First, we can assess the degree to which the US is unusual by inspecting the scatter plot. If US household income was somehow unusually inflated by health spending, then it its residual should be much larger than it is (especially in better models that incorporate some of the non-linearity observed in other countries).

Second, just where is this extra income supposed to come from? More likely households bear almost all of the incidence of this spending, meaning each dollar spent on health means approximately one dollar less available to households to use elsewhere. This is certainly what the literature on employer health insurance benefits tends to suggest.

Third, the timing of this is all wrong. It’s clear the arrow of causation goes overwhelmingly from income to health spending because changes in health spending clearly follow changes in income. Over the past few decades it takes an average of about 2-3 years for changes income to be fully reflected in changes in health spending due to the large role played by 3rd party payers, employers, etc (the current year elasticity is about 0.2 whereas the long-term elasticity is north of 1.6).

~RCA