Testing convergence in life expectancies: count regression models on panel data

Long-term growth convergence has extensively been investigated based on economic variables. Indicators of social development and health status are generally focused on their contribution to growth or on assessing national health care systems. Yet, as a general yardstick of well-being, life expectancy should be regarded as a criterion to measure crosscountry development patterns over long periods. Following a review of two approaches to estimating convergence, hypotheses and findings of recent studies on public health and growth are examined. Reformulating the analytical framework of both strands of research, discrete choice and parametric and semi-parametric Poisson regressions are applied to a three-decade panel of 132 countries. Determinants of achievements tend to impact differently across countries, with this distinction occurring particularly between negative and positive counts. Indications of convergence are tempered by results accounting for possible non-linear relationships, which further highlight the discrepancy between country groups with average life expectancy losses and gains.