OpenIDEO is an open innovation platform. Join our global community to solve big challenges for social good. Sign Up / Login or Learn more

Experience Pool: Crowd-sourcing farmer feedback about farm inputs' quality

Survey farmers about farm inputs' quality & widely disseminate resultant quality ratings, using incentivized locals to run back-end calls.

Photo of Ali Hasanain

Written by

EXPLAIN YOUR IDEA

Developing country farmers are often faced with poor quality farming inputs, or inputs whose quality is simply unknown to them. Although many development practitioners had tackled this problem in the past, an effective and practical solution had been elusive. In a recently concluded pilot project, our academic team partnered with the Pakistani government to proactively collect and disseminate quality data to farmers, relying on a centralized call centre to conduct farmer interviews after initial data collection by government agents equipped with smartphones. Not only did the project significantly improve farmers’ decisions regarding inputs, decreasing wastage and improving effectiveness, it also proved to be relatively cheap and scalable, with a 300% return on investment. Although economically viable, this previous pilot incurred the monthly expense of the call centre. For severely resource-constrained governments like Pakistan’s, it is important to lower recurring costs to reduce the fragility of the project in the face of future financial constraints. In order to do this, we intend to make one major modification in the existing program: we will incentivize other farmers located in the region to take out some time to conduct phone interviews. This will reduce costs, create work for locals during inactive periods (much like Uber), and the decreased social distance between the caller and the farmer will likely yield better information for our ratings platform.

WHO BENEFITS?

Primary beneficiaries are: a) farmers, especially those with small holdings, who purchase inputs from suppliers generally, and in the case of our pilot, from the artificial insemination market; and b) input providers who have hitherto been unable to distinguish themselves as being high-quality. Secondary beneficiaries are locals who conduct back-end calls for a piece rate. For more, please see the User Experience Map uploaded as a PDF attachment below.

WHERE WILL YOUR IDEA BE IMPLEMENTED?

Our initial tests will build on our previous pilot in Sahiwal district of Punjab Province in Pakistan, which is known for a large livestock stock. We will seek to expand into other markets and regions if the tests prove successful.

ARE YOU IMPLEMENTING IN AN ELIGIBLE COUNTRY?

  • Yes

EXPERTISE IN SECTOR

  • I’ve worked in a sector related to my idea for over a year

EXPERIENCE IN IMPLEMENTATION COUNTRY(IES)

  • Yes, for more than one year.

TELL US MORE ABOUT YOU!

I am a research fellow at Princeton, and an Assistant Professor of Economics at the Lahore University of Management Sciences, Pakistan. This project is in collaboration with Arman Rezaee, an Assistant Professor at UC, Davis. We have implemented tech solutions to dev problems in Pakistan for 6 years.

IS THIS IDEA NEW FOR YOU OR YOUR ORGANIZATION?

The idea of crowd-sourcing farmers' opinions about inputs, and then disseminating aggregated quality ratings to future customers of the input, is not new to us as a research team. Arman Rezaee and I, along with our co-author Yasir Khan and a group of bright research assistants, have worked with the local government to develop, pilot and then implement this idea locally, with great success. The additional idea of moving from a professional call-center to distributed locals incentivized to conduct these calls in their free time, is new to us.

HOW IS YOUR IDEA UNIQUE?

Many organizations are working to provide farmers with actionable information; early research demonstrated the value of cellphones in allowing Indian fisherman and Niger farmers to trade more efficiently. However, some initiatives to provide price, weather and crop advisories, such as Reuters Market Light in India, failed to impact farm results. We speculate that this may have been due to the information being insufficiently localized. In contrast, our implementation of a platform for crowd-sourcing and aggregating farm input quality information proved highly successful in the pilot we have completed. We believe that this may be due to the highly localized and economically relevant information being provided. Our divergence from other initiatives is in seeking to use distributed and part-time calling agents for our backend. We bring two advantages to bear on this effort: 1) having an existing platform allows us to test our innovation with minimal overhead effort in creating the information-sharing audience. 2) our existing experience strengthened our ability to evaluate the likely usefulness to farmers of information being shared on the platform.

WHO WILL IMPLEMENT THIS IDEA?

Arman Rezaee and I will implement the idea in close partnership. I will manage our relationship with a technology partner to develop our existing software platform further in the way proposed, and will take the lead in working with the Punjab Government, our close partner in implementation. Arman will play the lead on data analytics. Importantly, he will also lead our effort to calibrate the correct incentive levels to induce backend callers to conduct regular and high quality calls.

HOW HAS YOUR IDEA CHANGED BECAUSE OF BENEFICIARY FEEDBACK?

The input quality side of the project had already benefited from extensive beneficiary feedback. We focused here on studying the proposed volunteer caller. We surveyed 28 poor, rural respondents selected in a convenience sample. We described the project, focusing on the possibility of them conducting phone interviews with local farmers following a pre-existing script. We asked them about their willingness to use the app and conduct these calls. We got an overwhelmingly positive response, with over 85% (24 out of 28) claiming they would be interested in participating in the project, would install an app to do so, and be willing to undergo training. We were surprised that only 5 out of 24 were willing to volunteer their free time to make calls for free. We had anticipated being able to appeal to individuals' desire to see others' benefit much more. However, 22 out of 24 were willing to work in exchange for phone credit and every single respondent was willing to work in exchange for mobile money. Hence, our project now emphasizes incentivizing callers through piece-rates much more than previously. We will rely less than originally expected on callers volunteering time.

WHAT ARE SOME OF YOUR UNANSWERED QUESTIONS ABOUT THIS IDEA?

The first unanswered question is: what is the efficient piece-rate to offer callers in order for them to complete a call? Offering a low rate will likely reduce call quality and the app's success amongst callers. Offering a higher-than-necessary rate will be needlessly costlessly and reduce the ability to expand. Second, we don't know how well these amateur calling agents will conduct calls. Will farmers be more likely to hang up or stop app usage if the person on the other end of the line doesn't sound as professional as before? Third, we don't know how long we will retain the average caller. High turnover would make quality training less efficient. The reverse would also be true.

WHY DO YOU THINK THE PROBLEM YOUR IDEA SOLVES FOR HASN'T BEEN SOLVED YET?

Tech - Until recently, smartphone penetration in Pakistan's rural areas was negligible. Only since late 2014 have low-end smartphones in Pakistan dropped into the $60-80 range. The cheapest ones today are as low as $40. This means that the pool of people we can now tap into to make phone calls to farmers is much larger than in the past. Slow increase in farmer adoption of cellphone services - Most farmers in rural countries are still only adopting their first feature phones. Studies show that cellphone use increases familiarity with alphanumeric characters and literacy, but it takes time to move from adoption of basic cell functions to complex IVRs and Robocalls.

WHAT WOULD YOU ULTIMATELY LIKE TO ACHIEVE WITH THIS IDEA? WHAT IS YOUR NEXT STEP TO GET THERE?

If the pilot proposed here is successful, the next step is to scale it to ratings of other farm inputs, such as seeds, fertilizers and pesticides (which all suffer from tampering/adulteration in developing countries). Ultimately, the goal is to become a general product rating system with farmer-requested product categories that enables high-quality inputs to increase their market share thus generating increases in agricultural productivity in poor countries with low-capacity governments.

MEMBERS OF MY TEAM HAVE BEEN WORKING TOGETHER FOR:

  • More than a year

MY INTENDED BENEFICIARIES ARE:

  • Within 50 km of where our team does most of its work
  • Within 100 km of where our team does most of its work
  • Within in 500 km of where our team does most of its work
  • More than 500 km from where our team does most of its work

MY ORGANIZATION'S OPERATING BUDGET FOR 2015 WAS:

  • Between $100,000 and $500,000

Farmers are often faced with the problem of having to use a good or service before its quality is known. Was this seed better than the competition? Is that pesticide effective in my conditions? Farmers typically get one outcome data point for a complex combination of multiple inputs and other conditions. The inability to make the right choices about what input to use causes heavy losses in poor countries' agricultural sectors.

In developed countries, this fundamental problem is solved through experimental research and regulations controlling quality. In underdeveloped countries, the institutional capacity to adopt either of these solutions is typically lacking. Instead, farmers rely on their own experiences and what they learn through word of mouth. This data is usually both noisy and lagged.

Research in developed country markets has demonstrated that quality and other outcomes are substantially improved by the implementation of ratings systems. In underdeveloped countries, past attempts to replicate such systems have either been prohibitively costly, or simply not worked.

Our core solution to this problem was to create a Yelp-like system that solicited feedback proactively from farmers, aggregated and processed that information, and provided it back to the farming community as a ranking of the quality of service providers. Farmers could then use this information to switch suppliers or pressurize their existing supplier to improve the quality of supply.

Importantly, we ensured that the ratings information was truthful in two ways: (a) we conducted spot-checking of a sample of the reported information by reaching out directly to farmers and asking them to verify the information we had received; and (b) we did not start disseminating information until we had had the chance to build a large pool of responses, such that individual variations in ratings did not affect the overall rankings too much. 

Our pilot successfully demonstrated, in the Artificial Insemination of large ruminants, that such a system improved the service provision for farmers (the rate of successful pregnancies following Artificial Insemination rose 27%) , and thus their economic outcomes (we demonstrated a $32 per month increase in household income for farmers living on approximately twice that amount).

This heavily managed system was an initial proof of concept. Although it had highly positive rates of return, it would rely in the long-term on unreliable government funding. We now plan to pursue a modification of our original project design to lower costs and decrease dependence on public financing.

We intend to create an app that invites distributed smartphone users proximate to our farmers to conduct interviews with them. If viable, we will also create monetary incentives for these recruits (members of our team have worked in the past on optimizing incentives in real-time in a similar context).

Recruits will see available calls and be paid, possibly through mobile money, to make those calls.  Before the calls are made, the participant will be provided training to ensure they conduct high quality phone interviews. After the calls are made, we will implement random validation of call quality either by administrators or other calling agents.

In the long-term, we want to implement a more general ratings system where user-created ratings are systematically collectivized, and review requests are endogenously generated. Smartphone prices are falling rapidly, so it is now viable for individuals in even very remote areas to download and utilize an app that potentially pays them for their services.


View more

Attachments (5)

App screenshots.docx

We used this as additional material during our survey, to show respondents the type of app they may need to interact with.

Survey instrument.docx

We used this questionnaire in conducting our beneficiary survey.

User Experience Map.pdf

This User Experience Map documents the experience of a caller making phone calls for the program through the proposed app. It only indirectly discusses the experience of farmers who will benefit from the extra information. Details of that experience are already documented in the workflow images uploaded above.

264 comments

Join the conversation:

Comment
Photo of Dr. Sukumar Kar
Team

Food security refers to the availability of food and one's access to it. The World
Health Organization defines three facets of food security: food availability, food
access, and food use. India faces a threefold challenge to achieve food security; to
match the rapidly changing demand for food from a larger and more affluent
population to its supply; do so in ways that are environmentally and socially
sustainable; and ensure that the poorest people are no longer hungry. This challenge
requires changes in the way food is produced, stored, processed, distributed, and
accessed. Increase in production will have an important part to play, but they will be
considered as never before by the finite resources provided by Earth’s land, oceans and
atmosphere. Prevention of postharvest loss is increasingly cited as a means to
effectively contribute to available food supplies.

View all comments