Can Health Insurance Reduce Poverty?

Last registered on April 24, 2015


Trial Information

General Information

Can Health Insurance Reduce Poverty?
Initial registration date
April 24, 2015

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
April 24, 2015, 8:41 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.


Primary Investigator

McGill University

Other Primary Investigator(s)

Additional Trial Information

Start date
End date
Secondary IDs
This paper presents the pre-analysis plan of two randomized interventions aimed at evaluating the causal impact of health insurance on engagement in higher-health-risk higher-return activities, and poverty. In one treatment group, we offer free health insurance to small-scale farmers in Kenya. In another group, without any subsidies, we present information about the product to existing social networks, which contain early adopters who have experienced the product before. Our primary hypothesis is that health insurance will engage people in higher-health-risk higher-return activities. Our secondary hypothesis is that the second intervention increases trust in the product, and thus take-up and risk-taking behavior despite the absence of any subsidies.
External Link(s)

Registration Citation

Chemin, Matthieu. 2015. "Can Health Insurance Reduce Poverty?." AEA RCT Registry. April 24.
Former Citation
Chemin, Matthieu. 2015. "Can Health Insurance Reduce Poverty?." AEA RCT Registry. April 24.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details



In this treatment group, we offer free (or 90% subsidy) health insurance. In the privacy of people's homes, we 1) deliver and present an NHIF brochure containing all relevant information about the product, 2) explain the very basic concept of insurance with a cartoon, 3) offer assistance to register (filling the form and sending it on their behalf to NHIF). Note that participants still had to visit our office with the proper documents (national identification card for all adults and birth certificate for all children) for us to organize the rest of the registration.

Informal group meetings

For other randomly selected households, we organize meetings for them and their existing informal group. The intuition of our intervention is that close friends who have themselves experienced a reimbursement by NHIF are best placed to reinforce trust in NHIF. As one of our respondents put it: "I have no previous experience with insurance, but I have a friend who has NHIF. When that man's wife fell ill, NHIF paid the bill in full. Therefore, I trust the company and understand how it works". To find these close and respected friends, we use existing social networks in the following way.
To the general audience of the group meeting, we offer the same information and assistance to register as in the subsidy intervention. No subsidies are offered. We do not incentivize early adopters to talk. This intervention merely provides an environment in which to share a positive experience about NHIF.


To verify whether trust in the product can be increased by the informational content delivered in the meetings, in another location, we delivered the same information to other households on an individual basis (brochure, cartoon, and assistance to register as in the groups)

Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
High-health-risk high return activities

∙ Pesticide use
∙ Fertilizer use
∙ Cattle growing
∙ Total income from risky activities (crops using pesticide, avocado, macademia, cattle) per month per capita
(Multiplying price of each crop or livestock by quantity sold at market, sold to broker, or consumed at home)
∙ Total income from less risky activities (other crops, other livestock, formal sector work)
∙ Total income per month per capita

Health outcomes

∙ Hospitalizations in the last 2 years for any household member
∙ Hospital expenditures
∙ Work accidents in last month for any household member
∙ Any illness or disease in last month for any household member (malaria, typhoid, work / duties accident, other household accident, permanent disability, heart problems, fatigue, old age)
∙ Days of work lost due to these diseases

Informal insurance

∙ Total savings
∙ In an informal group?
∙ savings within the group
∙ hospitality, i.e., the amount one gets if one is hospitalized, increases
∙ Have you ever left a group or separated from a group and formed a new one?
∙ On a scale of 1 to 5, 1 being completely unlikely and 5 being completely certain, how likely is it that your group will dismantle within the next 5 years?
∙ How reliable are the other members of your group for paying their group contributions? 01. Not at all reliable 02. A little reliable 03. Somewhat reliable 04. Significantly reliable 05. Always reliable
∙ Size of group
∙ How much authority does your group's chairperson/executive have over your group? 01. No [...] 02. Little [...] 03. Moderate [...] 04. Significant [...] 05. Very significant [...]
∙ How much trust do you have in your group members?01. No [...] 02. Little [...] 03. Moderate [...] 04. Significant [...] 05. Very significant [...]
∙ Say you are registered for NHIF. Then you fall sick and have to go to the hospital, but you are covered. Will your group still pay-out the same amount of "hospitality" regardless of your coverage?

Effects of interventions

∙ Do you trust NHIF? 1. Distrust completely, 2. Somewhat distrust, 3. Somewhat trust, 4. Trust completely
∙ Do you know about NHIF?
∙ How much do you think NHIF costs per year?
∙ How do you pay your medical bills? (3 main sources maximum)
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
For the subsidy treatment, we randomly select 169 households to receive free (or 90% subsidy) health insurance outof our sample.
For the "informal group meeting" intervention, we randomly select 186 households living in another location (an administrative subdivision of radius 2 hours by foot). We ask them to identify the most important social group that they belong to (e.g., ROSCAs, clan or family groups, church groups). Upon approval by the chairperson (obtained in 92 percent of the cases), we then organize a presentation at the usual meeting time and place of the group.
For the information intervention, we randomly select 389 households.
Two control groups are considered. First, we consider the 283 households living in the same location as participants receiving an informal group meeting. Despite them not being invited to a meeting, it is possible that some of them attend. This is due to the fact that they belong to the same group as our participants. This group thus measures the potential spillovers inherent in organizing group meetings. Second, 359 households from another location formed a "purer" control group. They live much farther from the treatment group. This control group received no other interventions.
Experimental Design Details
Randomization Method
Randomization done in office by computer
Randomization Unit
Individual randomization.
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
The treatment is not clustered.
Standard errors will be clustered at the location level, an administrative geographical unit in Kenya.
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
Subsidy: 169
Informal group meeting: 186
Information: 389
Control group spillovers: 283
Control group: 359
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
In the baseline, monthly income per capita is 4301 Ksh (SD=5527). The minimum detectable effect size in a treatment group of 169 individuals versus 359 in the control group is 1450 Ksh, 26% income increase. This is enough to detect the massive returns due to pesticide or cattle growing. Pesticides can double yields (Behera and Singh (1999). In qualitative interviews, farmers from this area report that a 5 months bull is worth 8,000 to 10,000 Ksh, while a 3 year bull is worth 60,000 to 100,000 Ksh. In the conservative case of 60,000 (/12 months / average household size of 3.7)=1,351Ksh/month. These sample sizes are thus sufficient to detect the returns to these high-return technologies.

Institutional Review Boards (IRBs)

IRB Name
Microinsurance for the poor
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information


Is the intervention completed?
Data Collection Complete
Data Publication

Data Publication

Is public data available?

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials