Livestock—the principal store of wealth and source of livelihood for pastoralists living in the arid and semi-arid lands (ASAL) of the Horn of Africa— face tremendous risk from the frequent and catastrophic droughts that plague the region. Livestock losses can be especially catastrophic in this setting due to the poverty traps that characterize the system. Shocks can thrust once-prosperous households into chronic destitution.
- IBLI coverage has strong positive impacts on subjective, economic, and health-related indicators of well-being. The gains are especially pronounced in the midst of drought events.
- The marginal benefit/cost ratio of IBLI substantially exceeds that of unconditional cash transfers.
- These gains emerge despite IBLI’s imperfect coverage of purchaser’s risk exposure.
- Uptake of the product has been significant, with more than 40% of sampled households purchasing IBLI at least once.
In much of the world, insurance is commonly used to mitigate such shocks. Unfortunately, the high costs imposed by actuarial data collection, adverse selection, monitoring for moral hazard, and validating claims make conventional insurance policies unavailable in this geographically isolated and sparsely populated region. As a result, an overtaxed global humanitarian response system has, for decades, been the primary source of insurance against catastrophic drought events.
Over the past decade, index insurance has generated excitement among development organizations and national governments as a tool for extending access to formal insurance into environments that are inhospitable to conventional insurance, such as the ASAL. Index insurance bypasses many of the transaction costs associated with conventional insurance by basing policies on signals that are easy to observe and generally uninfluenced by individual action.
In addition, index products typically use a single index, further reducing transaction costs. The tradeoff is that index insurance only covers covariate shocks – the average losses within a community – leaving some household-specific, idiosyncratic shocks uninsured. For index insurance to successfully mitigate the impact of drought shocks on the insured, the index must be highly correlated with covariate risk and covariate shocks must present a substantial risk.
Index-Based Livestock Insurance
IBLI was first launched in Marsabit county of Kenya in January 2010. Since then, it has been expanded to include Isiolo (August 2013), Wajir (August 2013), Garissa (January 2015), and Mandera (January 2015) counties in Kenya, and the Borana region of Ethiopia (July 2012). The Government of Kenya is exploring the possibility of taking a variant of IBLI nationwide under a proposed Kenya Livestock Insurance Program.
The details of the IBLI contracts vary by region, but they are each developed to reflect deviations from historic averages of a remotely sensed and publically available Normalized Differenced Vegetation Index (NDVI) measure of rangeland vegetation density. As of this writing (April 2015), 10,067 IBLI contracts have been sold and 149,007 USD in indemnity payments have been made to insured pastoralists of these regions.
The IBLI research team launched annual longitudinal household surveys in the Marsabit and Borana regions in order to monitor factors leading to IBLI uptake and to rigorously evaluate the impact of IBLI coverage on a variety of indicators of well-being and behaviors. Both surveys include baselines collected before IBLI was available in the region. The Marsabit annual survey started in 2009 and has been collected for 5 different years, each with 924 households. The Borana annual survey, first collected in 2012, has now been collected from 515 households for 4 rounds.
A set of randomized experiments was implemented among the surveyed households in order to learn about the impacts of price and product outreach on consumers’ understanding and uptake of the IBLI products. Price incentives, in the form of randomly distributed discount coupons that reduced premiums from 10%- 100% for the first 15 covered livestock units were distributed among surveyed households before each sales season.
Randomized extension campaigns using a variety of games, videos, cartoons and radio broadcasts have also been implemented. These experiments generate experimental variation in IBLI purchases that can be used to untangle the complex web of factors leading to purchasing insurance in order to let us isolate the causal effect of IBLI uptake on a variety of household behaviors and well-being indicators.
Index insurance policies make indemnity payments according to index readings, rather than the actual losses that policy holders experience. The indices are intended to reflect area-average losses, but necessarily do so with error. Furthermore, heterogeneity between individuals within an index area can result in un-indemnified losses even if an index perfectly tracks average losses. The differences between insurable losses and indemnity payments, called basis risk, is a chief weakness of index insurance. IBLI (in Marsabit) appears to be the first index insurance product to have its basis risk studied rigorously, for the simple reason that most index insurance studies lack adequate longitudinal household data.
The basis risk faced by IBLI insured households is substantial. In Marsabit, IBLI covers only 62-77% of the herd mortality risk that households face. The remaining basis risk is partially due to index error, or differences between predicted and area-average livestock mortality rates. A much larger portion of basis risk arises from between-household variation in livestock loss rates. The figure above illustrates the high degree of heterogeneity between households across eight seasons in Central Marsabit. At best, a perfect index product could mitigate the risk associated with variation in the means across seasons. Households insured with a perfect index product would continue to face the risk associated with variation around the means.
Demand for IBLI has followed a similar dynamic in both the Borana and Marsabit pilot areas. The initial launch and associated commercialization and outreach was met with robust demand for the product. In the sales periods following the launch, there is a continued upward trend in cumulative adoption but there is also a substantial rate of disadoption (figure below). Some dis-adoption is not surprising as households experiment with the product, especially if they do not receive indemnity payments early on that build their trust in the underwriter. Logistical complications have also likely dampened demand, for example, causing cancellation of the 2nd and 5th Marsabit sales windows.
In addition to implementation factors due to the delivery channel, household and policy characteristics play an important role in demand for IBLI. Similar to the experience in other index insurance pilots elsewhere, household financial liquidity is positively correlated with demand, which is significantly, but inelastically, related to price. Households in regions with greater basis risk are more responsive to prices. Although in this region livestock ownership is commonly concentrated in the hands of men, while women customarily bear responsibility for milking herds and selling milk, there is no apparent gender difference in IBLI uptake.
Households also exploit ecological signals when making purchase decisions, buying less coverage when they foresee good rangeland conditions and more when they anticipate a difficult season. Spatially, demand is greater where average risk, especially covariate risk, is greater. Those who have been educated about IBLI, and therefore grasp the concept of basis risk, are sensitive to higher rates of idiosyncratic losses.
Production and Welfare Impacts
In spite of the necessarily incomplete coverage IBLI offers against herd losses, it appears to benefit – or would benefit – most households in the ASAL regions studied across a range of different indicators.
Even at the unsubsidized commercial premium rate, approximately equal to the actuarially fair rate with an additional 40% loading for firm overhead and profits, purchasing full IBLI coverage for all the seasons in the data cost, on average, the equivalent of 1.1% of herd size. In exchange, IBLI sharply improves the skewness of herd survival rates – increasing, on average, by 45.1% (from -1.185 to -0.651, t-stat=10.31) – by considerably reducing the risk of catastrophic losses. Simulations find that the improvement to skewness dominates, such that the majority of households are better off purchasing full IBI coverage than choosing the no insurance alternative.
In spite of the necessarily incomplete coverage IBLI offers against herd losses, it appears to benefit – or would benefit – most households in the ASAL regions studied across a range of different indicators. Even at the unsubsidized commercial premium rate, approximately equal to the actuarially fair rate with an additional 40% loading for firm overhead and profits, purchasing full IBLI coverage for all the seasons in the data cost, on average, the equivalent of 1.1% of herd size.
In exchange, IBLI sharply improves the skewness of herd survival rates – increasing, on average, by 45.1% (from -1.185 to -0.651, t-stat=10.31) – by considerably reducing the risk of catastrophic losses. Simulations find that the improvement to skewness dominates, such that the majority of households are better off purchasing full IBI coverage than choosing the no insurance alternative.
Looking across both drought and non-drought years in Marsabit, households with IBLI coverage increase investments in livestock veterinary and vaccination services, and reduce herd size (most likely reflecting a reduction of precautionary savings in response to an insurance alternative). These and other changes to production strategies among the insured seem to pay off, increasing the milk productivity of livestock and the total value of milk produced. There is also a positive impact on other indicators of well-being, including greater household income per adult equivalent (AE) and improvements to mid-upper arm circumference (MUAC), a strong predictor of child malnutrition.
Even in the absence of severe drought or indemnity payments, IBLI seems to improve purchasers’ wellbeing by reducing their reliance on costly ex ante risk reducing strategies or just providing improved peace of mind about drought risk exposure. By March 2014, IBLI had been available in the Borana Zone for 24 months and had been sold during 4 different sales periods, but had yet to make an indemnity payment. IBLI purchasers nonetheless exhibit significant and meaningful improvements in subjective well-being. These positive effects of IBLI coverage are large enough to overcome an observed and statistically significant negative impact of buyer’s remorse due to having spent money on insurance that did not pay off and thus, in retrospect, was an unnecessary expense.
IBLI as a Cost-Effective Social Protection Tool
IBLI’s positive impacts do not, in themselves, justify investing scarce development or social protection funds in the design and delivery of index insurance products. One needs to have a sense of the opportunity cost of those funds; that is, how the expected benefit/cost ratio stacks up against other prospective interventions. In order to better understand the value of IBLI, its costs and impacts were compared with those of the Hunger Safety Net Program (HSNP), an ongoing unconditional, targeted cash transfer program launched by the government of Kenya in northern Kenya in 2009. The overlapping spatial coverage and timing in the launch of HSNP and IBLI created an opportunity to design a comparative evaluation of the two alternative approaches to investing in social protection in risky ASAL settings.
Both IBLI coverage and HSNP participation increase household income from milk, income per AE and MUAC in children. By dividing the estimated local average treatment effects of each program by the total program cost born by the public per client served, we find that the two programs are comparably cost effective at their current scale. But for scale up, the average treatment effect per marginal cost of an additional client is more relevant than the total program cost per client, which includes the fixed costs of establishing the product and its delivery channel(s). In terms of marginal benefits/costs, IBLI is an order of magnitude more cost effective, reflecting the relatively large sunk costs of developing and marketing a new product in what was previously a non-existent market and the very low marginal costs to government or donors of additional insurees who purchase the product from private providers.
Index insurance shows considerable promise, especially in settings like the arid and semi-arid lands of east Africa where conventional insurance to cover potentially catastrophic herd losses does not exist. It is common knowledge, however, that index insurance products cannot practically provide complete risk coverage to policy holders. And uptake of some index insurance products has been low, raising questions about their attractiveness, scalability and sustainability.
At the same time, development institutions and organizations have invested millions in developing and piloting new index insurance products because there are potentially large benefits from even modestly reducing risk for agricultural households in developing countries. Rigorous empirical analysis of the IBLI product in both Ethiopia and Kenya and across multiple years provides convincing evidence that investments of this sort can have strong, positive impacts on a variety of indicators of wellbeing and that those impacts per dollar are at least on par with those from cash transfers programs and, at the margin, can be considerably higher.
This work also illustrates that index insurance is not a single-shot solution to poverty and vulnerability to disasters. But when designed well, index insurance products such as IBLI offer an important tool to help address uninsured risk exposure problems, especially in places where alternative risk management tools are costly and significant losses of livestock can cast households into poverty traps from which escape is difficult.
Publication made possible by support in part from the US Agency for International Development Cooperative Agreement No. AEGA-00-08-00008-00 through the Assets and Market Access Collaborative Research Support Program. Funding for this project was provided by the United States Agency for International Development. All views, interpretations, recommendations, and conclusions expressed in this paper are those of the authors and not necessarily those of the supporting or cooperating organizations.