Health Insurance Subsidies Reduce Mortality for Low Income Individuals
Martin Andersen
Mark Dusheiko
Simona Grassi
Abstract:
We estimate the effect of changes in the level of health insurance subsidies over time on mortality for eligible individuals in the Swiss Canton of Vaud between 2002 and 2011. Administrative longitudinal data for over 350,000 individuals applying for health insurance subsidies with detailed individual as well as household demographic, socio-economic, health insurance characteristics and month of death is used. The causal effects are identified using (i) A differences in differences design exploiting a 15% reduction in the level of the subsidy as well as a decrease in the minimum income eligibility threshold in 2005 for adults receiving partial subsidies, compared to adults qualifying for full subsidies from means tested social insurance, which remained unaffected during this period. (ii) An instrumental variables estimation of the effect of changes in the level of subsidy for partially subsidized individuals; the instrument exploits the exogenous variation in the level of subsidies over time due to annual changes in the allocation rules.
Estimates from the two approaches yield comparable results. We find using approach (i) that the reduction in the maximum subsidy reduced subsidies for the partially subsidized population by an average of CHF40 per month and significantly increased their 12-month mortality by 0.4% (95% CI: 0.3-0.5%). Using approach (ii) we estimate that a CHF100 increase in the monthly subsidy for partially subsidized individuals significantly reduces the probability of death in the next twelve months by 0.6% (95% CI:0.05-1.14%)). The mortality reduction is significantly larger for women (1.1%) and the elderly (1.2%)
We investigate whether the quality of insurance coverage is associated with mortality by using the exogenous subsidy variation as an instrumental variable for the level of deductible chosen by the individual. We find that higher deductibles significantly increase mortality with the highest deductible available on the Swiss market (CHF2500) increasing annual mortality by 7%. Using the estimated coefficient on the monthly subsidy in the mortality regressions, we infer a value of a statistical life-year of approximately CHF200000. This estimate, however, is quite noisy and we cannot reject values as low as CHF90000 in all specifications.