Long-term Effects of Index-Based Livestock Insurance

The Index Based Livestock Insurance (IBLI) product is a commercial insurance product that makes insurance payouts based on remotely sensed NDVI data. The product is sold during two sales windows each year and provides coverage for 12 months, which containes two potential payout periods. This study tests for the long-term impacts of IBLI.

2010-01-01

2015-09-30

Herd size and composition
Total household income
Maximum years of education within a household

Herd size is measured by the total number of animals a household herds, and the number of animals a household owns. We aggregate across animal types by cattle market-value equivalent.3 We construct variables of herd composition – represented by the share of the value each animal type among the total value of livestock holding.

Total household income is examined as a measure of material well-being of the household. We construct this measure by aggregating the income of a household from various income sources over the year. We winsorize the value at the 99th percentile to address the effect of extreme values. For comparability, US dollar is used as a currency.

Human capital accumulation is a long-term outcome. If a household changed investments in human capital due to IBLI uptake, it could have an impact on the maximum years of education. We collected information of highest level of education completed by each household member, so we take the maximum of it to construct the variable.

Herd management expenditure
Annual milk income per livestock market-value equivalent
Livestock loss
Distress sales of livestock (in response to indemnity payments)
Share of children working full-time, part-time, and studying fulltime
IBLI uptake in the past 12 months

Herd management expenditure is an important indicator of pastoral investment of a household. To measure this, we calculate the total annual expenditure on water, fodder, supplementary feeding, and veterinary expenses.

We use milk income per livestock market-value equivalent as a measure of productivity of livestock herding. IBLI was found to increase milk income per livestock market-value equivalent in the short-term. We aggregate the income per each animal unit from the milk sales throughout four sales seasons. For comparability, US dollar is used as a currency.

Information on the number of animals lost due to various reasons (e.g., drought, diseases, livestock raid) was collected in the survey. We create a measure at the animal type, referring to the number of losses per animal type, as well as an aggregate measure by their market values using the pooled average prices observed from all purchases and all sold animals.

This variable is constructed by the number of animals sold to cope with droughts. We aggregate the number of animals across different types by their market values using the pooled average prices observed from all purchases and all sold animals.

Children play important role in a household’s livestock production. A previous study finds that IBLI increases children’s participation in full-time schooling while decreasing part-time work. Using information on children’s primary and secondary activities over the 12 month periods, we construct children’s activity status ranging from full-time work, part-time work and schooling, and full-time schooling.

IBLI uptake is recorded for all sample households in each type of animal – camels, cows, goats, and sheep. We examine both the extensive margin of whether or not the household purchased any IBLI policy, as well as the intensive margin corresponding to the number of animals insured of each type in Tropical Livestock Units (TLU)

The original research design distributed discount coupons for insurance premium randomly within each community (sublocation in Kenya, kebele in Ethiopia) each sales season. Details of this intervention are specified in Jensen, Barrett, and Mude (2017) for Kenya and Matsuda, Takahashi, and Ikegami (2019) for Ethiopia, respectively.

In addition to the discount coupons, interventions to transfer knowledge about the IBLI product were offered in both countries. In Kenya, a selected subsample of communities played IBLI knowledge games designed to illustrate key features of IBLI products. In Ethiopia, learning kits – including skit tapes and cartoons – were provided to randomly chosen subset of the communities. We do not explore the effect of these interventions in our analysis since the immediate effect of these interventions on insurance uptake were shown to be minimal (Jensen, Mude, and Barrett, 2018; Takahashi et al., 2016).

Stratified randomization done by computer.

Treatments were randomized among participants within sublocation in Kenya and within kebele in Ethiopia. Treatments were re-randomized each sales season.

No

16 sublocations in Kenya, 17 kebeles in Ethiopia

The original sample in Kenya contained 924 households. During rounds 2-6, attrited households were replaced when possible. Round 7, targeted the subsample of 954 households that had been surveyed at least twice during rounds 1-6 as its roster and managed to survey 871. In Ethiopia, the baseline sample was 515. During rounds 2-4, attrited households were replaced when possible. Round 5 targeted the subsample of 515 households that had been surveyed at baseline as its roster and managed to survey 398.

Treatment status was rerandomized within sample cluster each sales season.

IBLI_Long_term_effects_Pre_Analysis_Plan