Insuring crops through photos

A farmer in Kenya with a small plot of land faces complex decisions during the growing season. Should I start planting on the same day as I do every year? Will the rain come in time for the crops to grow well?

Unforeseen weather circumstances increase the risk of significant loss of harvest and therefore of income. In the future, this farmer can better manage the risk of drought or other weather events with a new form of insurance. Any damage is documented in photos farmers take of their fields every couple of weeks. These help insurers assess the damage.

Insurance against damage is difficult for small-scale farmers, explains Francesco Cecchi, associate professor in the Development Economics group. He conducts research in countries such as Kenya, Ethiopia and India, where millions of farmers farm on less than half a hectare of land each. ‘Most of those farmers are uninsured because it is simply too expensive, and the chances of compensation are too small. The damage has to be verified in very remote areas, which is too much effort for the insurer, so it doesn’t happen. Consequently, if there is any doubt about a claim, the farmer doesn’t get compensation.’

Satellite data

It was previously thought that the problem could be solved by using satellite data, making it easier to assess damage in remote regions. ‘Unfortunately, that didn’t increase the interest in insurance among small farmers. They still didn’t trust insurance companies to pay out in cases of damage. More rain might have fallen on one side of a hill than the other, and satellite data often can’t detect that.’

Being uninsured is a growing problem, as has been clear in recent years. ‘The risks of extreme weather are quite significant for farmers who depend on the yield of such a small plot,’ Cecchi says. ‘They make annual risk assessments to maximize the chances of earning enough income for their families, and take precautions such as planting several different kinds of crop.’

‘Timing is crucial too. The ideal moment for planting is right before the rainy season begins, since these are usually crops that need every drop of rain. But with climate change, both the timing and the amount of rainfall are changing. This increases the risk that the crops won’t grow at all, with all the inevitable consequences of that. A failed harvest is catastrophic for people living on or just below the poverty line.’

Damage experts

Working with the International Food Policy Research Institute (IFPRI), Cecchi has therefore developed a new method to provide proof for insurers in cases of yield loss: Picture-Based Insurance. Farmers use smart phones to take photos of their eld and upload them to the insurer. ‘The farmers must take the exact same photo of their eld every few weeks. If sudden damage occurs, you’ll see a clear difference, and if there’s ongoing drought, for instance, you’ll see the damage increase gradually.’

When a claim is submitted, three damage experts review the photos. If they identify significant loss of harvest, the insurance company pays out. In 2018, Cecchi and his colleagues launched a pilot in Kenya, Ethiopia and India in collaboration with the IFPRI and the national governments in these countries. Farmers were offered free insurance: one group could try the new Picture-Based Insurance, and the other group received standard insurance. After three years, participants in the pilot could opt in to a paid version, priced the same as standard insurance.

In Kenya, those who had experienced Picture-Based Insurance were 30 likely to choose to purchase insurance. One of the farmers explained the main reason for choosing Picture-Based Insurance: ‘If misfortune strikes, I’ll have a much bigger chance of being compensated, avoiding losses and even making a small profit, so I can farm again next season.’

Algorithm

The researchers are now training an algorithm with the photos collected so far. It will automatically assess how much damage the crops have suffered. In 88 per cent of cases, the judgement is in line with that of the experts reviewing the photos. Cecchi: ‘That’s a fairly high level of reliability, but not high enough to completely do away with human assessment. The advantage is that experts will only need to review the doubtful cases, which saves a lot of work.’

In the meantime, Cecchi hopes that insurance companies will make use of Picture-Based Insurance. One company is going to offer it, but in a more limited form. ‘In the end, it’s up to them whether and how they will integrate Picture-Based Insurance into their products.’