Household Risk Preferences and Community-based Health Insurance Uptake in Rural Villages, Savannakhet Province, Lao People’s Democratic Republic: Field Experimental Data

Background: The risk preferences of individuals have an important role in many decisions under uncertainty. Buying insurance is a choice made under uncertain future outcomes. A community-based health insurance (CBHI) scheme has the primary objectives of reducing the health and financial risks related to unexpected catastrophic healthcare expenditures. It is assumed that the more risk-averse or loss-averse that individuals are, the more likely they are to favor the insurance.
Objective: This paper examines the association of rural and self-employed households’ likelihood of purchasing the CBHI scheme with their own risk preferences (risk aversions for gains and probability prospects, and loss aversion), which are revealed by the field experiment in the rural villages of Savannakhet Province, Lao People’s Democratic Republic (PDR).
Method: To attain this objective, first a structured questionnaire-based household survey is employed to collect 580 rural and self-employed households’ objective data. Moreover, an incentive compatible lottery choice field experiment (Tanaka et al, 2010) is conducted to assess their risk preferences, which allows us to test the validity of the expected utility theory (EUT) and prospect theory (PT) assumptions simultaneously. Second, probit regressions are applied to examine the association between their CBHI participation and their risk preferences by controlling their demographic and economic backgrounds.
Results: The findings of our study show that the probability of a household’s decision to enroll in the CBHI scheme is independent of the risk aversion towards gains but is significantly associated with the risk aversion towards probability prospects. A weak correlation between loss aversion and the choice to participate the scheme is found when CBHI ex-members are excluded and more demographic and economic related variables are controlled in the regression.