Improving trust and relaxing liquidity constraints to enhance uptake of weather insurance in Ethiopia

3ie evidence programme:  Agricultural Insurance Evidence Programme
Author(s): Temesgen K Belissa¹, Erwin Bulte³, Francesco Cecchi^2,3, Shubhashis Gangopadhyay^2,4,5 and Robert Lensink^*2,3
Institutional affiliation(s): Haramaya University, Ethiopia¹, Faculty of Economics and Business, University of Groningen, Groningen, The Netherlands², Development Economics Group, Wageningen University, Wageningen, The Netherlands³, India Development Foundation, Gurgaon, India⁴, University of Gothenburg, Sweden⁵ 
Grant-holding institution: University of Groningen, Groningen, The Netherlands
Main implementing agency: University of Groningen, Groningen, The Netherlands
Sex disaggregation: Yes
Gender analysis: No
Equity focus: Yes
Study type: Formative evaluation

Context

The intervention was implemented in three districts in the Rift Valley zone: Bora, Adami Tullu and Arsi Negele districts. The Rift Valley zone is a semi-arid plain plateau area with a low land agro-ecology. This zone is comprised of about 15 million smallholders and about 95 per cent of them are dependent on rain-fed agriculture. The area receives very low average annual rainfall during May to August and another short rainfall during October and November. Moisture stress and drought frequently cause devastating crop failure, rampant livestock mortality and herd collapse. Households in the area are smallholder subsistence farmers who mainly produce maize and wheat. They often face drought-induced income shocks that translate into erratic consumption patterns. Future drought shock predictions in Ethiopia are pessimistic with expected rises in temperature from 23.08 to 26.92℃ (Hulme et al. 2001). As a result, the wide crop-livestock mixed farming system in dry and semi-dry areas like the Rift Valley zone were projected to transform into extensive systems to respond to the risks of climate change (Meinke and Stone, 2005, Thornton et al., 2010). Hence, innovative drought risk management mechanisms like adoption of drought insurances were highly recommended for farm households in the area.

The agricultural sector accounts for, on average, about 42 per cent of the GDP, employs about 85 per cent of the rural labour force and contributes around 90 per cent of the total export earnings. Farmers face weather-related shocks every year and catastrophic risk every five years, resulting in significant welfare losses. About 82 per cent of the smallholder households in the study area have no access to formal financial service. Also, they do not have access to non-traditional risk coping mechanisms. Thus, drought insurance technologies can help to cover covariate losses and safeguard informal insurances. The co-existence of market-based insurance with the predominant social insurance institutions in the study area provides an interesting setting to undertake an adequate analysis for effective adoption of these technologies. However, as in many other localities where index insurance is offered, take-up is very low – approximately 7-8 per cent.

Intervention design

There were two types of interventions evaluated. The first was marketing and sales of a standard index-based insurance product through local Iddir-led cooperatives to increase trust. Iddir leaders informed households during monthly Iddir ceremony meetings about (1) nature and seasonality based weather variations and emerging climate change induced droughts; (2) the need for insurance to mitigate the adverse effects of these risks and (3) an IOU insurance working philosophy.

The second series of interventions focused on a premium payment system designed to promote financial inclusion and relax binding liquidity constraints. The IOU product allowed farmers to pay the premium after harvest.  During a second wave of intervention activities, a joint-liability IOU contract was tested that allowed farmers to buy an IOU in the next round if everybody from the same Iddir had paid the premium on the IOU.

Evaluation design and methodology

In the first phase of the intervention and evaluation, 144 Iddirs were cluster-randomised and assigned to six experimental arms.

1. Standard insurance through IDDIR promotions
2. IOU through IDDIR promotions with binding contract
3. IOU through IDDIR promotions without binding contract
4. Standard insurance through conventional channel (control)
5. IOU through the conventional channel with binding contract
6. IOU through the conventional channel no binding contract

In the second phase of the intervention and evaluation, the six arms were consolidated into four groups:

1. IOU through IDDIR promotions with a legally binding contract
2. IOU through IDDIR promotions with a joint liability contract
3. Standard insurance through conventional channel (control)
4. IOU through standard purchase, but need to sign the new legally binding contract

The estimation strategy utilised a simple post-treatment OLS model to estimate effect sizes.

Focus group discussions (FGDs) and in-depth stakeholder interviews were also conducted to better understand the reasons for uptake, and to triangulate quantitative results. FGDs were organised at 12 Farmers’ Training Centres with a total of 1,963 households. The number of households involved per discussion was around 15.

Primary evaluation questions

1. To what extent will the IOU lead to a substantial increase in uptake (say an increase of a minimum of 3 percentage points)?
2. To what extent will marketing via Iddirs lead to a substantial increase in uptake (also around 3 percentage points)?
3. Are there additional benefits in terms of uptake if the IOU and marketing via Iddirs are combined?

Primary findings

In the first wave of the study, the IOU intervention and combined intervention were statistically significantly associated with enhanced uptake. In the second wave, the IOU + Iddir marketing and IOU + joint liability contract were significantly associated with increased uptake.  Default rates were sizable, but no significant differences were detected when comparing the different treatment arms. There were no significant differences detected with respect to improved financial literacy.

Implications

The primary finding suggests that while trust in the standard insurance product might matter for adoption, marketing via Iddirs in and of itself is not sufficient to have a significant impact. The same holds for an IOU with a legal contract aimed at ruling out defaults. However, the combination of marketing via Iddirs and IOUs has a big impact on adoption. This study suggests that a combination of an IOU with a marketing treatment through a socially trusted customary channel will be very successful in enhancing uptake of index insurance.