How to do this Annotated Bibliography on Multi-generational Families

Growing Parental Economic Power in Parent–Adult Child Households: Coresidence and Financial Dependency in the United States, 1960–2010

Joan R. Kahn & Frances Goldscheider & Javier García-Manglano

Published online: 21 February 2013 # Population Association of America 2013

Abstract Research on coresidence between parents and their adult children in the United States has challenged the myth that elders are the primary beneficiaries, instead showing that intergenerationally extended households generally benefit the younger generation more than their parents. Nevertheless, the economic fortunes of those at the older and younger ends of the adult life course have shifted in the second half of the twentieth century, with increasing financial well-being among older adults and greater financial strain among younger adults. This article uses U.S. census and American Community Survey (ACS) data to examine the extent to which changes in generational financial well-being over the late twentieth and early twenty-first cen- turies have been reflected in the likelihood of coresidence and financial dependency in parent–adult child U.S. households between 1960 and 2010. We find that younger adults have become more financially dependent on their parents and that while older adults have become more financially independent of their adult children, they never- theless coreside with their needy adult children. We also find that the effect of economic considerations in decisions about coresidence became increasingly salient for younger adults, but decreasingly so for older adults.

Keywords Living arrangements . Intergenerational coresidence . Multigenerational households . Financial dependency

Demography (2013) 50:1449–1475 DOI 10.1007/s13524-013-0196-2

Electronic supplementary material The online version of this article (doi:10.1007/s13524-013-0196-2) contains supplementary material, which is available to authorized users.

J. R. Kahn (*) : J. García-Manglano Department of Sociology and Maryland Population Research Center, University of Maryland, College Park, MD 20742, USA e-mail: [email protected]

F. Goldscheider Departments of Sociology and Family Science, and Maryland Population Research Center, University of Maryland, College Park, MD 20742, USA

Introduction

Research on coresidence between parents and their adult children has increas- ingly challenged the long-held view that elders benefit most from this type of arrangement. Whether they focus on relatively contemporary households (Choi 2003; Kotlikoff and Morris 1990; Speare and Avery 1993; Ward et al. 1992) or on a broad sweep of more than a century (Ruggles 2007), studies have shown that intergenerationally extended households benefit the younger generation disproportionately. Such studies imply not only that the normal structure of parent–adult child households includes dependent adult children but also that there has been little if any change over time in this structure.

We expect to see changes in recent decades because of important changes in the factors affecting intergenerational coresidence for both generations. In addition to the improvements in health among older adults that have increased their ability to live independently, there have been substantial increases in their financial well-being, with the implementation and growth of Social Security in the United States and spread of private pensions (McGarry and Schoeni 2000), although this trend has slowed or even reversed in recent decades (O’Rand et al. 2009). Young adults in the US have experienced increases in education, and delays in marriage and childbearing, as well as high rates of union disruption which have led to large numbers of young adults who are unmarried and at increased risk of living with their parents (Furstenberg et al. 2004). Analyses of intergenerational households have not accounted for these changes.

The increasing financial well-being among older adults has not been matched, however, among the younger generation. Young adults have in fact experienced greater financial strain (Levy 1999; Preston 1984). Hence, the younger generation is likely to have become more financially dependent on their parents’ generation in recent decades, making it important to examine change during this period. Further, this pattern of increased financial independence among the elderly might have reversed, as the cohorts entering late middle age have suffered financial reversals during the Great Recession (Gustman et al. 2010).

Nevertheless, in many cases, parents are indeed dependent on their adult children. Thus, the question remains, Has the balance shifted such that many fewer parents need to coreside while the opposite is the case for young adults? The question of recent changes in coresidence patterns has not been systematically addressed in previous studies and highlights the importance of considering changes in the resour- ces of both generations when attempting to understand intergenerational coresidence.

This article examines the extent to which changes in financial well-being over the past half-century (between 1960 and 2010) have been reflected in both the likelihood of intergenerational coresidence as well as in the relative economic dependency of both generations in parent–adult child households. Using census and American Community Survey (ACS) data, we examine whether the economic balance of power in intergenerational households has changed—and if so, how. We examine change in the determinants of living in an intergenerationally extended household from the perspective of both the older parent generation and the younger adult-child genera- tion. Although not an ideal data source for studying the reasons for coresidence or the duration of such arrangements, the census and ACS data allow us to learn a great deal

1450 J.R. Kahn et al.

about changes in both the characteristics of who coresides as well as who is support- ing whom in coresidential households. Specifically, we examine the changing effects of the key socioeconomic characteristics of education, employment status, and income on the likelihood of living with either adult children or older parents. Moreover, we compare the economic resources of each generation within multigen- erational households to determine whether indeed the balance of financial power and dependency within intergenerational households has changed over time. Although we recognize that there are many noneconomic factors that encourage intergenerational coresidence, including health and companionship needs, our focus is on trends in how financial needs shape decisions to coreside.

Background

The shift in living arrangement patterns in the United States toward greater residential independence in the period since World War II is well documented (see, e.g., Costa 1999; McGarry and Schoeni 2000; Santi 1990; Schoeni 1997). Adults of all ages are increasingly living in simple households, either in two-adult, married-couple house- holds or in one-adult households, with children or alone (Kobrin 1976). The percent- age of elderly widows living alone rose from 18 % in 1940 to 62 % in 1990 (McGarry and Schoeni 2000).

These changes in living arrangements have been linked to demographic, economic, and normative changes. Demographically, mortality declines have resulted in a growing number of persons surviving into the later years of the life course, with surviving parents and children, thereby increasing the availability of relatives with whom to live (Schoeni 1997). However, the concomitant increase in good health of older persons and the availability of home-based services (Krivo and Mutchler 1989) have also increased their option of living independently and caring for their own needs.

In addition to changing demographic factors, we know from McGarry and Schoeni (2000) that between 1940 and 1990, expanding Social Security benefits and private pensions made independent living possible for many older persons. Lifestyle and normative changes may have reinforced these demographic and economic patterns (Alwin et al. 1985; Pampel 1983) by increasing the priority given to privacy, independence, and age-segregated leisure-time activities (Goldscheider and Goldscheider 1987). Few older persons want or expect to become dependent on their family, residentially or otherwise (Burch and Mathews 1987; Lopata 1973).

It is increasingly clear, however, that complex households reflect the needs not just of older people but also of younger adults. Ruggles (2007) showed that the decline in intergenerational coresidence between 1850 and 2000 was due primarily to increasing opportunities for the young and declining parental control over their children, rather than the rising economic independence of the older generation. In a study of the early 1980s, Speare and Avery (1993) also found that intergenerational coresidence depended more on the economic needs of the younger generation than on those of the aged. Hence, the fact that the economic position of young adults has been declining since the 1970s and 1980s (Easterlin 1978; Levy 1999) means that although it has become increasingly feasible for older persons to be independent and to purchase privacy, their resources have become more important to their adult

Growing Parental Economic Power in Parent–Adult Child Households 1451

children. Hence, our first hypothesis is that between 1960 and 2010, younger adults living in parent–adult child households became increasingly dependent, financially, on their parents.

In addition to expecting greater dependency among the young, our second hy- pothesis is that over time, economic resources will play an increasing role in the decision about whether to form a multigenerational household. This angle on family extension has been even less researched. Although there is substantial research on the determinants of intergenerational coresidence, relatively little has focused on how the determinants may have changed. Most studies focus on a single period of time (e.g., Glick and Van Hook 2002; Mutchler and Burr 2003; Schmertmann et al. 2000; White 1994). Those who have taken advantage of the long historical sweep made possible by the Integrated Public Use Microdata Series (IPUMS) (Ruggles et al. 2010) have taken the broadest possible view, so the challenges of measurement comparability overwhelm behavioral change (e.g., Ruggles 2007).

An important pair of studies, however, focused on the question of whether the effect of income for elderly widows has intensified, with Costa (1999) arguing that it has, and McGarry and Schoeni (2000) disagreeing. This is an important theoretical issue: the increasing importance of income for decisions about coresidence suggests that such a living arrangement may have become an undesired default for those unable to purchase their desired privacy, and also that the value of privacy may have increased relative to companionship and mutual exchange. By focusing on elderly widows, neither Costa (1999) nor McGarry and Schoeni (2000) considered changes in the characteristics of adult children or changes over time in the effects of each generation’s economic resources on the likelihood of coresidence. Without account- ing for adult children’s characteristics, it is difficult to know whether economic resources have become more important determinants of coresidence over time.

Given coresidence, however, the research on which generation is more likely to benefit (e.g., Choi 2003; Cohen and Casper 2002; Speare and Avery 1993) has established that it is the younger generation, not the older, that typically benefits. These studies have generally found, not surprisingly, that those experiencing financial difficulties are more likely to be financially dependent in an intergenerational house- hold, and that those at the youngest and oldest ages are also more likely to be dependent, as are the unmarried and, more surprisingly, sons. This research, however, has not focused on how the factors affecting financial dependency in intergenera- tional households might have changed over time. Thus, our analysis of this issue is exploratory; we have no expectations on how the effects of resources, age, marital status, or gender might have changed as predictors of experiencing financial depen- dency in such households.

In this analysis, we address these intergenerational issues by examining the determinants of coresidence from the perspectives of both older and younger gen- erations in order to assess whether, on the basis of changes in their characteristics, the young have become increasingly “needy” relative to older generations (our first question). Then we consider changes over time in factors affecting the likelihood that younger adults will live with their parents, or that older adults will live with their adult children or other younger relatives. This analysis provides valuable information on baseline trends in coresidence patterns across the adult life course during a period of rapid social, demographic, and economic change, and addresses our second

1452 J.R. Kahn et al.

question directly: have resources become more important in determining the likeli- hood of residing in a parent–adult child household? We then address the question of “who supports whom” by examining the actual resources of each coresiding gener- ation in order to better determine the direction of the flow of support within house- holds. We expect that younger adults will be more financially dependent on their parents and older adults less financially dependent on their adult children in 2010 relative to 1960, but that the trends in the effects of financial resources might have changed differentially between the generations.

Data and Measures

Our analysis of the determinants of intergenerational living arrangements and the relative financial position of the generations begins with an analysis of change between 1960 and 2010 based on U.S. census and ACS data. These data provide the best view available of long-term change, although with limited measures. The 1960 census is the earliest to provide detailed information on income, education, and employment on all members of the household, and the 2010 ACS is the most recent national survey to obtain this information.

Data

We use decennial data from the IPUMS (Ruggles et al. 2010), which provide nationally representative 1 % samples of households in the U.S. census between 1960 and 2000, and from the ACS (2010). Although both sets of data are subject to minor levels of undercount (Robinson 1988; U.S. Census Bureau 2001), they are far more representative than the sample survey data that constitute the basis for much recent research on parent–child relationships. For all years, we use the self-weighted subsamples generated for IPUMS users.

Given that our interest focuses on intergenerational coresidence among adult relatives, our working sample includes only individuals 25 years of age or older, living in households. This is the internationally recommended population to study for these questions, primarily because in most cases, these young adults have completed the nest-leaving process, at least insofar as it is connected with continuing education (Pew Social and Demographic Trends 2010; United Nations 2005).

To determine coresidential status, we first classify all individuals into gen- erations according to their relationship to the householder1: (1) grandparents and grandparents-in-law; (2) parents, parents-in-law, uncles, and aunts; (3) spouses, siblings, and relatives of similar age (defined here as no more than 15 years older or younger than the householder); (4) children, children-in-law, nephews, and nieces; (5) grandchildren; and (6) other (including nonrelatives). Our classification of multiple-generation households is similar to the existing MULTGEN variable in IPUMS, except that we only include adults aged 25 and

1 The U.S. Census Bureau’s change in 1980 from “head of household” to the less sexist “householder” term has no effect on our definition of multigenerational households because our determination is based on comparing the relationships of all household members with the designated householder.

Growing Parental Economic Power in Parent–Adult Child Households 1453

older.2 Hence, all households containing only parents and their children younger than age 25 in the original sample are here classified as one-generation households.

Next, we build our main dependent variables by assigning a multigenerational status to each adult in the sample. Those in “one-generation households” were living alone or with a spouse and/or a child younger than 25, a sibling, another relative who is no more than 15 years older or younger than themselves, or a nonrelative, but not with any other related adults aged 25 or older. In households with multiple adult generations, all individuals are assigned into one of the following categories: “multi- generation with parents,” if they were living with one or more older related adults (98 % of whom were parents or parents-in-law) and “multigeneration with adult child,” if they were living with one or more younger related adults (99 % of whom were adult children).

We use two separate approaches to assigning multigenerational status: one for householders and one for all other household members aged 25 or older. For the multigenerational status of individuals who are not the householder, multigenerational status is simply based on their relationship to the householder. However, because householders have relationship codes with every member of the household, we create a hierarchy of relationships in order to determine the householder’s multigenerational status. Most cases are coded unambiguously because the householders either did not live with an older or a younger adult relative (and are therefore coded as living in a one-generation household), or they lived with a member of only one other generation (older or younger). In the rare event (less than 1 % of adults) that a householder lived with both older and younger adult relatives, our hierarchy gives priority to older adults; hence we coded the householder as living with parents.

Table 1 shows the distribution on household generational status for all U.S. adults aged 25 and older for each decade between 1960 and 2010. We divide the sample into three age groups: young adults (aged 25–44), middle-aged adults (aged 45–64), and older adults (aged 65 and older). For these ages and years, we show the proportions living in one- versus multigeneration households. Figure 1 summarizes the trends.

Throughout the 50-year period and regardless of age, most adults lived in one- generation households. Whereas in 1960, 12.4 % of young adults, 16.6 % of middle- aged adults, and 26.7 % of older adults lived with another adult generation, by 2010, between 15 % and 20 % of all three groups lived in multigenerational households (Fig. 1). The steep decline in coresidence for the elderly during the 1960s and 1970s is noteworthy, as is the steady increase after 1980 for young adults; in fact, however, all three age groups saw increases after 1980, which is consistent with earlier findings showing an increase between 1980 and 1990 (Goldscheider et al. 1994).

As one might expect, when we distinguish by whether the individual is living with an adult child or older parent, the likelihood of living in one or the other type of multigenerational household differs substantially by age. Table 1 shows that among those who lived with relatives of a different generation, young adults were much more likely to live with a parent rather than an adult child (11.5 % vs. 0.5 %

2 Because our focus is on the relative resources of adults (aged 25 or older) living in multigenerational households, we do not consider other extended household forms (e.g., adult siblings who live together) even though they may be important in groups such as recent immigrants. For the same reason, we also do not control for the presence of dependent children, even though they may influence both the need for and desirability of coresidence.

1454 J.R. Kahn et al.

Table 1 Household generational status of U.S. adults, by age and census year (1960–2010): Adults aged 25 and older

1960 1970 1980 1990 2000 2010

Age 25–44

N 447,351 455,701 591,384 728,269 749,965 658,397

Generational statusa

One generation 87.6 90.1 90.9 88.2 87.6 82.1

Two generations with parent 11.5 9.2 8.6 11.1 11.6 16.9

Two generations with adult child 0.5 0.6 0.3 0.4 0.3 0.3

% 100.0 100.0 100.0 100.0 100.0 100.0

Age 45–64

N 347,134 402,481 432,730 462,314 587,994 826,710

Generational statusa

One generation 83.4 85.8 84.8 81.3 83.7 82.1

Two generations with parent 7.1 6.3 5.0 4.3 4.9 6.1

Two generations with adult child 9.0 7.6 9.9 14.1 11.1 11.2

% 100.0 100.0 100.0 100.0 100.0 100.0

Age 65 and Older

N 147,367 186,408 234,594 307,631 326,961 460,904

Generational statusa

One generation 73.3 80.5 84.5 84.2 81.8 80.8

Two generations with parent 0.9 1.3 1.1 0.8 0.8 0.9

Two generations with adult child 25.2 17.9 14.2 14.6 17.1 18.1

% 100.0 100.0 100.0 100.0 100.0 100.0

a Generational status is determined for each adult based on his or her relationship with other adults in the same household, regardless of the presence of children younger than age 25. A person is classified as living in a one generation household if s/he lives alone, with a spouse or a sibling, but not with any adult children or own parents or parents-in-law. A person is classified as living in a two generations with parent household if s/he lives with at least one member of an older generation (e.g., a parent, parent-in-law, or grandparent). A person is classified as living in a two generations with adult child household if s/he lives with at least one member of a younger adult generation (e.g., an adult child or adult grandchild).

Fig. 1 Trends in intergenerational coresidence by age, 1960–2010

Growing Parental Economic Power in Parent–Adult Child Households 1455

in 1960; 16.9 % vs. 0.3 % in 2010), given that few were old enough to have an adult child. Older adults were much more likely to live with an adult child than with a parent (25.2 % vs. 0.9 % in 1960, 18.1 % vs. 0.9 % in 2010) because few still had a living parent. The middle-aged were more evenly split between living with parents and adult children at each census. Based on these general patterns, we restrict our analyses of coresidence with parents to age groups in which individuals are likely to have living parents (aged 25–44 and 45–64); our analysis of coresidence with adult children is limited to age groups in which individuals are likely to have adult children (aged 45–64 and 65 and older).

In addition to examining “who coresides,” we also consider “who supports whom” within intergenerational households by comparing the income received by the members of each generation within these households. Our ultimate dependent variable is an indicator of financial dependency reflecting whether an individual (plus his or her spouse, if married) provides less than 40 % of the income earned by members of the two generations, combined. If so, that person (and spouse, if any) is considered to be financially dependent on the other generation. Without more precise data on the flow of support, we assume that if one generation provides substantially less than one-half of the household in- come, then it is likely to be the recipient of support from the other (donor) generation.3 There may not always be a donor and recipient in multigenera- tional households, but our goal is to see how the balance of economic resources within multigenerational households has changed over time.4

To create our measure of income dependency, we limit our focus to indi- viduals who are living in multigenerational households. We first calculate the income received (from all sources) by each generation, including the spouse’s income if either generation is married with a spouse present. This means that for each individual, we have his or her own/couple income as well as the income of the other generation in the household. We then sum the incomes from both generations to produce a measure of “multigenerational income” within that household. In more than 75 % of cases, multigenerational income equals total household income; the remaining households have other adults who receive income.5 Based on total multigenerational income, we determine wheth- er each generation’s share is less than 40 % of the total, indicating their dependency on the other generation. Hence, we are attempting to distinguish income dependency from the myriad other reasons to coreside, including health conditions and other noneconomic reasons, such as tastes valuing companion- ship relative to privacy.

3 We selected the 40 % threshold because it was sufficiently below the 50–50 mark and would therefore indicate an unequal sharing of financial support by the two generations. We explored other thresholds (e.g., 10 % and 25 %), but there were few differences, either in trends or determinants. 4 Throughout the period from 1960–2010, in only 15 % to 20 % of multigenerational households is financial support shared relatively equally (i.e., with between 40 % and 60 % of income provided by each generation). 5 In no more than 15 % of multigenerational households do other adults contribute more than 25 % of household income. However, the income of other earners has no effect on our intergenerational compar- isons because we focus only on income from the adult children and their parents (and spouses, if any).

1456 J.R. Kahn et al.

Other Measures

Other individual and household characteristics are used as potential correlates of each adult’s likelihood of living with and depending on adult relatives of a different generation. These variables are coded in the same way for all adults in all years, including both members of an intergenerational household pair. Marital status included four categories: (1) married, spouse present; (2) separated, divorced, or married, spouse absent; (3) widowed; and (4) never married. Race is coded into three categories: whites, blacks, and other. Because the 1960 census did not include a question on Hispanic origin, reflecting the small numbers of Hispanics in the United States at that time (Bean and Tienda 1987), we do not distinguish Hispanics in this analysis. Each individual’s nativity is derived from his or her place of birth, and we classify people into native (born in the United States, excluding outlying areas and territories) and foreign-born. Area of residence indicates whether the individual’s household was located in a metropolitan area. Formal education is measured by the highest grade completed at the time of the census and is grouped as follows: less than high school, high school graduate, some college, and college graduate or more. Employment status indicates whether the indi- vidual was currently employed at the time of the census or ACS interview. Total personal income from all sources is adjusted for inflation to reflect 1999 U.S. dollars, and is expressed in tens of thousands of dollars. Age is coded as a trichotomy: young adult (24–44), middle age (45–64), and older adult (65 and older).

The coresidence models include all adults aged 25 and older and incorporate these individuals’ characteristics. The income-dependency models are restricted to adults who live in multigenerational households, incorporating characteristics of both the younger and older generations. For both stages of the analysis, we present descriptive and regression results for only 1960, 1990, and 2010. We include results for all six census years in the tables in the online appendix (Online Resource 1).

Results

Coresidence Analysis

We present results for the coresidence analysis separately for two overlapping age groups of younger and older adults corresponding to the results from Table 1: adults aged 25–64 make up the sample at risk of living with a parent, and adults aged 45 and older are those at risk of living with an adult child. Sample characteristics for the coresidence analysis are included in Table 2, which shows the expected large changes between 1960, 1990, and 2010 in the distributions on many socioeconomic and demographic characteristics. (The full set of characteristics for the coresidence sample for the six census years appears online in Table S1).

Turning to the relationships between the covariates and intergenerational coresi- dence, Table 3 presents bivariate relationships for the likelihood of living in a multigenerational household, shown separately for those living with parents and adult children for the years 1960, 1990, and 2010. (The full set of bivariate relationships for the six census years appears online in Table S2.) Throughout the 50-year period, the trends show a small increase in the likelihood of living with parents (from 9.6 % in

Growing Parental Economic Power in Parent–Adult Child Households 1457

Table 2 Distributions on covariates by age and census year, (1960, 1990, and 2010): Adults aged 25 and older

Adults Aged 25–64 Adults Aged 45 and Older

1960 1990 2010 1960 1990 2010

N 794,485 1,190,583 1,485,107 494,501 769,945 1,287,614

Economic Characteristics

Education

Less than high school 54.3 15.5 9.9 70.9 29.7 12.9

High school graduate 27.8 33.9 34.5 16.0 35.1 38.9

Some college 9.5 27.6 24.5 7.4 19.3 21.3

College graduate or higher 8.4 23.1 31.1 5.7 16.0 27.0

Employment status

Not currently employed 36.8 25.0 27.8 50.5 54.7 49.1

Currently employed 63.2 75.0 72.2 49.5 45.4 50.9

Income

In 10 K of 1999 dollars 1.9 3.0 3.2 1.6 2.8 3.1

Below the median income 45.7 44.9 46.6 54.8 54.9 51.1

Above the median income 54.3 55.1 53.4 45.2 45.2 48.9

Other Characteristics

Age

25–44 56.3 62.3 48.4 NA NA NA

45–64 43.7 37.7 51.6 70.2 60.9 66.9

65 and older NA NA NA 29.8 39.2 33.1

Sex

Male 48.3 48.5 48.1 47.6 45.2 46.6

Female 51.7 51.5 51.9 52.4 54.8 53.4

Marital statusa

MSP 82.5 70.3 61.2 71.6 67.3 62.2

MSA/separated/divorced 5.9 13.9 17.5 5.5 11.5 18.3

Widowed 4.5 2.7 2.0 16.7 16.7 11.4

Never married 7.1 13.2 19.3 6.2 4.6 8.1

Race

White 90.3 85.6 81.1 91.4 88.4 84.3

Black 8.9 10.6 12.3 8.0 9.0 10.5

Other 0.8 3.8 6.7 0.6 2.6 5.2

Nativity

Native-born 92.9 89.3 81.4 86.7 90.3 84.9

Foreign-born 7.2 10.7 18.6 13.3 9.7 15.2

Area of residence

Nonmetropolitan/not identifiable 40.3 23.7 22.3 42.4 26.4 25.1

Metropolitan area 59.7 76.3 77.7 57.7 73.6 74.9

a MSP = married, spouse present. MSA = married, spouse absent.

1458 J.R. Kahn et al.

1960 to 11.3 % in 2010) and an even smaller decline in the likelihood of living with adult children (from 13.8 % in 1960 to 13.4 % in 2010). The percentages at the top of Table 3 show that for both types of coresidence, however, the trend was not monotonic. For young adults, there were declining levels of coresidence between 1960 and 1990, followed by a particularly large bump up in coresidence with parents between 1990 and 2010. For older adults, there was a slight increase in the likelihood of coresidence between 1960 and 1990 (from 13.8 % to 14.3 %) followed by a slight decline between 1990 and 2010 (from 14.3 % to 13.4 %).

These totals mask considerably larger changes over time for specific sociodemo- graphic subgroups, as well as different patterns for upward and downward coresi- dence. There were dramatic declines in the proportion of never-married adults who live with a parent (from 47.7 % in 1960 to 31.4 % in 2010), suggesting that the small increases observed for all younger adults reflect the increase in the proportions never married, and declines for those aged 65 and older and for widows who live with adult children (from 25.2 % to 18.0 %, and from 35.5 % to 26.6 %, respectively), although in each case, even lower levels appeared during the intermediate period.

We also see particularly strong increases over time, however, in coresidence with parents among more vulnerable subgroups. Those who are nonwhite, have low education, are not employed, or have lower than the median personal income were more likely to live with a parent in 2010 than in 1960. Moreover, both the race and education gradients in coresidence with a parent grew steeper by 2010, implying that disadvantage has played an increasing role in the residential choices of young adults. Finally, we see an interesting reversal in the effects of employment and income on coresidence with parents: in 1960, individuals with more resources (e.g., a job or higher income) were more likely than those with fewer resources to live with parents (perhaps because they could afford to offer support to their parents if they were in need). However, by 2010, individuals with fewer resources were more likely to live with parents (perhaps because they needed the support and their parents could now provide it). Consistent with this interpretation, we see, in contrast, a weakening of the negative gradients for income and employment on the likelihood of living with adult children. This suggests that over time, older adults’ own economic needs may be playing a less important role in their coresidence decisions, and their children’s economic needs may be playing a larger role.

The bivariate results suggest, in addition to the curvilinear patterns over time, a fundamental shift in the processes leading to intergenerational coresidence for youn- ger and older adults. Whereas socioeconomic disadvantage is playing a bigger role in the residential choices of young adults in the year 2010 than in 1960, it has become less central to the story for older adults.

Multivariate Analysis of the Changing Determinants of Coresidence

To assess the net impact of these factors, we now turn to the multivariate logistic regression results. Table 4 presents odds ratios from logistic regressions that predict the likelihood of living either with a parent (columns 1–3) or an adult child (columns 7–9) for

Growing Parental Economic Power in Parent–Adult Child Households 1459

Table 3 Bivariate relationships of coresidential status with covariates, by age and census year (1960, 1990, and 2010), for adults aged 25 and older: Likelihood of multigenerational coresidence

% Adults Aged 25–64 Living With a Parent

% Adults Aged 45 and Older Living With an Adult Child

1960 1990 2010 1960 1990 2010

N 794,485 1,190,503 1,485,082 494,501 769,902 1,287,602

% 9.6 8.5 11.3 13.8 14.3 13.4

Economic Characteristics

Education

Less than high school 9.0 9.6 13.3 16.5 18.8 25.0

High school graduate 10.8 9.7 14.0 8.3 14.4 14.3

Some college 9.8 8.3 11.3 7.3 11.6 11.4

College graduate or higher 9.7 6.4 7.8 5.2 9.0 8.3

Employment status

Not currently employed 7.9 10.0 15.3 19.6 15.2 16.4

Currently employed 10.6 8.1 9.8 8.0 13.2 10.6

Income

Below the median income 8.7 11.1 15.9 18.5 16.5 16.7

Above the median income 10.3 6.4 7.3 8.2 11.7 10.1

Other Characteristics

Age

25–44 11.5 11.1 16.9 NA NA NA

45–64 7.1 4.3 6.1 9.0 14.1 11.2

65 and older NA NA NA 25.2 14.6 18.0

Sex

Male 10.1 9.9 12.9 10.4 11.8 10.5

Female 9.1 7.2 9.9 16.9 16.4 16.0

Marital statusa

MSP 5.7 2.6 4.2 9.8 13.2 11.9

MSA/separated/divorced 20.6 13.8 14.5 16.1 14.9 14.1

Widowed 7.4 5.0 6.8 35.5 21.5 26.6

Never married 47.7 35.4 31.4 0.3 3.6 4.9

Race

White 9.5 7.6 10.2 13.5 13.0 12.1

Black 10.8 14.5 15.9 17.2 24.2 18.7

Other 13.2 13.0 16.4 21.5 26.4 25.2

Nativity

Native-born 9.9 8.3 11.2 12.3 13.5 11.4

Foreign-born 6.2 10.1 11.8 23.6 22.0 25.0

Area of residence

Nonmetropolitan area/not identifiable 9.3 7.0 10.0 13.0 11.0 10.5

Metropolitan area 9.8 9.0 11.7 14.5 15.5 14.4

a MSP = married, spouse present. MSA = married, spouse absent.

1460 J.R. Kahn et al.

T a b le 4

O d d s ra ti o s fr o m

lo g is ti c re g re ss io n s p re d ic ti n g m u lt ig en er at io n al co re si d en ce , b y ag e an d ce n su s y ea r (1 9 6 0 , 1 9 9 0 , an d 2 0 1 0 ): L ik el ih o o d o f li v in g w it h a p ar en t o r an

ad u lt ch il d

A d u lt s A g ed

2 5 – 6 4 L iv in g W it h a P ar en t

A d u lt s A g ed

4 5 an d O ld er

L iv in g W it h an

A d u lt C h il d

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g ea

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g ea

1 9 6 0 – 1 9 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

1 9 6 0 – 1 0 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

N 7 9 4 ,4 8 5

1 ,1 9 0 ,5 8 3

1 ,4 8 5 ,1 1 7

4 9 4 ,5 0 1

7 6 9 ,9 4 5

1 ,2 8 7 ,6 1 6

E co n o m ic

C h ar ac te ri st ic s

E d u ca ti o n (r ef . = le ss

th an

h ig h sc h o o l)

H ig h sc h o o l g ra d u at e

1 .1 5 * * *

1 .1 9 * * *

1 .2 9 * * *

* * * *

* * *

0 .6 2 * * *

0 .7 8 * * *

0 .6 6 * * *

* * *

* * *

* * *

S o m e co ll eg e

1 .0 1

0 .9 4 * * *

1 .0 5 * * *

* *

* * *

* 0 .5 4 * * *

0 .6 1 * * *

0 .5 5 * * *

* * *

* * *

C o ll eg e g ra d u at e o r h ig h er

0 .8 6 * * *

0 .7 6 * * *

0 .8 6 * * *

* * *

* * *

0 .5 1 * * *

0 .5 1 * * *

0 .4 0 * * *

* * *

* * *

E m p lo y m en t st at u s (r ef . = n o t em

p lo y ed )

C u rr en tl y em

p lo y ed

1 .1 8 * * *

0 .8 9 * * *

0 .7 9 * * *

* * *

* * *

* * *

0 .7 2 * * *

1 .0 5 * * *

0 .9 8 * *

* * *

* * *

* * *

In co m e

In 1 0 K

o f 1 9 9 9 U S D

0 .9 6 * * *

0 .8 6 * * *

0 .8 8 * * *

* * *

* * *

* * *

0 .9 2 * * *

0 .9 7 * * *

0 .9 7 * * *

* * *

* * *

O th er

C h ar ac te ri st ic s

A g e (r ef . = 4 5 – 6 4 )

2 5 – 4 4

1 .6 9 * * *

1 .7 0 * * *

2 .3 9 * * *

* * *

* * *

N A

N A

N A

6 5 an d o ld er

N A

N A

N A

1 .8 4 * * *

0 .8 2 * * *

1 .2 3 * * *

* * *

* * *

* * *

S ex

(r ef . = m al e)

F em

al e

0 .8 9 * * *

0 .6 3 * * *

0 .6 7 * * *

* * *

* * *

* * *

1 .0 5 * * *

1 .2 0 * * *

1 .2 7 * * *

* * *

* * *

* * *

M ar it al

S ta tu sb

(r ef . = M S P )

M S A /s ep ar at ed /d iv o rc ed

4 .5 4 * * *

6 .3 2 * * *

4 .0 5 * * *

* * *

* * *

* * *

1 .7 1 * * *

1 .0 0

1 .0 5 * * *

* * *

* *

* * *

W id o w ed

1 .8 1 * * *

2 .5 8 * * *

2 .1 2 * * *

* * *

* * *

* * *

3 .3 1 * * *

1 .6 7 * * *

1 .7 8 * * *

* * *

* * *

* * *

N ev er

m ar ri ed

1 5 .2 2 * * *

1 9 .8 8 * * *

8 .3 7 * * *

* * *

* * *

* * *

0 .0 3 * * *

0 .2 0 * * *

0 .3 1 * * *

* * *

* * *

* * *

Growing Parental Economic Power in Parent–Adult Child Households 1461

T a b le

4 (c o n ti n u ed )

A d u lt s A g ed

2 5 – 6 4 L iv in g W it h a P ar en t

A d u lt s A g ed

4 5 an d O ld er

L iv in g W it h an

A d u lt C h il d

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g ea

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g ea

1 9 6 0 – 1 9 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

1 9 6 0 – 1 0 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

R ac e (r ef . = w h it e)

B la ck

0 .8 6 * * *

0 .9 9

0 .9 5 * * *

* * *

* * * *

1 .1 6 * * *

2 .0 0 * * *

1 .7 5 * * *

* * *

* * *

* * *

O th er

1 .1 0 †

1 .7 0 * * *

2 .0 0 * * *

* * *

* * *

* * *

1 .8 1 * * *

1 .9 5 * * *

1 .6 8 * * *

* * *

N at iv it y (r ef . = n at iv e- b o rn )

F o re ig n -b o rn

0 .6 4 * *

1 .0 7 * * *

1 .0 6 * * *

* * *

* * *

1 .6 8 * * *

1 .5 0 * * *

1 .9 3 * * *

* * *

* * *

* * *

A re a o f re si d en ce

(r ef . = n o n m et ro p o li ta n ar ea )

M et ro p o li ta n ar ea

0 .9 9

1 .1 8 * * *

1 .1 7 * * *

* * *

* * *

1 .1 7 * * *

1 .5 2 * * *

1 .3 6 * * *

* * *

* * *

* * *

a B as ed

o n y ea r in te ra ct io n s w it h co v ar ia te s te st ed

o n p o o le d sa m p le s co m b in in g o b se rv at io n s fo r p ai rs

o f ce n su s y ea rs

1 9 6 0 , 1 9 9 0 , an d 2 0 0 0 .

b M S P = m ar ri ed , sp o u se

p re se n t. M S A = m ar ri ed , sp o u se

ab se n t.

† p < .1 0 ; * p < .0 5 ; * * p < .0 1 ; * * * p < .0 0 1

1462 J.R. Kahn et al.

three time periods: 1960, 1990, and 2010. Odds ratios for all census years between 1960 and 2010 are included in Table S3 in Online Resource 1. (Given our very large sample sizes, virtually all coefficients reach statistical significance.) Because our focus is on changes over time in the determinants of intergenerational coresidence, we test for year interactions with all covariates in pooled models comparing 1960 with 1990, 1990 with 2010, and 1960 with 2010 (interaction models not shown), and we report the significance levels of these interactions in columns 4–6 and 10–12 of Table 4. The effects of almost all covariates changed significantly over time, often monotonically.

Our story about the changing needs of older and younger generations can be seen most clearly in the effects of employment and income. In 1960, net of other factors, employed younger adults were 18 % more likely than the unemployed to live with parents (OR = 1.18), presumably because they were better able to provide them with support; and employed older adults were only 72 % as likely as the unemployed to live with adult children, presumably because they were less likely to need support. By 2010, however, employed younger adults were only 79 % as likely to live with parents as the unemployed (suggesting they did not need their now well-off parents’ support as much as their unemployed peers), and employed older adults were 98 % as likely to live with their adult children. The significant weakening of the negative effect of parents’ employment probably reflects both the strengthened economic position of all parents (regardless of employment status) and the growing vulnerability of the younger generation. A similar pattern can be observed for the effects of income between 1960 and 2010.

In summary, our findings provide a clear picture of the changing determinants of intergenerational living over the past half-century. Whereas in 1960, the neediness of the older generation played a large role in coresidence decisions, by 2010, these decisions were clearly driven more by the economic needs of the younger generation. The effects of income became more important for the younger generation but became less important for older parents. Reinforcing this interpretation, the effect of being employed changed sign: it changed from positively predicting coresidence with a parent in 1960, implying that financially stable young adults were providing residen- tial support to their parents, to negatively affecting coresidence, reinforcing the portrait of an increasingly needy younger generation. The weakening of the negative effect of employment for the older generation clarifies this portrayal as their own financial needs played a smaller role in coresidence decisions.

Our other findings also highlight important changes during this period. Although men were more likely than women to live with their parents in all census years, and women were more likely than men to live with their adult children, the gender gaps grew larger over time. Thus, by 2010, younger women were significantly less likely than men to live with parents compared with 1960, and older women became significantly more likely than men to live with adult children in 2010 than in 1960. These gender patterns are likely to be linked to the delays in marriage (especially if it is unmarried men who are more likely to live with parents than unmarried women), wage stagnation (which affected men more than women), and increases in marital disruption (especially if previously married mothers are more likely than previously married fathers to live with adult children). In spite of the changes in marital status, the differences in coresidence patterns by marital status have moderated over time: never-married adults are still more likely than others to live with parents, but the gap between the never- and the ever-married has narrowed. Similarly, whereas previously married adults are still more likely than married adults to

Growing Parental Economic Power in Parent–Adult Child Households 1463

live with adult children, again the gaps have narrowed, probably reflecting the wider options available for older widowed and divorced adults to live on their own.

There was a shift in race patterns of coresidence. In 1960, blacks were 14 % less likely than whites to live with parents (OR = 0.86), but in 2010, they were only 5 % less likely to do so (OR = 0.95). This change could reflect the greater distance between many blacks and their parents in 1960 compared with whites because of the Great Migration north between the two World Wars (Tolnay 1997), but by 2010, race differences in proximity between generations were likely to have declined. Nevertheless, the lower level of coresidence with parents among younger black adults compared with whites is something of a puzzle, given that studies based on surveys normally found greater parental coresidence among young black adults (e.g., Goldscheider and DaVanzo 1989).

Coresidence patterns by nativity have shifted as well, probably reflecting historical immigration patterns. Whereas in 1960, foreign-born adults were only 64 % as likely as the native-born to live with parents (OR = 0.64), by 2010, they were 6 % more likely to do so (OR = 1.06). And between 1960 and 2010, foreign-born adults became even more likely than the native-born to live with adult children (OR = 1.68 in 1960, and OR = 1.93 in 2010). This could be due to the shifting composition of the younger foreign-born population, with many more recent immigrants in 2010 than in 1960, who may have immigrated as children and still live near their parents. Immigrants became more likely to live with their parents and also more likely to live with their adult children. Again, it is not clear how much of this reflects their own needs versus those of their relatives.

Finally, there were interesting shifts in metropolitan/nonmetropolitan patterns: whereas in 1960, metropolitan-area residents were equally likely as nonmetropolitan- area residents (OR = 0.99, nonsignificant) to live with parents, by 2010, adults in metropolitan areas were significantly more likely to live with parents (OR = 1.17). This may reflect the higher cost of living in metropolitan areas. From the older adults’ perspective, metropolitan-area residents became increasingly likely over time to live with adult children compared with nonmetropolitan-area residents (OR = 1.17 in 1960, and OR = 1.36 in 2010).

Overall, then, our results suggest substantial changes in intergenerational coresi- dence patterns in recent decades, which indicate that there have been shifts in the patterns of need that produce multigenerational households. Not all are totally unidirectional, but none of the few curvilinear series of coefficients evident either in Table 4 or in Table S3 in Online Resource 1 seem either substantive or interpret- able. Hence, we focus on the overall pattern of change, which seems to support an argument that whereas in 1960, coresidence decisions primarily reflected the needs of older rather than younger adults, by 2010, it was clear that the financial needs of the younger generation became more important. Indeed, the results suggest a pattern of growing neediness among younger adults along with their increasing dependency on older relatives for support.

The results from the coresidence analysis in Table 4 suggest that the younger generation has grown needier over time, forcing the older generation to continue in the provider role later in life. The evidence based on shifting economic determinants of coresidence, however, is suggestive at best. To examine these relationships explic- itly, we now look at the relative incomes of coresiding generations to draw inferences about the flow of support across generations within households.

1464 J.R. Kahn et al.

Analysis of Financial Dependency Within Coresidential Households

Our analysis of financial dependency is restricted to the subsample of adults who were living in coresidential households. Characteristics of both coresiding genera- tions in these households are presented for 1960, 1990, and 2010 in Table 5, separately for adults aged 25–64 who were living with a parent, and adults aged 45 or older who were living with an adult child. (Distributions for all census years between 1960 and 2010 are presented in Table S4 in Online Resource 1.) Although it is based only on adults who live in intergenerational households, the results in Table 5 show many of the same trends as Table 2 (e.g., rising education, declining marriage among young adults). Moreover, the pattern of growing neediness of the younger generation suggested by our earlier analysis is clearly confirmed by our direct estimates of financial dependency within multigenerational households.

Table 6 shows bivariate relationships between the covariates and the likelihood of contributing less than 40 % of multigenerational income (income dependency), for 1960, 1990, and 2010 (results for all census years are presented in Table S5 in Online Resource 1). Trends in income dependency are plotted in Fig. 2 which shows that over time, despite fluctuations, income dependency increased sharply for those who lived with a parent (from 19.2 % in 1960 to 47.7 % in 2010) and declined sharply for those who lived with an adult child (from 54.4 % in 1960 to 26 % in 2010). In other words, older parents shifted from having lower incomes than their coresident adult children in 1960 to being the financial providers in 2010. Unlike the trends in coresidence, which varied substantially by subgroup, the trends in intergenerational dependency were pervasive, and the gradients by social and economic status persisted over time. As one might expect among coresiding adults, the more vulnerable subgroups (e.g., the youngest and oldest adults, the unmarried, the less-educated, and the unemployed) were much more likely to be financially dependent on the other generation in all years.

Table 7 presents odds ratios from logistic regressions for 1960, 1990, and 2010, predicting whether an individual (plus his or her spouse, if married) contributes less than 40 % of the total income received by both coresiding generations, implying financial dependency on the other generation. The models of financial dependency are presented separately from the adult child’s and parent’s perspectives (in columns 1–3 and 7–9, respectively), although all models include characteristics of both generations. Results from year-interaction tests for each covariate are presented in columns 4–6 and 10–12. Results for all census years from 1960 to 2010 are in Table S6 in Online Resource 1.

The regression results show that the flows of resources within multigenerational households reflect the characteristics of both generations in the household. The effect of socioeconomic resources such as education and employment on financial depen- dency is especially strong. For both parents and adult children in multigenerational households, having a higher education protects each generation from dependency on the other, and this effect has grown stronger over time. Controlling for the education of the parent or child with whom one lives, one’s own lower education is an increasingly important predictor of financial dependency, suggesting that individ- uals with fewer educational resources were at a greater relative disadvantage in 2010 than in 1960. Individuals living with a highly educated parent or child were

Growing Parental Economic Power in Parent–Adult Child Households 1465

Table 5 Distributions on covariates, by age and census year (1960, 1990, and 2010): Coresident adults of different generations

Adults Aged 25–64 Living With a Parent

Adults Aged 45 and Older Living With an Adult Child

1960 1990 2010 1960 1990 2010

N 76,303 100,542 148,593 68,400 110,130 160,512

Economic Characteristics

Child’s education

Less than high school 50.7 17.8 11.0 49.9 16.1 9.9

High school graduate 31.2 39.1 42.5 29.9 38.7 42.1

Some college 9.7 26.2 23.7 10.4 27.0 24.1

College graduate or higher 8.4 16.9 22.8 9.9 18.3 23.8

Parent’s education

Less than high school 86.0 44.9 25.1 84.4 39.8 22.5

High school graduate 8.5 31.7 41.5 9.6 35.3 42.4

Some college 3.6 14.0 17.2 3.9 15.2 18.1

College graduate or higher 2.0 9.4 16.2 2.1 9.8 17.1

Employment status

Both unemployed 22.2 20.1 24.5 14.6 18.1 23.0

Child unemployed, parent employed 7.9 9.2 12.2 6.5 9.0 13.1

Child employed, parent unemployed 49.3 40.4 39.1 56.8 40.7 36.6

Both employed 20.6 30.1 24.2 22.2 32.2 27.3

Other Characteristics

Child’s age

25–44 67.7 80.1 69.2 71.6 84.3 73.4

45–64 32.3 19.9 30.8 28.4 15.8 26.6

Parent’s age

45–64 35.7 50.9 47.4 45.8 59.1 54.2

65 and older 64.3 49.1 52.6 54.2 40.9 45.8

Child’s sex

Male 50.9 56.0 54.1 66.0 61.9 57.9

Female 49.1 44.0 45.9 34.1 38.2 42.1

Parent’s sex

Male 42.1 47.7 37.5 35.9 37.4 37.2

Female 57.9 52.3 62.5 64.1 62.6 62.8

Race

White 88.8 78.1 74.6 89.2 81.4 77.1

Black 10.0 16.1 14.9 9.9 13.8 13.3

Other 1.1 5.8 10.5 1.0 4.8 9.6

Child’s marital status

Unmarried 51.4 78.3 74.4 70.2 88.4 85.4

Married 48.6 21.7 25.6 29.8 11.6 14.6

Parent’s marital status

Unmarried 70.0 56.9 59.6 49.4 37.9 41.5

Married 30.0 43.1 40.4 50.6 62.1 58.6

1466 J.R. Kahn et al.

significantly more likely than others to be financially dependent, even controlling for their own education.

Over time, employment continues to be highly protective against financial dependency for both adult children and parents, especially when the other gener- ation is unemployed. Not surprisingly, unemployed adult children living with employed parents and unemployed parents living with employed adult children were increasingly more likely to be financially dependent than those living in households where neither generation was employed (the reference category). These effects are noticeably stronger in 2010 than in previous years, and they are consistent with the bivariate relationships in Table 6, which shows that in 2010, almost 90 % of unemployed adult children were economically dependent on their employed parents, compared with only 50 % of unemployed parents living with employed adult children. Interestingly, in 2010, in the case in which both gen- erations were unemployed, 58 % of adult children are economically dependent on their aging parents; this compares with only 17 % of parents who were econom- ically dependent on their adult children.

There were also interesting changes in the effects of the other covariates less closely tied to resources. We see a weakening of the relative financial position of young adults in 1990 and 2010 as they became more likely than middle-aged adult children to be financially dependent on their parents, compared with 1960, when they were less likely than the middle-aged to be dependent; similarly, we see that over time, parents became even less likely to be financially dependent on their young adult rather than middle-aged children. The financial position of elderly parents also strengthened such that by 2010, they were no longer so much more vulnerable to financial dependency than middle-aged parents: in 1960, they were almost twice as likely as middle-aged parents to be dependent on their adult children, whereas in 2010, they were only 7 % more likely.

Also, in contrast to our results for coresidence, the analysis of financial dependen- cy shows a decline in the significance of gender, at least among adult children. Whereas in 1960, adult daughters were 23 % more likely than adult sons to be financially dependent on their parents, the gap dropped to 10 % in 1990 and reversed direction by 2010, at which time sons were 5 % more likely than daughters to be

Table 5 (continued)

Adults Aged 25–64 Living With a Parent

Adults Aged 45 and Older Living With an Adult Child

1960 1990 2010 1960 1990 2010

Nativity

Both foreign-born 4.0 10.5 17.2 3.7 8.7 14.9

Child native-born, parent foreign-born 20.6 5.5 9.2 19.1 5.3 9.5

Parent native-born, child either 75.5 84.0 73.6 77.2 86.0 75.6

Area of residence

Nonmetropolitan area 39.0 24.1 22.8 39.6 24.8 23.9

Metropolitan area 61.0 75.9 77.2 60.4 75.2 76.1

Growing Parental Economic Power in Parent–Adult Child Households 1467

Table 6 Bivariate relationships of dependency status with covariates, by age and census year (1960, 1990, and 2010), for individuals living with a parent or an adult child: Likelihood of dependency (% contributing less than 40 % of multigenerational income)

Likelihood of Dependency

Among Adults Aged 25–64 Living With a Parent

Among Adults Ages 45 and Older Living With an Adult Child

1960 1990 2010 1960 1990 2010

N 76,303 100,542 148,593 68,400 110,130 160,512

% 19.2 43.7 47.7 53.4 27.0 26.0

Economic Characteristics

Child’s education

Less than high school 21.1 45.3 50.5 54.7 27.7 24.1

High school graduate 16.4 43.6 50.3 52.4 25.9 23.0

Some college 20.1 44.4 48.0 49.2 26.4 25.4

College graduate or higher 17.7 41.5 41.2 54.1 29.7 32.5

Parent’s education

Less than high school 18.2 34.1 32.5 55.4 37.4 40.5

High school graduate 20.9 43.7 45.7 45.9 24.1 25.9

Some college 27.0 56.7 58.5 41.9 16.6 17.5

College graduate or higher 43.1 70.4 64.7 30.2 11.4 16.0

Employment status

Both unemployed 22.0 48.4 58.3 32.7 20.9 16.8

Child unemployed, parent employed 58.6 82.1 87.4 9.7 5.1 3.6

Child employed, parent unemployed 5.3 20.2 20.6 73.5 46.6 50.1

Both employed 34.4 60.6 60.6 28.2 11.7 12.1

Other Characteristics

Child’s age

25–44 23.2 49.9 55.6 45.0 21.4 19.6

45–64 10.9 18.7 29.8 74.7 56.7 43.6

Parent’s age

45–64 30.4 57.5 60.6 34.4 16.1 15.9

65 and older 13.0 29.5 36.1 69.5 42.8 37.9

Child’s sex

Male 18.3 45.6 50.7 60.4 29.0 25.7

Female 20.2 41.4 44.1 39.9 23.8 26.3

Parent’s sex

Male 34.7 63.0 64.4 40.0 17.2 18.6

Female 8.0 26.2 37.6 60.9 32.9 30.3

Race

White 18.3 43.9 49.0 54.2 25.7 23.6

Black 26.7 47.7 51.1 46.2 28.1 26.2

Other 22.4 30.1 33.3 52.9 45.8 45.1

Child’s marital status

Unmarried 34.2 54.0 60.4 39.7 20.4 18.1

Married 3.4 6.8 10.9 85.8 77.3 72.1

1468 J.R. Kahn et al.

financially dependent. Based on the bivariate trends in Table 6, it appears that sons saw a greater increase in dependency over time than did daughters. Mothers, how- ever, remained more financially dependent on adult children compared with fathers (OR = 1.07 in 1960, and OR = 1.09 in 2010). Nevertheless, the gender of the child mattered less to a parent’s financial dependency in 2010 than in 1960: whereas in 1960, a parent was 75 % as likely to be financially dependent on a daughter as on a son, by 2010, the odds were even (OR = 1.0). Thus, for the younger generation, gender has come to play a smaller role in the flow of resources within intergenera- tional households.

Race differences have also narrowed over time such that black adult children are no longer more likely than white adult children to be dependent on their coresidential parents (OR = 1.21 in 1960 vs. OR = 0.93 in 2010). However, during the same period, black parents became even more likely than white parents to be financially dependent on their adult children (OR = 0.95 in 1960 vs. OR = 1.27 in 2010). These

Table 6 (continued)

Likelihood of Dependency

Among Adults Aged 25–64 Living With a Parent

Among Adults Ages 45 and Older Living With an Adult Child

1960 1990 2010 1960 1990 2010

Parent’s marital status

Unmarried 9.3 26.0 32.9 78.3 51.2 44.9

Married 42.3 67.2 69.4 29.2 12.2 12.6

Nativity

Both foreign-born 9.7 25.5 24.2 69.1 49.6 52.1

Child native-born, parent foreign-born 14.2 34.4 46.8 63.5 35.3 27.7

Parent native-born, child either 21.1 46.7 53.3 50.2 24.2 20.6

Area of residence

Nonmetropolitan area 22.4 44.8 51.2 50.3 25.3 21.1

Metropolitan area 17.2 43.4 46.6 55.4 27.6 27.5

Fig. 2 Trends in income dependency within intergenerational households by age, 1960–2010

Growing Parental Economic Power in Parent–Adult Child Households 1469

T a b le

7 O d d s ra ti o s fr o m

lo g is ti c re g re ss io n s p re d ic ti n g d ep en d en cy , b y ag e an d ce n su s y ea r (1 9 6 0 , 1 9 9 0 , an d 2 0 1 0 ), fo r in d iv id u al s li v in g w it h a p ar en t o r an

ad u lt ch il d :

L ik el ih o o d o f co n tr ib u ti n g le ss

th an

4 0 %

o f m u lt ig en er at io n al

in co m ea

L ik el ih o o d o f D ep en d en cy

A m o n g A d u lt s A g ed

2 5 – 6 4 L iv in g W it h a P ar en t

A m o n g A d u lt s A g ed

4 5 an d O ld er

L iv in g W it h an

A d u lt C h il d

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g eb

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g eb

1 9 6 0 – 1 9 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

1 9 6 0 – 1 9 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

N 7 6 ,3 0 3

1 0 0 ,5 4 2

1 4 8 ,5 9 3

6 8 ,4 0 0

11 0 ,1 3 0

1 6 0 ,5 1 2

E co n o m ic

C h ar ac te ri st ic s

C h il d ’s ed u ca ti o n (r ef . = le ss

th an

h ig h sc h o o l)

H ig h sc h o o l g ra d u at e

0 .6 3 * * *

0 .7 5 * * *

0 .8 4 * * *

* * *

* *

* * *

1 .1 9 * * *

1 .3 5 * * *

1 .2 6 * * *

* *

S o m e co ll eg e

0 .7 6 * * *

0 .7 1 * * *

0 .7 7 * * *

* 1 .0 3

1 .5 2 * * *

1 .4 3 * * *

* * *

* * *

C o ll eg e g ra d u at e o r h ig h er

0 .5 9 * * *

0 .5 7 * * *

0 .5 5 * * *

1 .3 0 * * *

1 .9 4 * * *

2 .2 5 * * *

* * *

* *

* * *

P ar en t’ s ed u ca ti o n (r ef . = le ss

th an

h ig h sc h o o l)

H ig h sc h o o l g ra d u at e

1 .3 6 * * *

1 .3 6 * * *

1 .4 3 * * *

† 0 .7 4 * * *

0 .6 8 * * *

0 .7 1 * * *

*

S o m e co ll eg e

1 .9 9 * * *

2 .1 9 * * *

2 .1 3 * * *

0 .6 0 * * *

0 .4 3 * * *

0 .4 8 * * *

* * *

* * * *

C o ll eg e g ra d u at e o r h ig h er

3 .1 1 * * *

4 .2 1 * * *

3 .6 9 * * *

* *

* *

* 0 .4 3 * * *

0 .2 6 * * *

0 .3 2 * * *

* * *

* * *

* *

E m p lo y m en t st at u s (r ef . = b o th

u n em

p lo y ed )

C h il d u n em

p lo y ed , p ar en t em

p lo y ed

2 .4 0 * * *

2 .4 0 * * *

3 .1 8 * * *

* * *

* * *

0 .4 0 * * *

0 .4 1 * * *

0 .2 9 * * *

* * *

* * *

C h il d em

p lo y ed , p ar en t u n em

p lo y ed

0 .1 3 * * *

0 .1 8 * * *

0 .1 4 * * *

* * *

* * *

5 .2 7 * * *

4 .6 2 * * *

5 .8 9 * * *

* *

* * *

* *

B o th

em p lo y ed

0 .5 0 * * *

0 .5 6 * * *

0 .4 9 * * *

* *

* * *

1 .1 6 * * *

1 .1 3 * * *

1 .1 2 * * *

O th er

C h ar ac te ri st ic s

C h il d ’s ag e (r ef . = 4 5 – 6 4 )

2 5 – 4 4

0 .9 2 *

1 .3 5 * * *

1 .3 7 * * *

* * *

* * *

0 .9 1 * *

0 .7 0 * * *

0 .7 5 * * *

* * *

† * * *

P ar en t’ s ag e (r ef . = 4 5 – 6 4 )

6 5 an d o ld er

0 .8 1 * * *

0 .8 1 * * *

0 .9 5 *

* * *

* * *

1 .9 3 * * *

1 .3 2 * * *

1 .0 7 * * *

* * *

* * *

* * *

1470 J.R. Kahn et al.

T a b le

7 (c o n ti n u ed )

L ik el ih o o d o f D ep en d en cy

A m o n g A d u lt s A g ed

2 5 – 6 4 L iv in g W it h a P ar en t

A m o n g A d u lt s A g ed

4 5 an d O ld er

L iv in g W it h an

A d u lt C h il d

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g eb

1 9 6 0

1 9 9 0

2 0 1 0

S ig n if ic an t C h an g eb

1 9 6 0 – 1 9 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

1 9 6 0 – 1 9 9 0

1 9 9 0 – 2 0 1 0

1 9 6 0 – 2 0 1 0

C h il d ’s se x (r ef . = m al e)

F em

al e

1 .2 3 * * *

1 .1 0 * * *

0 .9 5 * * *

* * *

* * *

* * *

0 .7 5 * * *

0 .8 8 * * *

1 .0 0

* * *

* * *

* * *

P ar en t’ s se x (r ef . = m al e)

F em

al e

0 .4 6 * * *

0 .5 8 * * *

0 .6 8 * * *

* * *

* * *

* * *

1 .0 7 * * *

1 .1 4 * *

1 .0 9 * * *

* *

C h il d ’s ra ce

(r ef . = w h it e)

B la ck

1 .2 1 * * *

1 .0 6 *

0 .9 3 * *

* *

* * *

* * *

0 .9 5

1 .2 7 * * *

1 .2 7 * * *

* * *

* * *

O th er

1 .0 3

0 .9 1 *

0 .8 9 * * *

1 .0 5

1 .4 7 * * *

1 .4 3 * * *

* *

C h il d ’s m ar it al

st at u s (r ef . = u n m ar ri ed )

M ar ri ed

0 .0 8 * * *

0 .1 0 * * *

0 .1 0 * * *

* *

* * *

4 .0 8 * * *

7 .1 4 * * *

8 .0 1 * * *

* * *

* *

* * *

P ar en t’ s m ar it al

st at u s (r ef . = u n m ar ri ed )

M ar ri ed

2 .0 3 * * *

3 .0 9 * * *

4 .0 6 * * *

* * *

* * *

* * *

0 .1 8 * * *

0 .1 7 * * *

0 .1 7 * * *

* * *

N at iv it y (r ef . = p ar en t n at iv e- b o rn , ch il d ei th er )

B o th

fo re ig n -b o rn

1 .2 5 *

1 .3 8 * * *

1 .7 8 * * *

* * *

* * *

0 .9 1

0 .5 8 * * *

0 .5 0 * * *

* * *

* * * *

C h il d n at iv e- b o rn , p ar en t fo re ig n -b o rn

1 .6 4 * * *

1 .8 6 * * *

2 .8 6 * * *

* * *

* * *

0 .6 2 * * *

0 .4 2 * * *

0 .2 9 * * *

* * *

* * *

* * *

A re a o f re si d en ce

(r ef . = n o n m et ro p o li ta n ar ea )

M et ro p o li ta n ar ea

0 .8 2 * * *

1 .0 0

1 .1 2 * * *

* * *

* * *

* * *

1 .1 0 * * *

1 .0 0

1 .0 0

* * *

a “M

u lt ig en er at io n al

in co m e”

is d ef in ed

as th e su m

o f to ta l in co m e re p o rt ed

b y th e ad u lt ch il d (a n d sp o u se , if m ar ri ed ) an d p ar en t (a n d sp o u se , if m ar ri ed ).

b B as ed

o n y ea r in te ra ct io n s w it h th e co v ar ia te s te st ed

o n p o o le d sa m p le s co m b in in g o b se rv at io n s fo r p ai rs

o f ce n su s y ea rs

1 9 6 0 , 1 9 9 0 , an d 2 0 1 0 .

† p < .1 0 ; * p < .0 5 ; * * p < .0 1 ; * * * p < .0 0 1

Growing Parental Economic Power in Parent–Adult Child Households 1471

patterns suggest that younger whites are no longer as advantaged over younger blacks as in the past, and also that the gains we have observed for older adults are less characteristic of blacks and other races (likely Hispanics) than for whites.

Our results also highlight the importance of marriage for both generations. When either adult children or parents are married, they are significantly less likely to be financially dependent on the other generation, regardless of the latter’s marital status. And conversely, when the other generation is married, this raises the likelihood of financial dependency, regardless of one’s own marital status. In addition, we find that immigrant families have become even more likely than native families to support their adult children into adulthood, with especially strong intergenerational support in families with foreign-born parents and native-born children. And finally, we find that the odds of being financially dependent on parents were higher in nonmetropolitan areas in 1960 and in metropolitan areas in 2010. These patterns probably reflect the rapidly rising cost of living in metropolitan areas as well as the movement of young adults from rural areas to metropolitan areas during this time period. Nonetheless, by 1990 and again in 2010, there was no difference in the likelihood of financial dependency on adult children between parents from metropolitan areas and those from nonmetropolitan areas.

Discussion

In this article, we examine the changing nature of intergenerational coresidence and financial support in the United States over the past half-century—a period of rapid social, economic, and demographic changes. Unlike previous studies that have focused primarily on only one age group, such as young single adults or elderly widows, we focus on the residential choices of both younger and older adults in order to understand how the needs of different generations influence their joint living arrangements. And unlike previous studies, we examine change over a recent, five-decade period by using U.S. census and ACS data from 1960 through 2010. We examine changes over time in the determinants of living with either an older or a younger generation from the perspective of younger adults (aged 25–64) and older adults (aged 45 and older), as well as the determinants of financial dependency within parent–adult child households, in each case assessing how these determinants have changed over the period.

We find that the patterns of intergenerational coresidence and resource flows within coresidential households have changed in dramatic ways, paralleling the general trends toward the greater economic security of older adults and the increasing financial strain experienced by younger adults. Consistent with our first hypothesis, we found that younger adults have become increasingly needy over time, as reflected in their likelihood of intergenerational coresidence. Our results suggest that the needs of the older generation played a much larger role in coresidence decisions in 1960 than in 2010, when these decisions were clearly driven more by the economic needs of the younger generation. Further, consistent with our second hypothesis, economic resources played a more important role in the decisions of young adults to coreside in 2010 than was the case in 1960. However, this was not the case for the older generation, for whom the effects of their own resources on the coresidence decision declined.

1472 J.R. Kahn et al.

These countervailing patterns highlight the importance of considering the financial well-being of both generations. The strengthening effect of income insecurity on the likelihood that young adults coreside with their parents is likely to be the result of the decline in impoverished elderly parents living with their relatively affluent children, rather than any increase in “tastes” for privacy. Similarly, the declining effect of resources on the likelihood that older parents coreside with their adult children is likely to reflect that their relatively impoverished adult children have come to live with them rather than any decline in “tastes” for privacy. Our direct measures of individual-level income provide a more robust test of the changing impact of income on coresidence than was possible in the studies by Costa (1999), which used state average Old Age Assistance benefit levels, and McGarry and Schoeni (2000), which used an imputed average Social Security benefit.

Our analysis of the determinants of being financially dependent in parent–adult child households shows that young adults have clearly become the more financially dependent generation compared with their parents. Although the determinants of financial dependency have not changed over time nearly as much as the determinants of coresidence, our results suggest increasing challenges for young adults, especially those with fewer economic resources. In addition to finding that younger adults are increasingly disadvantaged relative to older age groups, we also find that education played a larger role in financial dependency in 2010 than in 1960, further disadvan- taging the least educated.

Whereas nonemployment and nonmarriage remain strongly associated with de- pendency for both adult children and older parents, we find that gender became less predictive of dependency, in spite of the growing gender gap in coresidence: by 2010, young men were much more likely to coreside than young women, and older women became increasingly likely to live with adult children compared with older men. Daughters are no longer so much more likely than sons to be financially dependent on their parents. Surely, the increases over time in women’s employment and wages along with the relative stagnation of men’s wages have combined to reduce the gender gap in dependency. Although mothers remain more likely than fathers to be financially dependent on their adult children, they are now equally likely to depend on daughters as sons.

Finally, our findings regarding race suggest a mixed pattern with a narrowing of the race gap in financial dependency at younger ages (whereby young blacks are no longer more likely to be dependent than young whites) but a disturbingly larger race gap at older ages. Whereas in 1960, black parents were no more likely than white parents to be financially dependent on their coresidential adult children, by 2010, the race gap for parental dependency had increased: black parents in 2010 were 27 % more likely than whites to be financially dependent on their children. These changes parallel those in the analysis of coresidence.

Our findings with regard to gender, marital status, and race deserve additional research in analyses that are possible when using IPUMS. Does the changing gender pattern of coresidence reflect increases in female unmarried parenthood? Do the changes in effects for race reflect the changes in race/ethnic/nativity composition that occurred over the 50-year period covered by our analysis? Do the same patterns characterize households with combined incomes below the poverty line the same way as more financially well-off households?

Growing Parental Economic Power in Parent–Adult Child Households 1473

These questions are becoming particularly critical as the United States struggles with the prolonged economic strains caused by the Great Recession. Our results for the period 2000–2010 showed a particularly large increase (by almost 50 %) in young adults’ coresidence with parents, which may reflect the increases in youth unemploy- ment or housing foreclosures (Kochhar et al. 2011; Wiemers 2012). Middle-aged adults were not exempt from these trends as they too saw an increase of almost 25 % in the proportion living with parents; in contrast, elders aged 65 and older saw an increase of only 5 % in the proportion living with adult children. So although there clearly has not yet been a reversal of fortunes at older ages—the current elderly are even more likely to be providing coresidence and financial support to their children than in the past—the result for middle-aged adults may presage substantial change by 2020. Given that their behavior more closely resembled that of younger adults than the oldest adults, by the time the middle-aged have become elderly—with lower savings than earlier cohorts, a lower likelihood of receiving a generous private pension, and perhaps substantial cuts in the value of social (and health) security—a reversal does not seem totally implausible in the near future.

Thus, coresidence (and the support that it typically provides) clearly continues to be an important resource for families. It can increase financial well-being not just for those unable to live independently—such as the traditional categories of children, the disabled, and those with substantial caregiving demands—but also for struggling adults. Much more needs to be learned about the ways in which families do or do not provide such support. As our results indicate, the IPUMS is an extraordinary resource with which to address such questions.

Acknowledgements This research was supported in part by funds provided to the Maryland Population Research Center from the Eunice Kennedy Shriver National Center for Child Health and Human Devel- opment Grant R24-HD041041. The authors gratefully acknowledge the helpful comments from the anonymous reviewers. They also acknowledge the unpublished work by Goldscheider et al. (1994), which formed the conceptual basis for this article. Previous versions of this article were presented at the 2011 annual meeting of the Population Association of America, Washington, DC, and 2011 annual meeting of the Social Science History Association, Boston, MA.

References

Alwin, D. F., Converse, P. E., & Martin, S. S. (1985). Living arrangements and social integration. Journal of Marriage and the Family, 47, 319–334.

Bean, F., & Tienda, M. (1987). The Hispanic population of the United States. New York: Russell Sage. Burch, T., & Mathews, B. (1987). Household formation in developed societies. Population and Develop-

ment Review, 13, 495–512. Choi, N. G. (2003). Coresidence between unmarried aging parents and their adult children: Who moved in

with whom and why? Research on Aging, 25, 384–404. Cohen, P. N., & Casper, L. M. (2002). In whose home? Multigenerational families in the United States,

1998–2000. Sociological Perspectives, 45, 1–20. Costa, D. (1999). “A House of Her Own”: Old age assistance and the living arrangements of older

nonmarried women. Journal of Public Economics, 72, 30–39. Easterlin, R. (1978). What will 1984 be like? Socioeconomic implications of recent twists in age structure.

Demography, 15, 397–432. Furstenberg, F. F., Jr., Kennedy, S., McLoyd, V. C., Rumbaut, R. G., & Settersten, R. A., Jr. (2004).

Growing up is harder to do. Contexts, 3(Summer), 33–41. Glick, J., & Van Hook, J. (2002). Parents’ coresidence with adult children: Can immigration explain racial

and ethnic variation? Journal of Marriage and Family, 64, 240–253.

1474 J.R. Kahn et al.

Goldscheider, F., Biddlecom, A., & McNally, J. (1994, August). Dependency, privacy, and power in intergen- erational households: Changes in the living arrangements of the elderly in the U.S., 1940–90. Paper presented at the 1994 annual meeting of the America Sociological Association, Los Angeles, CA.

Goldscheider, F., & DaVanzo, J. (1989). Pathways to independent living in early adulthood: Marriage, semiautonomy, and premarital residential independence. Demography, 26, 597–614.

Goldscheider, F., & Goldscheider, C. (1987). Moving out and marriage: What do young adults expect? American Sociological Review, 52, 278–285.

Gustman, A. L., Steinmeier, T. L., & Tabatabai, N. (2010). What the stock market decline means for the financial security and retirement choices of the near-retirement population. Journal of Economic Perspectives, 24, 161–182.

Kobrin, F. (1976). The fall in household size and the rise of the primary individual in the United States. Demography, 13, 127–138.

Kochhar, R., Fry, R., & Taylor, P. (2011). Wealth gaps rise to record highs between whites, blacks, Hispanics (Pew Research Center, Social and Demographic Trends). Washington, DC: Pew Research Center.

Kotlikoff, L., & Morris, J. (1990). Why don’t the elderly live with their children? A new look. In D. Wise (Ed.), Issues in the economics of aging (pp. 149–169). Chicago, IL: University of Chicago Press.

Krivo, L. J., & Mutchler, J. E. (1989). Elderly persons living alone: The effect of community context on living arrangements. Journal of Gerontology, 44, S54–S62.

Levy, F. (1999). The new dollars and dreams. New York: Russell Sage Foundation. Lopata, H. (1973). Widowhood in an American city. Cambridge, MA: Schenkman. McGarry, K., & Schoeni, R. (2000). Social Security, economic growth and the rise in elderly widows’

independence in the twentieth century. Demography, 37, 221–236. Mutchler, J. E., & Burr, J. (2003). Living arrangements among older persons. Research on Aging, 25, 531–558. O’Rand, A. M., Ebel, D., & Isaacs, K. (2009). Private pensions in international perspective. In P. Uhlenberg

(Ed.), International handbook of population aging (pp. 429–443). Rueil-Malmaison, France: Springer Science + Business B.V.

Pampel, F. (1983). Changes in the propensity to live alone: Evidence from consecutive cross-sectional surveys, 1960–1976. Demography, 20, 433–447.

Pew Social and Demographic Trends. (2010). The return of the multi-generational family household (Report). Washington, DC: Pew Research Center. Retrieved from http://pewsocialtrends.org/2010/03/ 18/the-return-of-the-multi-generational-family-household

Preston, S. (1984). Children and the elderly: Divergent paths for America’s dependents. Demography, 21, 435–457. Robinson, J. G. (1988, April). Perspectives on the completeness of coverage of population of the United

States decennial censuses. Paper presented at the annual meeting of the Population Association of America, New Orleans, LA.

Ruggles, S. (2007). The decline of intergenerational coresidence in the United States, 1850 to 2000. American Sociological Review, 72, 964–989.

Ruggles, S. J., Alexander, T., Genadek, K., Goeken, R., Schroeder, M. B., & Sobek, M. (2010). Integrated public use microdata series: Version 5.0 [machine-readable database]. Minneapolis: University of Minnesota.

Santi, L. (1990). Household headship among unmarried persons in the United States, 1970–1985. Demog- raphy, 27, 219–232.

Schmertmann, C., Boyd, M., Serow, W., & White, D. (2000). Elder-child coresidence in the United States. Research on Aging, 22, 23–42.

Schoeni, R. (1997). Reassessing the decline in parent–child old-age coresidence during the twentieth century. Demography, 35, 307–313.

Speare, A., & Avery, R. (1993). Who helps whom in older parent–child families? Journal of Gerontology: Social Sciences, 48, S64–S73.

Tolnay, S. (1997). The Great Migration and changes in the northern black family, 1940 to 1990. Social Forces, 75, 1213–1238.

United Nations. (2005). Living arrangements of older persons around the world. New York: United Nations, Population Division, Department of Economic and Social Affairs.

U.S. Census Bureau. (2001). Preliminary estimates show improvement in Census 2000 coverage. Retrieved from http://www.census.gov/newsroom/releases/archives/census_2000/cb01cn03.html

Ward, R., Logan, J., & Spitze, G. (1992). The influence of parent and child needs on coresidence in middle and later life. Journal of Marriage and the Family, 54, 209–221.

White, L. (1994). Coresidence and leaving home: Young adults and their parents. Annual Review of Sociology, 20, 81–102.

Wiemers, E. (2012, May). The effect of unemployment on household composition and doubling up. Paper presented at the annual meetings of the Population Association of America, San Francisco, CA.

Growing Parental Economic Power in Parent–Adult Child Households 1475

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.