Living Alone in Later Life. a Multi-Country Comparative Analysis
Miguel Requena, UNED
David Reher, UCM
Mojgan Padjab, Umeå University
Glenn Sandström, Umeå University
Methods. Multilevel logistic regression techniques are used to estimate the weight of different factors behind the residential choices of elderly women. Decomposition exercises of intercountry differences are also presented.
Results. Basic theoretical expectations about micro-determinants of living alone are validated and compared with those found in the existing literature. As expected, the importance for living alone of not having children is greater in more familistic and less developed societies where institutions tend to be weaker. Population characteristics explain only a small part of the differences in living alone that characterize these societies.
Discussion. While economic development accounts for an important part of living alone among elderly women observed in the five countries, many of the observed differences are also explained by societal characteristics such as family systems and available policy options. Everywhere levels of living alone are likely to increase in the future, but strong differences in the pace of change are likely to emerge in view of different levels of development and familism.
This paper compares the determinants of living alone among women during later life in five countries with very different family systems, policy contexts, levels of development and socioeconomic characteristics. The set of countries used is taken from the developed and the developing world, spans every continent, and includes Spain, Iran, Indonesia, Kenya and Costa Rica circa 2011. In the developed world, the increasing incidence of single living among older persons has been associated with the general process of ageing. It is often seen as the result of modernization and the way it changes the traditional dependence on the family during the final stages of life, substituting it wherever possible with residential independence and institutional options for the overall management of aging. This mix of factors has been shown to increase the overall weight of this living arrangement, though persistent differences exist. Research on the developing world on this subject is less abundant but equally interesting given the rapid rhythm of demographic and social changes experienced in these countries. The inclusion of both developed and developing nations enables us to assess the determinants of single living in later life from the standpoint of the pace of modernization in different societies as well as the way it intermixes with the availability of institutional and familial sources of support for elderly persons and the degree of familism. This paper includes a framework for understanding living arrangements during later life from a comparative perspective.
Data. This analysis will make use of micro census data from the IPUMS database. The five countries have the requisite information needed for this type of analysis. The micro census data (circa 2011) include several household and socio-demographic characteristics (household size, age, sex, education and marital status). A unique aspect of this paper is that the analysis makes use of census data where estimates of children ever born (for women, not for men) are included. This variable, absent from most censuses circa 2011, provides us with an valuable though indirect estimate of the supply of direct kin available to people as they age, a key explanatory variable in our explanatory framework. The large sample sizes guarantee very small sampling errors and imply high reliability of estimates.
Variables. Our dependent variable is based on the living arrangements of older women coded as a straightforward dichotomous variable: living alone and living with others. Our main independent variable is past fertility (coded into two categories: childless women versus mothers). In the models used in this paper, other key covariates were included: age (coded into five-year groups), current marital status (never married, married, separated/divorced, and widowed), and educational attainment (primary or less, secondary, and university). Place of residence linked to city size and migratory status have been also added to the models as population controls.
Statistical analysis. Multilevel logistic regression techniques are used to estimate the weight of different factors behind the residential choices of elderly women. Pooled and category-specific and country-specific logistic regression models are specified making use of a categorical dependent variable (alone/not alone). In order to control for territorial heterogeneity in the five countries, multilevel logistic regression models are estimated including a random intercept for respective territorial demarcations. Non-linear regression-based decomposition models are also specified to assess the extent to which observed differences across countries are attributable to specific population characteristics or to other factors.
Results. Basic theoretical expectations about micro-determinants of living alone are validated and compared with those found in the existing literature. These affect both the supply of potential caregivers (fertility, mortality) and the willingness of the family and/or of individuals to go it alone. As expected, the importance for living alone of not having children is greater in more familistic and less developed societies where institutions tend to be weaker or even non-existent. Population characteristics explain only a small part of the differences in living alone that characterize these societies.
Discussion. In the coming years, it is likely that single living among the elderly will skyrocket in many societies, though growth may well be slowing down or coming to an end in the richest, most developed part of the world. Strong differences in the pace of change are likely to emerge in view of different levels of development and family systems. For, while economic development accounts for an important part of living alone among elderly women observed in the five countries, many of the observed differences between the five countries are also explained by societal characteristics such as family systems and available policy options. These findings have relevant implications for the management of ageing. The ability of the family and family systems to meet the challenge of aging will be discussed, specifically with respect to countries with medium and low levels of development where incipient aging and modernization is proceeding at the fastest pace.
Presented in Session 1115: Families and Households