Table 2 provides a description of the sample and the differences between within-household caregivers and non-caregivers in the individual-level variables included in our multivariate models. Relative to non-caregivers, caregivers were more likely to have lower levels of education, income (poor to low middle income), and wealth. Caregivers were also more likely to be older, female, living in larger households, not working, and to report poorer health relative to their non-caregiving counterparts.
Table 3 presents the distribution of care provision within the household, individual socioeconomic resources (education, income, and wealth), and country-level indicators across the 21 countries in our study. The average prevalence of care provision inside the household was 7% but varied cross-nationally from the lowest observed in Sweden (4%) to the highest in Portugal (12%). Several Southern and Eastern European countries also showed high prevalence of care provision within the household including Croatia, Italy, Czech Republic, Spain, Hungary, and Poland (9 to 10%).
Regarding macro-level socioeconomic resources, income inequality varied within regions as some Northern and Western countries showed relatively high levels (e.g., Luxembourg, France, Switzerland, and England with an average Gini index of 32 to 35) while other Northern and Western countries had relatively low income inequality (e.g., Sweden, Denmark, the Netherlands with an average Gini index of 27 to 29). Portugal showed the highest income inequality with an average Gini index of 36. Finally, average social expenditure as a percentage of GDP ranged from a low of 16 to 18 percent in Estonia, Ireland, Czech Republic, and Poland to 31 percent in Denmark.
Table 4 presents the results of our multivariate analyses of care provision within the household. M2 and M3 examined the association between individual-level socioeconomic resources and caregiving within the household, without (M2), and with (M3) individual-level covariates. Our unadjusted model, M2, showed that low education and income (specifically low medium income) were positively associated, whereas higher levels of education and income were negatively associated, with the incidence of caregiving within the household. Regarding wealth differences, wealthier households (50,000 or more Euros) showed a lower incidence of caregiving within the household relative to less wealthy households. In brief, providing informal care was more common among individuals with lower relative to higher SES, and this finding held for all SES indicators.
Adjusting for individual-level covariates (M3) did not change the directions but attenuated the strength of the associations between individual-level socioeconomic resources (education, income, and wealth) and care provision within the household. The addition of covariates significantly improved the model and was a better fit for the data based on the statistically significant likelihood ratio tests as well as the AIC and BIC statistics. Regarding our sociodemographic covariates, we found informal care provision to be positively associated with age, household size, and women were more likely than men to provide informal care within the household. Employment, full-time and part-time (relative to not working), was negatively associated with informal care provision. Regarding respondents’ health, we found that having one or more limitations with instrumental activities of daily living was positively associated with informal care provision. While having limitations with instrumental activities of daily living could imply that these older adults may also need help themselves, our finding suggested that these respondents might be reciprocating care they received either currently or in the past or were surrounded by people with care needs as network members typically have similar characteristics.
M4 and M5 show the associations between country-level characteristics and care provision within the household. Accounting for the country’s level of income inequality (M4) as measured by the Gini coefficient did significantly improve model fit, compared to M3. It was positively associated with care provision within the household. Furthermore, the direction and strength of associations between individual-level income and wealth, and care provision within the household as observed in M3, were maintained. Additional analyses, not shown, indicated that even after controlling for the country’s overall wealth, the direction and strength of the association between within-country income inequality and care provision within the household were unchanged. Thus, countries with higher levels of income inequality, on average, showed a slightly higher incidence of care provision within the household. The inclusion of the country’s level of social expenditure (M5) also significantly improved model fit compared to M3. Results showed that countries with higher levels of social expenditure, on average, exhibited a slightly lower incidence of care provision within the household. Thus, our findings suggested that socioeconomic inequalities at the macro-level were associated with an increased propensity for older adults’ provision of informal care, whereas higher social protection expenditure within countries was associated with a lower propensity for older adults’ informal care provision.
Across all models, the estimated country-level variance indicated there was significant between-country variation in care provision within the household among the countries included in our study. Notably, the addition of our macro-indicators (Gini coefficient and social spending as a % of GDP) substantially decreased the AIC and BIC (e.g., 1173 and 992 in M4, respectively when accounting for income inequality), compared to the empty model. Thus, the country-level indicators improve predictive power and explain some of the cross-national variation in care provision within the household. Yet, the variance remained statistically significant suggesting additional unobserved heterogeneity in informal care provision across countries.