Scaling picture-based crop insurance
Source: Flickr, Lee Tucker

By Berber Kramer and Matthew Krupoff

Picture-based insurance (PBI) offers a promising alternative to existing crop insurance products for smallholder farmers. To assess plot-level damage, PBI uses a stream of pictures taken on a regular basis from sowing to harvest by farmers themselves with their own smartphone cameras. The project has thus far been piloted in Punjab and Haryana States in India. Researchers at IFPRI have found that this approach is feasible, in that farmers are willing and able to take pictures of sufficiently high quality for loss assessment and that picture-based loss assessments prove to be more accurate indices of crop damage than standard weather indices. In addition, a formative evaluation provides evidence that the approach is also economically viable, in that PBI improves farmers’ willingness to pay for crop insurance without inducing moral hazard or adverse selection. Based on these findings, IFPRI is planning to implement and evaluate PBI on a larger scale in partnership with private sector actors, other research institutions, development practitioners, and government agencies in India.

On September 7th, 2017, IFPRI and the Borlaug Institute for South Asia (BISA) organized a workshop that brought together multiple stakeholders to discuss the challenges and opportunities associated with scaling up the PBI concept. These stakeholders included agronomists at Punjab Agriculture University (PAU), researchers at BISA, extension officers, and farmers who had experience using the mobile app. Participants identified five topic areas and discussed how to address each area.

Optimizing the picture-taking protocol

Under the PBI approach, farmers send pictures of insured crops on a regular basis using a smartphone app. Repeat pictures are used for two purposes: (i) to document that the crop was managed properly prior to the incidence of damage and (ii) to map greenness and texture index curves over time so that image processing algorithms can detect damage. This protocol clearly requires effort on the part of the farmers; although the formative evaluation showed that most farmers are willing and able to undertake this effort, workshop participants discussed ways to further improve both the app and the protocol itself. Expert agronomists pointed out that the transition stages in the wheat-growing cycle (from flowering to milking, for instance) were crucial periods in which farmers should be encouraged to submit an increased number of pictures. This led to ideas on incorporating a crop calendar into the app to notify farmers to take more pictures when they approach these crucial time periods.

Improving assessments of non-visible damage

Not all damage is visible in smartphone pictures, including heat stress, which is a risk that Indian farmers are increasingly exposed to due to climate change. Workshop participants thus discussed how to cover risks that damage crops in non-visible ways; among the tools discussed was weather index-based insurance. For example, it was suggested that PBI could act as a top-up product in a more comprehensive risk management bundle that includes weather index-based insurance.

Our formative evaluation showed an extremely poor correlation between weather indices and crop damage. Another weakness of weather index-based insurance is that weather does not affect all farmers in the same way. The relationship between weather and yields depends on the crop growth stage during which extreme weather events occur. Combining PBI and weather index-based insurance could help overcome this weakness, because weather indices could be defined over crop growth stages—measured from the pictures—as opposed to fixed calendar periods during which the crop might not be at risk. Together with our collaborators, IFPRI and BISA are further exploring these complementarities between PBI and weather index-based insurance.

Establishing Distribution Channels

One challenge for scaling-up any crop insurance product is finding a way to distribute the product through channels that farmers can trust. Currently, cooperative societies cover nearly all villages in the study region and offer loans to farmers for agricultural purposes. Working with these cooperative societies would be beneficial because of their strong connections with farmers, but a lack of centralized control and experienced staff would make it hard to use them as a platform to offer PBI. For instance, although these cooperatives exist in almost every village, the administrative system has at most one secretary for all of the farmers in the society. It is unclear whether these secretaries possess the financial literacy to effectively market the insurance product to farmers, or if they are even a trusted member within the village.

Another option would be to work with input cooperatives, such as the Indian Farmers Fertilizer Cooperative Limited (IFFCO), which is a more centralized cooperative society with strong ties to villages. These cooperatives also tend to have experience working with the insurance industry. In addition, farmers often interact with these input cooperatives as they are preparing for the planting season, presenting good timing to market the PBI product to farmers.

Agricultural extension services are also a potential distribution channel. The Krishi Vigyan Kendra (KVK) is one such network that has connections with farmers from across the country. However, one key hurdle is a lack of staff, with advisor-to-farmer ratios currently reported as 1:2000. Thus, it could be more promising to build on alternative initiatives that help improve farmers’ access to agricultural information, including non-governmental non-profit organizations and actors in the private sector. As such, IFPRI is building a partnership with the Centre for Agriculture and Biosciences International (CABI) to leverage CABI’s existing two-way advisory system in Haryana and to develop a bundled product that combines PBI with picture-based agro-advisories.

Complementarities with Other Risk Management Practices

Most index-based crop insurance programs face low take-up when they have non-subsidized insurance premiums; thus, insurance premium subsidies could help boost demand. The question is how to optimally design these premium subsidies. Participants therefore explored whether subsidies or discounts for farmers purchasing the PBI product could be conditioned on adopting practices that could help lower risk exposure, such as conservation agriculture (CA). One CA strategy involves leaving residues on the soil and sowing directly into them instead of burning the residues. This practice can reduce damage from extreme heat, heavy winds, and excess rainfall, thus reducing expected payouts from picture-based insurance and justifying the conditional subsidy or discount. In addition, residue management is visible in smartphone pictures at the time of sowing, enabling monitoring of CA practices.

The formative evaluation included an experiment in which we tested the effects of providing PBI coverage free of charge, conditional on not burning residues. Farmers who received PBI free of charge unconditional on their management practices burnt their paddy straw in more than 95 percent of the cases. The no-burning condition improved residue management - burning rates fell by a significant 15 percentage points. However, the vast majority of farmers still burnt their residues, and lost their option to receive insurance coverage free of charge as a result. Farmers reported that they did not follow the recommended practice of leaving residue on their land because this requires special sowing equipment to directly sow into residues, which they often did not have, and because they lacked knowledge on how to apply CA. These findings suggest that the intervention would have been more effective if combined with agricultural advisory services and linkages with service providers who can rent out the necessary equipment. 

Expanding Into Other Crops

The formative evaluation focused on PBI for wheat, as this is a crop that is grown by many farmers in Haryana and Punjab. However, agricultural production in the area has been moving toward a system of monoculture in which farmers are growing rice during the Kharif (monsoon) season and wheat during the Rabi (winter) season. Increased rice production has resulted in higher irrigation use; as a consequence, the water table in Haryana and Punjab has been receding. Participants emphasized that more diverse agricultural production is needed to restore an ecologically sustainable production system. The PBI product could contribute to increasing diversity by covering more diverse and high-risk crops.

Horticultural crops, for instance, are regarded as risky crops, but they also have high returns. Participants thus discussed how PBI can expand to include horticultural crops. PBI could offer an ideal solution to providing insurance for such crops due to the large heterogeneity in horticultural production, which makes it difficult to design standardized weather indices, or to measure losses through satellite imagery. Expert agronomists stated that damage to high-value crops (for example, through pests or diseases) can certainly be visible in smartphone pictures through the coloration of leaves. Given that image processing will need to trained to detect these risks, ground-level data through close-up pictures will have to be collected over a period of time.

Moving Forward

We believe that PBI could help improve insurance coverage, reduce exposure to risk, and encourage risk-averse farmers to invest in high-return activities. This would ultimately improve agricultural incomes and farming households’ resilience to various shocks. To further develop PBI, IFPRI is designing ways to scale up PBI based on the feedback and discussions from the workshop, while measuring the impacts on farmers and their families through a rigorous multi-year impact evaluation.

Photo credit:Flickr, Lee Tucker