New Health Insurance Solutions in Pakistan

described in Experimental Design

2014-12-09

2017-02-18

Insurance takeup (in general and by risk-type according to baseline health measures)
Awareness about insurance
Health service utilization / claim ratio
Out of pocket expenditures in case of disability/hospitalization
Asset indices (liqid vs. illiquid)
Saving levels
Child labor input
Coping in case of disability/hospitalization
Psychological well-being

NRSP in agricultural areas usually works with so called community organizations (COs), which consist of 18 to 20 member households. Members of these community organizations are eligible for NRSP credit products with joint liability on the CO level. Further, NRSP offers non-agricultural credit products with joint liability to so called credit groups which usually consist of three member households. All NRSP credit clients and their spouses are covered by mandatory health and accidental death insurance. With this project NRSP gives existing credit clients the possibility to insure additional members of its household. The insurance innovations are randomly assigned to some Revenue Villages (RV), whereas no innovation will be implemented in other villages. Importantly, the latter RVs (control villages) do neither receive access to nor information about any of the innovations. There are six different groups: (i) control group, (ii) awareness campaign, (iii) individual insurance, (iv) individual insurance with high coverage, (v) household insurance, (vi) household insurance with group eligibility. I.e. we essentially vary the extent of information about the existing (mandatory) product and provide additional insurance products with varying coverage (in terms of amount and insured individuals).
At the household level, we furthermore cross-randomize prize discounts per insured individual and information about peer uptake.

The six different groups on the village level are:
(i) Control group: Credit clients in this group of RVs receive no additional insurance service other than the already available mandatory health and accident insurance for client and spouse.
(ii) Awareness campaign: To control for the awareness effects created by the insurance innovation one group of RVs only receives information on the existing insurance coverage. Awareness sessions aim at providing exactly the same information (regarding the existing mandatory insurance) as the one provided in the process of marketing the new insurance innovations. The content of the awareness campaign was developed by NRSP and checked for conformity with the research design by University of Mannheim.
(iii) Individual insurance: Additional to the mandatory insurance for client and spouse individual members can be covered at the cost of 100 rupees / year and individual (i.e. for each additional member that voluntarily takes up insurance).
(iv) Individual insurance with high coverage: This product is equivalent to the individual insurance described above, except that it doubles the coverage limit for hospitalization to 30,000 rupees / year. Premium payments are PKR 150 per insured household member.
(v) Household insurance: Additional to the mandatory insurance for client and spouse all other household members (those eligible according to the screening criteria) together may be covered at the cost of 100 rupees x number of additional household members / year. Compared to the individual insurance (iii) this might improve the risk pool of insured persons and hence the financial viability of the scheme. Also, the household package might be easier to administer. At the same time clients benefit from full protection of the household. On the other hand for large households the total additional premium may appear rather large and they might not insure.
(vi) Household insurance with group eligibility: Additional to the mandatory insurance for client and spouse household insurance can be purchased. In order to make the innovation eligible for the group members, at least 50% of client households in the CO / credit group need to take up. If this threshold is not met, the innovation is not available for the whole group. Compared to the household insurance (v) this might further improve the risk pool of insured persons and hence financial viability of the scheme. Also, there could be interesting impacts on informal risk-sharing. Fully insured households might be less willing to help uninsured households. This scheme prevents such a situation by imposing full or no additional insurance. Also, there even might be an incentive for those willing to insure to help financing insurance for the whole group.

As default, NRSP will offer the insurance product (iii), (v) and (vi) at a cost of 100 rupees per year and individual and the product (iv) at cost of 150 rupees per year and individual. To estimate demand for the products at different prices, vouchers with discounts will be distributed randomly to client households. Each household receives a discount with the following probability distribution:
- No discount: 25%
- 10 rupees per insured individual: 25%
- 20 rupees per insured individual: 25%
- 30 rupees per insured individual: 25%
The discounts are drawn in private and are iid.

The second cross-randomization element is the order in which the members of the credit group / CO draw their respective discount voucher and take their final insurance decision. The important element here is that takeup decisions of clients taking their decision earlier are observable for those taking their decision later. This means there will be variation in peer behavior observed prior to the own take-up decision.

For discounts: individual lottery
For order of insurance decision:randomization done in office by a computer
For assignment to product type:randomization done in office by a computer

Villages (RVs) for insurance type offered
Household level for price discounts and order of takeup

Yes

500 villages

7,500 households, 40,000 individuals

83 villages in each of the six treatment arms (1,250 households, 6,700 individuals per treatment arm)