Asymmetric effect of income on the US healthcare expenditure: evidence from the nonlinear autoregressive distributed lag (ARDL) approach

Previous research that has investigated the relationship between income and Health Care Expenditure (HCE) assumes that the effect of income variation on HCE is symmetric. Additionally, while HCE consists of twelve different types of health services, most of the studies have only focused on the relationship between income and aggregate HCE. By applying the linear ARDL approach, introduced by Pesaran et al. Bounds testing approaches to the analysis of level relationships. J Appl Econom 16(3):289-326 (2001), this study expands the literature by estimating the income elasticity of HCE for all types of healthcare services. Our findings imply that while for some health services income elasticity is below unity, for some other services health spending tends to grow faster than GDP. We also applied the non-linear ARDL approach of Shin et al. Festschrift in honor of Peter Schmidt (2014), for the first time in the literature, to examine if the adjustment of income variation follows a non-linear path. This paper provides statistically significant and robust evidence that the effect of income variation on healthcare expenditure is not symmetric, despite what previous studies assume.