Introduction
Over our life cycle, humans usually need care from other people as children, but also as adults, in case of temporary impairments or permanent health conditions requiring help to develop their daily lives. By care, we refer to any activity intended to preserve the health and ensure the wellbeing of individuals and provide them necessary support and protection. Total care work has become a matter of growing interest at economic, social and political levels, as it is crucial for both individual and societal wellbeing. Hence, it is essential to fully understand both the levels and the trends of care activities. However, most of them are provided outside of the markets, that is, unpaid (Miranda Reference Miranda2011), because they take place within families and without any financial compensation, which makes their analysis challenging. Nevertheless, measurement is crucial to understand their importance. For example, the European Commission and Ecorys (2021) estimated that unpaid long-term care exceeds the value of the formal long-term care provided in the EU. This difficulty of measurement has huge implications for gender equality, as women have historically been the primary caregivers, both in the formal care market and especially in unpaid care. This has reinforced the invisibilization of women’s contributions to the wellbeing of our societies (Ferrand et al. Reference Ferrand, Pesando and Nowacka2014; Eurofound 2025).
In modern societies, most unpaid care occurs within the same household (e.g. parents undertaking childcare). However, it can also be provided to people residing in a different household (henceforth interhousehold care), which includes significant intergenerational relationships such as grandparenting of small children and care provided by adult children to their older parents (living in their own homes). Due to current demographic and living arrangements developments (Liu and Esteve Reference Liu, Esteve, Schneider and Kreyenfeld2021; Sobotka and Berghammer Reference Sobotka, Berghammer, Schneider and Kreyenfeld2021), we can expect that interhousehold relationships will become more important in the future (Bengtson Reference Bengtson2001; Esteve et al. Reference Esteve, Pohl, Becca, Fang, Galeano, García-roman, Reher, Trias-Prats and Turu2024).
The main difficulty in analysing unpaid care work refers to data availability and adequacy. Being invisible in the market, data to identify and quantify care providers, care activities, and care intensity come from surveys on how people spend their time, which can be collected based on two main methods. The first one, considered the gold standard (Juster and Stafford Reference Juster and Stafford1991; Van den Berg and Spauwen Reference Van den Berg and Spauwen2006; ILO 2018), is time diaries, in which respondents are asked to keep a record of the time devoted to all the different activities they perform during a given period (typically a day). Many countries around the world undertake time-use surveys (TUS) at the national level based on time diaries. In particular, Eurostat launched a Harmonized European TUS (HETUS), which currently has two rounds available (2000 and 2010) and an ongoing third one (2020).
The second method, generally known as the recall method, gathers information on time use (including care) based on stylized questions or activity lists. In this case, respondents answer specific questions about the time they have devoted to a list of tasks in a reference period (such as the previous day or week). There are different macro-surveys (at the national or cross-national level) providing results on time devoted to non-market activities (including care provision) based on the recall method. At the European level, the most broadly used is the Survey of Health, Ageing and Retirement in Europe (SHARE).
The two methods of collecting data on unpaid activities have their advantages and disadvantages, which need to be carefully considered according to the specific purpose of the analysis. In general, information collected by surveys based on time diaries is considered more reliable because it can be very accurate in the time reported in each activity, but less detailed than that derived from a well-targeted set of stylized questions. This is because time diaries are much more expensive and difficult to undertake, and they cannot be so specific on the type of activities asked, as this would entail a substantial respondent burden, which could discourage participation or affect data reliability. At the empirical level, surveys based on time diaries have been widely used to measure care, although there are criticisms of how adult care is included (Miranda Reference Miranda2011; Francavilla et al. Reference Francavilla, Giannelli, Mangiavacchi, Piccoli, Bettio, Plantenga and Smith2013; Vargha et al. Reference Vargha, Gál and Crosby-Nagy2017). Different national surveys based on the recall method have also been broadly employed to analyse intergenerational transfers of care. At the European level, SHARE is the most used dataset to investigate care transfers (Attias-Donfut et al. Reference Attias-Donfut, Ogg and Wolff2005; Deindl and Brandt Reference Deindl and Brandt2011; Albertini and Kohli Reference Albertini and Kohli2013; Alessie et al. Reference Alessie, Angelini and Pasini2014, among many others).
In this article, we aim to contribute to the current knowledge of the measurement of care activities by analysing unpaid care provided to people living in a different household (interhousehold care) in Europe, regardless of the recipient’s age. Specifically, we compare data on unpaid interhousehold care extracted from the two main databases available at the European level: wave 2 of HETUS (around 2010), based on time diaries, and wave 2 of SHARE, from 2007 (which collects information with stylized questions). We focus on interhousehold care because it mainly consists of two intergenerational exchanges (grandparental childcare and support to adult individuals) which are not necessarily provided daily and can imply more problems of accuracy if time diary surveys are used. Moreover, care provided between non-household members might be more objectively collected, as care done within the household is more sensitive to subjective identifications when there is multi-tasking (De Koker Reference De Koker2008). Finally, we use a broad definition of care to allow comparison across the two surveys. By understanding care as any form of help or support provided to others (including personal care and practical help in the house), we ensure conceptual consistency and comparability between the datasets.
Data availability allows comparing seven European countries present for both HETUS and SHARE (Austria, Belgium, France, Greece, Italy, Poland and Spain), with diverse welfare state models, as well as different socio-economic and demographic characteristics. This will let us examine whether divergences between the two survey methods are consistent across different care patterns, or whether they depend on the intensity and distribution of care within each country.
As SHARE restricts data collection to individuals aged 50+, we also limit our comparative analysis to this age group. Still, we will capture both grandparenting – most grandparents in Europe are older than 50 – and adult children providing care to their older parents. Descriptive and multivariate analyses are conducted to identify and assess the differences between the estimations of childcare and adult care resulting from both surveys. This comparison brings valuable insights into methodological issues to quantify interhousehold transfers of care, as well as important features of care patterns in different countries, offering researchers and policymakers a deeper understanding about how interhousehold care is organized around Europe. Moreover, as exposed by Emery and Mudrazija (Reference Emery and Mudrazija2015), research on intergenerational transfers in Europe has become too dependent on SHARE, in the sense that most studies are based on this dataset. This could compromise the findings, as the results are sensitive to specific features of this survey, such as changes in wording, the reference period, and the categorization of the respondent as caregiver or care recipient. In a previous comparison of SHARE with other European surveys using stylized questions, Tur-Sinai et al. (Reference Tur-Sinai, Teti, Rommel, Hlebec and Lamura2020) already exposed significant differences in the identification of caregivers within countries.
The rest of the work is organized as follows. The next section provides a brief overview of intergenerational care in Europe. Then, the different types of surveys and methods to collect data on unpaid activities are reviewed. Later, the two macro surveys (HETUS and SHARE) to be compared in this study are presented, together with the methodology employed for the comparative analysis, followed by the results and a final section discussing the main findings.
Intergenerational care in Europe
According to a recent study from Eurofound (2025), on average, 48.1 per cent of women and 42.3 per cent of men living in Europe declare that they provide unpaid care. These percentages vary across age groups, displaying a clear inverted U-shape: they are the lowest for the youngest and oldest and peak at ages 25–44, coinciding with the typical parenting period. Significant differences are also observed at the country level: while in Bulgaria, Greece, Cyprus and Romania, more than 60 per cent of people aged 16 and over provide informal care, in Luxembourg, France, and Hungary the share is below 30 per cent. Regarding types of care, Eurofound (2025) distinguishes childcare of own children, childcare of other children, and long-term care, with childcare of own children registering the highest intensity, followed by long-term care. However, the difference between the hours devoted to the childcare of own children and long-term care is already small in several countries (Denmark, Portugal, Estonia, Lithuania and Italy), and it does not even exist in France. In the current context of pronounced population ageing, with progressively higher life expectancy and persistently low fertility rates, long-term care could easily become the main category of unpaid care work in the coming decades.
The relationship between care and gender is deeply intertwined. Women disproportionately perform unpaid care work, in both number and intensity (on average, six more hours per week than men), and are more likely to have multiple care-giving responsibilities, simultaneously managing both childcare and long-term care duties (Eurofound 2025). This creates inequalities and impacts the economic and social opportunities of female caregivers (EIGE 2023). Although there is a convergence observed in gender disparities devoted to care over time, these trends have changed more slowly in recent years among European countries (Charmes Reference Charmes2022).
Unpaid adult care can take place for members living in the same household or not, for both children and adults. In this article, however, we focus solely on care provided to members living in different households, also known as interhousehold care. This care, support or help can be provided to relatives, friends or neighbours living elsewhere – most commonly older parents – who require assistance with personal care, household tasks or administrative matters (Daatland and Herlofson Reference Daatland and Herlofson2003; Haberkern and Szydlik Reference Haberkern and Szydlik2010), and in the case of children, it mostly refers to grandparental childcare.
In the literature, grandparental childcare and parental care from adult children have usually been assessed separately. Regarding grandparental childcare, ILO (2018) estimates for 30 high-income countries report that 25 per cent of children aged up to two years and 27 per cent of those aged three to five are cared for by non-parent unpaid caregivers (mostly grandparents) on a weekly basis. In Europe, almost half of grandparents care for their grandchildren, whereas around 25 per cent do it regularly (Zanasi et al. Reference Zanasi, Arpino, Bordone and Hank2023) and 11 per cent do it daily (Glaser et al. Reference Glaser, Price, Ribe Montserrat, Di Gessa and Tinker2013). No remarkable changes in such patterns have been observed over the last decade (Zanasi et al. Reference Zanasi, Arpino, Bordone and Hank2023), although significant differences arise in cross-country comparisons. In Northern European countries, where childcare is publicly and extensively organized, grandparenting is more common but less intensive (Glaser et al. Reference Glaser, Price, Ribe Montserrat, Di Gessa and Tinker2013; Zanasi et al. Reference Zanasi, Arpino, Bordone and Hank2023). On the contrary, in Southern and some Eastern European countries, there is an implicit social assumption that grandparents provide childcare for their grandchildren (Glaser et al. Reference Glaser, Price, Ribe Montserrat, Di Gessa and Tinker2013; Bordone et al. Reference Bordone, Arpino and Aassve2017) to fill the gap between childcare needs and the lack of adequate family policies (sufficient part-time jobs, formal childcare services and in-kind family benefits), and this can be much more intense.
Researchers have also analysed the role of adult children caring for their older parents, and while intrahousehold care has traditionally received greater policy and research attention due to its intensity and proximity, interhousehold care also represents a substantial and often underestimated share of unpaid care work. This form of support is particularly salient in European contexts where older generations tend to live independently but rely on adult children for regular help (Kaschowitz and Brandt Reference Kaschowitz and Brandt2017). It is worth noting that Europe has the highest share of older people living alone, although with remarkable cross-country differences (Jacobs et al. Reference Jacobs, van Groenou, Aartsen and Deeg2018; De Klerk et al. Reference De Klerk, De Boer and Plaisier2021; Cohen et al. Reference Cohen, Shiovitz-Ezra and Erlich2023). Moreover, gendered patterns persist: women are more likely than men to provide any type of care (intra- or interhousehold), but men’s participation is relatively higher in interhousehold arrangements, where tasks are often more flexible or less time-intensive (Haberkern et al. Reference Haberkern, Schmid and Szydlik2015). Similar to grandparental childcare, in Northern and Western European countries, exchanges of care and financial help between adult children and their parents not living in the same household are more common but less intense than in the South. According to Mudrazija et al. (Reference Mudrazija, Angel, Cipin and Smolic2020), such differences are strongly linked with the robustness of each country’s welfare state, and to co-residence patterns of adult children and their older parents, which is much higher in Southern European countries due to their weaker welfare states (Isengard and Szydlik Reference Isengard and Szydlik2012; Isengard et al. Reference Isengard, König and Szydlik2017). Previously, Deindl and Brandt (Reference Deindl and Brandt2011) also identified a link between the welfare regime and the intergenerational transfers between middle-aged children and their parents across Europe.
Overall, there is strong evidence of the importance of intergenerational transfers of care in Europe, including those involving members of different households, such as grandparental childcare and help to older individuals, but with major regional differences. Nevertheless, as in the case of other productive activities that take place outside the market, the main difficulty for a thorough analysis of this unpaid care work is identifying and quantifying it. The undeniable contribution of both types of care to individual and societal wellbeing (Barnay and Juin Reference Barnay and Juin2016; Pulgaron et al. Reference Pulgaron, Marchante, Agosto, Lebron and Delamater2016; Bom et al. Reference Bom, Bakx, Schut and Van Doorslaer2019; Coe et al. Reference Coe, Guo, Konetzka and Van Houtven2019; Rutigliano Reference Rutigliano2020; Danielsbacka et al. Reference Danielsbacka, Křenková and Tanskanen2022) justifies the need for accurate measures of such activities, which is the main goal of this study. The main issues related to such measurement are reviewed in the following section.
Issues in measuring unpaid care
To obtain an accurate care measurement, some preliminary considerations need to be made. First, it is necessary to distinguish between formal care, that is, paid care activities, and informal or unpaid care, which is help provided outside of the markets, mostly on the grounds of kinship but also involving non-kin acquaintances such as neighbours and friends (Cruz et al. Reference Cruz, Soeiro, Canha and Perrotta2023). Second, care involves both direct help, such as feeding a baby or nursing a dependent adult, and indirect care tasks, such as cooking, cleaning or shopping for those who cannot do it by themselves (ILO 2018). Third, childcare and adult care (especially long-term care) are very different in their specific needs, arrangements, and impact on both caregivers and the whole society (Bittman et al. Reference Bittman, Fisher, Hill and Thomson2005). Probably, this explains why childcare and adult care have been generally approached separately in the literature. In fact, usually, the term informal care has been specifically identified with help provided to people with disabilities and frailties, excluding childcare (OECD 2017; Zigante Reference Zigante2018; Li and Song Reference Li, Song, Gu and Dupre2019). In this article, following the recommendations by ILO (2018), we use the term unpaid care work to refer to any kind of activity (direct and indirect) addressed to provide help to a person in need (children or adults), when such activities are performed outside the market (without economic compensation).
As unpaid care is not provided through the markets, there are no observed indicators of its amount and its economic value. Quantifying unpaid care implies identifying both the number of care providers and the amount of time devoted (i.e. the intensity), which requires using surveys collecting data on people’s time use. There are two main approaches to data collection of time spent in different activities. First, those based on time diaries, where respondents record all the activities undertaken over a given period (typically a day or a week). Second, those based on the recall method, implemented through stylized questions or activity lists, which ask respondents to recall the amount of time they have allocated to specific activities over a given past period. As pointed out by Flemming and Spellerberg (Reference Flemming and Spellerberg1999), according to INSTRAW (1995), data collection based on the recall method tends to be better accepted by respondents (less difficulty in responses and lower burden), and it is not too expensive to put into the field. On the contrary, data reliability is generally lower because respondents ‘estimate’ how much time they spent in different tasks (Juster and Stafford Reference Juster and Stafford1991). On the other side, time diaries tend to be burdensome for respondents and very expensive to undertake. Moreover, they are not error-free because the specific period selected for the diary can, by chance, not be representative of the usual activity. This is especially important in the case of interhousehold transfers of care as they are not necessarily provided daily. However, time diaries are generally considered the gold standard in time-use data collection (Juster and Stafford Reference Juster and Stafford1991; Van den Berg and Spauwen Reference Van den Berg and Spauwen2006; Kan and Pudney Reference Kan and Pudney2008). In fact, this has been the method generally adopted by the national statistical offices of many countries to produce TUS. Nevertheless, there are several methodological decisions (such as the number and specific days of the week covered, the number of activities to record, the time interval for recording, and the type of interview, among others) that can compromise cross-country comparability. At the European level, an extra effort has been made by Eurostat to homogenize the TUS methodology, crystallized in HETUS.
In parallel, other macro-surveys (representative at a national level) containing information on unpaid work but based on stylized questions have also been carried out. Examples are the European Quality of Life Survey (EQLS), the European Health Interview Survey (EHIS) and the Survey on Health, Ageing and Retirement in Europe (SHARE) for Europe; the English Longitudinal Study of Ageing (ELSA) for the UK; the Health and Retirement Study (HRS) for the US; or the Generations and Gender Survey (GGS) for 25 countries across four continents.
Different studies have found that estimates of time spent on unpaid work derived from surveys based on stylized questions are higher than those obtained from surveys based on time diaries (Juster et al. Reference Juster, Ono and Stafford2003 for the US; Baxter and Bittman Reference Baxter and Bittman1995; Bittman et al. Reference Bittman, Fisher, Hill and Thomson2005 for Australia; Niemi Reference Niemi1993 for Finland). According to Bittman et al. (Reference Bittman, Fisher, Hill and Thomson2005), who compared Australian TUS with the National Survey of Disability, Ageing and Care for the same country, time diaries produce more robust data, but after adjustments to properly identify caregivers and the time they effectively spend on care activities. Tur-Sinai et al. (Reference Tur-Sinai, Teti, Rommel, Hlebec and Lamura2020) also pointed out the need to correctly identify the number of caregivers in surveys based on stylized questions. They compared the estimations obtained from three European macro-surveys, all of them based on the recall method (EHIS, SHARE and EQLS), finding significant differences for some specific countries. For example, in Spain, estimates from EQLS produced twice as many caregivers as SHARE, while the opposite happened in Germany and Denmark.
Rutherford and Bu (Reference Rutherford and Bu2018), after comparing caregivers’ responses to different care questions in ELSA for the UK, concluded that the most common stylized questions tend to underestimate both the scale and the scope of unpaid care provided. In another comparative work devoted to housework, Bonke (Reference Bonke2005) found that respondents in a Danish 2001 TUS reported more housework in diaries than in interviews. In other small-scale studies designed to capture particular care needs, differences in time devoted to care from both types of data collection have also been addressed, but showing that the recall method reports higher estimates (Van den Berg and Spauwen Reference Van den Berg and Spauwen2006). Hence, no systematic deviations from estimations from surveys based on time diaries and on recall questions are observed.
Moreover, the specific method of data collection is not the only challenge of measuring time devoted to unpaid work, as pointed out by Urwin et al. (Reference Urwin, Lau, Grande and Sutton2021) in a systematic literature review. Properly defining caregiving time, how to deal with intangible tasks (such as social and emotional care) and joint production (performing multiple tasks at the same time) have also been identified as important difficulties to face, without an optimal solution in the empirical field. For the specific case of childcare, after revising results from TUS in the US, Canada, Australia and the UK, Folbre and Yoon (Reference Folbre and Yoon2007) found inconsistencies between the definitions of primary and secondary childcare activities and underscored the need to define childcare more carefully in surveys’ design. Overall, data based on time diaries are generally preferred to those obtained from questionnaires based on the recall method, but this does not imply that time diaries are completely free from bias.
Data and methods
In this study, the HETUS and the SHARE surveys are being compared to assess differences in the measurement of interhousehold transfers of care provided by people over 50 years old. HETUS is a dataset compiled by Eurostat based on national time-use surveys conducted by the respective statistical offices, following Eurostat’s methodological guidelines and hence, ensuring full cross-country comparability of the data. The surveys cover all members of the sampled households, aged ten and older. We use the second and last wave available, which includes comparable estimates for 17 European countries, with samples collected between 2008 and 2015 (Eurostat 2021).
SHARE is a European cross-national panel database collected biannually since 2004. It is specifically addressed to individuals aged 50 and older. It collects comprehensive data on various aspects of individuals’ lives covering topics such as health, income, employment, and family dynamics, including intergenerational transfers of money and care, which can involve members of the same household (intrahousehold) or different households (interhousehold). Given the aim of our analysis, we use wave 2 (referred to 2007) because it contains the most detailed questions on the help received from and given to other people, including those not living in the same household (SHARE-ERIC 2024). Subsequent waves of SHARE have not included detailed questions on the time devoted to care that would allow the specific measurement of caregiving intensity in hours per day.
The comparison of both surveys restricts the analysis to individuals aged 50 years and older from HETUS, and to the seven countries with information on interhousehold transfers of care in the two surveys: Austria, Spain, Belgium, France, Greece, Italy and Poland. The concrete years of the HETUS surveys used in our analysis are as follows: Austria (2008–2009), Belgium (2012–2013), France (2009–2010), Greece (2013–2014), Italy (2008–2009), Poland (2012–2013), and Spain (2009–2010). This means that the time gap between the data from the two surveys for these seven countries ranges from a minimum of two years (Austria and Italy) to a maximum of seven years (Greece). Given empirical evidence that time-use patterns tend to evolve slowly (Sullivan et al. Reference Sullivan, Gershuny and Robinson2018; Najmi et al. Reference Najmi, Bostanara, Tabasi, Khan, Auld, Pinjari and Rashidi2025), this interval is unlikely to introduce significant bias in our analysis.
Table 1 shows the size and main characteristics of the final sample used. Being a cross-sectional survey, HETUS includes more interviews than in the case of SHARE, which implies that estimates can be more reliable (with smaller confidence intervals). Regarding the socio-economic characteristics, both samples are very similar in age, sex, and employment status distributions. Small differences are observed for partnership status, while differences are larger for educational level distributions. This could be partially explained by the fact that HETUS samples for Italy and Poland are much higher than for the other five countries, and people aged 50+ in these two countries might have a lower educational level than in the other countries. However, we are more interested in comparing surveys country by country. In order to reduce sample composition effects in the comparison of surveys, we will also implement a multivariate analysis controlling by age, sex, educational level, labour and partnership status which is described in the analytical strategy section.
Sample composition
Note: Employment status is dichotomized between Employed (currently working) and Not employed (all the rest, including unemployed, inactive, and retired).
As abovementioned, the two surveys we compare employ different methodologies for collecting time-use information. HETUS uses time diary data where respondents document their activities in 10-minute intervals over a 24-hours period, with Austria utilizing 15-minute intervals instead. Diaries also collect information about simultaneous activities, although this information is underreported, and, for that reason, we decided to use only the main activity reported. Conversely, SHARE collects information using stylized questions where respondents report the duration of time spent in specific activities in a previous period. Table 2 provides an overview of the primary distinctions between the questioning approaches and the nature of the data obtained from each survey. The detailed wording of the SHARE questions can be found in the Appendix.
Methodology and questions in HETUS and SHARE
Note: 15 minutes in the specific case of Austria
Using microdata from the two surveys, we estimate the time allocated to provide any kind of help to people living in a different household than the provider. We distinguish between help addressed to adults (adult care) and to children (childcare), which is mostly provided by grandparents, and the cumulative time spent on both types (total help), which constitute our three dependent variables. The period of reference of the time reported is minutes per day, as captured in the time use diary. In the case of SHARE, the questions of frequency (daily/weekly/monthly or less) and hours of a typical day/week/month/year are combined to estimate an average number of minutes per day, comparable to estimations from HETUS. For example, if someone answered caring 10 hours per week, this would be computed as 600 minutes divided into seven days, resulting in 85.7 minutes per day. We additionally adjusted the total amount of care provided to a maximum of 16 hours per day. This implies that for all individuals reporting more than 16 hours of care per day, the daily total was truncated to 16 hours. In the survey, individuals are asked a frequency question for up to four people they can take care of. To avoid simultaneous caring activities, we restricted the analysis to only one adult and one grandchild they can take care of. Nevertheless, in some cases, individuals provide a disproportionate amount of daily hours of caring (sometimes more than 24 hours). From the total sample of 17,800 individuals, 7,526 reported providing care to people living in a different household (41 per cent). Of these, 53 individuals reported more than 16 hours of adult care per day, and 83 more than 16 hours of childcare. In all these cases, we adjusted adult care and childcare to meet the maximum of 16 hours per day. When adding adult care and childcare, there were still three individuals who needed an additional adjustment, as the sum of both was more than 16 hours in total. In these cases, adult care and childcare were proportionally adjusted to meet the total amount of 16 hours.
We generated three different measures for each dependent variable (childcare, adult care and total care time). The first computes the average time devoted to these activities across the studied population (total), encompassing both individuals who reported engaging in the activity and those who did not report any involvement, following previous works (Vargha et al. Reference Vargha, Gál and Crosby-Nagy2017). The second consists of a participation rate in interhousehold care, computed as the proportion of individuals who dedicated any time to this kind of activity (proportion of caregivers). Finally, the third indicator computes the average time spent in interhousehold care exclusively by those who allocated any time to these tasks (intensity of care). All three measures are estimated separately for each survey, and results are presented by country, gender, and age group, using the weights provided in each dataset. Descriptive statistics for each variable are shown in Table A1 in the Appendix.
Analytical strategy
Initially, we present descriptive findings to explore general patterns of help provided to non-household members and to evaluate disparities between the two surveys. As previously indicated, all figures exhibit estimations categorized by age and country. We first outline the overarching trends in total help and then we analyse these trends by type of help and gender of the caregiver across the three proposed indicators.
Next, we run multivariate models for each country and gender, separately, to assess the variation between the estimations from the two surveys concerning the three dependent variables (childcare, adult care and total care time) and measures (total time, proportion of caregivers and intensity of care). We conduct ordinary least squares (OLS) regression models for the average time of care given by the studied population, logistic regressions for the proportion of caregivers, and a final OLS model with the time provided by caregivers (sample restricted only to caregivers). Across all models, we include controls for the respondent’s age (in ten-year intervals until age 80 and above), level of education (less than primary completed, primary completed, secondary non-tertiary completed, tertiary), employment status (employed or not) and partnership status (living in a couple or not), in order to minimize the compositional effects on our outcome variables due to the variation of these covariates.
Results
Descriptive results
The first comparison obtained from the two surveys refers to the average amount of total time devoted to providing care to non-household members (children and adults). We can see in Figure 1 that SHARE estimates are higher than HETUS for all the countries observed. Nevertheless, the pattern by age remains very similar across countries in both surveys, showing a peak at ages 60 to 69, which decreases afterwards. The exceptions are Austria in the case of SHARE estimates, and Belgium with HETUS values, where the peak is at ages 50 to 59. Poland, Italy, Greece, and Belgium are the countries reporting the highest amount of average minutes per day of interhousehold transfers of care in SHARE, while Poland, Spain, and Italy report the highest values in HETUS. This means that even if surveys differ greatly, in the case of Poland and Italy, for example, they continue to be above the other countries regarding the amount of care reported, independently of the collection method used. This is not the case of Spain, though, as its ranking in each survey is very different. The amount of care provided decreases with age for all countries and for both surveys. Among older adults, Italy shows the highest levels of care provision in SHARE, while Spain does so in HETUS.
Interhousehold care (total average for children and adults) provided by country in SHARE and HETUS.
Figure 2 depicts results differentiating the total average time devoted to care by the age of the recipient (adult care and childcare). The largest differences between surveys are observed for time provided to childcare (which we assume is mainly provided by grandparents in HETUS). In fact, estimations from SHARE reveal that the average time provided to childcare is higher than that devoted to care for adults. However, in HETUS, this difference between types of care is not so clear. In all countries, childcare differences between surveys are higher for age groups 60–69 and 70–79, and for women caregivers. Nevertheless, in the case of adult care, differences between surveys are small for some countries such as Poland and Spain, and almost negligible at older ages (70+) in practically all countries. It is also interesting to note that gender differences are clearer in SHARE than in HETUS, and that these are larger at younger ages, when women typically devote more time to care for both children and adults. By contrast, gender differences decrease or even reverse at older ages according to data from both surveys.
Interhousehold care (total average) provided for children and adults, caregiver’s gender, and country in SHARE and HETUS.
To further investigate the nature of the observed survey differences in the average time devoted to interhousehold care, we look into the proportion of identified caregivers, whom we also call participants, as well as the care intensity (the amount of time that each of them devotes to care activities). Together, they allow us to understand whether the observed differences in the average time provided to interhousehold care between HETUS and SHARE are due to capturing a different number of caregivers, a different average amount of care time (due to the use of a different method to report it), or both.
Figure 3 presents the proportion of caregivers resulting from HETUS and SHARE. As observed, SHARE shows higher estimates compared to HETUS in all countries, types of care, and genders. Nevertheless, the size of such differences varies by country, type of care, gender, and age. Spain shows fewer discrepancies between the two surveys. Also, differences in the 50–59 and the 80+ age groups providing childcare show, in general, the smallest discrepancies between surveys. In the case of adult care, differences between surveys decrease with age in all countries. It is interesting to note that in most countries, gender gaps in care provision, where women provide more time care, are higher at younger ages and larger in the case of childcare, a pattern that can be observed in both surveys.
Proportion of interhousehold care-givers, for children and adults, care-giver’s gender and country in SHARE and HETUS.
Figure 4 focuses on comparing the care intensity resulting from SHARE and HETUS, computed as the average number of minutes provided by each caregiver. Profiles show blanks when all the respondents of a given age and gender report zero minutes of interhousehold care activities, which only occurs in Austria and Belgium in the case of HETUS, and for Poland in SHARE. Here, we can see that the absolute differences in care intensity between both surveys are lower than those previously observed in the number of caregivers in Figure 3, and the levels of caregiving are much alike between countries for both surveys. Moreover, a few estimates from HETUS data result in higher values than from SHARE, especially in adult care provided by women. Regarding gender differences in care intensity, they are small in the case of childcare for some countries like France, Belgium, and Spain, especially in HETUS. In the case of adult care, it is important to note the reversing pattern of gender differences observed in estimates from HETUS for most countries, where men report providing more care than women, while this does not happen in SHARE. This is probably because when men provide adult care, they usually engage more intensively and report a remarkable amount of time.
Interhousehold care provided by each caregiver (intensity of care), for children and adults, caregiver’s gender and country, in SHARE and HETUS.
Overall, Figures 3 and 4 reveal that the observed differences in the results obtained from HETUS and SHARE for the total average amount of interhousehold care (Figures 1 and 2) mainly come from a difference in the proportion of caregivers captured by each survey. That is, SHARE identifies a higher number of individuals providing any kind of care to non-household individuals.
Multivariate analysis
To get a more precise estimation of the gap between the estimations of interhousehold care obtained from HETUS and SHARE, and control by compositional differences between the two surveys, we run regression models for the three types of care (childcare, adult care and total care), for each country and gender, controlling for several socio-demographic characteristics (age, level of education, employment status and partnership status). Table 3 reports the results of models of the minutes of interhousehold care provided for all respondents (total), Table 4 displays the estimated probability of being or not a caregiver, and Table 5 shows the estimates of the time provided only by caregivers (intensity of care). In the three tables, we report the coefficient for each survey in the models, which can be interpreted as the difference in the estimates from SHARE compared to HETUS. Full models with all control variables can be consulted upon request. In Table 4, where the model is a logistic regression, we report the average marginal effects (AME) of the coefficients in order to make them comparable across countries.
Coefficients of survey SHARE (reference HETUS) from the OLS in minutes per day devoted to interhousehold care among the overall population (total time)
Source: Own calculations from HETUS-2010 and SHARE-2007.
*** p < 0.001, **p < 0.01, *p < 0.05.
Models control by age, educational attainment, employment status and partnership status.
Average marginal effects (AME) of SHARE (reference HETUS) from the logistic models of the probability of being a care-giver of interhousehold care
Source: Own calculations from HETUS-2010 and SHARE-2007.
*** p < 0.001, **p < 0.01, *p < 0.05.
Models control by age, educational attainment, employment status and partner status.
Country acronyms refer to AT-Austria, BE-Belgium, ES-Spain, FR-France, GR-Greece, IT-Italy and PL-Poland.
Coefficients of SHARE (ref HETUS) from the OLS models for time in minutes per day devoted to interhousehold care among care-givers (intensity of care)
Source: Own calculations from HETUS-2010 and SHARE-2007.
*** p < 0.001, **p < 0.01, *p < 0.05.
Models control by age, educational attainment, employment status and partner status.
Country acronyms refer to AT-Austria, BE-Belgium, ES-Spain, GR-Greece, IT-Italy and PL-Poland.
Results from Table 3 confirm that SHARE consistently captures higher averages of care time provided than HETUS in all countries, genders and care types, with the only exception of care provided to adults by men in Spain. Nevertheless, the models confirm strong differences between countries, age of recipient and gender, which are larger for childcare and women and smaller for adult care done by men. Spain shows the lowest difference between surveys, for both men and women, while Belgium displays the highest.
Table 4 shows that the probability of being a caregiver of non-household members is significantly higher in SHARE with respect to HETUS, as all AMEs are positive for all countries, for any type of care, and for both genders. As in the previous model, Belgium shows the largest differences and Spain the lowest, for both men and women. In this case, though, differences between surveys are more similar for both types of care.
When the comparative analysis is restricted to care time provided only by caregivers, Table 5 shows that differences between surveys are not always significant and can be positive or negative depending on the country, gender, and type of care. Differences are significant and negative for adult care provided by men in all countries, meaning that SHARE reports less time for adult care than HETUS. In the case of adult care provided by women, only Austria and Italy show a significant coefficient, being negative in Austria while positive in Italy. For childcare, estimates are negative and significant only for women in Belgium and Austria. However, as shown in Figure 4, some age groups in HETUS in these two countries were blank, so comparisons should be made with caution. From the rest of the coefficients of childcare provided by caregiving women, we observe that estimates for SHARE are higher than for HETUS. In the case of male caregivers, only Spain, Greece and Italy show a significant positive coefficient. Moreover, we can see that survey differences, when significant, are still important in levels. For example, female caregivers in Poland report almost 50 minutes a day more of childcare in SHARE than in HETUS, and male caregivers in France report 2 hours a day more of adult care in HETUS than in SHARE.
Discussion
Survey differences
In this study, we compared the reported time to support or take care of individuals living in a different household (interhousehold transfers of care) from two surveys, HETUS and SHARE, in seven European countries. Each of these surveys uses different methods to collect information about the time devoted to such activities. While HETUS is based on a diary and captures all the activities performed during a single day, SHARE uses stylized questions that refer to the last 12 months. The results obtained from the two datasets differ greatly, but our goal was to understand the level of divergence and whether it varies by country, gender, and type of care (adult care or childcare). Additionally, we explored the implications of using one dataset over the other in the analysis of interhousehold care. The results for the average daily time provided for interhousehold care reveal significant disparities between surveys. For instance, according to SHARE, men aged 50 and older report providing between 13.9 and 42.7 minutes per day of care to non-household individuals more than in HETUS, while for women, the difference between surveys ranges from 29.7 to 69.3 minutes, among the different countries under study.
The multivariate analysis, controlling by educational level, employment and partnership status, confirms these differences by country, gender and type of care. The largest divergences are found in childcare, particularly when performed by women, whereas the smallest differences between surveys are reported for adult care, with even smaller discrepancies when done by men. Regarding childcare provided outside the household by middle-aged and older care-givers, we can assume that most of it refers to grandparental care, and our results confirm that there is a significant proportion of older individuals in Europe engaged in this activity (from 24 per cent to 60 per cent), but with a great variability in the intensity of care across countries (Zanasi et al. Reference Zanasi, Arpino, Bordone and Hank2023). Concerning care provided by men to adult members not living in the same household, this likely refers to a specific group that, when captured, is engaged rather intensively. This is particularly evident in countries with limited public support for care, such as Spain, Italy and Poland (Spasova et al. Reference Spasova, Baeten and Vanhercke2018; Barczyk and Kredler Reference Barczyk and Kredler2019). The willingness to provide care in these countries might be more reluctant, as individuals know that once they commit, it will likely involve more demanding tasks. At the same time, in countries where a higher public provision of care is widely available, more people agree to participate in these activities as they will be more sporadic and will imply a lighter burden (Brandt and Deindl Reference Brandt and Deindl2013; Verbakel Reference Verbakel2018). Hence, when there is a higher proportion of caregivers and an important share of them provide care less intensively. This is particularly evident in the case of grandparental care, where we observe larger differences between surveys in total per capita estimates. On the contrary, when a smaller proportion of caregivers is involved but with a higher intensity, as in adult care, especially if done by men (Labbas and Stanfors Reference Labbas and Stanfors2023), we observe smaller differences between the two surveys.
These variations can be due to two components explored in the results: the proportion of caregivers captured in each survey and the actual time reported by caregivers. The separate analysis of both indicators reveals that the two generally move in opposite directions. On the one hand, in the models of the proportion of caregivers, SHARE consistently shows higher estimates than HETUS, although the larger discrepancies observed in per capita values of childcare are reduced. This indicator suggests that HETUS would be missing, more than SHARE, sporadic caregivers. On the other hand, in the models of care intensity, HETUS reports higher estimates than SHARE in some countries and care activities, as in adult care provided by men. This suggests that once individuals in HETUS declare caregiving activities for non-household individuals, they may report greater levels than in SHARE. In the case of adult care, this outcome may be explained by the fact that when someone reports caregiving to non-household individuals in HETUS, it is more likely to refer to an intensive caregiver, and their time devoted to care would be more accurately measured through time diaries, which tend to be more reliable in capturing routine activities (Juster and Stafford Reference Juster and Stafford1991; Van den Berg and Spauwen Reference Van den Berg and Spauwen2006; Kan and Pudney Reference Kan and Pudney2008).
For men providing childcare and women providing adult care, the differences in care intensity between the two surveys are not significant in many countries. This indicates that while SHARE captures more caregivers than HETUS, once caregivers are captured, as these would be a mixture of intense and non-intense caregivers, both surveys show similar estimates of caregivers’ activity. However, this is not the case for grandparental childcare provided by men and women in Spain, Greece, and Italy, and by women in France and Poland, where SHARE still reports higher levels of care intensity. In these countries, this could indicate that the care provided is often performed simultaneously with other activities, lowering the reported time in HETUS. This is in line with previous comparisons done by Bittman et al. (Reference Bittman, Fisher, Hill and Thomson2005) with two Australian surveys where they stated that time-use surveys underreported the overall time spent on care provided, especially when caregiving is done simultaneously with other activities, as it is often the case of childcare.
Hence, our results underscore the significant impact that the choice of survey has on the measurement of interhousehold transfers of care among older individuals. While HETUS is considered a gold standard for capturing routine activities (Juster and Stafford Reference Juster and Stafford1991; Van den Berg and Spauwen Reference Van den Berg and Spauwen2006; ILO 2018), caution is needed when addressing occasional activities, such as interhousehold care. Our findings suggest that HETUS may underreport the number of people engaged in caregiving activities of childcare and adult care done more sporadically, as well as care intensity of care done simultaneously with other activities, resulting in biased averages for per capita values of the entire population. This has implications for research conducted at the aggregate level based on time-use surveys such as the age profiles estimated in National Time Transfer Accounts (NTTA) (Kluge Reference Kluge2014; Gál et al. Reference Gál, Szabó and Vargha2015; Zannella Reference Zannella2015; Rentería et al. Reference Rentería, Scandurra, Souto and Patxot2016; Sambt et al. Reference Sambt, Donehower and Verbic2016; Vargha et al. Reference Vargha, Gál and Crosby-Nagy2017, Reference Vargha, Binder-Hammer and Donehower2023; Hammer et al. Reference Hammer, Spitzer, Vargha and Istenič2020), as well as other studies estimating care work (Miranda Reference Miranda2011; Charmes Reference Charmes2019). In these cases, although interhousehold care represents only a small component of total unpaid care, its underestimation can undervalue the actual role of older individuals as caregivers.
Cross-country differences
Furthermore, the comparison of the two surveys has unveiled differences between countries and genders regarding participation and intensity of adult care and childcare performed outside the household, which can be used to refine conclusions derived from cross-country comparisons using SHARE. In fact, SHARE is a widely used survey for measuring and analysing care in Europe, but it has seldom been contrasted with outcomes from other surveys. To our knowledge, this has only been done for monetary transfers (Emery and Mudrazija Reference Emery and Mudrazija2015) and the number of adult caregivers (Tur-Sinai et al. Reference Tur-Sinai, Teti, Rommel, Hlebec and Lamura2020). However, the differences in intensity values among caregivers when comparing SHARE and HETUS suggest that both adult care provided by men and grandparental care provided by women should be interpreted with caution in certain countries. Specifically, in the case of men caregivers for other adults, SHARE might be underestimating the time these caregivers devote to these activities across all countries, but particularly in Austria, Belgium, and France, where men report similar care intensity levels as women using HETUS. For childcare performed by women, SHARE might be underreporting care intensity in Austria and Belgium, while overestimating the time provided in the rest of the countries. However, this discrepancy may also indicate that childcare (or grandparental childcare) in Spain, France, Greece, Italy, and Poland is being done more frequently as a simultaneous task with other activities, causing a less accurate report in HETUS.
Limitations
The limitations of our study are various, starting with the fact that we could not use the exact same year for both surveys. Nevertheless, we do not expect drastic changes in care patterns over small periods of time (Sullivan et al. Reference Sullivan, Gershuny and Robinson2018; Najmi et al. Reference Najmi, Bostanara, Tabasi, Khan, Auld, Pinjari and Rashidi2025). Additionally, there are small inconsistencies in the definition of care activities between the two surveys. For example, in SHARE, childcare refers uniquely to grandchild care, whereas in HETUS, childcare is defined as care provided to any child not living in the household, which includes other children who are not their grandchildren. This could lead to higher estimates of caregivers reported in HETUS than in SHARE, but as the contrary is observed, we expect minimal bias produced by this different specification. Additionally, SHARE reports information on care provided to multiple recipients, and we cannot know if this care is delivered simultaneously. To address this, we have only used the information on care provided to one adult recipient and one grandchild. In the case of HETUS, simultaneous care cannot be overcounted. This, again, could result in lower estimates for SHARE, suggesting that our observed differences between the two surveys may be conservative. Regarding simultaneous activities in HETUS, we could have considered secondary activities, but we chose not to because there are concerns about the reliability of that information (Durán Reference Durán2010). This is especially complex in the case of childcare for non-household children, as such care may occur at grandparents’ homes while they engage in other activities (e.g. cooking, cleaning). Consequently, this would underestimate the intensity of this care reported in HETUS compared to SHARE, which is the case in several countries, as we have already discussed.
Another limitation is that, due to the design of HETUS, it is not possible to perform a differentiated analysis by type of relationship between the caregiver and the recipient of care, which would certainly be of interest. One could reasonably expect reported differences depending on whether the care is provided to a parent, a sibling, a disabled child, or a neighbour, for example. Finally, as survey differences vary by country, international comparisons might also be subject to biases arising from the data collection processes or to differences in how these activities are performed in each country, of which we may not be aware.
Concluding remarks
To conclude, we have presented arguments regarding the implications of using two well-known and widely utilized surveys, HETUS and SHARE, to acknowledge and assess interhousehold caregiving and grandparental childcare in different European countries. A potential extension of this work would be to apply a Heckman selection model to account for differences in selection into caregiving across surveys. This could provide further insights into the survey effect.
While we reaffirm that time diary surveys like HETUS are the most suitable instrument to measure activities that are performed routinely almost every day, they lack the ability to accurately capture individuals who engage in more sporadic activities, as is the case of interhousehold care. Therefore, we highly recommend complementing time diary surveys with stylized questions that add sporadic dimensions, which are also essential for understanding the broader dynamics of caregivers over a longer time reference. Looking ahead, the development of a dedicated, care-focused survey that thoroughly measures the intensity, frequency, and context of both regular and occasional caregiving – including interhousehold and cross-generational exchanges – would represent a crucial step towards a more comprehensive understanding of care dynamics.
Acknowledgements
This work was financially supported by the Spanish Ministry of Science, Innovation and Universities, Spain [grant number PID2023-152779OB-I00, PID2020-114040RB-I00, PCI2021-121975, PID2020-118770RB-I00, and Ramon y Cajal grant numbers RYC-2017-22586 and RYC2018-024808-I]; the Catalan Research Agency [grant number 2021SGR00678 and CERCA programme]; and the European Commission [SUSTAINWELL project, a project of the HORIZON-CL2-2022-TRANSFORMATIONS-01-10].
Author contributions
An artificial intelligence (AI) tool was used solely to assist with language editing, including grammar and stylistic improvements. No AI tools were used for data analysis, interpretation, or generation of scientific content. The authors are fully responsible for all content presented in this manuscript.
Competing interests
The authors declare that they have no competing interests.
Ethical standards
This study uses publicly available secondary data from SHARE and HETUS. As the data contains no personally identifiable information, no formal ethical approval was required. All authors declare no ethical conflicts.
Appendix
Summary statistics of dependent variables of interhousehold care by sources
Source: HETUS 2010 and wave 2 of SHARE surveys for Austria, Belgium, France, Greece, Italy, Poland and Spain.
Questions from SHARE used to obtain the number of hours dedicated to helping members from outside the household
– General help to adult members (presumably):
[sp008] Now I would like to ask you about the help you have given to others. [During the time since last interview/In the last 12 months], have you personally given any kind of help listed on Card 38 to a family member from outside the household, a friend or neighbour? 1. Yes | 2. No
Card 38: 1. Personal care | 2. Practical household help | 3. Help with paperwork
Asked until three persons:
If sp008 = 1.Yes
[sp009_i] Which family member from outside the household, friend or neighbour have you helped [since the last interview/ in the last 12 months]?
[sp010_i] Which types of help have you given to this person?
Card 38: 1. Personal care | 2. Practical household help | 3. Help with paperwork
[sp011_i] In [the time since the last interview/the last 12 months], how often altogether have you given such help to this person? Was it …
1. Almost daily | 2. Almost every week | 3. Almost every month | 4. Less often
[sp012_i] About how many hours altogether did you give such help [on a typical day/in a typical week/in a typical month/in the last 12 months]?
– Looking after grandchildren:
[sp014] During [the time since the last interview/the last 12 months], have you regularly or occasionally looked after [your grandchild/your grandchildren] without the presence of the parents? 1. Yes | 2. No
If sp014 = 1. Yes
[sp015] From which of your children [is/are] [the grandchild/the grandchildren] you have looked after?
[sp016] On average, how often did you look after the child(ren) of [FLChild[i]] in [the time since the last interview/the last 12 months]? Was it …
1. Almost daily | 2. Almost every week | 3. Almost every month | 4. Less often
[sp017] About how many hours did you look after the child(ren) of [FLChild[i]] [on a typical day/in a typical week/in a typical month/in the last 12 months]?
