Households’ Willingness to Pay for Health Microinsurance and its Impact on Actual Take-up: Results from a Field Experiment in Senegal

Abstract

Community-based health insurance schemes (CBHIS) have been present in the region of Theis, Senegal, for many years. Yet, despite the benefits they offer, take-up rates remain low. This article measures the willingness to pay (WTP) for CBHIS premiums in such a context; our results highlight the role of income, wealth and risk preferences as determinants of WTP. We also provide an analysis of the predictive power of WTP on the actual take-up of insurance following our offering of membership to a sample of 360 households. Results show that WTP has a positive and significant impact on actual CBHIS take-up.

Acknowledgements

We acknowledge financial support from the International Labour Organisation (ILO) Microinsurance Innovation Facility, the Fonds National de la Recherche du Luxembourg, and the Carnegie Trust for the Universities of Scotland. We thank the GRAIM in Thiès, Ndeye Seyni Kane for her help during our field work, Olivier Dagnelie and Kyle McNabb. Any remaining errors are our own.

Notes

1. Health care in Thiès is organised according to a tiered system consisting of health huts (staffed by community health workers), health posts (staffed by nurses and certified midwives) and health centres (staffed by medical doctors, nurses and certified midwives). The health district of Thiès has one regional public hospital and one privately run mission hospital (St-Jean de Dieu). Data for this region show that the ratio of inhabitants to health centres is seven times greater than WHO standards, while the ratio of inhabitants to health posts is in line with international norms (ANSD, 2008).

2. For example, An Fagaru, a popular MHO in Thies, proposes the following coverage: 80 per cent of consultation at health posts; 50 per cent of expenses at health centre and hospitals (regional hospital and Saint Jean de Dieu hospital). The monthly per capita premium is 200 FCFA.

3. Starting bids are randomly drawn from 100, 150, 200, 250, 300 FCFA.

4. Any arrears on premiums can lead to exclusion for that member from coverage by the MHO. Whilst the rules are strict, the administrators of some MHOs have admitted allowing for a degree of flexibility.

5. Tests for random assignments of treatments across samples are provided in Bonan et al. (2012). Randomisation with respect to voucher assignment appears satisfactory. Some significant differences between the invited and not invited samples are discussed in this article.

6. We also use alternative ways of expressing wealth: (1) the DHS Wealth Index (Filmer & Pritchett, 2001; Rutstein & Johnson, 2004), which is a synthetic index obtained by the first principal component derived from the principal component analysis on the answers on housing and dwellings; (2) quintiles of the DHS Wealth Index. Our results hold when we use either one of these measures. Results are not shown but are available upon request.

7. These results are robust to different definitions of time and risk preferences. For risk preferences we consider the sub-samples of risk averse agents (always opting for the certain amount) for small and large stakes, for gains and losses. For time preferences we employ different time horizons and stakes; namely, we elicit two days, two weeks, one month and six months discount factors for small (1000 CFA) and large (10000 CFA) stakes, and we construct a dummy taking a value of 1 when the individual belonged to the more patient half of our sample for each time horizon. We use these different combinations of time and risk variables. Results are not shown, but are available upon request.

8. We argue in Bonan et al. (2012) that our sample size calculation was powered to detect statistically significant differences from the various groups. The power, for plausible pre-survey values of take-up for our different groups, is in almost all cases above 70 per cent. For the computations and more discussion on the power of our tests, we refer the reader to the Online Appendix.

9. One may argue that enumerators’ ability in conducting the survey and personal characteristics may drive part of such results. However, when we include enumerator fixed effects, the previous results do not change. Moreover, the dummies identifying each enumerator are jointly insignificant.

10. The interaction of WTP with the voucher treatment yields no significant impacts. Results are not shown, but are available upon request.

11. We also investigate the interaction of WTP with the baseline knowledge of insurance principles and find no significant effect on actual purchase (column 3).