1

Do informal risk-sharing groups reduce the challenge of providing weather indexed insurance products?

Evidence from a randomized field experiment in Ethiopia

Guush Berhane, Daniel Clarke,Stefan Dercon, Ruth Vargas Hill and

Alemayehu Seyoum Taffesse

The 2012 Research Conference on Micro-insuranceInstitute for Innovation and Governance Studies (IGS)

University of Twente, The NetherlandsApril 11 – 13, 2012

Authors’ affiliations:IFPRI: Ruth Vargas Hill, Alemayehu Seyoum Tafesse, Guush BerhaneUni. of Oxford: Stefan Dercon (also DFID), Daniel Clarke (also WB)

2

Introduction• Weather risk remains a major challenge to farming in the

developing world.

• Thin insurance possibilities. Informal insurance hampered by risk covariance!

• Classical information asymmetry problems & high implementation costs limit viability of traditional insurance.

• Index-based weather insurance offers new possibilities• However, demand remains invariably low --- basis-risk – a key

challenge

3

• Steps taken to mitigate basis risk still limited.

• In this work, we study how working with local traditional risk-sharing institutions – as Iddirs in Ethiopia - can help mitigate such challenge.

• In a carefully designed RFE, we study how some of the basis risk inherent in an index product can be mitigated.

• In so doing, we explore the possibilities in which traditional group savings and risk-sharing activities can be harnessed not just to mitigate basis risk but also examine if such mechanism can be strengthened so that they can be resilient in the face of ever changing climatic and environmental challenges.

Introduction

4

Research questions

• Specifically, we ask the following questions:

• 1) Can group contracts mitigate basis risk by increasing side-payments in the event of individual specific bad outcomes?

• 2) Do group contracts require ex-ante rules to effectively mitigate basis risk?

• 3) What are the mechanisms through which these processes work and what determines the direction of the outcome?

• 4) What are the overall welfare effects (results not yet ready)?

5

• Long run pilot—looking at group institutions takes time– first year in 2011, second year piloting now.

• 57 Kebeles (3-4 villages) selected around 3 weather stations in Oromia region of Ethiopia --- Shashemene, Dodota and Tibe

• Primary interest is to target risk-sharing groups -- conducted a network mapping exercise to ensure selection of villages with low prob. of network overlap between “treatment” and “control” villages.

• 110 villages selected: 60 villages randomly selected for treatment & 50 for controls

• Initially 150 villages selected, but 40 were dropped due to lack of capacity to market there

Weather index pilot in Ethiopia

6

Selling to groups vs. individuals, & mandating

• 60 treatment villages further split into:– 25 randomly selected to receive INDIVIDUAL contracts– 35 randomly selected to received GROUP (i.e., IDDIR)

contracts – groups decided how many policies to buy & they were encouraged to draw-up bylaws as to how to distribute group savings/payouts

• These 35 GROUP villages were further split randomly into two groups :– 18 mandated a certain format for the bylaws.– 17 non-mandated but encouraged

7

Mandated sharing-rules

• What did we mandate? – The group establishes regular savings to a common pot; – A 10% of any insurance payout in this group goes to this

pot– This pool is disbursed to members that experience

idiosyncratic basis risk, at zero interest rate.– Disbursement criteria is discussed and set by the group at

the beginning of the year– Members apply for the loan, group follows disbursement

rules!– Repayment is enforced by the rules

8

• Money was contributed by project as “savings”– Research goal: examine how money is disbursed – need to see

disbursements – and also show them we keep our words!– Discussing and agreeing on bylaws is a time-consuming process, it

helped to have a reason to do this

• Disbursement procedures– In July/August Iddirs received a promise of 800 Birr in October on

completion of bylaws discussion. • Mandated: 800 Birr on completion of mandated form agreeing to how payment

would be spent• Non-mandated: 800 Birr on completion of discussion, form could state that a

discussion would be held in the future on how to split payment

– Individual villages: In July/August 16 individuals were randomly selected in a public meeting to receive 50 Birr each in October

• Total flow of money into the village is the same, but who receives it is different

Provision of savings

9

• Village level meetings and trainings– First to iddir leaders and influential people– Second to everyone in the village – organized through iddirs leaders

and village elders

• Very few early season (May, June and July) polices were sold

• Discounts offered for late season policies (September):– Free insurance in Dodota and Bako Tibe– Price discounts in Shashemene: 40%, 60%, and 80% discounts

randomly allocated across villages

• 296 policies were sold in Shashemene (134 individuals and 435 Iddir members), about 13% of households

Insurance Marketing, Sales, and Take-up

10

• September rains were poor in Shashemene – triggered payout!

• Insurance payout was made at the end of October in Shashemene.

• Free savings payouts were also made at the end of October in all three study sites.

Payouts

11

Data

• Baseline survey: February –March 2011– 1760 households in 110 villages (16 households per village)

• Follow up survey I: December 2011, 5- 8 weeks after payouts were made– 1734 households in 110 village re-visited (very little attrition, 1.5%)– 138 iddirs in 110 villages

• Follow up survey II: February-March 2012

• Follow up survey III: February-March 2013

12

Baseline characteristics • High incidence of drought: 51% experienced drought shock in

last three years• Formal insurance an almost unknown concept:

– 10% had heard about traditional indemnity (car, life or health) insurance – No-one had heard of weather or crop insurance before

• Also: – Only 21% have heard of what a millimeter is – Only 7% had a bank account

• Initial interest in index-type insurance: 87% were interested in a weather indexed insurance policy described to them in the survey

• Indications of huge basis risk: only 32% thought rainfall measured at the nearest weather station accurately measured rainfall on their plots

13

Informal insurance very prevalent: only 5% did not belong to an iddir; 92% belonged to 1-5 iddirs

01

02

03

04

0P

erc

en

t

0 2 4 6 8d1_how many iddir are you or members of your household a member of?

Baseline characteristics

14

02

04

06

08

0P

erc

en

t

this village other village in this kebele nearby kebele elsewhered6_location of iddir

Close to 80% of iddirs span within the village

Baseline characteristics

15

• We compare outcomes between the control and the following treatment groups: – Individual and iddir– Mandated and non-mandated iddirs

• We estimate the ANCOVA for outcome variables of interest with baseline data:

• We estimate a difference in outcome equation for outcome variables of interest without baseline data:

• Stratification at location (weather station level) so dummies are included for this in all regressions

• Randomization at village level, so standard errors are clustered at the village level

Analysis

16

• Changed iddir rules– No clear difference between iddirs in “iddir” treatment and “individual” treatment

villages– We will see that this is because we are combining mandates and non-mandated iddirs

• Changing perceived ability to access cash for emergencies– Households from “iddir” villages are more likely to access cash for 1,000 medical

emergency. No more likely to access larger amounts or cash for non-emergent reasons.

– Households from “individual” villages perceive themselves as less able to access large (4000 ETB) cash loans. Further investigation needed to answer why this is.

• Actual transfers received since payouts (from those who did not directly receive an insurance payout, or “free savings” in individual villages): – Fewer transfers from individuals– More transfers from institutions (including iddirs) – largely from immediately after

payouts were made

Results: can groups be encouraged to reduce basis risk by increasing transfers?

17

loansloans for crop loss

1,000 Birr for medical emergency

4,000 Birr for medical emergency

4,000 Birr to start a business

individuals institutions

Iddir 0.061 0.066 0.065* 0.023 0.026 -0.017** 0.034***(0.046) (0.041) (0.033) (0.036) (0.028) (0.008) (0.009)

Individual 0.071 0.042 -0.018 -0.070** -0.061** -0.002 0.015(0.044) (0.031) (0.031) (0.033) (0.026) (0.010) (0.012)

Estimation method ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVAObservations 3629 3850 1733 1734 1730 4,993 4,983R-squared 0.198 0.013 0.049 0.014 0.038 0.129 0.121District dummies included to account for stratification. Robust standard errors in parentheses

Does your iddir provide Could you obtain… in a week Transfer received from …

Results: can groups be encouraged to reduce basis risk by increasing transfers?

18

• Changed iddir rules– Clear difference between mandated-iddir villages and other villages:

mandating changed the rules

• Changing perceived ability to access cash for emergencies– Non-mandated iddirs perceive themselves as better able to fund emergencies,

but the source of these funds is not iddirs, rather friends or assets. – Mandated iddirs are not more likely to finance medical emergencies, but are

more likely to rely on iddirs for the financing of these emergencies.

• Actual transfers received since payouts (from those who did not directly receive an insurance payout, or “free savings” in individual villages): – Mandating reduces transfers from individuals, non-mandating does not.– Mandating results in a larger effect on transfers from institutions (but point

estimates for mandating and non-mandating are not statistically significant from each other).

Results: are ex-ante rules important in encouraging transfers?

19

Results: are ex-ante rules important in encouraging transfers?

loansloans for crop loss

1,000 Birr for medical emergency

4,000 Birr for medical emergency

4,000 Birr to start a business

individuals institutions

Individual 0.07 0.042 -0.018 -0.070** -0.061** -0.002 0.015(0.044) (0.031) (0.031) (0.033) (0.026) (0.010) (0.012)

No mandate -0.013 0.024 0.089** 0.058 0.067** -0.011 0.025**(0.049) (0.050) (0.037) (0.041) (0.030) (0.009) (0.012)

Mandate 0.128** 0.102* 0.044 -0.009 -0.012 -0.023*** 0.041***(0.057) (0.053) (0.044) (0.045) (0.037) (0.009) (0.013)

Estimation method ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVA ANCOVAObservations 3629 3850 1733 1734 1730 4,993 4,983R-squared 0.204 0.016 0.050 0.015 0.041 0.129 0.121District dummies included to account for stratification. Robust standard errors in parentheses*** p<0.01, ** p<0.05, * p<0.1

Does your iddir provide Could you obtain… in a week Transfer received from …

20

• Change of rules for mandated iddirs

• Higher insurance demand: – Evidence that demand for insurance was lower in non-mandated iddir villages and higher in

mandated iddir and individual villages.

• Trust: – Not much change in overall trust (e.g., “most people can be trusted”) as a result of the

intervention, perhaps due to longer period needed.– No effect on trust in iddir leaders.

• Changing nature of payouts and information: households in iddir villages were: – less likely to have received a payout (esp. non-mandated)– more likely to have reported that the insurance or payout was big enough to help

(significant)– more likely to know others that received an insurance payout, or have others know about

their insurance payouts (esp. mandated)

Results: By what mechanisms are transfers increased?

21

Conclusions

• Lessons learned: – Local risk sharing groups can be encouraged to make payouts to members in the event of

idiosyncratic crop loss. We saw changed rules and increases in institutional payments– Not clear whether mandating helps– it ensures rules are changed, but does it do more

than this? – Seems to be some crowding out—transfers from individuals are lower, particularly in

mandated iddirs (although increased institutional transfers).

• Further analysis needed: – Did the increased institutional transfers reduce basis risk and improve welfare outcomes?– How does this develop in the long-run?– Better understand how this interacts with demand for formal insurance products

• Next steps, this season: – Continue with sharing rules and observe an additional season of insurance. – Included a feature to the index – i.e., gap insurance. A carefully designed crop-cutting

experiment is added to the index. – A lot of optimism this year – many policies already sold, particularly in area where

payouts made last year

22

Thank you