within villagecomparison offarmers whotake-up and donot take-up ingroups A,B,C,D5. Do differences in initial targeting based on farmer social networkand risk aversion characteristics increase diffusion within villages? A,B,C,D
InnovationsinMeasuringSocialNetworks• Most social network analyses in economics has used a network sample. Someevidence that this leadsto measurement error (Chandrasekhar,2011).• Omission of influential links• This study will take a censuses of all householdsin the community andestablishes how they are linkedbased on several different dimensionsof asocial network.• Type of network: Farmers within village• Type of SN links: Relatives,organizations,plot neighbors, financialties,people with whom they discuss agriculturalissues, friends.• Type of information: Frequency of communication,subject ofcommunication• Information on link: householdcomposition, assets, education
TimelineDate ActivityJuly 2013Sample Frame Enumeration inKwara and Selection of DataCollection FirmSeptember 2013Questionnaire Development andPilotingOctober 2013SN Census and HH Enumerationincluding adoption and input useDecember 2013 Detailed baseline SurveyMay 2014Fertilizer training and offersdepending on village treatmentstatusNovember-December 2014Detailed follow-up Survey +Adoption Survey
Key Questionsfor Discussion• What factors do you think are the most important in thefarmer’s decision to adopt?• Price incentives or other incentives?• What particularities about rice production in Kwara should wetake into consideration?• Is the more relevant stratification upland/lowland orirrigated/non-irrigated? Is there a substantial difference?• Your feedback can really help this research as it is at a earlystage!