Picture-based crop insurance: using farmers’ smartphone pictures to reduce basis risk and costs of loss verification

3ie evidence programme:  Agricultural Insurance Evidence Programme
Author(s): Francisco Ceballos, Berber Kramer, Azad Mishra, Miguel Robles, Mann S Toor
Institutional affiliation(s): International Food Policy Research Institute (IFPRI), HDFC Ergo, World Bank, Borlaug Institute of South Asia
Grant-holding institution: International Food Policy Research Institute (IFPRI)
Main implementing agency: International Food Policy Research Institute (IFPRI)
Sex disaggregation: Yes
Gender analysis: Yes
Equity focus: Yes
Study type: Formative evaluation

Context

Farmers in India are increasingly exposed to climate change and natural disasters, causing extreme hardship. Anticipating the possibility of such calamities, farmers underinvest in productivity-enhancing technologies. Policymakers hence seek ways to improve farmers’ resilience, and delivery of risk management strategies, including crop insurance, has become a major objective on policy agendas. Existing insurance products, however, face several challenges. On one hand, traditional indemnity insurance products suffer high transaction costs and moral hazard (Hazell, et al. 1986). On the other hand, index-based insurance, which could overcome such concerns by relying on indices that are easily measurable and outside the farmer’s control, suffer low demand due to basis risk, lack of trust, and poor product ownership (Cole et al. 2013).

The study was conducted in Haryana and Punjab, two states in northwest India. These two states are the second and third largest wheat producing states in India and the largest contributors to the central pool of food grains used to provide welfare entitlements to India’s poor. Farmers in these states also have higher agricultural incomes and larger landholding sizes compared to the rest of India (Government of India 2016). However, in the last decade, yield-growth has stagnated and farmers have struggled with depleted water tables and poor soil fertility (Ministry of Agriculture and Farmer’s Welfare 2015–2016).

Intervention design

This intervention seeks to leverage new technologies to improve coverage and affordability of a weather based index insurance (WBI). A picture-based insurance (PBI) product that assesses crop damage using mobile phone photos uploaded by farmers was implemented in Haryana due to its importance in cereal production, high smartphone ownership and homogenous production patterns, providing an ideal setting for a proof of concept.

Capitalising on recent advancements in remote sensing, image processing, machine learning and smartphone ownership, PBI assesses individual crop damage from a stream of crop pictures, uploaded by farmers themselves. Farmers enrol their plots by downloading a freely available, easy-to-use mobile application onto their own smartphones, submitting an initial picture of the plot along with the premium payment, and regularly uploading repeat pictures of this plot with the exact same view angle as the initial picture, from sowing to harvest. The resulting time series of pictures are processed and assessed for visible damage. Farmers with visible damage are compensated.

Evaluation design and methodology

A combination of qualitative and quantitative approach was employed for this study. Focus group discussions were conducted in Punjab and Haryana during the intervention design stage to gather qualitative data on the main risks that farmers face, along with their perceptions of and overall interest in PBI.

A quantitative study was conducted with more than 700 farmers from 50 villages in six districts in the states of Haryana and Punjab to assess the feasibility of offering pictured-based insurance.  In each of these 50 villages, the researchers conducted a village census of all farming households and subsequently selected 15 farmers per village, randomly, conditional on meeting the following criteria: (1) having less than 15 acres of operational farmland and (2) planning to grow at least two acres of wheat during the upcoming Rabi season. Administrative records around farmers’ activity were used to assess the feasibility of PBI.

Primary evaluation questions

1. What are the impacts of marketing PBI products, designed to complement and strengthen existing area-yield index-based insurance schemes, such as the PMFBY?
2. How cost-effective is picture-based loss verification in generating these impacts within this context?
3. What are the key mechanisms driving (a) impacts of marketing PBI products and (b) cost-effectiveness?

Primary findings

1. 80 per cent out of the full sample of 592 farmers uploaded at least one valid picture during the season.
2. Farmers were willing to pay significantly higher amount for the product offering a combination of WBI and PBI. Farmers perceived most hazard to be visible from smartphone pictures.
3. Number of trust indicators were lower for the PBI group, including trust in weather stations and the pay-outs given by insurance companies.

Implications

1. A product relying on picture-based loss verification can improve the willingness to pay for insurance coverage, due to improved engagement with farmers and minimised basis risk.
2. Expanding PBI coverage for more risky cash crops such as cotton or horticultural crops could have stronger impacts on agricultural investments, production and rural livelihoods.