Mitigation Index Insurance for Developing Countries: Insure the Loss or Insure the Signal?

Conventional agricultural index insurance indemnifies based on the observed value of a specified variable, such as rainfall, that is correlated with agricultural production losses. Typically, indemnities are paid to the policyholder after the losses have been experienced. This paper explores alternate timing for index insurance payouts. In particular, we explore the potential benefits of what we call “mitigation index insurance” in which the payouts of the insurance contract arrive before losses are incurred, in time to be used to take measures to mitigate, that is, reduce, eventual losses. For mitigation insurance to be of value, two conditions must be met. First, there must be a strong objectively measurable signal that is highly correlated with losses, but which is realized before the losses are incurred. Second, the signal must be realized in time for loss mitigation measures to be cost effective. We can think of numerous examples, and mention three. First, the onset of extreme rainfalls and profound flooding in coastal equatorial areas can be anticipated months in advance based on low sea-surface temperature readings, the so called El Niño phenomenon. Skees et al., proposed an El Niño-Southern Oscillation (ENSO) business interruption insurance contract that would provide indemnities to rural communities in coastal Peru before the eventual onset of torrential rains and catastrophic flooding that typically accompany the most severe El Niño events. The indemnities would afford communities the opportunity to implement adaptation strategies to mitigate the serious losses and disruptions that are almost certain to follow (Skees and Murphy, 2009). A second example is slow-onset disasters such as severe droughts, which in developing countries often lead to widespread famine, but not immediately. Relief agencies whose mission is to provide humanitarian assistance to victims of famine face tough questions about when to act. International donor response to a humanitarian crisis is often slow and inadequate, with funding mobilized only after evidence of a widespread famine become apparent. Often, by the time that a famine is recognized to exist, large-scale loss of life is already inevitable. The loss of life is not due to lack of early warnings, but rather the ability of international donors to mobilize funds quickly. An insurance contract that indemnifies relief agencies when drought occurs, before the onset of widespread starvation, would provide ready funding that would allow relief agencies to begin relief efforts in a more timely way. Chantarat et al. (2007) proposed to use weather index insurance to improve drought response for famine prevention, which pays claims based on realizations of a weather index that forecasts the prevalence and severity of food insecurity conditions in the targeted areas. A third example is replanting guarantee insurance. Crop yields are especially sensitive to the weather conditions that exist during the critical agronomic phase of germination, which occurs shortly after planting. A poor smallholder who invests in high quality seeds can quickly find that poor rainfalls shortly after planting have substantially reduced the maximum attainable yield at harvest. Given that it is still early in the planting season, the farmer can typically replant. However, if the farmer is poor and credit-constrained, he may lack the financial means to purchase new seeds, given that he spent what little cash he had on his original bag of seeds. In 2014, in Tanzania, Acre Africa launched a mobile-enabled weather index insurance contract that is bundled by seed companies into the bags of seed they sell. The insurance product indemnifies the registered smallholder if a drought occurs during the first three weeks after planting, with the farmer receiving a mobile money transfer for the full cost of quality seed so they can replant within the same season. In this paper we systematically compare the costs and benefits of mitigation index insurance with those of conventional index insurance using a stylized three-period, discrete choice, stochastic dynamic optimization model. We assume that insurance is purchased in period 0; a signal correlated with losses in period 2 emerges in period 1, at which time mitigation measures may be taken; and losses, if any, are realized in period 2. We assess the relative values of mitigation index insurance and conventional index insurance by deriving the individuals expected ex-ante welfare under three insurance scenarios: a) the individual purchases no insurance; b) the individual purchases conventional index insurance, which indemnifies in period 2 based on an index observed in period 2; and c) the individual purchases mitigation index insurance, which indemnifies in period 1 based on an index observed in period 1. Our analysis indicates that mitigation insurance can reduce moral hazard by providing incentives to undertake mitigation that are absent with conventional index insurance. We also find that the value of mitigation insurance rises as the precision of the period 1 signal rises, the losses avoided through mitigation rise, the cost of mitigation rises, and the individual’s initial wealth falls. We then turn to a multi-period dynamic stochastic model with a more refined treatment of time and explore how the relative benefits of mitigation index insurance vary with the point in time at which the indemnities are paid. In general, we find that the relative benefits of mitigation index insurance depend on two dynamically countervailing factors. The closer one is to the realization of the loss event, the more accurately the signal predicts eventual loss; however, it is also the case that the loss reduction obtained from mitigation falls, or equivalently the cost of mitigation rises. Thus the trade-off is: the longer one waits to mitigate, the better informed one is about its benefits, but the less effective it becomes. Information and mitigation costs thus have profound implications for the optimal timing of insurance indemnities. Mitigation insurance appears to be most promising if there is signal that predicts eventual losses with high precision, in time for cost-effective mitigation measures to be taken. We also expect mitigation insurance to be most valuable when credit constraints prevent the agent from ceasing the opportunity to mitigate when it becomes clear that it is beneficial to do so, a common situation among the poor of the developing world. We are currently investigating the impacts of credit and savings on the value of mitigation insurance, with plans to report the results in the working paper we present at the AAEA Annual Conference.