New Estimates of the Sandwich Generation in the 2013 Panel Study of Income Dynamics

Abstract

Purpose of the Study:

We use the Roster and Transfers Module in the 2013 Panel Study of Income Dynamics to obtain the first estimates of the prevalence of transfers to adult children and parents for United States men and women aged 35–75.

Design and Methods:

This article extends the current understanding of the sandwich generation by comparing recent transfers of time and money to parents and adult children for men and women and across ages between 35 and 75 years of age.

Results:

Over 30% of individuals with living parents and adult children provide transfers to two generations. The prevalence of transfers does not differ by age and the differences between men and women are small, though statistically significant. Conditional on providing time transfers, women provide more hours of help than men, particularly to their adult children. The number of hours given to children exceeds the number given to parents.

Implications:

These findings are the first to show that both men and women are likely to provide transfers to two generations and that transfers to two generations are common across adult ages. Our findings suggest a need to rethink the notion of the sandwich generation, which has focused on women in late middle age, to include men and women across younger and older ages.

Recent demographic trends toward increased longevity (Miniño, Murphy, Xu, & Kochanek, 2011) and young adults taking longer to transition to adulthood (Furstenberg, Rumbaut, & Settersten, 2005) have resulted in an increase over time in the number of people who may face competing demands on their time and finances from two generations of family (Wiemers & Bianchi, 2015). This so-called “sandwich generation” has received attention in the popular press but much less in the academic literature. The popular press has focused on the financial and emotional stress associated with giving money and time to children and parents particularly as middle-aged adults are working and saving for retirement (Hamm, 2014; Lieber 2010; Tarantine, 2014) but this evidence is anecdotal and there is still little current academic evidence describing the prevalence or consequences of providing support to multiple generations of family.

Evidence from the 1990s suggests that approximately 30% of women give transfers to two generations at the same time (Grundy & Henretta, 2006; Pierret, 2006; Parker & Patten, 2013). However, studies that estimate the size of this phenomenon use data from over 15 years ago and have typically had a narrow focus on women over a small number of age groups (Grundy & Henretta, 2006; Pierret, 2006). Work using more recent data on the sandwich generation has not examined differences by age and gender (Parker & Patten, 2013) and has only examined a limited set of transfers (Wiemers & Bianchi, 2015). To date, we know little about the contemporary prevalence of transfers of time and money to two generations across the United States population as a whole.

This article uses newly collected data in the 2013 Panel Study of Income Dynamics [PSID] to provide the critical first step of assessing the prevalence of support to multiple generations and the extent to which transfers flow upward and downward in the generations for both men and women and across a wide range of adult ages.

Design and Methods

Data and Measures

This article uses the Roster and Transfers Module in the 2013 PSID. The PSID began with a sample of roughly 18,000 people in 5,000 households in 1968 and has followed the original 1968 sample members as well as children who were born to or adopted by the original sample members, children of these children, and so on. This genealogical design implies that the study provides data on a sample of extended families at each wave. In each wave, the PSID includes information on respondent’s economic circumstances, demographic characteristics, health, and detailed consumption (see McGonagle, Schoeni, Sastry, & Freedman, 2012 for more information about the PSID).

The 2013 Roster and Transfers Module enhances the main PSID data by providing complete information on all of the parents, parents-in-law, and adult children of PSID respondents and their spouses—covering many individuals, such as in-laws, who are not included in the genealogical structure of the main PSID data (see Schoeni, Bianchi, Hotz, Seltzer, & Wiemers, 2015 for more information on the 2013 PSID Family Roster and Transfers Module).

In the module, respondents and their spouses (G1) were asked to report the name and key characteristics of all living parents, step-parents, and parents-in-law (G0) and all living children and step-children over age 18 (G2). We examine several of the characteristics reported by respondents (G1) about parents (G0) and adult children (G2) including children’s current work status (G2), from which we obtain the percent of children who are students, parental health (G0), a 5-point health scale which we collapse to poor health for individuals who report a parent in fair or poor heath, and parental family income (G0) in four brackets (<$25,000, $25,000–$50,000, $50,001–$75,000, $75,000+), which we collapse to income above or below $50,000.

As part of the module, respondents (G1) were asked about transfers of time and money to parents (G0) and adult children (G2) during the 2012 calendar year. Transfers of time include errands, rides, chores, babysitting, and hands-on care. Transfers of money include loans and gifts over $100. Individuals report whether a transfer was given and how many hours or dollars were given. Transfers of money are collected between one household and another but time transfers are collected at the individual level, with married/cohabiting respondents being asked whether time transfers were given mostly by the respondent, mostly by the spouse/cohabitor, or about equally. We use reports on who gave transfers to disaggregate hours provided by women and men. For each transfer in which the respondent reports equal giving, we allocate the hours to each member of the couple. For each transfer in which the respondent report that they (spouse) were mostly responsible we allocate the total number of hours to the respondent (spouse). We examine the prevalence, amount, and type of transfer to adult children and parents for men and women.

Sample and Methods

We begin with a sample of women and men aged 35–75 (4,688 women and 3,952 men). Because only individuals with a living parent and adult child are able to provide transfers to two generations, we begin by enumerating the population with at least one living parent or parent-in-law and at least one child over age 18 (2,246 women and 1,836 men). We show the prevalence of transfers to two generations among this population. For the remainder of our analyses, we restrict our sample to the 824 women and 606 men who engage in transfers with two generations. We describe the individual characteristics of these men and women as well as the characteristics of their parents and adult children, document the prevalence of time and financial assistance, and compare mean differences using t-tests. Finally, we describe the amount of transfers in terms of total dollars and total hours, disaggregating total hours into hours given mostly by men and mostly by women and compare means within groups. Tables show the results of t-tests of mean differences between men and women. All tables are weighted using the 2013 PSID individual cross-sectional survey weights to account for differential probabilities of selection into the sample and attrition (We used Version 3 of the Roster and Transfers data and conducted all analyses using Stata 13.).

Results

Table 1 shows the proportion of the population with living parents and adult children for men and women in three age groups 35–49, 50–64, and 65–75.

Table 1.

Percent of Women and Men with Children and Parents, by Age (PSID 2013 Family Roster & Transfer Module)

	Women				Men
	Overall	35–49	50–64	65–75	Overall	35–49	50–64	65–75
Both	44.9	41.6	59.0	17.7	44.3	31.0***	61.9†	32.1***
Child(ren) Only	26.9	3.6	28.7	75.7	20.9***	2.7	21.9***	61.6***
Parent or In-law only	24.6	52.9	8.0	0.9	31.1***	63.8***	11.7**	1.4
None	3.6	1.9	4.4	5.7	3.8	2.5	4.5	5.0
Married (%)	67.6	70.5	67.7	60.9	77.0***	75.4**	77.3***	80.3***
N	4,688	2,106	2,008	574	3,952	1,768	1,658	526

Notes: Weighted using 2013 individual weights. Unweighted N. †p < .10; *p < .05; **p < 0.01, ***p < 0.001.

Overall 45% of women and men have at least one living parent or parent-in-law and at least one child over 18. These probabilities vary predictably with age and the likelihood of having living parents and adult children reaches its peak between ages 50 and 64. There are important gender differences in the sample as well, with more older men and younger women in the population of individuals with parents and adult children. To our knowledge, contemporary information on living parents and children is not available in any other nationally representative survey for the full range of adult ages that we cover. Because information on living parents and children is not available in any other survey, it is difficult to directly assess the population representativeness of the sample of men and women with adult children and living parents in the PSID. However, Wiemers and Bianchi (2015) show that the demographic characteristics of a similar sample match Census, Current Population Survey, and National Survey of Households and Families quite closely. The PSID has also been shown to be population representative on a broad range of characteristics producing comparable estimates to the American Time Use Survey for time use behaviors, to the National Health Interview Survey for health status and health behaviors, and to the Current Population Survey for income (McGonagle et al., 2012).

Table 2 shows the transfers given to parents and adult children by men and women with adult children and living parents. Over 30% of these men and women provide transfers to two generations. Men are slightly more likely to provide transfers to parents only, while women are more likely to provide to both parents and children. Differences between men and women in the prevalence of providing transfers to two generations are small but statistically significant. While the prevalence of transfers to parents increases and to adult children declines with age, transfers to two generations are surprisingly stable across adult ages with a jump in the percent giving to two generations in middle age. As we noted above, the population of individuals with living parents and adult children takes an inverted U-shape with age so a higher proportion of men and women in late middle age give transfers to two generations than any other age group. The 824 women and 606 men who provide transfers to two generations make up the analytic sample for what follows.

Table 2.

Transfers to Multiple Generations Among Women and Men with Children and Parents (PSID 2013 Family Roster & Transfer Module)

	Women				Men
	Overall	35–49	50–64	65–75	Overall	35–49	50–64	65–75
Transfers to
Parents only	11.4	10.3	11.2	19.5	12.5	13.1†	12.0	13.7
Children only	32.3	33.9	32.3	23.8	31.6	32.9	32.5	23.9
Neither	20.8	21.9	19.4	26.8	23.3**	23.6	20.9	34.0
Both	35.5	33.9	37.1	30.0	32.7**	30.4	34.6	28.4
N	2,246	983	1,155	108	1,836	622	1,029	185

Notes: Weighted using 2013 individual weights. Unweighted N. †p < .10; *p < .05; **p < 0.01, ***p < 0.001.

Table 3 shows characteristics of the individual (G1), the parent (G0), and the adult child (G2) for the men and women providing transfers to two generations. On average, men are two years older and more likely to be married than women, but men and women have approximately the same number of children and parents. Characteristics of the parent generation are also quite similar and characteristics of the adult child generation are nearly identical for men and women.

Table 3.

Characteristics of Women and Men with Children and Parents Who Are Giving Transfers (PSID 2013 Family Roster & Transfer Module)

	Women	Men
Individual (G1)
Age group
35–49	35.1	26.4***
50–64	59.1	62.8
65–75	5.8	10.8**
Age	52.5	54.3***
Married or cohabiting	75.0	84.7***
# Children	2.3	2.2†
# Parents/in-laws	1.6	1.6
Parents (G0)
Age oldest	80.3	80.7
Married or cohabiting	48.4	51.3
At least 1 Par Inc<50K	84.4	84.7
At least 1 Par poor health	25.4	24.9
Children (G2)
Age youngest	24.7	24.7
Married or cohabiting	57.7	56.7
# Children (G3)	1.9	1.7
At least 1 child student	30.8	31.1
At least 1 child Inc <50K	88.4	87.2
Sample size	824	606

Notes: Weighted using 2013 individual weights. Unweighted N. Percentages based on nonmissing data. †p < .10; *p < .05; **p < 0.01, ***p < 0.001.

Table 4 examines the distribution of types of transfers to parents and adult children among those providing transfers to two generations. The rows do not sum to 100% because individuals are included in multiple categories of transfers. Overall, women are more likely to provide time and men are more likely to provide money to parents and adult children. Conditional on giving to two generations, women are 8% points more likely to provide time to parents and adult children than their male counterparts while men are approximately 5% points more likely to give money to parents and adult children.

Table 4.

Type of Transfers Among Women and Men with Transfers to Both Children and Parents (PSID 2013 Family Roster & Transfer Module)

	Women				Men
	Overall	35–49	50–64	65–75	Overall	35–49	50–64	65–75
Time to parent	79.5	79.9	79.6	77.2	71.5***	73.0*	72.6*	61.1
Time to child	72.7	74.9	71.9	67.8	65.2**	65.8*	66.6	55.9
Money to parent	42.6	47.1	40.0	43.0	47.0*	54.4†	43.6	48.6
Money to child	76.2	78.9	75.0	72.5	81.7*	87.7*	78.4	86.6
N	824	374	415	35	606	210	347	49

Notes: Weighted using 2013 individual weights. Unweighted N. †p < .10; *p < .05; **p < 0.01, ***p < 0.001.

In Table 5, we investigate the hours and dollar amounts of transfers. Table 5 shows the average hours and dollars given to parents and adult children conditional on providing a given type of transfer, distinguishing hours between those given mostly by men (women) and those given about equally. Overall, men and women are giving more hours of time and more money to adult children than to parents. There are also large gender differences in time given to parents and adult children. Overall and for each age group, women give more time than men. On average, women give 45 more hours per year to parents and 140 more hours per year to adult children than their male counterparts. Differences in time given to adult children are statistically significant. The gender differences grow with age and are accounted for almost entirely by additional hours that women provide mostly by themselves. Men spend relatively little time providing to adult children by themselves, particularly at older ages and differences between men and women are statistically significant.

Table 5.

Amounts of Transfers Among Women and Men with Transfers to Both Children and Parents (PSID 2013 Family Roster & Transfer Module)

	Women				Men
	Overall	35–49	50–64	65–75	Overall	35–49	50–64	65–75
Time to parent (hr/year)	315.8	319.5	306.2	393.3	271.3	300.6	202.0	664.7
Mostly woman (man)	243.3	220.5	258.1	229.9	176.9	149.3	139.8†	513.6
All about equally	72.5	99.0	48.1	163.4	94.4*	151.3	62.2	151.1
Time to child (hr/year)	562.0	676.0	498.8	482.6	422.2*	635.1	347.9†	323.8
Mostly woman (man)	320.5	378.8	283.9	326.3	111.8***	174.5*	89.4*	85.6
All about equally	241.5	297.1	214.9	156.3	310.4*	460.6*	258.4	238.1
Money to parent ($/year)	1,321	1,068	1,489	1,410	1,410	1,147	1,573	1,273
Money to child ($/year)	5,088	3,334	5,816	8,960	4,668	2,895	5,514	4,598

Notes: Weighted using 2013 individual weights. Unweighted N. †p < .10; *p < .05; **p < 0.01, ***p < 0.001.

Discussion

Prior work examining the sandwich generation has focused on women in late middle age (Grundy & Henretta, 2006; Henretta, Grundy, & Harris, 2001; Pierret, 2006; Wiemers & Bianchi, 2015). Using newly released data from the 2013 PSID, we provide the first evidence on providing help to parents and children simultaneously comparing men and women across adult ages. We find that over 30% of individuals aged 35-75 with adult children and living parents give either time or money to two generations and, in sharp contrast to the focus on women in the literature, we find that women are only slightly more likely than men to give to two generations though differences are statistically significant. These results are consistent with the broader literature on transfers which suggests that both men and women have high levels of involvement in providing time and money help to parents and adult children (Chesley & Poppie, 2009; Kahn, McGill, & Bianchi, 2011; Van Houtven, Coe, & Skira, 2013). Our estimates on transfers to two generations for women are similar to other estimates from the NLS-Young Women (Pierret, 2006) and the HRS (Grundy & Henretta, 2006). To our knowledge, we are the first to examine transfers to two generations by men. Gender differences in the prevalence of transfers though small, are statistically significant. But, conditional on giving, men provide different types of transfers than women which include more money and less time. Men also provide fewer hours, especially to children. We further elaborate on past work by examining a broader age range and find that the prevalence of transfers to two generations is nearly constant across adult ages though the population with living parents and adult children is highest in late middle age. Our findings by gender and age suggest that the concept of who might be in the “sandwich generation” should be broadened considerably beyond the middle-aged women traditionally examined in the literature.

There are several limitations to our analysis. We are unable to distinguish time help for chores with time help for personal care and babysitting. Prior research has shown that there are gender differences in the type of time provided (Kahn et al., 2011; Van Houtven et al., 2013). In addition, we have no information on emotional support to parents and children or on the potential for transfers in times of need. Despite these limitations, these data are one of the few sources that permit an analysis of giving over a broad range of ages and by gender, which our work shows is critical for getting a complete picture of the sandwich generation.

Understanding that competing demands are prevalent across adults ages for both men and women is important because, as the popular press and academic literature on transfers and caregiving suggests, helping family members may come with financial, employment, and health consequences (Barnett, 2014; Bittman, Hill, & Thomson, 2007; Feinberg, Reinhard, Houser, & Choula, 2011; Lee, Tang, Kim, & Albert, 2015; Van Houtven et al., 2013). However, we acknowledge that these processes are complex because help in families often runs in multiple directions and transfers may also enhance well-being (Arrondal and Mason, 2001; Bengtson, 2001; Bryson and Casper, 1999; Kahn et al., 2011; Pebley and Rudkin, 1999; Wang and Marcotte, 2007). Future work remains to understand the motivations underlying these giving behaviors, the multiple directions of transfers within families, how couples negotiate giving within the family, and the consequences of providing transfers to multiple generations for the health and well-being of both men and women over the life course.

Funding

This work was supported by the Alfred P. Sloan Foundation (2011-6-24) and the National Institute on Aging (P01AG029409). The authors also acknowledge administrative support from the California Center for Population Research at UCLA which receives core support from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R24-HD041022) and training support from the National Institute on Aging (T32-AG033533).