SEWA Wage Insurance Experiment

How to cite

Malani, Anup; Miller, Grant; Miller, Morgen; Mahajan, Aprajit; Torgovitsky, Alex; Tebaldi, Pietro. 2019. "SEWA Wage Insurance Experiment." AEA RCT Registry. AEARCTR-0004240. https://doi.org/10.1257/rct.4240-2.0

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

Anup Malani* , Grant Miller , Morgen Miller , Aprajit Mahajan , Alex Torgovitsky , Pietro Tebaldi

* Primary investigator

Registry ID
AEARCTR-0004240
DOI
https://doi.org/10.1257/rct.4240-2.0
First registered
Last updated
Status
On going
Trial dates
2022-03-01 – 2024-04-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/4240

Abstract

We present the motivation, design and analysis plan for the SEWA Wage Insurance Experiment (SWIE). The SWIE is a randomized controlled trial examining the demand for and impacts of providing so-called "hospi-cash'' insurance that provides a per-diem indemnity payment while a beneficiary is hospitalized. We brand this a wage insurance product because SEWA intends it to cover lost wages due to hospitalization, even though it can cover other costs and consequences of hospitalization as money is fungible. The SWIE offers roughly 200 villages access to two hospi-cash policies. The SWIE randomizes the prices to which villages are offered access to these two plans. The SWIE has two methodological innovations. First, it identifies sharp non-parametric bounds on demand using the approach in Tebaldi et al. (2019). Second, we develop a new approach to experimental design in order to choose prices so as to minimize the width of these non-parametric bounds. The SWIE also contributes to our understanding of the value of indemnity health insurance policies. The SWIE examines impacts of prices on hospi-cash insurance and health care insurance uptake and the impact of prices and uptake of hospi-cash on hospi-cash insurance claims and health care utilization. The evidence on uptake also estimates demand for indemnity insurance. The SWIE also estimates adverse selection into this wage insurance. Finally, the experiment also examine whether the hospi-cash insurance product improves risk-sharing within treatment villages.

Intervention

We offer households the ability to purchase two hospi-cash policies at certain prices (i.e., premiums). The two policies differ in their coverage.

Experimental design

This study will include three pilot surveys, a baseline survey, and an endline survey approximately 18 months after the end of the baseline.

The village and household samples for this study were assembled in two steps. First, we selected 214 villages in Ahmedabad and Gandhinagar districts in Gujarat state. Of these 214 villages, 5 were assigned to a pilot 1 that we used to determine which two hospi-cash policies we would study, 9 were assigned to the pilots 2 and 3 that we used to determine the price conditions for the experiment in the main study. The remaining 200 villages were used for the main study.

Our inclusion criteria for households in the 3 pilot studies and the main study was that households have at least one SEWA member currently living there. (For example, the adult children of a SEWA member who no longer lives with them were ineligible to participate in the study.) While this criteria affects external validity, it provides a cohesive sample that already trusted SEWA and its financial services arm, VimoSEWA, and is thus likely to be cooperative with our study.

There were no exclusion criteria for pilot 1. For pilots 2 and 3 and the main study, however, we excluded members that were subscribed to VimoSEWA's existing hospi-cash policy. We excluded these members because a member who already has a hospi-cash policy is unlikely to purchase a second policy. Moreover, we are interested in demand for hospi-cash amongst those who do not already have it. We also excluded members below the age of 18 and above the age of 54 from pilots 2 and 3 and the main study, as the hospi-cash policy is not available to individuals outside this age band.

After the pilots, each of the 200 villages in the main study were randomly assigned villages to one of the four possible price-pairs for the two products offered. SEWA will also continue to offer it's standard suite of non-hospi-cash policies to SEWA households.

In advance of the start of marketing, we will conduct a baseline survey of sample households. After marketing, we will gather data from SEWA on which households purchased a hospi-cash policy and which one they purchased. We will also obtain SEWA claims data for all sample households that enroll in a hospi-cash plan. Approximately 18 months after the end of the baseline, we will conduct an endline.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
The unit of of randomization is a village. Each of the 200 villages are randomly assigned to a price 2-tuple for each of the 2 chosen products. 1. Villages are first sorted into 2-tuples based on the similarity of certain village-level features 2. In each 2-tuple, villages are randomly assigned (without replacement) to each of the 4 possible price combinations for the two wage insurance policies. All 200 villages are offered SEWA’s standard health insurance product at the SEWA-set premium.
Clusters
200 villages
Observations
10,000 SEWA members
Arms
50 villages
Primary outcomes
1. Uptake of each of the 2 selected hospi-cash products 2. Uptake of health insurance products 3. Days of hospitalization, by each member adult member and by any minor members of the household. 4. Days of hospital treatment reimbursed under the hospi-cash policies
Secondary outcomes
1. Uptake of other insurance and/or financial products 2. Number of days of work in last month and average daily wage on days worked 3. Monthly income 4. Asset index 5. Savings 6. Monthly consumption expenditure (net of medical expenses) 7. Measures of within village consumption inequality (e.g. variance of log consumption for households in a village) 8. Monthly medical expenditures 9. Battery of health and wellness measures

Cited by

No citation data available for this registration yet.

Keywords