Parental Income and Wealth Loss and Transfers to their Young Adult Children

Abstract

As young people transition to adulthood, many rely on financial support from their parents to complete schooling and to live independently. Evidence suggests that there has been a gradual lengthening of the time young adults take to transition to adulthood. Young people attempting to move out of their parents’ home, complete college, or enter the workforce during the Great Recession faced uncertain economic times, increasing their need for financial support. At the same time, the income and wealth losses experienced by young adults’ parents may have disrupted transfers from them. We analyze the impact of large and unexpected declines in parents’ income and wealth during and immediately after the Great Recession on monetary transfers to their young adult children using data from the Panel Study of Income Dynamics (PSID) and the PSID Transition to Adulthood study. We find parents’ financial support of their young adult children declined during the Great Recession. The likelihood of receiving a transfer declined from 74% in 2005 to 57% in 2009. Parents’ loss of income was a factor in the amount of decrease but on average was relatively modest – a $10,000 parental income loss decreased transfers to their adult children by $109. However, parents experiencing large declines in income, those at the 75^th and 95^th percentile of income loss, reduced transfers to adult children by $1,150 and $1,700, respectively. Declines in parental transfers that reduce college completion rates, increase student loan debt and decrease likelihood of homeownership may have long term consequences for financial well-being.

It is widely recognized that the Great Recession, which began in December 2007 and officially ended in June 2009, exerted a profound impact on economic wellbeing in the United States (National Bureau of Economic Research, 2010). Data in the US from the Bureau of Labor Statistics (2017) showed that between 2007 and 2010, the unemployment rate for the non-institutionalized civilian population ages 16 and older increased by 5.0 percentage points – peaking at over 10% in 2009 - and did not recover to pre-recession levels until 2017. At the same time, stock market and house values plummeted (S&P Dow Jones Indices, 2017). These changes resulted in significant decreases in wealth and income for many Americans. Data from the Survey of Consumer Finance showed median net worth of families declined by 39% and median value of real income fell by 7.7% (Bricker, Kennickell, Moore, & Sabelhaus, 2012).

The large declines in household income and wealth during the Great Recession may have caused parents to reduce financial support to adult children. Many young adult children between ages 18-24 are not yet fully independent, have not completed the transition to adult roles in families, households, or the workforce, and rely on parents for financial support. Parental financial support during this time for education (Brown, Schulz, & Seshadri, 2006; Keane & Wolpin, 2001; Lovenheim 2011; Lovenheim & Reynolds, 2013; Zissimopoulos & Smith, 2010), home-buying (Engelhardt & Mayer, 1998; Guiso & Japelli, 2002; Lee, Myers, Painter, Thunell, & Zissimopoulos, 2018) and consumption (Rosenzweig & Wolpin, 1993) can have important short and long-term consequences for the wellbeing of these young adults.

Although there is a robust and growing literature on the impact of the Great Recession on households and families, only a few studies have examined how changing circumstances during the Great Recession impacted intergenerational family transfers (Cox & Way, 2011; Gottlieb, Pilkauskas & Garfinkel, 2014) and none have examined the impact of unexpected declines of parents’ income and wealth on their transfers to children at the critical juncture of transitioning to adulthood. In this study, we analyze the relationship between parental financial transfers to their young adult children and parents’ income and wealth from 2005 to 2011. We use rich panel data on transfers received by young adults from the Panel Study of Income Dynamics Transition to Adulthood Study (PSID-TA) linked with data from the main survey on their parents’ income, wealth, and other socioeconomic and demographic characteristics. The descriptive analysis highlights changes in transfers, parents’ resources, and the local economy that took place during the Great Recession and multivariate regression quantifies the effect of changes in parents’ income and wealth on changes in transfers to their young adult children. We exploit the largely unexpected nature of income and wealth losses experienced during and immediately after the Great Recession in our estimation strategy. We analyze how other types of assistance, specifically shared housing, change over this period to inform our results on financial transfers.

The data show about 74% of young adults ages 18 to 27 received financial transfers from parents in 2005 and 69% in 2007 – the years before the Great Recession. By 2009, the percentage of young adults receiving transfers fell 12 percentage points to 57% and another percentage point to 56% in 2011. After rising between 2005 and 2007, median transfer amount declined by $1,217 (from $5,642 to $4,425) between 2007 and 2009. Median transfer amount in 2011 fell further to $3,100. Our main empirical specification relates changes in financial transfers to changes in parents’ income and wealth removing the effect of unobserved time invariant characteristics from the estimates. Rich controls for changes in adult children’s circumstances isolate changes in parent’s resource from these potentially correlated changes. We find income loss was associated with a decline in financial transfers to young adult children, while wealth loss had no impact on financial transfers. Specifically, a $10,000 decline in parental income led to a $109 decrease in transfers to young adult children; however, large income losses, at the 75^th and 95^th percentile, were associated with transfer declines of $1,150 and $1,700, respectively. Income loss was not associated with an increase in co-residency among young adults and their parents.

Background

Theoretical Framework

Altruism and exchange motivations provide two main explanations for why parents provide financial transfers to children. Altruistic parents gain utility from providing financial support to their children through increasing the child’s welfare (Barro, 1974; Becker, 1974, 1991). Under the altruistic model, parents’ income and wealth are positively associated with both the probability of transfers and the amount of positive transfers, while children’s income and wealth have a negative effect on transfers (Cox & Rank, 1992). An alternative theory motivates financial transfers as an exchange for the provision of care to elderly parents (Bernheim, Shleifer, & Summers, 1985; Cox, 1987; Feinerman & Seiler, 2002). Under the exchange theory, parental resources are positively related to transfer probability, but the child’s income and wealth increase the “price” of her services and, thus, increase positive transfer amounts (Cox & Rank, 1992).

Much of the early empirical work on intergenerational transfers focused on testing alternative hypotheses for parents’ motivation for supporting their children. A positive association between a child’s need and parental transfers, consistent with altruism, was found; however, the effects sizes were smaller than the altruistic theory would predict (Altonji, Hayashi, & Kotlikoff, 1997). In contrast, other studies find a positive association between a child’s economic resources and the amount of transfers she received, which is more consistent with the theory of exchange (Cox, 1987; Cox & Rank, 1992). Results from some studies do not conform with either theory, such as those finding equal bequests among children (Hurd & Smith, 2002; McGarry, 1999; Menchik, 1980).

The goal of the present study is to estimate the effect of unexpected changes in parental resources on transfers to their young adult children. Our focus is on the impact of parents’ resources, thus we motivate our empirical models using a life cycle model of consumption (Yarri,, 1965) where transfers are a form of consumption, as described in Hurd, Smith, and Zissimopoulos (2011).¹ This model is agnostic about the motivation for transfers and can accommodate either altruistic or exchange motives. It has the advantage of generating clear predictions about the effect of parental resources on transfers. Specifically, expected changes in income or wealth should not have an effect on transfers (or any form of consumption); however, unexpected changes in parental income and wealth will affect financial transfers to children.² That is, transfers are expected to decline with unexpected decreases in income and wealth and previously positive transfers may become zero. Many parents experienced large and unexpected shocks to their income and wealth as a result of the Great Recession. The life cycle model predicts consumption, including transfers, is reduced in response to permanent shocks, but may not decrease if the decline in resources is perceived as transitory. Thus, the effect of shocks to parental resources on transfers to their young adult children is an empirical question and will depend on the permanent or transitory nature of the unexpected losses in income and wealth. Although a life cycle model of consumption applied to transfers does not explicitly generate predictions about the effect of a change in children’s resources on transfers, the empirical literature on intergenerational transfers demonstrates that a child’s need influences transfer decisions. Moreover, shocks to parents’ resources may be correlated with changes in children’s needs. Thus in empirical models, we control for a rich set of child characteristics and factors that may influence a young adult child’s need for transfers including marriage, schooling, employment and earnings.

Transfers Across Generations of Family Members

Empirical studies consistently find that parents’ level of resources influences the amount of money transferred to their young adult children (Altonji et al., 1997; Cox & Rank, 1992; Hurd et al., 2011; Schoeni & Ross, 2005; Zissimopoulos & Smith, 2011). Descriptive and inferential analyses reveal higher income parents tend give larger amounts of money to their children and transfer receipt decreases with the child’s age (e.g. Schoeni & Ross 2005; Zissimopoulos & Smith 2011). Early cross-sectional studies finding a positive relationship between transfers and parents’ resources, however, could not account for unobservable factors that may affect both parents’ transfer behavior and their income and wealth (e.g. Cox, 1987; Cox & Rank, 1992). For example, Keane and Wolpin (2001) assume parents draw on a “pre-determined transfer rule” in transfers for educational attainment. To account for unobserved heterogeneity across families or individuals, several studies use fixed effects or first difference models. Using fixed effects models, Lovenheim and Reynolds (2013) find increases in home equity (i.e. housing wealth) led to an increase in college attendance, presumably through more financial assistance from parents although transfers are not directly measured.

Transfers from parents to children may be especially important during the child’s transition to adulthood. Settersten and Ray (2010) discuss the gradual lengthening over recent decades of the time young adults take to transition to adulthood – complete schooling, live independently, enter the workforce. The authors argue financial support from parents has become increasingly important during the young adult’s transition period, particularly in the United States where there is emphasis on the importance of private markets and personal responsibility. Indeed, Sironi (2018) finds young adults in the US, young men especially, suffered greater declines in full-time employment and increases in low-paid work compared to countries with more generous welfare states (Norway and Germany). Empirical research demonstrates that young adults often rely on material assistance from their parents for several key milestones during the transition to adulthood, such as buying a home (Engelhardt & Mayer, 1998; Guiso & Jappelli, 2002; Lee et al., 2018) or attending college (Keane & Wolpin, 2001; Lovenheim, 2011; Lovenheim & Reynolds, 2013). While this line of research points to the importance of transfers during the transition period, most of these studies do not measure financial transfers directly (Lovenheim, 2011; Lovenheim & Reynolds, 2013) or analyze the effect of transfers on other outcomes, such as housing (Engelhardt & Mayer, 1998; Guiso & Jappelli,2002).

Impacts of the Great Recession on Individuals, Families and Households

A growing body of literature documents the impact of the Great Recession on household decisions and behavior along several dimensions, including household formation (Lee & Painter, 2013), having children (Astone, Martin, & Peters, 2015; Cherlin, Cumberworth, Morgan, & Wimer, 2013), spending and consumption (Christelis, Georgarakos, & Jappelli, 2015; Hurd & Rohwedder, 2011) and retirement (McFall, 2011). For example, Lee and Painter (2013) use multivariate analysis and show the likelihood of forming a new household decreases by 3 to 9 percentage points during a recession, particularly among young adults, ages 21 to 29.

Several studies draw on the largely unexpected changes to employment, income and wealth of the Great Recession to isolate the effects of the Great Recession from other factors. For example, Christelis et al. (2015) found that becoming unemployed reduced spending by 10%, while wealth losses during the same period had a negative, but much smaller, effect. McFall (2011) found unexpected wealth losses due to the Great Recession delayed expected retirement by 2.5 months among working adults ages 45 and older. Hyclak, Meyerhoefer, and Taylor (2015) found that middle aged and older men and women who lost their jobs or had their work hours reduced during the recession delayed medical procedures and filling prescriptions, and were at greater risk of depression and anxiety. The empirical evidence on the Great Recession describes and quantifies the effects of economic crises across a broad range of health and economic outcomes. Yet, the impact on the financial support provided by parents to young adult children remains largely unexplored in the empirical literature.

Few studies have examined parental financial transfers to their adult children during the Great Recession. Cox and Way (2011) found a positive association between becoming unemployed and receiving a private financial transfer, which was higher among individuals whose family members were unaffected by the recession. Gottlieb et al. (2014) focused on families with small children, finding that high unemployment increased transfers from family and friends to low-income families; however, their research did not include data on those transferring money and, thus, could not measure the impact of the givers’ changing circumstances on their behavior. Related, but not measuring financial transfers directly, Friedline, Nam, and Loke (2013) found a large negative effect on the personal savings of young adults ages 21 to 25 from families with the largest declines in net wealth during the recession.

Housing in the form of co-residency may be another form of help that parents provided in response to the shocks of the Great Recession. Several studies document an increase in adult children’s co-residence with parents and alternative living arrangements during the Great Recession (e.g. Cherlin et al., 2013; Weimers, 2014). Cherlin et al. (2013) found a significant effect of the recession on the likelihood of an adult ages 25 to 34 living with a parent, particularly among Hispanics and married people. Bitler and Hoynes (2015) found a negative (albeit small) effect of the unemployment rate on independent living; however, the co-residence patterns in their sample were not significantly different during the Great Recession than they were during recessions in the 1980s. Wiemers (2014) exploited changes in employment due to the recession and found that becoming unemployed almost tripled the likelihood an individual 25 years or older moved into a household with another adult (i.e. friend, family, etc.).

In sum, empirical literature to date has established the profound impact of the Great Recession on the U.S. population. The small body of research on parental financial transfers to adult children during the Great Recession suggests that parental support to young adults may have increased during the Great Recession. Yet, none of the studies quantified the impact of large changes to parents’ income and wealth on financial transfers to young adult children. No study focuses on the group of young people transitioning to adulthood, who may be particularly vulnerable to sustained long-term effects of the short-term shocks of the Great Recession. We contribute to the literature by addressing these gaps. We use panel data and exploit the large, unexpected decreases in parental wealth and income during and immediately after the recession to estimate the effect of income and wealth loss on transfers to young adult children.

Empirical Strategy

Data

We use nationally representative, longitudinal data on young adults, ages 17 to 27, and their parents from four waves (2005-2011) of the Panel Study of Income Dynamics (PSID) and the Transition into Adulthood (PSID-TA, hereinafter TA) study. The TA sample draws from children interviewed as part of the PSID Child Development Survey and follows them through young adulthood. The TA survey focuses on issues and questions central to young adults as they transition out of their parents’ care. Participants enter the study when they age into it (typically at 18) and remain in the sample until they age out of it (typically at age 26).

Participants in the TA sample provide detailed information on their time use, skills, responsibilities, education, employment, income and wealth, marriage and relationships, health and health behaviors, social environment, religiosity, aspirations, and subjective wellbeing. Relevant to this study, the young adults are asked whether they received financial support from parents or other relatives over the last year and its level.³ They are also asked about the purpose of the support: for rent or mortgage, tuition, vehicle, expenses or bills, personal loans, inheritances, and other large gifts.⁴ Respondents report the amount they received for each type of transfer.

We restrict our sample to child observations for which we have reliable parent matches, which excludes a total of 279 observations over all four waves.⁵ We also exclude those with very large transfers (above $175,000), 5 observations total. Our sample includes 2,070 children from the TA sample that participated in at least one of the four TA waves. The first year of the study, 2005, contains 793 observations, representing 713 unique children.⁶ By 2011, there are 1,950 observations representing 1,791 unique children. Each wave includes a significant number of children from the prior wave. In 2007, 63% of the observations are carried over from 2005, by 2011 repeated observations account for 78% of the total observations. Our final pooled sample consists of 5,537 child-year observations.

Data on the children in the TA sample is matched with their parents’ information in the main PSID and the restricted Geocode file. From these two datasets, we obtain information on parents’ wealth, income, education, health, and marital status. In our descriptive analysis, we augment the PSID data with annual median house value from the U.S. Census Bureau’s American Communities Survey (ACS) 1-Year Estimates and annual unemployment rate from the Bureau of Labor Statistics Local Area Unemployment Statistics (LAUS). Both sources provide data at the county level, which we match to the parents’ county of residence in PSID and use to describe changing local economic conditions.⁷ We discuss our key variables of interest and estimation strategy in the remaining part of this section.

Dependent Variables.

We analyze transfer likelihood and transfer amount. Transfer likelihood is constructed using data from the TA survey about whether the child received a transfer for any of the following over the last year: rent or mortgage, tuition, vehicle, expenses or bills, personal loans, inheritances, and other large gifts. The dichotomous variable equals one if the child received any type of transfer in the survey year and zero otherwise. Transfer amount is the sum of the dollar amount of each type of transfer, standardized to 2011 dollars.

Independent Variables.

Our primary variables of interest are parents’ income and wealth. Income and wealth measures were taken from the PSID family files for the parents’ households. Income is constructed using total household income from all sources (wages, business income, dividends, welfare, and child support). The wealth measure represents all forms of wealth (businesses, bank accounts, real estate, stocks, vehicles, and annuities/IRAs) net of debt and home equity.⁸ Income and wealth are standardized to 2011 dollars. In multivariate models, we include income as the natural log of total parental income. Due to the presence of negative values in the wealth variable, we transform parental wealth using inverse hyberbolic sine.⁹ Control variables include parents’ educational attainment, marital status, and health status, and characteristics of young adult children, student status, work status, earnings (2011 dollars), age, educational attainment, marital status, and gender. We include variables to account for the child’s need for transfers, including student and employment status. Young adults in the sample are identified as students if they were a full-time or part-time student at the time of the interview. Children are considered employed if they were working, full-time or part-time, at the time of the interview. Student and employment status are not mutually exclusive. Individuals can be students and/or workers in each sample year. We control for children’s age (using two-year age groups) due to differences in transfers associated with different parts of the young adult lifecycle.

Methods

We first estimate models of whether a transfer was received using a linear probability model and, conditional on transfer received, amount of transfers using ordinary least squares. We pool waves of data from 2005 to 2011.¹⁰ Independent variables and control variables are described above. Robust standard errors are clustered at the individual (child) level to account for correlation in the error term. However, unobserved factors that are correlated with the independent variables of interest, parents’ income and wealth, will produce biased estimates. We estimate changes in parents’ income on wealth on changes in transfers using first difference (FD) models to eliminate bias of unobserved factors correlated with changes in income and wealth and fixed over time. Under the assumption that the changes to income and wealth from the Great Recession were unexpected, the estimates are unbiased.¹¹ We estimate change in transfers received by a young adult child as a function of changes in parents’ income and wealth only (Model 1) and controlling for other characteristics of the parents, e.g. health changes and marital status (Model 2). However, estimates from the model may still be biased by time-varying characteristics of the adult children that are correlated with changes in parent’s income and wealth, thus we also include these as controls in regression models. We operationalize income and wealth changes with large values being larger losses. Limitations of these approaches and alternative approaches are addressed in the Discussion section.

Results

Descriptive Analysis

The descriptive analysis highlights changes in transfers, parents’ resources, and the local economy that took place during the Great Recession. Table 1 displays transfer likelihood and transfer amount for years 2005 to 2011. Transfers are common and the amounts are significant. Just under 62 percent of young adults receive a transfer from a parent and the average annual amount is $10,402. Transfer amounts are skewed and median amounts are about $4,200. Both transfer likelihood and amount declined over the period. The likelihood of a child receiving a transfer dropped from 73.5% in 2005 to 56.2% in 2011. Between 2007 and 2009, the year just preceding the recession and the year of the recession, there was a large decline, 12 percentage points. Median transfer (conditional on receiving a transfer) increased in the years before the recession, by $1,265 between 2005 and 2007, and declined between 2007 and 2009 by $1,217 and between 2009 and 2011 by another $1,325.

Table 1.

Transfer likelihood and conditional transfer amount by year (2011 dollars), 2005-2011

	Transfer Likelihood (%)	Mean Transfer Amount (Conditional)	Median Transfer Amount (Conditional)
All years	61.5	10,402	4,231
2005	73.5	10,371	4,377
2007	68.8	11,177	5,642
2009	56.8	10,711	4,425
2011	56.2	9,547	3,100

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys.

Note: Includes all children in the sample. Weighted results.

Our sample includes young adult children ages 17 to 27 (with the majority ages 18 to 26). Young adults may experience several different transitions during those ages, such as going to and graduating from college, and entering the workforce, that may impact both their needs and their parents’ giving over time. Thus, we describe transfer likelihood and amounts holding age constant and results are shown in Figure 1. As seen in Panel a, the likelihood of receiving a transfer declines with age. Notably, transfers for children in all age groups declined between 2007 and 2009, the height of the Great Recession. Panel b shows median transfer amount by age group. Median transfer amount declined during the recession years with the exception of children ages 20 to 21. This is an age at which students are in or are graduating from college and there may be a change in the composition of young adults entering and graduating from college over this period.

Fig. 1.

Transfer Likelihood and Median Transfer Amount (Conditional on Transfer Receipt)

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys.

Note: All dollar figures converted to 2011 dollars. Weighted results.

Parents may substitute shared housing, co-residence, for financial support to their young adult children. Table A1 in Appendix 2 displays the likelihood a young adult in our sample co-resides with his/her parents. Like financial transfers, the likelihood a young adult lives with his/her parents varies greatly with age, decreasing as the child gets older. Holding age constant, there are distinct patterns of co-residence in our sample. Two age groups (17 to 19 and 22 to 23) experienced dips in co-residence during the Great Recession, between 2007 and 2009; while those ages 20 to 21 saw an overall increase.

Table 2 displays mean values for the variables in our analysis by transfer receipt. Consistent with the literature, we find parents that gave money to their children are different than those that did not on most dimensions. Parents who gave have higher income and wealth, are more educated and healthier than those who did not give financial support to young adult children. Wealth is nearly 2.25 times higher and income is about 1.5 times higher among parents whose adult children received money relative to those that did not. Children that receive money from their parents are different than those that receive no financial assistance. In general, children receiving financial transfers from their parents are younger, more likely to be white than from a different racial/ethnic group, more likely to be in school, and less likely to be employed.

Table 2.

Variable Means Over Transfer Likelihood, 2005-2011

	Overall Mean	No Transfer	Transfer
Dependent Variables
Transfer likelihood	0.62
Transfer amount (conditional)	10,402
Parent Characteristics
Wealth (mean)	352,053	200,083	449,672
Income (mean)	102,995	78,746	118,801
Parent age	49.74	49.21	50.08
Married/partnered	0.68	0.61	0.72
Education
Less than 12 years	0.13	0.19	0.10
12 years	0.31	0.37	0.27
13-15 years	0.27	0.27	0.27
16 or more years	0.26	0.15	0.32
Head health status
Excellent	0.17	0.14	0.20
Very Good	0.35	0.34	0.32
Good	0.33	0.35	0.32
Fair	0.12	0.15	0.11
Poor	0.03	0.03	0.03
Child Characteristics
Age
17-19	0.34	0.23	0.41
20-21	0.28	0.26	0.30
22-23	0.20	0.23	0.18
24-25	0.13	0.20	0.08
26-27	0.05	0.08	0.03
Race
NH White	0.64	0.56	0.69
NH Black	0.16	0.22	0.12
Hispanic	0.16	0.19	0.14
Other	0.04	0.03	0.05
Female	0.51	0.50	0.52
Married	0.09	0.15	0.05
Employed	0.59	0.68	0.52
Earnings	10,365	14,570	7,737
Student	0.45	0.25	0.57
Number of Siblings	1.810	2.046	1.662
N	5,537	2,232	3,305

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys.

Note: differences in means for all covariates are statistically significant at the 5% level, except for child gender. Wealth is the sum of business, bank accounts, real estate, stocks, vehicle, annuities and IRAs less debt, plus home equity (PSID calculations). Income is sum of wages, business income, dividends, welfare, and child support (own calculation). Wealth, income, and earnings are standardized to 2011 dollars. Weighted results.

Figure 2 shows mean and median income and wealth of the parents over time alongside median house value and mean unemployment rate in a parent’s county over time. The latter two measures represent broader economic changes taking place in the parents’ county due to the impact of the Great Recession. Panel a shows the downward trend in mean and median parent wealth over time, and the declines in median house value over the same period. In Panel b, we show unemployment increased sharply in parent’s county of residency between 2007 and 2009 and remained high through 2011. Mean and median income declines appear to lag behind unemployment changes; however, income is reported over the prior year, so 2009 income reflects income between 2008 and 2009 and the lag is not unexpected. Both mean and median income decline between 2009 and 2011. The figure highlights the significant local economic changes experienced by parents in our sample and how they relate to changes in parental resources. Multivariate analysis examines whether these economic shocks resulted in fewer transfers to young adult children.

Fig. 2.

Mean and Median Parent Income and Wealth, Unemployment Rate, and Median House Value, 2005-2011

Source: 2005-2011 PSID and PSID-TA Surveys; 2005-2011 ACS 1-year estimates; 2005-2011 LAUS.

Multivariate Analysis of Parental Resources and Transfers to Young Adult Children

Table 3 reports the results from models using OLS estimation.¹² Consistent with the literature, and predictions from our theoretical model, we find a positive and statistically significant relationship between parents’ current resources and their giving. We find, all else equal, a 10% increase in income is associated with a 0.002 percentage point increase in the likelihood a young adult child receives a transfer. We also find that a 10% increase in income is associated with a $192 increase in the amount a young adult received from her parent conditional on a transfer being made. We find no statistically significant relationship between parents’ wealth the likelihood a young adult child received a transfer, but parental wealth is positively associated with amount of transfers, conditional on transfer receipt. ¹³

Table 3.

OLS Models: Transfer Likelihood and Transfer Amount Conditional on Transfer Receipt

	Transfer Likelihood	Transfer Amount (Conditional)
Parent Characteristics
Wealth (IHS)	0.002	96.3 *
Income (LN)	0.023 **	1,918.4 **
Education (ref. less than 12 years)
12 years	0.030	853.2
13-15 years	0.044	15.8
16 or more years	0.130 **	4,084.2 **
Child Characteristics
Age Groups (ref. 17-19)
20-21	−0.066 **	2,191.9 **
22-23	−0.099 ***	1,139.1
24-25	−0.243 ***	−3,933.0 **
26-27	−0.290 ***	−2,360.5
Employed	−0.072 ***	−2,150.1 **
Earnings (LN)	−0.003	−23.3
Student	0.169 ***	3,921.6 ***
Number of siblings	−0.034 ***	−491.1 †
Year (ref. 2007)
2005	0.013	−408.0
2009	−0.069 **	245.3
2011	−0.004	−107.3
Constant	0.386 ***	−14,530.5 *
R^2	0.202	0.133
N	5,537	3,305

Source: 2005-2011 Panel Study of income Dynamics (PSID) and PSID transition to Adulthood Surveys

Note: parental marital status and health, and missing indicators and child race, gender, marital status, and education level included in model. Robust standard errors are clustered at the individual (child) level. Weighted results.

^***p<0.001;

^**p<0.01;

^*p<0.05;

^†p<0.1

Estimated coefficients on parent and child characteristics are also as expected based on prior research. More educated parents were more likely to give money and give significantly larger amounts of money. We find that the likelihood of receiving a transfer was 16.9 percentage points higher for students. In addition, employed children were 7.2 points less likely to receive transfers than their unemployed counterparts. Moreover, unemployed children received $2,150 more than their employed counterparts. Surprisingly, the parameter estimate on child’s earnings is not significant in either model. Lastly, all else equal, the likelihood of a child receiving a transfer in 2009, the height of the Great Recession, was 7 percentage points lower compared to 2007.

Descriptive Analysis of Changes in Parental Resources

As demonstrated in Figure 2, parents on average experienced significant changes to their income and wealth during the Great Recession. Using 2005-2007 as the baseline period, Table 4 highlights the trends (amount and percent change) in individual changes in wealth and income experienced by parents in our sample, as well as the change in wealth (income) conditional on wealth (income) loss between periods t and t-1. We also show the probability that the parents lost income or wealth between two periods. Prior to the Great Recession (2005 to 2007), median wealth increased by $7,187, or 6.2% (Panel a). Of the 23.7% of parents that experienced decreases in their wealth, median wealth loss was $35,233. During the recession years, the likelihood of wealth loss increased by nearly 17 percentage points to 40.3%. At the median, households lost $8,827 (−6.4% of median wealth in 2007) between 2007 to 2009. Among the 40.3% that lost wealth, the median wealth loss was $78,893. Wealth loss was reduced in the following years, but had not yet reached the pre-recession net increases and likelihood of loss remained high by the end of the period. Changes in income (Panel b) follow a similar pattern; however, more dramatic changes occurred in the 2009-2011 period, when the likelihood of income loss increased from 29.5% to 43.5%. During the same period, the median change in income became negative at −$2,036, or −2.6% from median income in 2009.¹⁴

Table 4.

Parents’ Wealth and Income Changes, 2005-2011

Panel a: Parents’ Wealth Loss Likelihood and Change in Wealth between t and t+1, 2005-2011
	Wealth loss likelihood (%)	Median Wealth Change Amount	Median Wealth Change Rel. to Base Year (%)	Median Wealth Loss Amount (Conditional)
Total	31.1	−1,159	N/A	−59,079
2005-2007	23.7	7,187	6.2	−35,233
2007-2009	40.3	−8,827	−6.4	−78,893
2009-2011	40.2	−833	−1.0	−49,609
Panel b: Parents’ Income Loss Likelihood and Change in Income between t and t+1, 2005-20011
	Income loss likelihood (%)	Median Income Change Amount	Median Income Change Rel. to Base Year (%)	Median Income Loss Amount (Conditional)
Total	29.6	306	N/A	−17,257
2005-2007	26.4	2,200	2.7	−16,506
2007-2009	29.5	3,020	4.0	−19,040
2009-2011	43.5	−2,036	−2.6	−17,257

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys

Note: 2011 dollars. Weighted results.

At the same time, children experienced changes in the transfers they received from their parents. Figure 3 illustrates changes in children’s transfer status over the period, holding age constant at 20 to 24 years old. Specifically, the figure shows the increase in the percentage of young adult children who lost transfers during the Great Recession, from 25% in the 2005-2007 baseline to 34.6% in 2007-2009. This is much higher than in the pre-recession period despite the presumably higher likelihood that a young adult’s need for financial support increased during the Great Recession. The portion of children that lost transfers declined after the recession, to 28.5%, but did not yet recover to the pre-recession levels. Table 5 is the sample of young adults who had received a transfer and two-years later did not receive one (from Figure 3) and shows changes in transfer amounts for these young adults who ‘lost’ transfers. Conditional on transfer ‘loss’, the amount that a child ‘lost’ at the median increased during the recession period from −$1,382 in the pre-recession years to about −$2,170 between 2007 and 2009 and declined slightly −$2,096 by 2009 to 2011.

Fig 3.

Percent Receiving No Transfer in Time t+1, Conditional on Transfer Receipt in Time t, ages 20-24

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys

Note: Includes children ages 20-24. Weighted results.

Table 5.

Transfer loss likelihood and amount change conditional on transfer loss (2011 dollars), ages 20-24

	Transfer Loss Likelihood (%)	Median Transfer Change Amount (Conditional)
Total	30.7	−2,062
2005-2007	25.0	−1,382
2007-2009	34.6	−2,170
2009-2011	38.5	−2,096

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys

Note: Only non-zero transfer amounts reported. 2011 dollars. Weighted results.

Multivariate Analysis of Changes in Parental Resources

Table 6 displays the results from the first difference models, estimated using ordinary least squares. We estimate the effect of changes in parents’ income and wealth (income and wealth loss) on changes in transfer amount. The estimates of the change in transfer in all models amount show transfers declined with the loss of parental income. Prior to including any covariates (Model 1), on average, a $10,000 decrease in income led to a modest, but statistically significant, decline in the amount received by the young adult child of $111. Adding covariates to account for parents’ characteristics (Model 2) and changes in the child’s characteristics (Model 3) reduce the estimate slightly. The estimate implies a $10,000 decline in income on average led to $109 reduction in transfers to young adult children. Figure 4 illustrates the relationship at different point in the distribution of income and wealth changes. Figure 4 shows the predicted change in transfer amount at several points along the distribution of parental income loss. A parent with median income loss (about $470) decreased transfers to their child over the period by about $1,000, while a parent at the 75^th percentile (a loss of $16,600) decreased transfers by $1,150. For a parent with the most significant losses, at the 95^th percentile, transfers to their child declined by nearly $1,700.

Table 6.

OLS First Difference Model, Transfer Amount Change

	Model 1	Model 2	Model 3
Wealth loss (in 10,000s)	0.6	0.5	1.6
Income loss (in 10,000s)	−111.7 *	−111.9 *	−109.2 *
Child got married			1,026.8
Child employment change (ref. never working)
Always working			−353.6
Started working			−369.7
Stopped working			1,253.1
Child earnings (LN)
Child earnings loss (in 10,000s)			887.9 ***
Child student status change (ref. never in school)
Always in school			2,167.2 ***
Started school			914.0
Stopped school			−331.6
Parent health changes (ref. "about the same")
Better		450.1	355.4
Worse		−271.7	−265.3
Constant	−962.7 ***	−1,252.3 ***	−932.3
R^2	0.003	0.004	0.023
N	3,211	3,211	3,211

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys

Note: Missing indicators and controls for marital status and education, and child race and gender included. Robust standard errors are reported.

^***p<0.001;

^**p<0.01;

^*p<0.05;

^†p<0.1

Fig. 4.

Predicted Change in Transfers to Young Adult Children by Parental Income Loss at Median, 75^th and 95^th percentile of income loss

We find parental wealth loss had no effect on transfer behavior. It is possible that significant income or wealth loss experienced by parents precluded them from providing material support to their children. These parents may offer to share housing with their children instead, letting them move back into the parent’s home. Our data indicate that roughly 14.5% of young adults ages 20 to 24 that lived outside their parents’ homes in 2007 subsequently moved in with their parents in 2009. This is lower than the 20% who moved in with parents during the baseline period (2005-2007). To further explore this, we estimate a multivariate first difference model (as in Table 7) of the likelihood of living with parents in time t conditional on not living with parents in t-1. Results are provided in Table A2 in Appendix 2. OLS estimation shows no significant relationship between parents’ income and wealth losses and the likelihood their child moved into their home. Thus, we find no evidence that parents are substituting co-residency for non-resident financial support.¹⁵

Other important factors associated with changes in transfers over this period were changes in a child’s earnings and schooling. Both variables measure the child’s demand for transfers and indicate parents respond significantly to their children’s needs. Our first difference model shows average change in transfers received by young adults who entered school was $914 and who stayed in school was $2,167 more than their counterparts who were not in school in either time t or t+1, although only the estimate for the latter is statistically different than zero. A $10,000 decline in the child’s earnings led to an $888 increase in transfers.

Discussion

In this study, we analyze the financial gifts young adult children receive from their parents over the period 2005 to 2011 – years prior, during and immediately after the Great Recession. The transition to young adulthood is a time when individuals experience major changes. A successful transition to young adulthood will form a foundation for the individual over the rest of her life. Financial transfers from parents can be an important part of this transition, helping to finance education or consumption in the early stages of career development. Yet, the Great Recession reduced many parents’ ability to give money to their children while at the same time, young adults’ financial needs may have increased as unemployment rates increased.

We find many young adults received transfers from parents. In 2005, 74% of our sample received transfers. During the Great Recession, the percentage of young adults receiving parental transfers fell to 57%. A larger percentage of parents experienced income and wealth losses during and after the recession than before and mean and median income and wealth declined. A life cycle model of consumption applied to transfers predicts shocks to parental resources that are permanent, and not transitory, will result in lower transfers. We find that income loss has a small effect on the amount of money parents give children on average - a $10,000 parental income loss decreased transfers by $109, while wealth declines did not significantly affect transfers, consistent with the hypothesis that the shocks were perceived to be transitory. Parents may have found other ways to smooth transfers to children during the period, such as reducing consumption in other areas. The significant influence of our measures of the child’s need (schooling and income) on transfers further indicates evidence of the possibility of such accommodations. The finding of larger effects at other points in the income distribution- $1,150 and $1,700 at the 75^th and 95^th percentile of income loss respectively, highlight the heterogeneity in response that may be related to the perceived or actual transitory nature of the loss. The finding that income and not wealth impact transfers is consistent with Zissimopoulos and Smith (2010) who find that income is an important determinant of financial support in the short run, while wealth is more important for giving over multiple years. Moreover, the parents in our sample are on average age 50 and hold relatively less wealth, most of which is illiquid housing wealth. One limitation of first difference models is their tendency to bias estimates toward zero in the presence of measurement error. This may explain in part the relatively small magnitude of our parameter estimates.

Parents may substitute shared housing for material support in response to income loss. Income loss of parents, however, did not lead to more co-residency. Prior research demonstrates an increase in the percentage of young adults living with parents or friends during the period (Kahn, Goldscheider, & Garcia-Manglano, 2013; Weimers, 2014); however, these studies restricted their analyses to children 25 and older (Weimers, 2014) or cross-sectional Census data separated by decades (Kahn et al., 2013). South and Lei (2015) use PSID-TA data to model the determinants of moving out or back into the parental home among young adults. Their multivariate models reveal an increase in moving back in the parental home over the period, 2005 to 2011, with a negative (though insignificant) effect of family income, which is similar to our findings.

Our study has limitations, one being the measurement of transfers. The PSID-TA measures of transfers include parents and other relatives. We only measure changes in parents’ financial circumstances, however the majority of transfers received by young adults are from parents (Brown & Weisbenner, 2004; Gale & Scholz, 1994). Parental income and wealth loss may be correlated and including both in the models may affect the parameter estimates. We test the sensitivity of our results to the exclusion of income and wealth changes, in turn. The coefficients on wealth and income are robust to these specification changes. In the model including only income changes, a 10,000 loss in income resulted in a $108 decline in transfers (versus $109 in the main model). If changes in income and wealth partially reflect unobserved, time-varying characteristics that influence both parental resources and their giving behavior, the coefficient estimates from the primary specification would be biased. For example, wealth declines could reflect savings behavior, while change in income may capture parents entering or leaving the workforce when their child starts or stops college. We control for these changes in children enter/leaving school and the workforce to address the latter; however, these variables may not capture all children’s needs, such as student loan debt.

Our models utilize the variation from all changes in income and wealth over the study period. In sensitivity analyses, we included indicator variables for during and post-recession time-periods and interactions with income and wealth. As expected, transfers were lower in years 2009 and 2011 relative to 2005 and 2007 (−$1,416 and - $1,274 for years 2009 and 2011, respectively). We estimated that a $10,000 loss in income was associated with a $281 decline in transfers, larger than that reported in Table 6, however, the coefficient on the interaction between the post-period and income was not statistically different from zero. We cannot empirically distinguish between unexpected and expected income losses in the pre-recession period, recession period, or post-recession period. We hypothesized that income losses in the recession period would be unexpected and if perceived as permanent, would lead to lower transfers. However, even if the loss was unexpected, if it was perceived as transitory, there would be no effect on transfers. Thus, our estimated results are consistent with parents’ perception of income losses as transitory in the recession period.

We also explored an instrumental variables estimation approach to analyze the robustness of our estimates. We estimated a two-stage least squares model using changes in local unemployment and median house value in the parents’ county of residence as instruments. These estimates show a $10,000 income loss leads to a $1,150 decrease in transfers and a $10,000 wealth loss leads to a $55 decrease in transfers; however, both estimates are imprecisely measured and not statistically different than zero. Furthermore, first-stage estimates show the instruments are weak leading to unknown biases. We also utilized unemployment rates specific to in the parents’ industry of employment as instruments. The estimate on income loss increased to $766, however the estimate was not statistically different than zero and first stage estimates indicated the instrument was weak. Estimates from all models are available upon request.

This study examines short term consequences. There are multiple avenues for future research that would add to our understanding of the long-term impacts of financial transfers. For example, a recent study by Lee et al. (2018) found large positive transfers to adult children was associated with a higher likelihood a home purchase. This relationship was especially pronounced over the last decade, which has been characterized by a tightening of the mortgage market. A recent study by Haider and McGarry (2018) found that although a parents gave differentially to their children for college expenses, it was not compensated with difference in transfers to children later in life, potentially exacerbating within family differences in resources and outcomes. Further study of the relationship parental transfers and economic outcomes of their children over the life cycle will expand our understanding of factors driving intergenerational mobility across and within families.

We estimated a large, positive association between being in school and transfers. These transfers may support college completion and reduce burden of student loans, with long-term implications for earnings and financial security. Indeed, Zissimopoulos and Smith (2010) found annual parental transfers for college-age children in the U.S. was about 50% of average college tuition costs in 2005 and 30% of tuition plus room and board. Recent data from the Federal Reserve Bank of New York Consumer Credit Panel / Equifax on student loans for individuals under age 30 showed that between 2005 and 2009, the number of student loan borrowers increased significantly from 12.1 million to 16.2 million and in 2017, was 16.8 million. In 2005, student loan balances were $162.4 billion, and rose to $275.9 in 2009 and has been rising since. Between 2005 and 2009 defaults nearly doubled. The increase in student loan debt is in part due to a rise in college enrollment over this period. College enrollment increased about 10.6% from 2008 to its peak in 2011 (National Center for Education Statistics, 2017 table 303.70). Overall enrollment declined after 2011 through the end of 2017; however, enrollment in 4-year institutions continued to increase, albeit at a slower rate (2.4% between 2011 and 2017 compared to 12.7% from 2008 to 2011). In addition, recent evidence links increasing student loan debt to declining homeownership rates among young adults between 2005 and 2014 (Mezza, Ringo & Sommer, 2019). Changes in transfers from parents would have potentially affected young adults’ decision about college enrollment and levels of student debt. At the same time, the recession would have reduced employment opportunities. The impact of student loan debt and declines in homeownership may have long-lasting effects on financial well-being. There remain gaps in our understanding of the importance of transfers in the schooling decision, schooling completion and accrual of student loan debt, and long-term consequences. Further, there may be differences in these relationships across families. These are important avenues for future study.

There is little evidence that the gap in family support was filled with increases in public support for this age group. Poor employment prospects during the recession led to a rise college enrollment, while shrinking state budgets increased tuition rates (Barr & Turner, 2013; Long, 2014). These factors along with reductions in parental transfers to young adults likely impacted their ability to pay for college. Among 20 to 24 year olds that received a transfer in our sample, the median transfer amount declined by over $2,900 between 2007 and 2011 to $3,600. This amount is equivalent to 40% of the average in-state tuition at a public university in the 2010/2011 school year, about $7,600. The funding shortfall suggests a larger role for public assistance to pay for college during the period. Indeed, Congress passed legislation increasing aid to college students in response to the Great Recession; however, many states subsequently reduced their spending (Bettinger & Williams, 2014).

Conclusion

This study revealed a shrinking of private, familial safety nets for young adults over the Great Recession, in part due the decline in income of parents. The reduction in private transfers to young adult children may have a long-term impact on these young adults as they enter the workforce, gain financial independence and form their own households and this remains a key issue for study. Whether a rebound of parental income in the recovery period is accompanied by increased transfers that compensate for the loss, and the long term implications of the change in amount and timing of transfers, are future avenues for study.

Biography

Julie M. Zissimopoulos is Associate Professor, Vice Dean of Academic Affairs, Price School of Public Policy, and Director Aging and Cognition, Director Research Training, Schaeffer Center for Health Policy and Economics, University of Southern California. She is an expert in the economics of aging focusing on risk and costs of dementia, savings, labor supply, family caregiving and financial support. Her recently published research appeared in the Journal of the American Medical Association Neurology, Journal of Gerontology Social Science, Journal of the American Academy of Arts and Sciences, Journal of Health Economics. Dr. Zissimopoulos received her Ph.D. in economics from the University of California, Los Angeles.

Johanna Thunell is a Postdoctoral Scholar at the University of Southern California Schaeffer Center for Health Policy and Economics. Her main interests include health and labor policy, labor supply and caregiving, and aging. She is current involved in research on caregivers of people with Alzheimer’s disease. Dr. Thunell received her PhD in Public Policy and Management from the University of Southern California where she received the Henry Reining Dissertation award for best dissertation.

Stipica Mudrazija is a Senior Research Associate in the Income and Benefits Policy Center at the Urban Institute. He is also an adjunct professor at Georgetown University and an elected member of the National Academy of Social Insurance. He studies issues related to population aging, retirement, intergenerational transfers, health and caregiving. Before joining the Urban Institute, he was a postdoctoral scholar at the Edward R. Roybal Institute on Aging at the University of Southern California. He holds a doctorate in public policy from The University of Texas at Austin.

Appendix 1: PSID-TA Survey Questions

Questions about financial transfers in the PSID Transition to Adulthood Surveys, F36.

Whether the respondent received a transfer for rent, home purchase, tuition, expenses, bills or a loan:

The next questions are about financial help that you might have received during [previous year]. This could be in the form of money given to you or money paid on your behalf for goods or schooling. Did your parents or other relatives… [purchase a house or condominium for you, pay rent or mortgage on your behalf give you a personal vehicle, pay for tuition, cover expenses or bills, give you a personal loan]?

For positive responses, the amount of the transfers:

What was the value of [house, rent, vehicle, tuition, expenses, personal loan]?

Any other transfers not in the categories listed above:

Other than the amounts we just talked about, during the last two years, have you received any large gifts of money or property or inheritances of money or property

For positive responses, the amount of other transfers:

How much was it worth all together, at that time?

Appendix 2: Co-residency Results

Table A1.

Percentage of Young Adults Co-residing with the Parents by Child Age, 2005-2011

	17 to 19	20 to 21	22 to 23	24 to 25	26 to 27
All years	68.0	42.4	35.9	24.9	21.4
2005	71.3	35.2
2007	66.6	38.6	35.5
2009	65.6	43.3	32.6	23.8
2011	68.7	51.5	39.9	25.8	21.4

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys.

Note: Weighted results.

Table A2.

OLS First Difference Models, Probability of living with parents in Time t conditional on not living with parents in time t-1.

	Moved In
Change Variables
Wealth loss (IHS)	−0.001
Income loss (LN)	0.004
Child got married	−0.114 ***
Child employment change (ref. never working)
Always working	−0.043
Started working	0.040
Stopped working	0.009
Child earnings loss (LN)	−0.003
Child student status change (ref. never in school)
Always in school	−0.045
Started school	−0.056
Stopped school	−0.020
Parent health changes (ref. "about the same")
Better	−0.015
Worse	0.045
Constant	0.189 ***
R^2	0.037
N	1,542

Source: 2005-2011 Panel Study of Income Dynamics (PSID) and PSID Transition to Adulthood Surveys; 2005 American Communities Survey; 2007-2011 ACS 1-year estimates; Bureau of Labor Statistics Local Area Unemployment Statistics (LAUS).

Note: controls for parent age, marital status and education, and child race and gender included

^***p<0.001;

^**p<0.01;

^*p<0.05;

^†p<0.1