Abstract
Recent studies on catastrophic health expenditures often use arbitrary and universally cut-off shares
(e.g., 10 percent) of health expenditures on household income (or on total household expenditure) as the
main indicator of catastrophic health expenditure. However, some empirical studies in Thailand which
employed such indicators ended up with counter-intuitive findings, e.g., a substantial percentage of households
in the richest quintile were found to be prone to catastrophic health expenditures, and more often
than not, at higher rates than those with lower income.
This paper uses the same set of data to demonstrate that the counterintuitive results resulted from
such faulty indicators. Using sophisticated household-specific poverty lines as a measure of household catastrophic expenditure, the study finds that, unlike households in the two lowest quintiles, it is extremely
rare for those with higher-than-median income to have their post-health-expenditure income fall
below the poverty line, a result which is completely opposite to those derived from the former indicator.
However, since the head count of health-related impoverished households is not sentitive for those
households that are already below the poverty line, the study proposes a supplemental indicator, that is,
changes in the normalized poverty gap due to health expenditure, as an indicator that measures the healthexpenditure
impact on poverty.
