Living Alone in Later Life: A Global Perspective

THE PREVALENCE OF living alone during later life varies widely throughout the world, though everywhere its social relevance has grown substantially in recent decades. This living arrangement, more widespread among women than men aged over 65, is one of the most visible characteristics of societal aging currently underway. Explaining residential options among older persons and the different strategies implemented in societies of diverse cultural and historical traditions with different levels of development has important implications for our understanding and management of aging. Most of the research on this subject has concentrated on the developed world because it is there that levels of single living during later life tend to be highest as a result of the decline of intergenerational coresidence and the increase of solitary living among elderly people (Alter, Cliggett, and Urbiel 1996; Andorka 1995; Fauve-Chamoix 1996; Guinnane 1996; Pampel 1992; Ruggles 2001, 2007; Tomassini et al. 2004; Wall 1995). Less attention has been paid to the developing world owing to a lack of adequate data and lower levels of living alone. The result has been the absence of a global comparative perspective. While our goal here is not to explain all the variety in the world, we believe that a global viewpoint is essential, even if only as a starting point for the much-needed and more specific questions that will be the focus of future research on the subject. Given the potential importance for the health and wellbeing of the elderly of this particular living arrangement, its relevance for policy and for society is evident everywhere. We examine living alone among the elderly based on a large set of countries covering much of the world. The point of departure is the pioneering UN report on the subject (United Nations 2005) that was based on data taken mostly from the 1990s and made use of a number of earlier studies (Bongaarts and Zimmer 2002; Palloni 2001; Wolf 1995). Unlike this original study, our approach is based on more recent data (2010–11) and makes much more intensive use of micro census data. The heterogeneous nature of societies throughout the world makes a comparative viewpoint essential for understanding living arrangements among the elderly. Undertaking comparative empirical analysis, however, is not a straightforward task. The living arrangements of the elderly, and more specifically living alone, are the result of the preferences and resources people have, together with the constraints they face as they age. They reflect material conditions, health realities, the support they can expect to receive from their families, communities, and societies, and the extent to which they are willing or determined to live alone during the final stages of their lives. These preferences are generated within the context of prevailing cultural norms. In the developed world, this is always viewed as preference or choice, but it is still a part of prevailing norms. That is why behavior can be so different between those developed countries characterized by individualistic societies with sufficient material resources to realize preferences, and those developed countries characterized by familistic societies with few resources to implement alternative solutions. All of this takes place during a period of people's lives in which health conditions can change radically, making any general consideration of people above age 60 as elderly a misleading simplification. At least in the developed world, people entering this age group are likely to be in good health, continue to earn top-level salaries within their chosen professions, frequently prolong their working lives and even work after retirement, and are able to make significant material contributions to the wellbeing of their families. While the extent to which the situation is similar in the developing world is unknown, it is unquestionable that the health of older persons in their 60s will normally be much better than later in life. At some point in both developed and developing countries, failing health, disability, chronic illness, and decreasing personal autonomy will become more prominent. Thus, at the outset of old age, residential decisions appear to be determined overwhelmingly by personal preferences and expectations, while toward the end of life decisions are often made by others. The role of society in these decisions, expectations, and options is often decisive. Prevailing family systems (Goode 1963; Goody 1976, 1996; Hajnal 1982) and the extent to which non-family options for aging, normally related to policy and to publicly organized support and care (Shenfield 1957), are the key factors. They represent the two main options for managing the process of aging satisfactorily and are often, though frequently incorrectly, seen as mutually exclusive (Blome, Keck, and Alber 2009; Folbre and Wolf 2012; Kohli 1999; Motel-Klingebiel, Tesch-Roemer, and Von Kondratowitz 2005; Saraceno 2008). Family systems are important both because they are a key source of values and preferences in society and because of the willingness and ability of the family to provide the care needed by members as they age. Values are important because they guide people's expectations and behavior. In societies where family systems are strong and pervasive, people will perceive their own old age as tied to their families, who are the source of aid and of happiness. Where family systems are weaker, the value given to individualism and individual initiative will be greater, and this will lead to different strategies and realities when confronting personal aging (Reher 1998). Other things being equal, living alone will always be more prevalent in societies where individual values pertain and lower where they do not (Padyab et al. 2018). Early in the process of aging, even people living alone may perceive the family and family contact as a key part of the quality of their lives. When the progressive loss of personal autonomy sets in, they will expect their families to broker key decisions and to provide them with aid, often by means of coresidence with kin. Policy-based responses to aging are the other great force influencing the residential choices of the elderly. These responses are numerous, ranging from public pension systems and health care to institutions and instruments designed to provide the care needed to help people reach the end of their lives as comfortably as possible (Daly 2010; Steckenrider and Parrott 1998).1 Taken together they comprise the backbone of the welfare state wherever it exists. In historical terms, the welfare state is a relatively recent phenomenon, dating from the late nineteenth and especially the twentieth century. In many ways, the growth of these systems was an attempt to remove the role of care from the sphere of the family and transfer it to the public domain (Esping-Andersen 1999). It was part of the effort of the modernizing state to become the preeminent force in people's lives. The stronger the state's role, the smaller that played by the family both for expectations regarding care and care actually provided for the elderly. Because of the close link between modernization and economic development, it is not surprising that indicators of wealth and development tend to be positively linked with living alone and negatively linked with all forms of familial intervention.2 While some analysts see an inevitable pattern at work whereby familial interventions will eventually be replaced everywhere by policy-based responses to aging, in fact there are all kinds of patterns and combinations in existence throughout the world (Padyab et al. 2018). Any wide selection of countries will include some where there is little or no public intervention in later life and others where state-based interventions are paramount. Ultimately the process of societal aging will increase pressure everywhere, but it is likely to have a more negative effect on family-based interventions than on public ones (De Santis 2012). The family is poorly designed to confront the type of challenge posed by rapidly changing age structures and societal aging, and the resources it is able to marshal will be hard put to meet the realities of what has been called the “beanpole family” (whose living members come from several generations, but with few members in each generation) characteristic of modern life. At an individual level, there are other factors at work, some affecting the likelihood of being in a position to live alone and others affecting transitions out of single living into other forms of coresidence, whether institutional or not. Some of these factors influence the choices people make, while others affect the ways in which these preferences are shaped by forces beyond their control. Pertinent literature on household formation, based for the most part on the developed world, considers living arrangements as a means for achieving several domestic goods—privacy, companionship, domestic services, personal care, and consumption economies of scale, among others—whose specific combination is subject to the tradeoffs confronted by individuals when deciding with whom to live (Burch and Matthews 1987). Preferences related to these goods, their relative economic and social-psychological costs, and resources available to obtain them will determine the particular living arrangement chosen by individuals. With respect to the elderly, it has been suggested that the demand for these domestic goods depends on preferences for individual autonomy and privacy (Wister and Burch 1989; Palloni 2001), income levels (Michael, Fuchs, and Scott 1980; Macunovich et al. 1995; Wolf 1995), and the availability of kin (Bachrach 1980; Chevan and Korson 1975; Kobrin 1976; Pampel 1983; Soldo 1981; Wolf 1994). Since kin availability mainly refers to spouses and children, trends and patterns in marriage, divorce, and fertility constrain the stock of proximate kin available for potential coresidence and become factors relevant to explaining the phenomenon of living alone among the elderly. Several factors influencing choice come into play. Differences by sex in life expectancy among the elderly help explain the fact that in most societies, living alone is far more frequent among women than among men, leading to elderly populations filled with widows (Wolf and Soldo 1988). This is most clear in societies with reduced levels of mortality, where differences by sex in life expectancy are largest.3 Divorce is another important factor in the dissolution of unions whose frequency tends to vary sharply by levels of development and by cultural, religious, and legal contexts. These factors do not always operate in the same direction.4 It is also known that the more highly educated are the most likely to live alone. This may be because more highly educated people tend to have more material resources and may also have higher divorce rates. Yet a greater willingness to live alone among the more highly educated cannot be discarded either. In families prone to higher mortality or in those with very low fertility, the options to reside with kin will be more limited than in large families with many children (Kobrin 1976; Pampel 1983; Wister and Burch 1983; Wolf 1995; Iacovou 2000; Gaymu et al. 2006; Koropeckyj-Cox and Call 2007; Reher and Requena 2017). The supply of available kin, important as it is, may have little to do with the willingness of the family group to assume this role or with people's expectations. Income (Macunovich et al. 1995; Pampel 1992) and, particularly, people's ability to save for retirement or the existence of public pensions (McGarry and Schoeni 2000) can also influence their ability to avoid depending on others for support, especially during the first phase of old age. Later, however, matters related to health become paramount for this type of choice. There is a vast literature linking health in later life—and in particular chronic disease—with numerous socioeconomic characteristics (education, healthy and active lives, etc.), though reliable estimates of the timing of changes in health are generally beyond the control or the strategies of individuals. Once decline sets in, however, a person's material wellbeing will have much to do with the type of options he (or his kin) have available when making the key decisions for this stage of life. A number of these points are useful for understanding how the incidence of living alone changes as people age. When people are in the early stages of later life, the likelihood of living alone should increase, but after a certain age these increases will cease and declines will become the norm. The most evident reasons for these contrary trends are related to the important weight of spousal deaths—especially of men—together with divorce and relatively good health among the young elderly. All of these factors will contribute to the growing number of persons, especially women, living alone. As people grow older, however, the weight of failing health becomes the key factor as people are decreasingly able to care for themselves. This pattern has been found in some developed societies, yet we know almost nothing about it in the developing world. Even so, it is unlikely to differ substantially from the basic portrayal charted above. At all ages, especially among the oldest old, both family- and policy-based care become essential to people's wellbeing. This simple, straightforward explanation of changes as people age should be valid everywhere, though details and the exact mix of factors will differ. Figure 1 plots the percent of men and women aged 65 and over living alone in 61 societies in 2010–11. Containing data from different population counts, it provides a snapshot of the world at that time. The figure includes a broken line tracking the underlying trend and a solid line showing equal values for men and women. The variety of experiences is evident, with levels of older women living alone ranging from as high as 45–50 percent to as low as 5–10 percent and with levels among men varying from about 25 percent to below 5 percent. Despite this sex differential, a certain congruence between female and male levels is apparent (R2 = .60). The highest levels of single living appear to be concentrated in the richest societies, though this is not always the case (see Japan, for example), and the lowest levels are found in certain Asian and African countries. In most societies, levels of living alone among women are significantly higher than among men, and in some cases the differences are striking. Percentage differences tend to be highest in the richest societies, though the greatest relative differences are found in countries like Iran and Nigeria where levels among women are as much as 5 to 6 times higher than they are among men. In a small number of countries, living alone is more frequent among men than among women, though the differences tend to be small and are concentrated in Caribbean and African societies. With these exceptions, this pattern by sex is almost universal and seems to be relatively stable over time, as demonstrated by the data collected in the 1990s (United Nations 2005). These sex differentials tend to be related to the higher prevalence of widowhood among women. Our strategy for disentangling the multiple factors that determine living alone in later life is based on the extensive use of both country and micro-level data for large samples of countries from different areas of the world. Bivariable and multivariable analysis will provide empirical depth to many of the dimensions contained in the explanatory framework presented above. Our approach is based on the following premises: (i) Living alone in later life is a part of the process of population aging. (ii) Its importance will increase in much of the world in the near future, though in a limited number of mainly developed societies it could remain unchanged or even decrease. (iii) It reflects the role played by family and policy in the management of later life, together with the way health changes as people age. It also reflects individual characteristics such as preferences for living arrangements and marital status. Finally, (iv) it has implications for the health and wellbeing of the elderly, their families, and society. The analysis uses country-level data and the microdata needed to explore the individual determinants of living alone in different societies. The first data set consists of 61 countries. It includes estimates of levels of single living for persons above 65+ circa 2011 together with sex, 2011 levels of the Human Development Index (HDI), per capita gross domestic product (based on purchasing-power-parity, current international dollars and estimated for 2011), and the fertility corresponding approximately to that of the elderly women considered (i.e., TFR during 1970–75). The selected indicator for family systems (population 40–64 living in extended households) is taken from the Integrated Public Use Microdata Series, international (IPUMSi, see Minnesota Population Center 2017) and is available for 32 countries. An indicator of the differences in life expectancy by sex (female – male) has been generated based on life expectancy at age 60 taken from the UN World Population Prospects Database. Aggregate levels of living alone come from several sources: Eurostat for European countries; census data available in the IPUMSi database for African and some Asian countries; and national statistical offices for China (National Bureau of Statistics of China), Japan (Statistics Bureau of Japan), South Korea (Statistics Korea), Australia (Australia Bureau for Statistics), Canada (Statistics Canada), and the United States (US Census Bureau). Social, economic, and demographic indicators come from United Nations Development Programme (2011) for HDI, International Monetary Fund for GDP, and United Nations (2013) Department of Economic and Social Affairs for 1970–75 TFR. Given the large range of selected countries, including such demographic giants as China and India, the aggregate dataset constitutes a fairly representative sample of the entire world population. The micro census data come for the most part from the IPUMSi database and contain collected circa 2011 for 32 countries covering much of the world for come from a that is also available in Census data for and (2011) and South Korea from the for Statistics and Statistics In the analysis of our is based on the household of older people and as a straightforward living alone living with others. Our age into from to current marital into and and based on into or and The entire sample of census microdata here persons aged 65 and over living in and South and expectations about the process of household that levels of living alone among elderly people should be an important of levels of development, which in are to be with material strong family welfare and demographic We our 32 countries to their on the Human Development Index in our the is to others indicators because it different dimensions of development in a and with micro census data are into general and low levels of The high includes countries with an of development of It and a population 65+ of With a of development of another countries make the to and people The low with a of contain people 65+ living in countries. While the points between levels are the in this way are mostly with other including the one by the United Nations Development and for these of countries, including data on sample sex, age, marital and percent living alone. The only countries with micro census data and do not include those countries with only aggregate data. we make use of both and multivariable With country-level a of plots are presented including data on living alone and data on a number of the above. The of at the aggregate is by the of is mainly and for to avoid the of and that from this of aggregate data. Since the is alone living with is the most statistical for the analysis of 32 countries with micro census data. In this we several including the 32 countries and other one for each of of living alone for different sex, age, marital and for socioeconomic The of these are as of though the are available The of a are also in to the contributions to differences between of countries with different levels of development corresponding to and to other factors with specific to these populations effect or data the both in the prevalence of living alone among elderly people throughout the world and in some socioeconomic and demographic characteristics that the societal where options are given and decisions are with differentials in of people 65+ living alone, several indicators of kin and family systems are at the aggregate With respect to Figure 1 highest levels of living alone in and other developed countries. and the highest levels of elderly women and men living alone, between percent and percent for women and close to 25 percent for men. lower levels are found in and with and in the lowest of the global of living alone. high levels of female single living among elderly and not among make these two cases in with other countries with similar in part to their relatively high of older countries and and Caribbean countries here by and in an position between and societies. these the in recent in the of living alone among older persons countries and (United Nations 2005). These differentials point to the between levels of economic development and living alone. Figure the between 2011 per capita gross domestic product and of the population 65+ living alone. the between the two is relatively strong (R2 = the higher the per capita GDP, the higher the of elderly people living alone. some European countries from the pattern and above the trend levels of living alone higher than those by their of economic some extent these levels of living alone can be by higher of elderly widows in these countries. At least for elderly women, an indicator of their kin availability is the number of children they have The chosen indicator for kin availability was the fertility for the period or the census data when most of these elderly women still in their TFR is also an indicator of the timing of the demographic and an indicator of modernization (Reher As seen in Figure the lower the fertility during the first of the the higher the likelihood of living alone in later life. A decreasing linking both as as a solid (R2 = is India, and are cases the trend with levels of elderly people living alone below the from their very high fertility or for 32 countries on the aggregate between of the population 40–64 old living in extended and the population 65+ living alone are presented in Figure The indicator is a of the of at the societal level, both the of family systems other social institutions and the of strong family values in a given The figure a straightforward between the the of older people living alone tends to as the of extended among At the aggregate level, the incidence of extended in for percent of the in living alone at old Figure 5 the of the population 65+ living alone estimated from a by sex, age, marital and within the 32 countries with census tend to the (see on the In the of living alone among older persons are higher in more developed countries than in those with levels of development and, in these are higher than in the developed countries. The position of some European countries, with levels of living alone higher than from their of development, is also in a similar with an estimated for elderly people of living alone higher than in European countries or South When for the between and low levels of development in the of living alone and still the lowest the world of living alone by country and of development estimated from a by sex, age, marital and to the same weight to all countries. for Statistics and Statistics multivariable for countries at each of in to the of living alone for older people by sex and age and by marital and in different of here within the and reflect the of any given all other factors have been While there is no between men and women in their likelihood of living alone in developed countries, a lower likelihood among women is in and countries. other factors at lower levels of development the of living alone among older women is lower than among men. An of living alone by age can be in the of countries. While the increase in the likelihood of living alone as people age is much stronger at higher levels of development, everywhere the first increase with age and decrease. The point of however, is later in more developed countries than in developed ones the oldest old the likelihood of single living is still higher than it is for the in the more developed countries, but far below the same in the other two of countries. people in countries are more likely to live alone at an age than people in developed countries. for living alone by sex and age at levels of