Welfare within Families beyond Households: Intergenerational Exchanges of Practical and Financial Support in the UK

Families extend well beyond households. In particular, connections between parents and their adult offspring are often close and sustained, and transfers may include financial assistance, practical support, or both, provided by either generation to the other. Yet this major engine of welfare production, distribution, and redistribution has only recently become the focus of research. Who are the beneficiaries and to what extent are the patterns of exchange socially stratified? This article discusses findings from a programme of research analysing data from two nationally representative longitudinal studies, the British Household Panel Study and its successor Understanding Society, which record help given by, and received by, respondents through exchanges with their non-coresident parents and offspring in the UK. Some families exhibit a high tendency to provide mutual support between generations; these tendencies persist over time. Financial and practical support are generally complementary rather than substitutes. Longer travel time between parents and their offspring makes the provision of practical help less likely, whilst social class, social mobility, and ethnicity exhibit complex patterns of association with intergenerational exchanges. The resulting conclusion is that exchanges within families are an important complement to formal welfare institutions in the UK and that social policies should be designed to work with the grain of existing patterns of exchange, enabling family members to continue to provide help to one another, but ensuring that those who are less well supported by intergenerational assistance can access effective social protection.


Welfare within Families beyond Households
The UK is often classified as having a 'liberal' or 'individualistic' welfare regime, which relies on market mechanisms and a largely residual state safety net to meet people's needs and ensure an adequate standard of living [1][2]. This is said to contrast with the 'familialistic' welfare states of, for example, Spain, Italy, Japan, and Korea, in which there is greater reliance on families, including extended family networks, to provide financial support and to meet needs for housing and care throughout the life course [3][4][5]. However, it is recognised that just as some state supports are available in 'familialistic' regimes, so too are family exchanges in more 'individualistic' settings [6][7][8]. Indeed if we consider the five traditional pillars of the welfare state-education, healthcare, 'social services', 1 housing, and income support-it is apparent that families, everywhere, provide substantially for their members in all of these respects. Welfare in this broad sense is produced, distributed, and re-distributed within families.
The question is therefore not whether families play an important role in welfare, but rather what these family networks look like and the types and intensities of support that they provide. The focus in this article is on family ties that extend beyond the household, specifically exchanges between parents and their adult children who live separately. These intergenerational links are of particular interest because such vertical flows of support are important in understanding the transmission of advantage and disadvantage between generations and their possible relationship with social mobility. They are also relevant to debates about the 'burdens' imposed by an ageing population relative to the contributions they make to supporting their children and grandchildren [9,10]. The concentration in this article is on non-coresident parents and offspring because, in the UK, most adults of different generations live in separate households, although intergenerational households do also play an important, and under-recognised, role in supporting both older people and young adults, particularly among some ethnic minority populations [11,12].
Work by anthropologists and sociologists on family and kin networks has produced valuable insights [13], as have the large body of work by demographers on availability of kin, intergenerational exchanges, and provision of informal care [14][15][16], and the modelling of bequests and lifetime financial transfers by economists [17,18]. This study aims to build on these contributions by examining exchanges of both ' cash and care' and studying transfers both u[pwards] (from adult children to parents) and downwards (from parents to their offspring). Moreover, previous quantitative analysis has sometimes overlooked the complex structure of dyadic data; the programme of research on which this article draws has also contributed to developing a range of improved statistical tools to address these methodological challenges.
Hence the following questions: What is the extent and direction of intergenerational exchanges of ' cash and care'? Who benefits and in what ways are the exchanges socially stratified? And what does this imply for ways in which the welfare state and social policies should intersect with within-family intergenerational informal welfare? The next section introduces the data and definitions employed; the results are then discussed under the themes, further elaborated below, of mutuality, complementarity, persistence, and stratification; and the final section reviews the implications of these findings for public policy.

Data and Definitions
The analysis reported here derives from a number of projects carried out as part of the Dyadic Longitudinal Analysis of Intergenerational Exchanges programme (DyLAnIE) led by Fiona Steele. All the projects used the British Household Panel Study (BHPS) for 2001 to 2006 and/or its successor, the UK Household Longitudinal Study-UKHLS (also known as Understanding Society), for 2011-2013 to 2017-2019. Both are longitudinal, nationally representative surveys of private households in the UK. Analysis for DyLAnIE focuses on the family network module, which was fielded in selected years. Respondents (aged 16 or over) were asked whether they had one or more living non-coresident parent, the ages of these parents, how far away they lived (measured by the travel time), and a sequence of questions about the help they 'regularly or frequently' give to, or receive from, their parent(s). Respondents were also asked whether they had one or more child who did not reside with them. The focus in this article is on adult children (the terms ' offspring' and ' adult children' are used interchangeably) rather than dependent children living with another parent or carer. Again, respondents with a non-coresident adult child were asked a sequence of questions about any exchanges of help with them. Further details, including some significant limitations of these data, are given in the Appendix.
The nine types of help included in the module are shown in Table 1, together with the percentage of adults with non-coresident parents or offspring who are providing/receiving that type of help, using the most recent UKHLS wave (2017-2019). Help with personal needs is asked only of offspring about help given to parents, and of parents about help received from offspring. Conversely, help with childcare is asked only of parents about help given to offspring, and of offspring about help received from parents.  Table 1 shows that substantial proportions of those with non-coresident relatives are giving or receiving some kind of help. Nearly three-fifths (59%) of parents report giving help to any of their non-coresident children, and 40% of children acknowledge receiving help from a non-coresident parent (columns 3 and 2 respectively). This difference is likely to reflect a degree of reporting bias-we are more likely to want to acknowledge help that we give than help we receive-but also the fact that the two groups of respondents are not matched pairs and are not exact mirror images of one another, since ' child' respondents are asked about up to two non-coresident parents collectively, whilst 'parent' respondents are asked about all of their non-coresident children collectively. Turning to columns 1 and 4, two-fifths (43%) of adult children report giving help to their non-coresident parents, and slightly fewer parents-but still over onethird them (36%)-report receiving help from their non-coresident children.
In subsequent analysis, practical help-a broad category, in which the first eight types of assistance are included-is distinguished from financial help. Because the question about financial help is framed in terms of 'regular or frequent' help, it is likely to capture relatively small sums (for example, help with bills), rather than occasional, potentially larger, exchanges (for example, help towards the deposit for a house, or paying for care). Other work has investigated the scale and distributional effects of larger and broader types of inter-vivos transfers [17,18]. As far as regular financial assistance is concerned, Table 1 indicates it is much more likely to flow from parents to children than vice versa.

Mutuality and complementarity
Are those who are giving help also receiving it? Kuha et al. [19] model the tendency to give help to parents and the tendency to receive help from parents as continuous latent variables (that is, as ' sliding scales' of helpfulness), based on the eight different types of help. In addition, this study takes account of the fact that a substantial proportion of each group do not give, or receive, any help by specifying binary latent variables (that is, an ' on/off') for being a 'giver' or a 'receiver' of help at all. The specific types of help that men and women tend to give vary, as do the types of help given by people living nearby and living further away. For example, if they are helping at all, men are more likely to help with decorating, gardening, or house repairs than are women. Therefore giving this kind of help is a stronger signal of underlying helpfulness if observed in a woman than in a man. These variations in the strength of the signal of the underlying tendency to help provided by specific types of help given or received is recognised in the model. This approach is superior to measuring help only as a binary variable (any help vs no help) or as a count of number of types of help provided, which implicitly gives all types of help an equal weight.
Findings indicate that giving help to non-coresident parents and receiving help from them are fairly strongly associated, even when the giving and the receiving are measured at the same point in time. 2 Based on the ' on/off' part of the estimate of helping tendencies, the odds of receiving help from your parents are about 7 times higher if you are also giving help to your parents. Based on the 'sliding scale' part of the estimate, the marginal correlation between giving and receiving is positive and fairly large (+0.23), and increases if the age of the parent is taken into account (to +0.50). Such a degree of reciprocity between generations belies the simple notion that help flows either downwards or upwards.
On the other hand, the flipside of strong association between giving and receiving tendencies is that some respondents-both children and parents-will be neither giving nor receiving any help. This has significant implications for public policy, which are developed further in Section 4.
Reciprocity in giving and receiving between parents and their offspring is an important part of the story. But there are also differences in tendencies to give and receive that are related to the needs and capacities of the two parties. For example, parents who are older or living alone are more likely to be receiving help from their offspring, whilst offspring who are single and who do not have young children are more likely to be giving help. This does not look like straight forward reciprocity ('I give to you if you give to me') but rather a relationship of mutual recognition ('I recognise your needs and you recognise mine').
In further analysis, financial and practical help were examined separately [20,21]. 3 Previous studies have suggested some substitution between types of help provision: If a child is not able to provide physical assistance to a parent, she or he might pay for the parent to receive a service instead (such as gardening), or, conversely, a child with limited financial resources might nevertheless be able to provide some practical help [22]. The same logic also applies from the parental perspective. One might therefore expect a negative correlation between giving practical help and giving financial help. However, the results suggest the reverse. Children giving practical help to parents is positively correlated with giving financial help to parents (+0.48), as is receiving practical and financial help from parents (+0.47). These correlations remain strong even after allowing for differences in age, household income, and travel time between respondents and parents (+0.48 and +0.42 respectively). 4 Within families across households, it seems that practical and financial help are treated as complementary rather than as substitutes.
The likelihood of providing (or receiving) practical and financial help is, however, lower when parents and their offspring live further apart (Figure 1) and although the correlation between practical and financial help remains strong, the balance between them shifts [20]. Offspring who live less than 15 minutes away from their parents are more than three times as likely to provide regular practical help to their parents, and to receive help from them, than offspring who live more than two hours away. 5 Admittedly, the association may not be strictly causal: A third factor, such as emotional closeness, might give rise to both a preference for living near to one another, and a tendency to provide practical help, rather than the geographical proximity being the cause of helping per se. Additionally there are associations between social class, social mobility, and geographic mobility, explored further in section 3.3 below. But the strength of the association between distance and tendency to give or receive is large in comparison to other characteristics. For example, the marginal probability of providing practical help differs between men and women only 3 to 5 percentage points, and an increase in age of parent from 60 to 90 only alters the predicted probability of the offspring providing practical help by 17 percentage points. These compare to a gap of 42 percentage points in the predicted probability of providing practical help to parents between a child living less than 15 minutes away compared to living more than two hours away, and a gap of 33 percentage points in the predicted probability of their receiving practical help.
It is interesting that the provision and receipt of financial help also becomes less likely with greater geographical separation, despite there being no impediment to transferring money at a distance. However, the decline is much less steep than is the case for practical help, and does not reach statistical significance at conventional levels. On the other hand, there is certainly no evidence for increasing provision (or receipt) of financial help when parents and offspring live further apart, so again it does not appear that financial help is being used as a substitute for practical help. There is just less helping altogether when travel times are greater.
Overall, these sets of results confirm a high degree of mutual assistance within some families, with practical and financial help provided both from children to parents and from parents to children. Conversely, there are some families in which mutual aid is not the norm, and this is associated, among other things, with greater geographical separation.

Persistence
So far, this article has reported results on exchanges observed at a point in time. Given changes in needs and capacities over the life-course, one might expect the association between receiving and giving help over time to be even stronger. Offspring may be more likely to need support as they first make a transition to independence, with their (young) parents potentially in a good position to provide it, and parents may be more likely to need support as they age, with their (now older) offspring possibly in a better circumstances to supply it. The persistence of tendencies to give and receive practical and/or financial support are examined by observing the same people at five points in time in over a 14-year   [23]. 6 Controlling for other characteristics (gender, age, age of any children, partnership and employment status, travel time to nearest parent), the correlation between giving help to parents at the previous time point (two to five years previously) and giving help now is strong and positive (+0.68), as the equivalent correlation for receiving help from parents is (+0.55). 7 These lagged variables (i.e. previous help given or received) have stronger associations with giving or receiving help now than any of the other characteristics included in the models-other than the presence of a dependent child in the household which is associated with a high likelihood of receiving help from parents. So this suggests that 'helpers' tend to continue to help and 'recipients' continue to receive over time. Moreover, the random effect and residual correlations between giving and receiving help to and from parents are positive and moderately large, suggesting that offspring with a higher tendency to give help are also more likely to receive help, either contemporaneously or at a different point in time. Specifically, the random effect correlation which captures characteristics like the emotional bond between parents and offspring or their mutual expectations (technically, unobserved fixed characteristics) is +0.35, and the residual correlation which captures the 'inherent' association between giving and receiving at a point in time not otherwise explained by the characteristics included in the model that vary over time is +0.47.
In general, changes in respondents' circumstances (partnership transitions, the arrival of a child, employment transitions) are not significantly associated with increases or decreases in the likelihood of giving help to parents, although in some cases they are associated with changes in the likelihood of receiving help from parents. People who have suffered a recent partnership dissolution are more likely to receive help, as are those who have a youngest child under the age of 2.
Persistence in giving help to parents is also apparent in Figure 2. If respondents are grouped into ' cohorts' according to their age in 2001, and then followed as they get older over the next 16 years, the percentage of each cohort giving help to parents is found to be quite steady. For example, the percentage of respondents aged 18-29 at the start (blue line) who are giving help ranges around 35%, while among respondents aged 50-69 at the start (grey line), it ranges around 65%. However, it is also noticeable that successive cohorts have become less likely to provide help: At average age 55, 62% of the oldest cohort were providing help to their parents, but by the time the middle cohort (orange line) had reached that same average age, only 49% of them were providing help-the orange line is below the grey line. Similarly, at average age 39, 43% of the middle cohort were providing help, but by the time the youngest cohort had reached that 6    same age, only 34% of them were providing help: The blue line is below the orange line. If this pattern is sustained, future generations of parents can expect considerably less help from their offspring than current generations enjoy. By contrast, there are no major cohort effects in receiving help from parents (Figure 3): Successive cohorts are as likely to be receiving help (at equivalent ages) as were their predecessors. There is however a clear age gradient-the older you are, the less likely you are to be receiving help from your parents.
These findings together can be interpreted as suggesting that tendencies within families to give and receive help intergenerationally persist over time. The balance between giving and receiving shifts with age and can be affected by changes in the needs or capacities of either party, although generally only slightly. Generations born more recently appear to be less inclined to provide practical help to their parents than generations that came before.

Stratification
Do patterns of exchange within families across generations vary by social stratifications such as ethnicity and social class? Table 2 summarises the predicted probabilities of giving and receiving practical and financial help to and from parents by broad ethnic classification of respondent. 8 In columns 3 and 4 one can see that there is relatively little variation between ethnic groups in receipt of practical and financial help from parents. Asian/Asian British offspring seem to be the least likely to be receive help, and Black/Black British the most. 9 However, considering help provided to parents (columns 1 and 2), the variation is much larger. Controlling for other characteristics, Asian and Asian British offspring are three times more likely to be giving financial help to their parents compared to their White counterparts (15 per cent compared to 5 per cent), and they are also substantially more likely to be giving practical help (56 per cent compared to 42 per cent). Black and Black British offspring are also much more likely than White offspring to be giving financial help to their parents, and to be giving practical help.
These patterns merit further investigation, but considering the evidence presented here, it appears that there is a stronger net upwards flow, i.e. from offspring to parents, within Asian/Asian British families, than there is for the other broad ethnic groups . Asian pensioners have higher rates of poverty relative to Asian people of working age than is the case for White or Black families [25], so the stronger net upwards flow within Asian/Asian British families is consistent with the idea that giving flows from those with capacity to those with relative need. However it also seems likely that there are social and cultural norms in play [26].
In respect of social class, Table 3 shows that respondents classified as being in the managerial or professional social class behave somewhat differently to those in intermediate or lower social classes [27]. They are less likely to provide 8 Calculated from probit regression models, with a wide set of covariates including respondent's household income [15 Table 4]. 9 It is important to keep in mind here that our definition excludes help provided in the form of co-residence, which also varies by ethnicity [24]. practical help to their parents (41 per cent compared to 48 per cent for the other two classes) but more likely to provide financial help (7 per cent compared to 5 per cent for the other two classes). At the same time, they are less likely than respondents in the other social classes to receive either practical or financial help from their parents-marginally less likely in the case of practical help (36 per cent compared to 38 per cent), and substantially in the case of financial help (10 per cent compared to 13 per cent for intermediate social class, and 16 per cent for lower social class). Altogether, respondents in the least advantaged social class are three times as likely to be receiving financial help from their parents than they are to be giving financial help to their parents, whilst for respondents in the managerial and professional classes, this ratio is only one-and-a-half. On the other hand, these same respondents in the least advantaged social class are also much more likely to be 'net givers' of practical help to their parents than their higher social class counterparts. It should be noted that these percentages do not adjust for covariates, and it is likely that some of the differences are due to differences in the financial resources of respondents in different social classes, and in particular their resources relative to those of their parents, as well as other factors such as geographical distance.
It has already been shown that offspring who move further away from their parents are less likely to provide or receive practical help, and that there is little evidence of substitution of financial help for practical help at a distance. In relation to social, rather than geographical, mobility, 10 however, greater ' distance' between parents and offspring does not have the same dampening effect: On the contrary, upwardly mobile offspring are more likely than their immobile counterparts to provide financial help to parents. Similarly, downwardly mobile offspring are more likely than the immobile to receive financial help from parents. Interestingly, this pattern is repeated for practical help, although not for 'horizontally mobile' offspring (i.e. those who move within the intermediate class), who provide help at a similar rate to the upwardly mobile but receive it at a similar rate to the downwardly mobile. 10 Social mobility is ascertained by comparing the NS-SEC 5 classification of the current or most recent occupation of the respondent to the classification of the parent's occupation when the respondent was aged 14 (as reported by the respondent retrospectively). If the social class of both parents is available, the higher of the two is used. This is a measure of absolute social mobility rather than mobility relative to a cohort. The sample is restricted to respondents aged 30 to 60, a period of life in which occupational social class is generally well-established. Note: Analysis sample is UKHLS waves 3, 5, 7 and 9, respondents with at least one non-coresident parent in the UK and not living with a parent. Average predicted probabilities for each person-wave observation based on estimates from random effects multivariate probit models, for each ethnic classification in turn, with other covariates set at observed values. Other covariates as for Figure 1.
Sample sizes are number of person-wave observations. Source: Authors' calculations. These patterns of exchange are consistent with the idea that within families, the party with relatively greater resources tends to help the party with relatively fewer resources-children who have advanced beyond their parents' social class being more likely to provide financial and practical help, and children who have fallen on hard times being more likely to receive financial and practical help. The former is also compatible with an explanation based on lagged reciprocity: Possibly, the parents of upwardly mobile offspring have sacrificed their own living standards in order to promote their children's ascent, and the offspring feel a sense of obligation to repay that investment. In either case, there is no suggestion that increased social distance alienates the generations from one another in the same way that geographical distance appears to be associated with lower rates of exchange.

Welfare Policies Working with the Grain of Intergenerational Exchanges
Despite the UK's classification as a liberal, rather than a familialistic, welfare state, the evidence presented here suggests that routine support of both a practical and financial nature within families across households is widespread. While this study considers only one type of relationship-between parents and their adult children-it is likely that other members of an extended family such as siblings, cousins, aunts and uncles are also providing help. The tendency to help persists over time for a given individual, but successive cohorts appear to be becoming less inclined to provide help to their parents. The mutuality of help within families is striking: People who are receiving help from their parents are more likely than others to be providing help as well, both at the same time, and over the longer term. By the same token, people who are not receiving help from their surviving parents are less likely to be providing help: Some families have a high tendency towards mutual help and some families do not.
This leads to a first important implication: That the welfare state should recognise mutual help where it exists, without assuming that all families are, or should be required to be, in a position to provide it. It is undesirable for state provision of cash or services to supplant or displace mutual support provided within families, but it is also undesirable for state provision to leave those who do not benefit from such mutual support exposed to significant risks and unmet need. This is not straightforward to achieve. On the one hand, tax and benefit systems (in the UK) do not generally recognise the existence of the family beyond the immediate household, never mind record whether they are engaged in mutually beneficial exchanges. Moreover, the incorporation of such information into our current social security system would be likely to result in perverse effects because it is so heavily reliant on means-testing: Families who are supporting one another across households would be penalised through a reduction in their benefit entitlement. On the other hand, adult social care assessments do take into account whether the person with care needs has a family member living with them or nearby, but the result is all too often a heavy burden being placed on the family carer with little discretion.
Instead, the focus may be better directed towards supporting mutual help where it is provided, for example with benefits and services for carers, and exempting modest within-family cash transfers from means-tests. At the same time, we need to ensure that the minimum guarantee for those who are not able to access intergenerational familial support is comprehensive and at an adequate level-including for older people who do not have any surviving children, and young people who do not have any surviving parents (not included in the analysis, which is restricted to those with at least one non-coresident parent/child). Young adults who are not in close contact with their parents-importantly this would include most children who have been in local authority care-are also potentially in need.
Broadly speaking, the results in this study are consistent with an interpretation that where more help is provided in one direction than in another it is because the recipient has relatively higher needs and the donor has relatively higher capacity. Unfortunately, this study cannot model this directly, because there is limited information about the member of the pair who is not a survey respondent. However some nuances can be identified, to which supportive social policy should be sensitive. For example, after controlling for income, employment status, and a range of other characteristics, the estimates suggest that Asian and Asian British offspring are much more likely than other ethnic groups to be providing help to their parents, and are somewhat less likely than other groups to be receiving help from their parents. Conversely, offspring from a lower social class background are more likely than other social class groups to be receiving help from their parents, particularly those who are downwardly socially mobile.
A second key implication of these findings is that it is mistaken to think of financial and practical help as substitutes for one another, in general. Although there are sound theoretical grounds for the assumption that support in cash and support in kind could be used to achieve the same ends, this does not appear to be how families in the UK operate in practice. One caveat here is that major one-off financial transfers were not observed nor was the biggest type of investment in practical support included, namely, intergenerational co-residence. But as far as routine support is concerned, parents who are providing practical support are also more likely to be providing financial support, and the same is true for their offspring. This is particularly noteworthy when one examines the effect of geographical distance between the generations: The understandable diminution in the likelihood, for either party, of providing practical help when there is greater travel time between them is not accompanied by an increased propensity to offer financial help. Families at a larger geographical distance from one another appear to be less close, on average, in other respects as well.
This leads to a third implication: That public policies which disrupt family networks through obliging offspring to live further away from their parents potentially carry a heavy, and largely hidden, cost. Causation cannot be attributed definitively-children who are already less inclined to give or receive help may be more likely to move away-but we do know that for at least some families, geographical distance occurs through force of circumstance rather than through choice, and this may well lead to a reduction in mutual support. Local authorities in areas with high housing costs, such as London, frequently place homeless households in accommodation out-of-area, despite the National Homelessness Advice Service warning that, "Being close to friends and family can be crucial to basic, everyday support, and forms a key basis for enabling the household to settle and sustain accommodation in the medium term [28 p.7]." More widely, housing regeneration schemes which 'decant' existing tenants into alternative accommodation and issue compulsory purchase orders to leaseholders often have the effect of dispersing extended families. This has been challenged in planning inquiries and judged, on at least one occasion, to threaten the residents' right to respect for private and family life [29]. This is not a new insight-Young and Willmott's classic study of slum clearance came to the same conclusion in 1957 [30]-but it seems to be a lesson that policymakers still need to learn. More generally, regional housing and economic planning (or the lack of it) which results in young people being obliged to move to other parts of the country in order to find affordable housing or to get work, creates distance that may make it harder for families to sustain routine intergenerational support, in both directions. Geographical mobility has its drawbacks.
For too long, public policy has treated 'the family' as synonymous with 'the household', yet the links within families across generations and households are strong, and, in many cases, mutually beneficial. It is high time the relevance of this aspect of the sociology of the family was given greater recognition in designing social protection and housing policies, to ensure that they work with the grain of these intergenerational exchanges rather than cutting across them.