Speaker
Description
Keywords: risk, agriculture, technology adoption, digital technologies, Kenya
Introduction
Farmers in developing countries face a host of climate-related risks that make their incomes volatile, undermine their food security, and hamper investments in agriculture (Dercon, 2005). Agricultural insurance has the potential to transfer some of these risks away from smallholder farmers, thereby protecting households’ incomes and assets from weather shocks, improving access to credit, and increasing investments in agriculture (Karlan et al. 2014; Jensen et al. 2017; Hill et al. 2019; Janzen and Carter, 2019). Yet, insurance supply and demand remain limited across the global south. Classic incentive problems stemming from asymmetric information, such as moral hazard and adverse selection, along with high monitoring and transaction costs, restrict the supply of indemnity-based insurance (Alderman and Haque, 2007; Sanrtos and Barrett, 2011). Index-based insurance, introduced to lower transaction costs and solve problems associated with asymmetric information, grapple with low demand due to high basis risk (discrepancies between actual losses and insurance payouts determined by the index), product complexity combined with low financial literacy, liquidity constraints, and lack of trust (Clarke et al. 2016; Hill et al. 2016; Gine et al. 2008; Casaburi and Willis, 2018).
Objective
The objective of this paper is to evaluate the impacts of a digital innovation on crop insurance take-up and agricultural investments among smallholder farmers in Kenya. We analyze a picture-based insurance solution, designed to lower basis risk by indemnifying insurance claims based on visible damage detected from smartphone images of an insured crop (Ceballos et al., 2019). Reducing basis risk, combined with a more participatory product that is easier to explain than index-based insurance, could enhance farmer trust, product understanding, demand, and impacts. By providing insurance providers with a stream of pre- and post-damage pictures of insured crops taken from planting to harvest, the product also reduces monitoring costs, addressing potential concerns around classic incentive problems stemming from asymmetric information (Ceballos et al, 2021). This could translate into a more commerically viable insurance solution with increased demand whilst keeping monitoring costs low.
Methods
To evaluate the impacts of this innovation, we use a cluster randomized trial in which 191 villages from seven counties in western and eastern Kenya were randomized into one of three treatment arms: a control group, in which no free insurance trials were provided; an index-based insurance treatment, in which up to 20 farmers per village were provided with free trials of rainfall insurance; and a treatment arm in which up to 20 farmers received free trials of picture-based insurance. Across these three treatment arms, 3200 farmers were surveyed at baseline, midline and endline to measure the impacts of the two types of insurance products on insurance demand, perceptions, and agricultural investments. We cross-randomized whether farmers were also offered seeds of improved (drought-tolerant) maize and sorghum varieties. A few months prior to endline, farmers in the two insurance-related treatment arms were offered an opportunity to purchase the insurance product of which they had previously received an insurance trial (rainfall index-based and picture-based insurance, respectively). Insurance was offered at that time at actuarially fair or, for a randomly selected subset of farmers, subsidized premiums.
Findings
We find that providing picture-based crop insurance significantly increases demand relative to weather index-based insurance, but particularly among farmers who were randomly selected to receive an insurance premium subsidy. These effects are mediated by significantly improved perceptions of insurance in terms of trust, perceived quality, and satisfaction with provided coverage among farmers offered picture-based insurance, compared to those who were offered rainfall insurance. The increased demand for picture-based insurance is concentrated among women smallholder farmers. Whereas in the rainfall index insurance treatment, less than 20 percent of women purchased insurance, this increases to 31 percent in the picture-based insurance treatment. We observe no significant effects of offering picture-based insurance on take-up among men smallholder farmers. As a result, the picture-based product increases women’s insurance uptake beyond take-up rates observed among men.
Despite increased take-up, intent-to-treat estimates reveal no significant impacts on agricultural investments, including expenditures on seeds, fertilizer, other inputs, and hired labor, and on yields. Local average treatment effects, estimated by instrumenting insurance coverage using the randomly assigned treatment, show significant positive effects of insurance coverage on amounts spent on seed, fertilizer and hired labor. However, this crowding-in of inputs occurs only in the cross-randomized treatment where farmers were not offered drought-tolerant maize and sorghum varieties. This suggests that farmers perceive improved varieties, a technology to manage moderate risks, to be a substitute for crop insurance, which is in theory designed to manage more extreme risks. An important research question therefore remains how to design insurance products in ways that farmers perceive them to be complementary, rather than substitutes, for other risk management instruments they might use.
Conclusion
Our findings imply that an easy-to-understand insurance solution with low basis risk can increase demand to levels that are potentially more commercially viable than index-based insurance, but insurance programs would still benefit from insurance subsidies to reach larger numbers of farmers. Although the use of smartphone technology could have potentially reduced interest among women farmers, who have less access to smartphones, we find that this did not deter them from taking up insurance, and that in fact, the use of smartphone pictures for claims settlement increases demand among women more than among men. Whilst the innovation increases demand for insurance, in this context, it was not sufficient to increase smallholder farmers’ investments in agriculture, potentially because farmers perceive insurance to be a substitute for other risk management technologies in which they can invest. This will limit the potential impacts of such financial instruments on agricultural production, incomes, food security, and other nutrition-related outcomes.
