When caring comes at a cost: Psychological wellbeing of unpaid and paid carers and the role of social expenditure

Abstract

This study examines whether, and under what conditions, unpaid and paid care work are associated with reduced psychological wellbeing. The article begins by laying out a shared theoretical framework for understanding the psychological consequences of care among both unpaid and paid carers. It then tests the empirical implications of this framework, conducting multi-level model analysis of European Quality of Life Survey and European Social Survey data and: (1) disaggregating care work based on (a) the care recipient – i.e., adults or children – for unpaid carers and (b) the level of occupational professionalization for paid carers; and (2) examining the potential intervening role of social expenditure. Findings demonstrate that unpaid caring for adults (though not children) is associated with a marginal decrease in psychological wellbeing, but that this dynamic is limited to countries with smaller welfare states. Among paid care workers, only paraprofessionals are found to have lower levels of psychological wellbeing than comparable non-care workers – but here again increased social expenditure appears to have a significant buffering effect. Together, results reinforce the need for robust social spending to mitigate negative psychological consequences of care, while adding important nuance regarding the relevance of the type of care work being performed.

Keywords

care work psychological wellbeing paid work social expenditure unpaid work

Scholarship on unpaid and paid care largely remains in two solitudes, notwithstanding numerous commonalities between these types of work.¹ On the one hand, unpaid care work, which is typically defined as caring for dependent family members, neighbours, or friends (Folbre, 2012), is found to have a negative association with the carer’s ability to participate in the paid labour market and to be closely tied to cultural values (England et al., 2002; Jacobs et al., 2019). On the other hand, research on paid care work – most often defined as employment involving face-to-face interactions with children, the elderly, or people with complex healthcare needs (England et al., 2002) – largely focuses on the social and financial devaluation of such jobs and the physical demands they often entail (Duffy et al., 2021; Meagher et al., 2016).

At present, however, less is known about the conditions under which paid and unpaid care work may shape psychological wellbeing, as well as the extent to which structural context may play a buffering role. This article aims to contribute to the existing cannon of care work scholarship by providing a cross-national quantitative analysis of the extent to which unpaid and paid care are associated with psychological wellbeing. In doing so, it tests whether these dynamics may be shaped by (1) the type of care work being performed and (2) national-level factors, with a particular focus on social expenditure.

Existing theory, in particular the Stress Process Model (Pearlin, 1989; Pearlin et al., 1990), suggests that care work is likely to negatively affect psychological wellbeing due to its resultant stressors and strains (e.g. low levels of perceived control and high levels of stress tied to “role capture”). Yet empirical research has come to more mixed conclusions: studies on unpaid care work suggest that caring for children has negative and positive implications on psychological wellbeing (Henle et al., 2020); while research on paid care work has found that, notwithstanding high rates of burnout and turnover in many paid caring jobs, there are non-pecuniary benefits (i.e. positive associations from helping others in need) that may improve job satisfaction for certain types of care workers – with context and policy environment playing key roles (Hebson et al., 2015; Lightman and Kevins, 2019). Social expenditure is likely to be one such contextual factor, with social programmes that decrease unpaid care burdens potentially shaping the impact of care work on wellbeing (Collins, 2019; Rodriguez-Loureiro et al., 2020).

This article uses a common theoretical framework to assess care workers’ attendant psychological wellbeing outcomes, while also integrating a multi-dataset analysis of structural context across a wide range of countries. The analyses underscore that notwithstanding national variation and a potentially endogenous relationship between psychological wellbeing and performing care work, there are benefits to disaggregating care and integrating the potential intermediary role of social expenditure. Results suggest that caring for children has no perceivable relationship to psychological wellbeing, while caring for adults is associated with a marginal decrease in psychological wellbeing – though this dynamic is limited to countries with smaller welfare states. In the case of paid care workers, data indicate that only in the case of paraprofessionals (e.g., medical assistants, nursing aides, or primary school teachers) is there a correlation with lower levels of psychological wellbeing – arguably connected to a context of high demands combined with low levels of control within these jobs (Karasek, 1979) – but here again this dynamic disappears in countries with higher levels of social expenditure.

While our reliance on observational, cross-sectional data limits our ability to determine causality, these findings reinforce prior suggestions that there is a need for robust social spending and defamilializing welfare state policies to mitigate the negative association between psychological wellbeing and providing unpaid and paid care. They also add important nuance regarding the relevance of both the type of unpaid care recipient and the type of paid care work performed, and provide compelling reasons to bridge the scholarly divide between those focusing on unpaid and paid care.

Unpaid and paid care work

Scholarship on care work largely focuses on labour market outcomes and is typically divided between those who examine unpaid/informal care and those who examine paid care. In the case of unpaid caring, existing research often centres on the disproportionate burden that falls to women (and especially lower-class, immigrant, and racialized women) in families (Charmes, 2022; Collins, 2019).² Unpaid care work is typically measured in terms of type and amount of time spent on care, and may affect the carer’s ability to participate in the paid labour market as well as the type/quality of employment opportunities available to them (Hagqvist, 2018; Jacobs et al., 2019; Ragnarsdóttir et al., 2023). As a result, unpaid care work is thought to be a key factor associated with intersectional disparities in income, occupational prestige, and upward mobility (Esping-Andersen and Schmitt, 2020; Gencer et al., 2024). Cross-nationally, countries that have higher levels of provision of publicly funded childcare, more generous paternity leave policies, and more egalitarian gender attitudes, typically also have increased gender parity in unpaid caring (Charmes, 2022; Hook et al., 2023).

In the case of paid care work, existing research often points to an economic disadvantage tied to care jobs, as paid care workers receive lower wages than comparable workers in non-caring jobs (England et al., 2002). Yet scholars also note that working conditions, prestige, and pay are closely connected to level of occupational professionalization within paid care. Not surprisingly, outcomes are highly variable between higher status care jobs (e.g., doctors, professors) that are often socially valorized and well compensated and lower status care jobs (e.g., personal support workers, childcare assistants) that are often temporary and low wage (Folbre et al., 2021; Hebson et al., 2015). Many lower status care jobs are also highly repetitive, physically demanding, and/or afford workers limited autonomy (Glavin and Peters, 2015).

Ultimately, in assessing common themes within the literatures on unpaid and paid care, it is notable that in both cases the association with “women’s work” is intimately connected to its social and economic devaluation – and that the broad focus is most often on negative implications for vulnerable populations (Folbre et al., 2021; Meagher et al., 2016). Yet, while much of the existing literature on care work is theoretical, qualitative, or focused on a single country of analysis, there is growing recognition of a need for cross-national comparative analyses. Existing work points, for instance, to the need for additional research on the similarities and differences between unpaid and paid care (Folbre, 2012; Lightman, 2024), as well as further investigation into the direction of causality between care and outcomes of interest (Dang et al., 2022; Oshio and Usui, 2017). Notably, the potential psychological consequences of care work are increasingly being highlighted, reflecting growing recognition of the need to support carers’ mental health and wellbeing – both in private homes and in the paid labour market (Browning and Penning, 2020; Parbst and Wheaton, 2023).

The psychological implications of performing care

The literature connecting care work and wellbeing points to certain commonalities across countries. These include, for example: the stressors experienced by individuals providing unpaid care, particularly over extended time periods and in contexts of minimal social support (Aneshensel and Avison, 2015; Collins, 2019); and the high demands of many paid care work jobs that involve multi-tasking and struggles with work-life balance (Folbre, 2012; Lightman, 2024). Yet existing research also demonstrates that many crucial factors depend on national context. These include immigration frameworks, cultural values tied to caring, and historical legacies of state involvement or free-market orientation in the health and social care sectors (Jacobs et al., 2019; Lightman, 2024).

In theorizing the relationship between care work and psychological wellbeing, Pearlin’s (1989) Stress Process Model (SPM) provides a particularly instructive starting point. Broadly, the SPM identifies the ways that wellbeing is influenced by role-related stressors, tied to both family and work demands. In its inception, the SPM distinguished itself from related frameworks by underscoring the importance of social context in shaping how stressors – such as major life events, chronic strains, and daily demands – impact psychological wellbeing at the individual level (Aneshensel and Avison, 2015). Pearlin (1989) also suggested that the impacts of stressors, and the degree to which they proliferate, was tied to both individual resources and the degree of social support received. In what follows, we apply this model to (unpaid and paid) care work and work to reconcile its application with existing empirical care research.

The application of the SPM has more frequently been applied to unpaid/informal caregiving. Indeed, several studies on the topic have adopted a modified version of the SPM, emphasizing the importance of incorporating both macro-level factors (e.g., welfare state regime type) and ascriptive characteristics as determinants of psychological wellbeing (Browning and Penning, 2020; Pearlin et al., 1990). In particular, the SPM framework draws attention to the stresses experienced by individuals providing unpaid care, highlighting negative psychological burdens associated with factors such as a loss of leisure time, challenges in balancing paid and unpaid work, and the spillover of caregiving responsibilities to multiple aspects of one’s life (Glavin and Peters, 2015; Nomaguchi and Milkie, 2020).

Cross-sectional scholarship in this tradition typically suffers from concerns about endogeneity. For example, it cannot be ruled out if people with lower levels of psychological wellbeing are more likely to take on unpaid care burdens, perhaps because they are less likely to be in the labour market or are earning less (e.g. Dang et al., 2022; Johnson and Lo Sasso, 2006). Past empirical research nevertheless provides suggestive evidence that not all unpaid care work is experienced as a drain on psychological wellbeing. A number of studies find that some unpaid caregivers report gains in quality of life, and that most caregivers report at least some benefits from caring (De Oliveira and Hlebec, 2016; Maguire et al., 2019). One likely explanation for mixed findings here is importance of the type of unpaid care being provided: hours spent on eldercare and housework are generally found to be more stressful than those spent providing childcare, for which positive effects of assisting in child development are often a component of the care (Henle et al., 2020; Nomaguchi and Milkie, 2020).

And while the SPM is less commonly applied to paid care workers, there are good reasons to believe that similar dynamics play out here as well. Studies examining, for example, social workers, nurses, and teachers find that competing workplace demands, stress, and burnout are highly correlated with occupational prestige, particularly in jobs that require high levels of emotional labour – though here again, it is found that structural context often plays an important moderating role (Huang et al., 2022). Indeed, research in line with the SPM suggests that paid carers face disproportionate stressors at work, while often also coping with complex family dynamics and/or unpaid care burdens (Glavin and Peters, 2015; Meagher et al., 2016). Similarly, other studies find that paid care work takes a considerable toll on psychological wellbeing, especially over prolonged time periods and in scenarios of: (1) high job demands, coupled with low financial remuneration and/or limited control over decision-making; and (2) limited welfare state support, e.g. a lack of supportive parental leave policies or protective employment standards legislation (Karasek, 1979; Nomaguchi and Milkie, 2020).

Here again, however, other empirical work suggests that paid care may have positive associations with psychological wellbeing as well (Hebson et al., 2015), especially in a context of what Lopez (2006) terms “organized emotional care” – that is, where workers have greater autonomy and decision-making power. Lightman and Kevins (2019), for example, find that paid care work is associated with positive effects when disaggregated by level of professionalization. They use European Social Survey data and find that for care workers with non-professional jobs, assisting others may act as a psychosocial resource, leading to greater job satisfaction than for comparable non-care workers.

In sum, for both unpaid and paid care providers, the overall psychological consequences of care work remain unclear, with outcomes likely connected to the type of care recipient, the level of occupational professionalization, the amount of time spent providing care, and the national context. The SPM theoretical framework largely focuses on potential negative psychological implications of providing care (e.g. stress, burnout, social isolation, marginalization, and distress); yet empirical research provides more mixed conclusions, and the type of care work is likely a key factor underlying this variation (England et al., 2002; Hebson et al., 2015; Lightman and Kevins, 2019). At the same time, however, the SPM framework provides broader insights on the social structural origins of stressful life experiences and the importance of social supports (Aneshensel and Avison, 2015; Pearlin et al., 1990). Thus, from a macro perspective, there are good a priori reasons to expect that the wellbeing of unpaid and paid care providers, in comparison to non-carers, may be shaped by the welfare state.

Social expenditure, psychological wellbeing, and work in care

Cross-national research demonstrates that institutional structures, and in particular the welfare state, can have strong associations with psychological wellbeing. Such studies typically find a positive relationship between state intervention and life satisfaction (Grönlund and Öun, 2018). This research aligns neatly within the SPM framework, with its emphasis on the social structural origins of mental health: Parbst and Wheaton (2023), for example, suggest that social protections play a dual role in the SPM, both by reducing stressors and increasing coping resources.

Indeed, an array of studies suggest that social expenditure is a key factor associated with individuals’ quality of life and overall wellbeing (Collins, 2019; Parbst and Wheaton, 2023). Mental health inequalities, in particular, are found to be lower in countries where social expenditure is higher (Sjöberg, 2023). Park et al. (2020), for example, find that OECD countries can have a significant impact on population mental health by investing a larger proportion of total expenditure on social services. Typically, larger welfare states lead to more and better non-professional or paraprofessional paid care work jobs in the public sector, where working conditions are arguably more supportive of psychological wellbeing than in the private sector – due to factors such as increased unionization, improved scheduling control, and/or higher wages (Folbre et al., 2021; Nomaguchi and Milkie, 2020).

Welfare state scholars, for their part, note the challenges in measuring and comparing the range and trajectory of family policy constellations across Europe (Korpi et al., 2013; Lohmann and Zagel, 2016). Much of this literature focuses on individualizing – also called de-familializing – policies. Such policies outsource care from the family to formal institutions supported by the state, thereby reducing unpaid care burdens for families and, in the process, improving psychological wellbeing, especially for mothers (Collins, 2019; Rodriguez-Loureiro et al., 2020). Browing and Penning (2020) conclude that family care regime has a direct association with self-rated mental health, while Somogyi et al. (2021) find that familializing policies are negatively correlated with parental mental health.

Conceptually, increased social spending may be directed towards either familializing or defamilializing policies, thereby increasing or decreasing care burdens. Yet an association between social spending and increased gender equality is found across a wide range of countries (Amate-Fortes et al., 2024). In their review of the literature on family policies across Europe, Lohman and Zagel (2016) note that both familializing and defamilializing policies can have varied effects on patterns of care work. While not explicitly testing causality, they suggest that policy context may influence wellbeing, as state-funded policies such as maternity/parental leave, care services including public child and home care, and generous family benefits all tend to reduce gender-related and intergenerational dependencies of individuals towards their families. In addition, larger welfare states typically have higher levels of formal care coverage that reduce unpaid care burdens, and public social expenditure is often associated with improved work-life balance (Collins, 2019; Park et al., 2020). Yet Grönlund and Öun (2018) note that family policies supportive of dual-earner social roles can contribute to both increased feelings of work-family conflict and psychological wellbeing among women, emphasizing the deeply gendered nature of how social policies are experienced.

Together, then, existing cross-national research demonstrates that social expenditure may be closely connected to the breakdown between unpaid and paid forms of care within and across countries, with contingent effects on psychological wellbeing (Lightman, 2024; Nomaguchi and Milkie, 2020). Yet the relationship between social expenditure, unpaid and paid care work, and psychological wellbeing is far from clear, especially in light of cross-national variation, endogeneity concerns, differences in policy context, and data availability issues. While we are unable to conclusively address all of these issues, the present study sets out to offer some initial insights into these dynamics via cross-national empirical research.

Hypotheses

Taking the SPM framework (Pearlin, 1989) as our theoretical starting point, we anticipate that care work will be associated with lower psychological wellbeing due to a resultant increase in stressors, but that compensating differentials will act as a psychosocial resource (i.e., providing a buffering effect) in the case of (1) unpaid care for children and (2) non-professional care workers. Our hypotheses thus build from the SPM framework, while incorporating past research which suggests that different types and conditions of care work will be more or less likely to have a negative association with psychological wellbeing for carers. Figure 1 provides a visual diagram whereby we adapt the SMP to paid and unpaid care, identifying key stressors and buffers, at both the individual and contextual levels, that may affect psychological wellbeing for carers. We note both differences and similarities across these two types of care, as well as the potential role of social expenditure.

Figure 1.

Care stressors and buffers associated with psychological wellbeing.

Turning to our hypotheses, we note first that scholars have demonstrated that the effects of unpaid caring are likely tied to the type of care recipient, with unpaid caring for adults often taking a greater psychological toll than unpaid caring for children. This is because unpaid care for adults is typically found to be unpredictable and demanding (time intensive), while providing childcare is more often viewed as a combined chore and pleasure (Henle et al., 2020; Nomaguchi and Milkie, 2020). Second, paid caring is more likely to be associated with negative psychological wellbeing (e.g., tied to burnout, role overload, and the blurring of work-nonwork boundaries) at higher levels of professionalization; the underlying argument here is that any compensating differentials generated by care work, relative to employment in non-care work jobs with comparable levels of status/income/autonomy, are likely to be larger among non-professionals (see Lightman and Kevins, 2019).

Time spent on unpaid care work for adults will be more strongly correlated with lower levels of psychological wellbeing than time spent on unpaid care work for children.

Employment in the paid care sector will be correlated with lower levels of psychological wellbeing at higher levels of occupational professionalization.

The next step in the analysis is to examine cross-national differences in the relationship between caring and psychological wellbeing, homing in on the potential importance of variation in social expenditure. In particular, we anticipate that the welfare state may weaken the link between caring and psychological wellbeing via two potential mechanisms: a reduction of stressors (e.g., reducing role overload or inter-role conflict due to the demands of unpaid caring or a proliferation of paid care jobs with high demands/levels of workplace burnout); and an increase in coping resources (i.e., state support for social care through programs, tax allowances, more stringent work/non-work boundaries, and/or resultant changes to societal values) (Nomaguchi and Milkie, 2020; Parbst and Wheaton, 2023). If this is indeed the case, then the logic behind the individual-level hypotheses laid out above would suggest that social expenditure should especially matter for the types of care that the literature suggests will be most affected by reduced psychological wellbeing: namely, unpaid care for adults and paid care at higher levels of professionalization.

Higher social expenditure will be associated with a stronger reduction in the correlation between unpaid care work for adults and psychological wellbeing, relative to its effect on the relationship between unpaid care work for children and psychological wellbeing.

Higher social expenditure will be associated with a reduction in the correlation between more professionalized paid care work and psychological wellbeing.

Data

We test these hypotheses using two major cross-national datasets: the European Quality of Life (EQL) Survey (European Foundation for the Improvement of Living and Working Conditions, 2018), to examine unpaid care work; and the European Social Survey (ESS, 2014), to examine paid care work. For both datasets, we focus on the most recent round that included the survey items required to conduct the investigation.³

The analysis of unpaid care work uses the fourth wave of the EQL, which was fielded in 2016 and 2017; after accounting for missing information, the dataset included 20,046 respondents from 22 European countries. The analysis of paid care work, in turn, uses the seventh wave of the ESS, fielded between 2014 and 2015; after accounting for missing information, the censored dataset included 13,304 respondents from 20 European countries. Online Appendix (OA) Table 1 presents the number of observations per country included in the analysis from each dataset.

We rely on separate datasets for the two stages of the analysis as there is no single cross-national survey that would allow us to examine psychological wellbeing vis-à-vis both unpaid and paid care work. We nevertheless sought to maintain comparability across the two components of our investigation by maximizing consistency across the dependent and independent variables, controlling for potentially relevant contextual changes between the survey dates (captured via national-level data), and conducting additional robustness checks (discussed in detail below).

Dependent variables

For both sections of the analysis, the dependent variable was an index capturing psychological wellbeing. The index is derived from an established measure developed by the Center for Epidemiologic Studies (see Radloff, 1977). As the required questions are included in both datasets, this allows us to analyse comparable measures of psychological wellbeing among unpaid and paid care workers.

We constructed our EQL and ESS indices using Item Response Theory (IRT) graded response models, which enables us to account for the ordinal nature of the index items (see Treier and Hillygus, 2009). Each index was created using eight survey items, with responses re-coded where necessary to ensure that higher values indicate higher levels of psychological wellbeing. We exclude respondents who have not responded to all questions. These eight questions asked respondents about the frequency with which they felt various feelings (e.g., depressed, cheerful, restless, happy, lonely, sad, tense) in recent weeks. Full question wordings and response options are presented in OA Table 2.

The EQL psychological wellbeing index and the ESS index demonstrate reasonable reliability given how well-established the measure is (Cronbach’s α = 0.870 and 0.782, respectively). They also have comparable means (0.030 and 0.004, respectively) and standard deviations (0.923 and 0.893, respectively). We thus follow a long line of research using these psychological wellbeing indices, including via EQL and ESS data specifically (e.g. Reibling et al., 2017; Topp et al., 2015).

Explanatory variables

The key explanatory variables at the individual level capture the extent to which respondents engaged in a given type of care work.

In the case of unpaid caring, the EQL asked respondents “On average, how many hours per week are you involved in any of the following activities outside of paid work?” Responses were divided into time spent caring for children (“Caring for and/or educating your children” and “Caring for and/or educating your grandchildren”) and caring for adults (“Caring for disabled or infirm family members, neighbours or friends under 75 years old” and “Caring for disabled or infirm family members, neighbours or friends aged 75 or over”). Respondents who recorded doing more than 120 hours of unpaid care per week were excluded to minimize the risk of spurious outliers. On average, the number of hours spent caring for children was higher (mean = 11.4; standard deviation (SD) = 23.9) than the time spent caring for adults (mean = 3.0; SD = 13.4).

Paid care work, in turn, was assessed using International Standard Classification of Occupations (ISCO-08) codes. Reflecting past work (e.g. Lightman and Kevins, 2019), respondents were divided up based on (1) whether or not they are engaged in paid care work, and (2) the level of professionalization of their job, as designated by the ISCO code (i.e., distinguishing between professional, paraprofessional, and non-professional occupations). The full care work classification is available in OA Table 3 and includes roles such as medical doctors and teachers (at the professional level), dental assistants and physiotherapists (at the paraprofessional level), and housekeepers and care workers (at the nonprofessional level). After excluding the non-employed and those of non-working age (under 18 or over 70) from the sample, 18.3% of respondents were employed in care work, with the following division by level of professionalization: 32.0% non-professional; 36.7% paraprofessional; 31.3% professional care workers.

Finally, the key explanatory variable at the national level was social expenditure. This was measured as total public and mandatory private expenditure on social policy/programs as a percentage of GDP, and includes expenditure in areas such as family policy, state pensions, and incapacity-related benefits. Data was taken from the OECD (2020) to align with the year prior to the survey fielding, with the distribution of expenditure similar in the unpaid care work sample (mean = 24.2; SD = 5.3; min = 14.7; max = 32.2) and the paid care work sample (mean = 24.7; SD = 4.7; min = 14.5; max = 32.1).

Control variables

Controls were selected to reflect standards in the literature (Duffy et al., 2021; Hook et al., 2023) and were broadly equivalent in the two sets of analyses. The basic individual-level controls included: household income decile and household size (to account for variation in the potential meaning of household income); gender (measured with a binary woman/man division); age and its square; education level (less than upper-secondary; upper- or post-secondary; tertiary); place of birth (native-born/immigrant); minority status (whether or not the respondent was originally from the EU in the EQL data – the best available proxy; self-assessed minority status in the ESS data); self-assessed health (excellent; good; fair; bad; very bad) to control for potential endogeneity; part-time employment (30 or fewer hours a week); and whether or not the respondent lived with a spouse.

Several additional individual-level controls were then added to reflect factors that might uniquely relate to unpaid and paid carers. For unpaid care, the models incorporated the number of children in the household (0; 1; 2; 3+), whether or not the respondent lived with a disabled adult, and the respondent’s employment status (employed; unemployed; student; homemaker; on disability; retired). For paid care workers, additional controls were added for union membership, working in the public sector, working in a supervisory role, and being self-employed.

Finally, both sets of models also included national-level controls for: logged GDP per capita, to capture potential (non-linear) dynamics tied to economic development levels; inequality, as measured by the Gini coefficient; and a country’s unemployment rate. All national-level control data were taken from Eurostat (Eurostat, 2020). OA Tables 4 and 5 present descriptive statistics for every variable included in both sets of analysis.

Analyses

Our main analysis relies on multi-level models – using maximum likelihood estimation and including both population and design weights – with respondents nested in their respective countries to reflect the structure of the survey data. This approach should return more conservative estimates for both our direct and interactive effects (see Hox et al., 2017), which is especially important given the limited number of country clusters. All regressions include random intercepts for each country and random slopes for the key individual-level variables, to incorporate cross-country variation that would otherwise be unaccounted for in the models. We then conclude our analyses by testing alternative model specifications to ensure that our main results are not impacted by modelling choices.

Turning first to analyse unpaid care work, Figure 2 presents the predicted marginal effect of unpaid caring on psychological wellbeing when holding all other variables – i.e., those not in a given interaction – at their mean (see OA Table 6 for underlying regression results). As such, the lines in the figure indicate the predicted effect size of performing one hour of unpaid care, broken down by the type of care recipient (across the two panels) and the level of social expenditure in a given country (along the x-axis). All marginal effects plots also include 95% confidence intervals to illustrate statistically significant effects at the p < 0.05 level. Here and below, extreme values are excluded from the marginal effects plots and the x-axis is restricted to values between the 10^th and 90^th percentile of social expenditure, so as to foreground more typical marginal effect sizes.

Figure 2.

Predicted marginal effect of unpaid care work (95% CIs) by social expenditure rate, disaggregated by type of care recipient.

Results suggest that while unpaid caring for children has no perceivable relationship with psychological wellbeing (with a direct effect statistically indistinguishable from zero), unpaid caring for adults is associated with a decrease in psychological wellbeing (β = −0.00191, p < 0.01). When we take into account the interaction, however, we note that this dynamic is limited to countries with relatively small welfare states – becoming statistically indistinguishable from zero once social expenditure reaches about 27% of GDP. Even in low social expenditure countries, however, marginal effect sizes are modest: the predicted effect size at the 10^th percentile value of social expenditure, for example, is approximately −0.004. Moving across the 10^th to 90^th percentile value of time spent on unpaid adult care (from 0 to 5), for instance, would therefore be associated with a decrease in psychological wellbeing of 0.020 (i.e., just over 2% of a standard deviation). The findings here thus provide some support for our expectation that care for adults would be more strongly associated than care for children with lower levels of psychological wellbeing (as per H1), but the size of the marginal effect is small. The association with social expenditure, in turn, reflects the expected dynamics as well, as we only find evidence of a moderating effect regarding care for adults (as per H3).

Turning to paid care work, Figure 3 presents the predicted marginal effect of being employed in the care sector (as compared to all other jobs), broken down by level of occupational professionalism (across the three panels) and social expenditure levels (across the x-axis). The figure thus adopts the same approach to illustrating interactions as Figure 2, presenting us with the marginal effect of paid care work across our three professionalism levels as well as a range of social expenditure levels (for underlying regression results, see OA Table 7).

Figure 3.

Predicted marginal effect of paid care work by social expenditure rate, disaggregated by level of occupational professionalism.

Here again, the negative association with care work is restricted to countries with lower levels of social expenditure – dropping off entirely at about 26% of GDP – and only for certain types of care: in this case, paraprofessionals are the only group for which care work was correlated with lower levels of psychological wellbeing (β = −0.171, p < 0.05). As an illustration, the marginal effect sizes at the 10^th percentile value of social expenditure suggest that for paraprofessionals, being employed in care work would be associated with a 0.505 decrease in psychological wellbeing (i.e., 57% of a standard deviation) versus comparable workers in non-caring occupations. Yet no similar results are noted among professional care workers. This offers only mixed support for the expectation that being in paid care work would be associated with lower levels of psychological wellbeing among more professionalized care workers (H2). Similarly, results suggest that higher social expenditure levels are associated with a smaller care penalty on psychological wellbeing (as per H4), though once again this applies only to paraprofessional paid care workers.

Finally, additional analyses were used to confirm the robustness of the findings on both unpaid and paid care work. First, we minimize the possibility that key results were not simply artefacts of the precise control set included in the full regressions, via models using alternative specifications with restricted sets of national- and individual-level variables (see Models 1, 2, and 3 in OA Tables 6 and 7). Second, remove-one jack-knife analysis was used to assess whether results were driven by any particular country in the sample (see Model 1 of OA Tables 8 and 9). Third, results were tested to confirm that they were robust to changes to the modelling strategy, using alternative regression models that included cluster robust standard errors and excluded random slopes (see Models 2 and 3 of OA Tables 8 and 9). Finally, we examined whether our main results were robust to using an additional categorical control variable for welfare state types (Grönlund and Öun, 2018; Korpi et al., 2013), so as to control for potential differences in the meaning of social expenditure across various welfare regime types (see Model 4 of OA Tables 8 and 9). In no instance were our primary findings meaningfully affected.⁴

Conclusions

This study uses cross-national analyses to provide insights into the relationship between care work, psychological wellbeing, and national social expenditure. In doing so, it builds on existing theory highlighting the potential negative and positive effects of care work on psychological wellbeing. While many studies find that care work is associated with high levels of burnout and anxiety and worsened work-life balance, other studies provide empirical findings of positive associations, especially for lower paid care workers who may feel that they are “making a difference” (e.g. Henle et al., 2020; Lightman and Kevins, 2019).

Our starting point was the Stress Process Model (Pearlin, 1989), which highlights the role of stressors, resources, and coping mechanisms in shaping wellbeing. At the same time, however, our theoretical framing also reflects empirical work suggesting that compensating differentials may play a role in the case of unpaid care for children – due to positive associations not seen in unpaid care for adults – and non-professional care workers – due to nonpecuniary benefits tied to helping others that may be more important for these workers. Our expectations thus reflect both theoretical and empirical studies, which suggest that, under certain conditions, care work may also have positive associations with psychological wellbeing (Lightman and Kevins, 2019; Nomaguchi and Milkie, 2020).

While it is difficult to draw out causal mechanisms via cross-sectional studies – a limitation which is particularly acute in this research area given that decisions to provide care may be endogenous (Dang et al., 2022) – our findings suggest that social expenditure can play a role in shaping the relationship between care work and wellbeing for certain carers. In the case of unpaid care work, caring for children is not found to be associated with psychological wellbeing, indicating that both positive factors (e.g., assisting in child development and play) and negative factors (e.g., a loss of leisure time and challenges balancing paid and unpaid work) may be cancelling each other out. Yet, for individuals providing unpaid care for adults, the models point to a small decrease in psychological wellbeing – albeit only in countries with smaller welfare states. Thus, for unpaid carers, social expenditure may act as a protective factor (e.g., through defamilializing programs, policies, and tax allowances) for those doing the challenging work of assisting adults in need.

With paid care work, results suggest that only in the case of paraprofessional workers (e.g., medical assistants, nurse or midwifery assistants, teaching aides) did care work have a negative association with psychological wellbeing. Marginal effect sizes here were notably larger than what we found regarding unpaid care work, though once again only in countries with smaller welfare states. Results thus broadly align with the study’s expectations, but with one exception: there is no evidence of reduced wellbeing among professional care workers. For these workers, it may be that increased autonomy, prestige, and decision-making power, more likely at higher levels of occupational professionalization, play a larger role than anticipated and thereby buffer any negative psychological association specific to care. Lower psychological wellbeing among paraprofessionals, in turn, may be driven by the mix of high demands and low control within their professionalization level, as suggested by Karasek’s (1979) Job Strain Model – indicating that research within this tradition may have greater explanatory power in the paid care market.

Taken as a whole, results suggest that social expenditure is a pertinent policy lever that may help to address negative psychological impacts for carers. But the data also highlight major variation tied to the type of care recipient and the type of paid care job being performed. For policymakers, this suggests that programs and tax allowances may be best used to target those doing care work, both paid and unpaid, that are the most at risk of poor psychological wellbeing – specifically, unpaid carers for adults and paraprofessionals care workers. For scholars of care work, the findings underscore the importance of focusing on the potential psychological implications of care, and the ways in which structural and policy context can play a meaningful role.

Drawing from these insights as well as the limitations of our study, we conclude by highlighting several promising avenues for further research. First, research that employs familism/defamilism indices in place of social expenditure (e.g. Keck and Saraceno, 2012) would allow for a more nuanced understanding of the contextual effects discussed above. The development of new and/or updated familialism/defamilialism measures that could account for any association between gender/family regimes and wellbeing would bolster existing findings, while also recognizing that both types of policies may simultaneously be in place in a given country, and that social spending in some cases may, in fact, subsidize a familial caregiving scheme (Lohmann and Zagel, 2016). Second, despite our theoretically-derived focus on care as a determinant of psychological wellbeing, it is also possible that the causal arrow runs in the opposite direction: it may be that those with lower/higher levels of psychological wellbeing are more likely to engage in care work (for research highlighting care-related endogeneity issues, see: Dang et al., 2022; Johnson and Lo Sasso, 2006). Future work using panel data would therefore be especially insightful, given that individual fixed effects cannot control for preferences and other time-invariant characteristics that might affect psychological wellbeing (Oshio and Usui, 2017). Finally, additional research is needed to assess the disaggregated impact of different social policies (e.g., specific policies on eldercare, childcare) on the psychological wellbeing of carers and to analyze the specific mental health implications of both paid and unpaid care on populations disproportionately performing this work, such as women, minorities, and indigenous peoples.

Supplemental Material

Supplemental Material - When caring comes at a cost: Psychological wellbeing of unpaid and paid carers and the role of social expenditure

Supplemental Material for When caring comes at a cost: Psychological wellbeing of unpaid and paid carers and the role of social expenditure by Naomi Lightman, and Anthony Kevins in Journal of European Social Policy

Footnotes

ORCID iDs

Naomi Lightman

Anthony Kevins

Funding

This work was supported by the Social Sciences and Humanities Research Council of Canada (grant number 435-2021-0486) and by the British Academy/Leverhulme Small Research Grant (grant number 231164).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Amate-Fortes

Guarnido-Rueda

Martinez-Navarro

, et al. (2024) Are European social spending policies effective in the fight against gender inequality? Journal of Contemporary European Studies 32(1): 157–171.

Aneshensel

Avison

(2015) The stress process: an appreciation of Leonard I. Pearlin. Society and Mental Health 5(2): 67–85.

Browning

Penning

(2020) Family caregiver mental health: linking family care regime, intersectionality, and stress process frameworks. Innovation in Aging 4(Suppl 1): 500.

Charmes

(2022) Variety and change of patterns in the gender balance between unpaid care-work, paid work and free time across the world and over time: a measure of wellbeing? Wellbeing, Space and Society 3: 100081.

Collins

(2019) Making Motherhood Work. Princeton University Press.

Dang

H-AH

Hiraga

Nguyen

(2022) Childcare and maternal employment: evidence from Vietnam. World Development 159: 106022.

De Oliveira

Hlebec

(2016) Predictors of satisfaction with life in family carers: evidence from the third European quality of life survey. Teorija in Praksa 53(2): 1.

Duffy

Baughman

Smith

(2021) The flip side of turnover: employment transitions and occupational attachment among low-wage care workers in the United States. Feminist Economics 27(4): 62–89.

England

Budig

Folbre

(2002) Wages of virtue: the relative pay of care work. Social Problems 49(4): 455–473.

10.

Esping-Andersen

Schmitt

(2020) Multi-dimensional couple bargaining and housework allocation. Acta Sociologica 63(1): 3–22.

11.

ESS (2014) European social survey round 7 data.

12.

European Foundation for the Improvement of Living and Working Conditions (2018) European Quality of Life Survey Integrated Data File, 2003–2016.

13.

Eurostat (2020) European labour force survey.

14.

Folbre

(2012) For Love or Money: Care Provision in the United States. New York, NY: Russell Sage Foundation.

15.

Folbre

Gautham

Smith

(2021) Essential workers and care penalties in the United States. Feminist Economics 27(1-2): 173–187.

16.

Gencer

Brunnett

Staiger

, et al. (2024) Caring is not always sharing: a scoping review exploring how COVID-19 containment measures have impacted unpaid care work and mental health among women and men in Europe. PLoS One 19(8): e0308381.

17.

Glavin

Peters

(2015) The costs of caring: caregiver strain and work-family conflict among Canadian workers. Journal of Family and Economic Issues 36(1): 5–20.

18.

Grönlund

Öun

(2018) The gender-job satisfaction paradox and the dual-earner society: are women (still) making work-family trade-offs? Work 59(4): 535–545.

19.

Hagqvist

(2018) Toward gender equality in practice? Cross-national patterns of change in the gendered division of housework over two decades. Journal of Comparative Family Studies 49(3): 355–377.

20.

Hebson

Rubery

Grimshaw

(2015) Rethinking job satisfaction in care work: looking beyond the care debates. Work, Employment & Society 29(2): 314–330.

21.

Henle

Fisher

McCarthy

, et al. (2020) Eldercare and childcare: how does caregiving responsibility affect job discrimination? Journal of Business and Psychology 35(1): 59–83.

22.

Hook

Paek

, et al. (2023) National work–family policies and the occupational segregation of women and mothers in European countries, 1999–2016. European Sociological Review 39(2): 280–300.

23.

Hox

Moerbeek

Van de Schoot

(2017) Multilevel Analysis: Techniques and Applications. Routledge.

24.

Huang

Xie

Cheung

, et al. (2022) Job demands and resources, positive and negative affect, and psychological distress of social workers in China. Frontiers in Psychiatry 12: 2617.

25.

Jacobs

Van Houtven

Tanielian

, et al. (2019) Economic spillover effects of intensive unpaid caregiving. PharmacoEconomics 37(4): 553–562.

26.

Johnson

Lo Sasso

(2006) The impact of elder care on women's labor supply. INQUIRY: The Journal of Health Care Organization, Provision, and Financing 43(3): 195–210.

27.

Karasek

(1979) Job demands, job decision latitude, and mental strain: implications for job redesign. Administrative Science Quarterly: 285–308.

28.

Keck

Saraceno

(2012) Multilinks database on intergenerational policy indicators. Journal of Contextual Economics–Schmollers Jahrbuch(3): 453–461.

29.

Korpi

Ferrarini

Englund

(2013) Women's opportunities under different family policy constellations: gender, class, and inequality tradeoffs in western countries re-examined. Social Politics 20(1): 1–40.

30.

Lightman

(2024) Converging economies of care? Immigrant women workers across 17 countries and four care regimes. Journal of Industrial Relations 66(1): 79–103.

31.

Lightman

Kevins

(2019) Bonus or burden? Care work, inequality, and job satisfaction in eighteen European countries. European Sociological Review 35(6): 825–844.

32.

Lohmann

Zagel

(2016) Family policy in comparative perspective: the concepts and measurement of familization and defamilization. Journal of European Social Policy 26(1): 48–65.

33.

Lopez

(2006) Emotional labor and organized emotional care: conceptualizing nursing home care work. Work and Occupations 33(2): 133–160.

34.

Maguire

Hanly

Maguire

(2019) Beyond care burden: associations between positive psychological appraisals and well-being among informal caregivers in Europe. Quality of Life Research 28(8): 2135–2146.

35.

Meagher

Szebehely

Mears

(2016) How institutions matter for job characteristics, quality and experiences: a comparison of home care work for older people in Australia and Sweden. Work, Employment & Society 30(5): 731–749.

36.

Nomaguchi

Milkie

(2020) Parenthood and well‐being: a decade in review. Journal of Marriage and Family 82(1): 198–223.

37.

OECD (2020) Social expenditure: aggregated data. Paris.

38.

Oshio

Usui

(2017) Informal parental care and female labour supply in Japan. Applied Economics Letters 24(9): 635–638.

39.

Parbst

Wheaton

(2023) The effect of welfare state policy spending on the equalization of socioeconomic status disparities in mental health. Journal of Health and Social Behavior 64(3): 336–353.

40.

Park

Han

Torabi

, et al. (2020) Managing mental health: why we need to redress the balance between healthcare spending and social spending. BMC Public Health 20(1): 1–8.

41.

Pearlin

(1989) The sociological study of stress. Journal of Health and Social Behavior: 241–256.

42.

Pearlin

Mullan

Semple

, et al. (1990) Caregiving and the stress process: an overview of concepts and their measures. The Gerontologist 30(5): 583–594.

43.

Radloff

(1977) The CES-D scale: a self-report depression scale for research in the general population. Applied Psychological Measurement 1(3): 385–401.

44.

Ragnarsdóttir

Kostecki

Gornick

(2023) Accounting for the value of unpaid domestic work: a cross-national study of variation across household types. European Sociological Review 39(2): 262–279.

45.

Reibling

Beckfield

Huijts

, et al. (2017) Depressed during the depression: has the economic crisis affected mental health inequalities in Europe? Findings from the European Social Survey (2014) special module on the determinants of health. The European Journal of Public Health 27(suppl_1): 47–54.

46.

Rodriguez-Loureiro

Vives

Martínez Franzoni

, et al. (2020) Health inequalities related to informal employment: gender and welfare state variations in the Central American region. Critical Public Health 30(3): 306–318.

47.

Sjöberg

(2023) Work-retirement transitions and mental health: a longitudinal analysis of the role of social protection generosity in 11 countries. Scandinavian Journal of Public Health 51(1): 90–97.

48.

Somogyi

Van Lancker

Ciccia

, et al. (2021) The relationship between familizing and individualizing policies and mental health in parents in Europe. Social Sciences 10(2): 55.

49.

Topp

Østergaard

Søndergaard

, et al. (2015) The WHO-5 Well-Being Index: a systematic review of the literature. Psychotherapy and Psychosomatics 84(3): 167–176.

50.

Treier

Hillygus

(2009) The nature of political ideology in the contemporary electorate. Public Opinion Quarterly 73(4): 679–703.