Helping older parents in Europe: the importance of grandparenthood, gender and care regime

ABSTRACT

In ageing societies, it is not uncommon to enter grandparenthood while one’s own parents are still living. Thus, a pertinent question is whether it matters for provision of help to parents if adult children have grandchildren they look after. Earlier studies addressing help to more generations conclude that providing help to one generation increases the likelihood of helping another, and not the contrary. Here, we investigate whether gender and welfare state context make a difference for this finding by using data from the Survey of Health, Ageing and Retirement in Europe and the Norwegian Life course, Ageing and Generation study. The 14 countries included represent four different care regimes. Overall, we find that adult children who have grandchildren they look after on a weekly basis are more inclined to provide weekly help and care to parents compared to those without such frequent grandchild care responsibilities. However, the results suggest that gender and care regime matter. Helping more generations seems easier in some contexts than in others, and the contrasts across regimes are considerably greater for women than for men. When there are few alternatives to family care, many (women in particular) may have to prioritise which generation to help.

KEYWORDS:

• Grandparenting
• intergenerational relationships
• parent care
• sandwich generation

Introduction

A large number of studies on intergenerational help exchanges address parent and grandchild care (e.g. Laditka and Laditka 2001; Ogg and Renaut 2006 for parent care; and Hank and Buber 2009; Igel and Szydlik 2011 for grandchild care). Independent of the research focus, whether it be support to the older generation or to the younger, information about the surrounding family structure is typically missing. Consequently, we seldom know if the adult children in question have grandchildren or if the grandparents studied have parents still living. Such information is, however, of major relevance for understanding helping patterns in ageing societies where a substantial proportion of middle-aged men and women have both parents and grandchildren, and consequently may encounter care expectations from multiple generations. Here, we will attempt to remedy part of this shortcoming by taking a multigenerational perspective and analysing the provision of help to parents by daughters and sons across Europe. Some of these adult children only have parents they help, others have become grandparents and may therefore (also) be engaged in grandchild care, and others again may not be helping either generation.

Intuitively, one might think in zero-sum terms and assume that men and women with grandchildren they look after, are less likely to help their parents compared to those without grandchild care responsibilities. As time is a finite resource, engagement in one activity often takes place at the expense of involvement in another. However, the few existing studies addressing provision of help to more than one generation do not give any indication of a zero-sum phenomenon in the family: helping one generation does not seem to reduce the likelihood of helping another (Grundy and Henretta 2006; Fingerman et al. 2011; Vlachantoni et al. 2019; Železná 2018). Actually, Grundy and Henretta (2006), as well as Vlachantoni et al. (2019) and Železná (2018), have concluded that providing help and care to parents increases the probability of helping younger generations, not the contrary. The explanation suggested is in line with the family solidarity model (Bengtson and Roberts 1991). In families with strong solidarity, family members tend to help both generations instead of prioritising, whereas those in families with weak solidarity are more likely to not provide any help at all (Grundy and Henretta 2006).

In the present article, we build on the conclusions from the work referred to above and address two key dimensions: gender and care regime. Our first aim is to investigate whether it matters for provision of help and care to parents if adult children have grandchildren they look after. The second, and main, aim is then to understand if it makes a difference whether we base our analysis on men or women and if welfare state context is important for the results. To answer our questions, we employ data from two European ageing surveys: the Survey of Health, Ageing and Retirement in Europe (SHARE) and the Norwegian Life course, Ageing and Generation study (NorLAG). Together, the two studies include 14 countries representing different welfare states and care regimes.

Background

The debate that followed in the wake of Elaine Brody’s writings about ‘women in the middle’ (1990) centred on the likelihood of being a member of the so-called ‘sandwich generation’ – torn between the needs of older and younger family generations. Being ‘sandwiched’ was commonly interpreted as having both parents in need of help and minor children (and a paid job). Several scholars argued that this is a rather uncommon situation (e.g. Uhlenberg 1993; Rosenthal et al. 1996) and it was pointed out that a more common squeeze in ageing societies would involve ageing parents and young grandchildren (Soldo 1996). However, few have provided actual figures showing the shares of individuals in such potential family ‘squeeze’ constellations. One exception is Vlachantoni and colleagues (2019), who have reported that around 30% of 55-year-olds in Britain have both parents and grandchildren. Furthermore, Leopold and Skopek’s (2015) estimates, based on the median age for birth of first grandchild and for the death of the second parent, suggest that there is a large probability of having parents when entering grandparenthood, but that this probability is greater in some countries than in others, and greater for women than for men. This brings us to the first of our two key dimensions: gender.

The importance of gender

The reason why women are more likely than men to simultaneously have both parents and grandchildren has to do with gender differences in fertility patterns. Women are usually younger than men are when becoming a parent. Thus, they are also on average younger when entering grandparenthood – and the younger individuals are when their first grandchild is born, the more likely they are to still have parents. As for family involvement, women’s pivotal role as carers and kin-keepers (Rosenthal 1985) has been repeatedly documented in previous research. In their role as daughters they are more likely to care for parents compared to men (e.g. Laditka and Laditka 2001; Ogg and Renaut 2006), and as grandmothers they are more inclined to look after grandchildren (Hank and Buber 2009; Luo et al. 2012).

Although women have a greater likelihood of having parents still living when grandchildren are born and of being involved in family care provision, they are not necessarily more inclined than men are to help ageing parents if they also have grandchildren they look after. It could be argued that it is easier for women to help both parents and grandchildren, as they are usually more involved in family care than men are, and thus perhaps more used to juggle different care responsibilities. On the other hand, women are more likely to provide frequent and more demanding care compared to men. Findings from earlier studies on parent care show that daughters spend more time helping and caring (Grigoryeva 2017), they provide more types of help (Laditka and Laditka 2001), they are more likely to help for a longer period of time (Szinovacs and Davey 2013) and are more often involved in personal care (Haberkern et al. 2015) compared to sons. All these factors may make it difficult for women to provide help to more than one generation. To our knowledge, only Vlachantoni et al. (2019) have analysed help to multiple generations separately for men and women, finding no gender differences in their British sample. The importance of gender may, however, vary across welfare states, depending on care policies – the second key dimension of our study.

The importance of care regimes

That welfare systems and care regimes matter for both parent and grandchild care has been documented in several earlier studies (e.g. Kalmijn and Saraceno 2008; Igel and Szydlik 2011; Bordone et al. 2017). A rather clear European north–south gradient has emerged from this research. The likelihood of providing at least some help and support is reported to be greatest in the north and then gradually decreasing towards southern Europe. When accounting for the amount, the picture is reversed, with provision being more frequent in the south (Igel and Szydlik 2011; Brandt 2013) and often more demanding (i.e. personal care) (Brandt et al. 2009).

Scholars who have attempted to cluster countries and identify care regimes based on policies targeting care for both children and old people, have found it a challenging endeavour, as the two seldom follow the same logics (Leitner 2003; Bettio and Plantenga 2004; Saraceno and Keck 2010). Thus, different typologies have been proposed, with the number and characteristics of regimes depending on which indicators are considered, as well as on the number of countries with available data.

One way of grouping nations is to order them on an underlying dimension of familialism or extent of public services and transfers to children and old people (Leitner 2003; Saraceno and Keck 2010). Countries with a high degree of familialism, meaning that neither public alternatives to family care nor financial support for providing such care is offered, are then grouped together. At the other end of the spectrum are the more de-familialised countries where the broad availability of universal public care services have reduced families’ care responsibilities and dependencies. A third type of regime has been identified as the ‘supported familialistic’ regime (Saraceno and Keck 2010), characterised by offering financial compensations to family members with care responsibilities (e.g. cash-for-care payments).

Although there is ample evidence showing that policies shape the provision of both help to parents and grandchild care, it is not clear whether or how the association between the two differs across regime (in strength and direction). The studies we are building on here, have not addressed this issue. Vlachantoni et al. (2019) only included Britain, Grundy and Henretta (2006) employed data from two so-called liberal welfare states (the United States and the United Kingdom), and Železná (2018), using data from Europe, only controlled for country when assessing the provision of grandchild care. Thus, the question remains: is it easier to provide help to more than one generation in de-familialised welfare states, where family help is less extensive and demanding because of the availability of care services, than it is in welfare states where care is regarded a family responsibility? In strong familialistic welfare states, care provision may be too intensive to help more than one generation and family members may therefore have no option other than prioritising. Or they may be forced to help more generations because of the lack of formal care alternatives. Such mechanisms may also vary between men and women, which brings us to reflect upon the combination of our two key dimensions.

The combination of gender and care regimes

In all types of care regimes, no matter the level of public services, women provide more care to both younger and older family generations than men do, but the availability and the level of care services seem crucial for the size of this gender gap (Leitner 2003; Da Roit et al. 2015). De-familialised care policies relieve the family from its responsibilities by providing other care options, thus reducing the gendered division of family care. Familialistic policies, on the other hand, uphold the caring function of the family, and as care provision is mainly the responsibility of women, such policies contribute to a strengthening of the gender gap (Leitner 2003).

Only a few studies have investigated how macro-level factors may influence the gendered division of family care across different regimes (Schmid et al. 2012; Haberkern et al. 2015). These provide important insight by showing how gender differences in provision of parent care are greater in so-called familialistic countries than in welfare states characterised by de-familialisation. When services are universal and extensive, more family members are involved, but on a less intense and frequent level, with smaller gender differences as a result. When there are few alternatives to family care, on the other hand, women are considerably more likely than men to take on care responsibilities, thus strengthening the gender gap. According to Da Roit et al. (2015), adding economic and cultural factors to the analysis does not seem to alter this conclusion.

Our main concern here is to investigate whether gender and care regime make a difference for the likelihood of helping parents if adult children have grandchildren they look after. If the availability of care services is more important for women’s helping behaviour and if it is easier to combine help to older and younger generations when service levels are high, we should expect greater differences across regimes for women than for men. It could also be that findings from earlier research represent a more universal phenomenon – provision of family help is, in any case, simply not a zero-sum game. If so, our analyses should reveal similar patterns across both gender and care regimes.

Data and method

Sample

To investigate our questions, we use data from SHARE and NorLAG. In order to have data collected around the same time, we employ wave 2 of both surveys, which was conducted in 2006/2007 for SHARE and in 2007/2008 for NorLAG. In combination, the two studies comprise data from 14 nations: Austria, Belgium, Czech Republic, Denmark, France, Germany, Greece, Italy, Netherlands, Poland, Spain, Sweden and Switzerland (SHARE) and Norway (NorLAG), including more than 40,000 respondents aged 50 and older.

Data collection procedures of the two surveys differ somewhat. SHARE uses (predominantly) computer-assisted personal interviews (CAPI), whereas NorLAG collects data through telephone interviews (CATI), alongside self-administered postal questionnaires (for more details, see Börsch-Supan et al. 2013 for SHARE and Slagsvold et al. 2012 for NorLAG). Survey questions for family help and care are almost identical, which makes it feasible to merge the two data sets for the purpose of our analyses. We here include respondents with own parents still living who provided information on the questions we use for our analyses (n = 8129 adult children, of whom 3899 have grandchildren).

Dependent variable: help to parents

Our dependent variable is provision of regular help to parents. Regular help is here defined as practical help and/or personal care provided approximately once a week or more often during the past year. In both SHARE and NorLAG, practical help includes help with household chores, house repair, gardening, transportation, shopping and similar, whereas personal care means help with eating, getting up, dressing, bathing, etc. Respondents in both studies could name up to three different receivers. As for the frequency, NorLAG respondents were only asked about regular help (i.e. at least monthly), meaning that help provided less often was not to be considered. A follow-up question about frequency (number of times per month) was asked to all providers. In SHARE, the question did not have the same restriction but instead had a follow-up question with the following answer categories: almost daily/almost weekly/almost monthly/less often. For our analyses, the answers ‘almost daily’ and ‘almost weekly’ in SHARE and four or more times per month in NorLAG are here considered as frequent (weekly) help provision in order to harmonise the two measures.

Independent variable: grandchild care

Our main independent variable is provision of regular grandchild care to grandchild/ren (yes/no). In SHARE, the question about grandchild care was asked in the interview, in NorLAG it was part of the postal questionnaire. The questions in the two surveys were organised somewhat differently. In SHARE, respondents were first asked if they had looked after grandchildren regularly or occasionally (last 12 months), and then they were to specify how often (almost daily/almost weekly/almost monthly/less often). In NorLAG, the question about provision of grandchild care had the following answer categories: daily/weekly/monthly/several times a year/less often/never. All respondents reporting on caring for grandchildren (almost) weekly or more often are here regarded as providers of regular (weekly) grandchild care.

Control variables

We include the following controls: respondents’ age, partner status, education, employment and health status, which are characteristics that may influence provision of both help to parents and grandchild care. Partner status indicates whether or not the respondents lived with a partner (spouse/ cohabitant) at the time of the interview. Education is categorised into three groups based on the International Standard Classification of Education (low = ISCED 0–2, medium = ISCED 3–4, high = ISCED 5–6) (ISCED 1997 levels, which is used for data up to 2013). For employment, we separate full-time workers from those working part-time (with ‘not employed’ as the reference category). Finally, health status, which is measured by a similar question in SHARE and NorLAG on health limitations, indicates whether respondents are limited in daily activities because of health problems (yes/no).

Gender and care regimes

Our key dimensions in the analyses are gender and care regime. Taken together, we have data for 4482 daughters and 3647 sons in 14 European countries that are clustered in different regimes. The categorisation we use is based on the work by Saraceno and Keck (2011), who consider care for young children (parental leave and formal childcare) and for the old (residential care and home-based care among 65+), as well as acknowledgement of unpaid family work through contributions, in all the countries included in our analyses (except Switzerland). The first group covers Scandinavia (Denmark, Norway, Sweden), characterised by strong de-familialisation and weak supported familialism (the de-familialised regime, n = 2534). The Southern European countries (Greece, Italy, Spain) on the other hand, have low scores on most indicators (i.e. weak de-familialisation and weak supported familialism). Poland shares many characteristics with the countries in Southern Europe (Saraceno and Keck 2010; 2011), including a Catholic tradition (Saxonberg and Szelewa 2007), and is thus included in this familialised regime (n = 2090). The remaining countries are divided into two clusters (a similar split, although based on fewer countries, is made by Bettio and Plantenga (2004)). One is characterised by weak de-familialisation, but strong supported familialism and includes Germany, Austria and the Czech Republic (n = 1195). The other consists of countries that are harder to categorise, and are thus described by Saraceno and Keck (2011) as ‘internally divergent’ (France, Belgium and the Netherlands). For example, countries in this regime may be characterised by strong de-familialisation in the case of eldercare and strong familialisation for childcare, or the other way around. Switzerland is also included in this ‘mixed’ regime (n = 2310).

Models

We estimate multivariate logit models predicting the impact of regular (weekly) grandchild care on regular (weekly) help to older parents. In order to account for non-independence of observations on the country level, we use clustered standard errors. We first show an overall model before presenting separate models for daughters and sons, as well as for the four care regimes. Finally, we assess the association between grandchild care and parent help for daughters and sons separately within the four regimes.

Results

Descriptive statistics

In our sample of adult children aged 50 and older, 52% of the daughters and 43% of the sons are grandparents, confirming the gender differences in demographic patterns discussed above. The shares not only vary between men and women but also across care regimes due to the countries’ different demographic profiles (i.e. life expectancy, age at first birth and childlessness). Table 1 provides descriptive statistics for the variables included in our analyses, by gender and care regime. With few exceptions, gender differences are significant along all variables and in all regimes. Daughters are more likely to help parents and to look after grandchildren, compared to sons. An important reason for the latter is obviously also that more daughters than sons actually have grandchildren. Sons, on the other hand, are more likely to live with a partner, to be full-time employed, and to have completed higher education (except in the ‘de-familialised’ regime, where more daughters than sons have a high educational level). Finally, the share reporting health limitations is larger for daughters than for sons (except in the ‘supported familialistic’ regime).

Table 1. Descriptive statistics by gender and care regime (%).

	De-familialised regime		Mixed regime		Supported familialistic regime		Familialistic regime
	Daughters	Sons	Daughters	Sons	Daughters	Sons	Daughters	Sons
Regular help to parents	18	11	23	12	24	12	20	8
Regular grandchild careª	19	11	21	12	18	11	24	14
Age (mean)	56.3	56.8	55.9	56.3	56.7	56.3	55.8	56.9
Have partner	72	82	76	85	75	83	85	92
Education
High	40	33	26	30	17	29	13	18
Medium	48	55	58	57	75	68	52	58
Low	12	12	16	13	8	2	35	24
Employment
Full time	46	74	18	50	31	58	20	51
Part time	27	8	35	17	19	8	13	11
Not employed	27	18	47	33	50	34	67	38
Health limitations	29	19	35	30	40	40	28	24
Have grandchildren	58	47	50	42	59	50	44	34
N	1374	1160	1274	1036	687	508	1147	943

Sources and notes: NorLAG and SHARE wave 2; ªrefers to the whole sample, including individuals without grandchildren. Regular help to parents and regular grandchild care is help/care that is provided at least weekly. De-familialised regime: Norway, Sweden and Denmark; Mixed regime: Netherlands, Belgium, Switzerland and France; Supported familialistic regime: Germany, Czech Republic and Austria; and Familialistic regime: Italy, Spain, Greece and Poland.

The association between grandchild care and help to parents

The overall results for our entire sample (Table 2) show that the correlation between regular grandchild care and provision of regular help to parents is positive and statistically significant, also when controlling for various characteristics of the respondents (including gender and care regimes). Respondents with grandchildren they look after on a weekly basis are more likely to provide help to parents (at least weekly) than those who are not engaged in such frequent grandchild care. This finding confirms previous research findings: helping one generation is linked to a higher likelihood of helping another.

Table 2. Logistic regression models for provision of regular (weekly) help to parents.

	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)
Regular (weekly) grandchild care	0.34*** (0.063)	0.19** (0.065)
Gender (1 = female)		0.77*** (0.066)
Care regime (ref.: De-familialised regime)
Mixed regime		0.22 (0.232)
Supported familialistic regime		0.23* (0.088)
Familialistic regime		0.00 (0.199)
Age		0.02*** (0.007)
Partner		−0.03 (0.079)
Education (ref.: low)
Medium		0.37** (0.131)
High		0.32* (0.145)
Employment (ref.: not employed)
Part time		−0.09 (0.085)
Full time		−0.19* (0.078)
Health (1 = limited in daily activities)		−0.11 (0.070)
Constant	−1.69 (0.095)	−3.77 (0.327)
N	8129	8129

Sources and notes: NorLAG and SHARE wave 2; ***p < 0.001, **p < 0.01, *p < 0.05.

Turning to the control variables, daughters are generally more inclined to help older parents than sons are, which is in line with earlier research. Differences across regimes are modest, but compared to the Scandinavian ‘de-familialised’ regime, providing regular parent help is significantly more likely in the ‘supported familialistic’ regime (i.e. Germany, Austria and the Czech Republic). The higher the respondents’ age, the more likely they are to provide help and care (presumably caused by parents’ corresponding higher age and therefore greater needs). Higher-educated respondents are more inclined to help than the lower educated are, whereas full-time employed are less likely compared to the non-employed. Finally, there is no pronounced effect of health limitation on help provision to parents.

Our main concern here is to investigate whether the positive correlation between grandchild care and help to parents applies to both genders and to different care regimes. Thus, we perform separate analyses for daughters and sons, as well as for the four regimes (Table 3). The analyses include all controls, although not presented here (results available upon request). The results show a positive and significant association between grandchild care and help to parents among daughters, not among sons. It should, however, be noted that the two coefficients do not differ significantly. The analyses by gender also show that regular help provision in the ‘supported familialistic’ regime is more likely than in the ‘de-familialised’ Scandinavia, but only for daughters, not for sons. Moving on to the separate analyses for care regimes, the association between grandchild care and help to parents turns out to be positive and significant in all regimes, with the exception of the ‘familialistic’ Southern Europe (and Poland). Although not significant, it should be noted that the correlation is negative in this latter regime. The gender difference for provision of help/care to parents is confirmed for all four regimes.

Table 3. Logistic regression models for provision of regular (weekly) help to parents, by gender (model 1)/by care regime (model 2).

	Model 1: Gender		Model 2: Care regime
	Daughters	Sons	De-familialised regime	Mixed regime	Supported familialistic regime	Familialistic regime
	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)
Provision of regular (weekly) grandchild care	0.19* (0.076)	0.21 (0.131)	0.18* (0.090)	0.35* (0.160)	0.35*** (0.069)	−0.08 (0.086)
Gender (1 = female)	–	–	0.69*** (0.165)	0.65*** (0.091)	0.78*** (0.085)	1.04*** (0.099)
Care regimes (ref.: De-familialised regime)
Mixed regime	0.23 (0.258)	0.21 (0.252)	–	–	–	–
Supported familialistic regime	0.28*** (0.068)	0.12 (0.233)	–	–	–	–
Familialistic regime	0.12 (0.183)	−0.22 (0.275)	–	–	–	–
Constant	−2.96 (0.374)	−4.21 (0.641)	−4.22 (0.676)	−3.38 (0.758)	−4.07 (0.885)	−3.62 (0.731)
Number of observations	4482	3647	2534	2310	1195	2089

Sources and notes: NorLAG and SHARE wave 2; including controls as shown in Table 2. ***p < 0.001, **p < 0.01, *p < 0.05.

Finally, we estimate separate models for daughters and sons within each of the regimes, again controlling for all characteristics reported above. These analyses show that the most important differences in the association between grandchild care and help to parents are found when accounting for the combination of gender and care regime (see Figure 1 reporting average marginal effects and Table 4 reporting logit coefficients). For sons, the correlation is positive in all four regimes, but not significant. For daughters, the correlation is significant in all regimes, but with opposite directions. In the ‘de-familialised’ regime, as well as in the ‘mixed’ and the ‘supported familialistic’ regime, the association is positive. In the ‘familialistic’ regime, on the other hand, it is negative. Models including interaction effects (available upon request) confirm that daughters in the ‘familialistic’ Southern Europe and Poland stand out as significantly different from daughters in the three other regimes.

Figure 1. Grandchild care influences on regular help to parents by gender and care regime. Sources and notes: NorLAG and SHARE wave 2; average marginal effects from logistic regressions with clustered standard errors, including controls as shown in Table 4.

Table 4. Logistic regression models for provision of regular (weekly) help to parents, by gender and care regime.

	De-familialised regime		Mixed regime		Supported familialistic regime		Familialistic regime
	Daughters	Sons	Daughters	Sons	Daughters	Sons	Daughters	Sons
	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)	Logit coeff. (clustered S.E.)
Provision of regular grandchild care	0.21* (0.095)	0.07 (0.204)	0.41** (0.125)	0.17 (0.303)	0.29*** (0.074)	0.45 (0.265)	−0.18** (0.057)	0.30 (0.253)
Age	0.03 (0.018)	0.03 (0.023)	0.02* (0.007)	0.03 (0.022)	0.03*** (0.003)	0.05 (0.039)	0.01 (0.020)	0.03 (0.022)
Partner	0.05 (0.181)	0.22 (0.246)	−0.15 (0.143)	−0.41** (0.147)	−0.21 (0.265)	−0.20* (0.082)	0.20 (0.175)	−0.38 (0.247)
Education (ref.: low)
Medium	0.15 (0.286)	0.28 (0.181)	0.75* (0.309)	0.73*** (0.195)	0.08 (0.151)	13.30*** (1.121)	0.21 (0.160)	0.39 (0.327)
High	0.12 (0.377)	0.12 (0.430)	0.64*** (0.140)	0.73 (0.381)	0.46*** (0.128)	13.04*** (0.865)	−0.08 (0.354)	0.51 (0.310)
Employment (ref.: not employed)
Part time	0.00 (0.099)	−0.35 (0.525)	−0.19 (0.134)	−0.25 (0.228)	0.21 (0.265)	−0.82 (1.288)	0.11 (0.166)	−0.14 (0.596)
Full time	−0.01 (0.094)	−0.04 (0.458)	−0.66** (0.236)	−0.15 (0.291)	−0.38*** (0.100)	0.20 (0.337)	−0.13 (0.133)	−0.24 (0.484)
Health (limited in daily activities)	−0.28 (0.301)	0.49* (0.241)	−0.05 (0.105)	−0.06 (0.155)	0.09 (0.147)	−0.62 (0.677)	−0.21 (0.221)	−0.24 (0.289)
Constant	−3.44 (0.864)	−4.268 (1.050)	−2.66 (0.619)	−3.74 (1.431)	−2.64 (0.269)	−17.79 (2.493)	−2.29 (1.072)	−3.94 (1.856)
Number of observations	1374	1160	1274	1036	687	508	1147	943

Sources and notes: NorLAG and SHARE wave 2; including controls as shown in Table 2. ***p < 0.001, **p < 0.01, *p < 0.05.

Table 4 includes all control variables for the separate analyses (by gender and care regime). Of special interest are employment patterns and effects on provision of help and care to parents, as the two spheres may compete and reconciliation may differ across regimes. In Central Europe (the ‘mixed’ and ‘supported familialistic’ regimes), full-time employed daughters are less inclined to provide help to parents than the non-employed. In the de-familialised (Scandinavian) regime, as well as in the familialistic Southern Europe, there is no effect of employment, neither for daughters, nor for sons. It should be noted though, that daughters in the latter regime are considerably less likely to be employed (67% are non-employed compared to 27% in Scandinavia, see Table 1).

In our analyses, adult children providing grandchild care are compared to those who do not, either because they do not have grandchildren or because they have, but do not look after them on a regular (weekly) basis. We have also run models comparing the two latter groups. Overall, adult children without grandchildren are somewhat more likely to help parents compared to those who have grandchildren without providing grandchild care. When separating the analyses by gender and regime, we note that this finding only holds for some regimes (the ‘supported familialistic’ and the ‘familialistic’ regimes). In the remaining two, the correlation is positive, but smaller and thus not statistically significant (results available upon request).

Discussion

In ageing societies where a considerable share of men and women are likely to become grandparents when their parents are still living, a pertinent question is if help to one generation takes place at the expense of help to another. Thus, we set out to analyse whether it matters for adult children’s provision of parent help if they also have grandchildren they look after, and whether gender and care regime make a difference for the results. Overall, there does not seem to be any trade-off as looking after grandchildren at least weekly is correlated with a higher probability of providing help to parents (weekly or more often). However, this result depends on gender and care regime, meaning that in some contexts, zero-sum situations may exist.

When separating the analyses by gender, we find that the positive association between help to younger and older generations only holds for daughters, however, the correlation is rather modest. This may seem surprising, given the kin-keeping role of women, including their experience in combining tasks and juggling different responsibilities (Martire and Stephens 2003). On the other hand, women are more often involved in frequent, intense caregiving (e.g. Laditka and Laditka 2001), which may make it more difficult to combine help to different generations. Thus, the moderate result could be caused by these two competing mechanisms.

Turning then to care regimes, we find evidence for a positive correlation between grandchild care and help to parents in all regimes, except in the one characterised by familialisation (Southern Europe and Poland). In this regime, there seems to be a negative tendency (although not significant), indicating that the positive correlation between help to older and younger generations, reported in earlier studies (as well as in our overall model) may not be a universal phenomenon.

Taking into account the combination of gender and care regime reveals a more insightful picture. Among sons in all regimes, looking after grandchildren is positively associated with help to parents, but not statistically significant. For daughters, the correlation is significant in all regimes, but with contrasting directions. In the ‘de-familialised’, the ‘mixed’ and the ‘supported familialistic’ regimes, providing regular grandchild care is associated with a higher likelihood of also helping parents regularly. In the ‘familialistic’ regime on the other hand, the correlation is negative, meaning that daughters who have grandchildren they look after are less likely to help their parents compared to those without such grandchild care responsibilities. These results confirm previous findings indicating that welfare state context seems to matter more for daughters’ help provision than for sons’ (Schmid et al. 2012).

In previous research, scholars have turned to the family solidarity model (Bengtson and Roberts 1991) to explain the positive association between help to younger and older generations (e.g. Grundy and Henretta 2006). Our results indicate that such an explanation does not necessarily hold for all contexts. No matter the strength of family solidarity, in some welfare state, it may be difficult to help parents if adult children (daughters in particular) also have grandchildren they provide care to.

The finding for daughters in the ‘familialistic’ regime supports this line of arguing. Helping one generation may become too encompassing when there are few alternatives to family care, thus reducing the possibility of helping other family members. In this regime, a substantial share of adult children with parents in need of help, provide support on a daily basis (Ogg and Renaut 2006). The same is the case for grandchild care (Herlofson and Hagestad 2012). Also, helping parents more often includes personal care, which is typically more intense than practical help. As it may be difficult to meet such care responsibilities, it could be that many are left with no other option than to prioritise which generation to help. In care regimes with a higher level of formal services, on the other hand, the family is relieved of some of its responsibilities, making it easier to help more than one generation.

To facilitate interpretations, we decided to present single models for each gender and regime group. In addition, we estimated overall models including interaction effects in order to check whether significant gender and regime differences persist. We also performed separate analyses for practical help and personal care, which confirmed the overall findings described above. However, when separating the two types of support, the reported result for daughters in the familialistic regime is no longer statistically significant, which could mean that it is the combination of the two (practical help and personal care) that hinders multigenerational help provision. Further robustness checks, including different modelling strategies (e.g. nested models) and operationalisations (e.g. intensities), provided no doubts concerning our main results (analyses available upon request).

It should be noted that because of gender differences in demographic patterns, women are more likely than men are to simultaneously occupy the role of adult child and grandparent. They are, in general, also more inclined to help younger and older generations. As a consequence, more women than men are in a position where they may have to face multiple family responsibilities. The question is then whether or not they perceive this situation as a squeeze. Do they feel torn between the needs of both younger and older generations? Do they feel obliged to help or is providing help a voluntary choice? Future research should look more into the effects of various types of care provision among men and women in different care regimes (including also other dimensions, such as socio-economic status and migration background). Perhaps looking after grandchildren brings joy and energy, which makes it easier to fulfil caring responsibilities towards older parents. It has indeed been noted that the grandparent role ‘is like the dessert of life’ (Rosenmayr 2007). On the other hand, it could be that helping more than one generation renders family care providers exhausted and overburdened. In some contexts, they may even feel forced to prioritise one generation over the other, as the care load becomes too heavy.

Because of population ageing and the ongoing retrenchment of many welfare states, families’ care responsibilities are likely to increase (further) in the years to come. Policymakers should, however, bear in mind that such a development may come at a cost, not only for individuals, but also for families and the society as a whole.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes on contributors

Katharina Herlofson is a senior researcher in the department of Ageing Research and Housing Studies at Norwegian Social Research (NOVA), Oslo Metropolitan University. She has a PhD in sociology from the University of Oslo. Her main research interests are centred on gender, intergenerational ties and life course patterns in ageing societies. She has been involved in several national and international research projects on ageing and intergenerational solidarity, and is now a co-investigator for the Norwegian Life Course, Ageing and Generation Study (NorLAG).

Martina Brandt is Professor of Social Structure and Sociology of Ageing Societies at the Technical University Dortmund and a Senior Fellow of the SHARE central coordination in Munich. Currently, she is the director of the Institute for Sociology at TU Dortmund and leads the university-wide research focus on ‘Life in Ageing Societies’. She is interested in the links between family, work and health across the life course and social policies. Her recent quantitative research focuses on support networks and wellbeing of older people across different welfare states.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The research presented here was carried out with financial support from the Research Council of Norway (Norges Forskningsråd) [Grant N° 236997]. The analyses are based on data from NorLAG Wave 2 (DOI: 10.18712/NSD-NSD1461-V1) and SHARE Waves 1 and 2 (DOIs: 10.6103/SHARE.w1.600, 10.6103/SHARE.w2.600). See Slagsvold et al. (2012) and Börsch-Supan et al. (2013) for more details about the two studies. The NorLAG surveys are financed by the Research Council of Norway, The Ministry of Health and Care Services, Ministry of Labour, Ministry of Children, Equality and Social Inclusion, Ministry of Local Government and Regional Development, NOVA and Statistics Norway. The NorLAG dataset is part of the ACCESS Life Course infrastructure project funded by the National Financing Initiative for Research Infrastructure at the Research Council of Norway [Grant N° 195403] and NOVA (norlag.nova.no). The SHARE data collection has been primarily funded by the European Commission through FP5 [QLK6-CT-2001-00360], FP6 [SHARE-I3: RII-CT-2006-062193, COMPARE: CIT5-CT-2005-028857, SHARELIFE: CIT4-CT-2006-028812] and FP7 [SHARE-PREP: N°211909, SHARE-LEAP: N°227822, SHARE M4: N°261982]. Additional funding from the German Ministry of Education and Research, the Max Planck Society for the Advancement of Science, the U.S. National Institute on Aging [U01_AG09740-13S2, P01_AG005842, P01_AG08291, P30_AG12815, R21_AG025169, Y1-AG-4553-01, IAG_BSR06-11, OGHA_04-064, HHSN271201300071C] and from various national funding sources is gratefully acknowledged (www.share-project.org).