Tommy Bengtsson

This paper deals with socioeconomic differences in mortality among infants, children, adults, and elderly during the period of mortality decline in the nineteenth century. It uses multilevel Cox regression controlling for shared unobserved factors at the family level, and also compares the magnitude of the observed differences between socioeconomic groups with the impact of unobserved factors at family level using a new method for converting the variation stemming from unobserved factors into Median Hazard Ratios (MHR), directly comparable with relative risks. Longitudinal micro-level data for an economically transforming rural area in southern Sweden during the period 1815-1894 is used, and occupations are coded and classified using international standards (HISCO, HISCLASS). Together with information of landholding this provides high quality information on social position. The results do not support the idea that socioeconomic differences grew considerably in magnitude over the nineteenth century. For children there were clear socioeconomic differences already in the beginning of the nineteenth century, but they remained fairly constant, or even declined, over the nineteenth century. For other age groups, we do not find much in terms of social differences. For infants, other family-level variables, observed as well as unobserved, seem to have been more important than socioeconomic status in determining mortality. For adults and elderly, the models perform less well in explaining mortality, but it seems reasonable to conclude that socioeconomic status was not of prime importance for mortality in these ages, and certainly did not increase in importance over the nineteenth century.