Time, money, and entry into parenthood: The role of (grand)parental support.

Objective: The study's objective is to understand how parental propensities to provide support, as predicted by parental characteristics, shape adult daughters' and sons' entry into parenthood in the United States. Background: Much research explores the influence of parental support on adult children's fertility, but the evidence is mixed and primarily focuses on European contexts. Theoretical approaches suggest that to best understand how parental support shapes adult children's outcomes, it is important to account for different forms of parental support, that is, time and money, and variation in parental characteristics. Method: This study combined different data from the Panel Study of Income Dynamics: the 2013 Roster and Family Transfers module, main interview data file, and the Childbirth and Adoption History File. We implemented a two-step analysis strategy. In the first, we built two different measures of propensities to receive parental support (PPS) in the form of time and money. In the second, we used discrete-time logistic regression models to analyze the effects of these propensities to receive parental support on adult daughters' and sons' fertility.
Results: We find a positive and consistent effect of all types of PPS measures on adult daughters', but not adult sons', likelihood of entry into parenthood. The fertility decisions of adult daughters are highly responsive to the prospect of receiving parental support in the form of time or money. Conclusions: Our results reflect the importance of informal support for women's entry into parenthood and highlight gender differences in the perceived and actual costs of becoming parents.

INTRODUCTION

Since the 1960s, the amount of time and money US parents invest in their children has increased dramatically (Nomaguchi & Milkie, 2020). However, rising job insecurity together with the absence of paid parental leave and supportive work‐family policies have made it more difficult for would‐be parents to meet the increasing demands of parenthood, contributing to delays and declines in childbearing in the United States (Collins, 2019; Guzzo & Hayford, 2020). Parents, and particularly mothers, experience increasing levels of work‐family conflict, and are overburdened with the competing demands of their job and household responsibilities (Collins, 2019). At the same time, improvements in mortality and health have lengthened the healthy years (grand)parents share with their adult children and grandchildren (Margolis & Wright, 2017), and older adults today have greater financial security than previous generations (Keister & Deeb‐Sossa, 2001). Thus, parents can play a crucial role in their adult children's likelihood of entry into parenthood by supporting them with practical and financial resources.

There is ample evidence that parents provide resources to their adult children when the latter have children. Yet, it is unclear whether the decision to have a first child is shaped by the likelihood that their parents will provide them with time or financial support. Levels and types of parental support vary across children's lives, and are responsive to major life course events and transitions such as the entry into parenthood (Fingerman et al., 2020a). Parents devote significant practical and financial resources to both their adult children and their grandchildren (Albertini et al., 2007; Dukhovnov & Zagheni, 2015). Although adult children who are not yet parents cannot perfectly predict how their own parents would support them as grandparents, they may anticipate this support by observing their own parents' characteristics and behaviors that signal an (in)ability to provide different types of support, such as geographic proximity or financial resources (Rutigliano, 2020).

Little research considers how parental propensities to provide support, as proxied by their parents' characteristics and behaviors, shape adult children's decision to become parents in the context of the United States (see Lehrer & Kawasaki, 1985 for an exception). Instead, most research is conducted in the European context and uses unidimensional proxies of future parental support. In the European context, mixed results have been obtained on the association between parental support and adult children's fertility (Sear, 2017), as some studies find a positive association (Aassve et al., 2012; Pink, 2018; Rutigliano, 2020; Schaffnit & Sear, 2017b), others reported no association (Kertzer et al., 2009), and a few find a negative association (Schaffnit & Sear, 2017b). Most of these studies use unidimensional proxies of future grandparental childcare support (see Rutigliano, 2020 for an exception), such as whether they provided childcare to their existing grandchildren (Thomese & Liefbroer, 2013), the geographical distance between parents and adult children (Pink, 2018), or whether the parents are alive (Schaffnit & Sear, 2014). However, focusing on a single parental characteristic might be a poor predictor of future parental support. Furthermore, most studies exclusively focus on grandparent childcare as a source of parental support, to the exclusion of financial support (see Schaffnit & Sear, 2017a; Waynforth, 2012 for exceptions). In short, research is needed that examines how multidimensional predictors of future financial and practical parental support shape adult children's entry into parenthood in the US context.

This study combines the 2013 Roster and Family Transfers module as well as the main interview data of the Panel Study of Income Dynamics (PSID) with consolidated fertility histories from the Childbirth and Adoption History File to address three research questions: (1) How does parents' propensity to provide support shape their adult children's entry into parenthood in the United States? (2) Does parental support affect adult daughters' entry into parenthood differently than adult sons'? (3) Does parents' propensity to provide financial support influence adult children's fertility differently than their propensity to provide time support? Building on existing literature, we formulate theoretical expectations about the relationship between parental propensity to provide support and adult children's entry into parenthood in the United States. Following Rutigliano's study (2020), we apply a two‐step methodological approach to create indicators of parental propensities to provide time and financial transfers, test how these indicators influence the entry into parenthood of their adult children, and test whether these patterns differ by gender and by transfer type.

This study makes several contributions to the literature. First, instead of focusing on a single proxy of parental support, we account for the multiple and interrelated parental characteristics that influence their capacity to provide support to their adult children. Second, we add to a growing literature that accounts for multiple forms of parent‐to‐offspring support by studying time and money transfers from parents to their adult children, and how these transfers influence the adult children's fertility decisions. Third, we highlight gender differences in adult children's responses to parental support. Finally, we focus on how parental support influences their adult children's entry into parenthood, that is, the decision to have a first child, which has received far less attention than higher‐order birth transitions.

BACKGROUND

Intergenerational transfers: Parents and adult children

As a product of demographic changes, parent–adult child ties have strengthened over the past 40 years (Seltzer & Bianchi, 2013). Illustrating the salience of these ties, most intergenerational transfers occur between parents and children in the United States (Fingerman et al., 2013; Swartz, 2009). These transfers take multiple forms, including caregiving, housework, housing, financial support through loans and monetary gifts, and emotional support (Fingerman et al., 2020a; Gans & Silverstein, 2006).

While intergenerational transfers have a strong reciprocal element, over most of the life course intergenerational support primarily flows from parents to their children (Albertini et al., 2007; Attias‐Donfut et al., 2005; Kalmijn, 2019). Intergenerational transfers of time follow a pattern of “delayed flow reversal,” in which parents are net givers until they reach older ages, at which point they become net receivers (Kalmijn, 2019; Silverstein et al., 2002). Several studies estimate this crossover point to occur when parents reach their mid‐70s (Albertini et al., 2007; Dukhovnov & Zagheni, 2015; Kalmijn, 2019). Regarding intergenerational financial transfers, in Western countries it is still under debate whether this same pattern of flow reversal holds, as several studies find that parents tend to remain net givers of financial support even at older ages (Albertini et al., 2007; Seltzer et al., 2012). Regardless, because most potential grandparents are younger than the average flow reversal threshold, we restrict our focus to the association of downward transfers (parents‐to‐children) with adult children's entry into parenthood.

How do (grand)parental transfers reduce the costs of parenthood for adult children?

Grandparents today have substantially more healthy years of overlap with their grandchildren and financial resources than in the past, and therefore potentially contribute more support to their adult children during their entry into parenthood (Keister & Deeb‐Sossa, 2001; Margolis & Wright, 2017). US grandparents provide support to their adult children and grandchildren in multiple ways (Meyer & Kandic, 2017). Starting with time, childcare is one of the most important time transfers parents provide to their adult children in the United States (Dukhovnov & Zagheni, 2015; Luo et al., 2012). A recent US child time diary study showed that in a typical week, 50% of young grandchildren and 20% of teenage grandchildren spent time with their grandparents (Dunifon et al., 2018). While time spent with grandchildren is not a perfect indicator of childcare, these estimates closely matched survey measures of grandparental childcare that show that about 60% of grandparents had ever provided childcare to their grandchildren over a 10‐year period (Luo et al., 2012). In addition to providing childcare, parents contribute to their adult children's parenting needs by helping with errands, housework, and home repairs (Dukhovnov & Zagheni, 2015; Dunifon et al., 2018; Meyer & Kandic, 2017). The distribution of time provision among grandparents is not equal; several studies find (grand)mothers to be more involved in intergenerational ties than grandfathers, especially concerning grandparental childcare provisions (Fingerman et al., 2020b; Hank & Buber, 2009). These findings suggest that, in the context of the United States, parents are a potentially significant source of time support to their adult children in raising the next generation.

Turning to financial support, the existing literature on how intergenerational support affects adult children's entry into parenthood has primarily focused on large, one‐time parental wealth transfers in anticipation of parenthood, such as buying a new house or passing on real estate (Cox & Stark, 2005; Leopold & Schneider, 2011). However, we also know from the broader literature on parental financial transfers that parents often provide smaller, regular cash transfers in response to their adult children's economic needs or major life course transitions (Attias‐Donfut et al., 2005; Fingerman et al., 2015). For example, McGarry (2016) found that in the United States, 14.5% of adult children with a residential child received parental financial transfers, and the likelihood of receiving financial support increased with the number of children in the adult child's household. In this way, parents provide financial support to their adult children that can subsidize the costs of parenthood, such as buying clothes or toys or contributing to daycare costs. These studies show how adult children's entry into parenthood can elicit an increase in financial transfers from parents to their adult offspring.

Anticipating parental support and adult children's entry into parenthood

How do adult children who have yet to become parents anticipate future levels of parental support? While adult children who already have children can anticipate how their parents will help them from observing their current levels of support, childless adult children cannot. Instead, they may make inferences about the likelihood of future parental support from their parents' observable characteristics, such as their physical health, current income levels, or spatial proximity. Below, we describe how different parental characteristics and behaviors may shape adult children's entry into parenthood, how the gender of the adult child is likely to moderate that relationship, and how adult children's characteristics might predict both their entry into parenthood and their parental support.

Parental characteristics and behaviors: An observable signal for future (grand)parental support?

The level and type of support parents can provide to their adult children as grandparents is a function of multiple parental characteristics and behaviors, including their geographic proximity, vital status, current transfer levels, gender, age, health status, employment status, family size, and financial situation (Fingerman et al., 2020a; Meyer & Kandic, 2017; Seltzer & Bianchi, 2013). The key mechanism is that each of these factors might be interpreted by the adult child as a signal of their parents' availability and willingness to help as prospective grandparents. For instance, parents and children who live closer to one another have higher levels of contact, which may lead the adult child to expect that their parents would be available and willing to provide support. The same can be argued for employment status, as a parent who is still working may signal that they are unavailable to provide practical support, yet they may have more resources to provide financial support if needed.

Building on the idea that parental characteristics and behaviors can signal future (grand)parental support, a growing literature has explored how the likelihood of parental support influences adult children's entry into parenthood. For instance, several studies find that parental geographic proximity, and thus potential for childcare, has a mixed association with adult children's entry into parenthood (García‐Morán & Kuehn, 2017; Hank & Kreyenfeld, 2003; Mathews & Sear, 2013; Pink, 2018; Rindfuss et al., 2007; Schaffnit & Sear, 2014). Another indicator of future parental time support is parents' current level of grandparental childcare provided to their other adult children who already have children. However, only one study finds a positive association and only when the existing grandchild is Age 3 or older (Aassve et al., 2012), while others find no statistically significant association (Kaptijn et al., 2010). Turning to financial support, only one study has used current levels of financial transfers from parents to children to predict adult children's entry into parenthood, and found a negative relationship (Schaffnit & Sear, 2017b).

These inconsistent findings suggest that focusing on a single parental characteristic or behavior might provide a mixed signal for future parental support. For example, having parents in close geographic proximity might negatively predict parental support if one parent needs medical help, or it might positively predict parental support if both parents are healthy. Similarly, if parents are already providing childcare to the adult child's sibling, it may signal that they are capable and willing to provide support, but they simply may not be available to provide further care. Also, adult children's receipt of financial transfers may indicate their precarious economic situations and may not be a clear indicator of future grandparental support. Highlighting the importance of the interplay between parental characteristics, Rutigliano (2020) takes a novel approach to the study of parental support and adult children's entry into parenthood. In her study, Rutigliano constructs a measure of anticipated parental support that jointly accounts for multiple parental characteristics to investigate the role of grandparental childcare on entry into parenthood across different European welfare regimes. The findings suggest that grandparental childcare propensity is positively associated with having a first child, and the relationship is stronger in countries where work‐family public policies are weaker.

Building on these findings, we argue that in the US context, it is important to go beyond using single measures of parental characteristics and behaviors to proxy parental support due to the considerable variability in older parents' resources and constraints (Fingerman et al., 2015; Margolis & Wright, 2017). Furthermore, the positive effects of parental support on adult children's fertility is stronger in countries where public childcare is less prevalent (Aassve et al., 2012; Rutigliano, 2020). Therefore, we adopt Rutigliano's approach (2020) to construct measures of parental support propensity that account jointly for multiple parental characteristics. We also extend this approach to not only consider parental time support, but also financial support, which has received less attention within this literature. We hypothesize that in the US context, adult children with parents who can provide financial and time support are more likely to become parents themselves.

Gender differences in adult children's perceived and actual costs of parenthood

To date, research on parental support and first‐birth transitions in high‐income settings has primarily focused on adult daughters (see Schaffnit & Sear, 2017a for an exception). Yet, parental support is likely to shape adult daughters' and sons' entry into parenthood differently because the costs of parenthood are unequally distributed between men and women. Despite recent trends toward greater gender equity, women still perform the majority of childcare and housework (Nomaguchi & Milkie, 2020), and women prospectively anticipate these cost differences (Liefbroer, 2005). Furthermore, mothers and adult daughters are more involved in intergenerational support exchanges, especially when focusing on stereotypically feminine forms of support (Fingerman et al., 2020b). While it is beyond the scope and feasibility of this study to distinguish between paternal and maternal support, these gendered differences in intergenerational ties could suggest that adult daughters might be more responsive to any parental support. Taken together, we expect that adult daughters' entry into parenthood will be more strongly associated with anticipated parental support than adult sons' entry into parenthood.

We also anticipate that adult daughters and sons will respond differently to different types of parental support—that is, time versus money. Specifically, while parental financial support is beneficial for the adult child's entire household, parental time support is particularly helpful for an adult daughter, who is likely to experience substantial time pressure as she takes on a larger share of parenting responsibilities. Thus, in this scenario, parents may alleviate their adult daughter's time constraints by helping her with housework, providing her with free and reliable childcare, and running errands. Thus, we expect gender differences in the association between parental support and entry into parenthood to be greater for time than for financial support.

Other adult children's characteristics associated with both parental support and entry into parenthood

The focus of this study is on how anticipated parental support may shape entry into parenthood among adult daughters and sons in the context of the United States. We emphasize the moderating effect of the adult child's gender because we believe it to be understudied and possibly important given the well‐known gender differences in the costs of parenthood and intergenerational ties. Yet, it is important to note that there are other adult children's characteristics that may be associated with both differences in parental support and entry into parenthood. As such, we control for adult children's socioeconomic status (as measured by educational attainment), race/ethnicity, partnership status, birth cohort, and age to capture potentially confounding differences in fertility and parental support (Fingerman et al., 2015; Guzzo & Hayford, 2020; Sarkisian & Gerstel, 2004). Nevertheless, because of sample size limitations, it is beyond the scope of this study to explore the moderating effects of these variables in a gender‐stratified analysis.

DATA AND METHOD

Data: Panel study of income dynamics

This study uses the 2013 Roster and Family Transfers module (R&T), as well as the main interview data of the PSID with consolidated fertility histories from the Childbirth and Adoption History File (CAH) (https://psidonline.isr.umich.edu/). The R&T module was collected as a supplement to the 2013 PSID main interview (Schoeni et al., 2015). The R&T data identify parents, parents‐in‐law, children, stepparents, and stepchildren of PSID heads and wives. In the PSID, respondents are referred to as heads and wives, and the terminology of heads and wives is also applied to cohabiters in partnered households (this terminology became gender‐neutral starting in the 2017 PSID wave). The R&T module collects data on recent transfers of time and money (over $100) between the households of parents and their adult children. We focus on downward transfers from parents to their adult children. It is important to stress that the measures of time and financial transfers are collected at the inter‐household and not at the inter‐individual level (see Figure B1 in the Supporting Information for an illustration). For example, for families with parents in an intact partnership, the yearly amount transferred is the total of the transfers provided by the mother and the father to the adult child's household. For families with parents living in different households, data on transfers are collected for the household of each parent separately.

A two‐step approach

We face two main challenges when studying how parental support influences adult children's entry into parenthood. First, we cannot observe parental support before the birth of the grandchild. Second, we cannot use past transfers as predictors for future transfers because parental support changes in response to the major life course transitions and needs of the adult child. To overcome these issues, we implement a two‐step regression approach (see Rutigliano, 2020). In the first step, we analyze how parental characteristics predict downward time and money transfers in a sample of parents and adult children who are already parents. That is, the first step answers the question of how parental characteristics are associated with downward financial and time transfers among the subsample of adult children who are already parents. In the second step, we use the predictions from the first‐step models to generate a measure of the predicted propensity for parental support in a sample of parents and childless adult children at the time of survey (Step 2a) and assess whether this propensity predicts the subsequent fertility of the adult children (Step 2b). We run models separately for adult sons and daughters and by types of transfers, and only consider the transfers from the household of the adult children's own parents. In both the first and second step, we impute all missing values using chained equations (using the ‐mi impute chained‐ command and 50 imputations in Stata 16). In the following sections, we explain in more detail the purpose of each step and how we implement them.

First‐step specification: Model and variables

The main goal of the first step is to estimate how different parental characteristics predict downward time and money transfers among a sample of parents and adult children who already have children. We select a sample of adult children in the R&T module who are between ages 18 and 55 at the time of the R&T interview and who have at least one child under the age of 18. The final sample of the first step consists of 3040 adult daughters and 2219 adult sons (2738 mother–adult daughter pairs, 2319 father–adult daughter pairs, 1922 mother–adult son pairs, 1645 father–adult son pairs). The age selection for the first step maximizes the likelihood of adult children still coresiding with their own children (Daw et al., 2019). The key idea behind this step is to identify a set of observable parental characteristics that adult children may use to assess how supportive their parents will be during their own entry into parenthood.

Predictor variables

Building on the existing literature on predictors of intergenerational transfers in the presence of grandchildren we discussed above (see Fingerman et al., 2020a for a review), we include the following variables to predict parental propensity to provide support: parents' age (linear and quadratic terms), self‐reported health (poor/fair, good, very good/excellent), partnership status (intact, re‐partnered, single), annual household income (less than 25k, 25–50k, 50–75k, 75k+), homeownership status (yes or no), employment status (in or out of the labor force), the adult child's number of siblings, and geographical distance from the adult child's household (coresident, within 30 miles, within 200 miles, greater than 200 miles). Finally, we include the adult child's age to control for how the age difference between the parent and adult child shapes transfers (Fingerman et al., 2020). Table C1 in the Supporting Information summarizes the means, proportions, and minimum and maximum values of all the variables included in the first‐step models.

Downward transfer measures

We create time and money transfer variables using measures of interhousehold assistance from the R&T module. Among other transfer questions, PSID households' heads and wives are asked about short‐term transfers from their parents. For downward time transfers, respondents are asked: “Families sometimes help each other with activities such as errands, rides, chores, babysitting, or hands‐on care. In 2012, did your parent(s) spend time helping you? About how many hours in 2012 did your parent(s) spend helping you?” The total reported hours ranged from 1 to 8760. Similarly, for downward money transfers, respondents are asked: “In 2012, did your parent(s) give you any money, loans or gifts of $100 or more? About how much did your parent(s) give?” Total reported financial transfers received in 2012 ranged from $100 to $999,999,997. Appendix B in the Supporting Information provides detailed explanations on how we construct the transfer variables.

In our models, we dichotomized the transfer amounts because both time and money transfer variables were right‐skewed and zero‐inflated. Time transfers therefore take a value of 1 if either parent spent any time helping the adult child's household during 2012, and of 0 otherwise. Likewise, money transfers are equal to 1 if either parent gave any money, loans, or gifts worth at least $100 to the adult child's household during 2012, and are 0 otherwise. We separate these dichotomized measures by the household from which they were given—adult daughter's mother, adult daughter's father, adult son's mother, and adult son's father (see Appendix B in the Supporting Information). Whenever an adult child's mother and father live together, the household transfer amount is the same because the transfer variables capture interhousehold exchanges. As a robustness check, we replicate the analysis using the log‐transformation of the transfer amounts. The results show very similar findings both in terms of sign and statistical significance (see Table E1 in the Supporting Information).

First‐step models

We proceed with the first‐step analysis by estimating a series of models predicting any downward transfer for each parent‐adult child pair (i.e., mother–adult daughter, father–adult daughter, mother–adult son, and father–adult son) and type of transfer (time and money). We employ logistic regression models (using the ‐logit‐ command in Stata 16). Specifically, we fit the following models separately for adult daughters and sons: where is the probability that the transfer is 1, is equal to for time transfers and for money transfers, is the parental household‐adult child household, and is a vector of explanatory variables that capture characteristics and is the corresponding vector of coefficients.

Second‐step specification: Model and variables

The second step explores the relationship between parental support propensities in terms of time and money and adult children's entry into parenthood. The goal of this step is twofold: first, to build synthetic propensity for parental support (PPS) measures that assign to the adult children's household the likelihood that they will receive help from their own parents (Step 2a); and second, to identify how these measures influence adult children's entry into parenthood (Step 2b). We select adult children between ages 18 and 45 who had not had a child by the end of 2012, regardless of their partnership status. The parenthood status of the adult children as well as their fertility events is determined using the consolidated fertility histories from the CAH file. We consider any first births that occurred between 2013 and the last available interview in 2017. We first describe in detail how the PPS measures are constructed, and then discuss the fertility model specification.

Step 2a: PPS measures

The PPS measures are constructed by applying the first step's vectors of coefficients to the second step's sample, using the same set of explanatory variables . In other words, we use the estimates produced in the first step to compute out‐of‐sample predictions based on the second‐step sample of childless adult children in the 2013 wave. The PPS measures are calculated using variables collected in the R&T module, which ensures that any time‐varying variables are measured prior to the adult child's entry into parenthood. We build a PPS measure for each parent–adult child pair—that is, mother–adult child, father–adult child—by type of transfer, and separately for adult daughters and sons. Each of these indices includes household‐level (e.g., income) and individual‐level characteristics (e.g., health status). While each PPS measure is specific to the parent–adult child pair, it is important to stress that coresident parents have the same household variables. We obtain up to four PPS indices for each adult child: maternal and paternal PPS‐T (time); maternal and paternal PPS‐M (money). We do not calculate PPS indices if the parent is either missing or absent. In Appendix D (Supporting Information), we work out a simplified example to illustrate how PPS measures are calculated.

Our goal is to capture adult children's overall likelihood to receive support from their own parents. Therefore, to simplify the analysis, we take the maximum between the maternal and paternal indices for each type of transfer. The other advantage of taking the maximum is that it makes the PPS measures comparable across different parental household structures, that is, coresident parents, divorced parents, only one living/available parent. Thus, each adult child has only two indices: PPS‐M, which captures the adult child's household propensity to receive financial transfers from their own parents; and PPS‐T, which captures the adult child's household propensity to receive time transfers from their own parents. In supplementary analysis, we replicated the analysis using the paternal and maternal indices instead of the maximum (see Table E1 in the Supporting Information). The direction of the results using the paternal and maternal indices is very similar and not statistically different from our main results.

In the second step, we have a special case of missing data for our PPS indices, which occurs when the adult child does not have a parent because the parent is either dead or absent. This type of missing data can be defined as appropriately missing data for cases in which the missing value simply does not exist. For these cases, we follow Allison's recommendation (2002, footnote 4) to employ the dummy variable adjustment method. Thus, we first create an indicator variable that equals 1 if the data are appropriately missing, and that equals 0 otherwise. Second, we assign observations with appropriately missing data for the variable to a constant . We choose the mean value of each of the PPS indices as a value for . While such a method is inappropriate for survey nonresponse data, it produces unbiased estimates for appropriately missing data (see Daw et al., 2020 for a similar application to multigenerational PSID data).

Step 2b: Entry into parenthood models

We employ discrete‐time event‐history models using logistic regression. The dependent variable is defined as a binary variable that equals 1 in the year in which the adult child reports a first birth, and 0 otherwise. Observations are right‐censored at the earliest of the following events: the death of the respondent, Age 45, or the last recorded interview. Formally, the hazard of having a first child can be defined in logit form as: where is the probability that a transition occurs in year for the adult child . where is a set of indicator variables for each year to capture the time in years that has elapsed since the adult child entered the observation period, is a vector of time‐varying and time‐constant covariates for the adult child , is the propensity to receive parental support measured at Wave 2013, and is equal to for time transfers and for money transfers. To test for gender differences, we estimate the second‐step models jointly for adult daughters and sons, including an interaction term between and the adult child's gender. When testing for gender differences, we cluster the SEs of all models by 2013 household identifiers to account for statistically interdependent observations within couples. We do not report those results but indicate whenever they are significant throughout the results section.

The final sample is composed of 1013 men (264 births, 26%) and 1045 women (311 births, 30%). The key explanatory variable is the PPS measure previously described. We run separate models for adult daughters and sons. In the models, we include several control variables. First, we include a categorical variable for the time that has elapsed since the adult child entered the observation period in years (1, …, 5) and the adult child's birth cohort (1963–1974, 1975–1979, 1980–1984, 1985–1989, 1990–1995). We control for the adult child's race/ethnicity (non‐Hispanic white, non‐Hispanic Black, Hispanic, other race/ethnicity) and educational attainment at Wave 2013 (less than high school, high school diploma, some college, 4‐year college degree). We include several time‐varying variables: partnership status (single, cohabiting, married, first‐year cohabiting) and the region of residence (Northeast, North Central, South, West, outside of US mainland). We include a set of variables specific to the parent from whom the PPS measure is derived: parent's age and parent's gender. We include a parsimonious set of control variables for two reasons. First, our main explanatory variable is constructed by accounting for several family structure and parental characteristics. Second, we selected control variables to avoid conditioning on intermediate variables. Table 1 summarizes the means, SDs, and proportions of all variables included in the second‐step models.

TABLE 1

Descriptive statistics for adult daughters and sons in the second‐step sample

	Adult daughters			Adult sons
	Mean/Prop	Min.	Max.	Mean/Prop	Min.	Max.
DV: First‐birth event	0.07			0.05
PPS‐T	0.51	0.05	0.94	0.45	0.02	0.94
PPS‐M	0.26	0.03	0.58	0.23	0.01	0.73
Duration
Year 1	0.25			0.25
Year 2	0.22			0.22
Year 3	0.20			0.20
Year 4	0.17			0.17
Year 5	0.16			0.16
Birth cohort
63–74	0.09			0.11
75–79	0.15			0.17
80–84	0.29			0.29
85–89	0.33			0.34
90–95	0.13			0.09
Race
Non‐Hispanic white	0.73			0.77
Non‐Hispanic black	0.12			0.12
Hispanic	0.11			0.08
Other race/ethnicity	0.03			0.03
Education
<HS	0.03			0.04
HS	0.15			0.23
SC	0.25			0.25
BA+	0.56			0.48
Region
Northeast	0.18			0.17
North Central	0.26			0.25
South	0.32			0.32
West	0.23			0.24
Outside of US mainland	0.01			0.01
Partnership status
Single	0.45			0.53
Cohabiting	0.08			0.09
Married	0.41			0.35
First‐year cohabiting	0.05			0.03
No alive/present parent	.01			0.01
Parent's age (PPS‐T)	57.05	33.00	88.00	57.93	34.00	84.00
Parent's age (PPS‐M)	57.39	33.00	95.00	58.17	29.00	88.00
PPS‐T parent = mother	0.79			0.77
PPS‐M parent = mother	0.70			0.65
N	1045			1013
N person‐years	4028			3917

Notes: Authors' estimates based on data from the Panel Study of Income Dynamics. Family sample weights are applied. Percentages may not total 100 due to rounding.

Abbreviations: BA, Bachelor's degree; Duration, time that has elapsed since the adult child entered the observation period; DV, dependent variable; HS, high school; Max., maximum; Min., minimum; PPS, propensity of parental support; Prop, proportion; SC, some college/2‐year college degree.

RESULTS

The results section is organized as follows. We begin by presenting descriptive evidence for the associations between PPS measures and entry into parenthood. We then discuss the regression model results for the associations between PPS‐T and PPS‐M with entry into parenthood for adult sons and daughters. For the sake of brevity, we do not discuss the regression results of the first‐step models, because the goal of the first step is not to test associations between parental characteristics and downward transfers, but to properly build our PPS measures. We include the full regression results model in Table C2 (PPS‐T) and Table C3 (PPS‐M) in the Supporting Information to show the findings are aligned with the existing literature on predictors of parental transfers.

Descriptive results: Second‐step models

We first descriptively assess the relationship between PPS measures and the likelihood of having a first child. We expect adult children who have a higher propensity to receive parental support will be more likely to have transitioned to parenthood during the observation period. To describe these associations, we compare differences in the distribution of PPS measures between adult daughters and sons who have and have not become parents during the observation period. Figure 1 displays the density plots for the two different PPS measures, PPS‐M and PPS‐T, for adult daughters and sons. We provide simple, two‐tailed t tests to indicate whether the densities are statistically different between adult children who stay childless and adult children who become parents by Wave 2017.

FIGURE 1

PPS measures for adult daughters and sons. Authors' estimates based on data from the Panel Study of Income Dynamics. PPS‐M, propensity for parental support in money; PPS‐T, propensity for parental support in time

Starting with PPS‐T (Figure 1(A)), adult daughters who become mothers display higher PPS‐T values than adult daughters who remain childless. These differences are statistically significant (t test, p value = .00). In contrast, adult sons who become fathers display PPS‐T values similar to those of adult sons who remain childless, and these differences are not statistically significant (t test, p value = .17). Turning to PPS‐M (Figure 1(B)), we find statistically significant differences in the distributions of our PPS‐M measure between adult children who do and do not become parents. These differences hold for both adult sons and adult daughters and suggest that adult children who become parents have higher PPS‐M values than their counterparts who remain childless.

These descriptive findings suggest that adult daughters who have parents with the resources and characteristics to provide both time and financial support are more likely to transition to parenthood. While this pattern also holds for adult sons, the association is statistically significant for financial support only, and does not hold for time. We now turn to the regression results to assess whether these descriptive findings hold after accounting for the timing of fertility and adult children's characteristics.

Regression results: Second‐step models

In this section, we discuss results from the discrete‐time, logistic event‐history models to investigate the associations between adult children's propensities for receiving parental support and their likelihood of entry into parenthood. Results are presented as average yearly marginal effects of PPS‐T and PPS‐M measures on first‐birth transitions for adult sons and daughters, separately. Furthermore, we show the average yearly predicted probabilities of having a first child for both PPS‐T and PPS‐M measures for adult daughters only. We do not show the average predicted probabilities of PPS‐T and PPS‐M for adult sons, as they are not statistically significant. Both the average marginal effects and predicted probabilities are calculated holding covariates at observed values using results from discrete‐time logistic event‐history models (see Tables C4 and C5 in the Supporting Information). To compare predictions and marginal effects across logistic regression models, we implement Mize et al.'s general framework (2019), and use seemingly unrelated regression estimates to combine estimates from multiple models.

Figure 2 shows the average marginal effects of PPS‐T (Panel A) and PPS‐M (Panel B) on having a first birth for adult daughters and sons. Starting with time support, the average marginal effect of PPS‐T for adult daughters is positive and significant (AME = 0.066, SE = 0.026), whereas the average marginal effect of PPS‐T for adult sons is almost null and is not statistically significant (AME = −0.002, SE = 0.024). Turning to financial support, the average marginal effect of PPS‐M for adult daughters is positive and significant (AME = 0.109, SE = 0.052), whereas the average marginal effect of PPS‐M for adult sons is positive and not statistically significant (AME = 0.026, SE = 0.040). To put these effect sizes into context, we compare them to differences in average predicted probabilities of entry into parenthood between married and cohabiting adult daughters and sons (about 0.05 for adult daughters—specifically, 0.049 in the PPS‐T model and 0.051 in the PPS‐M model, and about 0.088 for adult sons in both PPS‐T and PPS‐M models). Partnership status is a well‐known predictor of entry into parenthood, with married couples being more likely to become parents than their cohabiting counterparts. Thus, the average marginal effect of PPS‐T for adult daughters is 1.29 time larger than the gap between those married and those cohabiting (1.29 = 0.067/0.051). The average marginal effect of PPS‐M for adult daughters is 2.22 times larger than the gap between those married and those cohabiting (2.22 = 0.109/0.049). This comparison suggests that the effect sizes of PPS‐T and PPS‐M are large and statistically significant for adult daughters and negligible and nonsignificant for adult sons.

FIGURE 2

Average yearly marginal effects of PPS‐T and PPS‐M on the probability of entry into parenthood for adult daughters and sons. The average marginal effects are calculated per 1‐year time interval using models presented in Table C4 (adult daughters) and Table C5 (adult sons) in the Supporting Information. The difference in average marginal effects for PPS‐T between adult daughters and sons is 0.069 with SEs of 0.028. The difference in average marginal effects for PPS‐M between adult daughters and sons is 0.100 with SEs of 0.047. PPS‐M, propensity for parental support in money; PPS‐T, propensity for parental support in time

We perform two additional tests to investigate whether these patterns of associations are different between adult daughters and sons and depending on the transfer type. We test for gender differences in the average marginal effects of PPS‐T and PPS‐M and find the gender differences to be statistically significant for both measures. Then, to test for transfer type differences within gender, we compare the average marginal effects of PPS‐T and PPS‐M and find that they are not statistically different for adult daughters or sons. This first set of results shows that the descriptive association between PPS measures and entry into parenthood are still significant once we account for sociodemographic controls and timing‐related variables for adult daughters, but not for adult sons. We next turn to the average predicted probabilities to better assess the effect size of the relationship between PPS measures and entry into parenthood for adult daughters.

Figure 3 shows average yearly predicted probabilities of having a first birth at different values of both the PPS‐T (Panel A) and PPS‐M (Panel B) measures for adult daughters. Starting with parental time support, adult daughters with higher PPS‐T values are more likely to become mothers. Illustrating the effect size, adult daughters with parental characteristics that suggest a moderately low likelihood of providing any time support (PPS‐T = 0.320, −1SD below the mean) have a .065 yearly predicted probability of entry into parenthood. In contrast, adult daughters who have parents with the resources and characteristics to signal a moderately high probability of providing any time support (PPS‐T = 0.688, +1SD above the mean) have a .089 yearly predicted probability of entry into parenthood. This means that, on average, adult daughters with a PPS‐T value of 1SD above the mean are 37% more likely to become parents than adult daughters with a PPS‐T value of 1SD below the mean. Turning to financial parental support, Figure 3 shows that adult daughters' likelihood of entry into a first birth positively responds to increasing values of the PPS‐M measure. Specifically, the average yearly predicted probability of having a first birth among adult daughters varies from approximately .066 when PPS‐M is moderately low (PPS‐M = 0.156, −1SD below the mean) to approximately .087 when PPS‐M is moderately high (PPS‐M = 0.350). Thus, on average, adult daughters with a PPS‐M value of 1SD above the mean are 33% more likely to become parents than adult daughters with a PPS‐M value of 1SD below the mean. In summary, we find that adult daughters who are more likely to receive either any time transfer (PPS‐T) or any money transfer (PPS‐M) from their parents are also more likely to become mothers. For adult sons, we find that the probability of receiving any time (PPS‐T) or any money (PPS‐M) transfers from their parents does not significantly affect their predicted probability of having a first birth. We return to these gendered results in the discussion.

FIGURE 3

Average yearly predicted probabilities of entry into parenthood at different levels of PPS‐T and PPS‐M measures for adult daughters. The average predicted probabilities are calculated per 1‐year time interval using models presented in Table C4 in the Supporting Information. For PPS‐T, −2SDs = 0.136, −1SD = 0.320, Mean = 0.504, +1SD = 0.688, +2SD = 0.872. For PPS‐M, −2SDs = 0.060, −1SD = 0.156, Mean = 0.253, +1SD = 0.350, +2SD = 0.446. See Table 1 for additional descriptive statistics for PPS‐T and PPS‐M. PPS‐M, propensity for parental support in money; PPS‐T, propensity for parental support in time

DISCUSSION

This study fills a critical gap in our understanding of how likelihood of parental support shapes adult children's fertility outcomes in the US context. Combining longitudinal data from the PSID main interview with the CAH and R&T supplementary files, we first construct measures of parental propensities to provide time and money support and then examine how these measures shape entry into parenthood differently for adult sons and daughters and by type of support. We expected that (a) having parents with the capacity to provide time and money support would be positively associated with entry into parenthood; (b) adult daughters would be more responsive than adult sons to parental propensity to provide support; and (c) adult daughters would be more responsive to time parental support than financial support.

Across different specifications and for both time and money transfers, we find that for adult daughters, parental propensity to provide support positively influences entry into motherhood. In contrast, while our descriptive results for adult sons suggested a positive association between fertility and the parental propensity for financial support, we find no such statistically significant pattern in our regression analyses. Furthermore, we find that these gender differences in the association between the propensity for parental support and fertility are statistically significant. This highlights how men and women respond differently to their parents' propensity to support them as they enter parenthood. While we expected adult daughters to be more responsive to time support, we find no statistically significant difference in the associations of adult daughters' fertility with parental propensities to provide time versus money. This suggests that both types of support equally influence adult daughters' entry into parenthood.

These findings contribute to the literature on intergenerational transfers and work‐family outcomes in several ways. First, the gender differences we find reflect the unequal strain of parenthood for men and women, as well as persistent societal beliefs about gendered family responsibilities. The women in our sample are responsive to any form of parental support, yet we find no such statistically significant relationship for men. While we can only speculate about the reasons for these gendered findings, one possible explanation is that women correctly anticipate that they will shoulder a larger share of the costs of parenthood and are thus more responsive to the possibility of receiving parental support. The null findings for men go hand in hand with notions of good fatherhood that are centered around breadwinning rather than caretaking (Nomaguchi & Milkie, 2020). If men are not expected to take on an equal share of parenting, parental support may simply be less relevant to their entry into parenthood. Echoing this interpretation, Liefbroer (2005) found that compared to men, women anticipate that the costs of parenthood in terms of career opportunities and partnership quality will be much larger, and that the rewards will be smaller. Additionally, some research suggests that women are more likely than men to spend income and financial transfers on children, and therefore money received from parents may be more relevant to women's childbearing decisions (Armand et al., 2020; Bhupal & Sam, 2014). Another potential explanation is that men fail to adequately anticipate the costs of parenthood. Qualitative studies on parenthood expectations have shown that men focus on their parenting role when their child is older, and overlook the more taxing early childhood phase (Dayton et al., 2016). Finally, research shows that emotional support is another important parental resource that may positively influence adult children's fertility outcomes (Mathews & Sear, 2013; Schaffnit & Sear, 2017a; Waynforth, 2012). Yet, emotional support tends to be gendered, and stronger between adult daughters and mothers (Fingerman et al., 2020b). Thus, it is possible that gender differences in emotional support may partially explain some of our gendered findings on the associations between parental support and adult children's fertility. While our study cannot disentangle these potential mechanisms, our results clearly show that there is a gendered relationship between anticipated parental support and entry into parenthood in the United States. Furthermore, our results echo the Europe‐based findings that grandparents matter most for adult children's entry into parenthood in contexts with weak family policies (Rutigliano, 2020).

Second, building on intergenerational ties research, we define and measure parental support while acknowledging that parents' propensity to support their adult children is a function of several interrelated factors. Because of data limitations, previous research had often resorted to using a single measure to proxy future parental capacity to provide help. We argue that these unidimensional measures might conflate parental resources and constraints, and thus provide a weak signal of future parental transfers. This could explain why studies using similar measures in different contexts have generated such a range of findings (see Table A1 in the Supporting Information). In this study, we leverage rich data from the R&T module that capture various parental characteristics, together with a two‐step approach to construct several measures of propensity to receive parental support. In supplementary analysis (available from the authors upon request), we replicated our study using unidimensional proxies of parental support, such as geographic proximity or vital status, and consistent with previous literature found mixed and mostly nonsignificant results. We cannot directly compare these additional results to past research in Europe because of differences in transfer measures and sampling strategy. Nevertheless, we speculate that these results highlight the importance of our methodological approach rather than differences across contexts in the association between parental support and adult children's entry into parenthood. Thus, we conclude that our approach allows us to draw a clearer conclusion and reconcile previous findings on the positive role played by parents in their adult children's entry into parenthood.

Third, our findings add to a growing literature that argues that studies of parental support and fertility should account for different forms of transfers, such as time and financial help (Sear, 2017). Because the R&T module provides quantitative measures of time and money transfers, we were able to disentangle how different types of parental support influence adult children's fertility decisions. Furthermore, our survey measure of time transfer is not restricted to childcare, and includes activities such as errands, rides, and chores. It is significant that a simple measure of time support that does not differentiate between tasks (i.e., grandparental childcare vs. running errands) matters for adult daughters' fertility. We find a similar statistical relationship with financial support. This is an interesting finding, as most of the previous literature had focused on the influence of large life‐cycle transfers, rather than recent and smaller financial transfers. The few studies that included a similar measure to ours found a negative and statistically significant association with adult children's fertility (Schaffnit & Sear, 2017b). This likely captures adult children's negative financial situations rather than a causal relationship between parental financial transfers and adult children's fertility. Indeed, in additional analysis mirroring this approach (available from the authors upon request), we replicated our study by using last year's financial transfers to childless adult children as a proxy for parental propensity to provide financial support and found a negative and statistically significant association for adult daughters. With our two‐step approach, which accounts for some of the endogeneity between parental support and fertility decisions, we found a positive and statistically significant relationship for adult daughters.

Although this study is among the first to analyze how the propensity of parental support shapes adult children's entry into parenthood, we must acknowledge some limitations. First, we do not have a sufficient sample size to explore how race and education might moderate the relationship between likelihood of parental support and adult children's entry into parenthood (see Table C6 in the Supporting Information). While we controlled for these variables, previous research suggests that parental support and childcare preferences vary widely across racial and educational lines in the United States (Fingerman et al., 2020a; Sarkisian & Gerstel, 2004). Thus, this is an important issue that should be addressed with a different data set. Second, as we measured parental characteristics at the 2013 interview, we were unable to capture changes in parental circumstances that occurred thereafter. To partially address this, our models control for the time that has elapsed since the 2013 wave and time‐varying variables at the adult child's household level, which together control imperfectly for possible changes. Nevertheless, this remains a limitation of using the R&T module, which provides a more representative picture of parental resources and family structure but is a one‐off supplementary survey. Third, our study is the first to focus on the associations between parental propensities to support in terms of time and money and adult children's entry into parenthood in the United States. While we test for differences in the role of money and time, we do not explore whether they are substitute or complementary resources for adult children. This is an important question that future work should address. Fourth, although our data have rich measures of directed time and money transfers, we do not have specific information on how that time is spent or where the money goes. Thus, we cannot test whether there are specific activities or investments, which influence the adult child's fertility decisions more than others. This is particularly important for time investments, as we cannot separate the role of childcare provision from other types of activities. Future research, with different and richer data, should address this point. Fifth, as the PSID is not representative of new waves of immigrants, the findings cannot be generalized to these subgroups of the US population. Finally, because we do not have parent‐specific data about both time and money transfers, we cannot test how the gender of the (grand)parents influences adult child's fertility decisions. Previous findings suggest that the gender of the (grand)parent may play an important role independently, but also in relation to the adult child's gender (Fingerman et al., 2020b). Specifically, grandmothers—and maternal grandmothers in particular—are found to be more involved than grandfathers in the care of grandchildren, especially at high levels of childcare intensity (Hank & Buber, 2009; Thomese & Liefbroer, 2013). Furthermore, adult daughters value their own mothers' social support more strongly (Leahy‐Warren et al., 2012). Future research should further explore how maternal and paternal social support matters differently for adult daughters' and adult sons' fertility decisions.

Our study highlights that having parents with the capacity to provide time or money transfers affects entry into motherhood in the United States. These findings reflect the challenges women face in a country where the state is silent on work‐family reconciliation issues, and families are left to find individual solutions to navigate their work and family lives.

CONFLICT OF INTEREST

The authors declare no potential conflict of interest.

Appendix S1: Supporting information

Click here for additional data file.