Public transfers and the balance of intergenerational family support in Europe Open Access

With the rapid growth of older population and limited public resources available to support their needs, policymakers are increasingly interested in understanding the exchange of support between family generations and the effects of public policies on family-transfer behavior. Such information would improve the assessment of the true impact of public transfers on the overall flow of family intergenerational support and provide answers to key questions of policy interest, such as who the true (as opposed to intended) beneficiaries are of public support or whether public support displaces or encourages private support. Consequently, intergenerational family support and its nexus with public policy have come to the forefront of social science research. The objective of this study is to estimate the net value of transfers exchanged between parents and adult children, and examine its relationship with some of the major public transfers that contribute to intergenerational redistribution of resources.

Previous research has provided important insights on family transfers and their relationship with public transfers in Europe. Family transfers of money flow from parents to adult children, irrespective of age (Attias-Donfut et al. 2005; Albertini et al. 2007; Litwin et al. 2008; Mudrazija 2014). Non-financial transfers follow a more complex pattern. Not counting grandchild care, adult children overall provide more support to their aging parents than they receive from them (Albertini et al. 2007; Deindl et al. 2014). However, these transfers exhibit a strong age gradient, as only parents aged 70 and older are net recipients (Albertini et al. 2007). Accounting for grandchild care, parents appear to become net recipients of support from their adult children even later, past age 80 (Mudrazija 2014).

Notwithstanding the general patterns, there are important cross-national differences in the likelihood and intensity of family support that cannot be fully accounted by individual characteristics of parents and children (Albertini et al. 2007; Igel and Szydlik 2011; Brandt and Deindl 2013). Some researchers have attributed these differences to variation in welfare regime characteristics across European countries (e.g. Albertini et al. 2007; Leopold and Raab 2011; Albertini and Kohli 2013), thereby implying that differences in principles, structure, and generosity of public systems of intergenerational redistribution of resources are important determinants of family-transfer behavior. Others have examined the relationship between the key characteristics of welfare regimes, in particular the level of public assistance such as provision of social services or social policy expenditures, with family support more directly. Their studies find that public assistance is positively associated with the likelihood of parents providing financial and time support to their adult children, but negatively with the intensity of support (e.g. Brandt and Deindl 2013). Adult children's provision of time-intensive support to older parents across Europe is increasingly being replaced by public assistance while children specialize as providers of less intensive help (Brandt et al. 2009; Igel et al. 2009). Overall, the results of these studies are consistent with a decrease in the intensity (i.e. magnitude) of family support, but in the context of specialization of families and the public sector for the provision of certain types of support, which may result in an overall higher incidence of family transfers.

Empirical evidence from various countries is largely consistent with this notion. Künemund and Rein (1999) focus on the incidence of family support in Europe, and find evidence of crowding in of private by public support. Similarly, Motel-Klingeibel et al. (2005) conclude that public transfers do not adversely impact provision of family transfers using data for England, Germany, Israel, Norway, and Spain. Conversely, studies focused on estimating the impact of changes in public transfers on the intensity of family transfers find some evidence of partial displacement of family transfers. Jensen (2003) finds a moderate (25–30%) crowding out of child-to-parent financial transfers by public pensions in South Africa. Kang and Lee (2003) estimate the magnitude of the displacement of family financial transfers by public transfers in Korea to be potentially in excess of 70%. Studying the effects of unemployment insurance benefits on family transfers in the United States, Schoeni (2002) finds evidence of partial (24–40%) crowding out, while Villanueva (2005) finds more modest (8–11%) effects in his study of the United States, Germany, and the UK.

While much important work in this area of research has been done, some of the key questions of policy interest remain unanswered or only partially answered. For example, although studies of unidirectional transfers of time or money between parents and children (e.g. Attias-Donfut et al. 2005; Kalmijn and Saraceno 2008) provide important insights about the frequency and intensity of family support, they cannot capture the simultaneity of changes in different types of family transfers and are not able to describe the pattern of rebalancing of family intergenerational support in response to variation in public support. Exploring these issues in the context of balance of family intergenerational support holds promise of advancing the field, and few existing studies applying this approach reveal its potential usefulness. Using data on financial and non-financial transfers to calculate the balance of intergenerational exchange between older parents and adult children, Litwin et al. (2008) established that parents are net providers of support to children until advanced old age, while Mudrazija (2014) found that the magnitude of parents–child redistribution of resources over the adult life cycle was moderated by the generosity of the welfare state.

Furthermore, when studying a single type of family transfers and/or unidirectional flow of intergenerational support, it is important to account for the fact that contextual factors such as geographic distance between parents and adult children may determine whether financial and practical support are functional substitutes or complements, which in turn can significantly affect the observed prevalence and magnitude of any individual type of transfer. Such studies, therefore, need to explicitly deal with the issue of complementarity or substitutability of various types of transfers (e.g. Bonsang 2007; Antman 2012). In contrast, the alternative approach pursued in this study focuses on exploring the impact of public policies on the resultant overall net intergenerational redistribution of resources at the family level. In other words, this analytic approach is primarily interested in the issue of who is the net beneficiary in the exchange of intergenerational family support at different stages of the life cycle in the context of public intergenerational redistribution of resources, and what is the value of the net benefit (i.e. the results of exchange), rather than focusing on the issues of whether, how, and why different types of family support interact (i.e. the process of exchange).

The current study uses data on 13 European countries to estimate the net value of transfers between parents and adult children – defined as the monetary value of financial and non-financial transfers that parents give to children, minus the monetary value of transfers they receive from children – and to determine its relationship with public social spending. Previous studies estimated the association of different elements of social policy expenditures with financial transfers between parent and children (Zissimopoulos and Smith 2010), and of total social policy expenditures and social services employment with the provision of support between parents and children (Deindl and Brandt 2011; Brandt and Deindl 2013). While building on their approach, this research is unique in exploring the public–private nexus of transfers in the context of net exchange of support between parents and children, which results in important modifications of country-level indicators included in the study. As the focus is on the intergenerational redistribution of resources, only those elements of social spending that redistribute between generations are of substantive interest and, therefore, some social policy expenditures such as unemployment benefits are not substantively interesting. Moreover, social spending variables for all countries are standardized to match the age distribution of population of a single country (Germany), which allows capturing more accurately the true redistributive impact of different elements of social spending across countries. Otherwise, the observed differences in social spending could at least partially reflect different age composition of population across countries rather than different levels of spending for any particular generation.

Finally, in order to accurately describe the link of public policies and net transfers between parents and children across different societies, it is important to account for the nested structure of data. Using multilevel model, this study aims at capturing the effects of the social context (i.e. public policies across countries) accounting for individual and family characteristics. The basic advantage of this approach over the standard regression approach is that it provides unbiased coefficient estimates and correct standard errors. Previous studies of public–family transfers link in Europe successfully applied similar estimation strategy (e.g. Brandt et al.2009; Brandt and Deindl 2013).

Over the years, multiple explanations have been proposed in the economic literature to describe the nature of family transfers. Need-based theories of giving like altruism (Barro 1974; Becker 1974), old-age security hypothesis (Leibenstein 1957; Caldwell 1976), parental repayment (Becker and Tomes 1976), or warm-glow giving hypothesis (Andreoni 1990) suggest that relative needs determine who transfer donors and recipients are going to be. Similar ideas can be found in the sociological literature where, for example, contingency theory contends that sharing resources across generations is contingent upon the recipient's needs for assistance (e.g. Fingerman et al. 2009). Various reciprocity-based theories represent the main alternative. In a narrower sense, exchange hypothesis (Cox 1987) suggests that intrafamilial transfers are akin to ‘payment’ for services, generally of commensurate value and happening simultaneously or within a fairly short time. In a broader sense, reciprocity also includes ‘repayment’ that can happen with a substantial time lag, without symmetry in value, and can be nominally made to a third party (Kohli and Künemund 2003).

The focus of the framework is on the exchanges between parents (P) aged 50 and older and their adult children (AC) aged 18–49, as well as the flows of resources between them and the government.¹ The thickness of the arrows represents the assumed approximate magnitude of transfer flows. The generation of parents aged 50 and older overall provides more financial and non-financial support to their adult children's generation, either by supporting them directly or indirectly by taking care of grandchildren (GC). The most salient feature of the framework is that public redistribution is an integral part of the overall intergenerational flow of resources. Government collects taxes and makes transfers, which are then to a large extent reallocated from working-age population to children and older-age population (Mason et al. 2006). Given that relatively more adult children than older parents are of working age, this results in an indirect flow of resources from adult children to older parents by means of taxes and public transfers for pensions and other social expenditures. This remains true even if one would consider public expenditures for young children as transfers benefiting their parents, that is, adult children (Lee and Mason 2011).

The horizontal axis right of the origin represents increasing governmental redistribution to older generation, and left of the origin to younger generation. The vertical axis represents net transfers from parents to children, which are increasingly positive above the origin, and increasingly negative below the origin. Focusing on the vertical axis, where government intergenerational redistribution of resources is age neutral, that is, equal to zero, parents and children can generally assume one of the three positions with respect to their exchange: first, parents can be net donors of transfers to children as represented with the point on the top dashed line where n > 0, second, they can be net recipients of transfers, which is represented on the bottom dashed line where n < 0, or, finally, their exchange can be in equilibrium as depicted on the middle dashed line that goes through the origin (n = 0). Regardless of the initial position of parents and children with respect to their exchange, the nature of the relationship with government redistribution of resources follows the same pattern: as government increases redistribution to older generations, net transfers from parents to children increase, too, whereas increased government redistribution to younger generations is associated with decreasing net transfers between parents and children.

The described framework gives rise to the following two research hypotheses:

1. Net transfers between parents and adult children benefit more individuals with greater need as indicated by their demographic, socioeconomic, and health status; and

2. The redistribution of resources between parents and adult children due to public spending is associated with a secondary redistribution from the public policy beneficiary to the other person.

Data for this study come from the second (2006−2007) wave of the Survey of Health, Ageing, and Retirement in Europe (SHARE), a cross-national panel study of individuals aged 50 and over (Börsch-Supan 2013). The decision to use the 2006–2007 wave rather than the more recent waves of the SHARE reflects the preference to focus on pre-financial crisis transfer patterns given that the crisis had very uneven impacts across European countries, and it is still unknown if the current trends represent only an aberration from the historic trends or rather a new normal. The initial sample included 14,543 children who provided financial and/or non-financial support to their parents or received it from them representing 13 countries: Austria, Belgium, Czech Republic, Denmark, France, Germany, Greece, Italy, the Netherlands, Poland, Spain, Sweden, and Switzerland.³ Of them, 562 children younger than 18 are excluded because the analytic focus is on adult children. As the information on practical help such as transportation, household chores, or administrative paperwork is available only for non-coresident parents and children, the analysis further excludes 2176 coresident adult children. Five cases are omitted due to suspect values – age difference between parents and children of 12 years or less – following a similar approach in prior research (e.g. Bonsang 2007). Finally, to limit the impact of large transfers made as a part of estate planning strategy, that is ‘early bequests’ (Arrondel and Laferrere 2001; McGarry 2001), the sample is trimmed on the dependent variable at the 98th percentile, following the approach applied in prior literature examining intergenerational transfers in Europe and the United States (Zissimopoulos and Smith 2010). The final sample includes 11,570 children from 8020 families.

The net value of transfers is defined as the monetary value of transfers from parents to their non-coresident child minus the monetary value of transfers they receive from the child. Therefore, this is a measure of disbalance (i.e. redistribution) in the exchange of support rather than a measure of the overall volume of exchange.⁴ The measure includes financial transfers given/received in the 12 months prior to the interview,⁵ time transfer of personal care, practical household help, administrative paperwork help, and grandchild care. Following Mudrazija (2014), grandchild care is included only if provided weekly or more frequently and if the annual total is at least 500 hours. The rationale for this decision is to include only those transfers that could plausibly be considered a substitute for formal childcare services, and therefore represent transfers from older parents to adult children with intrinsic economic value for the latter. This is consistent with the empirical evidence that intensive grandchild care is related with economic outcomes such as mothers’ employment (Di Gessa et al.2016). Moreover, determining the directionality of non-intensive and infrequent grandchild care is not straightforward since it cannot meaningfully substitute for formal childcare services, and its primary benefit may be bonding between family generations, that is affectual solidarity (Silverstein et al. 1998), which benefits grandparents, their adult children, and grandchildren.⁶

In the absence of direct information on the financial value of non-financial support, time transfers that are originally reported in hours are monetized using two alternative measures: (1) the national (or, when absent, the appropriate sectoral) legal minimum hourly wage rate in each country and (2) the national average hourly wage rates across countries. The first measure provides a conservative estimate of the financial equivalent of non-financial transfers and is in the primary focus of this study. The second measure provides a possible upper-bound estimate. Taken together, these measures allow establishing an approximate range of net-transfer values.⁷ Missing information on net transfers between parents and children and the predictors of interest is imputed following the multiple imputation procedure used by Leopold and Raab (2011).⁸ The resulting values of net transfers are adjusted for relative purchasing power parity (PPP) across the 13 countries.

The key predictor of interest is a ratio of two major government policies that redistribute resources to the generation of older persons (i.e. older parents) and younger persons (i.e. their adult children), respectively: spending on old-age and survivors’ insurance (OASI) benefits and family policy, both measured as a share of gross domestic product (GDP) in each country. OASI benefits and family policy spending are standardized to match the population structure of the reference country, Germany, by accounting for differences in the relative sizes of young, adult, and old-age population.⁹

The model also controls for various demographic and socioeconomic characteristics of parents and children. Parental control variables include age, sex, years of education, marital status, number of children, annual income, and gross financial wealth,¹⁰ difficulties in the activities of daily living (ADL) and instrumental activities of daily living (IADL), and limitation in usual activities during the six months prior to the interview. While ADL and IADL difficulties capture health issues that are chronic in nature, the indicator of recent health limitation primarily captures acute conditions. Annual income and gross financial wealth values are PPP-adjusted, and standardized using the square root of the household size, a common approach in the prior literature (e.g. Litwin and Sapir 2009). Children's controls include their sex, marital status, indicator of any children they may have, full-time employment indicator, and an indicator whether they live within 25 kilometers from their parents.¹¹ Unlike the parents, the majority of children in the sample did not experience marital transitions such as divorce, separation, or death of a spouse; therefore, children's marital status variable collapses these categories.

The analysis starts with an overview of values of net transfer between parents and children by age groups (50–59, 60–69, 70–79, and 80 and older) and a description of relative differences in each country's spending on OASI benefits and family policy. This is followed by an overview of sample means for parents and children. Next, a model of net transfers between parents and children is estimated with a multilevel, random intercept, model. The model includes three levels – child level, parents (i.e. household) level, and country level – reflecting the nested structure of the data. Therefore, the intercept for the latter two levels is considered random, while parameters associated with the other variables in the model area treated as fixed.¹² Model is fitted for the outcome variable calculated using the two alternative measures for monetizing the non-financial transfers to examine the sensitivity of results to the alternative monetization approaches. Other sensitivity tests are performed as well.

The results in Table 1 reveal that the net transfers from parents to adult children across the majority of European countries follow a similar pattern: they are positive for the age groups 50–59 and 60–69, lower but still positive for the age group 70–79, and for the oldest group of parents (80 and older), they continue decreasing and in most countries become negative, suggesting that old parents ultimately become net recipients of support from adult children. This general pattern is valid regardless of the monetization approach used for non-financial transfers. The difference in net-transfer values between the youngest and the oldest age groups, based on the minimum wage rate amounts for non-financial transfers, is generally the largest in Southern Europe (ranging from €8975 in Spain to about €6911 in Italy) and the smallest in the Northern European countries of Denmark and Sweden (€1138 and €1245, respectively), with the results for the Continental European countries generally between the two extremes.¹³ Therefore, the North–South geographic gradient in family-support patterns in Europe that has been noted in prior research (e.g. Brandt et al. 2009; Deindl and Brandt 2011; Igel and Szydlik 2011; Albertini and Kohli 2013) finds further support in data on net transfers.

Table 2 shows the levels of spending on OASI benefits and family policy, and their ratio across European countries. To facilitate comparisons between countries, as earlier explained, the levels of spending, measured as a share of GDP, are standardized using the population distribution of the reference country, Germany. As a result, in countries with a younger population structure than Germany standardized spending levels on family support are lower and on OASI benefits higher than using unadjusted spending levels, and vice versa. Results reveal that in years prior to the recent recession OASI-to-family spending ratio was generally smaller in Northern and Continental Europe than in Southern Europe, with the exception of Poland that had a spending profile that was much more similar to countries such as Italy and Greece than its neighbors.

Table 3 presents sample characteristics. Parents’ average age is 66 years and 60% of them are female. Approximately two-thirds of parents are currently married, and the rest are mostly widowed. Parents in this sample have on average 2.6 children, and they have 10.5 years of education. Annual standardized income in parental households is less than €22,000, while standardized gross financial wealth is over €39,000. Over two-thirds of parents reported some limitation in usual activities during the six months prior to the interview, but majority did not report any ADL or IADL difficulties. Their adult children are most likely to be married or partnered, but over one-fourth has never been married/partnered. About two-thirds of adult children work and have children of their own. Majority of children (70%) live within 25 kilometers from their parents.

Table 4 presents the results of the multilevel model of net transfers between parents and children. The results show that higher OASI-to-family spending ratio is associated with larger net transfers from parents to children: one point increase in the ratio is associated with an increase in net transfers of €181 using minimum wages to monetize non-financial transfers and €397 using average wages. This suggests that either parents provide increased support to adult children or children decrease their support to parents when public intergenerational redistribution of resources becomes relatively more favorable for parents than their children, and vice versa.

Consistent with the theoretical framework and descriptive results, net transfers decrease as parents (and children) age. Furthermore, estimated coefficients for other parental and children's characteristics in the model are congruent with the redistribution from those with relatively more resources and/or less need toward those with fewer resources and larger need. Net giving to adult children is larger for wealthier and better educated parents and smaller for parents in poor health as well as widowed or divorced parents. Presence of grandchildren is strongly positively associated with the net transfers from parents to adult children. Geographic proximity of children and parents is also associated with larger net transfers. Finally, children who suffer a loss of their partner through divorce, separation, or death seem to receive larger net support from their parents than married or partnered children, but this result is just below the standard significance thresholds and therefore tentative at best.

Given that the outcome measure in this analysis is a compound measure that requires determining financial value of various types of non-financial transfers, it is important to know if and to what extent different approaches to monetizing non-financial transfers affect the model results. As apparent from the multilevel model results (Table 4), monetizing time transfers using average hourly wage rates instead of minimum hourly wage rates generally leads to an increase in the magnitude of model estimates. However, the inference from the results obtained using the two alternative approaches of monetization does not change as both the directionality and the statistical significance of estimated coefficients remain consistent.

Furthermore, the SHARE does not collect the information on the intensity of practical help for coresident children, and there is no clear agreement in the literature whether coresidence primarily benefits parents or children and how exactly its benefits (such as implicit rent or shared meals) should be divided between them. For these reasons, coresident children were excluded from this study. However, since the rates of coresidence of adult children and parents vary from fairly modest in Northern Europe to very high in Southern Europe, it is possible that this constraint affects the findings. To explore this possibility, missing information on practical help for coresident children is imputed using the available data on time transfers between parents and children who lived in the same building, but in separate households, following the approach of Leopold and Raab (2011). Model results using these data reveal no substantive difference from the results obtained for the sample that includes only non-coresident children.

Finally, the estimated multilevel model implies random intercepts at parent/household and country levels. Statistical tests indeed confirm that the three-level model is superior to the models with fewer levels given the exact nested data structure.

In the wake of population aging, declining old-age support ratios, and related fiscal pressures, researchers and policymakers have started exploring the role of family support on the well-being of individuals and the effects of the interplay of public and family-support systems on the intergenerational redistributive effectiveness of public policies. This study contributes to the understanding of the public–family nexus of intergenerational support by exploring the association between public policies aimed at intergenerational redistribution of resources and the net-transfer value of financial and non-financial transfers from parents to adult children.

Findings provide support for the two research hypotheses. Consistent with the first hypothesis, multilevel model estimates suggest that children (parents) with greater needs, as indicated by their demographic, socioeconomic, and health profiles, benefit more from the exchange of support with their parents (children). Therefore, differences in relative needs of parents and children appear to be the key determinant of the direction and the magnitude of their net giving and this is congruent with the utility-optimization process between parents and children that was described in the theoretical framework section of the paper.

Net transfers model results also provide support for the second research hypothesis and the prediction of the theoretical model that public intergenerational redistribution of resources would be associated with a secondary redistribution from the generation benefitting from the public policy to other family generations. Growth in public support for older persons relative to the generation of their adult children (e.g. by increasing OASI benefits or decreasing family spending) is associated with an increase in net transfers that parents’ generation gives to their children's generation. This happens either because parents have more resources to help their children or because their children do not need to provide as much support to parents as they otherwise would. Conversely, public redistribution that benefits relatively more adult children than parents is associated with lower net parents–child transfers, consistent with either less giving from parents to adult children (e.g. no need for intensive grandchild care if the government provides appropriate childcare/daycare services) or increased children's provision of support to parents as a result of more resources that children have.

This study faces several limitations. The analysis relies on cross-sectional data that do not allow distinguishing between age and cohort effects and can capture only immediate reciprocation between parents and children. In the absence of longitudinal data that follow parents and children over their entire adult life, it is not possible to fully describe family-transfer behavior. Furthermore, the study focuses on exchanges between parents and non-coresident children. While this approach does not seem to substantively bias the inference, as evident from the sensitivity analysis, it limits generalizability to the population of parents and their non-coresident adult children. Also, the measure of net transfers is incomplete and partly arbitrary by design. First, it does not account for emotional transfers and coresidence between parents and children although they are an important component of family support (Temkin-Greener et al. 2004; Kohli and Albertini 2007). Unfortunately, the SHARE has very limited measures of emotional transfers, but even if it had a comprehensive list of emotional transfers’ measures, their monetization would be particularly challenging. On the other hand, coresidence information is collected, yet as previously explained, it is not clear who benefits from coresidence and by how much, which makes it exceedingly difficult to consider coresidence in the context of net exchange of support between parents and children. Second, the measure of monetary value of non-financial transfers does not account for the elasticity of demand for such transfers. Without this information, it is not possible to make a fully precise assessment of the economic value of time transfers between parents and children. Finally, self-reported transfer measures have been found to exhibit a systematic bias: survey respondents are likely to systematically over-report transfers they give and under-report transfers they receive (Brown and Weisbenner 2002; Mason et al. 2006). Consequently, the net transfers measure may be upward biased.¹⁴

Despite the limitations, this study makes important contributions to the research on the public–family nexus of intergenerational transfers. It highlights the importance of examining net transfers between family generations. Majority of prior research focuses on unidirectional transfers, which provides valuable information about the overall level of support provided between family generations, but it fails to capture their net redistribution of resources. Analyzing net family transfers, therefore, provides different, yet complementary information that allows estimating the impact of public policies on intergenerational redistribution of resources.

This study also empirically examines the association of some of the key elements of public spending affecting intergenerational redistribution, namely OASI and family spending, with the net value of transfers between parents and adult children, and establishes that public redistribution of resources between parents and children is associated with a secondary redistribution in the opposite direction at the family level. Overall, the results suggest that the need for support and the ability to provide it are important determinants of family-transfer behavior, providing further support to the prior literature that identified need to be among the most important operative motives of family giving (e.g. Kalmijn and Saraceno 2008). As more data become available, future studies can build on this foundation to identify with a higher degree of precision the magnitude of family intergenerational redistribution across the life cycle associated with variations in different public policies and to explore individual and family characteristics that may be modifying this relationship. Together with the findings from this study, such information would be essential for improving the evaluations of the effectiveness of policy intervention and the distributional impact of various public policies.

The author is grateful to the Editor and anonymous reviewers for their helpful comments and suggestions on earlier versions of this manuscript. This paper uses data from SHARE Wave 2 (doi:10.6103/SHARE.w2.500), see Börsch-Supan et al. (2013) for methodological details.

No potential conflict of interest was reported by the author.

Stipica Mudrazija is a Research Associate in the Income and Benefits Policy Center at the Urban Institute. His research examines issues related to population aging, retirement, health and long-term care for older adults, intergenerational support, social stratification, and comparative social policy.