Gilbert Brenes-Camacho, CCP, Universidad de Costa Rica
The variance of age at death is assumed to diminish as health improves in a population. Medical technology, sanitation infrastructure and better healthy habits make induce deaths to get concentrated in late life. However, different social, economic and institutional conditions at early and old ages could determine inequalities in the timing of death. The paper explores how socioeconomic status predicts both the mean age at death and its variance using a lognormal accelerated failure time models with equations for an ancillary parameter. I use three datasets harmonized through the Gateway to Global Aging Data: HRS, MHAS, and KLOSA. The most important predictors in the analysis are mother’s education, income, having public health insurance, and receiving a public pension. As expected, people with more education live longer. Mother’s education predicts neither the mean nor the variance of age at death in the three countries. Americans and Mexicans –but not Koreans– with higher income live longer too. On the contrary, having public retirement pension income is the only consistent predictor of longer lives and smaller variances at age at death, even after controlling for income. This result highlights that robust social security systems have an impact beyond merely financial security.
Keywords: Mortality and Longevity, Longitudinal studies , Health and Morbidity, Population Ageing