Understanding the Social and Economic

Determinants of Elderly Health in India: An

Empirical Analysis
Submitted by – Aditya Suthar, 24227707101
Abstract:
India as a nation is undergoing a demographic transition, with its elderly population growing
at a rapid pace. The elderly population (60 years and above) is projected to more than
double from 100 million in 2011 to 230 million by 2036. It therefore becomes crucial to
understand the factors that shape elderly health, for both policy and welfare planning.
This study examines and analyses the social as well as economic determinants of self-
perceived health among individuals aged 60 years and above, using data from the 75th
Round of the National Sample Survey (NSS). An Ordered Logit Regression Model is employed
to analyse the relationship between perceived health status and key factors such as
economic independence, living arrangement, and family composition, while controlling for
age, gender, education, sector, physical mobility, and marital status. The results reveal that
economic independence is strongly associated with better perceived health. The numbers of
sons show a more positive effect than the number of daughters, reflecting the traditional
norms within Indian families. Control variables such as age, gender, education, sector and
physical mobility also influence perceived health. The findings highlight the need for
integrated policies that promote financial autonomy, community-based elderly care, and
gender-inclusive caregiving practices.

1. Introduction
The world has seen a decline in both fertility rates and mortality rates, along with an
increase in the life expectancy, leading to an increase in the share of elderly population. In
East and North Asia, over a third of the population is expected to be 60 years or older by
2050, whereas in North and Central Asia one in four persons will be 60 years or older.
(UNESCAP, 2023).
India is undergoing a rapid demographic transition, with the elderly population (60 years and
above) projected to more than double from 100 million in 2011 to 230 million by 2036. This
transition indicates that by 2036, nearly one in seven Indians will be aged 60 years or older,
representing a fundamental restructuring of the country's population composition. (Press
Intelligence Bureau, 2025).
This transition, while signifying a crucial and positive development in terms of improving
healthcare and longevity, also presents some challenges. An important challenge that India
faces in this context is the challenge of providing adequate care and support for the elderly -
this includes taking care of both the physical as well as emotional and social well-being.
Indian society has undergone gradual changes owing to urbanization, modernization,
increasing individualism, consumerism and migration. The traditional living and caregiving
arrangements have weakened as a result. This raises an important concern – that the elderly
may face isolation, loneliness as well as economic insecurity.
Health among the elderly is a complex issue – influenced not by a single factor but by
multiple interrelated ones. Among these, the social and economic factors take a seat as the
most crucial of them all – especially in a country like India, where good healthcare facilities
and social protection are a bit limited.
Economic independence gives adults (especially elderly people) a sense of security – not
only in living their life comfortably, but also in a healthier manner by providing the power to
purchase medicines and avail good health facilities on their own. This autonomy helps them
build a healthier lifestyle.
Additionally, social support (or the lack thereof) can also affect the health among the elderly.
Social ties provide a sense of comfort – both mentally and emotionally which directly or
indirectly improves health. Strong social ties are linked to better physical health, higher life
satisfaction, reduced loneliness, and fewer depressive symptoms (Kahn, Hessling & Russel,
2003). Conversely, lack of such support may lead to poor health outcomes, loneliness and a
more depressive lifestyle.
These issues are both real and significant. Therefore, the need to study these effects in detail
becomes an important issue.
The objectives of the paper are as follows –

• To understand the relationship between economic independence and elderly health

in India. (The economic question)
• To see how living arrangements and family composition – measured through number
of sons, numbers of daughters – relate to elderly health. (The social question)
Our analysis draws upon data from the 75th NSS Round on Social Consumption: Health to
estimate these effects among individuals aged 60 years or more.
By analysing both social and economic factors, the study attempts to contribute to the
existing literature on the issue by providing a broad understanding of ageing and health in
India. The paper highlights the fact that despite the change in social and economic structure,
factors such as living arrangement, economic independence as well as family composition
continue to play an important role.
2. Literature Review
Several studies have examined the health status of the elderly and the various factors that
influence it. The existing literature has highlighted the role of both social and economic
factors in determining elderly health. This section aims to highlight a few such findings,
which provide strong evidence in favour of the variables used in this paper.
Evidence from the English Longitudinal Study of Ageing (ELSA) highlights a clear wealth
gradient in self-reported health among individuals aged 50 or more. Median wealth among
those in good health was nearly three times higher than among those reporting poor or fair
health, and this disparity persisted across all age groups. (Banks et al., 2003).
Consistent with this, research in Pakistan found that both economic independence and
social support were positively associated with the overall quality of life in older adults (Naz
et al., 2014).
Household income also plays an important role. Arber and Ginn (1993), using data from the
General Household Survey (GHS), showed that higher household income was linked to lower
levels of self-reported poor health and functional limitations, even after controlling for other
socioeconomic characteristics.
Social support has been examined as another key determinant of elderly health. Studies
demonstrate that stronger social ties are linked to better physical health, higher life
satisfaction, reduced loneliness, and fewer depressive symptoms (Kahn, Hessling, & Russell,
2003). Similarly, Berkman and Glass (2000) report that perceived availability of support can
influence physiological responses, such as blood pressure under stress, while social
engagement contributes to maintaining cognitive functioning in later life.
Number of children have an interesting impact on elderly health. Following is the result of a
paper written on the subject - In 15 of 24 countries, having more children (especially 4+)
correlated with poorer health. The strongest links were for depression (mental health) and
chronic conditions (physical health). Few or no children predicted poorer health in only a
minority of countries (e.g., China, Israel, France) (Antczak, Quashie, Mair, Arpino, 2023). In
simpler words, this paper suggests having fewer children is better from health perspective.
Living arrangements matter. Older adults without a spouse have been shown to experience
worse health outcomes and greater levels of disability (Tyagi & Paltasingh, 2017). Moreover,
increasing age itself has consistently been associated with poorer health status and declining
quality of life (Tyagi & Kapoor, 2010; WHO, 2005). Living arrangements of the elderly were
found to be a significant predictor of their perceived health status even after controlling for
other demographic and socio-economic confounders. (Ghosh, S., & Husain, Z., 2010)
At a broader level, demographic shifts underscore the urgency of addressing elderly health.
By 2025, nearly three-quarters of the global elderly population will reside in developing
countries, where health systems are strained (Shrivastava, Shrivastava, & Ramasamy, 2013).
This raises important challenges for preventive healthcare and social policy.
The rural-urban divide is also significant. Recent work by Das, Kundu, and Hossain (2023)
shows that older adults in rural settings are more vulnerable in terms of both healthcare
access and food security compared to their urban counterparts. In India, this disparity is
particularly stark: a large share of deaths in rural areas still occur without medical attention,
with the poor, elderly, and infants being the most at risk (Gupta, 2004). Another study was
carried out by house-to-house survey of all people aged over 60 years in an urban slum and
a village in the field practice area of a teaching hospital. A large number of unmet health
needs, such as unoperated cataract, uncontrolled hypertension, uncorrected hearing
impairment and tobacco use, exist in marginalized groups. Health interventions for these are
needed in developing countries. Preventive services such as tobacco cessation campaigns
among the elderly should also get priority (Thakur R, Banerjee A, Nikumb V, 2013).
Gender inequality is also a factor that comes into play when looking at elderly health. There
is evidence for older women being more vulnerable to mental health problems, elder abuse,
and experiencing a lower quality of sexual life. Despite this, they have lesser access to health
care services and receive low-quality long-term care (Nair S, Sawant N, Thippeswamy H,
Desai G, 2021)
For marital status however, the evidence isn’t all that strong. While the proportion of
unmarried persons in the near future is likely to be higher than today, findings suggest that
certain unmarried groups create a social environment that compensates (in ways that
improve health) for the loss or absence of a spouse so that recent changes in marriage
patterns need not have adverse effects on the future health status of the elderly (Goldman,
Korenman, Weinstein, 1995)
Education, in some way also seems to impact the elderly health. Education provides
opportunities for social interaction and the formation of new friendships, which are vital for
combating social isolation and loneliness among older adults (Zhang, Rahim, 2024).
The studies taken together highlight the multifaceted nature of elderly health, influenced
and shaped by both economic and social factors. While these studies provide valuable
insights into these variables and their importance, much of the existing research has focused
on high-income countries or small samples. This leaves a gap in understanding the Indian
context. Indian society – characterised by traditional family support systems needs to be
studied accordingly. The gradual transformation of traditional support system provides an
opportunity to reassess how living arrangements, family composition and economic
independence influenced perceived health among the elderly.

2.1 Gaps in the Literature and Contribution

The existing literature, including some studies from India, have established a strong link
between economic factors (like wealth and income) and elderly health. Similarly, the
importance of living arrangement is also documented.
However, a significant gap remains in the analysis of family composition. While existing
studies have done a tremendous work in quantifying various factors, they have largely
overlooked the distinct and differential impact of the number of sons versus the number of
daughters on elderly health.
This distinction is particularly crucial in the Indian social context, where sons have
traditionally been viewed as the primary providers of old-age security, while daughters' roles
have been culturally defined differently. It is unclear from the existing literature how these
deeply-rooted, gendered expectations for caregiving are reflected in the self-perceived
health outcomes of the elderly.
This paper aims to fill this specific gap by using the 75th NSS Round to (1) confirm the
importance of economic independence, and (2) provide empirical analyses to compare the
association between the number of sons living and the number of daughters living on the
perceived health of the elderly in India.

3. Data and Methodology

3.1 Data
This study uses data from the 75th round of the National Sample Survey (NSS), conducted by
the Ministry of Statistics and Programme Implementation (MOSPI) during July 2017 to June
2018. The survey, titled Household Social Consumption: Health (Schedule 25.0), was
designed to collect nationally representative information on the health sector in India.
3.1.1. Purpose and scope
The primary objective of the survey was to generate reliable estimates on morbidity,
hospitalisation, utilisation of public and private healthcare services, and expenditure on
medical treatment. It also collected information on the status of individuals aged 60 and
above, covering aspects such as health conditions, economic independence, and living
arrangements - which makes it particularly suitable for analysing elderly health outcomes.
3.1.2 Sampling and coverage
The NSS follows a stratified multi-stage sampling design, with villages (in rural areas) and
urban blocks (in urban areas) serving as the first-stage units. The ultimate stage units were
households, from which all members were enumerated. The survey covered the entire
Indian population, excluding only a few inaccessible regions. Data were collected over four
sub-rounds: July-September 2017, October-December 2017, January-March 2018, and April-
June 2018, to ensure seasonal representativeness.
3.1.3 Unit of analysis
The unit of analysis for this study is the individual aged 60 years or above. The dataset
provides detailed information on each household member’s demographic and
socioeconomic characteristics, health status, and access to care.
In particular, we used the Level 3 - Demographic particulars of household members and
Level and Level 11 - Particulars of economic independence and state of health of persons
aged 60 years and above. Level 11 was used to merge 3 more variables (Gender, General
Education and Marital Status) into our data for a more comprehensive and robust analysis.

3.2 Data Merging and Preparation

For the purpose of combining the demographic information with the elderly health module,
The Demographics and Health datasets from NSS 75th Round were merged at the individual
level. First, the common household and person identifiers present in both files were
identified. Based on this, a unique individual ID was constructed by concatenating FSU,
district, household sample codes and the member serial number. This unique ID was
generated separately in both datasets, and then used to perform a left join of the health
dataset with selected demographic variables (gender, marital status, general education).
After merging, duplicate IDs and missing values were checked to ensure that the merge was
successful and that each individual in the health dataset received the correct demographic
attributes. The merged dataset was then used to construct the final analytical sample for
regression.

3.3 Variables
From the entire data, there were in total 11 variables of interest from regression
perspective.

Dependent Variable
Own Perception of Current State of Health
This variable represents the respondent’s self-assessed health status, collected in three
categories: 'poor', 'good/fair', and 'excellent/very good'. To switch the original coded order,
the variable was recoded for the analysis such that poor < good/fair < excellent/very good,
ensuring that higher values correspond to better health. All interpretations in this paper are
based on this recoded variable.

Independent Variables
1. State of Economic Independence
This variable indicates the extent to which an elderly individual is financially self-sufficient. It
is classified into three categories — fully independent, partially independent, and not
independent. This variable captures the level of financial autonomy and the ability to meet
personal needs without relying on others.
2. Living Arrangement
This variable identifies the composition of the household and with whom the elderly person
resides. It provides insights into the social environment and available family support. The
original NSS variable had seven categories, which were collapsed into three for clearer
interpretation and to avoid very small cell counts. The categories are as follows:
1 = living with spouse
2 = living without spouse but with other household members (such as children or relatives)
3 = living alone
3. Number of Sons Living
Represents the total number of sons currently alive. The variable ranges from 0 to 10. This
variable serves as a proxy for family composition and potential support systems within
traditional Indian households.
4. Number of Daughters Living
Represents the total number of daughters currently alive, ranging from 0 to 10. This
inclusion allows the analysis to explore whether gender differences in family caregiving roles
affect elderly health.
Control Variables
To control for individual heterogeneity and contextual differences, the following variables
were included:
• Age: Measured as a continuous variable ranging from 60 to 115 years.
• Sector: Indicates whether the respondent resides in a rural (1) or urban (2) area.
• Physical Mobility: Captures the physical ability of the elderly, coded as —
1 = Physically immobile (confined to bed)
2 = Confined to home
3 = Able to move outside but only in a wheelchair
4 = Physically mobile
• Gender: Categorised as male (1), female (2), and transgender (3).
• Marital Status: Classified as never married (1), currently married (2), widowed (3),
and divorced/separated (4).
• General Education: Captures the educational attainment of the respondent, coded
from 0 (illiterate) to 16 (postgraduate and above)

3.4 Methodology
Missing Data Handling: This study employed listwise deletion for the analysis. This means
that for all the models analysed in the study, any individual with a missing value for any
variable was excluded from that specific regression. Importantly, such missing values were
present in number of sons living (around 1000 missing values) and number of daughters
living (around 4000 missing values).
The analysis was conducted in two stages.
First, descriptive statistics were used to examine the distribution of key variables and to
provide an overview of the socioeconomic and demographic characteristics of the elderly
population.
Second, to explore the association between perceived health status and the explanatory
variables, an ordered logistic regression model was employed, given the ordinal nature of
the dependent variable. The model estimates the likelihood of reporting better perceived
health as a function of economic independence, family structure, and living arrangements,
while controlling for other covariates.
All statistical analyses were carried out using the R software environment.
Why Ordered Logit? The dependent variable, Own Perception of Current State of Health, is
ordinal in nature with three categories - “poor”, “good/fair” and “excellent/very good”. The
jump from “poor” to “good” is not the same type of change as the jump from “good” to
“excellent”. Treating this variable as simple numeric or as purely categorical would risk losing
the actual structure within the response scale. Ordered logit is therefore more suitable
because it formally accounts for the ordinal nature of the outcome and estimates the
likelihood of moving to a higher category of perceived health.
To examine the determinants of perceived health status among the elderly, three separate
regression models were estimated. Each model uses perceived health status as the
dependent variable, while incorporating both social and economic explanatory factors.
The three models differ in their key social variable of interest:
• Model 1 - includes the number of sons as the main social variable,
• Model 2 - includes the number of daughters, and
• Model 3 - focuses on living arrangement.
In all models, economic independence is included as the main economic variable. Additional
demographic and socioeconomic characteristics, such as age, gender, education, marital
status, and place of residence are controlled for. This approach enables the study to assess
how different social and economic dimensions relate to the perceived health status of
elderly individuals.

3.5 Data Limitations

• A substantial number of missing values for the number of daughters variable (around
4000 missing).
• Variables used, especially for the social question are mere proxies and thus may not
be the whole truth.
• Since the data is cross-sectional in nature, it fails to capture causal relationships or
changes in health status over time.
• Variables like mental health status, stress or loneliness – which are crucial
components of elderly wellbeing could’ve helped form a better analysis.
• The dependent variable depends on self-assessment, which can be subjective in
nature and influenced by social norms and personal expectations.

4. Results
4.1 Descriptive statistics
Following is the table that provides the descriptive statistics of the dependent variable,
along with the main independent variables.
Table 1: Descriptive Statistics of Key Variables

Variable Categories Frequency Percentage

Own Perception Excellent/Very good 3529 8.3%
of Current
Health Status
(Dependent
Variable) Good/Fair 29423 68.8%
Poor 9803 22.9%

State of Fully dependent on others 20222 47.3%

Economic
Independence Partially dependent on others 10129 23.7%
Not dependent on others 12404 29%

Number of Sons 0 2668 6.4%

Living
1 15248 36.6%
2 15350 36.9%
3 5698 13.6%
4 or more 2632 6.3%

Number of 0 6753 17.4%

Daughters
Living 1 14827 38.1%
2 10808 27.8%
3 4295 11%
4 or more 2118 5.6%

Living Living with spouse 29128 68%

Arrangement
Living without spouse (others) 12792 30%
Living alone 842 2%

In addition to these main variables, certain control variables were also chosen. A slight
majority of the respondents were male (51.2%), while 48.8% were female. Most of the
elderly were currently married (68.6%), followed by widowed individuals (30.5%). About
55% resided in rural areas, indicating a slightly higher rural representation in the sample. In
terms of physical mobility, nearly 90% were physically mobile, while only a small proportion
reported mobility restrictions. Educational attainment was generally low, with almost half of
the respondents (47.2%) being illiterate. These findings highlight the demographic and
socioeconomic characteristics of the elderly population in the study, reflecting substantial
variation across gender, marital status, and education levels.
4.2 Exploratory Analysis
Among our main variables of interest, exploratory plots were generated for the number of
sons and number of daughters to see whether their relation with the dependent variable is
linear or not.
First, we plotted the number of sons living against the average code of perceived health. The
result is the following -

The plot indicates that the relation is non-linear in nature. Compared to having 0 sons,
having 1 son has a higher average code (which means better health outcome) but this
average reduces as the number of sons increase. For the case after 7 sons, the increase is
simply due to the fact that the number of observations for these is very low. To test this, a
smoothed curve was also created.

The smoothed curve also confirms that the relationship is non-linear.

For the number of daughters, similar plot was made and the results are the following -
This plot also indicates a similar non-linear relationship. It is notable that the average code,
despite sharing a non-linear relation – doesn’t seem to change much as the number of
daughters increase. A sharp decline in this number after 7 daughters is also due to the very
fact that number of observation under that category is very low.
In contrast, economic independence and living arrangement were already categorical
variables by design, therefore no functional form assessment was required for those.
The plots are primarily illustrative of how the variables relate to the dependent variable. The
non-linear patterns we see here (especially for number of sons) are confirmed again in the
regression results, which is the next section.

4.3 Regression Analysis

The first regression model examines the association of perceived health status of the elderly
with the state of economic independence and number of sons living, controlling for mobility,
demographic, and socioeconomic characteristics. The odds ratio (OR) with 95% confidence
interval (CI) of ordered logit regression of predictor variables on perceived health status is
presented in the Table below.
Table 2: Ordered Logit Regression Results - Economic Independence and Number of Sons

Independent Variables Odds Ratio Coefficient (β) Std. Error p-value

State of Economic Independence
(Ref: Fully Dependent on others)
Partially Dependent 1.53 0.42 0.0287 0.0000
Not Dependent 2.47 0.90 0.0321 0.0000
Number of Sons Living (Ref: 0 sons)
1 Son 1.44 0.36 0.0482 0.0000
2 Sons 1.37 0.31 0.0483 0.0000
3 Sons 1.24 0.22 0.0538 0.0000
4 or More Sons 1.10 0.10 0.0625 0.1021
Physical Mobility
(Ref: Confined to Bed)
Confined to Home 2.64 0.97 0.1059 0.0000
Wheelchair Only 3.46 1.24 0.1557 0.0000
Physically Mobile 14.57 2.67 0.0990 0.0000
Sector (Ref: Rural)
Urban 1.09 0.09 0.0242 0.0001
Gender (Ref: Male)
Female 1.16 0.15 0.0268 0.0000
Transgender 0.35 -1.03 0.0001 0.0000
Marital Status (Ref: Never Married)
Currently Married 1.10 0.10 0.1676 0.5433
Widowed 0.98 -0.01 0.1684 0.9200
Divorced/Separated 1.25 0.23 0.2594 0.3736
Age 0.94 -0.05 0.0017 0.0000
General Education 1.02 0.02 0.0029 0.0000
Results show that economic independence has a strong and significant positive association
with perceived health. Compared to the people who are fully dependent on others,
individuals who are partially dependent are 1.53 times more likely to report better perceived
health (β = 0.42, p < 0.001), while those individuals who are not dependent on others are
2.47 times more likely to report better health (β = 0.90, p < 0.001). This suggests that as an
elderly individual moves away from being totally dependent towards being partially (and
fully independent), the financial autonomy realised translates into a better health status.
The number of sons show an interesting relationship. As can be seen from the table, the
relationship is positive but non-linear in nature. Having one son increases the odds of
reporting better health status by 44% (as OR = 1.44, p < 0.001), while having two sons
increases the odds by 37% (OR = 1.37, p <0.001). After this, the effect diminishes fast. Having
three sons increases the odds by only 24% (OR = 1.24, p < 0.001), while having 4 or more
sons reduces this increase in odds by only 10% (OR = 1.10, p = 0.1021). This p-value in the
last case signifies that the increase in odds is not statistically significant at 5% level of
significance. This pattern suggests that number of sons have diminishing returns to elderly
perceived health. After having a certain number of sons, this effect is no longer significant.
The odds ratio of physical mobility signifies that it is one of the strongest predictors of
elderly health. Compared to those who are confined to bed, the odds of reporting better
perceived health are 2.64 times higher for those confined to home (β = 0.97, p < 0.001), 3.46
times higher for those using a wheelchair (β = 1.24, p < 0.001), and substantially higher at
14.57 times higher for those who are physically mobile (β = 2.67, p < 0.001). This is
intuitively expected – a person who is not able to function from a mobility perspective is
bound to report their health as worse compared to an individual who maintains functional
independence in late part of their life.
A significant rural-urban divide (or difference) is observed. Urban residents have a 9% higher
(or 1.09 times) chance of reporting better health compared to their rural counterparts, and
this relation is statistically significant (β = 0.09, p < 0.001). The reasons can be manifold, a
few of them include better access to healthcare, better access to more nutrition, easier
access to hospitals etc.
A gender difference is also seen in the result. In gender terms, females are 16% more likely
(OR = 1.16, p < 0.001) to report better health outcomes than males, while transgenders see
a sharper dip – a whopping 65% decline (OR = 0.35, p < 0.001). Note that the result is not
accurate for transgenders, as there were only 2 transgenders in our entire analysis. All in all,
this result does signify the deep-seated gender disparity that exists in an Indian society.
Among the other controls, age has a significant negative effect (β = -0.05, p < 0.001). The
logic follows directly – as a person gets older, they tend to become more prone to disease
and getting sick due to weakening immune system etc.
General education also shows a significant positive relationship (β = 0.02, p < 0.001). This
shows that attainment of a higher education corresponds to better perceived health status
in the elderly. The reasoning could again be manifold, but the main reason could be an
increases awareness of one’s own self. An educated person is more likely to go to the doctor
when he/she is sick compared to uneducated person simply due to an increased awareness
about the importance of early detection of disease.
The one variable which ends up very insignificant is marital status. Compared to those who
were never married, currently married elderly are 10% more likely to report better health,
although the effect is not statistically significant (β = 0.10, p = 0.54). This result is correctly in
line with the paper mentioned before, by (Goldman [Link], 1995) whose findings suggest that
certain unmarried groups create a social environment that compensates (in ways that
improve health) for the loss or absence of a spouse so that recent changes in marriage
patterns need not have adverse effects on the future health status of the elderly.
The second regression model examines the association of perceived health status of the
elderly with the state of economic independence and number of daughters living, controlling
for mobility, demographic, and socioeconomic characteristics. The odds ratio (OR) with 95%
confidence interval (CI) of ordered logit regression of predictor variables on perceived health
status is presented in the Table below.
Table 3: Ordered Logit Regression Results - Economic Independence and Number of Daughters

Variable Odds Ratio Coefficient (β) Std. Error p-value

State of Economic Independence
(Ref: Fully Dependent on others)
Partially Dependent 1.55 0.44 0.0299 0.00
Not Dependent 2.50 0.92 0.0330 0.00
Number of Daughters Living
(Ref: 0 daughters)
1 Daughter 1.01 0.01 0.0334 0.73
2 Daughters 0.97 -0.02 0.0353 0.52
3 Daughters 0.84 -0.17 0.0441 0.00
4 or More Daughters 0.92 -0.08 0.0555 0.13
Physical Mobility
(Ref: Confined to Bed)
Confined to Home 2.82 1.03 0.11 0.00
Wheelchair Only 3.69 1.30 0.16 0.00
Physically Mobile 15.98 2.77 0.10 0.00
Sector (Ref: Rural)
Urban 1.07 0.06 0.0251 0.0069
Gender (Ref: Male)
Female 1.18 0.16 1.18 0.00
Transgender 0.00 -11.12 0.00 0.00
Marital Status (Ref: Never Married)
Currently Married 1.38 0.32 0.16 0.051
Widowed 1.23 0.21 0.16 0.20
Divorced/Separated 1.80 0.58 0.26 0.02
Age 0.94 -0.05 0.001 0.00
General Education 1.02 0.02 0.003 0.00
Compared to the people who are fully dependent on others, individuals who are partially
dependent are 1.55 times more likely to report better perceived health (β = 0.44, p < 0.001),
while those individuals who are not dependent on others are 2.50 times more likely to
report better health (β = 0.92, p < 0.001). In this interpretation, the Beta coefficients as well
as the odds ratio remain similar to the first regression. This model reaffirms the strong
influence of economic independence on self-perceived health status.
However, the results for the number of daughters living show a different pattern from the
previous model (where we took the number of sons as the proxy variable). Compared to
elderly people with no daughters, those with one daughter are 1% more likely to report
better perceived health (OR = 1.01, p = 0.73). However, the high p-value suggests that this
association is statistically insignificant. Similarly, the analysis shows a more or less
insignificant effect for having a daughter, as the OR ranges from 0.84 to 1.01. This result
reflects the underlying socio-cultural dynamics within Indian society. An elderly man’s
daughter is most likely married and lives with her in-laws, taking care of them. The social
construct, in that sense, doesn’t allow daughters to take care of their own parents.
Interestingly, previous studies have found that elderly parents tend to expect and favour
non-financial support from daughters, such as emotional or social assistance, but rely
comparatively more on sons for financial support (Srinivasan, Sharada & Sebastian, Irudaya
Rajan & Nanda, Aswini & Bedi, Arjun., 2023). This societal dynamic could help explain the
limited direct association between the number of daughters and perceived health outcomes
in this study.
The control variables continue to have the same relationship (or association) as the previous
regression, reaffirm their influence on perceived health. Physical mobility continues to be a
strong predictor of perceived health status, while Gender and Marital Status maintain
patterns consistent with previous model. Additionally, sector of residence continues to show
the same result – urban elderly are more likely to report better self-perceived health than
their rural counterparts. Lastly, both Age and General Education have same signs as before
(Age having a negative relationship and General Education having a positive relationship).
These two regression results taken together, highlights the gendered nature of family
support in India. While sons seem to play a strong role in influencing elderly health
outcomes, daughters’ roles, though important in emotional terms, are rather less directly
reflected in perceived health of the elderly. This crucial result reflects not only the cultural
expectations but also the structural realities around caregiving and residence patterns in
Indian families. To understand this, it becomes extremely important to examine and analyse
another crucial social variable – living arrangement – which tells with whom the elderly
individual resides, and see how this affects their care, health, and overall well-being.
The third regression focuses on this exact topic – by introducing living arrangement as a
proxy for social support. Following are the results of the regression –
Table 4: Ordered Logit Regression Results - Economic Independence and Living Arrangement

Variable Odds Ratio Coefficient (β) Std. Error p-value

State of Economic Independence
(Ref: Fully Dependent on others)
Partially Dependent 1.56 0.44 0.028 0.00
Not Dependent 2.48 0.91 0.032 0.00
Living Arrangement (Ref: With Spouse)
No Spouse, With Others 0.93 -0.07 0.053 0.16
Living Alone 0.55 -0.58 0.092 0.00
Physical Mobility
(Ref: Confined to Bed)
Confined to Home 2.64 0.97 0.103 0.00
Wheelchair Only 3.40 1.22 0.152 0.00
Physically Mobile 14.68 2.68 0.097 0.00
Sector (Ref: Rural)
Urban 1.10 0.10 0.024 0.00
Gender (Ref: Male)
Female 1.18 0.16 0.026 0.00
Transgender 0.45 -0.79 0.000 0.00
Marital Status (Ref: Never Married)
Currently Married 1.26 0.23 0.146 0.11
Widowed 1.22 0.20 0.141 0.14
Divorced/Separated 1.56 0.45 0.234 0.05
Age 0.94 -0.05 0.002 0.00
General Education 1.02 0.02 0.003 0.00

The results here suggest the following – Compared to elderly people living with their
spouses, those who live alone are 45% less likely to report a better, positive health outcome
(OR = 0.55, p < 0.001). Since the variable is statistically significant, this suggests that living
alone has a strong negative association with perceived health among the elderly. This could
be due to the lack of day-to-day care, companionship and emotional support. On the other
hand, those not living with their spouse, but with other members (like children or someone
else) show a slightly lower likelihood of reporting better health – at around 7% less likely (OR
= 0.93, p = 0.16), though this is statistically insignificant.
This result highlights an important point – the importance of living arrangement. Living with
someone (spouse, children or other members) ends up increasing the odds of reporting
better health compared to living alone, suggesting the importance that companionship plays
in improving health.
As with the previous models, economic independence continues to show a strong and
significant association with perceived health. Compared to an individual who is fully
dependent on others, the person not dependent is 2.48 times more likely to report better
health (β = 0.91, p < 0.001). This result, just like the ones we found in the previous two
regression reaffirms the importance of financial autonomy in shaping (and improving)
perceived health status among the elderly.
The control variables exhibit consistent patterns with earlier models, reaffirming their
expected influence. Physical mobility remains a key determinant of perceived health, while
age, education, gender, marital status, and sector of residence continue to show similar
directions of association.
Summary of results -

Variable Effect Direction Significant?

Economic Positive Yes
independence
Number of sons Positive but Yes
diminishing
Number of Close to 0 No
daughters
Living alone Negative Yes

5. Conclusion and Discussion

Our study examined the social and economic determinants of self-perceived health among
the elderly in India using the 75th Round of the National Sample Survey (Social
Consumption: Health). Focusing on individuals aged 60 and above, we employed ordered
logistic regression to understand how key variables – economic independence, family
composition (number of sons and number of daughters), and living arrangement – shape the
perception of health among them.
The findings clearly reveal that economic independence is one of the most crucial and
significant predictors of perceived health status. Elderly individuals who are not dependent
on others are significantly more likely to report better health outcomes compared to fully
dependent counterparts. Even partially dependent individuals are better off than those who
are fully dependent. This indicates an important detail - while having complete financial
autonomy is important, even having some degree of financial autonomy still benefits elderly
individuals. A financially independent individual is in a better position to access appropriate
health facilities, in time, without relying on someone else - leading to improved health
outcomes.
Including family composition - through number of sons and number of daughters - offers
important insights on how they affect elderly health. The results for the two variables are
not the same. Number of sons shows a positive and significant relationship, but with a
diminishing marginal impact. Having more sons is better than having none - but the odds of
reporting better health decline with every additional son. Moreover, for the category “4 or
more sons”, the variable is positive but statistically insignificant. Thus, the common notion of
“having more children means more support” is not necessarily correct.
For the number of daughters, the results are different. The association between number of
daughters and perceived health status is sometimes positive and sometimes negative – but
statistically insignificant. A paper on this subject concluded that with an increase in number
of children in a family, daughters may free-ride by allowing others to support their parents,
reducing their own share of support, while sons provide relatively more in economic support
and household care (Jiguang Z, Yuncan W and Yunxing S, 2023). Thus, while emotional
support is likely provided by daughters, it does not translate into improved perceived health
status for the elderly.
Control variables were chosen and incorporated in a way that ensured that the model
remains as true to its objective as possible. The study’s control variables – including age,
education, gender, physical mobility, marital status, and sector of residence – continue to
show statistically significant and theoretically expected associations with perceived health.
Age exhibits a negative relationship, indicating that health naturally declines with age.
Education shows a positive relationship, reinforcing that better educated individuals tend to
possess better health awareness and therefore better outcomes. Physical mobility emerges
as a strong predictor - indicating that the ability to move independently not only reflects
physical capability but also translates into better social participation and confidence.
Moreover, elderly individuals in urban areas are more likely to report better perceived
health, highlighting disparities in access, infrastructure, and awareness between rural and
urban India.
Taken together, these findings highlight the multifaceted nature of elderly health – it is not
only determined by biological or medical factors, but also depends on a broader social,
economic and familial environment. Economic independence provides security, while family
and living arrangements provide a social framework within which the elderly experience
care, support, social life and emotional wellbeing. All these factors jointly influence
perceived health status among the elderly.
Future research may extend this analysis by incorporating longitudinal data, which would
allow us to test how transitions in family structure, financial autonomy and living
arrangements affect health over time. In addition, qualitative work on intra-household
bargaining and support provision could deepen understanding of gendered caregiving
norms. Finally, the role of community-level factors - such as neighbourhood trust, access to
primary care, and social participation spaces - deserves further examination to identify
mechanisms through which social environment interacts with individual determinants of
elderly health.

5.1 Policy Implications

Our results, while signifying the importance of economic and social factors, also carry
important implications for policy. With India’s growing elderly population, the following
changes can go a long way in ensuring a healthier elderly population.
Strengthening Economic Security: The analysis highlights that the elderly individuals who
are financially independent report better perceived health. Currently, economic schemes for
the elderly in India include pension schemes like Pradhan Mantri Shram Yogi Maandhan
(PM-SYM) and the Indira Gandhi National Old Age Pension Scheme (IGNOAPS), healthcare
coverage under the expanded Ayushman Bharat-Pradhan Mantri Jan Arogya Yojana (AB-
PMJAY). Such existing schemes should be expanded in both the coverage and benefit levels
to enhance financial independence.
Encouraging Gender-Inclusive Elderly Support: The result indicates that number of sons
have a positive and significant association with elderly health, while number of daughters
have a statistically insignificant effect. This reflects the traditional expectation that sons
should provide financial and caregiving support to ageing parents. This imbalance should be
addressed, encouraging daughters to participate equally in elderly care. Awareness
campaigns and incentives for shared caregiving responsibilities can help reshape social
norms around elderly care.
Promoting Family and Community-Based Care: As the result shows, living alone can have
detrimental effects on elderly health. Strengthening community care infrastructures such as
day-care centres and respite homes can help provide social and emotional support to elderly
who lack care from their family. These can complement traditional family systems while
simultaneously addressing the issue of elderly isolation.
Improving Access to Healthcare in Rural Areas: We found that elderly people living in the
urban areas have higher odds of reporting better perceived health status compared to their
rural counterparts. This highlights the need to improve healthcare accessibility in rural
regions through mobile health units and better medical infrastructure.
Integrating Health and Social Protection: As found in our study, elderly health is not solely a
biological or a medical issue, but also a social and economic one. Therefore, elderly health
should adopt a more integrated approach – linking healthcare, pensions, and social
participation under a unified policy framework.

5.2 Limitations of the Study

Beyond the data-related constraints described earlier, the study has important conceptual
limits. Additionally, the analysis is subject to survivorship bias - the elderly captured in the
survey are those who survived to the date of data collection; those with the worst health
may have died earlier and are therefore missing from the dataset, which could understate
the true negative effect of age and poor health.
First, the analysis has been carried out at a national, aggregated level and does not explore
heterogeneity across states, caste groups, religion or socio-economic strata. Given India’s
diversity, the determinants of elderly health may differ substantially across these groups,
which this paper does not unpack. Second, while the three separate models allow
comparison across different social proxies (sons, daughters and living arrangement), the
study does not estimate any combined model or interaction effects (for instance, whether
the effect of economic independence differs by living arrangement). These analytical choices
were deliberate for clarity, but limit how far we can generalise about mechanisms. Future
work could extend this by incorporating state-fixed effects, subgroup analysis or qualitative
evidence on caregiving practices. Third, reverse causality is possible: healthier elderly may
remain financially active longer, which complicates interpretation.
India should prepare for ageing like it prepared for fertility decline: as a structural
transition. Both pension architecture and caregiving models need to be redesigned to
match demographic reality.

6. References
Antczak R, Quashie NT, Mair CA, Arpino B. Less Is (Often) More: Number of Children and
Health Among Older Adults in 24 Countries. J Gerontol B Psychol Sci Soc Sci. 2023 Nov
14;78(11):1892-1902. doi: 10.1093/geronb/gbad123. PMID: 37622727; PMCID:
PMC10645313.
Arber, S. and Ginn, J. (1993) Gender inequalities in health in later life. Soc. Sci. Med. 36:33–
46.
Banks, J. and Casanova, M. (2003) Work and retirement. In: Health, wealth and lifestyles of
the older population in England: the 2002 English Longitudinal Study of Ageing (ed. M.
Marmot, J. Banks, R. Blundell, et al.). London: Institute for Fiscal Studies.
Berkman, L.F. and Glass, T.A. (2000) Social integration, social networks, social support and
health. In: Social epidemiology (ed. L.F. Berkman and I. Kawachi), pp. 137–73. Oxford: Oxford
University Press
Das J, Kundu S, Hossain B. Rural-urban difference in meeting the need for healthcare and
food among older adults: evidence from India. BMC Public Health. 2023 Jun 26;23(1):1231.
doi: 10.1186/s12889-023-16126-4. PMID: 37365536; PMCID: PMC10294478.
Ghosh, S., & Husain, Z. (2010). Economic independence, family support and perceived health
status of the elderly: Recent evidence from India. Asia Pacific Population Journal, 25(1), 47-
77.
Goldman N, Korenman S, Weinstein R. Marital status and health among the elderly. Soc Sci
Med. 1995 Jun;40(12):1717-30. doi: 10.1016/0277-9536(94)00281-w. PMID: 7660185.
Gupta, I. (2004). Medical attention at death: Evidence from India. Journal of Health
Management, 6(1), 73–84.
Jiguang Z, Yuncan W and Yunxing S (2023) Raising sons or daughters for old age? Influence of
children's gender on intergenerational family support in rural families. Front. Public Health
10:1063390. doi: 10.3389/fpubh.2022.1063390
Kahn, J. H., Hessling, R. M., & Russell, D. W. (2003). Social support, health, and well-being
among the elderly: What is the role of negative affectivity? Personality and Individual
Differences, 35(1), 5–17. [Link]
Nair S, Sawant N, Thippeswamy H, Desai G. Gender Issues in the Care of Elderly: A Narrative
Review. Indian J Psychol Med. 2021 Sep;43(5 Suppl):S48-S52. doi:
10.1177/02537176211021530. Epub 2021 Jul 5. PMID: 34732954; PMCID: PMC8543621.
Naz, Sumara. (2014). Relationship between Economic Independence, Social Support and
Quality of Life among Elderly People. Journal of the Indian Academy of Applied Psychology.
40.
Shrivastava, Saurabh & Bobhate Shrivastava, Prateek & Ramasamy, Jegadeesh. (2013). Role
of self-care in management of diabetes mellitus. Journal of diabetes and metabolic
disorders. 12. 14. 10.1186/2251-6581-12-14.
Thakur R, Banerjee A, Nikumb V. Health problems among the elderly: a cross-sectional study.
Ann Med Health Sci Res. 2013 Jan;3(1):19-25. doi: 10.4103/2141-9248.109466. PMID:
23634324; PMCID: PMC3634218.
Tyagi R, Paltasingh T. Determinants of Health among Senior Citizens: Some Empirical
Evidences. Journal of Health Management. 2017;19(1):132-143.
doi:10.1177/0972063416682613
Tyagi, R, & Kapoor, S. (2010). Functional ability and nutritional status of Indian elderly. The
Open Anthropology Journal, 3(1), 200–205.
Zhang F and Abd. Rahim N (2024) The impact of education on the elderly well-being: A
meta-study. International Journal of Infrastructure Research and Management 12(2):68–78
[Link]