1

Precision Development (PxD)Jul 2021

Leveraging evidence-based mobile phone communication to improve farmer productivity and environmental resilience in Odisha, India

Prepared for SRP’s “Reimagining opportunities for ICT-driven innovation in agriculture and its transformative potential for rice smallholders” workshop

2

➔ PxD overview

➔ Odisha: Ama Krushi programme

➔ Our impact

➔ The future

3

Mission

We provide actionable information and other scalable services to people in poverty to empower them to sustainably improve their well-being

Vision

An end to information poverty

Goal

Positively impact 100 million of the world’s poor

4

Why

● Smallholder farmers: 70% of world’s poor● Global food demand continues to grow● Climate change, soil and water constraints

● Information exists, but farmers don’t have access

● Mobile phones enable access at scale, and farmer-level data enables customization

Opportunity

What we do: Provide Quality Agricultural Advice

5

Farmer profile information

• Location • Agro-ecological zone• Socio-demographics• Crop variety• Water management

Customized content

● Input recommendations

● Management advice● Market information● Weather-related

contentAgricultural data

• Soil type• Rainfall • Market prices• Pest/disease outbreaks

6

How: Leverage Mobile Phones Through Multiple Channels

● SMS: Regular written messages timed to crop calendar● Voice: Regular push calls timed to crop calendar● IVR: Users pull content from automated menu options● Q&A Hotline: Users pose questions that are answered by

agronomic experts

● Mobile apps: More advanced content (e.g. video, photo, etc.) and apps (e.g. WhatsApp, Telegram, etc.)

● Learning Platforms: Self-directed, interactive content for smartphone users

Two-way communication

Available on demand

Customized

Timed to crop calendar

Cost-effective

Key Features:

Key Elements Differentiating PxD’s Model

7

Low-cost communication

Human-centered design

Customization & targeting

Iteration & learning (A/B tests, RCTs)

Data science

Free to the user

PxD reached 5.0 million users in Q1 2021

8*Graduated users are supported by services that PxD built or contributed to in the past, but which are now managed by partners without PxD’s direct involvement.

4.1 million users reached across active programs in nine countries

0.9 million graduated users*

Founded in 2016 by four co-founders - including professors at Harvard, Chicago, and Brown - with expertise in impact evaluation, business, technology, and agricultural development

200+ employees with mix of technologists, data scientists, agronomists, researchers, and program managers

Global non-profit organization with operations in ten countries in Africa, Asia, and Latin America

9

Average cost per user

PxD’s Costs have Fallen Rapidly over Time

10

➔ PxD overview

➔ Odisha: Ama Krushi programme

➔ Our impact

➔ The future

69% ● Average rice yields in Odisha have been found to be 25% lower than the national average* and only 50% of the potential yields **

● Farmers found to be lacking knowledge on the latest and most appropriate seed varieties as well as on best practices

● For the small fraction of farmers whose knowledge is supplemented with advice, that information may be a) not timely, b) too infrequent, c) not relevant, or d) not trusted.

11

of farmers in the state of Odisha, grow rice in one or more agricultural seasons.

* Das, S. (2012). Rice in Odisha. Metro Manila, Philippines: IRRI** Statistics, D. o. (2015-16). State of Indian Agriculture. New Delhi: Government of India

12

Model

Reach

Content

Partners

Build - Operate - Transfer programme with the State Govt of Odisha (GoO); 2018 - 2021

Serving 1.35 million farmers across all 30 districts in Odisha & growing

Content across 20+ crops, livestock & fisheries

GoO, PxD, BMGF and JPAL-South Asia

Ama Krushi: A customized two-way IVR helpline available to farmers 24x7

13

Engaging with farmers at different touchpoints to maximise relevance of information

14

Farmer characteristics through profiling

(location, dialect, crops); agronomic data (soil,

weather)

Agronomic information (input

recommendations, farm management advice), behavioral economics

Outbound voice calls, SMS & WhatsApp.

Inbound calls via agents & IVR

A/B tests, Randomised Control Trials, Field &

remote polling

More on process: www.precisionag.org

/kharif-in-odisha/

All year round process

15

Kharif: Main growing season

Rabi: Secondary winter season

Content designed in collaboration with local university & dept experts (Weekly Content Review Committee for livestock & agriculture); integrated messaging across different bodies

Farmer engagement data shared weekly (e.g. in Crop Weather Watch Group meetings attended by policy makers at various levels within the government); Participation in government pandemic response

Collaborations with SC & ST department, Pradan (a GoO partner), FARD to serve the diverse needs of farmers and provide targeted outreach (eg: through designated community resource persons)

Ama Krushi training sessions conducted at block-level with extension workers & community leaders (“training the trainer model”); extension officers involved in farmer registration

Capacity Building

Content Design & Validation

Data for Decision-making

Supplementing GoO’s gender and inclusion strategy

16

Institutionalisation within the Department of Agriculture & Farmers’ Empowerment (GoO)

17

➔ PxD overview

➔ Odisha: Ama Krushi programme

➔ Our impact

➔ The future

18

PxD’s Theory of Change: Indicators to Monitor and Evaluate

ACQUIRE/INCREASE

User Engagement

REDUCE

Frictions and Barriers

CHANGE

Farmer Behavior

Registration and referrals

Listening and pick-up rates

Content accessed

Questions recorded for Q&A hotline

User ratings and satisfaction

User comprehension of PAD content

Ability to answer agronomic-related questions

Metrics of aspirations and motivation

Adoption of recommended inputs

Adoption of recommended farming practices

IMPROVE

Farmer/Household Welfare

Crop yields/

Livestock production

Crop losses avoided

Farmer net incomes

Benefit-Cost Ratio

Impact of Digital Agricultural Advisory Services from Rigorous Evaluations

19

4% average yield gains

22% average increase in recommended farming practices

10:1 benefit-cost ratio

Large variation in impact estimates across studies

Source: “Realizing the potential of digital development: The case of agricultural advice.” Science, December 2019.

20

Evidence from other PxD locations

- SMS trials in Kenya show messages

increased knowledge by 21-28% on

use of lime, adoption increased by

20-30% across various trials in

Kenya, Rwanda.

- SMS trials in Kenya find improved

adoption (5-23%) of practices

focused on addressing fall

armyworm (FAW)

Push calls double the percentage of farmers applying recommended quantity of MOP and urea (fertilizers) during basal application

Voice messages increased knowledge (5% relative to control) of improved cotton practices

Audio messages increase the comprehension of redesigned soil health cards (SHCs) by 37% as compared to the control group that only received the SHC

PxD’s own RCT in Gujarat, India

Simple, targeted messages can impact knowledge & adoption of sustainable practices

Flood tolerant (FT) seeds are evidenced to increase farmers’ investment and yields in low lying areas of Odisha, India. Yet, adoption remains low.

Two voice messages focused on highlighting the benefits of FT seeds significantly increased adoption and knowledge among farmers with low land:

● 25% increase in adoption● 7.5% increase in knowledge● Est cost-benefit ratio of 17: 1

○ Estimated income increase of 72 INR per hectare against a marginal cost of 4.1 INR

21

The volume and framing of messages are important for influencing behaviour

Last Rabi season, we tested sending farmers fewer messages. Messages were:1. Focused on high-impact

practices2. Framed using a standardized,

logical template

Preliminary findings:● 5.5 pp increase in pick-up rates● 1.8 pp increase in listening rate● 7.6 pp increase in adoption of

seed treatment

22

23

➔ PxD overview

➔ Odisha: Ama Krushi programme

➔ Our impact

➔ The future

24

PxD’S Future PlansScale and Impact

Reach 100 million users, iterate and improve impact per farmer, and refine evidence base

New Technologies

Communications: WhatsApp, Telegram, photos, videos, chatbots, AR, VR

Data: weather forecasts, remote sensing, satellite, drone imagery, machine learning

New End Users

Extension workers, input suppliers, crop off-takers, etc.

New Sectors

Addressing information poverty more broadly, including education, nutrition, gender, etc.

Commercial Partnerships

PartnershipsPartnershipsFor-profit agro-businesses (without excluding the poorest, or losing farmers’ trust)

Geographic Expansion

Africa (additional states in Nigeria; DRC, West Africa, North Africa, etc.)

South Asia (additional states in India, Pakistan, Bangladesh; Afghanistan, etc.)

Latin America and Caribbean (Brazil, Colombia, Mexico, Peru, etc.)

Southeast Asia (Indonesia, Vietnam, Thailand, Myanmar, Philippines, etc.)

Empowering users with quality information at their fingertips.

precisiondev.org@PrecisionXDev

Annex

27

Current ProjectsReducing Crop Burning in India● Partner: The Nature Conservancy ● Program: Advice to farmers on how to re-use crop

residue instead of burning it

Sustainable Coffee in India● Partners: Coffee Board, Walmart Foundation ● Program: Advice to farmers on environmentally

sustainable coffee growing practices

Climate Smart Agriculture in Colombia● Partner: Rare and The Nature Conservancy● Program: Advice to farmers on how to increase

productivity while adopting more environmentally sustainable farming practices

Reducing Soil Erosion in Rwanda● Partner: One Acre Fund, Rwanda Agricultural Board● Program: Advice to farmers on tree planting and tree

survival, and messages to extension workers to encourage tree distribution

Climate Smart Agriculture in Rwanda● Partner: Root Capital● Program: Advice to coffee farmers on climate smart

agriculture and adapting to climate change

Past Projects

Addressing Climate And Environment

28

Addressing Gender Issues

Background

• 40% of ag labor force in developing countries are women

• Lower use of modern agricultural practices• Lower mobile phone ownership• Lower access to in-person extension

PxD’s Approach

• Developing women-oriented content, e.g. horticulture, livestock husbandry & dairy

• Leveraging existing social networks to reach more women, e.g. self-help groups (SHGs)

• Encouraging joint decision-making in agriculture

Limited evidence on complementary vs. supplementary effects with in-person extension

● PAD’s trial (Cole & Fernando, 2021) do not find significant differences in behavior when digital extension is complemented by in-person training

○ Context: cotton farmers in Gujarat, India, treatment arm attended a single intensive training day○ Research is underway to understand the impact of clubbing digital extension with more frequent

training programs● No difference in impact between fully digital vs. extension worker intermediated

model (Fabregas et al, 2019)● Multiple trials document the “reminder” or “reinforcement” effect

○ SMS reminders boost early repayment of loans by 4 pp in Kenya○ SMS reminder on existing initiative to adopt different bean in Kenya doubled adoption (1.8 pp)

● Account for user heterogeneity, digital divide: service design matters. 29

Influencing factors: frequency, complexity, costs, user types

Diffusion of advice through farmer social networks

30

Evidence from PxD’s work shows some evidence of positive spillovers:

● SMS messages to particular farmers enrolled in farmer groups in Rwanda increased lime adoption for all farmers within the group, including those who did not receive messages. (Fabregas et al, 2019).

● Some evidence of reduced crop losses due to pest attacks (4%) for peers of treated cotton farmers in Gujarat, India (Cole & Fernando, 2020)

In HARIT, 44% of farmers report discussing recommendations from the service with others, almost always with farmers in the same village

31

How PxD has Responded to COVID-19Ongoing operations

Research and data collection

New types of information

● Uninterrupted PxD services, while traditional in-person extension is suspended

● Continuing to add thousands of new farmers every month

● Started two new country programs (Nigeria and Colombia) remotely

Surveys of farmers, agro-dealers, & extension workers in India, Pakistan, & Kenya (results & analysis on website)

Uganda - COVID-related health informationIndia - COVID-related market disruptions:

● Which agricultural activities are exempt from lockdowns● Which markets remain open and which crops are being sold

at what prices● Alternative organic inputs to use while markets are

inaccessible● Organic inputs to use while markets are inaccessible● Storage of crops that cannot be sold at markets

Kenya - education-related information to address school closures

32

Who: Our end users and our partners value PxD’s services

“I find the Krishi Tarang service very useful. It has become a source of timely and correct information on coffee management. This recommendation I think has really helped improve the quality of my coffee this season.”

Mallesh Gowda, Chikmagalur District, Karnataka State, India

“The messages from MOA-INFO have given me instructions that have helped me to increase my yields. Before I started receiving these messages I did not know how to deal with Fall Armyworm, but now I know what chemicals to apply and I have managed to invite over 20 farmers to the platform.”

Javan Akwabi Waliaro, Kenya

33

“PAD has been playing a critical role as a partner for the DoA’s flagship program Extension 2.0. They are helping us to deliver timely, accurate and viable advisory to farmers. PAD is at the forefront for devising effective and efficient communication models.” Sharjeel Murtaza, Department of Agriculture, Punjab, Pakistan

“In a country like Bangladesh, agricultural technologies take a lot of time to reach the fields of smallholder farmers. PAD helps us evaluate the effect our services have on farmers. We highly value PAD’s contribution to our services as now we know better what actually works.” Shah Mushfiq Rahma, mPower Social Enterprises

Who: Our end users and our partners value PxD’s services (cont.)