Tailoring Advice and Incentives to Enhance Consumer Welfare from Livestock Insurance

How to cite

Morsink, Karlijn; Barrett, Chris; Harrison, Glenn; Jensen, Nathan; Gidey, Tagel Gebrehiwot. 2022. "Tailoring Advice and Incentives to Enhance Consumer Welfare from Livestock Insurance." AEA RCT Registry. AEARCTR-0010345. https://doi.org/10.1257/rct.10345-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Karlijn Morsink* , Chris Barrett , Glenn Harrison , Nathan Jensen , Tagel Gebrehiwot Gidey

* Primary investigator

Registry ID
AEARCTR-0010345
DOI
https://doi.org/10.1257/rct.10345-1.1
First registered
Last updated
Status
Completed
Trial dates
2021-08-12 – 2023-03-31
Countries
Ethiopia ()
Registry record
http://www.socialscienceregistry.org/trials/10345

Abstract

We aim to investigate how individual-specific tailored advice about expected consumer surplus from index-based livestock insurance (IBLI), combined with incentives for insurance agents aligned with the tailored advice, impact take-up and consumer welfare from IBLI. To do so, we conduct a cluster-randomized controlled trial with 2400 pastoralists in 240 zones in Borena, Ethiopia. Zones were randomly assigned to three treatment arms. In the first arm pastoralists received status quo insurance promotion activities from the insurance company via insurance agents and insurance agents received a status quo sales incentive for their activities. In the second arm pastoralists received tailored financial advice based on elicited risk preferences, actual herd size, and subjective beliefs about herd losses and weather conditions and insurance agents received status quo sales incentives. In the third arm pastoralists received tailored financial advice, and insurance agents received incentives linked to the expected welfare generated by pastoralist's decisions. The interventions were delivered by Oromia Insurance Company (OIC) in collaboration with the International Livestock Research Institute (ILRI), the Ethiopian Environment and Climate Research Centre (ECRC) at the Policy Studies Institute (PSI), Cornell University, Center for the Economic Analysis of Risk (CEAR) at Georgia State University, and Utrecht University.

Intervention

Tailored Advice about Expected Consumer Surplus from Insurance:
- The individual-specific tailored advice is calculated based on an Expected Utility Theory (EUT) model of expected consumer surplus from index insurance demand with subjective beliefs about losses, building on the method developed by Harrison, Morsink and Scheneider (2020). We combine information about product characteristics with individually estimated risk preferences that were elicited at baseline, and real-time information on the herder's current herd size and expectations about losses, and weather realizations. In case a pastoralist owns a particular animal, we calculate, in real time, the expected consumer surplus (ECS) from purchasing IBLI for each potential share of the number of animals of this type, and select the share which generates the highest expected consumer surplus, conditional on a significant difference between purchase and non-purchase. If this share is 0, we advise the herder not to purchase insurance, while if this share is larger than 0, we advise the herder to purchase insurance. If there is no significant difference we inform the herder that our advise is inconclusive.

Sales and Welfare Incentive:
- The standard sales incentive structure that OIC uses to incentivize sales activities by VIPs consists of a commission of 8\% of the total insurance premium sales of the individual and a bonus that is increasing in size conditional on the total insurance premium sales, and is maximum USD 33 per VIP. In the zones assigned to the welfare incentives the objective is to incentivize VIPs to promote the tailored advice that maximizes ECS, which varies across individual herders, rather than to promote maximum sales.

Therefore, the VIP welfare incentive structure was designed to maximize incentives for the VIP if the pastoralist's purchase decision follows the advice exactly, and to decrease the VIP incentive gradually as the pastoralist's purchase decision differs from the tailored advice. To test the impact of these different incentive structures, the average expected incentive per pastoralist in each zone was designed to be similar in the zones with sales incentive and the zones with welfare incentives, so that differences can be attributed to the incentive structure, rather than to differences in the magnitude of the incentive.

Experimental design

We conduct a cluster-randomized controlled trial with 2400 pastoralists from 240 zones in 80 kebelles (3 zones per kebelle) in Borena in Ethiopia. Each kebelle has one VIP and this VIP conducts the training activities in each of the three zones in the kebelle. The three zones in each kebelle are randomized into one of three treatment arms:

T1: No tailored advice for pastoralists + Sales incentive for VIPs (Status quo)
T2: Tailored advice for pastoralists + Sales incentive for VIPs
T3: Tailored advice for pastoralists + Welfare incentive for VIPs

It is thus important to note that each VIP conducts activities in three zones, and the VIP incentive structure consists of a standard sales incentive in two zones, but a welfare incentive in one zone.

In addition,

Our treatment for pastoralists consists of two components: An informational video about the welfare framework used to produce the individual-specific advice, and the actual individual-specific tailored advice itself. Therefore, any effects of our treatments may result from either the informational video, individual-specific advice, or both. To test this, during our last sales season of Jan-Feb 2023, we randomized the control group respondents at the individual level to one of the following treatment arms: control group (T1-1) who neither watched the informational video nor received individual-specific advice, a group that watched the informational video but did not receive individual-specific advice (T1-2), and the remaining third were assigned to a group that received individual-specific advice without watching the informational video.

T1-1: No individual-specific advice for pastoralists, no informational video + Sales incentive for VIPs (Pure control)
T1-2: Informational video, no individual-specific advice for pastoralists + Sales incentive for VIPs
T1-3: Individual-specific advice, no informational video for pastoralists + Sales incentive for VIPs

Treatment assignments for T2 and T3 were kept the same.

Design details

Randomization method
Randomization is conducted at the zone level, with stratification at the kebelle level, implying that each VIP serves three zones, each with one of the three treatment assignments: T1(No tailored advice for pastoralists + Sales incentive for VIPs (Status quo)), T2( Tailored advice for pastoralists + Sales incentive for VIPs), and (T3: Tailored advice for pastoralists + Welfare incentive for VIPs). This stratification strategy was chosen to allow us to incorporate strata fixed effects controlling for VIP characteristics, variation in weather shocks, and variation in product characteristics. Randomization was executed after the baseline survey and before the delivery of the intervention.
Randomization unit
Randomization is conducted at the zone level
Clusters
240 zones
Observations
2400 pastoralists
Arms
800 pastoralists by each treatment arm
Primary outcomes
Insurance Demand - Extensive margin: purchase or non-purchase - Intensive margin: number of animals insured (in numbers and in Tropical Livestock Unit (TLU)) Adherence to advice - Extensive margin: follow the advice when advised to purchase or not advised to purchase - Intensive margin: the absolute value of the difference between the advised number of animals to insurance and the actual number of animals insured, conditional on being advised to purchase or advised not to purchase. Expected consumer surplus
Secondary outcomes
Welfare(Pastoralist's subjective well-being, Herd size, Distress livestock sales during drought, Livestock losses) Mechanisms( Insurance agents' effort)

Cited by

No citation data available for this registration yet.

Keywords