Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Out-of-pocket health care spending imposes a substantial financial burden on households in many countries. To describe the magnitude of the problem, the literature has employed the concept of catastrophic payment, defined as out-of-pocket spending over a pre-specified threshold as a percentage of either household income or non-food expenditures. While the incidence of catastrophic payments provides information on the overall magnitude, it does not convey whether it is lower or higher income households that experience catastrophic payments more often. To overcome this shortcoming, recent studies have proposed and applied the concentration index for the incidence of catastrophic payments. One implicit assumption underlying the concentration index is that catastrophic payments matter more for poor households than for better-off ones. To test this implicit assumption, we empirically examine whether poor households are more likely to perceive the financial burden of out-of-pocket spending even after controlling for the out-of-pocket share in non-food expenditures. This question, while seeming obvious, is indeed a very interesting one because, unlike other ill-health outcomes in health-related inequalities, catastrophic payments by definition have already taken into account household's ability-to-pay. We analyzed data from the Korea Labor and Income Panel Study, a household panel study from South Korea between 1999 and 2005. For the dependent variable, we constructed a binary variable of whether a household reported a perceived financial burden of out-of-pocket spending. Standard logit models that relate the out-of-pocket share to the perceived financial burden may suffer from two methodological issues: unobserved heterogeneity in household vulnerability to out-of-pocket spending and endogeneity of the out-of-pocket share. To address these issues, we employed panel data methods. Specifically, we estimated a random-effects logit model including the out-of-pocket share and household income levels. The coefficients of primary interest are the ones on a set of household income dummy variables with the reference being the highest income group. Positive and statistically significant coefficients on these variables would indicate that differential vulnerability effects exist by household income level when the out-of-pocket share and other things being equal. In addition to these main effects, we examined whether these differential vulnerability effects vary with the magnitude of out-of-pocket share, by introducing interaction terms in the model. Poor households are more likely to perceive the financial burden of out-of-pocket spending, even after controlling for the out-of-pocket share and other household characteristics. Furthermore, these differential vulnerability effects are more pronounced with small increases in out-of-pocket spending. Therefore, catastrophic payments matter more for poor households than for better-off ones. Our study lends support to the use of the concentration index of the incidence of catastrophic payments. Furthermore, defining catastrophic payments simply based on the out-of-pocket share without accounting for differential vulnerability effects may lead to invalid estimates for financial burden of out-of-pocket spending.