The impact of education on instrumental activities of daily living: evidence from Europe using SHARE data

The ageing population across Europe is growing rapidly, posing substantial challenges to healthcare systems and national economies. As life expectancy increases, the demand for elderly care services and the pressure on public health systems intensify, raising concerns about the sustainability of economic growth and social welfare (Yasuoka, 2019). Older adults typically require more intensive and prolonged healthcare services, which drives up healthcare expenditures and exerts pressure on public finances (OECD, 2021). This demographic shift also affects labour markets by reducing workforce participation and increasing dependency ratios, which can hinder economic growth and strain social welfare systems (Bloom et al., 2010). Age-related declines in functional abilities are increasingly threatening older adults' capacity to manage instrumental activities of daily living (IADLs) ^[1]. About 30% of older Europeans report difficulties with performing at least one IADL.

Beyond its well-documented links to labour income share and gender gaps (Çelik, 2022), formal education is robustly associated with healthier ageing. European panel data (SHARE) show that higher education predicts slower subsequent increases in ADL/IADL limitations and physical decline from ages 50 to 80, consistent with cumulative advantage over the life course (Leopold and Engelhardt, 2013). Cross-national analyses of eight harmonised cohorts similarly find large, persistent educational inequalities in healthy-ageing trajectories, with tertiary education associated with substantially higher baseline and maintained healthy-ageing scores (Wu et al., 2020). At the cognitive level, an integrative analysis across international cohorts indicates that education delays the onset of accelerated cognitive decline in patterns consistent with Alzheimer's-related pathology, aligning with cognitive-reserve theory (Stern, 2012; Clouston et al., 2015). Finally, a recent global systematic review and meta-analysis shows that each additional year of schooling reduces adult all-cause mortality risk by ∼2%, with benefits extending into older age (Sørensen et al., 2024).

The relationship between education and health has been widely studied. Education can enhance individuals' health by improving their knowledge, shaping healthier behaviours and increasing access to healthcare resources. Higher educational attainment is often linked to better cognitive functioning, healthier lifestyles, and greater ability to manage chronic conditions, all of which contribute to improved health outcomes and functional independence in later life (Cutler and Lleras-Muney, 2012). Moreover, individuals with higher education tend to be employed in safer workplaces, live in communities with stronger support systems and associate with social networks that emphasize healthy behaviours, all of which promote improved health outcomes (Brunello et al., 2016). Despite existing evidence, the impact of education on IADL difficulties has not yet been thoroughly examined from the perspectives of gender and cultural differences. Examining this relationship for older adults is important. First, older adults are more likely to experience chronic conditions and functional limitation more than any other age group. Second, Europe suffering from a significant increase in older adults. Projections suggest that by 2050, nearly one-third of Europe's population will be aged 65 or older (European Commission, 2021).

Despite extensive work on education–health links, there is little causal evidence on how schooling affects older adults' functional independence (IADLs), especially in pan-European settings and with heterogeneity by gender and culture; most quasi-experimental studies emphasise mortality, smoking, obesity or self-rated health rather than IADLs (Galama et al., 2018). We address this gap by exploiting compulsory-schooling reforms in an IV-Probit framework using harmonised SHARE data (2006–2022) by estimating gender- and culture-specific effects. The European countries are an essential test bed because it is among the world's oldest regions demographically, with rapidly rising old-age dependency ratios and significant long-term care and fiscal implications, making evidence on preserving IADL independence directly policy-relevant.

This article examines the impact of years of education on IADLs in 7 European countries (Austria, Denmark, Belgium, Spain, Italy, France and the Czech Republic) using the Survey of Health, Ageing and Retirement in Europe (SHARE). It covers the period 2006–2022 for people aged 50 and over. Europe is ageing rapidly: the EU old-age dependency ratio is projected to climb from about 34% in 2019 to roughly 55% by 2050, implying many fewer workers per older adult (European Commission, 2024). The European Commission's 2024 Ageing Report anticipates sustained pressure on pensions, health and especially long-term care; in risk scenarios, ageing-related costs could rise by about 4% points of GDP by 2070 (European Commission, 2024). Against this backdrop, we consider two pathways, compression of morbidity (more healthy, independent years) versus expansion of morbidity (more years with limitations), and estimate how education's effect on IADLs can tilt outcomes toward the former. We exploit the compulsory schooling as instrumental variable. We estimate the impact of education on IADLs by gender and across different European regions.

The rest of this article structured as follows: Section 2 presents the literature review. Section 3 presents data. Results are presented and discussed in Section 4. Section 5 discussion. Section 6 concludes the article.

Population ageing poses significant economic and social challenges, particularly through rising healthcare costs and declining functional independence among older adults. Kopecky (2023) demonstrates that population age structure significantly influences bilateral trade flows, with ageing societies experiencing diminished trade capacity due to a shrinking working-age population. The ability to perform IADLs is essential for maintaining autonomy and reducing care needs. Zhao et al. (2022) argue that educational interventions not only enhance health literacy but also empower individuals to better manage chronic conditions and navigate complex care systems, thereby supporting integrated care outcomes. Dolgikh and Potanin (2024) document that higher educational attainment significantly boosts individual productivity and human capital formation, suggesting that better-educated seniors possess the cognitive, financial and informational resources necessary to manage daily tasks more effectively. This aligns with broader findings that integrated care models benefit from incorporating educational strategies to reduce dependency and improve self-management (Correia de Matos et al., 2025). Opoku et al. (2025) analyse labour-force participation among Ghanaians aged 60 and over and find that each additional year of formal education increases the probability of continued workforce engagement, an indicator of preserved capacity to perform complex IADLs independently.

The impact of education on health is well established in the economics literature. Some studies find a positive impact (Lleras-Muney, 2005; Brunello et al., 2016; Fonseca et al., 2020) while others find no impact (Galama et al., 2018). Clark and Royer (2013) examine the effect of education on adult mortality and health using data from Britain, specifically leveraging changes in compulsory schooling laws as an instrumental variable. Their analysis draws on administrative and survey data to estimate the impact of additional schooling on long-term health outcomes. They find that, contrary to much of the existing literature, increased education does not lead to significant improvements in mortality rates or most health measures. These findings challenge the assumption that education universally enhances health and suggest that the relationship may be more context-dependent than previously thought.

Cavelaars et al. (1998) examine educational differences in self-reported morbidity across 11 Western European countries using harmonised national health surveys (1985–1993); it finds that lower education is consistently associated with higher prevalence of less-than-good self-rated health, chronic conditions, long-term disabilities and longstanding illness, with inequality magnitudes – measured by the Relative Index of Inequality – generally larger in Sweden, Norway and Denmark and smaller in Spain, Switzerland and (West) Germany.

Leopold and Engelhardt (2013) examine the impact of educational attainment on trajectories of physical health in later life using SHARE Waves 1–2 (ages 50–80), covering ADL, IADL, mobility, chronic diseases, self-rated health and grip strength; it finds that higher education is linked to slower decline and that the education-health gap widens with age for physical functioning (ADL/IADL, mobility) and grip strength, while remaining roughly constant for chronic diseases and self-rated health.

Brunello et al. (2016) examine the impact of education on self-reported health, focusing particularly on the role of health behaviours as mediators. Using data from individuals aged 50 and older across 12 European countries. They show that education has a protective effect on self-reported health and influences health behaviours such as smoking, drinking and exercise.

Fonseca et al. (2020) examine the causal relationship between education and health outcomes using data from SHARE. Their findings suggest that an additional year of education significantly reduces the probability of reporting poor health and lowers the likelihood of experiencing limitations in both ADLs and IADLs. However, the study also notes that education does not have a statistically significant effect on the incidence of specific health conditions such as cancer, stroke or psychiatric illness.

Xie et al. (2022) investigate how perceived health competence and prior health education influence health-promoting behaviours among older adults living in rural China. The study finds that individuals with greater confidence in managing their health and more exposure to health education are more likely to engage in beneficial practices such as regular physical activity, nutritious eating, stress reduction, maintaining social connections and taking proactive steps to care for their health.

Kosorukova et al., (2025) examine the treatment effect of higher education on multiple health outcomes in Russia using age-cohort models and two identification strategies, parametric (multivariate recursive probit) and nonparametric matching; it finds for women significant negative effects of higher education on obesity and hypertension and a positive effect on reporting high self-rated health, while for men it finds positive effects on the likelihood of heart disease, hypertension, allergy and high self-rated health, with effect sizes differing by birth cohort and partial support for a cohort-diminishing pattern.

Despite extensive work on education and later-life health, causal evidence on functional independence (specifically IADLs) in pan-European settings is scarce. Existing studies seldom make IADLs the primary endpoint and rarely examine whether effects differ by gender or cultural context, where theory suggests strong effect modification. Coverage is also narrow in countries and waves, limiting policy relevance as population ageing accelerates. This study addresses these gaps by centring IADLs, leveraging cross-country schooling variation for identification, and testing gender and culture-specific heterogeneity within a harmonised, multi-wave European framework.

We contribute to the literature in two ways. First, while Fonseca et al. (2020) examine the impact of education on health using self-reported health, ADLs and IADLs as health indicators, they used IADLs as a secondary or supplementary indicator of health. However, this is the first article to use IADLs as a primary health outcome. By focusing specifically on IADLs, which reflect the ability to live independently and manage complex daily tasks, this study provides a more targeted understanding of how education influences functional health in older age. Second, we divide European countries into two cultural groups, individualistic and collectivistic, based on Hofstede's (1984) cultural dimensions. This classification enables us to investigate whether the effect of education on functional health, measured through IADLs, differs across cultural contexts.

This study draws on the Survey of Health, Ageing and Retirement in Europe (SHARE), a rich, multidisciplinary panel dataset covering health, socioeconomic status, social networks and other facets of later-life well-being across 28 European countries and Israel (Börsch-Supan et al., 2013). Fielded in nine waves between 2004–2005 and 2021–2022, SHARE offers detailed, harmonized measures that enable cross-national comparisons of ageing trajectories. For our analysis, we focus on wave 2 (2006), 4 (2010) 5 (2013), 6 (2015), 7 (2017), 8 (2019) and 9 (2022) examining adults aged 50+ in 7 European nations (Austria, Spain, Italy, France, Denmark, Belgium and the Czech Republic). Other waves and countries are excluded due to incomplete data ^[2].

Education is defined as the total number of years of formal, full-time schooling completed by the respondent, ranging from 0 (no formal education) to 25 (equivalent to advanced postgraduate study). This measure is based on self-reported educational attainment and has been standardised across countries to account for variations in national education systems.

The IADL index captures the respondent's level of difficulty with certain daily tasks. Participants were asked: “Please tell me if you have any difficulty with these activities because of physical, mental, emotional or memory problems.” The activities assessed include making telephone calls, taking medication, managing finances, preparing a hot meal and shopping for groceries. For each of the five activities, a response indicating difficulty is coded as 1, while no difficulty is coded as 0. The total index score represents the sum of difficulties across these tasks, ranging from 0 (no difficulties) to 5 (difficulty with all activities). We make it dummy variable, individuals who reported no difficulties in any of the five activities were coded as 0, while those reporting difficulty in at least one activity were coded as 1.

A range of control variables is employed to isolate the effect on IADLs. These include age, age squared, marital status dummies, employment status dummies, household size, number of chronic conditions, wave dummies and country controls. Further details on these variables are provided in the  Appendix.

About 53.5% of our sample are women. The average age is around 67.48 years during the survey periods. Summary statistics are presented in Table 1.

Table 1 presents the summary statistics of the study sample, disaggregated by cultural context and IADL status. The results indicate clear differences between individuals with and without IADL difficulties. In both collectivistic and individualistic countries, respondents with IADL impairments are notably older, with mean ages of 75.5 and 73.4 years, respectively, compared to 66.2 and 65.2 years among those without difficulties. Educational attainment also differs substantially: in collectivistic contexts, individuals without IADL difficulties report an average of 10.7 years of education, compared to 8.5 years among those with limitations. A similar pattern is observed in individualistic countries, where the respective averages are 12.0 and 10.6 years. Gender composition also varies, as men represent a smaller share of the population with IADL difficulties across both cultural groups. Household size is generally lower among respondents with IADL difficulties, suggesting a greater likelihood of living alone or in smaller households. Interestingly, the number of children does not differ markedly by IADL status, though it is slightly higher among those reporting difficulties in collectivistic contexts. Overall, the descriptive statistics highlight strong associations between older age, lower educational attainment, smaller household size and the presence of functional limitations, providing a foundation for the subsequent analysis.

To examine the causal effect of education on difficulties in IADLs using a probit model with a potentially endogenous regressor. Let $IADLit∈{0,1}$ indicate whether individual i at time t reports any IADL limitation. Define the latent-index representation.

Because education may be correlated with unobservable that also affect functional limitation (reverse causality and selection), we instrument $EDit$ with exposure to compulsory-schooling reforms.

The First-stage equation is:

Where $CSit$ is an indicator for cohort-by-country exposure to compulsory-schooling laws that raised minimum schooling. Validity requires (1) relevance: $CSit$ shifts schooling (strong first stage) and (2) exclusion: $CSit$ affects IADLs only through schooling. Compulsory-schooling instruments are standard in the literature for generating exogenous variation in education Angrist and Krueger, 1991; Lleras-Muney, 2005; Jürges et al., 2013).

We estimate IV-Probit using both maximum likelihood (ML) and Newey's two-step (TS) estimators. Under the ML approach, we assume a joint normal distribution for the disturbances in the outcome and first-stage equations:

So that $ρ≠0$ captures endogeneity of $EDit$⁠. We report (1) the estimated correlation parameter via $ρˆ(orathρˆ)$ and the likelihood-ration test of $H0:ρ=0$ (test of exogeneity).

For ML, and (2) the Wald test of exogeneity for the two-step estimator. For instrument strength, we report standard first-stage diagnostics (coefficient on $CSit$ and first-stage F-statistic).

Our parameter of interest is the average partial effect (APE) of an additional year of education on the probability of any IADL difficulty. For a probit index, the APE is:

Where $Φ(.)$ is the standard normal density. Consistent with standard practice, APEs are evaluated at the observed covariates (including $EDit$⁠) and standard errors are obtained by the delta method (Newey, 1987). All specifications include country fixed effects and wave fixed effects to absorb time-invariant cross-country heterogeneity and common shocks across SHARE waves. Standard errors are clustered at the individual level to accommodate heteroskedasticity and serial correlation within persons over time.

We use the cross-country variation of compulsory schooling laws over time as an instrument variable of education. Our hypothesis is that different compulsory schooling laws can affect education differently across birth cohorts and across countries in an exogenous way, given that the laws can change over time and/or by country. Since individuals in our sample are aged 50 years and older, we consider compulsory schooling laws that would impact individuals in 1960 and 1970s as shown in Table 2.

Table 2 shows the results obtained from probit IV-Probit estimator estimations.

The probit estimates, presented in the second column of Table 3, indicate a slight negative correlation between education and IADLs. However, as previously discussed, this relationship may be affected by endogeneity. To address this issue, we employ the IV-Probit estimator. In the IV-Probit model, the first stage confirms a positive and significant correlation between years of education and compulsory schooling. They show that the instrumental variable coefficient is 0.334 and statistically significant, indicating that the instrument is strongly correlated with the endogenous regressor (education), thus supporting the validity of the IV approach.

The second stage shows that years of education decrease IADL difficulties. An additional year of education decreases IADLs by 13.9%. Education enhances cognitive functions, which are crucial to manage complex daily tasks; it can also positively influence health through several pathways. Individuals with more years of education tend to have higher health literacy, better problem-solving skills and greater access to health-promoting resources, all of which contribute to healthier lifestyles and improved self-management of chronic conditions (Wilkie et al., 2024).

In this section, we conduct a heterogeneity analysis by gender. Given the well-documented differences in health outcomes, life expectancy, and social roles between men and women in later life, it is plausible that the effect of education on IADLs may vary by gender. Results shown in Table 4.

The results in Table 4 show that for women, the coefficient on education is −0.178, indicating that each additional year of education reduces IADLs. In contrast, the effect for men is smaller and statistically insignificant. A plausible interpretation of Table 4 is that education reduces IADL limitations mainly for women because (1) it boosts health literacy, planning and navigation of care; (2) it expands economic resources and intra-household bargaining power and (3) among the cohorts observed, women start from a higher baseline risk of functional limitation and historically lower schooling, so each additional year yields larger marginal returns. This aligns with evidence showing women have higher IADL limitation risk than men across European regions, the classic so protective factors like education can translate into larger gains for women (Scheel-Hincke et al., 2020).

In this section, we investigate whether the effect of education on IADLs differs across cultural contexts. Drawing on Hofstede's (1984) cultural dimensions, we classify European countries into two groups: individualistic (Spain, Italy, France and the Czech Republic) and collectivistic (Austria, Denmark and Belgium). This classification enables us to explore whether societal values and norms influence the relationship between educational attainment and the ability to perform IADLs. We classify cultural context using Hofstede's individualism–collectivism construct from the 6-D model of national culture, which conceptualises individualism as looser social ties and prioritisation of personal autonomy, and collectivism as tighter in-group bonds and role obligations; in our analyses we treat culture as a continuum but, for exposition, we refer to “more individualistic” versus “more collectivistic” environments using Hofstede's country scores as a transparent, theory-grounded operationalisation for testing whether the education–IADL association varies with prevailing social norms. Results shown in Table 5 (Hofstede et al., 2010).

A concise reading of Table 5 is that schooling offers a larger protective effect against IADL difficulties in the collectivistic group (Spain, Italy, France and the Czech Republic) but not in the individualistic group (Austria, Denmark and Belgium). A plausible mechanism is that more generous and universal services in many individualistic/welfare-heavy systems compress education-related health gradients – so extra schooling yields smaller marginal gains for functional independence – whereas in more family-oriented or navigation-intensive contexts, education better translates into health literacy, self-advocacy and access to appropriate care, producing stronger marginal effects. Cross-European evidence shows sizable cross-country variation in education-health inequalities shaped by institutions and welfare regimes, consistent with this pattern (Mackenbach et al., 2008).

To ensure the reliability and validity of the estimated effects of education on IADL difficulties, this section examines the robustness of these results by assessing their sensitivity to variations in the econometric specification. This approach strengthens the credibility of the results and helps rule out potential biases arising from omitted variables, measurement error or model misspecification.

Our mean results cover individuals who are aged 50–100. In this section, we restrict the sample to individuals aged 55–75. This narrower age range reduces heterogeneity in health and life circumstances, allowing for a more focused assessment of education's impact on IADL difficulties during late midlife and early old age. The results in Table 6 are not different from those described in the main regression.

We next consider the definition of education. In the main results, we use the years of education. Alternatively, we can use the education level as a measurement it which is based on ISCED, which classifies education into four categories. Individuals with no formal education or who did not complete the first level of education are classified as having no education (ISCED level 0). Those who completed primary or lower secondary education (ISCED levels 1–2) are grouped under low education. Respondents who attained upper secondary or post-secondary non-tertiary education (ISCED levels 3–4) fall into the medium education category. Finally, those who completed tertiary education (ISCED levels 5–6) are classified as having higher education. The results in Table 7 confirm that education continues to decrease IADL difficulties.

This study provides robust evidence that education plays a significant role in reducing difficulties in IADLs among older adults in Europe. The IV-Probit results confirm that higher educational attainment decreases the probability of experiencing functional limitations, underscoring education as a key determinant of healthy ageing. These findings are consistent with prior studies documenting the positive effects of education on health outcomes. For example, Cutler and Lleras-Muney (2012) argue that education improves health knowledge, shapes healthier behaviours, and enhances access to healthcare resources, while Brunello et al. (2016) show that education influences self-reported health through healthier lifestyles. Similarly, Fonseca et al. (2020), using SHARE data, find that additional schooling reduces the likelihood of poor health and functional limitations, though they treated IADLs as a secondary outcome. By focusing on IADLs as the primary health outcome, our study adds targeted evidence on functional independence in later life.

The heterogeneity analysis provides further nuance. The results suggest that education's protective effect is more pronounced for women than for men, which contrasts with earlier work highlighting stronger educational health returns for men due to cumulative advantages in income and occupational status (Delaruelle et al., 2018). This difference may reflect gendered patterns in health behaviours and social roles across Europe, or it may indicate that women benefit more from education when it comes to managing daily functional challenges. Our cultural analysis also shows important variation: education significantly reduces IADL difficulties in collectivistic societies but has an insignificant effect on individualistic ones. This partially aligns with Hofstede's (1984) cultural dimensions, which suggest that collectivistic societies place greater value on social and family support networks. Our results imply that education interacts with these cultural frameworks to shape ageing outcomes, adding to recent work by Zidkova et al. (2025), who find that cultural characteristics are strongly linked to population health differences across Europe.

Overall, this study confirms the established view that education improves later-life health outcomes but extends the literature by showing that its benefits are not uniform. The findings highlight that both gender and cultural context moderate the education–health relationship, pointing to the need for more nuanced, context-sensitive policies. In doing so, this research contributes to a growing literature that sees education not only as an economic investment but also as a crucial determinant of health and independence in older age.

This study examined the impact of education on IADLs among older adults in Europe, using data from the SHARE survey. The findings demonstrate that higher educational attainment is associated with fewer IADL difficulties, thereby supporting functional independence in later life. The heterogeneity analysis reveals that the protective effect of education is evident for men but not for women, and that it is significant in individualistic cultural contexts, while less so in collectivistic ones.

From a policy perspective, our findings point to education as a durable lever for healthy ageing. By strengthening functional independence in later life, schooling can curb demand for healthcare and long-term care, easing pressure on families and public budgets. In practical terms, widening access to high-quality education, especially for groups at greater risk of functional decline, offers a cost-effective way to mitigate the fiscal and social challenges associated with population ageing. Nonetheless, this study has several limitations. The use of compulsory schooling reforms as instruments may not capture all dimensions of education, such as quality or informal learning. Cultural classifications are simplified and may overlook within-country variation. In addition, while SHARE provides rich longitudinal data, the findings may not generalise beyond European contexts.

Future research could address these limitations by exploring alternative measures of education, such as lifelong learning, adult training or digital literacy, and by considering broader indicators of cultural and institutional variation. Comparative studies that extend beyond Europe would help assess whether these results hold in different socio-economic and cultural settings. Further investigation is also warranted into the mechanisms behind the gender differences identified here.

In summary, education emerges as an important determinant of functional independence in later life, though its benefits are uneven across gender and cultural contexts. Deeper exploration of these dynamics will be essential to inform both education and ageing-related policies in an era of demographic change.