This study examines the time series behavior of infant mortality rates within a long memory approach with non-linear trends using data for 37 countries. The main results show significant differences both in the degree of integration and non-linearities among the analyzed series. Furthermore, non-linearities in the time trends are found in most of the cases, in contrast with the main assumption of linearity used in the literature. Finally, the results on the integration order of the series have important policy implications in many areas, such as on international convergence in mortality rates, on the income and infant mortality relationship, and, on whether health policy interventions will have transitory or permanent effects on infant mortality rates.

Article.