Precision Agriculture for Development (PAD)

The vast majority of the three billion smallholder farmers and their families worldwide, including 102 million in Bangladesh, live in poverty because of poor agricultural productivity [IFAD, 2014]. Harnessing technology, data science, and behavioral economics, PAD empowers smallholder farmers without access to quality agricultural information by providing them with timely, customized, and actionable advice through voice and SMS messages. To date, we have reached 2.9 million farmers in seven countries, including Bangladesh.

With 40% of the Bangladeshi workforce in agriculture, four-fifths of rural households owning a mobile phone, and farmer yields projected to decrease by 25-50% in coastal districts due to climate change, our digital extension has the potential to complement the government extension service and positively impact millions of people in Bangladesh at low cost. [World Bank Data, 2019, Bayes, 2019, Salehin M. et al., 2018].

An estimated 3 billion people live in smallholder farmer households [FAO, 2015]. In Bangladesh, 102 million people live in rural communities, and 35% of them live in poverty [World Bank Data, 2019]. A major cause of low income in rural communities is poor agricultural productivity. In Bangladesh, the rice yield gap is 50-60%, and is increasing  because of a rapidly changing climate and environmental degradation due to soil salinity and flooding [IFPRI, 2013, Salehin M. et al., 2018, Timsina, J. et al., 2018 ].

Adopting simple, sustainable agricultural practices can increase yields. For example, several agronomic trials have found that appropriate use of agricultural lime [Opala, P. et al., 2018] [Owino, K., 2015] [Pretty, J. et al., 2006], hybrid seeds [Kisinyo, P. et al., 2015], fertilizer [Duflo, E. et al., 2008], pest, disease, and weed management [Das, S.R., 2012], and resource-conserving practices such as reduced or zero-tillage [Duflo, E. et al., 2008] resulted in mean yield increases up to 79%. However, adoption of these agricultural practices remains low due to poor access to quality agricultural information. Taking advantage of high mobile phone ownership in Bangladesh, PAD can disseminate relevant, environmentally-focused advice to dramatically improve yields, income and climate resilience.

Our beneficiaries are smallholder farmers and their households. The majority are poor, with many farming less than 2 hectares and living at subsistence levels. They are often food insecure, and suffer from nutritional problems such as insufficient calories. There are 3 billion target beneficiaries globally, and 1 billion in the seven countries where PAD operates, including 102 million in Bangladesh. Bangladeshi farmers are incredibly vulnerable to climate change, with flooding, poor soil quality, and soil salinity threatening crop yields. However, adoption of sustainable farming practices can mitigate these losses, making them an ideal population to serve. [IFPRI, 2013].

PAD’s mobile advisory service provides timely and relevant information to farmers, customized by crop, soil type, agro-ecological zone, etc. Our content covers topics such as fertilizer, pesticide, and seed variety recommendations, as well as planting and harvest advice, and is developed and tested based on expert agronomist advice and farmer focus groups to determine their most urgent information needs. Our service includes a Q&A feature, which allows farmers to ask agronomists questions, addressing their direct and immediate needs. PAD also conducts A/B tests and farmer surveys to improve our service, making it easier to use and responsive to user feedback.

In Bangladesh, we propose to build a digital advisory system focused on mitigating yield loss caused by climate change and environmental degradation. Our team of agronomists, utilizing the principles of human centered design, will generate a package of practices to address critical challenges confronting farmers. For example, to address soil salinity, we will send messages encouraging farmers to adopt salt-tolerant varieties, such as Boro rice, encourage the adoption of alternative sowing methods, such as sowing on the ridge of a furrow, and message farmers to add nutrients to the soil at optimal times. To mitigate pest-related crop losses, the second largest cause of crop loss in Bangladesh after climate-related shocks, we will recommend sustainable and safe use of pesticides [IFPRI, 2013].    

Building on PAD’s global operational experience, our solution will deliver customized agricultural advice directly to farmers’ mobile phones through a variety of delivery channels, including: 1) regular (i.e. weekly) outbound push voice calls, timed to each stage of the agricultural season; 2) regular outbound push SMS messages; 3) inbound Q&A services where farmers can dial in and pose questions to be answered by agronomists within 24-48 hours; 4) inbound interactive voice response (IVR) hotlines where farmers dial in and pull content from an automated menu options using their keypads; and 5) mobile apps to provide advanced content such as video and photos for smartphone users.

Our solution uses data and rigorous research to deliver continuous improvements in quality and targeting, while maximizing scale and minimizing cost. For example, we use experimental approaches to compare message content and framing, and use datasets from a network of partners (e.g., farmer characteristics, soil fertility, weather, etc.) to customize advice and improve targeting based on user engagement. Leveraging scale, we use big data analytics and machine learning to deepen our ability to customize and target over time. Operating at scale across multiple settings, we can amplify the impacts we deliver to farmers by making our learnings available to the broader agricultural extension community.

While there are over 150 mobile agriculture solutions globally, PAD has achieved both scale (2.9m farmers) and high levels of farmer engagement (>80% user listening rates) through its unique approach combining four components:

1. Design: We deliver simple, effective messages that farmers can understand and act on, drawing on human-centered design to optimize ease of use, behavioral economics to craft messages that influence farmer behavior, and social learning theory to facilitate diffusion of information across farmer networks. Existing mobile advisory services are often not user-friendly, and content is often difficult to understand. 

2. Customization: Our systems allow us to customize messages based on geographic conditions (soil types, agro-ecological zone, etc.), market conditions (input and output prices and availability, etc.), and farmer-specific information (education, experience, risk tolerance, demographics, etc.), to ensure that information is useful, timely and actionable.

3. Evidence: We use the latest research techniques - including RCTs, A/B testing, big data analytics and machine learning (enabled by our scale), and two-way user feedback - to continuously monitor and iteratively improve our impact and cost-effectiveness over time. PAD’ significant expertise in these areas sets us apart, with economists and data scientists in-house, and through collaborations with researchers at leading universities.

4. Scale: By working through partners providing services to farmers at scale - including governments, non-profits, and agri-businesses - PAD’s innovative model can reach scale rapidly, with extremely low customer acquisition costs and rapidly falling costs per farmer per year, from $5 in 2017 to less than $2 in 2019.

While new agricultural practices effectively increase yields, adoption is low, partly due to low access to information. PAD’s digital agricultural extension 1) makes timely and relevant information accessible to a very large number of farmers, leading to high farmer engagement; 2) achieves high farmer comprehension due to simple, easy-to-understand messaging and human-centered design; 3) shifts farmer behavior towards yield-improving farming practices through actionable advice using insights from behavioral economics and social learning; and therefore 4) raises farmers’ productivity and incomes. 

Randomized controlled trials (RCTs) support our theory of change: in India, a mobile phone-based voice service increased comprehension of fertilizer recommendations delivered via Soil Health Cards by four-fold [Cole & Sharma, 2018]; of six studies in East Africa, five found that SMS advice designed by PAD on agricultural lime in areas with acidic soil increased lime adoption by 10-18% [Fabregas, R. et al., 2019]; two out of four RCTs in Kenya and India found yield increases of 8% for cotton farmers, 28% for cumin farmers, and 11.5% for sugarcane farmers [Cole & Fernando., 2018, Casaburi, L. et al., 2014]. 

Our solution is highly cost-effective. We deliver mobile advice at roughly $2-5 dollars (voice) per farmer per year. Estimates from the RCTs above, along with evidence from agronomic trials, suggest benefit-cost ratios of 6:1 to 10:1, accounting for income gains [Fabregas, R. et al., 2019; Cole & Fernando, 2018]. These estimates will increase over time as we absorb the fixed cost and improve our service via iterative testing.

We currently serve 2.9 million people. In one year that will increase to 4.5 million. In five years, we will be serving 10 million people.

In the next year, PAD hopes to expand its farmer reach by more than 50%, from 2.9m to 4.5m, through a combination of incorporating more farmers into our current services across the seven countries in which PAD operates, launching new programs in existing countries such as Bangladesh, and developing new and emerging partnerships in our pipeline in 1-2 new geographies. At the same time, our goal is to improve the quality of our advice through continuous iteration, expand our advisory content to include fisheries, animal husbandry, and new types of crops, and deepen our engagement with our existing farmer populations.

In the next five years, PAD will continue to expand our reach to 10m farmers by developing additional partnerships in our pipeline, increasing our geographic coverage, and continuing to diversify our content. In Bangladesh, our goal is to expand our reach from 50,000 farmers through a partnership with mPower to at least 1 million farmers through partnerships with 2-3 additional government, non-profit, or agri-business partners. We will conduct rigorous impact evaluations across multiple sites to increase the external validity of our research, add to our evidence base, and identify areas of program improvement. We will expand our solution to new channels, such as videos and images accessible through mobile apps to take advantage of rising smart-phone ownership, and provide farmers content in their preferred medium (voice, SMS, app). We have already developed a smartphone app in India and plan to expand to other locations in the future.

Expanding our advisory service in a new context implies large fixed operational costs. The key barriers we identify are:

1. Human resources - In order to operate and scale, we will need to identify a qualified team that can support our expansion. 

2. Logistics and legal - Under our current partnership with mPower in Bangladesh, we operate directly from their office. For this project, we will need to identify our own office space and complete legal registration. 

3. Identification of beneficiaries - The success of our advisory service depends on our ability to identify farmers that could benefit from mobile-agricultural extension. We will need to obtain farmers’ names, contact information, and farmer characteristics to customize content. This access to personally identifiable information poses an additional challenge related to data privacy. 

4. Quality of available local data - Our ability to customize advice depends on the availability, granularity, and quality of input data (e.g., market prices, soil fertility, weather forecast, etc). The cost of obtaining/validating/gathering relevant data is somewhat unpredictable.  

5. Technology - Our solution uses a set of proven technologies to improve farmers’ lives. However, bringing this solution to a new population implies dedicated efforts to adapt or complement our existing technology to ensure that our advisory service is optimally designed for the Bangladesh context and that it covers all the requirements of this population.

1. Human resources - We will hire a diverse, local team of researchers, agronomists, data scientists, and implementation specialists. Additionally, the program will be supported by existing, experienced staff familiar with the local context. 

2. Logistics and legal - We will need to identify our own office space and complete legal registration with the government of Bangladesh. 

3. Identification of beneficiaries - We will develop partnerships with governments, non-profits, telecom companies, and agri-businesses in Bangladesh that work with the farmers we aim to serve. In other geographies, these partnerships have allowed us to obtain databases of farmer contact information. To ensure the privacy of the data we receive, we will use data encryption and two-factor authentication systems.

4. Quality of available local data - we will reach out to a wide network of agencies and organizations that have invested in building relevant databases, including government agencies with data on local markets, local soil quality, and farmer characteristics, agricultural universities and technical organizations conducting agronomic trials in Bangladesh, and other groups working on local weather forecasting models.

5. Technology -  We will work to enhance our existing technology systems that we have built in India, Kenya, and other settings by creating the ability to randomize advice between different farmer groups, tweak content, message and system designs more quickly in response to farmer feedback, and develop different versions of advisory messages for A/B testing.

We plan to bring our mobile-phone agricultural extension service to Bangladesh, reaching 1 million Bangladeshi farmers in the next five years. To achieve this, PAD will 1) identify public and private partners with large farmer databases; 2) establish a quality, diverse team in the country; 3) build and customize our advice to suit local needs; 4) identify and gain access to relevant data (e.g., soil fertility, input availability, market prices) for improved customization; and 5) measure the impact of our service on farmer engagement, input adoption, yield, and income. 

Bangladesh presents a great market opportunity for PAD. Mobile phone ownership is high even among rural households, an already-large yield gap is increasing for wheat and rice due to a rapidly changing climate and environment [Bayes, 2019, Salehin M. et al., 2018], and the ratio of government extension agents to farmers is high at 4500:1 [USAID]. These circumstances underscore an enormous potential for a low-cost, scalable mobile phone-based extension that delivers relevant and dynamic agricultural advice. PAD’s data-driven approach focusing on nimble and iterative development could help increase incomes for millions of smallholder farmers in the country.  Furthermore, the Department of Agriculture Extension (DAE) and Agricultural Information Service (AIS) have expressed interest in our digital extension model.

Our plan builds on our existing relationship with mPower, with whom we have designed and tested agricultural advice sent to 50,000 farmers in Bangladesh, and our experience in other geographies providing advice to millions of farmers facing similar challenges to those in Bangladesh.

We have 180 full-time staff based in Kenya, Ethiopia, India, and Pakistan. We have 12 full-time global staff.

Our team is comprised of economists, data scientists, researchers, and agronomists, and draws from a global pool of talent. Our diversity strengthens our innovative approach, merging economic theory, machine learning, agronomy, evaluation techniques and international perspectives that enable us to provide customized, specific, and easy to understand advice to farmers. 

PAD’s senior leadership team has expertise that draw on years of experience working in and studying agriculture in sub-Saharan Africa and South Asia, business, development economics and behavioral economics, technology, data science, and monitoring and evaluation. PADs co-founders and board members are 2019 Nobel Laureate for Economic Sciences Michael Kremer (Harvard University), Shawn Cole (Harvard Business School), Daniel Björkegren (Brown University), and Heiner Baumann (PAD). PAD is led by Chief Executive Officer Owen Barder, who brings to the organization more than three decades of experience as a development practitioner, scholar and advocate. PAD’s full-time staff represent more than fifteen nationalities and cohere decades of applied experience in developing countries. 

PAD is a registered not-for-profit based in Boston, USA (12 staff), with offices in Kenya (9 staff), Ethiopia (5 staff), Pakistan (4 staff), and several states in India (34 staff). Our projects in Bangladesh, Uganda, and Rwanda are supported by staff in our global and country offices.

We partner with government agencies, private companies, cooperatives, and nonprofits. This includes mPower in Bangladesh, One Acre Fund in Kenya and Rwanda, the government of Odisha, India, Rwanda’s Agriculture Ministry, Kenya’s Ministry of Agriculture, The Nature Conservatory, Agriculture Transportation Agency of Ethiopia, The Uganda Coffee Agronomy Training Program, and the government of Punjab, Pakistan. We are in the process of creating partnerships with dairy companies and cooperatives in Kenya and India, and coffee cooperatives in Karnataka, India. 

Our engagement varies according to the program and partner needs. In some programs, we provide indirect services by supporting partner organizations to improve the services they are delivering to farmers. For example, in Bangladesh, we have been supporting mPower to iterate and improve the delivery of their Agro360 and GeoPotato services. In other settings, we provide direct services, by developing, implementing, and monitoring our own mobile advisory services. In these programs, we find partners that can support our work and allow us to amplify our impact, such as non-profits or governments that are already working with the farmer populations we want to reach. For the Tiger Challenge, we are proposing a direct implementation.

We think the Tiger IT Foundation can help PAD overcome our previously mentioned barriers by providing operational, technological, and legal support as we implement our solution in Bangladesh.