Association for Public Policy Analysis & Management

Weather Index Insurance (WII) has been recently paid great attention especially in developing countries where standard indemnity-based insurance is rarely available for smallholder farmers living in remote areas. Despite its potential merits, the uptake rate of such products has been low (Gine and Yang, 2009; Cole et al., 2013). Moreover, unlike the choice patterns observed for conventional insurance that risk adverse agents are more likely to purchase, the demand for WII appears to decrease as the level of risk aversion increases (Gine et al., 2008; Cole et al., 2013). This puzzling empirical evidence about WII participation patterns requires a need for better understanding of WII demand factors such as risk attitudes.

This paper fills the gaps by eliciting three different risk parameters from Prospect theory (PT) and examining the effects of these parameters on rainfall insurance demand among 239 smallholder producers in Kenya’s arid and semi-arid lands. We first randomly assigned one of two WII products (one has a lower basis risk probability than the other) to the farmers and provided a brief explanation of the corresponding assigned product. Then, they participated in a field experiment in which we elicited parameters for risk aversion, loss aversion and non-linear probability weighting using the methodology developed by Tanaka, Camerer, and Nguyen (2009; hereafter the TCN game). Following the TCN game, we elicit demand curves at seven different prices using a multiple price list auction (Anderson et al., 2007). Combining these data with subjects’ demographic information and the level of knowledge on basis risk that we collected before and after the main experiment, respectively, we examine how index insurance demand responds to farmers’ three risk parameters and other characteristics.

We first find that PT explains our index insurance demand data better than Expected Utility theory (EU), which suggests that a single risk parameter based on EU may fail to extensively explain insurance purchase decisions of smallholder producers. Our preliminary results show that farmers are likely to buy more insurance when they are less risk averse and less loss averse. A negative coefficient on risk aversion can be attributed to competition between insurance schemes and farmers’ other risk-hedging strategies. However, because our sample farmers have never been introduced index insurance before and we control the variable of saving through informal saving groups, we think that farmers may view purchasing standalone WII as adopting new technology rather than a risk hedging tool. Also, we find evidence that insurance premiums matter (Clark, 2011) for farmers in that this upfront payment will become sunk costs when insurance payout is not realized (not only due to no rainfall shocks but also basis risk). For the same WII product with the same expectation of basis risk events, farmers who have high loss aversion are less likely to purchase insurance when they face higher insurance prices at which the amount of premium that could be a sunk cost is more substantial. Our results imply that different strategies such as a group index insurance contract need to be designed.