In the literature dedicated to the "health as a luxury good" question, health care expenditure (HCE) is hypothesized to be a function of GDP without considering any other relationships. In this paper, we argue that this could be a bilateral relationship: good health is considered as an input of the macroeconomic production function, stimulating the GDP. A modified version of the Granger (1969) causality test proposed by Toda and Yamamoto (1995) is investigated between GDP per capita and HCE per capita in United States for comparing the periods of 1965_1984, 1975_1994, 1985_2004 and 1965_2004. Results show these three periods have different causal relationships. At the beginning for 1965_1984, there exists a bilateral relationship. For the following period, there is a unidirectional relationship from HCE to GDP, and for the 1985_2004, a unidirectional GDP_HCE is significant. From the start to end of periods (1965_2004), a unidirectional relation from HCE to GDP is existed.
