With the rapidly aging population in China, many older people are considered to have an “empty nest” [1
]. The empty nest families mean that elderly do not live with their children or do not have a child. In fact, one-child policy in China was considered one of the major catalysts contributing to the large number of empty nest elders [2
]. Since 2016, China cancelled its one-child policy and fully implemented the policy of allowing each couple to have two children as an active response to the aging population. Nevertheless, most of the new generation would not be able to have another child even if they wanted to because of their disadvantaged positions in the increasingly competitive job market and the increased costs of raising a child [3
]. Some demographers have predicted that the ‘two-child’ policy will not lead to a baby boom [5
In the traditional Chinese family pattern, when parents cannot take care of themselves, they can live with their children and receive assistance [8
]. Recently, the “Data Analysis of the Sampling Survey of the Aged Population in Urban/Rural China 2010” report, published by China’s National Committee on Aging, indicated that the proportion of elderly people living alone was rising. More importantly, empty nest elders accounted for 49.3% of the entire older population, including 54.4% of urban empty nest elders and 45.6% of rural empty nest elders [9
]. Compared with urban areas, the empty nest phenomenon has become increasingly serious in rural areas. Urbanization is now a global phenomenon and will inevitably affect the living patterns of the elderly in rural China. Based on the National New Type Urbanization Planning (2014–2020) report, there are approximately 100 million rural people who will move into cities from the country’s farming regions [10
]. If this trend continues, the traditional family relations of being supported by children will certainly be challenged.
Whether there was a causal relationship between empty nest and Chinese elders’ health is an important issue that is well worth studying. Although there have been several studies on this topic, the findings have been equivocal and even conflicting, and the mechanisms between the two have remained unclear. Some studies have found that an empty nest has a protective effect on elders’ health [11
] and that living alone provides a great opportunity for elders to enjoy their leisure time [13
]. However, other studies have supported the hypothesis of a strong negative relationship between empty nest and elders’ health [15
], especially regarding mortality risk [17
] and changes in self-rated health [18
]. Despite these divergent findings, there is strong evidence that empty nest is associated with elders’ health.
To more thoroughly assess the effects of empty nest on elderly health, physical health (ADL), cognitive ability (MMSE) and psychological health were adopted as indicators of health in this study. Some researchers have shown that empty nest elders have a better health condition than elders who live with their children [19
]. In contrast, some studies have revealed that an empty nest has a significant adverse effect on elders’ health. A study from China showed that empty nest elders had worse self-care ability and lower mental health scores [22
]. In addition, the prevalence of deterioration in physical function of the elderly has been associated with a low income, single status, poor living conditions, number of chronic diseases and poor family functional conditions. Some studies have indicated that an empty nest is negatively related with elders’ cognitive ability [23
], especially for female, aged, rural, single elders and elders without health insurance [25
]. In general, when children move out of their homes, old parents will be more frustrated, depressed and anxious. If these symptoms cannot be eliminated in time, they can weaken elders’ immunity and, more seriously, further increase their risk of cognitive disability and Alzheimer's disease [26
]. The relation between empty nests and the psychological health of older people has attracted the attention of clinical researchers in recent years [27
]. Some Chinese studies also found that empty nest elders had more discomfort, anxiety, and depression [21
]. However, another study suggested that empty nest elders could enjoy more freedom when their children lived away from them and that they had more time to enjoy life, make friends and entertain [28
Other studies have explored the indirect relationship between empty nests and elderly health [22
]. Living resources, availability of medical treatment and social activities have been recognized as the three main potential mediating variables of this relationship. Empty nest elders were in worse condition because they did not have a sufficient pension (as farmers did not previously receive a pension in China) to support themselves; this situation led elders in impoverished mountainous regions in China to depend on their children [30
]. When children live far away from their parents, they may not have the opportunity to provide sufficient material goods and assistance for their parents [31
]. Living resources, a determining factor of health, have appeared to be significantly associated with an empty nest. Empty nest elders from low-income families have shown a tendency toward non-visiting and non-hospitalization after adjusting for need and predisposing factors [30
]. Concurrently, previous studies have shown that younger age has strong associations with more social activities, which could prevent depression [34
]. For the elderly, depressive symptoms can spontaneously appear when social support provided by work disappears and their children are unable to provide support for them in time [35
Previous studies had some limitations. First, the endogenous problem was not adequately solved by the researchers who did explore the causality between an empty nest and elders’ health. Previous studies usually analyzed the effects of baseline living status on these elders’ health, after controlling for baseline health conditions [18
], but ignored that the elderly could move in with their children if they felt unwell. Second, these studies have devoted more attention to the physical function of the elderly, although integrated studies on physical health, cognitive ability and psychological health have not been well conducted, especially in developing countries such as China [37
]. Third, the effects of an empty nest on the health of the elderly may be heterogenous, which may have caused the deviation in prior understanding of the needs of elderly care services. For example, urban elders and rural elders have different ideas, life environments and family structures, and the relationship would certainly be biased if these two groups were combined [1
]. Finally, almost no studies have examined the mechanism between empty nests and overall elderly health. Therefore, this study hypothesized that an empty nest had a negative effect on elders’ health in China, aimed to analyze the relationship between an empty nest and the overall health of the elderly, and explored the mechanisms behind how an empty nest influences the health of the elderly in urban and rural China by employing the method of instrumental variables, which was considered the best strategy to address the problem of endogeneity inherent to these types of data.
A significant difference between empty nest elders and non-empty nest elders was found in the primary data analysis. The results of Hausman’s test proved that endogeneity existed (p < 0.05); therefore, the LIML model was used to address this problem. Concurrently, the relationship and mechanism between an empty nest and elder’ health were also investigated in an urban sample and a rural sample.
3.1. Sample Characteristics
Analysis of variance was employed to compare the differences in ADL, MMSE and psychological health between empty nest elders and non-empty nest elders, and the sample characteristics are shown in Table 2
. Of the total sample, elders living with children accounted for most of the respondents. In 2008 and 2011, empty nest elders’ ADL (M2008
= 2.882, M2011
= 2.846), MMSE (M2008
= 2.938, M2011
= 2.924) and PH (M2008
= 2.876, M2011
= 2.888) scores’ logarithm were significantly higher than those of non-empty nest elders. Moreover, non-empty nest elders were about six years older than the empty nest elders (M2008
= 79.594, M2011
= 81.569), which may be another reason why the empty nest elders were healthier.
Compared with the group of non-empty nest elders, empty nest elders had significantly more years of education (M2008 = 2.689, M2011 = 2.886), were more often their own main source of income (M2008 = 0.250, M2011 = 0.281), more often had a spouse (M2008 = 0.624, M2011 = 0.589), and smoked (M2008 = 0.235, M2011 = 0.214) and exercised (M2008 = 0.367, M2011 = 0.405) more often. However, non-empty nest elders had a significantly higher income (M2008 = 8.696, M2011 = 8.724).
Based on the results of previous studies, the variables, including “living resources”, “availability of medical treatment” and “social activities”, were used as potential moderators of the relationship between empty nests and elders’ health. Specifically, a lack of living resources can definitely damage elders’ health [25
]. If the elders could not obtain medical treatment when they were sick, this would also directly influence their health. Additionally, empty nest elders’ ability to engage in social activities may be related to their psychological health. A previous study showed that the psychological health of empty nest elders cannot be guaranteed and that their communications with friends could also be reduced because of a lack of support from social networks [60
]. The results of the statistical analysis showed that inadequate living resources (M2008
= 0.759, M2011
= 0.772) and underutilization of medical treatment (M2008
= 0.955, M2011
= 0.909) were significantly more prevalent in empty nest than in non-empty nest elders. However, empty nest elders engaged in more social activities (M2008
= 2.949, M2011
= 2.924) in this sample. Empty nests may have an indirect influence on elders’ health through these three moderators.
3.2. Empty Nest Elders’ ADL
, Table 4
and Table 5
present the findings of the LIML analyses compared with the 2SLS analyses. In this study, the F values in the weak identification test of the LIML model were 13.547 for the urban group and 46.875 for the rural group, which means there was no need to worry about weak instruments. Additionally, when the number of instrumental variables is more than that of endogenous variables, over-identifying restriction tests should be required. In this study, the p
values of over-identifying restriction were 0.863 for the urban group and 0.138 for the rural group, meaning there was no problem with over-identification. After resolving the endogeneity problem, the results of the second stage of the LIML model showed that urban empty nest elders had significantly poorer health compared with non-empty nest elders (β = −0.617, p
< 0.001). Furthermore, similar results were obtained in the 2SLS analyses (β = −0.616, p
< 0.001), which further verified the robustness of the regression. When elders lived with their children, they could receive more daily care from their children; however, empty nest elders had to rely on themselves, and this placed a substantial amount of pressure on their health. In the rural sample, an empty nest also had a negative influence on elders’ ADL (β = −0.238, p
< 0.001). In addition, education years (β = −0.002, p
< 0.01) and exercising (β = 0.051, p
< 0.001) were also significantly associated with rural elders’ ADL.
In the urban sample, elders’ ADL scores seemed to decline gradually as age progressed (β = −0.008, p < 0.01), that is, young elders had a better health condition. However, income (β = −0.050, p < 0.001) and being ill in the past two weeks (β = −0.081, p < 0.001) were negatively related to elders’ health, whereas having a spouse (β = 0.159, p < 0.001) and exercising (β = 0.089, p < 0.001) were positively related to elders’ health.
3.3. Empty Nest Elders’ Cognitive Ability
The effects of an empty nest on urban and rural elders’ MMSE scores are shown in Table 4
; as expected, an empty nest had a negative relation with cognitive ability in urban elders (β = −3.585, p
< 0.001) and rural elders (β = −2.438, p
< 0.001). Furthermore, in the urban sample, elders who were male (β = 0.224, p
< 0.01), younger (β = −0.045, p
< 0.01) and those whose major source of income was mainly themselves and their spouses (β = 0.260, p
< 0.01) had significantly higher cognitive ability. Meanwhile, spousal support (β = 0.931, p
< 0.001) and exercising (β = 0.326, p
< 0.001) also influenced cognitive ability significantly. However, elders who had been ill in the past two weeks had significantly poorer cognitive ability (β = −0.206, p
< 0.01). In the rural sample, an empty nest had a dramatic positive impact on elders who were male (β = 0.291, p
< 0.001), younger (β = −0.046, p
< 0.001), married (β = 0.540, p
< 0.001), whose main source of income was themselves (β = 0.226, p
< 0.001) and who had a low income (β = −0.195, p
< 0.001). Besides, elders who had not been ill in the past two weeks (β = −0.124, p
< 0.001) and exercising (β = 0.204, p
< 0.001) had higher cognitive ability.
3.4. Empty Nest Elders’ Psychological Health (PH)
Psychological health is another important part of elders’ health, and the effects of an empty nest on urban and rural elders’ psychological health are shown in Table 5
. The results of the LIML regression indicated that an empty nest had a significant influence on urban elders’ PH (β = −3.751, p
< 0.001) and rural elders’ PH (β = −2.595, p
< 0.001). The results of the 2SLS regression further demonstrated that an empty nest had a significantly negative influence on elders’ PH. Meanwhile, we also found that an empty nest had a greater effect on urban elders’ PH than on rural elders’ PH.
In the urban sample, elders who were older (β = −0.037, p < 0.001), single (β = 0.979, p < 0.001), sick in the past two weeks (β = −0.238, p < 0.001) and had a higher income (β = −0.287, p < 0.001) had significantly poorer PH. However, exercising (β = 0.309, p < 0.001) and having a main source of income of themselves and their spouses (β = 0.303, p < 0.01) improved the PH of elders. Similar results were also found in the rural sample.
3.5. Heterogeneity Analysis
The previous results showed that an empty nest had an effect on elderly health. From the results of regressions, elders with different gender, age, marital status and exercise status had significant differences in health. Hence, interaction analysis was performed to estimate the association between two factors. In Table 6
, significant differences between rural and urban elders were not observed with respect to gender (p
> 0.05) and age (p
> 0.05), but they were noted among marital status (p
< 0.001) and exercise status (p
< 0.001). The proportion of rural elders who were single (p
< 0.001) or did not exercise (p
< 0.001) was significantly higher than that of urban elders. In terms of gender, the proportion of women who were over 75 years of age (p
< 0.001), single (p
< 0.001), or did not exercise (p
< 0.001) was significantly higher than that of male elders. For senior elders, the proportion of elders who were single (p
< 0.001) or did not exercise (p
< 0.001) was significantly higher. For single elders, the proportion of elders who did not exercise was significantly higher (p
showed that characteristics of the individual were different among different subgroups, it might lead to the effects of an empty nest on subgroups were different. Therefore, the effect of an empty nest on elders’ health among different subgroups was further analyzed.
3.5.1. Activities of Daily Living (ADL)
showed that in the urban sample, in terms of gender, female empty nest elders had significantly lower ADL scores (β = −0.472, p
< 0.01), but for male elders, regardless of whether they lived with their children, their ADL scores were not related to the empty nest variable (p
> 0.1). For age, an empty nest had a weak association with 65–74 years of age elders’ ADL (p
> 0.1), but had a strong association with 65–74 years of age elders’ ADL (β = −0.681, p
< 0.01). In terms of marriage, an empty nest was not associated with elders’ ADL when they lived with a spouse (p
> 0.1), but for single (including unmarried, divorced or widowed) elders, an empty nest had a negative impact on their ADL (β = −0.448, p
< 0.001). In addition, urban empty nest elders who did not exercise had poor physical health (β = −1.031, p
< 0.01), whereas for elders who did exercise, an empty nest was not related to elders’ ADL. In the rural sample, gender, age, marital status and exercising still had a significant influence on elders’ ADL. Elders who were female (β = −0.237, p
< 0.001), were aged more than 75 years (β = −0.287, p
< 0.001), were single (β = −0.200, p
< 0.001) or did not exercise (β = −0.320, p
< 0.001) were more negatively affected by an empty nest.
3.5.2. Mini-Mental State Examination (MMSE)
showed that in the urban sample, an empty nest could impair the cognitive ability of elders who were female (β = −2.727, p
< 0.001) and 65–74 years of age (β = −3.966, p
< 0.001). However, in terms of marriage, an empty nest was weakly associated with the MMSE scores of married elders (p
> 0.05), whereas it had a significantly negative effect on the MMSE scores of single elders (β = −2.623, p
< 0.001). For elders who did not exercise, an empty nest had a strong relationship with their MMSE scores (β = −6.175, p
In the rural sample, an empty nest had a significantly negative influence on elders, male (β = −2.672, p < 0.001) and female (β = −2.355, p < 0.001). Meanwhile, an empty nest had also a significant negative influence on the MMSE scores of elders who were over 75 years of age (β = −2.906, p < 0.001) or single (β = −2.171, p < 0.001). Moreover, regardless of whether elders exercised (β = −1.550, p < 0.01) or not (β = −2.738, p <0.001), an empty nest impaired their cognitive ability.
3.5.3. Psychological Health (PH)
In Table 7
, Empty nests had a strong adverse impact on elders’ psychological health, especially for female (β = −3.008, p
< 0.001) over 75 years of age (β = −4.315, p
< 0.001) and single (β = −2.749, p
< 0.001) urban elders; however, for married urban elders, the influence of an empty nest was positive (β = 2.315, p
< 0.01). For elders who did not exercise, an empty nest had s significant negative influence on elders’ psychological health (β = −6.664, p
In the rural sample, over 75 years of age (β = −3.121, p < 0.001), or single (β = −2.311, p < 0.001) showed negative associations with an empty nest. The effect of empty nest on elders’ psychological health was no significant difference between male (β = −2.807, p < 0.01) and female (β = −2.538, p < 0.001). Moreover, Elders who did not exercise (β = −2.967, p < 0.001) were more negatively affected by an empty nest.
3.6. Mediating Effect Analysis
Based on the results above, an empty nest had a direct negative effect on elders’ health. One thing worth noting is that an empty nest is a choice of living arrangement, and many factors may be significantly associated with it. Therefore, we hypothesized that there may be a mediating effect between an empty nest and elders’ health. Based on previous studies, this study aimed to explore mediating variables from the material well-being, health services and social activities perspectives. First, material well-being assistance was considered, as if the elders lived alone, they may receive less material assistance from their children. Furthermore, elderly people are accustomed to saving money, and they may only buy basic living goods to keep from starving. However, if elders did not have enough money or basic living goods to remain healthy, then an empty nest could directly impair their health. Second, the utilization of medical services plays a crucial role in keeping old people healthy. If elderly people lived alone, their accessibility and utilization of medical services could be reduced. For example, it may be inconvenient for them to go to hospitals without being accompanied by their children, and the distance to the hospital and the complicated medical procedures may present a challenge and barrier to receiving medical care. Third, social activities were included because for elderly people, participating in social activities not only improves their health, but more importantly, they can receive spiritual support by making friends with other people, further promoting a good state of mind. If they live alone, they may have to perform more housework and thus have less time to participate in social activities.
As shown in Table 8
, an empty nest not only had a direct influence on elders’ health but also had an indirect influence on elders’ health. For urban elders, “living resources” (Z = −2.715, p
< 0.01) and “medical treatment” (Z = −2.527, p
< 0.01) had significant mediation effects between an empty nest and elders’ ADL, accounting for 10% of the total effect. For rural empty nesters, “living resources” (Z = −3.211, p
< 0.001), “medical treatment” (Z = −2.082, p
< 0.05) and “social activities” (Z = −3.371, p
< 0.001) also had a strong mediating effect, accounting for 24% of the total effect. In addition, “social activities” played the most important role, accounting for 15.87%.
Furthermore, these three mediator variables also had significant mediation effects on the relationship between an empty nest and the MMSE and PH scores of elderly people, although the degrees of the effect declined. In urban and rural elders, “living resources” (in rural, 3.61%; in urban, 4.36%) and “medical treatment” (in rural, 4.12%; in urban, 4.84%) acted as significant mediators. In the rural sample, the “social activities” variable exerted a main mediating effect (Z = −3.381, p < 0.001, M = 9.08%). In terms of elders’ PH, “social activities” showed a main mediating effect between an empty nest and rural elders’ PH (Z = −3.361, p < 0.001, M = 5.79%). In contrast, in urban elders, “medical treatment” contributed to the main mediating effect (Z = −3.616, p < 0.001, M = 8.33%).
In this study, sample characteristics showed that empty nest elders’ ADL, MMSE and PH scores’ logarithm were significantly higher than those of non-empty nest elders. These results may be influenced by selection bias, as healthier elders were capable of taking care of themselves and living independently may have been their best choice for enjoying the leisure of retirement, whereas elders with poor health had to accept assistance from their families. After selecting “whether old parents talk with their family when they are upset” and “ownership of housing” as effective instrumental variables to address the endogeneity problem, we explored the causality between an empty nest and elderly health. To the best of our knowledge, this may be the first study to explore whether an empty nest influences the overall health of Chinese urban and rural elders, and this study also identified the mechanisms between an empty nest and elderly health.
It was found that an empty nest had a significant negative influence on urban (β = −0.617, p < 0.001) and rural (β = −0.238, p < 0.001) elders’ physical health, especially for disadvantaged elders (female, older, single). Moreover, urban (β = −3.585, p < 0.001) and rural elders’ (β = −2.438, p < 0.001) cognitive ability was affected by an empty nest. Exercising could offset the negative influence of an empty nest on cognitive health. Furthermore, the results revealed that an empty nest had a significantly negative influence on urban elders’ PH (β = −3.751, p < 0.001) and rural elders’ PH (β = −2.595, p < 0.001).
Previous studies showed that living arrangements were relevant to elderly health [19
], and some studies demonstrated that elderly health was a key factor influencing elderly living arrangement [36
]. In this study, solving the endogeneity problem was essential. However, the interaction between elderly health and living arrangement has still not received enough attention in previous studies. To assess the true effects of an empty nest on elderly health, the instrumental variable method, which is an appropriate solution to addressing the endogeneity problem, was used in this study. A recent study used “the first child of the old person is a son and has survived” as an instrumental variable to resolve the endogeneity between an empty nest and elderly health, but along with changes in procreation and conception and the decreasing number of newborns, this instrumental variable does not have long-term applicability [25
]. “Talks to family” and “ownership of housing” have more extensive applicability.
This study examined the causality between empty nests and elders’ physical health, cognitive ability and psychological health using a large-scale, nationwide longitudinal survey of elders in China in 2008 and 2011. A key finding of this study was that an empty nest had a significantly negative influence on urban and rural elders’ ADL, MMSE and PH scores. Elders who lived with their children could receive daily care, economic support and spiritual consolation from their children and therefore had better health conditions [16
]. However, empty nest elders received less help from their families and society, especially in the case of certain elders, such as female, older, single elders and who may receive less help and thus have relatively worse health conditions.
Regarding elders’ cognitive ability, when elders lived with children, they received daily life assistance from their children, and their communication with their family may have also helped mitigate the deterioration in cognitive ability. However, when elders lived alone, they spent more time and energy taking care of themselves, and their cognitive ability received less attention. Another important finding was that exercising could offset the negative influence of an empty nest on health. Exercising was also a type of social interaction, as during exercise, elders could communicate with each other and continually accept new things, which may help in reducing deterioration in cognitive ability [65
]. Furthermore, the findings also revealed that an empty nest had a significantly negative influence on elders’ psychological health, especially that of urban elders. A possible reason for this finding may be that when elders began to live alone, they might feel abandoned by their children. It could produce emotional problems, including discomfort, anxiety, pain, and even depression [21
The effects of an empty nest on elders’ health differed between male and female elders. For females, elders who were older (aged 75 years and over), single and did not exercise were more significantly influenced by an empty nest, and this may be because they were in a disadvantaged position regarding their health. For elderly people with an older age, it is clear that they are relatively vulnerable to illness due to a deterioration in physiological function [68
]. In addition, the old rural elderly would be more harmed by an empty nest. For rural elders, farming had been an important part of their daily lives, and living with children would directly affect old men’s workloads, further influencing their health. In China, females usually provide daily care services, and when elderly people live alone, they may receive less assistance from their families [69
]. From the perspective of family structure, empty nest couples have to depend on each other, and thus spouses play a very important role in providing daily care and spiritual support [71
]. Compared with married empty nest elders, single empty nest elders may live alone because they had no family members around to provide emotional support [73
]. Although China has made great economic advances, its social security system for elderly people remains under-developed [74
]. To prevent tragic outcomes in empty nest elderly people, the screening and treatment of depressive symptom should be strengthened; additionally, the responsibility of the family should be intensified because the family still provides an important role in supporting elders, especially because a specific elderly care system had not been established in China [75
There was a difference in the effects of an empty nest between urban elders and rural elders. Compared with rural elders, urban elders’ health was more influenced by an empty nest. In rural China, the long-term farming lifestyle, although the elders are out of the labor market, have made them used to farming work such as planting crops, which not only provides food but also keeps them healthy. In addition, rural people may have closer neighborhood relationships than urban people, which may help alleviate the psychological pressure of rural elders [76
Another finding was that an empty nest also had an indirect influence on empty nest elders’ health through living resources, availability of medical treatment and social activities. The exploration of potential mediating effects was another important contribution of this study. To date, there have been no studies investigating the effect of the mediator variables “living resources”, “availability of medical treatment” and “social activities” on the relationship between an empty nest and elders’ health. The results showed that all three variables played important mediating roles in the relationship between empty nest and elders’ health. Results of Sobel test reported that in urban elders, the effects of the mediator variables “living resources” and “availability of medical treatment” were greater on the relationship between an empty nest and elders’ health; while in rural elders, the effect of mediator variable “social activities” was more important.
This study provided empirical evidence regarding empty nest elders: empty nest elders should not only receive attention in terms of their material needs but also regarding their participation in more social activities for relaxation and entertainment. In the long term, an elderly care system should be established as soon as possible in China to take care of the increasing number of elderly people, given the lower birth rates and the longer life expectancies in the future. It is clearly impossible to sustainably enhance the quality of life of elderly people by relying only on family members. This study also emphasized the importance of developing a social security system for aged people that not only provides economic support and medical aid for empty nest elders but also pays attention to empty nest elders’ spiritual needs and social activities to allow elderly people to maintain a good health status.
There are some limitations of this study that should be acknowledged. First, the data were based on the best available data on the population; the analysis used convenient data rather than data specifically designed for empty nest elders, and thus related details may not be included. Second, mediator variables preclude subjectivity and variability. In this study, only three variables were chosen as mediator variables based on the data and our knowledge, and there may be other potential variables that were neglected.